Research Programs – Organometallic Chemistry

2.4 Advances in Metal-Carbene Chemistry Department of Organometallic Chemistry by Alois Fürstner

ABSTRACT: The major lines of research in this Department continue to be: (i) alkyne metathesis, (ii) iron catalyzed C-C-bond formation, (iii) π-acid catalysis using platinum, gold and rhodium complexes, and (iv) unorthodox catalytic addition reactions. All areas are prospering, including the application of the in-house methodology to target-oriented synthesis; yet, it was the field of ruthenium-catalyzed addition chem- istry which led to the most perplexing and (hopefully) significant results. For the unexpected intervention of discrete metal carbenes as reac- tive intermediates, the major findings in this area are discussed together with our recent contributions to the related field of rhodium carbene chemistry.

Ruthenium. cis-Delivery of H2 to a π-system of an unsaturated cal outcome is astounding, if one considers that conventional trans- substrate is the canonical course of metal catalyzed hydrogenation hydroboration is the textbook example for a cis-addition process via reactions. This stereochemical paradigm remained basically unchal- a four-membered transition state under frontier-orbital control. All lenged since the pioneering work of Sabatier until our group re- newly discovered trans-hydrometalation reactions break this fun- ported the semi-reduction of internal alkynes with the aid of damental stereochemical rule; importantly, they are robust, distin- [Cp*Ru]-based catalysts. The reaction clearly violates this funda- guished by excellent functional group compatibility, and have mental rule and affords E-alkenes by direct trans-hydrogenation already stood the test of natural product synthesis in a number of (Scheme 1). Answering the question as to how this unorthodox demanding cases (see below).1 transformation might proceed and whether it is a singularity or the Scheme 2. Mechanism of trans- and gem-Hydrogenation manifestation of a more general reactivity mode became a top E-alkene Cp* 1 H priority in our laboratory. Ru Cll H + R R Scheme 1. Prototypical trans-Hydrogenation Reaction alkyne H2 A O O Cp* O Cp* [Cp*RuCl(cod)] cat., H2, CH2Cl2 O Cl Ru O O O O H concerted Cl Ru R H E = H B E:Z 98:2 H H R trans-hydrogenation R H R H2 A combined experimental and theoretical approach provided stepwise trans-hydrogenation - compelling evidence that trans-hydrogenation can actually be gem hydrogenation Cp* Cp* reached by two distinctly different yet interconnected pathways H 2 Ru Ru σ overreduction Cl H Cll (Scheme 2). A -complex of type A is initially formed, which R D R evolves via the rate-determining H-delivery to the activated triple isomerization R R C H H H H bond into a metallacyclopropene B. It is at this stage that the reac- H tion pathway bifurcates: thus, B can transform in a concerted pro- 2 cess into the desired E-alkene E by passing through a low-lying genuine carbene reactivity stereo-determining transition state; this product-forming step is Only in the case of the trans-hydrogenation can the metallacy- strongly exergonic and therefore almost certainly irreversi- clopropene B evolve by a second pathway, in which both H-atoms ble. The computed barrier for the formation of the Z-alkene of H2 are transferred to one and the same C-atom of the substrate is notably higher, which explains the experimentally observed (Scheme 2). This geminal delivery, which entails formation of a excellent E/Z ratios. discrete metal carbene C, is without precedent in the literature. Reagents other than H2 able to form σ-complexes similar to A Computational studies at the DFT and the CCSD(T)) level sug- should undergo analogous trans-addition to alkynes. In fact, we gest that the trans- and the gem-pathway have similar barriers, but were able to accomplish closely related trans-hydroboration, trans- polar substituents in vicinity to the reacting triple bond foster hydrosilylation (previously described by the Trost laboratory), carbene formation and allow regioselectivity to be imposed on this trans-hydrogermylation and trans-hydrostannation reactions, remarkable transformation (Figure 1). Moreover, it has been un- which are equally paradigm-changing processes.1 The stereochemi- ambiguously shown by spectroscopic means (PHIP NMR) that the

