Research Collection
Doctoral Thesis
Discovery, Development and Study of Carbenoid Mediated Reactions From Olefin
Author(s): Sarria Toro, Juan M.
Publication Date: 2014
Permanent Link: https://doi.org/10.3929/ethz-a-010350749
Rights / License: In Copyright - Non-Commercial Use Permitted
This page was generated automatically upon download from the ETH Zurich Research Collection. For more information please consult the Terms of use.
ETH Library Diss. ETH No. 22272
Discovery, Development and Study of Carbenoid Mediated Reactions From Olefin Metathesis to Cyclopropanation
A thesis submitted to attain the degree of DOCTOR OF SCIENCES of ETH ZURICH (Dr. sc. ETH Zurich)
Presented by Juan Manuel Sarria Toro
M. Sc. – Chemistry, Universidad Nacional de Colombia
born on 24.08.1984 citizen of Colombia
Accepted on recommendation of
Prof. Dr. Peter Chen, examiner Prof. Dr. Antonio Togni, co-examiner
Zürich, 2014
“Es mejor ser rico que pobre” Antonio Cervantes - Kid Pambelé
Acknowledgments – Agradecimientos – Захвалница – Danksagung
First and foremost I would like to thank my supervisor Prof. Dr. Peter Chen for giving me the chance to join his research group. During my stay he managed to show me subtly the difference between impossible and just difficult. Thanks to his encouragement and trust I have discovered that more often than not, the limit of what is chemically possible is only imposed by our lack of understanding. For these years of personal and professional development I will be always in his debt.
I would also like to thank my co–examiner Prof. Dr. Antonio Togni for playing a crucial role in my connection to ETH. One could rightfully say that all of this is his fault.
I am very grateful to PD. Dr. Andreas Bach, for being always there to solve all my questions, it is very reassuring to know that there is always someone who knows what needs to be done.
Dr. Sebastian Torker had the great challenge of taking my rebel character and focus it in a productive direction, thanks for not giving up. It was then thanks to Dr. Tim den Hartog’s obsessive attention to detail and rigorousness that the best of this PhD was possible. For his partnership, friendship and delicious dinners, organized together with the endlessly cheerful Dr. Elena Melillo, I will always be grateful.
Dr. Eric Couzijn and Dr. Krista Lynn Vikse contributed with inspiring academic and non– academic conversations to the completeness of my PhD. Krista’s help with the proof reading of this document was invaluable.
My stay in the Chen group would have not been the same without the cheer, the musical taste and the fun of my lab mates Augustin Armand Senghor Tchawou Wandji and David Hans Ringger, best luck to you in your future career.
Former and present group members have cheerfully shaped my image of what research and life abroad are; I am very happy to have met so many different cultures and the points of view that come along with them. Merci beaucoup Dr. Marc-Etienne Moret for trusting me his hood, there is certainly magic happening there, Dr. Daniel Serra for explaining to me the true nature of the T–shaped intermediate, Dr. Déborah Mathis for patiently guiding me though my very first Ski slope and Dr. Laurent Batiste for being such a good travel partner. Díky moc Marek Bot for opening my eyes to the true facts of Charlie.
i
Acknowledgments – Agradecimientos – Захвалница – Danksagung
Mulţumesc mult Dr. Mihai Răducan for gains. 谢谢 Yanan Miao for enthusiastically taking over my workplace, I am sure you will make good use of it. Hvala puno Inesa Semić for keeping the Balkan beat going in the group.
Ich bin auch sehr dankbar meiner deutschsprachigen Schweizer Kollegen, die während den Winter und Sommer Ausflüge mir ein Teil ihrer wunderschönes Land gezeigt haben. Danke schön Joël Gubler, Raphael Oeschger, Lukas Fritsche, Stefan Jungen und Stefan Künzi (Viel Erfolg bei der Cyclopropanierung).
I am especially thankful to our secretarial team, which has been always ready to solve my day–to–day questions about Switzerland and its ways. Merci vielmal! Mirella Rutz, Barbara Loepfe and Anke Witten.
The technical help provided by the permanent ETH staff during those years is greatly acknowledged. Particularly from our always ready–to–help Armin Limacher and from Carmela Pansino at the HCI–shop.
Quisiera también agradecer de forma especial la sincera amistad de mis queridos amigos hispanoparlantes, quienes han constituido durante estos años mi familia en Europa. Gracias por ser ejemplo viviente de que el idioma en lugar de crear barreras se disfruta mejor reduciéndolas. Gracias por la ayuda, la compañía, las risas, las fiestas, los chistes, pero por sobre todo por ser ejemplo de excelencia. Muchas gracias a Dr. Esteban Mejía, Luisa Duque, Dr. Luis Ladino, Elisa Ladino, Dr. Rafael Rodriguez, Dr. Vanessa Landaeta, Dr. Gustavo Santiso, Nelli Sanne, Dr. Fernando Rascón, Dr. Fernando Cortés, Sofia Riaño, Mónica Angarita y Dr. David Plaza. Gracias a los locos que desde pequeño me aguantan y todavía se acuerdan de mí: Omar Avella, Nicolás López y Juan Esteban Forero.
Merecen un especial reconocimiento mis mentores en la Universidad Nacional de Colombia Prof. Dr. Marco Fidel Suarez Herrera y Prof. Dr. Luca Fadini. Por su confianza, su motivación y su entrega estaré siempre en deuda.
Изразито ми је драго да ми се пружила прилика да упознам много изванредних људи са којима имам пуно тога заједничког. Прелепо ме је изненадило откриће да невезано за удаљеност наших земаља или језичких разлика, свеједно делимо исте страхове, наде и циљеве. Хвала пуно Ивани Фуртули, Илији Кобилијанском (знаш да припадаш овде), Драгани и Штефану Хоесер, др. Селени Милићевић и др. Марку Септону, Петри и Марку Танасковић, Милици и Томасу Рајц, Бојани и Бошку Ступар, Драгани Аврамовић и ii
Acknowledgments – Agradecimientos – Захвалница – Danksagung
Вукашину Нешовићу, Ани Јокановић и Анђелији Вукић што сте поделили са мном оно што нас чини људима.
Не могу да изразим колико сам захвалан на прилици да упознам др. Марију Јовић током доктората. Само захваљујући њеној љубави, инспирацији и охрабривању мој ум је остао разуман све ове године. Хвала што си ме узела за руку и показала ми да постоји диван свет спреман да га заједно откријемо.
Finalmente quiero agradecer a mi familia, a mis padres Miguel Angel y Yolanda, por traerme a la vida en más de una ocasión y a mis hermanas Angela Yolanda y Martha Patricia por apoyarme en todas y cada una de mis locuras. Gracias a su fe ciega y apoyo incondicional he podido seguir mis sueños que hoy están aquí plasmados. Este trabajo es de ustedes. Muchas gracias.
iii iv
Published Parts of this thesis
Reviewed Articles
1. “A Palladium-Catalyzed Methylenation of Olefins Using Halomethylboronate Reagents” Tim den Hartog, Juan Manuel Sarria Toro, and Peter Chen Organic Letters, 2014, 16, 1100-1103
2. “A lithiomethyl trimethylammonium reagent as a methylene donor” Tim den Hartog, Juan M. Sarria Toro, Erik P. A. Couzijn and Peter Chen Chemical Communications, 2014, 50, 10604-10607
3. “Cyclopropanation of styrenes and stilbenes using lithiomethyl trimethylammonium triflate as methylene donor” Juan M. Sarria Toro, Tim den Hartog and Peter Chen Chemical Communications, 2014, 50, 10608-10610 Poster Presentations 1. “New metathesis catalysts based on asymmetric phosphinosulfonate bidentate ligands for the A-ROMP of cycloolefins”
Presented at the XXIVth International Conference on Organometallic Chemistry (ICOMC XXIV), Taipei, Taiwan, July 18-23 2010.
