Cycloaddition Chemistry of a Silylene‐Nickel Complex toward Organic π‐Systems: From Reversibility to C−H Activation

Abstract The versatile cycloaddition chemistry of the Si−Ni multiple bond in the acyclic (amido)(chloro)silylene→Ni0 complex 1, [(TMSL)ClSi→Ni(NHC)2] (TMSL=N(SiMe3)Dipp; Dipp=2,6‐iPr2C6H4; NHC=C[(iPr)NC(Me)]2), toward unsaturated organic substrates is reported, which is both reminiscent of and expanding on the reactivity patterns of classical Fischer and Schrock carbene–metal complexes. Thus, 1:1 reaction of 1 with aldehydes, imines, alkynes, and even alkenes proceed to yield [2+2] cycloaddition products, leading to a range of four‐membered metallasilacycles. This cycloaddition is in fact reversible for ethylene, whereas addition of an excess of this olefin leads to quantitative sp2‐CH bond activation, via a 1‐nickela‐4‐silacyclohexane intermediate. These results have been supported by DFT calculations giving insights into key mechanistic aspects.

The importance of cycloadditionr eactions of carbon-metal pbonds in catalysis cannot be overstated, paramount in processes such as alkene metathesis and cyclopropanation. [1] In these systems, highly reactivec arbon-metal multiple bonds can undergo formal [2+ +2] cycloaddition reactions with carboncarbon or carbon-heteroatom multiple bonds, typically through a[ 2 + +1] addition of the unsaturated species at the metal center,w ith subsequent chemistry leading to cyclopropyl or metathesized products. [1d, 2] Thus, such systems have been studied extensively since the seminald iscoveryo fastable carbene-metal complex by Fischer et al. [1a, 3] Notably,N -heterocyclic carbenes (NHCs), as well as other stable carbene systems, are broadly utilized as ligands in transition-metal chemistry, but their carbene-metal bonds are typically unreactivet oward CÀX p-bonds (X = C, heteroatoms). [4] More recently,N-heterocyclic silylene (NHSi)-transition-metal complexes, which contain a dative Si II !M s-bond, have seen considerable attention, [5] with examples in which the divalent silicon centeri si nf act directly involved in bond activation processes. [6] Examples of SiD!M multiple bonds have also seen considerable precedent in the literature. [7,8] Nevertheless, examples of cycloaddition chemistry of these moieties are somewhat sparse. Addition of alkynes and phosphalkynes to threefold-bonded SiÀOs, [9] and ketones and carbodiimides to threefold-bonded SiÀWs pecies have been reported, [10] somewhat comparable with the wide-ranging andv ersatile cycloaddition chemistry of homonuclear EÀE multiple bonds (E = Si, Ge, Sn). [11,12] Well-defined examples of the cycloaddition chemistry of SiÀMd ouble bonds are limited to reports from Sekiguchi et al. (Figure 1), in the [2+ +2] addition of alkynesa nd benzonitrile to aS i =Ti bond. [13] These remarkable reports are reminiscent of keys teps in the metathesis reactions of classical Schrock-type carbene complexes. [2] The exciting synthetic utility of SiÀMm ultiple bonds in this regard thus warrants considerable furtheri nvestigation, and could pave the way to new functional silicon-containing organic molecules which are otherwised ifficult to prepare. Indeed, metalsilylene complexes have been highlighteda sp otential keyi ntermediates in important catalytic processes such as hydrosilylation, [7c, 14] whereas unsaturated four-membered sila-metallacycles have also been inferred as intermediates in the catalytic ring-expansion of silacyclopropanes. [15] We wished to gain further insights into the chemistry of such metallacyclesb ye mploying the previously reporteda cyclic silylene-Ni 0 complex, TMS L(Cl)SiD!Ni(NHC) 2 1 ( TMS L = [(Dipp)(SiMe 3 )N] À ;D ipp = C 6 H 3 - [a] Dr. iPr-2,6;N HC = [DC{N(iPr)C(Me)} 2 ]), which possesses ad egree of SiÀNi multiple-bondc haracter. [16] We envisaged that cycloaddition chemistry with unsaturated organicc ompounds mayb e possible utilizing 1.H erein, we demonstrate that the SiÀNi multiple bond in 1 readily undergoes [2+ +2] cycloaddition reactions with ar ange of unsaturated CÀXb onds (X = C, N, O) The furtherc hemistry of isolated four-membered nickelasilacycles reveals both reversibility in this cycloaddition process for ethylene, as wella st he facile and stoichiometric activationo fi nert CÀHb onds. The computationally derived mechanism for the latter process with ethylene operates via ar eactive 1-nickela-4silacyclohexane intermediate, formed through af ormal [2+ +2+ +2] cycloaddition reaction of two ethylene molecules with 1.