Research Programs – Organometallic Chemistry resulting carbenes are kinetically competent intermediates rather Several attempts at harnessing genuine carbene reactivity by al- than thermodynamic sinks off the catalytic cycle; they evolve via kyne gem-hydrogenation have already been successful.2 Most associative H2-dependent processes into the desired E-alkene (and notably, it is possible to intercept the carbene primarily formed possible by-products). The spectroscopic evidence is in excellent with tethered olefins (Scheme 3); this allows either cyclopropenes accord with the computational results.2 or cyclic olefins to be formed; it is the substitution pattern of the substrate that determines the course of the reaction.4 In any case, the new “hydrogenative cyclopropanation” stands in striking con- HO OMe 1 trast to the hydrogenolytic cleavage of cyclopropanes commonly used in organic synthesis. Likewise, the “hydrogenative metathesis” [Cp*Ru(coδ)Cl] is an entirely new manifold to be distinguished from classical enyne p-H CD Cl 2, 2 2 metathesis, because it delivers cyclic olefins rather than 1,3-dienes as the product. The available data allow a fairly detailed mechanis- Me Cl Ru O tic picture to be drawn, especially with regard to the fate of the H 2 O secondary carbene formed during the actual metathetic C−C bond H H cleavage.4 δ C = 340.9 ppm

2 = trans-Hydrometalation. As mentioned above, the concerted JH,H -16.0 Hz pathway of trans-hydrogenation (A → B → E) finds correspond- ence in ruthenium catalyzed trans-additions of pinacolborane 1 Figure 1. Formation of a Pianostool Ru-Carbene by gem- (pinBH) or R3EH (E = Si, Ge, Sn) to internal alkynes. During the Hydrogenation; Structure of the Complex in the Solid State report period, these reactions were subject to extensive scrutiny. Scheme 3. Hydrogenative Cyclopropanation and Hydrogena- They are distinguished by excellent stereo- as well as regioselectivi- 5 tive Metathesis ty, especially when working with propargylic substrates: spectro- scopic, crystallographic and computational evidence suggests that a H2 (1 bar) H H nascent hydrogen bond between the protic substituent and the [Cp*RuCl]4 cat. polarized [Ru−Cl] unit of the catalyst locks the substrate in place; MeO [Ru] MeO 93% MeO at the same time, the –Cl ligand steers the incoming reagent via a

hypervalent interaction with the R3E− group (Scheme 4). These hydrogenative cyclopropanation synergistic effects impose directionality on the ligand sphere of the

H2 (1 bar) H H loaded catalyst F, which ultimately translates into excellent levels of cat. [Cp*RuCl]4 selectivity.5 MeO [Ru] MeO 93% MeO Scheme 4. Directed trans-Hydrometalation and Novel Down- hydrogenative metathesis stream Chemistry

H H OAc H OH

R1 R2 K R1 R2 L From a conceptual viewpoint, the formation of discrete metal O H carbenes by hydrogenation of an alkyne is arguably of the highest H ER3 significance. In a formal sense, the triple bond behaves as a 1,2- R1 R1 H OH H OH dicarbene synthon: one “carbene” intercepts the [Cp*Ru] frag- R3E-H Ru R1 R2 1 2 − R R ment, whereas the vicinal “carbene” site inserts into the H H bond Cl ER3 Me 2 (Figure 1). To the best of our knowledge, the geminal hydrogena- R OH O H R2 F G H tion of a stable carbogenic compound is a new reactivity mode.1,2 H OH This counterintuitive entry into ruthenium carbenes raises new H OH proximal / trans delivery 1 2 R R = R1 R2 questions and opens exciting chemical opportunities. Even though E Si, Ge, Sn COOMe F we were able to isolate and even crystallize numerous examples, it is J I not intuitive whether such pianostool ruthenium complexes exhibit Fischer-carbene or Schrock-alkylidene character. For two repre- sentative complexes, however, has it been possible to record the solid-state 13C NMR spectra and to analyze the chemical shift tensors of the carbene signals.3 Details apart, this advanced spectro- scopic technique drew the portrait of a family of metal carbenes that amalgamates purely electrophilic behavior with characteristics more befitting metathesis-active Grubbs catalysts. Moreover, we were able to show that less electron-rich CpR ligands facilitate carbene formation by gem-hydrogenation.3