2. “Influence of the anchoring group of a bidentate ligand on the activity and selectivity of a ruthenium metathesis catalyst”
Presented at the Gordon Research Conference on Organometallic Chemistry, Salve Regina University, Newport, RI, USA, July 10-15, 2011.
3. “Gas phase study of transition metal complexes containing N-C ylides”
Presented at the International Symposium on Reactive Intermediates and Unusual Molecules (ISRIUM), Ascona, Switzerland, July 8-13, 2012.
Presented at the XXVth International Conference on Organometallic Chemistry (ICOMC XXV), Lisbon, Portugal, September 2-7, 2012.
v
vi
Table of contents
Abstract ...... xi Zusammenfassung ...... xiii 1. Introduction ...... 1 1.1. Scope ...... 1 1.2. Olefin metathesis ...... 1 1.1.1 Discovery of olefin metathesis and mechanistic proposals ...... 1 1.1.2 Schrock’s high oxidaton state early transition metal alkylidene complexes. .. 4 1.1.3 Grubbs Ruthenium based catalysts...... 6 1.1.4 Heterogeneous systems ...... 8 1.1.5 Applications of olefin metathesis ...... 9 1.1.6 Stereoselective olefin metathesis ...... 11 1.3. Methylenation Reactions ...... 12 1.3.1. Three–membered rings ...... 12 1.3.2. Bonding in three–membered rings ...... 14 1.3.3. Carbene equivalents or carbenoids for the synthesis of three–membered rings ...... 15 1.1.7 Nucleophilic methylenation ...... 17 1.1.8 Electrophilic methylenation ...... 19 2. Ruthenium catalyzed alternating ring opening metathesis polymerization (A- ROMP) ...... 23 2.1. Introduction ...... 23 2.1.1. Studies on the mechanism of olefin metathesis ...... 23 2.1.2. Rationally designed A-ROMP catalysts ...... 29 2.1.3. Optimization of the catalysts for chemo and stereoselectivity ...... 30 2.2. Phosphinosulfonate system ...... 35 2.2.1. Synthesis of a ruthenium metathesis catalyst bearing an asymmetric phosphino sulfonate ligand ...... 35 2.2.2. Structural characterization of complex 17 ...... 36 2.2.3. Alternating ROMP catalyzed by complex 17 ...... 38 2.2.4. Calculation of the catalyst commitment ...... 40 2.2.5. Symmetrically substituted phosphino sulfonate complex ...... 43 2.2.6. Chloride replacement with bulky sulfonate anions ...... 44 2.2.7. Polymerization experiments with sulfonate complexes ...... 48 2.2.8. Attempts to enhance the chemoselectivity of the phosphino sulfonate systems...... 50 2.3. Discussion ...... 53 2.4. Conclusions ...... 56 3. Lithiomethyl trimethylammonium salts as methylene donors ...... 59 vii
Table of contents
3.1. Introduction ...... 59 3.1.1. Yildes: bonding and reactivity ...... 59 3.1.2. Nitrogen ylides ...... 60 3.2. Lithiomethyl trimethylammonium reagents as methylene donors...... 63 3.2.1. Soluble tetramethylammonium salts ...... 63 3.2.2. Bonding in the lithiomethyl trimethylammonium ion ...... 66 3.2.3. Reactivity of lithiomethyl trimethylammonium triflate with aldehydes, ketones and imines ...... 68 3.2.4. Reactivity of lithiomethyltrimethylammonium triflate with styrenes ...... 70 3.2.5. Reactivity of lithiomethyl trimethylammonium reagents with transition metals ...... 75 3.3. Discussion...... 81 3.4. Conclusions ...... 84 3.5. Acknowledgments ...... 85 4. Palladium catalyzed methylenation of olefins using halomethylboronate reagents ...... 87 4.1. Introduction ...... 87 4.1.1. Nucleofuge and electrofuge alternatives ...... 87 4.1.2. Palladium catalyzed olefin transformations ...... 89 4.1.3. Palladium–catalyzed cyclopropanation of olefins ...... 90 4.2. Cyclopropanation of olefins using halomethylboronates ...... 92 4.2.1. Our hypothesis ...... 92 4.2.2. Methylenation of norbornene ...... 92 4.2.3. Methylenation of other olefins ...... 95 4.3. Mechanistic considerations ...... 97 4.3.1. Preliminary mechanistic investigations ...... 99 4.4. Discussion...... 102 4.5. Conclusions ...... 104 4.6. Acknowledgments ...... 105 5. Conclusions and Outlook ...... 107 6. Experimental Part ...... 111 6.1. General Remarks ...... 111 6.2. Ruthenium Catalyzed Ring Opening Metathesis Polymerization (Chapter 2) .. 112 6.2.1. Polymerization experiments ...... 112 6.2.2. Synthesis of ligands and complexes ...... 113 6.3. Solution phase chemistry of lithiomethyl trimethylammon reagents (Chapter 3) 119 6.3.1. Tetramethylammonium salts ...... 119 6.3.2. Notes on the influence of transition metal contamination on the methylenation reactions ...... 121 viii
Table of contents
6.3.3. Methylenation of aldehydes, ketones and imines ...... 121 6.3.4. Cyclopropanation of styrenes and stilbenes ...... 124 6.3.5. Kinetic measurements ...... 127 6.3.6. Reaction of lithiomethyl trimethylammonium reagents with transition metal complexes ...... 131 6.4. Palladium catalyzed methylenation of olefins using halomethylboronates reagents (Chapter 4) ...... 132 6.4.1. Synthesis and characterization of products from our methylenation reaction ...... 132 6.4.2. Synthesis and characterization of authentic samples ...... 134 6.5. X-Ray crystallography ...... 136 7. Appendices ...... 139 7.1. Crystal structure of complex 21 ...... 139 7.2. Crystal structure of complex 23 ...... 139 7.3. NMR spectra of the isolated mixture of 24 and 25...... 140 7.4. Computational Details ...... 140 7.5. Evidence of formation of 4-fluoro-1-cyclopropylbenzene reaction mixture (71f) ...... 144 7.6. 1H spectrum of complex 81 ...... 146 7.7. Confirmation by 1H NMR of the formation of exo-tricyclo[3.2.1.02.4]octane (87) ...... 146 7.8. Confirmation by 1H NMR of the formation of cis-bicyclo[6.1.0]nonane (91a) .... 147 7.9. Confirmation by 1H NMR of the formation of (2-cyclopropylethyl)benzene (92a) ...... 147 7.10. Confirmation by 1H NMR of the formation of cyclopropylbenzene (71a) ...... 148 7.11. NMR spectra of (2-cyclopropylethyl)benzene (91b) ...... 148 8. Bibliography ...... 151 Curriculum Vitae ...... 163
ix
x
Abstract
Carbenoids are versatile reagents in the synthetic organic chemist’s toolbox, capable of undergoing a wide range of transformations ranging from olefin cyclopropanation to transition metal–catalyzed olefin metathesis. This dissertation combines the work in the field of ruthenium–catalyzed alternating ring opening metathesis polymerization (A- ROMP) with the quest for new reactivity in the area of olefin cyclopropanation.
