Main

Highly regioselective catalytic protocols have served as powerful tools for the transformation of both simple and complex molecules into structures of profound utility across diverse fields such as pharmaceuticals and materials science1,2,3,4,5,6. One critical tenet in synthetic chemistry has been the effective incorporation of multifunctional synthetic handles within organic molecules7,8,9,10,11,12,13,14. These handles can be used to introduce structural diversity by serving as general reactive coupling partners or reagents, substantially broadening the range of strategies available to synthetic practitioners (Fig. 1a). Therefore, the ability to selectively install versatile handles into densely functionalized substrates offers compelling advantages, such as allowing for general arrays of diverse late-stage modifications15,16. As such, these approaches represent powerful strategies for enabling library syntheses and facilitating accelerated exploration of chemical space.

Fig. 1: Concepts and strategies for the installation of synthetic handles within arenes and their utility.
Fig. 1: Concepts and strategies for the installation of synthetic handles within arenes and their utility.
Full size image

a, Generic utility of multifunctional synthetic handles within chemical transformations. b, Regiochemical challenges associated with C(sp2)–H functionalization reactions. Pinacolboronic esters have proven broad synthetic utility in a variety of transformations; however, installation by metal-catalysed C(sp2)–H borylation has required innovative strategies for regiocontrol. Bpin, 4,4,5,5-tetramethyl-1,3,2-dioxaborolane. c, Synthetic utility of benzylic trimethylsilanes as masked anion equivalents. Nu, nucleophile. d, Overview of mechanistic pathways for the proposed ruthenium-catalysed C(sp2)–H functionalization leading to regiodivergent outcomes. 1°, primary; 2°, secondary; 3°, tertiary. e, Regiodivergent silylmethylation enabled by a single ruthenium catalyst, [(tBuCN)5Ru(H2O)](BF4)2, providing site-selective access to both ortho- and meta-silylmethylated arenes. f, Reaction conditions for both meta- and ortho-silylmethylation.

Given their synthetic versatility, boronic esters and acids have emerged as exemplar multifunctional handles for a plethora of synthetic manipulations17,18,19. The C(sp2)–H borylation of arenes represents an efficient example of their synthetic preparation, which has been achieved using metal-catalysed20,21,22, metal-free23 and radical-mediated processes24 (Fig. 1b). However, maintaining control over regioselectivity within these reactions presents a primary challenge and has necessitated the development of innovative strategies25.

Alternatives have emerged to complement the utility of boron-based compounds, for example, silicon-based synthetic handles have attracted interest for their ability to serve as useful pro-nucleophiles and coupling partners26,27,28,29. The synthetic applicability of silyl groups is analogous to that of boronic esters with potentially wide synthetic suitability for a varied selection of synthetic transformations. Benzylic silanes in particular hold considerable promise due to their ability to serve as masked benzylic anion equivalents (Fig. 1c)30,31,32,33,34. This reactivity allows the silyl group to be readily unmasked using fluoride or alkoxide reagents, unveiling species capable of acting as general nucleophiles towards a diverse set of electrophiles. Moreover, the utility of benzylic silanes extends beyond anion reactivity, with this class of compounds additionally serving as robust precursors for benzylic radicals35,36. However, classical synthetic strategies for their preparation have required the use of highly reactive halosilanes, such as chlorotrimethylsilane, and organometallic reagents37,38,39,40,41.

Considering the growing synthetic utility exhibited by silyl synthetic handles, we questioned whether ruthenium-catalysed C(sp2)–H functionalization might serve as an effective method for installing silylmethyl groups. We reasoned that, with a ruthenium catalyst, we could achieve site-selective installation of this class of synthetic handle at both ortho and meta sites within arenes bearing N-based directing groups42,43. This approach relied on the emerging evidence supporting the ability to obtain high levels of site selectivity in ruthenium-catalysed C(sp2)–H functionalization reactions. For instance, prevailing studies suggest that aryl halide and primary alkyl halide electrophiles preferentially react to give ortho-C(sp2)–H functionalization, while secondary and tertiary alkyl halides favour meta addition (Fig. 1d)44,45,46,47,48. The regiocontrol in these transformations is therefore commonly predicated on the structure of the electrophile, thus providing a framework for precision functionalization. We posited that if these observations hold universally, we could strategically use primary and secondary halo(silyl)methane reagents for a regiodivergent set of C(sp2)–H silylmethylation reactions. While differentiating between ortho and meta positions was feasible based on substrate design, we recognized the challenge of obtaining good reactivity with sterically encumbered electrophiles (for example, SiMe3, A-value = 2.5) as well as limiting the formation of over-addition products.

