Regioselective and Enantiospecific Synthesis of Dioxepinesby (Cyclopentadienyl)ruthenium-Catalyzed Condensations ofDiazocarbonyls and Oxetanes

L8o Egger,a Laure Gu8n8e,b Thomas Bgrgi,c and J8rkme Lacoura,*a Department of Organic Chemistry, University of Geneva, Quai Ernest Ansermet 30, 1211 Geneva 4, Switzerland

Fax: (++41)-22-379-3215; phone: (++ 41)-22-379-6062; e-mail: jerome.lacour@unige.chb Laboratory of Crystallography, University of Geneva, Quai Ernest Ansermet 24, 1211 Geneva 4, Switzerlandc Department of Physical Chemistry, University of Geneva, Quai Ernest Ansermet 30, 1211 Geneva 4, Switzerland

Received: May 19, 2017; Revised: June 13, 2017; Published online: August 16, 2017

Supporting information for this article can be found under https://doi.org/10.1002/adsc.201700638.

Abstract: 1,4-Dioxepines result from the decompo-sition of a-diazo-b-keto esters in the presenceof oxetanes using the catalytic combination ofthe (cyclopentadienyl)ruthenium complex[CpRu(CH3CN)3][BArF] and 1,10-phenanthroline.The regioselective [4++3] insertions follow an SN1-like mechanism and occur yet enantiospecifically(es 74%). The retention of configuration was ascer-tained by vibrational circular dichroism (VCD) andsolid state analyses. Furans, products of [4++1] inser-tions, are only observed as traces in the above pro-tocol. To promote their formation under CpRu cat-alysis, it is necessary to use a two-step process withg-halogenated alcohols as substrates.

Keywords: diazo compounds; insertion; metal car-benes; oxetanes; oxonium ylides; retention of con-figuration

Seven- to eleven-membered cycles, also calledmedium-sized rings, are important building blocks,being present in a large variety of biologically rele-vant natural and medicinal products.[1] Due to strainand entropy factors,[2] their synthesis is often challeng-ing but it can be approached with confidence throughcycloaddition, ring-closing metathesis, coupling orring expansion reactions among others.[3] In the par-ticular context of ring-expansions,[4] the use of reac-tive substrates such as readily accessible oxetanes[5]

can be advantageous.[6] In fact, the high nucleophilici-ty of oxygen lone pairs[7] and the important ring strain(&25 kcalmol@1)[8] permit the facile formation of re-active oxonium ylide intermediates that decomposespontaneously into higher ring systems regio- and ste-reoselectively.[9] For instance, under Cu cataly-

sis[9a,b,g,h,l] or photochemical conditions,[10] diazo re-agents react with oxetanes in [4++1] processes to formsubstituted furans (Scheme 1a, [1,2]-shift mechanism).On the contrary, under Rh(II)[11] or Pd(II)[9i] catalysis,decompositions of a-diazo-carbonyls in the presenceof oxetanes (solvent) afford 15-membered macrocy-cles by [3++4++ 4++ 4] condensation (Scheme 1b). Re-cently, it has been shown that combinations of 1,10-phenanthroline (phen) and [CpRu(CH3CN)3][X] {X=PF6 1a or X=BArF, tetrakis[3,5-bis(trifluoromethyl)-phenyl]borate 1b} efficiently catalyze the decomposi-tion of a-diazo-b-keto esters and promote selective1,3-C–H insertions into THF[12] and condensation re-actions with ketones, lactones and cyclic carbonates.[13]

With oxiranes (epoxides), [3++3] insertions lead to theformation of a large variety of 1,4-dioxene moieties.[14]

Herein, in a new development, condensations of oxe-

Scheme 1. Oxonium ylide formation by decomposition ofdiazo reagents in the presence of oxetanes and subsequentreactivity (a, b or c).