Chapter 2 deals with the structural modification of a ruthenium–catalyst framework that has previously been employed to obtain alternating norbornene–cyclooctene copolymers. Substitution of the original anchoring phenolate for a sulfonate group provided a new catalyst that showed increased activity, chemo– and stereoselectivity.
Substitution of the chloride anion for bulky sulfonates increased the stereoselectivity as expected, but also diminished the activity and, surprisingly, also the chemoselectivity. The effects of the new anchoring group could be explained in terms of geometric parameters derived from experimentally–obtained molecular structures, and the increased intrinsic reactivity could be rationalized by DFT–calculated “catalyst commitment”
Chapter 3 describes the synthesis, characterization and reactivity studies of soluble lithomethyltrimethylammonium reagents, obtained by simple deprotonation of a tetramethylammonium salt in which the anion provides solubility in organic solvents.
This class of reagents were first described by Wittig in 1947 but their insolubility limited their usability and made them difficult to study. The soluble variants reported here are
xi
Abstract promising alternatives for the methylenation of aldehydes, ketones and imines, and for the unprecedented cyclopropanation of styrenes and stilbenes by nucleophilic reagents.
Chapter 4 describes the discovery and initial development of a palladium–catalyzed cyclopropanation of electron–rich olefins where the methylene donor is a commercially available, bench–stable halomethylboronate reagent.
Optimal conditions for the cyclopropanation of highly–strained norbornene were found using both Pd0 and PdII precatalysts. Other types of olefins could also be cyclopropanated, however lower conversions and the formation of –hydride elimination products were observed. Preliminary observations point towards a possible Pd0–PdII ‘diverted Heck’ mechanism for this methylenation. Further mechanistic work is underway to find ways to improve catalyst stability and selectivity.
xii
Zusammenfassung
Carbenoide sind vielseitige Reagenzien in dem organischen Synthesechemiker Werkzeugkasten, die eine grosse Auswahl Transformationen von Olefin- Cyclopropanierung bis zur übergangsmetall-katalysierten Olefin–Metathese durchführen können. Diese Doktorarbeit versucht die Ergebnisse zum Thema alternierend Ruthenium–katalysierten Ringsöffnungsmetathese–Polymerisation (A- ROMP) mit der Suche nach neuen Reaktivität im Bereich der Olefin-Cyclopropanierung zu verbinden.
Kapitel 2 befasst sich mit der strukturellen Modifikation der Grundstruktur des zuvor verwendeten Ruthenium-Katalysators, mit dem alternierenden Norbornen–Cycloocten Copolymere erhalten worden. Austausch der Befestigungsgruppe von ursprünglichen Phenolat mit einem Sulfonat, ergab einen neuen Katalysator, welcher erhöhte Aktivität, Chemo-und Stereoselektivität zeigte.
Ersetzung des Chlorids durch sterisch anspruchsvollen Sulfonate erhöhte, wie erwartet, die Stereoselektivität, jedoch abnahm die Aktivität und unerwarteterweise auch die Chemoselektivität. Die Eigenschaften des neuen Katalysators lassen sich durch den geometrischen Parameter der experimentell erhaltenen molekularen Strukturen erklären. Die erhöhte intrinsische Reaktivität wurde mit Hilfe von berechneten (DFT) „Katalysator Engagement“ verstanden werden.
Kapitel 3 beschreibt die Synthese, Charakterisierung und Reaktivitätsstudien eines löslichen Lithiummethyltrimethylammonium Reagenzes, erhalten durch die einfache Deprotonierung eines Tetramethylammoniumsalzes, bei welchem das Anion die Löslichkeit in organischen Lösungsmitteln bewirkt. Solche Reagenzien wurden zuerst von Wittig im Jahr 1947 beschrieben, obwohl ihre Unlöslichkeit die Nutzbarkeit einschränkte und dadurch die Untersuchungen erschwerte. Die in dieser Arbeit beschriebenen löslichen Varianten sind vielversprechend Alternativen für die
xiii
Zusammenfassung Methylenierung von Aldehyden, Ketonen und Iminen, sowie für die neuartige Cyclopropanierung von Styrolen und Stilbenen durch nucleophile Reagenzien.
Kapitel 4 beschreibt die Entdeckung und anfängliche Entwicklung einer Palladium– katalysierten Cyclopropanierung von elektronenreichen Olefinen, in der der Methylen- Donor ein kommerziell erhältliches, luftstabiles halomethylboronate Reagenz ist. Optimale Bedingungen für die Cyclopropanierung hochgespannter Norbornen wurden mit Pd0- und PdII- Katalysatoren gefunden.
Olefinen anderer Art wurden auch cyclopropaniert worden, jedoch sowohl niedrigere Ausbeute als auch die Bildung von -Hydrid-Eliminierungsprodukten wurden beobachtet. Erste Beobachtungen verweisen auf einen möglichen Pd0–PdII „umgeleiteten Heck“ Mechanismus für diese Methylenierung hin.
xiv
1. Introduction
1.1. Scope
This dissertation is divided in two main topics. First, in chapter 2, the development of a ruthenium metathesis catalysts bearing a phosphino–sulfonate ligand for the alternating ring opening metathesis polymerization (A-ROMP) of norbornene and cyclooctene will be discussed. The synthesis and structural characteristics of this type of catalysts and how they influence the obtained polymer will be described. The relative activity of the novel system will be compared to the previous phosphino–phenolate framework and explained by means of density functional calculations. Lastly, the attempts of controlling the stereoselectivity of the catalysts by replacement of the chloride anion with sterically demanding sulfonates will be reported.
In the second part of this dissertation, chapters 3 and 4, two novel approaches to the synthesis of three–membered rings will be discussed. The first one involves the generation of a soluble ‘nitrogen ylide’ derived from a tetramethylammonium salt possessing a solubilizing anion. This reagent can perform nucleophilic addition to aldehydes, ketones, imines and styrenes to yield epoxides, aziridines and cyclopropanes efficiently. The second approach uses the advantages of Pd catalysis to transfer a methylene unit from commercially available halomethylboronate salts to a number of different olefins generating cyclopropanes.
1.2. Olefin metathesis
The exchange of methylene termini of a pair of olefins mediated by a transition metal catalyst is known as olefin metathesis (Scheme 1.1).