In this report, we present ruthenium-catalysed procedures for the site-selective installation of these useful silyl synthetic handles at ortho and meta sites within arenes bearing N-based directing groups and demonstrate their utility through a series of synthetic transformations. Regioselective installation was achieved using (bromomethyl)trimethylsilane for ortho selectivity and bis(trimethylsilyl)chloromethane for meta selectivity (Fig. 1e). In each case, optimized reaction conditions allowed for both excellent levels of reactivity as well as regioselectivity (Fig. 1f). These synthetic inventions underscore the powerful capabilities of ruthenium catalysis and the broad utility of silicon-based synthetic handles that serve to expand the breadth of downstream structures.

Results

Identification of ortho-silylmethylation reaction conditions

To discover reaction conditions for ortho-silylmethylation, we systematically assessed key reaction variables such as choice of (pre)catalyst, base, additives and solvent. Principally, ortho-alkylation reactions using halide electrophiles have used (pre)catalysts based on [(p-cymene)RuCl2]2 and its derivatives due to their widespread commercial availability. However, this class of (pre)catalyst possesses a substantial barrier to activation and consequently necessitates the use of high temperatures45 (≳100 °C) or light irradiation46. For ortho-silylmethylation, we selected (bromomethyl)trimethylsilane (3) as electrophile and 2-phenylpyridine (2a) as a model substrate (Fig. 2). Reaction at 80 °C in THF using [(p-cymene)RuCl2]2 or analogues such as [(C6H6)RuCl2]2 gave only low levels of reactivity (17% and 14% yield of 4a, respectively, Supplementary Table 7). Key for high levels of reactivity was the use of [(tBuCN)5Ru(H2O)](BF4)2 (RuAqua, 1) as (pre)catalyst, which gave 4a in excellent yield (93%). We recently reported the robust air and moisture stability of 1 as well as its diverse reactivity enabled by its more labile ligand sphere49.

Fig. 2: Application of ruthenium-catalysed ortho-silylmethylation to a range of arenes.
Fig. 2: Application of ruthenium-catalysed ortho-silylmethylation to a range of arenes.
Full size image

Reaction scope using (bromomethyl)trimethylsilane (3). Reactions were performed on a 0.25-mmol scale using 5 mol% [Ru] unless otherwise noted. All yields correspond to isolated compounds. a10 mol% [Ru] and 60 mol% PhP(O)O2K2 were used. b36-µmol scale. c180-µmol scale. d87-µmol scale. TBDMS, tert-butyldimethylsilyl; PMB, para-methoxybenzyl; Ac, acetyl; Boc, tert-butyloxycarbonyl; WHO, World Health Organization.

While polar aprotic solvents such as N-methyl-2-pyrrolidone (NMP) have been used previously to facilitate ruthenium-catalysed ortho-alkylation reactions47, we found that ortho-silylmethylation could be performed in a range of solvents, including ethyl acetate, dioxane and THF (Supplementary Table 6). Crucial to both high levels of reactivity and selectivity was the addition of NaI. We observed that in the absence of NaI, considerable meta-silylmethylation occurred as well as the formation of over-addition products (Supplementary Tables 1–4).

Applicability of ortho-silylmethylation

To assess the generality of the developed ortho-silylmethylation reaction, a selection of arenes bearing N-heterocycle groups were subjected to the reaction conditions (Fig. 2). Free benzylic alcohol 2b and analogues with silyl (2c), para-methoxybenzyl (2d) and acetyl (2e) protecting groups all underwent selective ortho-silylmethylation in high-to-excellent yields. Similarly, 4-substituted free aniline 2h and phenol 2i were well tolerated, giving the corresponding silylmethylated species 4h and 4i in good yields (55% and 62%, respectively). 1,2,5-Trisubstituted arene 2j was also ortho-silylmethylated on a 1-mmol scale to give 4j in excellent yield (79%), highlighting the ability of the developed protocol to functionalize polysubstituted substrates. Arenes bearing meta electron-withdrawing (2k) and electron-donating (2l) substituents underwent productive functionalization, albeit with 2l requiring increased catalyst loading (10 mol%). Amides 2m,n and esters 2o,p, including those containing structural units derived from biologically active compounds, also participated in ortho-silylmethylation, generally in good-to-excellent yields.