As mentioned, 1 shows somed egree of backbonding from nickel to silicon,resulting in aS i ÀNi interaction with some multiple bond character (WBI = MBO = 1.29;W BI:W iberg bond indices;M BO:M ayer bond order). [16] This, alongside the relative polarity in this bond, led us to hypothesize that 1 should be reactive towards unsaturated CÀXb onds (X = C, N, O), given the prominence of such chemistry in reactive carbene-transition-metal complexes. Initial efforts towards this end focused on phenyl acetylene, and related reports for Ti=Si bonds from Sekiguchi et al. [13] Deeply red-purple-coloreds olutions are immediately obtained upon addition of one molar equiv of phenyla cetylene to 1,w ith quantitative formation of as ingle product suggested by 1 HNMR analysis of the reactionm ixtures. However,X -ray analysis of suitable single crystalso btained from reaction mixtures indicated that CÀHa ctivation of the acidic acetylene protonh ad in fact occurred at Si II ,y ielding aN i 0 p-complex of a( phenyl)(silyl)acetylene derivative( 2, Scheme 1). [17] Similarly,t he CÀHa ctivation product 3 was also obtained in the reactiono f1 with acetophenone, due to enolization of this ketone ( Figure S40, Supporting Information), indicating that the toleranceo f1 towards relatively acidic CÀH moieties is low.
To circumvent formation of an enolized product, 1 was reacted with p-CF 3 -benzaldehyde, as well as the relatedi mine, Nbenzylideneaniline. We found that in both cases the desired [2+ +2] cycloaddition products 4 and 5 were quantitatively formed, respectively.B oth compounds show highly unsymmet-rical environments for their NHC and TMS Ll igands in their 1 HNMR spectra, due to the rigid four-membered ring at their core. The molecular structure of each speciess hows activation of their formerlyC ÀO/N multiple bonds, and NiÀSi bond lengthsi nk eeping with single bonds (Figures S40 andS 41, Supporting Information), considerably lengthened relative to that in 1.A lthough meaningful 29 Si NMR datac ould not be obtained for 4,d ue to solubility issues in solvents with which 4 does not react, the 29 Si NMRs pectrum of 5 shows am arkedly highfields hift for its formerlyS i II center (1: d = 123.2; 4: d = À65.4 ppm).
Given that relative atomiccharges derived from an NPA(natural population analysis) of 1 indicates ap ositive relative charge at silicon (NPA Si =+ +1.14;N PA Ni = À0.59), it's not surprising that in the aforementioned cases the heteroatom binds silicon, formingp lanar and cyclic [SiNiCX] cores (X = Oo rN ). This is in contrast to the reactivity of Sekiguchi's titanium-silylene complex (Figure 1), which forms both regio-isomers in the reaction with benzonitrile. [13c] Observing the frontier orbitals of 1, one can see that the LUMO represents the p*-orbital of the SiÀ Ni bond, considerably weighted towards Si II ,w hereas the HOMO is af illed 3d orbital at Ni 0 .T hus, am echanism of initial oxygen/nitrogen donation to silicon, followed by Ni!Cn ucleophilic attack can be proposed. This wasc orroboratedb ya DFT analysis, in which the most favorable reaction coordinate involves ac oncerted [2+ +2] cycloaddition,d irectly leading to 4 and 5 in as ingle step ( Figure S45, Supporting Information). Interestingly,o bservingt he HOMOÀ1a nd the LUMO+ +1o f1 (À3.05 and À0.38 eV,r espectively), which are close in energy to the HOMO and LUMO (À2.84 and À0.45 eV,r espectively), [16] it is clear that these orbitals may too be involved in the reactivity of 1,b oth being of p-symmetry;t hese are notably similar to those orbitals in ap reviously reported silylene-Pt complex. [18] Compound 1 was treated with acetylene and ethylene to generate nickelasila-cyclobutene and -cyclobutane derivatives. Indeed,t he former is particularly interesting given previousi nvestigationsi nto the metallacyclobutene-vinyl carbene equilibrium for the 'all-carbon' system. [19] Addition of approx. one molar equivo fe ither ethylene or acetylene to solutionso f1 in diethyle ther or toluene, respectively,a tÀ78 8Cl ed to an immediate color change to bright yellow. 