Adv. Synth. Catal. 2017, 359, 2918 – 2923 V 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim2918

COMMUNICATIONS DOI: 10.1002/adsc.201700638

tanes 2 and a-diazo-b-keto esters 3 under CpRu-catal-ysis are reported affording 1,4-dioxepines 4(Scheme 1c). To the best of our knowledge, it is thefirst report of 7-membered heterocycle formation inthis type of reaction. Of interest, the 1,4-dioxepinesare obtained as single regioisomers via SN1-like mech-anisms that proceed nevertheless with a certain levelof enantiospecificity (es 74%); the retention of config-uration being ascertained by vibrational circular di-chroism (VCD) analysis.

As just mentioned, efficient decomposition of diaz-ocarbonyls can be provided by the catalytic combina-tion of [CpRu(CH3CN)3] salts 1 and diimine ligands.In many instances, in the presence of carbonyl groupsand cyclic ethers, rather unexpected reactivities werereported.[12–14] These results led us to consider the re-activity of other Lewis bases with the catalytic combi-nation, and oxetanes in particular. The first experi-ments were achieved by dissolving 2-naphthalenyloxe-tane 2a (1.0 equiv.) in a CH2Cl2 solution of complex1b and phen (2.5 mol% each). In practice, diazoace-toacetate 3A (2.0 equiv.) was added in one portionand the mixture was warmed up to 60 88C. From thebeginning, dinitrogen release could be observed andthe reaction mixture was stirred until full conversionof oxetane 2a (16 hours, TLC monitoring). 1H NMRspectroscopic analysis of the crude reaction mixtureled to the identification of ring-expanded 7-mem-bered ring 4aA in low yield (18% NMR).[15] In viewof the importance of 7-membered heterocycles in syn-thesis and medical chemistry, it was decided to studythe reaction further and improve the reactivity. How-ever, despite many attempts (see the Supporting In-formation), the optimization of the reaction wasrather difficult and only a moderate increase in yieldwas achieved. Eventually with a milder procedure, de-creasing the temperature down to 30 88C and using5 mol% of catalyst led to full conversion of starting2a in 16 hours and 4aA was obtained with a yield of35% (Table 1, entry 1). The structure of 4aA was con-firmed by X-ray diffraction analysis (Table 1, Support-ing Information).[16] With these conditions in hand,various diazocarbonyls were reacted. Using diazo re-agents with longer alkyl ester chains, such as ethyl 3Band benzyl 3C, slightly higher yields were obtained inthe presence of oxetane 2a (entries 2 and 3). Not sur-prisingly, the sterically hindered adamantyl group wasfound to have a negative effect on the formation of4aD as an oxetane conversion of only 50% was ob-served after extended reaction time (3 days, entry 4).p-Chlorophenyl substrate 3E was tolerated as product4aE was obtained with a similar yield to that of themethyl derivative 4aA (entry 5). With tosyl or mesylgroups instead of ester moieties, formation of 1,4-di-oxepines 4aF and 4aG was achieved in 39% and 38%,respectively (entries 6 and 7). Furthermore, introduc-tion of an a-keto ester group in place of the acetyl

moiety led to diester dioxepine 4aH in 19% yieldonly (entry 8). The reason for the lower reactivitycould be the lesser nucleophilicity of the enolate in-termediate (vide infra, Scheme 2, species D with R2 =CO2Et). Alternatively, 4aH was found to be particu-larly sensitive to acidic conditions and the lower yield

Table 1. Diazocarbonyl reactivity.[a]

[a] Reaction conditions : 1b (5 mol%), phen (5 mol%),CH2Cl2, 30 88C, 16 h unless otherwise noted, c= 0.5 M;conversions (conv.) determined by 1H NMR spectrosco-py. ORTEP view of the crystal structure of 4aA. Thermalellipsoids are drawn at 50% probability. Ad=1-adaman-tyl; Ts=p-toluenesulfonyl and Ms=methanesulfonyl.

Adv. Synth. Catal. 2017, 359, 2918 – 2923 V 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim2919

COMMUNICATIONS asc.wiley-vch.de

could be the result of a degradation during purifica-tion (silica gel chromatography). Finally, trisubstitutedenol ether 4aI was obtained in 35% yield using an al-dehyde-containing diazo ester reagent (entry 9).