We recognized that it was key to demonstrate the value of this synthetic invention in both the context of broad functional group tolerance and its applicability across a diverse range of structures. Therefore we opted to look beyond pyridine-based N-heterocycles and explored the reactivity of pyrimidine, pyrazole, oxazoline, imine and related structures. Such broad applicability would be key to enable streamlined syntheses of a wide array of structures and thus allow accelerated discovery and exploration of chemical space. Accordingly, pyrimidine 2q was subjected to the established reaction conditions, selectively giving the expected product 4q in good yield (65%). Similarly, pyrazoles 2r–t underwent successful ortho-silylmethylation to give 4r–t in moderate-to-good yields (36–69%). The formal ortho-silylmethylation of ketones was facilitated using imines as temporary directing groups. Specifically, ketimines 2v–x derived from 4-(dimethylamino)aniline were used to give the corresponding ortho-functionalized products. After acid hydrolysis, successful isolation of the ortho-silylmethylated ketones 4v–x was achieved in appreciable yields (45–52%).

One powerful application of any C–H functionalization methodology is its capacity to functionalize targets bearing substantial functional group density and diversity. In particular, late-stage functionalization of biologically active compounds has served as a valuable tool for the rapid generation of medicinally relevant analogues. The ortho-silylmethylation protocol was therefore applied to a range of pharmaceutical targets (Fig. 1, bottom). Anti-retroviral compound atazanavir (2y) underwent successful ortho-silylmethylation to give 4y in a low but useful yield (25%). Imidazo[1,2-a]pyridine-containing zolimidine (2z) and zolpidem analogue 2aa similarly reacted to selectively give 4z and 4aa in yields of 36% and 16%, respectively. Lastly, the well-known anxiolytic diazepam (2ab) containing the tetrahydrobenzo[e][1,4]diazepin-2-one directing group underwent silylmethylation to give 4ab in good yield (40%), with the transformation tolerant of the halide handle within 2ab.

Identification of meta-silylmethylation reaction conditions

In contrast to the ortho reactivity obtained with (bromomethyl)trimethylsilane (3), the use of bis(trimethylsilyl)chloromethane (5) necessitated an elevated temperature (100 °C) and an aqueous solvent mixture (iPrOH–H2O, 3:2) for good levels of meta-selective reactivity (Supplementary Tables 9–16). Akin to ortho-silylmethylation, the use of RuAqua (1) as the (pre)catalyst proved effective for meta-silylmethylation using 5, allowing for the use of a single (pre)catalyst for both ortho- and meta-silylmethylation reactions. At the outset, we determined functional-group tolerance through a robustness screen50 under the optimized reaction conditions. For this screening process, we used 2-phenylpyridine (2a) as a model substrate and bis(trimethylsilyl)chloromethane (5; Supplementary Table 16). This rapid survey of additive compounds highlighted the tolerance towards amide, carbamate and amine groups, amongst others.

Applicability of meta-silylmethylation

To further evaluate the generality of the developed meta-silylmethylation reaction, the optimized reaction conditions were applied to a selection of arenes bearing N-heterocycles (Fig. 3). Specifically, a range of substrates bearing pyridyl directing groups were subjected to the established reaction conditions. These reactions resulted in successful meta-silylmethylation using 5 to give the corresponding products 6a–m, generally in good-to-excellent yields. Despite the presence of the sterically demanding geminal trimethylsilyl groups within electrophile 5, installation could still be achieved at the meta position of para-substituted arylpyridines, as exemplified by the formation of products 6c–g. A range of ester-containing arenes were successfully meta-silylmethylated using 5 to give 6h–j in good yields (45–77%). The functional-group compatibility highlighted in the robustness screen was corroborated in the successful synthesis of tertiary amide 6k in high yield (70%). The methodology was successfully expanded to include other N-heterocycles, specifically, arylpyrimidines, which gave the corresponding meta-silylmethylated compounds 6o,p. Moreover, a selection of arylpyrazoles underwent successful functionalization, giving products 6q–t in moderate-to-good yields (38–50%).

Fig. 3: Application of ruthenium-catalysed meta-silylmethylation to a range of arenes and utility in pharmaceutical synthesis.
Fig. 3: Application of ruthenium-catalysed meta-silylmethylation to a range of arenes and utility in pharmaceutical synthesis.
Full size image

Reaction scope using bis(trimethylsilyl)chloromethane (5). Reactions were performed on a 0.25-mmol scale using (5 mol%) [Ru] unless otherwise noted. All yields correspond to isolated compounds. The reaction yield for the [NH4][Ce(NO3)6] oxidation of 6x on a gram scale was 57%, for Pinnick oxidation the yield was 70% and for amidation using cyclopropylamine the yield was 72%. aIsolated as the desilylated compound following reaction with excess tetrabutylammonium fluoride. b10 mol% [Ru] was used. Cy, cyclohexyl.