Scheme 1.1 Prototypical olefin metathesis reaction
1.1.1 Discovery of olefin metathesis and mechanistic proposals
Started by Ziegler and Natta in the early 1950s, the studies on the polymerization of olefins with transition metals ultimately lead to the discovery of what we know today as olefin metathesis. Originally discovered in the late 1950s at the industrial laboratories of Du Pont, Standard Oil and Phillips Petroleum, the first reports of cycloolefin polymerization with supported molybdenum or titanium seem to deal with different reactions.1 Initially, the polymerization of norbornene with a Ziegler-Natta catalyst was 1
Discovery, Development and Study of Carbenoid Mediated Reactions thought to be a simple insertion polymerization.2 Eleuterio’s work on the disproportionation of propene to produce a mixture of ethylene and 2-butenes3 and the polymerization of cyclic olefins4 marked an early understanding of the unique nature of this transformation. It was later recognized by Calderon that these reactions were in essence equivalent.5 He was the first one to use the name of “Olefin Metathesis”. Early industrial applications used ill-defined systems commonly composed of combinations of tungsten and molybdenum sources and alkylaluminums to form homogenous catalysts. One of these systems was used for reaction injection molding (RIM) polymerization of endo-dicyclopentadiene producing a polymer with unprecedented strength and stiffness.6 Heterogeneous catalysts were also employed. Immobilized molybdenum based catalysts are at the core of the Shell Higher Olefin Process (SHOP)7 and the “Tri Olefin Process”.8 The latter being an example on how the equilibrating nature of metathesis can be exploited. Depending on market demands and available feedstock the process can operate to produce ethylene and 2-butene from propylene, as initially developed, or, in reverse, to synthesize propylene from other olefins.
Even though the metathesis reaction has been employed since its discovery and industrial applications quickly arose, the mechanism of metathesis took much longer to be understood. The mediation of several novel organometallic intermediates was proposed during the late 1960s and early 1970s by renowned investigators (Figure 1.1). However these proposals did not offer explanation for all of the observations encountered to that date. Most importantly they could not predict the actual thermodynamic ratios at equilibrium of the products formed in cross metathesis reactions.
Figure 1.1 Intermediates proposed for the olefin metathesis mechanism. (a) Calderon (1968),9 (b) Pettit (1971)10 and (c) Grubbs (1972).11
In 1971 Yves Chauvin together with his student Pan Hérisson proposed a mechanism which could account for the observed disproportionation, polymerization and telomerization (chain transfer) of olefins that were so far encountered12 (Scheme 1.2). Their proposal consisted of the coordination of an olefin to a metal-carbene complex, from which a rearrangement would form a metallacyclobutane intermediate. A further 2
1. Introduction rearrangement was proposed to form a new olefin and place a new carbene moiety on the metal center. As seen on the left-hand side of scheme 1.2, depending on the orientation of the olefin and the substituents on the metal-carbene, the metallacyclobutane intermediate can lead to products that are identical to the starting materials, this reaction, now known as degenerate metathesis, was first accounted for by Chauvin.
Scheme 1.2 Representation of Chauvin’s mechanism for the olefin metathesis.
Multiple experimental evidence fitting Chauvin’s mechanism quickly followed. For instance, Casey and Burkhardt13 reported the generation of diphenylethylene when a tungsten diphenylmethylidene complex was reacted with isobutylene. Later Grubbs and coworkers14 distinguished between the mechanisms proposed by Calderon, Grubbs and Chauvin based on the expected ratios of deuterium found in ethylene produced by reaction of a mixture of 1,7-octadiene and 1,7-octaniene-1,1,8,8-d4 with three different catalysts based on tungsten or molybdenum. Their observations clearly favored Chauvin’s mechanism even though the true active species could not be identified at that time.
The first advance towards a discrete, well-characterized system for olefin metathesis began in 1978 with Fred Tebbe. His studies on the structure and reactivity of a titanium- methylene-aluminum complex,15 now known as Tebbe’s reagent, demonstrated for the first time the ability of metal alkylidenes to perform degenerate olefin metathesis 3
Discovery, Development and Study of Carbenoid Mediated Reactions presumably via metallacyclobutane intermediates16 (Scheme 1.3, a). In this study an analogous four–membered ring, a metallacyclobutene, was isolated by reaction of Tebbe’s reagent with diphenylacetylene (Scheme 1.3, b). Shortly afterwards Grubbs reported the isolation of a true metallacyclobutane intermediate when Tebbe’s reagent reacted with neopentene in presence of a pyridine base17 (Scheme 1.3, c). Addition of
(CH3)2AlCl to this complex regenerated Tebbe’s reagent and the free olefin. On the other hand, when exposed to a different olefin the products of metathesis could be identified; confirming that this metallacyclobutane was as indeed an intermediate in the metathesis of olefins.
Scheme 1.3 (a) Degenerate metathesis catalyzed by Tebbe’s reagent. (b) Isolation of a metallacyclobutene. (c) Grubbs’ isolation of a metallacyclobutane complex.
1.1.2 Schrock’s high oxidaton state early transition metal alkylidene complexes.
Before Wilkinson’s report18 on the synthesis of stable metal-alkyl complexes, it was believed that such species could not be prepared and isolated due to the intrinsic instability of the metal-carbon bond. Nonetheless, Wilkinson demonstrated that the absence of -hydrogens in the alkyl chains enhanced the stability of those complexes driving the research efforts towards the preparation of new organometallic systems. Inspired by the work of Schmidbaur19 on the synthesis of pentaalkyl phosphorus and arsenic derivatives, Schrock started his studies on the preparation of pentaalkyl tantalum complexes.20 Methyl and benzyl derivatives were stable and isolable as predicted, however attempts to prepare penta(neopentyl) derivatives led to the isolation of the first non–heteroatom stabilized neutral alkylidene species (Scheme 1.4). The mechanism by which such complex was formed was later recognized as -hydrogen activation or sigma bond metathesis. At first, a strong interaction of the electron pair from the C-H bond and
4
1. Introduction the metal center, i.e. an agostic interaction,21 is favored by the increased steric crowding provided by the bulky alkyl groups. Ultimately, a nucleophilic carbon center, from one of the neopentyl ligands, performs an intramolecular deprotonation yielding neopentane and the observed alkylidene complex. The recognition of the generality of this reaction produced a family of similarly crowded, electron deficient, alkylidene complexes of tantalum and niobium.22
Scheme 1.4 Schrock’s synthesis of a tantalum alkylidene complex.
These new metal alkylidenes exhibited an electron poor character at the metallic center together with a nucleophilic carbene carbon resembling the reactivity of the well–known phosphorus ylide. These complexes rarely yielded metathesis products, in contrast, when reacted with olefins the resulting metallacyclobutane intermediates underwent rearrangement into tantalocyclopentanes as the result of the formation of a tantalum(III) complex which in turn reacted with two olefin molecules.23
The replacement of chlorides with alkoxy ligands in these complexes “turned on” metathesis activity.24 Judicious exploration led Schrock to develop his current framework of molybdenum, tantalum and tungsten complexes characterized by a bulky imido ligand to prevent bimolecular decomposition together with electron deficient alkoxy anions to direct the reactivity towards metathesis25 (Figure 1.2).
Modular optimization of all the substituents in this framework has provided tailor–made catalysts for all types of metathesis reactions. In figure 1.2, c, a recently developed catalyst for the enantioselective tandem ring–opening ring–closing metathesis is shown.26 The resulting family of catalysts is characterized by its very high activity, and as a result it has found its way into commercialization in recent times. However, the high oxophilicity of early transition metals reduces the tolerance of the related catalysts to molecules containing polar groups and limits the use of polar solvents as reaction media. The desire to expand the scope of olefin metathesis in a different set of conditions eventually led to the development of Grubb’s ruthenium based catalysts.