1 HNMR experiments carried out in parallel indicated that as ingle highly unsymmetrical speciesi sf ormed in both cases. Similart oc ompounds 4 and 5,w eproposed that the source of this asymmetryw as the formation of metallacycles, locking the formed speciesi na single conformer with an asymmetricalS i-center.S tructural analysiso ft he products from these reactionm ixtures confirmed that cycloaddition of acetylene and ethylene had occurred,forming nickelasila-cyclobutene and -cyclobutane derivatives 6 and 7,r espectively( Scheme2,F igure 2). Although no such speciesh ave been crystallographically characterizedf or Ni, [20] metallosila-cyclobutenes are known for Ti and Pd, [13,21] the Ti derivatives being generated through [2+ +2] cycloaddition (see above). The four-membered core of both 6 and 7 is planar,a si n4 and 5.T he CÀCd istance in the core of 6 (1.344 (3) ), however,i sc onsiderably contracted relative to that in 7 (1.556 (3) ), indicative of double-bond character.F urther analysiso fb ond lengths in nickelasila-cyclobutene 6 indicates typical CÀSi, SiÀNi, and CÀNi single bonds, thus indicating that there is no degree of formation of the vinyl carbene complex 6',w hich was found to be 32.8 kcal mol À1 higheri n energy than cyclic 6 (Scheme2). The SiÀCa nd CÀCb ond lengths in the cyclic core of 6 also compare well with the acetylene derived titanasila-cyclobutene reported by Sekiguchi, [13(b)] pertaining to af ormal metallacyclic structure, whereas related distances in both the SiMe 3 and nBu substituted derivatives reported by the same group pertain to ad egree of metal-silylidene alkyne p-complex character (that is ac ontracted TiÀSi distance and an elongated SiÀC alkyne distance). [13a] This observation is most likely caused by the increased steric profile of the alkyne substrates in the latter.I na na ttempt to observe as imilar trend foro ur system, 1 was reacted with 1,4-dimethoxy-2butyne, generating am uch more sterically crowded derivate of 6,n amely 6-OMe (Scheme 2). However,w ef ound that 6-OMe is essentially isostructural to 6 (see Figure S43 in the Supporting Information). Bulkiera lkynesd id not show any reaction with 1,e ven after heating.
The facile reactiono f1 with both ethylene anda cetylene is reliant on the aforementioned LUMO+ +1i nt his complex. That is, aD FT mechanistic analysis based on model complexes of 6 ( Figure S46, Supporting Information) and 7 ( Figure 3) suggests that both are formed through an initial h 2 -complex at this nickel-centered frontier orbital. This initial step is in fact reminiscento ft hat for the reaction of carbene-transition-metal complexes with alkenes and alkynes in now well-established multiple bond metathesis processes, [2] and contrasts with the concerted [2+ +2] mechanism for reactions with polar substrates in the formation of 4 and 5.F ollowing this initial [2+ +1] cycloaddition, as pontaneous ring expansion to Si proceeds, generating the metallasila-cyclobutane and -cyclobutene complexes 6 and 7. [22] Notably, complex 7 is only 2.5 kcal mol À1 lower in energy than 1,w ith a4 4.9 cal mol À1 K À1 entropic barrier to the formation of intermediary IM1 (Figure 3). This is born out experimentally:d espite the apparent CÀCs ingle bond present in the cyclic core of 7,d issolution of pure crystals of this compound in C 6 D 6 led to the generation of small amounts of both 1 and C 2 H 4 .T hus, to our surprise, the cycloaddition of ethylene to 1 is in fact reversible. Allowing as ample of 7 dissolved in C 6 D 6 to stand for 24 ha ta mbientt emperature led to the formationo fa1:1m ixture of 1 and as ingle new compound, 9 (see below), resulting from 1 reacting with two molar equiv of ethylene, given that no free ethylene could be observed in the 1 HNMR spectrum of this mixture. Although a computational investigation for the reaction of 1 with ethylene suggested that 7 may readily undergo af urtherc ycloaddition event with ethylene to form the six-membered metallasilacycle 8 (Scheme 3, Figure 3), we were surprised to find that the experimentally observed product from the reaction of 1 with excess ethylene is the alkene-Ni 0 p-complex 9. [23] It seems likely that compound 9 is formed throughasequential Nimediated b-hydride elimination/reductive eliminationr eaction from intermediary 8 (Scheme 3). [24] Circumstantial evidencef or the intermediate generation of 8 came from the isolationo f one or two crystalso ft his compound by the low-temperature reactiono f1 with excess ethylene, followed by storagea t À30 8Cfor two weeks. Althought his compound was highly unstable, allowing only for the collection of preliminaryc rystallographic data, the molecular structure, ascertaining the connec-Scheme2.The [2+ +2] cycloaddition reactions of ethylene and acetylene derivatives with 1,and the calculated energy for the isomerization of metallacyclobutene derivative 6 to silene-carbene complex 6'.  Chem.E ur.J. 2020,26,[1958][1959][1960][1961][1962] www.chemeurj.org  Figure S22 in the Supporting Information.