After this first screen of diazo reactivity, the reac-tion was performed with various oxetanes(Table 2).[17] 1-Naphthalenyl substrate 2b afforded thecorresponding 7-membered ring 4bB with a loweryield than 4aB, due probably to the increased sterichindrance (Table 2, entry 1). A higher yield of 4cBwas however obtained with 1-phenyloxetane (entry 2,yield 54%, Table 2).[18]

Globally, this increased reactivity was confirmedwith phenyl substituents carrying electron-withdraw-ing (F, Cl) or electron-donating (Me) groups at thepara position of the aromatic ring (4dB–4fB, entries 3to 5, 55–61%); the reactivity being lower only in thepresence of a bulkier mesityl group (4gB, entry 6,41%). Finally, p-tolyloxetane 2f was reacted with

three other a-diazo-b-keto esters. While similar re-sults were obtained with isobutyl and 2-methoxyethylchains (4fJ–4fK, entries 7 and 8, 52% and 53% re-spectively), a slightly lower yield was observed withthe trichloroethyl carboxylate (Troc) ester (4fL,entry 9, 40%).

Additionally and of importance for the mechanisticdiscussion, the enantiospecificity of the reaction wasexamined. Enantioenriched oxetane (R)-2c (ee>99%) was synthesized under reported conditions.[9l]

Treatment under the optimized conditions led to 1,4-dioxepine (++)-4cB with exactly the same yield [Eq.(1)]. Chiral stationary phase HPLC analysis indicatedan enantiomeric ratio of 87:13 for 4cB and hencea chirality transfer (enantiospecificity) of 74%.

Interestingly, in terms of stereospecific transforma-tions, only SN2 ring expansions of oxetanes have beenreported.[9d,e] Care was thus taken to determine theabsolute configuration of product (++)-4cB. It was es-tablished by vibrational circular dichroism (VCD). Acomparison between measured VCD spectra for both(++)-4cB and (@)-4cB, obtained enantiomerically pureby semi-preparative CSP HPLC separation of (:)-4cB with the theoretical spectrum of the most stableconformers (Boltzmann distribution) of (R)-4cB wasperformed (Figure 1). A good agreement was ach-ieved between experimental and theoretical spectraof (++) and (R)-configurated 4cB. This finding is inagreement with what was observed in the X-ray crys-tallographic study of (++)-4cB (see the Supporting In-formation). Due to the lack of heavy atoms in thestructure, the Flack parameter could not be deter-mined with good precision but its value was found tobe around zero [x=@0.13(16)]. These results clearlyindicate that the formation of 4cB occurs witha global retention of configuration.

At this stage, a mechanistic rationale can be pro-posed (Scheme 2). First, ruthenium salt 1b reacts withphen to generate complex [CpRu(phen)(CH3CN)][BArF]. Then, upon dissociation of the acetonitrileligand, catalytically active 16-electron species A isformed. After nucleophilic attack of diazo reagents 3onto A and dinitrogen extrusion, metal carbenes oftype B are generated. Intermediates B behave as elec-trophilic Fischer carbenes and nucleophilic attacks ofLewis basic oxetanes yield metal-bound oxoniumylide intermediates C. Due to ring strain and the elec-trophilic activation of the aromatic rings, regioselec-

Table 2. Oxetane reactivity.[a]

[a] Reaction conditions : 1b (5 mol%), phen (5 mol%),CH2Cl2, 30 88C, c=0.5 M.

Adv. Synth. Catal. 2017, 359, 2918 – 2923 V 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim2920

COMMUNICATIONS asc.wiley-vch.de

tive C–O bond cleavage occurs to form stabilized car-bocationic intermediates D. The neighboring carbonylgroups then trap the carbocations leading to fast in-tramolecular 7-membered ring formation (D!E).After decomplexation, catalyst A and the 1,4-dioxe-pines are released. Importantly, this mechanistic ra-tionale explains not only the global SN1-like regiose-lectivity but also the observed (partial) retention ofconfiguration with substrate (R)-4cB. In fact, the

proximity between the carbonyl groups and the sec-ondary carbocationic center in D ensures rapid 7-endo-trig cyclization.[19] This avoids, to a large extent,the scrambling of the stereotopic faces of carbeniumions D (through internal C–C rotations) and the race-mization.

