Mechanistic Divergence in the Hydrogenative Synthesis of Furans and Butenolides: Ruthenium Carbenes Formed by gem‐Hydrogenation or through Carbophilic Activation of Alkynes

Abstract Enynes with a tethered carbonyl substituent are converted into substituted furan derivatives upon hydrogenation using [Cp*RuCl]4 as the catalyst. Paradoxically, this transformation can occur along two distinct pathways, each of which proceeds via discrete pianostool ruthenium carbenes. In the first case, hydrogenation and carbene formation are synchronized (“gem‐hydrogenation”), whereas the second pathway comprises carbene formation by carbophilic activation of the triple bond, followed by hydrogenative catalyst recycling. Representative carbene intermediates of either route were characterized by X‐ray crystallography; the structural data prove that the attack of the carbonyl group on the electrophilic carbene center follows a Bürgi–Dunitz trajectory.

Abstract: Enynes with at ethered carbonyl substituent are converted into substituted furan derivatives upon hydrogenation using [Cp*RuCl] 4 as the catalyst. Paradoxically,t his transformation can occur along two distinct pathways,each of which proceeds via discrete pianostool ruthenium carbenes.In the first case,h ydrogenation and carbene formation are synchronized ("gem-hydrogenation"), whereas the second pathway comprises carbene formation by carbophilic activation of the triple bond, followed by hydrogenative catalyst recycling.Representative carbene intermediates of either route were characterized by X-ray crystallography;t he structural data prove that the attacko ft he carbonyl group on the electrophilic carbene center follows aBürgi-Dunitz trajectory.
After ac entury of intense research by the scientific community on catalytic hydrogenation, our group has recently been able to identify an entirely new reactivity mode.S pecifically,i tw as shown that alkynes can undergo gem-hydrogenation, ar eaction in which both H-atoms of H 2 are delivered to one and the same acetylenic C-atom while the adjacent position is concomitantly transformed into adiscrete metal carbene. [1,2] At the current stage of development, [Cp*RuCl] 4 is the precatalyst of choice;m oreover,h eteroatom substituents in vicinity of the triple bond are often necessary to render the reaction efficient. Detailed spectroscopic and computational data indicate that the resulting pianostool ruthenium complexes basically exhibit aF ischercarbene character with ac ertain overtone reminiscent of Grubbs-type catalysts. [3] This view is corroborated by the fact that they participate in either intramolecular cyclopropanation or metathesis reactions,d epending on the chosen substrate. [4] Fort heir largely electrophilic nature,s uch complexes should be able to participate in various other catalytic transformations too.I ft his is the case, gem-hydrogenation might eventually evolve into an attractive alternative to diazoalkane decomposition, which is arguably the most common gateway to highly reactive late-transition metal carbenes. [5,6] Thef oray along these lines outlined below was inspired by ar ecent publication describing an innovative entry into highly substituted furan derivatives (Scheme 1). [7] Specifically,d iazo compounds A were shown to react with catalytic [CpRu(MeCN) 3 ]PF 6 to generate transient cationic ruthenium carbenes B,w hich get trapped by the tethered ester group to give the corresponding heterocycle C.W e reasoned that this type of transformation might be emulated by gem-hydrogenation of enyne D,i nw hich the propargylic -OR substituent directs carbene formation to the distal acetylenic site. [1,2] Akin to B,t he resulting intermediate E might furnish furan F,even though E is aneutral rather than cationic entity.