This ruthenium-catalysed method was further applied in the synthesis of the pharmaceutical compound losmapimod (8), which holds promise for the treatment of facioscapulohumeral muscular dystrophy51. Initially, arylpyridine 2ac was meta-silylmethylated using the established reaction conditions to selectively give a multigram quantity of product 6x in good yield (2.4 g, 54%). This compound 6x was then oxidized to a carboxylic acid, followed by amide coupling with cyclopropylamine to give 7. Application of the ortho-silylmethylation conditions to 7 using silane 3 in NMP selectively gave losmapimod (8) in 52% yield (Fig. 3, bottom).

Synthetic utility of benzyltrimethylsilanes

Selective C(sp2)–H silylmethylation offers a synthetically useful step for the generation of benzylic trimethylsilanes that can be further elaborated by several strategies to produce diverse compounds. To highlight the synthetic utility of the products formed from both ortho- and meta-silylmethylation reactions, these were subjected to several different functionalization reactions (Fig. 4a). ortho-Silylmethylated compound 4a was subjected to the conditions reported by Reidl and Bandar for nucleophilic aromatic substitution (SNAr) coupling with 4-cyanopyridine (9), which gave coupled product 10 in good yield (55%)33. Similarly, benzylic anion reactivity was achieved using the conditions of Das and O’Shea30, which enabled nucleophilic addition into aldehyde 11 to give homobenzylic alcohol 12 in excellent yield (82%). Similar reactivity was achieved when applied to ketone 13, which gave the corresponding alcohol 14, once again in high yield (70%). para-Selective C(sp2)–H benzylation of (diacetoxyiodo)benzene 15 was achieved by reaction of 4a in the presence of excess trimethylsilyl triflate to give 16 in moderate yield (29%)52. The benzylic trimethylsilane 4a also served as a suitable precursor for Peterson-type olefination using imine 19 to give alkene 20 in excellent yield and diastereoselectivity (74%, >99:1 E/Z)34. The ortho-silylmethylation reaction could also serve as a valuable procedure for the selective formal ortho-methylation of arenes by reaction of 4a with excess tetrabutylammonium fluoride to give ortho-methyl compound 21 in excellent yield (90%).

Fig. 4: Synthetic utility of benzylic trimethylsilane and benzylic bis(trimethylsilane).
Fig. 4: Synthetic utility of benzylic trimethylsilane and benzylic bis(trimethylsilane).
Full size image

a, Reactions showing the silyl group as a useful ortho-synthetic handle for diverse transformations. b, Reactions showing the bis(silyl) group as a useful meta-synthetic handle for diverse transformations. Benzylic anion reactivity was unveiled using fluoride or siloxide (KOSiMe3), reacting through SNAr towards cyanopyridine or with pyridine N-oxide. Unmasking the silyl group also allowed for nucleophilic addition into carbonyl electrophiles. The reaction of 4a with excess tetrabutylammonium fluoride gave ortho-tolylpyridine 21, the product of formal ortho-methylation, while the reaction of 6a gave meta-tolylpyridine 30, the product of formal meta-methylation. Reaction conditions for the direct arylation of 4a to give 18: 17 (1.0 equiv.), 1 (10 mol%), KOAc (30 mol%), K2CO3 (2.0 equiv.), NMP (1.0 M), 35 °C, 24 h. Reaction conditions for the synthesis of 28 in a one-pot sequence: 17 (1.0 equiv.), 1 (5 mol%), CyCO2Li (10 mol%), Li2CO3 (1.2 equiv.), iPrOH–H2O (3:2, 1.0 M), 100 °C, 4 h. 18-c-6, 18-crown-6; DMSO, dimethylsulfoxide.

Analogous to the reactivity observed using the ortho-silylmethyl synthetic handle within 4a, the geminal bis(trimethylsilyl)methane group in 6a could also be used as a pro-nucleophile (Fig. 4b). To illustrate, 6a underwent reaction with pyridine N-oxide (22) to give the coupled product 23 in good yield (61%)32. Application of Reidl and Bandar’s conditions enabled SNAr with 9 to give the coupled compound 24 in high yield (71%)33. Iterative SNAr using 9 and 1,4-addition with acrylamide 25 was achieved under similar conditions to give 26 in appreciable yield (46%). Formal arene meta-formylation to produce 27 was realized following conversion of the gem-silyl handle within 6a using excess ammonium ceric nitrate. Direct olefination of the gem-silyl handle using imine 19 gave E-alkene 29 in high yield and excellent diastereoselectivity (78%, >99:1 E/Z)31. Similarly to the ortho-silyl group, the meta-silyl handle in 6a also underwent protodesilylation using excess tetrabutylammonium fluoride to give meta-methyl compound 30 in excellent yield (90%). This reactivity therefore demonstrates a viable strategy for the formal C(sp2)–H meta-methylation of arenes. Taken together, the meta-selective silylmethylation reaction provides generic access to a broad suite of functional groups, thus offering a transformative approach to the diversification of aromatic compounds.