Also of mechanistic importance, acid-sensitivefuran 5c could be isolated from the reaction of oxe-tane 2c with 3B (Scheme 3). In fact, careful analysisrevealed a proportion of 5–10% of 5c in crude mix-tures (GC-FID estimation).[20] The formation of 5ccan be explained by a second route starting from in-termediates C or D (Scheme 3). At that stage, directC-alkylation gives rise to furan 5c, of a structure simi-lar to that described by Nozaki and Noyori.[9a,b] More-over, 5c is isolated as a single diastereomer, an obser-vation that tends to indicate a highly stereoselectiveprocess as described in the literature.[9a,b,l,21] The pre-ferred formation of 4cB over 5c can be explained byBaldwinQs rules since 7-endo-trig cyclizations are fa-vored over 5-endo pathways.[19]

Finally, Bull and co-workers have described recent-ly the formation of 1,4-dioxenes in two steps using b-halogeno alcohols as substrates.[22] In a first step,CpRu-catalyzed O–H insertions of a-diazo-b-ketoesters occur, followed by intramolecular cyclizationsunder basic conditions. Using this protocol, only O-al-kylations of ketoenolate intermediates are observedleading to a large variety of dioxenes. In view of theefficiency of the method, care was taken to apply itfor the synthesis of 1,4-dioxepines. Reactions underCpRu catalysis of alcohols 6c and 6h with 3B affordedhalogenated keto esters 7c and 7h in good yields, 90%and 79% respectively (Scheme 4). These compoundswere then treated with NaH (1.2 equiv.) in DMF.However, instead of forming 1,4-dioxepine products,reactions afforded oxolanes 5c and 5h exclusively(62% and 70% yields). Clearly, the ketoenolate inter-mediates undergo C-alkylation preferentially to formthe corresponding furans. According to BaldwinQsrules, both 5-exo and 7-exo-tet cyclizations are fa-

Figure 1. Experimental VCD spectra (CD2Cl2, 298 K) of(++)-4cB (black) and (@)-4cB (red). Calculated spectrum of(R)-4cB (green).

Scheme 3. Formation of furan 5c.

Scheme 2. Mechanistic proposal.

Adv. Synth. Catal. 2017, 359, 2918 – 2923 V 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim2921

COMMUNICATIONS asc.wiley-vch.de

vored; the reactivity under BullQs conditions indicatesa clear bias toward the 5-membered ring formation.[19]

In conclusion, the combination of [CpRu(CH3CN)3][BArF] and 1,10-phenanthroline leads to a new reac-tivity between a-aryloxetanes and carbenes derivedfrom a-diazo-b-keto esters. 1,4-Dioxepines are ob-tained in moderate yields but with a good enantiospe-cificity in an original SN1-like transformation, as de-termined by VCD and X-ray diffraction analyses.BaldwinQs rules for the ring closure can be applied toexplain the preferential formation of dioxepines 4 vs.furan 5.

Experimental Section

General Procedure

In a 2-mL screw-cap vial equipped with a magnetic stirringbar, 1,10-phenanthroline (3 mg, 16 mmol, 5 mol%) and[CpRu(CH3CN)3][BArF] (18 mg, 16 mmol, 5 mol%) were dis-solved in 0.60 mL of dry dichloromethane. The vial wasflushed with nitrogen and capped. The resulting deep red so-lution was stirred for 10 minutes at 25 88C before the additionof oxetane 2 (0.32 mmol, 1 equiv.) and the desired diazo de-rivative 3 (0.64 mmol, 2 equiv.). The solution was stirred at30 88C until full conversion (1H NMR monitoring). The crudemixture was purified by column chromatography (pentane/Et2O, SiO2) to afford 1,4-dioxepine 4.