5
Discovery, Development and Study of Carbenoid Mediated Reactions
Figure 1.2 (a) Schrock’s framework for metathesis active alkylidene catalysts. (b) Early example. (c) Catalyst optimized for use in enantioselective ring closing metathesis.
1.1.3 Grubbs Ruthenium based catalysts.
In his personal account,27 SonBihn Nguyen relates one of the most frustrating aspects of the early days of research using Schrock’s metathesis catalysts: their synthesis. The required skill for their preparation and the sensitivity of the resulting complexes was a major driving force in the search for alternative, more robust catalytic systems. Having
28 noticed Natta’s report on the polymerization of cyclobutene by RuCl3 in protic media, Grubbs engaged in the search for ruthenium based molecular catalysts for olefin metathesis. Soon Grubbs reported the ring opening metathesis polymerization (ROMP) of norbornene derivatives in aqueous media under atmospheric conditions,29 an unprecedented feat that demonstrated not only the stability, but also the reactivity of the assumed organometallic intermediates in such conditions.
The first molecularly well–defined alkylidene ruthenium–based metathesis catalyst was reported by Grubbs in 1992 (Figure 1.3, a). This complex was synthesized by reaction of
30 (PPh3)3RuCl2 and diphenylcyclopropene. It showed a remarkable stability towards protic solvents and polar functional groups, however its activity was limited to the ROMP of highly strained olefins. Replacement of PPh3 ligands for the more bulky and more electron donating tricyclohexyl phosphine (PCy3) in the complex increased the reactivity dramatically and afforded catalysts capable of performing more demanding reactions like ROMP of low-strained olefins, cross metathesis and ring closing metathesis of functionalized dienes (Figure 1.3, b).31 Even though the applicability of these catalysts was quickly recognized, their synthesis, although much more convenient than for Schrock’s complexes, was not fully scalable. The focus on the use of more efficient alkylidene sources led to the synthesis of a benzylidene derivative of Grubbs’ complexes by reaction of the same ruthenium precursor with phenyldiazomethane at low temperature.32 This
6
1. Introduction complex, now known as the first generation Grubbs catalysts (Figure 1.3, c), could be obtained in kilogram scale, making its commercialization feasible. The availability of such robust catalysts triggered the application of olefin metathesis in all aspects of organic chemistry.
N N
P Ph P Ph P Cl Cl Cl Cl Ru Ph Ru Ph Ru Ph Ru Cl Cl Cl Cl Ph P P P P
(a) (b) (c) (d)
Figure 1.3 (a) First well-defined ruthenium metathesis catalyst, (b) variation with PCy3, (c) first generation and (d) second generation Grubbs catalysts.
Mechanistic studies by Grubbs33 and others34 suggested that the active species during olefin metathesis by ruthenium alkylidene complexes was a 14e- complex formed by dissociation of one phosphine ligand. Attempts to increase the effective concentration of this intermediate resulted in accelerated catalyst decomposition. The replacement of one phosphine ligand with an N-heterocyclic carbene35 (NHC) afforded catalysts showing dramatically increased activity and stability (Figure 1.3, d). At the time, it was argued that the NHC ligands, thought to be excellent sigma donor ligands with limited accepting capabilities, would promote the dissociation of the phosphine trans to the NHC ligand, while the ligand’s large steric demand would prevent bimolecular decomposition by forcing the catalyst molecules to stay apart from one another. Mechanistic studies together with theoretical calculations later showed that this interpretation was wrong (see section 2.1.1)
Based on the established ruthenium framework many different variants have been developed. Hoveyda introduced in 1999 a catalyst with a chelating isopropoxy group attached to the benzylidene moiety36 (Figure 1.4, a), this substitution improved the stability of the first generation systems. Later, in 2000 the same concept was applied to the second generation systems37 (Figure 1.4, b). In 2002, Grela reported the inclusion of an electron withdrawing group in the para position with respect to the ether substituent38 (Figure 1.4, c), in this way the oxygen-ruthenium bond was destabilized facilitating the activation of the catalyst. In 2001, Grubbs replaced the remaining
7
Discovery, Development and Study of Carbenoid Mediated Reactions phosphine ligand with pyridines (Figure 1.4, d) which are more easily displaced by an olefin.39 This complex was found to initiate more than a million fold faster than the regular second generation Ru-catalyst. Given this leap in reactivity the catalyst received the name of third generation Grubbs catalyst.
Figure 1.4. (a) First generation and (b) second generation Hoveyda-Grubbs catalysts, (c) Grela’s complex and (d) Third generation Grubbs catalyst.
1.1.4 Heterogeneous systems
Convenient catalyst recovery and recyclability, and the possibility of incorporating static catalysts in flow reactors are just some of the benefits of using heterogeneous catalysts for organic transformations in the laboratory or in industrial processes.40 Ever since the discovery of olefin metathesis heterogeneous systems have played a crucial role. The
3-4 initial reports by Eleuterio used MoO3 or WO3 supported on silica or alumina. Important advances were made by British Petroleum with their Re3O7/Al2O3 system which can perform olefin metathesis at room temperature.41 Even though these kinds of systems have facilitated important industrial applications their actual molecular composition remains ill-defined. Contrary to the case of the well-defined molecular catalysts, rational design of heterogeneous systems is limited by the knowledge of the surface active sites. One approach that attempts to combine the advantages of heterogeneous systems with the abilities of the homogeneous catalysts consists of the immobilization of molecular catalysts on well-defined surfaces by covalent bonding with reactive sites. This methodology is known as surface organometallic chemistry (SOMC). Parallel with the development of the early transition metal molecular catalysts, the grafting of these
8
1. Introduction complexes on silica systems provided the earliest examples of supported, highly active heterogeneous systems.42 However, the limited tools available at the time restricted the understanding at the molecular level of catalysts produced. More recently, the incorporation of more capable techniques has allowed the complete characterization of silica-supported metal alkylidene complexes.43 The intrinsic modularity of the well- defined supported catalysts has allowed their continuous improvement, in a fashion reminiscent to their homogeneous counterparts44 (Figure 1.5).
Figure 1.5 (a) First, (b) second and (c) third generation of well-defined silica supported alkylidene complexes.
A wide variety of anchoring groups have been used to attach ruthenium based catalyst to solid supports. Covalent bonding of the phosphine or NHC ligand in first or second generation systems, attachment via anionic group or the incorporation of the benzylidene into the solid phase have been used to immobilize active catalysts in polymeric supports.45 More recently, using SOMC principles, second generation Grubbs systems have been incorporated into hybrid organic-silica materials.46
1.1.5 Applications of olefin metathesis
Large scale industrial applications of olefin metathesis are mainly focused on the production of small to medium sized commodity olefins. The unmatched global demand for propene has driven the reverse-operation Phillips “Tri Olefin Process”8 to convert a mixture of ethene and 2-butene into the more valuable propene. The process uses a mixture of WO3/SiO2 and MgO at high temperature, the latter being an olefin isomerization catalyst. As an alternative, the Meta-4 process was developed using
47 Re2O7/Al2O3 for the same reaction in a liquid phase at 35 °C and 60 bar. Medium sized linear internal olefins (C11-C14) are produced by olefin metathesis in the third step of the Shell higher olefin process. These olefins are further transformed into detergent alcohols or alkylates.7
9
Discovery, Development and Study of Carbenoid Mediated Reactions On a slightly smaller scale, industrial production of polymers via ring opening metathesis polymerization (ROMP) has been focus of intense research since the early days. The ready availability of the required monomers from petrochemical refining helped the commercial success of ROMP. On the contrary, acyclic diene metathesis (ADMET) has so far remained as an academic curiosity (Scheme 1.5).