Mechanistic considerations

Both ortho- and meta-selective C(sp2)–H functionalization reactions using ruthenium catalysts have been proposed to proceed through intermediate cyclometallated species (Fig. 5a). Mechanistic hypotheses have suggested the involvement of mono- and biscyclometallated ruthenium(II) species (for example, Int-I to Int-III) as key reactive intermediates that are formed before reaction with halide electrophiles45,46,47. Based on density functional theory calculations, Ackermann and co-workers proposed that monocyclometallated ruthenium(II) complexes (that is, with additional acetate and 2a coordination, Int-II) react favourably by inner-sphere single-electron transfer (SET) with 1-bromohexane (Gibbs energy of activation, ∆G‡ = 16.6 kcal mol−1) and tert-butyl bromide (∆G‡ = 15.6 kcal mol−1)53. These values contrast with those determined for biscyclometallated species (for example, Int-III). For this class of complex, the reactions with primary and tertiary alkyl bromides were calculated to have considerably higher barriers (∆G‡ = 22.9 and 22.1 kcal mol−1, respectively). Therefore, these calculations suggest a marked difference in the reactivity of the monocyclometallated and biscyclometallated species.

Fig. 5: Mechanistic considerations for the key ruthenacycle intermediates responsible for reactivity.
Fig. 5: Mechanistic considerations for the key ruthenacycle intermediates responsible for reactivity.
Full size image

a, Proposed ruthenium intermediates include monocyclometallated ruthenium species with and without the interaction of other arenes. Alternatively, two C(sp2)–H activation steps give a bisruthenacycle that can serve as a reactive intermediate. b, Stoichiometric reactivity of monocyclometallated ruthenacycle 31. Reaction conditions: (i) 3 (7.0 equiv.), KOAc (10 equiv.), (CD3)2CO (0.1 M), 40 °C, (ii) with added 2j (1.2 equiv.), (iii) with added 34 (1.2 equiv.); (iv) 5 (5.0 equiv.), KOAc (10 equiv.), (CD3)2CO (0.1 M), 40 °C, (v) with added 2j (1.2 equiv.), (vi) with added 34 (1.2 equiv.). Reactivity was only observed when an additional equivalent of arene bearing ortho-C(sp2)–H bonds was present. c, Independent synthesis of monoruthenacycle 36 to confirm the absence of its formation in mechanistic experiments. An X-ray structure of 36 is shown. with thermal ellipsoids at the 50% probability level.

To probe these mechanistic proposals in the context of ortho- and meta-silylmethylation, monocyclometallated ruthenium(II) complex 31 was subjected to a series of control experiments (Fig. 5b). When complex 31 was heated at 40 °C with ortho-selective bromide 3, no reaction was observed (Fig. 5b(i)). In contrast, reaction in the presence of added arylpyridine 2j gave the ortho-silylmethylation product 4j (Fig. 5b(ii)). The addition of arylpyridine 2j allows for further cyclometallation and formation of a biscyclometallated species (that is, complex 40), which was observed in the reaction mixture (Supplementary Figs. 15 and 16). Similarly, to mimic the intermediate species proposed by Ackermann and co-workers, we conducted the reaction with added 2-(2,6-difluorophenyl)pyridine (34; Fig. 5b(iii)). The ortho-fluorine substitution within 34 precludes cyclometallation while still allowing for pyridine coordination; however, no silylmethylation was observed.

Similarly for meta-silylmethylation reactivity, in the stoichiometric reaction of monocyclometallated ruthenium(II) complex 31 at 40 °C with meta-selective chloride 5, neither product 6l nor its cyclometallated analogue (for example, 36) was observed (Fig. 5b(iv)). However, in the presence of added arylpyridine 2j, the same reaction gave the meta-silylmethylation product 6l (Fig. 5b(v)). This outcome was not replicated when the same experiment was conducted with 34 (Fig. 5b(vi)). The independent synthesis of the cyclometallated ruthenium(II) complex bearing product 6l facilitated direct comparison of the reaction spectra to exclude the formation of this type of species in the absence of added arylpyridine 2j (Fig. 5c). Importantly, these experiments were reproducible with analogous monocyclometallated complexes bearing distinct substituents (Supplementary Fig. 17). These mechanistic observations were suggestive of a key role for the formation and reactivity of a biscyclometallated species, which stands in contrast to previously proposed pathways.