Acknowledgements

We thank the University of Geneva and the Swiss NationalScience Foundation for financial support. We acknowledgethe contributions of the Sciences Mass Spectrometry (SMS)platform at the Faculty of Sciences, University of Geneva.

References

[1] a) J. D. Rainier, in: Metathesis in Natural Product Syn-thesis, Wiley-VCH, Weinheim, 2010, pp 87–127; b) I. V.

Hartung, H. M. R. Hoffmann, Angew. Chem. 2004, 116,1968–1984; Angew. Chem. Int. Ed. 2004, 43, 1934–1949.

[2] G. Illuminati, L. Mandolini, Acc. Chem. Res. 1981, 14,95–102.

[3] a) G. A. Molander, Acc. Chem. Res. 1998, 31, 603–609;b) L. Yet, Chem. Rev. 2000, 100, 2963–3008; c) A. Dei-ters, S. F. Martin, Chem. Rev. 2004, 104, 2199–2238;d) A. Hussain, S. K. Yousuf, D. Mukherjee, RSC Adv.2014, 4, 43241–43257.

[4] a) J. R. Donald, W. P. Unsworth, Chem. Eur. J. 2017,23, 8780–8799; b) A. Sapegin, A. Osipyan, M. Krasavin,Org. Biomol. Chem. 2017, 15, 2906–2909; c) L. G.Baud, M. A. Manning, H. L. Arkless, T. C. Stephens,W. P. Unsworth, Chem. Eur. J. 2017, 23, 2225–2230;d) Z.-L. Li, X.-H. Li, N. Wang, N.-Y. Yang, X.-Y. Liu,Angew. Chem. 2016, 128, 15324–15328; Angew. Chem.Int. Ed. 2016, 55, 15100–15104; e) J. E. Hall, J. V. Mat-lock, J. W. Ward, K. V. Gray, J. Clayden, Angew. Chem.2016, 128, 11319–11323; Angew. Chem. Int. Ed. 2016,55, 11153–11157; f) C. Kitsiou, J. J. Hindes, P. I’Anson,P. Jackson, T. C. Wilson, E. K. Daly, H. R. Felstead, P.Hearnshaw, W. P. Unsworth, Angew. Chem. 2015, 127,16020–16024; Angew. Chem. Int. Ed. 2015, 54, 15794–15798.

[5] a) J. A. Bull, R. A. Croft, O. A. Davis, R. Doran, K. F.Morgan, Chem. Rev. 2016, 116, 12150–12233; b) O. A.Davis, J. A. Bull, Synlett 2015, 26, 1283–1288.

[6] a) Z. Zhang, J. Wang, Tetrahedron 2008, 64, 6577–6605;b) M. P. Doyle, D. C. Forbes, Chem. Rev. 1998, 98, 911–936; c) A. Padwa, M. D. Weingarten, Chem. Rev. 1996,96, 223–270; d) A. Padwa, S. F. Hornbuckle, Chem.Rev. 1991, 91, 263–309.

[7] J. A. Burkhard, G. Wuitschik, M. Rogers-Evans, K.Mgller, E. M. Carreira, Angew. Chem. 2010, 122, 9236–9251; Angew. Chem. Int. Ed. 2010, 49, 9052–9067.

[8] H. K. Eigenmann, D. M. Golden, S. W. Benson, J. Phys.Chem. 1973, 77, 1687–1691.