In line with our expectations,p roduct 2 was formed in almost quantitative yield (! 95 %) upon stirring of asolution of 1 and [Cp*RuCl] 4 (2 mol %) in CH 2 Cl 2 under an atmosphere of H 2 (1 bar) for 3h at ambient temperature (Scheme 2). Treatment of the crude material with silica released the corresponding butenolide 3. [8] Alternatively, 2 can be alkylated with allyl iodide to give product 4. [9] In other cases such as 5-7,t he furan itself was sufficiently stable for isolation. In line with our previous investigations, [2][3][4] different propargylic substituents (for example,-OR, -OSiR 3 , -OMOM) were found to instigate gem-hydrogenation. Gratifyingly though, even substrates devoid of such directing groups led to the formation of butenolides 10 and 11; [10] in these cases,t he ester itself might serve as ad irecting group, fostering regioselective carbene formation by gem-hydrogenation. [2a] This effect is not always sufficient, however,a s illustrated by the formation of alkene 12 through transhydrogenation of the corresponding alkyne substrate. [11] This outcome is not overly surprising since trans-hydrogenation has previously been shown to be afacile process downstream of an initial gem-hydrogenation event (even though 12 could very well originate from ac oncerted pathway). [1,2,12] Although these results are fully consistent with the formation of pianostool ruthenium carbenes by gem-hydrogenation in the first place,w es ought to confirm this mechanistic interpretation. To this end, we resorted to substrate 13,w hich is expected to undergo regular gemhydrogenation but should be resilient to cyclization because heterocycle formation comes at the prize of dearomatization of the phenyl ring (Scheme 3). Indeed, the reaction stopped at the stage of the pianostool ruthenium carbene complex 14. Theu se of para-hydrogen as the reagent leads to am assive amplification of the 1 HNMR signal of the methylene group flanking the carbene center as ar esult of the PHIP effect (PHIP = p-hydrogen induced polarization). [13,14] This spectroscopic signature provides unambiguous proof that these two H-atoms originate from the same H 2 molecule by pairwise delivery,acharacteristic trait of gem-hydrogenation. [2] Thestructure of this sensitive complex in the solid state is highly informative (Figure 1). It suggests that the directing effect of the methyl ether substituent emanates from its interaction with the Lewis-acidic Ru center at ad istance of 2.18 ;the actual C1ÀRu1 carbene bond length (1.883 (2) ) is in the expected range. [2] Thea rguably most remarkable structural feature,however, is the orientation of the carbonyl oxygen atom O1 relative to the carbene center C1:the O1 À C1 distance (2.983(4) )i sw ell below the sum of the van-der-Waals radii of these atoms (ca. 3.22 )a nd the Ru1ÀC1ÀO1 angle of 116.08 8 shows that the ester carbonyl approaches the trigonal carbene center along aB ürgi-Dunitz trajectory. [15,16] In view of this ideal geometric predisposition for an outersphere attack, it is reasonable to believe that ring closure stalls because of the unfavorable thermodynamic rendering of this model compound, since loss of aromaticity of the benzene ring would not be compensated by the enthalpic gain of an emerging isobenzofuran. TheB ürgi-Dunitz angle is well recognized as af undamental principle of (dynamic) stereochemistry originating from the particular shape and occupancy of the relevant frontier orbitals. [17] Yet, complex 14 seems to be the first case of an electrophilic transition metal carbene and its nucleophilic reaction partner,f or which this prominent effect has been explicitly recognized as astructuredetermining element manifest in the crystallographic data. [18] Next, we attempted to extend the novel hydrogenative furan formation to substrates carrying nucleophiles other than an ester group.The readily available diketone derivative 15 gave the expected product 16 but required ar eaction temperature of 70 8 8C (Scheme 4). Surprisingly,small amounts of ketone 18 were also isolated in one run from the crude mixture,w hich seems to indicate incidental oxidation of an intermediate of type G featuring the carbene site proximal to the directing -OMe substituent. Such ar egiochemical course violates the mechanistic rationale outlined above for the ester series and is inconsistent with all other available information on alkyne gem-hydrogenation, [1][2][3][4] not least with the X-ray structure shown in Figure 1. Formation of G might be   explained, however,b ya ssuming that the p-acidic metal fragment activates the substrate to the extent that attack of the ketone onto the triple bond is faster than binding of H 2 ,as necessary for gem-hydrogenation to occur. This type of mechanism has ample precedent in the literature for many different carbophilic catalysts, [19] although rigorous proof for the intervention of discrete carbenes is exceedingly rare. [20] To test this hypothesis,e nyne 15 was reacted with stoichiometric amounts of [Cp*RuCl] 4 in the presence as well as in the absence of H 2 :e ither set-up afforded the very same pianostool ruthenium carbene 20 in less than 30 min reaction time.The structure of this remarkable complex in the solid state shows that the ruthenium atom no longer interacts with the adjacent -OMe group ( Figure 2);r ather,i ti st ightly ligated to the electron rich "enol ether" site of the newly formed furan ring as manifested in the observed distances as well as in an elongated C2ÀC3 bond. Thedistinctive up-field shift of the NMR signals of C2 (d C = 132.3 ppm), C3 (d C = 89.7 ppm), and the carbene center C1 (d C = 266.7 ppm) suggest that this bonding situation persists in CD 2 Cl 2 solution. [21] Despite the stabilizing interaction, 20 is ac ompetent intermediate on the way to product and by-product alike.I t reacts with H 2 at 70 8 8Ctogive furan 16, [22] whereas stirring of asolution in air afforded the furyl ketone 18 (Scheme 4).