Reactivity of biscyclometallated ruthenium(II) complexes

Following the observed inactivity exhibited by monocyclometallated ruthenium(II) complexes, we next examined the stoichiometric reactivity of a biscyclometallated ruthenium(II) species. Thus, we initially synthesized biscyclometallated complex 33 from its monocyclometallated precursor 37 (Fig. 6a). This complex was treated with excess (bromomethyl)trimethylsilane (3) in [D6]benzene at 29 °C, tracking the reaction progress by 1H NMR spectroscopy (Fig. 6b). Notably, the product 4ac and its ruthenium(II) coordinated analogue 38 were observed, without the need for any additives or base, and in less than 1 h.

Fig. 6: Stoichiometric reactivity of a reactive bisruthenacycle.
Fig. 6: Stoichiometric reactivity of a reactive bisruthenacycle.
Full size image

a, Synthesis of bisruthenacycle 33 from monoruthenacycle 37 and comparison of the CVs of both species (see Supplementary Figs. 41 and 42 for full CVs). Fc, ferrocene. b, Stoichiometric reaction of bisruthenacycle 33 with (bromomethyl)trimethylsilane (3). c, Temporally resolved Evans NMR experiment for the reaction of 33 with bis(trimethylsilyl)chloromethane (5). A capillary containing [D6]benzene was inserted within the NMR tube containing the reactants. The consumption of bisruthenacycle 33 was monitored over time by 1H NMR spectroscopy and the divergence in the δ(solution) versus δ(capillary) signals measured from the maximum peak positions at each time point. D, diffusion coefficient, measured by diffusion-ordered NMR spectroscopy; µ, magnetic moment, measured in Bohr magnetons µB (µB = 9.27 × 10−24 J T−1); calc, calculated; exp, experimental.

We similarly explored the use of biscyclometallated ruthenium(II) complex 33 to understand the reactivity exhibited in the meta-silylmethylation reaction using bis(trimethylsilyl)chloromethane (5). However, in complex 33, the site where electrophile 5 would initiate carbon–carbon bond formation is impeded by a methyl substituent (para to ruthenium) and thus no reactivity was anticipated. Despite this, when 33 was treated with excess 5 at room temperature, we noted the formation of a paramagnetic species in solution. This was identified by characteristic 1H NMR resonances that fell outside the typical sweep width (that is, δ1H = 72.0, –14.8 and −19.3 ppm, see Supplementary Fig. 34).

To garner information on the number of unpaired electrons within the newly formed paramagnetic ruthenium species, we used the Evans NMR method (Fig. 6c)54. The reaction of 33 and 5 in [D6]benzene was conducted in a J Young NMR tube incorporating a sealed inner capillary containing [D6]benzene to serve as a reference. We monitored the progress of the reaction in real time using 1H NMR spectroscopy and observed a clear divergence in the resonance frequency of the [D6]benzene residuals (that is, the signal from the reaction mixture versus the signal from the capillary). From this, we determined a spin-only magnetic moment μeff of 1.74 μB, which is supportive of a single unpaired electron (Supplementary Equation (1))54. Given the theoretical electronic configurations of an octahedral ruthenium(II) species (d6, diamagnetic) and a singly oxidized octahedral ruthenium(III) species (d5, paramagnetic), these observations are coherent with the formation of a ruthenium(III) species. Based on the diffusion-ordered NMR spectra and mass spectrometry data, the reaction of 33 and 5 generated a putative ruthenium(III) dimer 39 (Supplementary Figs. 37–39).

Given the limitations imposed by the densely substituted biscyclometallated species 33 to undergo meta-selective functionalization, we questioned the feasibility of producing an unsubstituted analogue. Biscyclometallated ruthenium(II) complex 40 was synthesized from its monocyclometallated analogue 31 and arylpyridine 2j in useful yield (35%; Fig. 7a). Comparative cyclic voltammograms (CVs) of monocyclometallated ruthenium(II) complexes 31 and 37 and biscyclometallated ruthenium(II) complexes 40 and 33 showed distinctly different redox potentials. Specifically, the Ru(III)/Ru(II) redox potentials in 40 and 33 are E1/2 = −0.37 and −0.51 V, respectively, versus ferrocene/ferrocenium, while the Ru(III)/Ru(II) redox potentials in monocyclometallated 31 and 37 are E1/2 = 0.49 and 0.29 V, respectively, versus ferrocene/ferrocenium. The distinct differences in the redox potentials underscore the enhanced reducing properties of the biscyclometallated ruthenium(II) complexes compared with their monocyclometallated analogues (ΔE1/2 ≈ 0.8 V) and are consistent with the observed reactivity towards electrophiles.

