Facile Access to Dative, Single, and Double Silicon−Metal Bonds Through M−Cl Insertion Reactions of Base‐Stabilized SiII Cations

Abstract Silicon(II) cations can offer fascinating reactivity patterns due to their unique electronic structure: a lone pair of electrons, two empty p orbitals and a positive charge combined on a single silicon center. We now report the facile insertion of N‐heterocyclic carbene (NHC)‐stabilized silyliumylidene ions into M−Cl bonds (M=Ru, Rh), forming a series of novel chlorosilylene transition‐metal complexes. Theoretical investigations revealed a reaction mechanism in which the insertion into the M−Cl bond with concomitant 1,2‐migration of a silicon‐bound NHC to the transition metal takes place after formation of an initial silyliumylidene transition‐metal complex. The mechanism could be verified experimentally through characterization of the intermediate complexes. Furthermore, the obtained chlorosilylene complexes can be conveniently utilized as synthons to access Si−M and Si=M bonding motifs bonds through reductive dehalogenation.


Introduction
The presence of al one pair of electrons, two vacant orbitals and ap ositivec hargeo nt he silicon centerm akes silyliumylidene ions an incredibly versatile and promising class of lowvalent silicon compounds. [1] They offer al arge, yet untapped synthetic potential in organosilicon chemistry with the possibility to form up to three new bonds in as ingle reaction. [2] They are promising candidates for the activation of small molecules, transition metal free catalysis [3] and can act as synthons for novel (low-valent) silicon compounds. Further, with the presence of as tereochemically active lone pair,t hey can also function as ligands in transition metal complexes. So far,n oo necoordinate Si II cation has been isolated [4] and most reported examples are three-coordinate and utilize two Lewis bases for their stabilization (e.g. NHCs). [5] This brings the drawback of a generally reduced reactivity by blocking the empty p-orbitals.
Hence, both amount andd iversity of reported reactivities lag behind those of silylenes, where common reactivity patterns include insertion reactions into varioust ypes of (strong) bonds. As taggering number of examples for the insertion into EÀH ( E= H, N, O, S, C, B, …) and EÀHalogen bonds have been reported in recent years. [6] Similarly,t he coordination chemistry of silylenesw ith transition metals is ac ontinuously expanding researchf ield with variouscatalytic applications. [6d, 7] In contrast, even as the number of isolable base-stabilized silyliumylidenes continues to grow, [4a, 8] reportedr eactivities remain scarce. [5,9] Only few reactivity studies with small molecules [10] have been found and EÀHb ond activation reactions are limited to SÀH, OÀHa nd acidic CÀHb onds. [8g, 11] The chemistry of Si II cations as transition metal ligands has seen some progress in recent years. [12] Reported examples include complexes with coinage metals [12d] and group 6a nd 8 metal carbonyls, [12b, e] butn of urther reactivity of these complexesh as been reported to date. Importantly,t he synthesis of new types of complexes with silicon-basedl igands and substituentsi so fh igh interestf or the development of improved catalysts. [7a-c, 13] With their intriguing synthetic potential, silyliumylidenes are uniquely suited for the facile synthesis of various types of SiÀM( multiple) bonds (e.g. through salt metathesis or formation of coordination complexes followedb ya bstraction/ migrationo fstabilizing Lewis bases). This was elegantly demonstrated by Filippou et al. with the direct synthesis of am olybdenum silylidyne complex. [12c] For silylene complexes,avariety of follow-up chemistry is known. [6d, 7] For instance, multiple insertion reactions into metal-chloride bonds of ac oordinated transition metal fragment have been reported. For example, Jutzi and co-workers disclosed the insertion of Decamethylsilicocenei nto aH g ÀX bond, furnishing silyl-substituted Hg compounds (I, Scheme1A). [14] The group further reported analogous insertion reactions into NiÀCl and AuÀCl bonds [15] and related reactivities with PtÀCl bonds were also reported by Lappert et al. [16] Recently,K ato,B aceiredo, and co-workersr eported the insertion of ac hlorosilylene ligand into the RhÀCl bond of ac oordinated [RhCl(COD)] fragment (II), forming the corresponding RSiCl 2 ÀRh(COD) compound III. [17] For silyliumylidenei ons and their transition-metal complexes, no analogous reactivity has been observed so far.I nf act, insertion reactions into EÀHalogen bonds have not been reported at all.