Scheme 1.5 Top. Ring opening metathesis polymerization (ROMP). Bottom: Acyclic diene metathesis.
Polymers of cyclopentene (polypentenomer), cyclooctene (Vestamer®), norbornene (Nosorex®) and dicyclopentadiene (Metton®, Telene® and PentamTM) quickly became available. Although these polymers can be used by themselves or in combination with other polymers or via functionalization of their olefins, as for example by vulcanization into rubber, the high price associated with these processes sets these polymers apart in the niche of specialty materials. In that respect, olefin metathesis offers unusual possibilities. For instance, it enables the incorporation of end-groups after polymerization by use of chain transfer reagents producing telechelic polymers for specialized applications.48 Self-healing polymers in which a vesicle provides monomer after a crack reaches a critical point, take advantage of the living nature of ROMP to regenerate the structure.49 Furthermore, biocompatible polymers made by ROMP are strong candidates for corneal tissue engineering.50
In fine chemistry, the most widely used application of olefin metathesis is ring closing metathesis (RCM).51 The possibility to create rings ranging from 5 to 20 or more carbon atoms with evolving ethylene as driving force for product formation has made this technique a crucial part of many total syntheses of biologically relevant molecules.52 However, the synthesis of increasingly bigger cycles must be performed at higher dilutions, catalysts loadings and temperatures in order to avoid dimerization and to compensate for catalyst decomposition. The combination of RCM with other types of metathetic transformations like ROM or CM in one pot are elegant approaches for the construction of complex molecules from relatively simple precursors.
10
1. Introduction 1.1.6 Stereoselective olefin metathesis
The complexity generated by contemporary synthetic approaches gives rise to products which may contain pro-chiral groups. Enantiopure variants of both early transition metal and ruthenium based catalysts have been made in order to prepare enantiomerically enriched products. The rigidity of the framework offered by early transition metal catalysts was exploited by replacing the alkoxy ligands with chelating chiral bisalcoholates (see for example see figure 1.2, c). These complexes were first shown to perform very well for the catalytic kinetic resolution of dienes via ring closing metathesis.53 Development of more user-friendly chiral systems formed in situ by combination of a commercially available optically pure binaphthol ligands and a suitable metal alkylidene precursor has been reported.54 Ruthenium catalysts bearing bidentate NHC ligands have also been reported to perform asymmetric ROM/CM reactions55 (Figure 1.6, a). These catalysts benefit from the inherent robustness of the ruthenium alkylidenes allowing them to operate under normal atmospheric conditions.
Figure 1.6 Examples of (a) enantioselective and (b) and (c) Z-selective metathesis catalysts.
Although very important, enantioselectivity is not the only isomerism that can be influenced during the process of olefin metathesis. The resulting olefin is often a mixture of the Z and E geometric isomers. Formation of the thermodynamically more stable E isomer is favored in most of the cases. Nonetheless, many important synthetic targets require the formation of the more challenging Z isomer. The first advances in this area were achieved by tuning the steric demand on monoaryloxy W and Mo alkylidene complexes, these complexes showed high selectivity towards Z-olefin formation56 (Figure 1.6, b). In a similar way, steric arguments guided the development of ruthenium catalysts bearing bidentate phosphine ligands which displayed a high degree of chemoselectivity for the alternating ROMP of cycloolefins; these catalysts were also optimized to induce some degree of Z-selectivity.57 Catalysts with chelating NHC ligands
11
Discovery, Development and Study of Carbenoid Mediated Reactions have also been shown to influence the stereochemical outcome of different olefin metathesis processes58 (Figure 6, c). Replacement of one59 or two60 of the chlorides by sulfur ligands in traditional second generation Grubbs systems has also yielded highly Z- selective catalysts (Figure 1.7).
Figure 1.7 Most recent examples of Z-selective catalysts based on sulfur-containing ligands.
1.3. Methylenation Reactions
1.3.1. Three–membered rings
Among the cyclic molecular scaffolds, three–membered rings are of particular importance due to their reactivity and abundance in molecules with important biological activities. The smallest cyclic hydrocarbon, cyclopropane (figure 1.8, a), and its derivatives are found in many relevant molecules, both naturally occurring and synthetically prepared.61
Figure 1.8 Left: Simplest examples of chemically relevant three membered rings: (a) cyclopropane, (b) oxirane, (c) thiirane and (d) aziridine.
One of the first compounds derived from cyclopropane that have found their way into industrial production are the members of the pyrethroid family. In 1924, Staudinger and Ružička isolated and characterized esters of (+)-trans-chrysanthemic acid as the active component in the defense mechanism against insects of certain pyrethrum flowers62 (figure 1.8).
12
1. Introduction
Figure 1.9 Structures of naturally occurring (a) (+)-trans-chrysanthemic acid and synthethic (b) allethrin and (c) permethrin.
It was quickly recognized that the synthetic derivatives of chrysanthemic acid were more potent, less toxic and more resistant to UV radiation than the natural variants,63 turning this type of compound into the most common household insecticide to date,64 after DDT was banned in 1972 in the USA by the Environmental Protection Agency (EPA).65 It comes as no surprise that the synthesis of cyclopropanes,66 and their subsequent use as intermediates,67 has become an established and intensively active area of research.
Three–membered rings containing one oxygen atom, i.e. epoxides, (figure 1.8, b), are important molecules that are not only found in nature,68 but are also regularly used as intermediates in the synthesis of more elaborate functionalities. Their rich chemistry,69 particularly the versatile ring opening,70 makes epoxides a desirable target or intermediate in many synthetic routes.71
Figure 1.10 Naturally occurring (a) moth sex pheromone (–)-Posticlure72 from Orgyia postica and (b) antifungal ambuic acid73 from Pestalotiopsis microspora.
The smallest nitrogen–containing heterocycles, the aziridines, also constitute an important class of compounds. Many of the discovered naturally occurring aziridines show antitumor activity, making them important targets for total synthesis.74 The cleavage of the aziridine ring can afford amines, amino acids, amino alcohols and other relevant nitrogen–containing molecules.75 The stereoselective synthesis of aziridines has been recently reviewed.76
13
Discovery, Development and Study of Carbenoid Mediated Reactions
Figure 1.11 Natural aziridines (a) (2S,3S)-(+)-aziridine-dicarboxylic acid,77 an antibacterial isolated from Streptomyces MD398-A1 and (b) Mitomycin K,78 an antibacterial and anticancer agent.