The molecular structureo f9 contains as ilyl-substitutede thylene unit in the coordination sphere of Ni 0 ,t he silyl group bearing the TMS L, Cl, and Et ligands. The CÀCa nd NiÀCb onds in the cyclic core of 9 are in keeping with those in related alkene-Ni p-complexes. The 1 HNMR spectrum of 9 in C 6 D 6 is very complex, both due to the asymmetricals ubstitution at the alkene and the silyl center,l eadingt od iasterotopic proton couplings (Figures S26 and S27,Supporting Information). This does, however,f urtherc onfirmt he connectivity in this species. The two Si IV centers yield very similar resonances in the 29 Si NMR spectrum of 9 at d = 5.2 and 7.8 ppm, the latter corresponding to the formerly Si II centera ss hown through at wodimensional 1 H, 29 Si HMQC NMR experiment ( Figure S31). Notably,t his reaction was shown to be reproducible for other alkenes, as shown by the formation of alkene p-complex 9-Ph in the reaction of 1 with two molarequivs. of styrene. The molecular structure of 9-Ph is essentially isostructural to that for 9 ( Figure S44), and is therefore also similartopreviously reported alkene-Ni p-complexes. It is worthy of note that the trans-conformation in 9-Ph is exclusively formed, most likely due to steric interactions between itss ilyl and phenyl substituents. These reactions, and particularly that with ethylene, whose CÀ Hb onds are relatively inert, perhaps point towards potential synthetic applicationsf or these remarkablec ycloaddition reactions, particularly when one notes that an asymmetric silicon center is generated. Synthetic utility of the described cycloaddition reactions was furtherd isplayed when 1 was reactedw ith an excess of 2butyne, which proceeded throught he reductivee limination of silole 11 (Scheme 3, Figure 4), similarly to ap reviously reported 6-membered platinasilacycle which eliminates silole when heated to 120 8C. [25] Nevertheless,s uch heterocycles are typically not obtained in the direct reactions of silylenes with acetylene derivatives, for which the [2+ +1] reaction products are more commonly encountered. [26,27] The molecular structure of 11 is in agreement with previously reported siloles, containing ap lanar SiC 4 ring and two short C=Cb onds (d(C16ÀC17) = 1.352 (2) ; d(C18À C19) = 1.347 (2) ). As with the formation of CÀHa ctivation products 9 and 9-Ph,h eterocycle 11 is likely formed througha [2+ +2+ +2] cycloaddition reaction of the SiÀNi bond in 1 with two molar equivalents of 2-butyne, proceeding via 1-metalla-4sila-cyclohexadiene derivative 10.I ndeed, it has previously been shown that platina-sila-cyclobutene speciesc an undergo such ar ing-expansion reactioni nt he presence of excess alkyne, albeit without silole elimination. [21c] AD FT investigation employing acetylene in place of 2-butyne suggests that this is the most energetically favored reaction coordinate, with the acetylene derivative of 10 (i.e. IM3",F igure S46 in the Supporting Information) lying 55.6 kcal mol À1 lower in energy than 1 and free acetylene. Reductive elimination and formation of a p-complex of the liberated silole derivative (IM4",F igure S46) is furtherfavored by 27.9 kcal mol À1 .
In summary,w eh ave investigated the [2+ +2] cycloaddition chemistry of the SiÀNi multiple bond in 1 towards unsaturated organic compounds, leading to ar ange of four-membered nickelasilacycles, including rare examples of metallasilacyclobutene species. Markedly,w eh ave found that the addition of ethylene is reversible, whereas the reaction with excess ethylene proceeds through a[ 2 + +2+ +2] cycloaddition reaction, leading finally to sp 2 -CH bond activation. Further, the liberation of fragments containing newly formed CÀCb onds has also been shownpossible, reminiscent of keyintermediary steps in established catalytic processes.