Herein,w en ow report the first reactivity studies regarding insertionr eactions of aS i II cation into transition metal-chloride bonds. Reactions of NHC-stabilized silyliumylidene ions with dimeric, chloro-bridged transition-metal precursors lead to coordination of the Si II cation to the metal fragment, followed by insertiono ft he silyliumylidenel igand into the MÀCl (M = Ru, Rh) bond,f urnishingN HC-stabilized transitionm etal silylene complexes (Scheme 1B). The complexes have been fully characterized by multinuclear NMR spectroscopy and SC-XRD (single crystal X-ray diffraction) and the insertion mechanism has been investigated theoretically and verified experimentally. Furthermore, we present af acile access route to SiÀMa nd Si=Mb onds through stepwise reduction of the isolated complexes with KC 8 ,i nitially furnishing silyl-substituted complexes, followed by the formation of the corresponding Si=Ru double bond through additional reductive dehalogenation.I mportantly,while thesetypes of insertionreactions are generally accompanied by an increase of the silicon oxidation state from II to IV,n os uch change occurs for silyliumylidene ions (cf. Scheme 1).

Results and Discussion
Insertion of aS i II cation into aR u ÀCl Bond While exploring the coordination chemistry of NHC-stabilized Si II cations, we investigated the reaction of the Tipp-substituted silyliumylidene ion 1a [8g] with the transition metal precursor [RuCl 2 (p-cym)] 2 (Scheme 2, p-cym = 1-Me-4-iPr-benzene). Addi-tion of cold acetonitrile to am ixture of 1a and the precursor at À40 8Cl ed to an immediate color change of the solution to deep red. At about À20 to À15 8C, the color of the solution rapidly changed to orange. Even at À40 8C, ac olor change to orange can be observed within 2hours. The 29 Si NMR of the orange solution displays one resonance with an expected downfield shift at 17.6 ppm (from À69.5 ppm (1a) [8g] ), indicating the formation of as ingle coordination product. Interestingly,t he corresponding 1 HNMR (cf. Supporting Information,Figure S8) showed ah ighly asymmetrics peciesw ith four separate septets and eight doublets( corresponding to four chemically unique iso-propyl groups) and two distincts ignal sets for the NHCs. The 13 CNMR showed two resonances for the carbene carbon atoms at 169.9 and 154.8 ppm, indicating the possible migration of one NHC to the transition metal.
The complex rapidly decomposesa tr oomt emperature in solution to am ixture of products, making further investigation and functionalization difficult. Nevertheless, crystals suitable for SC-XRD analysisc ouldb eo btained by storing ac oncentrated solution of 2 in MeCN at À35 8C. Figure 1s hows the solidstate structure of 2,u nambiguously confirming the asymmetric natureo ft he complex and the shift of one NHCt ot he metal. The half-sandwich complex with ap iano-stool configuration  features an NHC-stabilized aryl-chlorosilylene ligand with atetrahedral coordination sphere aroundt he silicon center and a Si1ÀRu1b ond length (2.409 (1) )t ypical for SiÀRu bonds. [18] The SiÀC NHC (1.970(4) )a nd RuÀC NHC (2.077(4) )b ond lengths are in the typical range for SiÀC NHC and RuÀC NHC bonds.
It is worth notingt hat attempts to stabilize the complex by employing the significantly bulkier m-terphenyl (2,6-(2,4,6-Me 3 -C 6 H 2 ) 2 -C 6 H 3 )s ubstituent were unsuccessful and no reaction could be observed, presumably due to its large steric hindrance.S imilarly,w ee nvisioned the introductiono faCp* ligand (Cp* = 1,2,3,4,5-pentamethyl-cyclopentadienyl) on the metal. The p-cymene ligand is often aw eak spot in such complexes, as it can be relativelyeasily cleaved from the metal. Unfortunately,n or eactiono ft he relatedp recursor [RhCl 2 (Cp*)] 2 with 1a could be observed, most likely due to the increased steric demand of the Cp* substituent.