Since p-bond activation by the carbophilic catalyst does not require assistance by an eighboring directing group, [23] substrates of type 21 (R = alkyl, aryl) devoid of donor substituents at the propargylic position should react analogously (Scheme 5). As the formation of furans 23 a,b and 24 shows,t his is indeed the case as long as the substituent R shields the transient carbene center in 22.Ifnot, dimerization with formation of at etrasubstituted alkene 26 becomes competitive.T hus,s ubstrate 21 c (R = Ph) gave am ixture of the monomeric furan 23 c and olefin 26c (R = Ph, E/Z ca. 1:1), [24] whereas 21 d (R = n-Bu) afforded only traces of the monomeric furan derivative but furnished the product 26d (R = n-Bu) as the major product in the form of as ingle isomer.The E-configuration of the central double bond could only be assigned by crystallographic means. [21,25] It is assumed that the triple bond of asecond substrate inserts into the [Ru= C] bond of 22 in an enyne metathesis fashion to furnish av inylcarbene 25, [26] which is then interrcepted by the adjacent ketone to close the second furan ring of the resulting dimer 26.
Insufficient steric shielding of the carbene center can open yet another sidetrack. TheN MR data of carbene 22a (R = tBu, d C = 285.3 ppm) correspond to those of 20 and its structure must hence be similar, [22] but the ruthenium complex

Angewandte Chemie
Communications derived from enyne 21c (R = Ph) and [Cp*RuCl] 4 is strikingly different. Whereas 20 and 22a (R = tBu) are both rather sensitive burgundy-red compounds,avery robust green solid material is formed that does not react with H 2 even at 70 8 8C, is stable towards air and moisture for extended periods of time, and does not catalyze furan formation either;i ti sh ence an off-cycle product. Thes tructure of this complex was determined by X-ray diffraction. As shown in Figure 3, 27 incorporates ab ridging rather than terminal carbene moiety;t he dimer persists in solution as indicated by the markedly upfield-shifted carbene resonance (d C = 189.9 ppm). [27,28] Thed istances between C1 and the two Ru atoms are uneven, as are the distances between the bridging chloride Cl2 and the metal atoms.Ru2 benefits from the same stabilizing interaction with the proximal enol site of the furan ring discussed above for 20,whereas Ru1 carries an additional terminal chloride ligand.
Thes tatic picture of the structure in the solid state, however, provides an incomplete description of this complex. Rather,t he 13 CNMR signals of the two Cp* rings show massive line broadening at ambient temperature;areversible dyotropic process 27/27' ' is the likely cause for the equilibration of the two different metal subunits, [29] adynamic behavior that can be frozen out by lowering the temperature to À50 8 8C. At the same time,t he rotation of the phenyl substituent comes to ah alt. Inspection of the space-filling model of 27 shows that this phenyl ring is "sandwiched" between the lateral Cp* rings.A ctually,i ti sp lausible that the complex draws some of its remarkable stability from this peripheral interaction, even though this aspect needs further scrutiny.In any case,o nly slim substituents will be able to intercalate analogously.O nly two related complexes are known in the literature:indeed, they carry H/alkenyl and Me/Me substituents at ab ridging ruthenium carbene site between the Cp* ligands. [4,30] Sterically demanding branched alkyl residues in lieu of the phenyl group are unlikely to fit into the groove formed by the Cp* rings and will hence prevent bridging carbene formation from occurring;t he structure of the complex 20 discussed above bears witness of this notion.
Thes ubtlety of the transformations described above is deemed remarkable:hydrogenation of enynes carrying atethered carbonyl group with the help of [Cp*RuCl] 4 as the catalyst invariably affords highly substituted furan products (Scheme 6). Depending on the nucleophilicity of the carbonyl substituent, however, the reaction takes place along two distinctly different pathways,e ach of which involves pianostool ruthenium carbenes as the key intermediates.T hese reactive species evolve,however, on the opposite ends of the central alkyne subunit of the substrate.Inthe first case,the pcomplex initially formed binds H 2 ,w hich in turn triggers the unorthodox gem-hydrogenation of the triple bond with concomitant formation of ar uthenium carbene; [1][2][3][4] hydrogenation and carbene formation are hence synchronized. Attack of the tethered ester carbonyl onto the electrophilic carbene center delivers the heterocyclic product and, at the same time,regenerates the catalyst.
Alternatively,d irect attack of ak etone onto the ligated alkyne outperforms gem-hydrogenation. [31] Thee nsuing fiveexo-dig cyclization places the emerging ruthenium carbene away from the incoming nucleophile.I nt his p-acid catalysis scenario,c arbene formation precedes the actual hydrogena- tion reaction, which is necessary to regenerate the catalyst and ensure turn-over.I nsertions of either free carbenes or metal carbene complexes into H 2 are known; [32,33] to the best of our knowledge,h owever, the furan synthesis described herein is the first example in which carbene hydrogenation is essential for the release of the desired product and catalyst recovery alike;ittherefore keeps the actual catalytic carbene formation up and running. [34] From the conceptual viewpoint, the chemistry described herein outlines new strategic roles for catalytic hydrogenation chemistry.O ngoing studies in our laboratory try to leverage some of the opportunities that this unconventional reactivity paradigm may provide.