Research Programs – Organometallic Chemistry

OH HO H O Me was the formation of the macrocyclic precursor 4 by alkyne me- O OH HO O OH tathesis (this laboratory) followed by hydroxy-directed trans- OH H hydrostannation (this laboratory) and subsequent methoxycar- O O brefeldin A OH ACIE 2015, 54, 3978 O bonylation of the resulting stannane 5 (this laboratory). Treatment B aspicillin dihydrocineromycin 2017, 49, 202 ACIE 2015, 54, 6339 of compound 6 thus formed with Cs2CO3 in MeOH triggered a O Synthesis O cascade comprised of (i) deacetylation, (ii) stereoselective O O C19H39 O HN transannular Michael addition with formation of the challenging OH OH O O HO O C H HO 8 17 medium-sized ring, (iii) β-elimination with concomitant cleavage O O OH OH F typhonoside of the carbonate, (iv) attack of the released alkoxide onto the prox- OH Chem. Eur. J. 2018 24 9667 disciformycin B HO , , OH O imal ester with formation of a butenolide ring 8, and (iv) front-side O OH HO HO Chem. Eur. J. 2018, 24, 109 MeO attack of MeOH from the medium onto this Michael acceptor, OH O OH O which is rendered particularly reactive by the bridgehead alkene OH F H O HO moiety. Final ether cleavage then furnished the target compound in O O O 6 sinulariadiolide excellent overall yield.

Cl JACS 2019, 141, 805 O acid A OH Rhodium. Controlled decomposition of diazo compounds with paecilonic MeO ACIE 2017, 56, 6161 Cl O O (chiral) dirhodium tetracarboxylate complexes has gained tremen- F a O OH 14-fluoroprostaglandin 2 H Chem. Eur. J. 2018, 24, 9667 N dous importance in (asymmetric) catalysis. The transient rhodium N H OH carbenes, however, had basically defied direct experimental obser- O NMe O O O O N vation until our group was able in 2016 to present crystal structures Ph O O of several prototypical members of this elusive series (cf. last pro- nannocystin Ax OMe rhizoxin D gress report). NMR data showed that the structures in solution are JOC 2018, 83, 6977 OMe ACIE 2019, 58, 248 very similar to those in the solid state. The crystal structure of the donor-acceptor carbene 11 derived The resulting products G provide ample opportunity for down- from [Rh2((S)-PTTL)4] (12), a chiral catalyst with an excellent stream functionalization. In addition to the known repertoire of 7 track record, is deemed particularly relevant (Figure 2). Although organoboron, -silicon and –tin chemistry, we developed new meth- crystallographic and spectroscopic data can only draw the portrait ods that allow such substrates to be converted into the polyketide of the ground state, it is likely that the forces which impose the motifs H and I by C-methylation or methoxycarbonylation, respec- peculiar and − at first sight − unexpected “all-up” conformation tively. Likewise, stereodefined fluoroalkenes J as well as acyloins K onto the chiral ligand sphere are also operative in the selectivity- came into reach (Scheme 4). Each of these transformations has determining transition state. already served as a key step in natural product synthesis.

carbene moiety removed for clarity Scheme 5. Total Synthesis of Sinulariadiolide O O OAc O O OAc MeO 1. cat. O [Mo] MeO [Cp*RuCl] O HO OH OTMP 2. Zn, HOAc Bu3SnH 76% O HO 66% 3 4 N H

O OAc O O 1. Pd(OAc)2 cat., MeO O O OAc TFA cat. O CO, BQ, O CH2Cl2 MeO O O O MeOH, 57% Rh OMe OH Rh

2. COCl2, pyridine, 97% O COOMe O 5 6 HO SnBu3 O 11 MeO C4 MeO O MeO O H O Cs2CO3 O O O O O O Figure 2. X-ray structure of a chiral dirhodium carbene CH2Cl2 / aq. MeOH Cs O H MeO O O 7 O 8 MeOH O Under this premise, it was possible to rationalize the stereochem- MeO O OH ical course of a cyclopropanation reaction effected by this famous

9 R = Me 7 O O BBr catalyst. Our model predicted the correct isomer; therefore, it also 10 R = H 3 RO seemed legitimate to analyze why the level of induction is only H O Sinulariadiolide modest (78% ee). An essentially C4 symmetric “all up” array entails O two notably different rhodium faces (Scheme 6): although the The arguably most involved case concerns the nor-cembranoid crystallographic data show that the “achiral” pore between the tert- sinulariadiolide (10) (Scheme 5).6 Specifically, a transannular butyl groups is narrower than the aperture of the “chiral” calyx, it approach was conceived, which allowed the tricyclic scaffold com- remains large enough that CH2Cl2 can enter and coordinate to Rh2. prising a central nine-membered ring to be forged. Key to success If a diazo derivative reaches Rh2 and gets decomposed to the corre-