Fig. 7: Synthesis and reactivity of a bisruthenacycle bearing available meta sites for functionalization.
Fig. 7: Synthesis and reactivity of a bisruthenacycle bearing available meta sites for functionalization.
Full size image

a, Synthesis of bisruthenacycle 40 from 31 and comparison of the Ru(II/III) redox potentials for both monoruthenacycles 31 and 37 and bisruthenacycles 40 and 33 (see Supplementary Figs. 39–42 for full CVs). The X-ray structures of 31 and 40 are shown with thermal ellipsoids at the 50% probability level. b, Stoichiometric reactivity of bisruthenacycle 40 with (bromomethyl)trimethylsilane (i) and bis(trimethylsilyl)chloromethane (ii). Reaction conditions (i) towards 4j: (bromomethyl)trimethylsilane (7 equiv.), C6D6–(CD3)2CO (3:2, 0.05 M), 40 °C. Reaction conditions (ii) towards 6l: bis(trimethylsilyl)chloromethane (18 equiv.), C6D6–(CD3)2CO (3:2, 0.05 M), 50 °C.

Having synthetically established access to 40, we treated this complex with ortho-silylmethylation reagent 3 at 40 °C, monitoring the reaction progress by 1H NMR spectroscopy (Fig. 7b, top). The formation of functionalized product 4j was observed in conjunction with the consumption of complex 40. Similarly, the reaction of meta-silylmethylation reagent 5 and complex 40 was also tracked using 1H NMR spectroscopy (Fig. 7b, bottom). In this instance, product 6l was observed alongside uncharacterized paramagnetic species, concurrent with the consumption of complex 40. The reactivities observed highlight the ability of the halide reagents 3 and 5 to readily engage with biscyclometallated complexes 40 and 33. In stark contrast, no reactivity was observed when these reagents were applied to monocyclometallated ruthenium(II) analogues 31 and 37. These findings are therefore supportive of a key role for biscyclometallated species during catalysis.

To gain further mechanistic insight, we measured the kinetic orders of the components in the reactions using variable-time normalization analysis55. For the meta-silylmethylation reaction, an order of 1 on the Ru catalyst, 0.2 on arene, 0.4 on electrophile 5 and −0.5 order on LiCl were observed (Fig. 8a), while for the ortho-silylmethylation reaction, an order of 1 on the Ru catalyst, 0.5 on arene, 0.7 on electrophile 3 and 0 order on NaI were observed (Fig. 8b).

Fig. 8: Variable-time normalization analysis for ortho- and meta-silylmethylation and overall mechanistic hypothesis.
Fig. 8: Variable-time normalization analysis for ortho- and meta-silylmethylation and overall mechanistic hypothesis.
Full size image

a,b, Normalized reaction plots to determine the kinetic orders for the ruthenium catalyst, arene, electrophile, lithium chloride and sodium iodide in the meta-silylmethylation (a) and ortho-silylmethylation (b) reactions. The progress in both reactions was monitored by gas chromatography with flame ionization detection using biphenyl as internal standard. c, Mechanistic overview of both ortho and meta selectivity in the silylmethylation reaction, diverting post-SET.

Mechanistically, both pathways proceed by sequential C–H activation, forming Int-III via the monocyclometallated intermediate Int-I (Fig. 8c). Int-III was confirmed as the active catalytic intermediate through comparative stoichiometric experiments, showing the need for biscyclometallation over activation by N coordination (Int-II). The measured kinetic orders on the arene substrate of 0.2 and 0.5 for meta- and ortho-silylmethylation, respectively, are consistent with a reversible C–H activation step. This was also evidenced by the observation of H/D exchange in reactions carried out in the presence of D2O (Supplementary Information). While Int-III possesses considerably more reducing power than its predecessor Int-I, the reduction potentials of typical alkyl halides remain out of reach for an outer-sphere mechanism56. It is therefore proposed that the reaction proceeds via an inner-sphere complex, where the oxidation potential of the electrophile is reduced to an accessible value57. This pathway agrees with the partial orders measured for the electrophiles, which are indicative of a reversible reaction of the alkyl halide with a catalytic intermediate preceding the rate-determining step. The following SET to Int-IV was confirmed by Evans NMR experiments as well as by the observation of 1H NMR chemical shifts consistent with its formation. The divergence in reactivity originates from the propensity for the alkyl radicals to be sequestered by Ru in Int-IV (ortho) to form Int-VII or to attack the arene system (meta) to give Int-V.