Formation of complex 2-mechanistic insights
To elucidate the mechanism of formation of chlorosilylene complex 2,w ep erformed DFT calculations at the B97-D/def2-SVP level of theory ( Figure 2). In af irst step, the coordination of as ilyliumylidene moiety to each transition metal center leads to the splitting of the dimer,f orming the silyliumylidene complex 2'.T his also indicates why no reactionc ould be observed at all for the significantly bulkier m-terphenyla nd Cp*: the initial coordination step is blockedd ue to their large steric hindrance, which completely stops any product formation. After the coordination,t he insertion reaction of the low-valent silicon into the RuÀCl bond occurs with concomitant 1,2-migration of one NHC moiety to the transition metal.W eh ave previously observed ar elated NHC migration reaction involving NHC-stabilized silyliumylidene ions with the formation of a [(IMe 4 ) 2 Au]Cl complex from as ilyliumylidene gold complex. [12d] This migration/insertion reaction is similartothe mentioned in-sertion reaction of ac hlorosilylenel igand into aR h ÀCl bond (II!III,S cheme 1). [17] However,akey distinction to the insertion reactions of silylenes is that in the case of the Si II cation, the formal oxidation state of the silicon center does not change:h ere, the insertion reactionl eads from [RÀSi II ] + to [RÀ Si II ÀCl],w hereas silylenes [R 2 Si II ]y ield silyl-substituted complexes[ R 2 ClSi IV ÀM] (cf. Scheme 1).
Based on the calculated reaction profile we presumed that the deep red specieso bserved at low temperatures during the synthesis should be the silyliumylidene complex 2'.I ndeed, low-temperature 29 Si NMR analysis (À30 8C) showedaweak resonancea tc onsiderably higherf ield (À21.1 ppm vs. + 17.6 ppm for 2)t hat immediately vanished upon warming and even disappeared at low temperatures within 2hours. This upfield shifted resonance is expected for aS i II center with two coordinated NHC moieties and is in line with our previously reported group 6s ilyliumylidenec omplexes (Cr: + 6.3 ppm;M o: À17.3 ppm;W :À30.5 ppm and the relatedi ron complex (+ 5.4 ppm)). [12e] To further reinforce the suggestion that 2' is in fact the intermediate observed at low temperatures,w ec alculated the 29 Si NMR shifts for 2 and 2':w ef ind that the calculated chemical shifts (19.8 ppm for 2 and À23.4 ppm for 2' (HCTH407/def2-SVP//B97-D/def2-SVP)) are in good agreement with the experimentally observedv alues.
Due to the relatively rapid insertion reaction occurring even at low temperatures, we were unable to structurally characterize 2'.H owever,b ased on these results we hypothesized, that the insertion/migration reactionf rom 2' to 2 occurs so rapidly to reduce the considerable steric congestion at the silicon center and that reducingt he size of the aryl substituent could enable us to isolate the intermediate silyliumylidenec omplex. Consequently,w eu tilized 1b [8k] and performed the same reaction (Scheme 3). Indeed, 29 Si NMR analysis of the resulting redorange solution showed ar esonance at À20.5 ppm, considerably upfield shifted compared to 2 (17.6 ppm) and very close to the À21.1 ppm for 2'.H owever, 3 decomposesi ncredibly quickly at room temperature (even faster than 2)a nd slowly at À35 8C, preventing furtherc haracterization and analysis( especially through SC-XRD). Hence, to stabilize the desired complex, we also attempted the reaction with [RhCl 2 (Cp*)] 2 ,w hich proceeds instantly even at À40 8C. 29 Si NMR analysiso ft he deep red solution showed ar esonance at À24.2 ppm (d, 1 J SiÀRh = 66.9 Hz), indicating the formation of the desired complex 4.W hile 4 is somewhat more stable in solution than 3,i t still decomposes rapidly (for detailsconcerning the decomposition, see Supporting Information). Still, we were ablet oo btain single crystals of 4 through quick diffusion of Et 2 Oi nto a MeCN solution at À35 8C. The solid-state structure (Figure 3) revealed as ilyliumylidene complex with ag eometry comparable to the chlorosilylene complex 3,e xcept that in 4 both NHCs are still located on the silicon centera nd both chlorides are still bound to the metal.T he compound features al ong Rh1ÀSi1b ond length of 2.426(2) with typical SiÀC NHC bond lengths (1.958(7) and 1.944 (7) ). The angle between the coordinated NHCs (93.9(3)8)i sc omparable to uncoordinated silyliumylidene ions (e.g. 1a:9 4.3(1)8). [12e] Attempts to convert 4 into the chlorosilylene complex analogous to 2 throughp rolonged stirring failed due to the low stability of 4 in solution. No conversion could be detected after 12 hours at À35 8Ca nd at highert emperatures only decomposition products were observed.