1.3.2. Bonding in three–membered rings
While on the basis of geometric arguments one would expect the cyclopropane ring to be one of the most unstable molecules, its experimentally measured ring strain is just slightly larger than that of cyclobutane; 27.5 kcal/mol vs 26.5 kcal/mol respectively.79 Additionally, the chemical reactivity of cyclopropane resembles more closely that of ethylene than other cycloalkanes. In this respect both ethylene and cyclopropanes can: interact with neighboring -electron systems (resonance); form metal complexes; add strong acids, halogens and ozone; and engage in cycloadditions. Although ethylene normally reacts much faster, there are no fundamental differences in the reaction types these two functional groups can perform. Much work has been devoted to the study of the bonding situation in the cyclopropane subunit.80 Two independent, albeit mathematically equivalent, bonding models were developed in order to explain the particularities of the cyclopropane ring. In 1949, A. D. Walsh derived the molecular orbitals (MO) for cyclopropane by combining three independent methylene units, each with a sp2 hybrid orbital pointing towards the center of the ring and an independent p orbital differing only in their symmetry properties81 (Figure 1.12 top). This description is very similar to the common MO picture of a C=C double bond in which the interaction of two sp2 atomic orbitals (AO) generate a MO and the two perpendicular p AOs form the bonding MO (Figure 1.12 bottom).
The second description, based on the valence bond model, predicts the C–C bonds to result from the overlap of two sp5 orbitals at each C atom.82 The direction of these orbitals does not correspond with the direction of bonding (line connecting two carbon atoms) and thus gives the impression of bent bonds (Figure 1.13 top). It has been experimentally confirmed that electron density is the greatest off the C–C axis.83 The same construction can be applied to ethylene giving rise to two “banana” bonds on each side of the C–C axis, and because of this it has sometimes been called the smallest cyclic hydrocarbon.84
14
1. Introduction
Figure 1.12 Top: Walsh MO description of cyclopropane. Bottom: common MO description of ethylene.
As a consequence of the increase in double bond character (p character) of the C–C bonds in cyclopropane, the C–H bonds are enriched in s character. The C–H bonds in cyclopropane are therefore stronger and shorter than in other alkanes; i.e. the bond strengths and distances are somewhere between those of the C–H bond of olefins and saturated hydrocarbons.
Figure 1.13 Valence bond model of cyclopropane (top) and ethylene (bottom).
The bonding description in the other three–membered rings; oxirane, thiirane and aziridines; can be extrapolated from the one in cyclopropane by replacing one methylidene unit with an oxygen, sulfur or nitrogen atom respectively.
1.3.3. Carbene equivalents or carbenoids for the synthesis of three–membered rings
As shown in scheme 1.6 left, one general approach to the synthesis of three–membered rings, especially of cyclopropanes, is the direct addition of a methylene carbene to an unsaturated substrate. For this purpose, the most efficient way to generate free carbenes in solution is the thermal85 or photolytic86 decomposition of diazomethane (scheme 1.6, right).
Scheme 1.6 Left: Prototypical methylene addition to unsaturated compounds. Right: Generation of methylene from diazomethane.
15
Discovery, Development and Study of Carbenoid Mediated Reactions Although this may seem like an optimal reaction, the handling of diazomethane is complicated by severe health and safety risks. Due to its high nitrogen content, and the production of thermodynamically stable molecular nitrogen upon decomposition, diazomethane is intrinsically explosive. Solutions of diazomethane have been reported to explode upon distillation, exposure to sharp edges, drying over sharp KOH pellets, and exposure to light.87 Additionally, diazomethane and the precursors from which it is prepared, are highly toxic.88 In fact, these compounds have been found to be potent carcinogenic agents.89 Despite these drawbacks, when necessary, diazomethane can be used safely in the laboratory with the appropriate equipment (diazald kit). In situ generation of diazomethane under phase–transfer conditions offers a method in which the concentration of diazomethane never builds up to dangerous levels.90
In order to avoid the use of diazomethane and its precursors, reagents known as carbenoids are more commonly employed. A carbenoid, as defined by Closs and Moss, is a species which shows reactivity ‘Qualitatively analogous to those of carbenes without necessarily being a free divalent carbon species’.91 A wide range of reagents can be covered by this term, see figure 1.14. Although operating by markedly different mechanisms, all the carbenoids shown can be used for the synthesis of one or more of the three–membered rings discussed before (figure 1.8).
Figure 1.14 Some common methylene donors for the synthesis of three membered cycles: (a) iodomethyllithium,92 (b) iodomethyzinc iodide,93 (c) bis(iodomethyl)mercury,94 (d) dimethylsulfoxonium methylide95 and (e) transition metal carbenoid compounds.
Carbenoids can be classified in two main groups on the basis of their dominant electronic character. Nucleophilic carbenoids like (a) and (d) predominantly add to unsaturated compounds, yielding intermediates from which intramolecular ring closure affords a cyclic product. On the other hand, electrophilic carbenoids like (b), (c) and (e) are normally subject to attack by -systems, forming the cyclic products in one concerted step. The mechanism of reaction for the most important representatives of these two families will be discussed in the following sections.
16
1. Introduction 1.1.7 Nucleophilic methylenation
All nucleophilic methylenation reactions follow the same general mechanism depicted in scheme 1.7. The carbenoids employed in this type of reaction are derived either from deprotonation or lithium–halogen exchange of a suitable starting material like dihalomethanes,92 trimethylsulfonium halides,95 ammonium salts96 or enolizable carbonyl compounds.97 In addition to the nucleophilic carbon of the carbenoid, a leaving group needs to be present to enable the intramolecular ring closure. The leaving group can be either incorporated into the carbenoid or be already present in the substrate. Due to the stepwise nature of this methylenation, this transformation is not stereospecific, nonetheless many stereoselective variants have been reported.
Scheme 1.7 General mechanism for the nucleophilic methylenation of unsaturated compounds. (M) denotes a metal which may or may not be necessary, depending on the nature of the carbenoid.
Sulfur ylides: Corey-Chaykovsky reagent
Following a 1961 report by Johnson,98 E. J. Corey and M. Chaykovsky developed in 1962 the use of dimethylsulfonium and dimethyl(oxo)sulfonium methylides for the synthesis of epoxides from aldehydes and ketones.95, 99 Over the years their methodology has become a routine protocol in organic chemistry.100
Scheme 1.8 Corey-Chaykovsky reaction.
Although mainly used for the synthesis of epoxides, sulfur ylides can also be applied in the synthesis of other three–membered rings. The additions of sulfur ylides to benzylaniline101 and to chiral sulfinimines102 to afford the corresponding aziridines, have been reported (Scheme 1.9). Despite these promising reports addition of ylides to imines remains underdeveloped. Unlike carbonyl compunds (for epoxidation) and ,- unsaturated compounds (for cyclopropanation), N-alkyl and N-aryl imines are rather unreactive towards nucleophilic attack by ylide agents.103
17
Discovery, Development and Study of Carbenoid Mediated Reactions
Scheme 1.9 Sulfur ylide mediated methylenation of Left: benzylaniline101 and Right: chiral sulfimines.102
Sulfur ylides readily add to ,–unsaturated carbonyls affording cyclopropanes in excellent yields.104 This methylenation reaction belongs to the more general class known as Michael–initiated ring closures (MIRC). This reaction is not limited to conjugated carbonyls; in principle, any olefin activated by an adjacent electron withdrawing group can be attacked by an ylide. Addition to ,–unsaturated esters, amides, imides, nitriles, conjugated nitroolefins and vinylsulfones has been reported.104 On longer conjugated systems 1,6 addition is preferred over 1,4 addition.
Scheme 1.10 Some examples of Michael-initiated methylenation with sulfur ylides.104 Left: Cyclopropanation of an ,,,-unsaturated conjugated ketone. Right: cyclopropanation of a conjugated nitrile.