Conclusion

We have developed distinct site-selective synthetic methods for the installation of silylmethyl synthetic handles using ruthenium (pre)catalyst 1. The versatility of the resulting silylmethyl products has been highlighted through their application to several different transformations, including protodesilylation of the installed handle to allow for formal arene ortho- and meta-methylation. Moreover, the regiodivergent installation of silylmethyl groups enables an alternative synthetic route to the active pharmaceutical losmapimod (8). This type of synthetic strategy using silyl synthetic handles allows broad composability for the potential synthesis of derivative compounds along the synthetic path. Given that the PubChem database contains more than 6 million (hetero)aromatic compounds bearing a suitable N(sp2) for ortho-C(sp2)–H cyclometallation, these synthetic inventions provide substantial utility. Mechanistic investigations conveyed the key role of reducing biscyclometallated ruthenium(II) species for reactivity with both primary and secondary electrophiles, with Evans NMR experiments revealing the formation of a putative ruthenium(III) species that is likely reflective of relevant intermediates formed during the catalytic process. By expanding the mechanistic understanding within ruthenium-catalysed C–H functionalization, these findings provide long-awaited insights that will aid the development of future protocols.

Methods

General procedure for the meta-silylmethylation of arenes

Arene (0.250 mmol), bis(trimethylsilyl)chloromethane (59.0 mg, 0.300 mmol, 1.20 equiv.), [(tBuCN)5Ru(H2O)](BF4)2 (9.0 mg, 13.0 µmol, 5 mol%), lithium carbonate (22.0 mg, 0.300 mmol, 1.20 equiv.) and lithium cyclohexanecarboxylate (3.4 mg, 25.0 µmol, 10 mol%) were added to a vial, which was sealed (crimp-capped) and purged for ~40 s with N2. N2-sparged isopropanol–water (3:2, 250 µl) was injected into the vial using a syringe and the reaction mixture was heated at 100 °C for 4 h. The reaction mixture was then cooled to room temperature, the cap was removed and the solvent removed in vacuo before direct purification by flash column chromatography.

General procedure for the ortho-silylmethylation of arenes

Arene (0.250 mmol), (trimethylsilyl)bromomethane (41.8 mg, 0.250 mmol, 1.00 equiv.), [(tBuCN)5Ru(H2O)](BF4)2 (9.0 mg, 13.0 µmol, 5 mol%), potassium carbonate (69.0 mg, 0.500 mmol, 2.00 equiv.), potassium phenylphosphonate (17.5 mg, 75.0 µmol, 30 mol%) and sodium iodide (37.0 mg, 0.250 mmol, 1.00 equiv.) were added to a vial, which was sealed (crimp-capped) and purged for ~40 s with N2. Anhydrous THF (500 µl) was added using a syringe and the reaction mixture was heated at 80 °C for 18 h. The reaction mixture was then cooled to room temperature, the cap was removed and the solvent removed in vacuo before direct purification by flash column chromatography.

Procedure for the synthesis of bis[2-(2,5-difluorophenyl)-5-methylpyridine]ruthenium(II) bis(acetonitrile) (40)

In a glove box under an atmosphere of purified argon, 2-(2,5-difluorophenyl)-5-methylpyridineruthenium(II) tetrakis(acetonitrile) hexafluorophosphate (123.0 mg, 0.200 mmol), 2-(2,5-difluorophenyl)-5-methylpyridine (45.1 mg, 0.220 mmol, 1.10 equiv.), potassium carbonate (82.8 mg, 0.600 mmol, 3.00 equiv.), potassium acetate (14.0 mg, 0.100 mmol, 0.500 equiv.) and acetone (2.0 ml) were stirred in a sealed vial (crimp-capped) at 70 °C for 18 h. The reaction mixture was then cooled to room temperature and filtered (2.5 µm polytetrafluoroethylene filter). The solvent was concentrated in vacuo and pentane (20.0 ml) was slowly added whilst stirring until a solid precipitated. The precipitate was collected by filtration and washed with pentane to give bis[2-(2,5-difluorophenyl)-5-methylpyridine]ruthenium(II) bis(acetonitrile) (42.0 mg, 0.070 mmol, 35%) as an amorphous red solid. Note that complex 40 is air-sensitive and should be stored in a glove box under an atmosphere of purified argon.