Reactivity of silyl-substituted silyliumylidene ions
Silyl groupsh ave proven to be excellent substituents for the stabilization of elusive main group species because of their tuneable steric demand as well as their strong s-electron-donating properties. [19] Consequently, we attempted the same conversionsw ith our recently reported silyl-substituted silyliumylidenes [8k] 5 in the hope of furnishing analogous silyliumylidene or chlorosilylenec omplexes with increased stabilityi n solution to allow further functionalization. Reaction of 5 with [RuCl 2 (p-cym)] 2 and [RhCp*Cl 2 ] 2 (Scheme 4) furnished the orange to red chlorosilylene complexes 6 and 7,r espectively. Only 5c did not react in ac lean fashion with [RhCp*Cl 2 ] 2 , giving am ixture of products containing the desired complex with less than 40 %( cf. Supporting Information Figure S56). Purificationa ttempts were not successful. This can presumably be attributed to the increased steric demando fthe bulkier NHCs together with the Cp* ligand,t hus favouring side reactions. 29 Si NMR analysis of 6-7 (see Ta ble 1) revealed resonances close to 2,c learly indicating the formation of the analogous chlorosilylenec omplexes.F urthermore, 1 Ha nd 13 CNMR spectra show formation of asymmetrics pecies with clear signal sets forN HCs bound to both silicon and the metal. Generally, reactions with the rhodium precursor give higher yields than the analogous ruthenium reactions due to higher stabilityo f the Rh complexes in solution. While complexes 6 still slowly  decompose at room temperature in solution( 6a being the most stable of all ruthenium complexes with full decomposition after roughly 12 hours), 7a and 7b are stable for at least two weeks.
To furthere lucidate and strengthen our proposed reaction mechanism, we used the tBu 3 Si-substituted silyliumylidene triflate 5a-OTf (insteado fc hloride) and carriedo ut the same reaction:t he corresponding complex 7a-OTf could be obtained (cf. SupportingI nforamtion, Figures S46-S48), excluding any relevant involvement of the anion in the reactionm echanism. This reactivity also further underscores the hypothesis that the Si II cation indeed inserts into the MÀCl bond.

Access to SiÀMs ingle and Si=Mdouble bonds
As complexes 2, 6 and 7 exhibit ah alide counterion and one halide bound to the silicon andt ransition metal each, we thought them to be ideal precursors for the synthesis of SiÀRu and SiÀRh multiple bonds through reductived ehalogenation. We utilized the tBu 3 Si-substitutedc omplexes 6a and 7a for furtheri nvestigations due to their significantly increased stability in solution. After treatment of 6a and 7a with one equivalent of potassium graphite (Scheme 5), we werea ble to isolate the unexpected paramagnetic silyl-substituted complexes 8 (brightg reen) and 9 (grey-black) in moderate and good yield, respectively.E PR analysis of 8 and 9 revealed only as ingle band in both cases (cf. Supporting Information,Figures S64 and S67). No hyperfine coupling to a-o rb-silicon could be observed. The g-values (8: g = 2.1062, 9: g = 2.1003) are in line with other paramagnetic ruthenium and rhodium complexes. [22] We successfully confirmedt he compositiono f8 and 9 throughS C-XRD analysis ( Figure 4, left and center). Formation of these complexes most likely takes place through1,2-migration of the metal-boundc hloride to silicon under dissociation of the silicon-boundN HC. As expected, the chloride counterion was the first halide to be removed through reductive dehalogenation.
We further attempted the reaction of 6a with two equivalents of KC 8 in the hopes of furnishing aS i =Ru bond. Indeed, two-electron reduction of 6a or additional reduction of 8 with 1K C 8 yielded the ruthenium silylene complex 10 (Scheme 5, 63 %f rom 8,4 1% from 6a). During the reaction, an intense color change from bright green (8)t od eepr ed (10)c an be easily observed. We also attempted the reduction of complex 7a (or 9;c olor change from black to purple) to as imilar Si=Rh species. While 29 Si NMR and mass spectrometry analysis(for details, see Supporting Information) suggest that formation of the analogouscomplex takes place (albeit in as ignificantly less clean fashion), we have been unable to obtain satisfactory analytical data so far.