[9] a) H. Nozaki, H. Takaya, S. Moriuti, R. Noyori, Tetra-hedron 1968, 24, 3655–3669; b) H. Nozaki, S. Moriuti,H. Takaya, R. Noyori, Tetrahedron Lett. 1966, 7, 5239–5244; c) S. Cenini, G. Cravotto, G. B. Giovenzana, G.Palmisano, S. Tollari, Tetrahedron 1999, 55, 6577–6584;d) C. Gronnier, S. Kramer, Y. Odabachian, F. Gagosz,J. Am. Chem. Soc. 2012, 134, 828–831; e) E. D. Butova,A. V. Barabash, A. A. Petrova, C. M. Kleiner, P. R.Schreiner, A. A. Fokin, J. Org. Chem. 2010, 75, 6229–6235; f) C. A. Malapit, A. R. Howell, J. Org. Chem.2015, 80, 8489–8495; g) K. Ito, M. Yoshitake, T. Katsu-ki, Tetrahedron 1996, 52, 3905–3920; h) K. Ito, T.Fukuda, T. Katsuki, Synlett 1997, 387–389; i) M. Kita-mura, M. Kisanuki, K. Kanemura, T. Okauchi, Org.Lett. 2014, 16, 1554–1557; j) S. K. Pawar, D. Vasu, R.-S.Liu, Adv. Synth. Catal. 2014, 356, 2411–2416; k) R. L.Sahani, R.-S. Liu, Chem. Commun. 2016, 52, 7482–7485; l) M. M. C. Lo, G. C. Fu, Tetrahedron 2001, 57,2621–2634.

[10] a) W. Kirmse, P. Van Chiem, V. Schurig, TetrahedronLett. 1985, 26, 197–200; b) K. Friedrich, U. Jansen, W.Kirmse, Tetrahedron Lett. 1985, 26, 193–196.

[11] D. Rix, R. Ballesteros-Garrido, W. Zeghida, C. Bes-nard, J. Lacour, Angew. Chem. 2011, 123, 7446–7449;Angew. Chem. Int. Ed. 2011, 50, 7308–7311.

Scheme 4. 5- vs. 7-membered ring cyclizations.

Adv. Synth. Catal. 2017, 359, 2918 – 2923 V 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim2922

COMMUNICATIONS asc.wiley-vch.de

[12] C. Tortoreto, T. Achard, W. Zeghida, M. Austeri, L.Gu8n8e, J. Lacour, Angew. Chem. 2012, 124, 5949–5953; Angew. Chem. Int. Ed. 2012, 51, 5847–5851.

[13] a) C. Tortoreto, T. Achard, L. Egger, L. Gu8n8e, J.Lacour, Org. Lett. 2016, 18, 240–243; b) M. Austeri, D.Rix, W. Zeghida, J. Lacour, Org. Lett. 2011, 13, 1394–1397.

[14] T. Achard, C. Tortoreto, A. I. Poblador-Bahamonde, L.Gu8n8e, T. Bgrgi, J. Lacour, Angew. Chem. 2014, 126,6254–6258; Angew. Chem. Int. Ed. 2014, 53, 6140–6144.

[15] 1H NMR monitoring of the crude reaction mixture in-dicates products of degradation and of polymerizationof 2. Their isolation was not possible.

[16] CCDC 1551193 and CCDC 1551194 contain the supple-mentary crystallographic data for this paper,products 4aA and (++)-4cB, respectively. These datacan be obtained free of charge from The

Cambridge Crystallographic Data Centre viawww.ccdc.cam.ac.uk/data_request/cif.

[17] Alkyl-substituted oxetanes were unreactive under theoptimized reaction conditions.

[18] The reaction was scalable up to a 1 g scale (8 mmol)(see the Supporting Information).

[19] a) J. E. Baldwin, J. Chem. Soc. Chem. Commun. 1976,734–736; b) H. B. Buergi, J. D. Dunitz, Acc. Chem. Res.1983, 16, 153–161.

[20] Overlapping 1H NMR signals of 1,4-dioxepine 4 andfuran 5 rendered the spectroscopic analysis difficult fora quantitative study.

[21] A selective and complete decomposition of the seconddiastereomer upon purification might be an alternativeexplanation.

[22] O. A. Davis, R. A. Croft, J. A. Bull, J. Org. Chem. 2016,81, 11477–11488.

Adv. Synth. Catal. 2017, 359, 2918 – 2923 V 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim2923

COMMUNICATIONS asc.wiley-vch.de