Research Programs – Organometallic Chemistry sponding metal carbene, a racemic background reaction will ensue this ion, which provides more space. The long Bi−O bonds render and the ee of the resulting product necessarily drops.7 the chiral pocket about the rhodium center in 13 more confined than that of its homobimetallic analogue 12. Scheme 6. Concept for an Improved Catalyst Design

O R R N R H unselective! O O O O O O selective O Rh Rh Rh O O channel Rh steric blocking O necessary R

R = tBu [Rh2(S-PTTL)4] (12) R

O R tBu R Figure 3. X-ray Structure of [BiRh((S)-PTTL)4]⋅EtOAc H O N O O Scheme 7. Superior Performance of the Heterobimetallic O open but O O Bi Rh more confined O unreactive Precatalyst 13 O chiral site ? Bi O O N Rh 2 OMe R MeOOC S-PTTL) R O 13: 84% 95%) [BiRh( 4] (13) MeO 14 (ee 12: 92% 79%) Ph (ee 12 or 13 cat., CH2Cl2 MeO 15 Replacement of the tert-butyl groups by even bulkier substitu- ents should improve the outcome; earlier empirical catalyst optimi- zation has shown that this is indeed the case. We pursued a less Importantly, the new precatalyst 13 leads to cyclopropanes with conventional approach to catalyst optimization, which builds upon notably higher optical purity than those obtained with the tradi- recent insights into structure and bonding in heterobimetallic tional dirhodium congener 12 (Scheme 7).8 This lead finding is the 8 paddlewheel carbene complexes. In a collaboration with the Neese basis for ongoing work in our laboratory which intends to explore II group, we had shown that formal replacement of one Rh center of the new design and its synthetic implications in more detail. the bimetallic core for BiII enhances the electrophilic character of the resulting carbenes to a significant extent, although the rate of metal-carbene formation is manifestly slower.9 More important in REFERENCES the present context is the fact that the Bi site lacks any notable (1) Fürstner, A. J. Am. Chem.Soc. 2019, 141, 11. Lewis acidity and proved incapable of decomposing ethyl diazoace- (2) Guthertz, A.; Leutzsch, M.; Wolf, L. M.; Gupta, P.; Rummelt, S. M.; tate. Even though the ionic radius of Bi is larger than that of Rh and Goddard, R.; Farès, C.; Thiel, W.; Fürstner, A. J. Am. Chem. Soc. 2018, the Bi center hence certainly more exposed, any deleterious back- 140, 3156 (3) Biberger, T.; Gordon, C.; Leutzsch, M.; Peil, S.; Guthertz, A.; Copé- ground reaction should cease (Scheme 6). Furthermore, the differ- ret, C.; Fürstner, A. Angew. Chem. Int. Ed. 2019, 58, 8845. ent radii impart a conical shape on the heterobimetallic core, which (4) Peil, S.; Guthertz, A.; Biberger, T.; Fürstner, A. Angew. Chem. Int. likely translates into a narrower chiral pocket and hence potentially Ed. 2019, 58, 8851. improves the level of asymmetric induction. As these factors might (5) Roşca, D.-A.; Radkowski, K.; Wolf, L. M.; Wagh, M.; Goddard, R.; synergize, it seemed worthwhile to pursue this design concept.8 Thiel, W.; Fürstner, A. J. Am. Chem. Soc. 2017, 139, 2443. In line with our expectations, the structure of the heterobimetal- (6) Meng, Z.; Fürstner, A. J. Am. Chem.Soc. 2019, 141, 805. (7) Werlé, C.; Goddard, R.; Philipps, P.; Farès, C.; Fürstner, A. Angew. ⋅ ⋅ lic analogue [BiRh((S)-PTTL)4] EtOAc (13 EtOAc) in the solid Chem. Int. Ed. 2016, 55, 10760. state shows that the co-crystallized EtOAc is bound to rhodium, (8) Collins, L. R.; Auris, S.; Goddard, R.; Fürstner, A. Angew. Chem. whereas the bismuth center is unligated (Figure 3). Once again, the Int. Ed. 2019, 58, 3557. ligand sphere adopts the a,a,a,a-conformation, presumably be- (9) Collins, L. R.; van Gastel, M.; Neese, F.; Fürstner, A. J. Am. Chem. cause this arrangement places the tert-butyl groups as the most Soc. 2018, 140, 13042. bulky substituents at maximum distance from each other; their orientation toward the Bi center follows from the larger radius of