With the additional reductive step, 10 is no longerp aramagnetic. The 29 Si NMR exhibits as ignificantly downfield shifted resonance at 240.6 ppm, which falls in the expected range of Si=Mb onds with at hree coordinate Si center [23] and indicates the multiple-bond character of the Si=Ru bond. The observed resonance is even more downfield shifted than the previously reporteds tructurally relateda ryl-chlorosilylene complexes Cp*(R 3 P)(H)Ru=SiCl(aryl)( aryl = Tipp (221.7 ppm), m-terphenyl (205.0 ppm)). [23a] No signal splitting analogous to complexes 6 and 7 could be observed in the 1 H/ 13 CNMR spectra.T he carbene carbon atom of the metal-bound NHC also exhibits as ignificantly more downfield shifted resonance at 188.5 ppm compared to the 172.1 ppm observed for 6a.

Computational studies
To better understand the bonding situation and the electronic structure of the isolated complexes, we also carriedo ut DFT calculations (for details, see Supporting Information). The calculated metric parameters ( Table 2, SupportingI nformation Ta ble S10) show good agreement with the experimentally observedv alues, indicating the validity of the computational method.A nalysis of the Natural Bond Orbitals (NBO, Supporting Information Table S3-S9) revealed that the SiÀMb ond polarity can change in different complexes:f or example, the SiÀ Ru bond is polarized towards the metal centeri nc omplexes 2, 8,a nd 10.I nc ontrast, the bond is polarized towards the Si atom in 6a.C ombinedw ith the very long Si-Ru bond distance in 6a (2.499 (1) ), we conclude that the SiÀRu bond in 6a is more dative in nature while it exhibitsa ni ncreased covalent character in the other complexes. Natural Population Analysis (NPA, Ta ble 2a nd Supporting Information Table S10) shows that for the aryl-substituted complexes 2 and 4 the central Si atom bears am ore positive charget han in the silyl-substituted complexes 6a, 7a and 8-10.T his can presumably be attributed to the stronger s-donating properties of the silyl moieties compared to aryl groups. In general,t he rutheniumc enter in complexes 2, 6a and 8 exhibits am ore negative charge than  the Rh atom in 4, 7a and 9.T he Ru centeri nc omplex 10 exhibits the highest negative charge (À0.73) out of all complexes. This increased negative charge is most likely the consequenceo ft he double bond character of the Si=Ru bond in 10 suggested by the NBOs (cf. Supporting Information, Ta ble S9). The Wiberg Bond Index (WBI) andM ayerB ond Order (MBO) also support the double bond character, as both WBI and MBO for complex 10 are significantly higher than in the other complexes. These results agree well with the experimentally determined SiÀMb ond lengths. The calculated frontier orbitals also confirm the validity of the Si=Ru bond in 10 ( Figure 5), where the HOMO (highest occupied molecular orbital) and LUMO (lowest unoccupied molecular orbital) correspond to the bonding and anti-bondingo rbital of the SiÀRu p-bond. Additionally, we were unablet of ind similar orbitals for the other investigated complexes (cf. Supporting Information, Figure S88-S93), in which the HOMO and LUMO are associated with the metal d orbitals andt he p-system of the Cp* or p-cymene ligands.

Conclusions
In summary,w eh ave used NHC-stabilized Si II cationsa saconvenient entry point for the isolation of Si!M, SiÀMa nd Si=M moieties via the insertion of silyliumylidenesi nto MÀCl (M = Ru, Rh) bonds with simultaneous silicon-to-metal NHC-migration, followed by reductive dehalogenation. This work significantly expands the still-young field of silyliumylidenet ransition metal coordinationc hemistry and showcases the ease with whichr elatively bulky aryl-and silyl-substituted silyliumylidenes inserti nto MÀCl bonds, forming chlorosilylene transition metal complexes.T his is an important distinction to previously reported MÀCl insertion reactions of low-valent silicon compounds, where the insertion leads to Si IV compounds.T he mechanism of formation was investigated theoretically and predicted to include an initially formed silyliumylidene transition metal complex followed by insertion of the Si II cation into the MÀCl bond with concomitant1 ,2-migration of as iliconbound NHC moiety to the metal. This could be verified experimentally through NMRa nd XRD characterization of the silyliumylidene complexes.
The presence of multiple halideso nt he isolated chlorosilylene complexes gives as imple accessr oute to SiÀMs ingle and Si=Md oubleb onds through successive reductived echlorination. The possible utilization of this synthetic approacht o accessv arioust ransitionm etal silylidene and silylidyne complexesi sc urrently under investigation in our laboratory.