Halomethyllithium carbenoids
Since their introduction as nucleophiles,105 halomethyllithium carbenoids have been used as an alternative to diazomethane and sulfur ylides in the homologation of aldehydes and the synthesis of epoxides and aziridines.92 These carbenoids are simply prepared by addition of an alkyllithium to a dihalomethane at very low temperatures (scheme 1.11, left). Although initially believed to be unstable above -110 °C, more recent studies point to a less prohibitive temperature restriction of -78°C.106 More practical procedures involving the in situ preparation of the carbenoid in the presence of the electrophile, allow for even higher operating temperatures of up to 0 °C.
18
1. Introduction
Scheme 1.11 Left: Preparation of halomethyllithium carbenoids. Right: Use of these carbenoids for (a) Synthesis of epoxides107 and (b) aziridines.108 1.1.8 Electrophilic methylenation
Electrophilic carbenoids are organometallic species in which the carbon atom has a net electropositive character, and therefore is susceptible to attack by a nucleophile: in the case of cyclopropanation reactions, an olefin. The mode of addition of these reagents is a concerted, asynchronous process in which the key transition step possesses a butterfly geometry (scheme 1.12).
Scheme 1.12 General electrophilic methylenation of olefins.
Halomethylzinc reagents: the Simmons-Smith reaction
Although synthesized 30 years earlier by G. Emschwiller,109 it was the discovery by Howard Ensign Simmons, Jr. and Ronald D. Smith at the Du Pont Central Research laboratories that iodomethylzinc iodide reacted with olefins producing cyclopropanes stereospecifically93 that marked the development of the currently most popular method for the synthesis of this class of compounds, particularly for large scale applications.
The original Simmons and Smith procedure involved the activation of Zn with Cu at high temperature in a hydrogen atmosphere. Since then, more convenient variants have emerged. The most commonly used activation procedures involve the treatment of Zn
110 111 112 powder with CuSO4, Cu(OAc)2 or CuCl. In all of these cases an ethereal solvent is required (Et2O, THF or DME). Nonetheless, solvent coordination reduces the electropositive character of the carbenoid impairing its reactivity. A major breakthrough was achieved by Furukawa who replaced metallic Zn by soluble diethylzinc.113 Iodomethane replaces one of the alkyl groups and generates a reactive species which
19
Discovery, Development and Study of Carbenoid Mediated Reactions performs cyclopropanation efficiently in a wide variety of non-coordinating solvents, e.g. hexane, dichloromethane, toluene and benzene. Replacement of one alkyl ligand in diethylzinc by an anionic oxygen ligand like carboxylate,114 trifluorosulfonate,115 phenolate116 or phosphate117 has been reported to generate very effective cyclopropanating reagents by withdrawing electron density from the metallic center, and consequently increasing the electrophilic character of the carbenoid. These zinc organometallics have been employed in countless synthetic procedures in organic chemistry.118
Scheme 1.13 Top: original Simmons-Smith cyclopropanating reagent. Variants by Middle: Furuwaka and Bottom: Charette.
Transition metal catalyzed decomposition of diazomethane
The inherent safety risks associated with the bulk thermal or photolytic decomposition of diazomethane to generate free methylene has encouraged the development of alternative and safer ways to transfer the methylene moiety from diazomethane to olefins. Transition metal catalyzed decomposition of diazomethane allows the use of lower concentrations of this hazardous chemical. Almost all metals have been reported to react with diazomethane yielding a variety of products including cyclopropanes when olefins are present.
Depending on the metal used, two main mechanisms have been proposed for the methylene transfer to olefins.66b, 119 A more traditional outer sphere mechanism,120 in which the olefin does not coordinate the metal center, in resemblance to the Simmons- Smith reaction, is proposed for reactions mediated by Fe, Ru, Rh, Cu and Co (Scheme1.12 left). In this mechanism, an initial attack by diazomethane on the metal complex produces a metal methylidene intermediate after nitrogen extrusion. This species can then engage in a [2+1] concerted addition with the olefin producing cyclopropane and regenerating the metal catalysts.
20
1. Introduction
Scheme 1.12 Proposed (left) outer and (right) inner sphere mechanism for the methylene transfer in transition metal catalyzed decomposition of diazomethane.
Alternatively, a stepwise inner sphere mechanism, in which the olefin coordinates the metal center, is proposed mainly for Pd catalysts.121 In this mechanism, an unsaturated metal complex is initially coordinated by the substrate. Subsequent attack by diazomethane generates a carbene complex which then can undergo [2+2] cycloaddition to form a metallacyclobutane intermediate. From this intermediate cyclopropane is formed by reductive elimination with concomitant regeneration the metal catalyst.
21
22
2. Ruthenium catalyzed alternating ring opening metathesis polymerization (A-ROMP)
2.1. Introduction
2.1.1. Studies on the mechanism of olefin metathesis
Since the mechanistic proposal by Hérrison and Chauvin for the olefin metathesis12 considerable research effort has been devoted to experimentally support this mechanism. As discussed in the introduction, it quickly became clear that metal– alkylidene complexes were the active catalysts.13, 15, 122 The intermediacy of metallacyclobutanes was also confirmed as a result of the isolation and characterization of several examples which mediated olefin metathesis to varying degrees.16-17, 123 Even though these initial findings provided strong support for Chauvin’s mechanism, the fine details of every step were the source of debate over the following years. In the case of ruthenium catalysts, Adlhart and Chen have studied all the possibilities within a set of constrains reasonable for a transition metal catalyzed metathesis reaction124 (scheme 2.1).
Scheme 2.1 Possible pathways for the ruthenium olefin metathesis. Taken from (Adlhart and Chen, 2004)124.
Scheme 2.1 summarizes all the reasonable pathways for ruthenium–catalyzed olefin metathesis. Starting from a well–defined 16e- complex (A), i.e. the first generation Grubbs
23
Discovery, Development and Study of Carbenoid Mediated Reactions catalyst31 where the benzylidene was simplified by a methylidene, olefin coordination can take place directly either cis or trans with respect to the methylidene to form intermediate (Ba)cis or (Ba)trans respectively. In a dissociative pathway, initial phosphine loss leads to 14e- intermediate (B), which can then be coordinated by the olefin to form one of the intermediates (C), (C1)cis or (C2)cis depending whether one of the chloride anions is in the trans or in one of the two possible cis configurations. Direct formation of a metallacyclobutane from this intermediates leads to structures (D), (D1)cis and (D2)cis respectively. Structure (Da)cis can be formed directly from (Ba)cis, and (Da)trans can be formed after phosphine recoordination to intermediate (C). Calculations from Adlhart and Chen predicted the minimum energy pathway to go through the structures (A), (B), (C) and (D). Over the years, different groups have provided experimental evidence also favoring this particular pathway.
Grubbs offered convincing evidence for an initial dissociative mechanism in his phosphine exchange studies.125 He showed that olefin coordination is preceded by phosphine dissociation to form a reactive 14e- intermediate. More importantly, it was recognized for the first time that second generation systems (where one phosphine has been replaced by an N-heterocyclic carbene) showed a lower rate of initiation, in contradiction to the common belief. The increased reactivity seemed to arise from their intrinsic reactivity after the initiation event. Isolable 14e- complex reported by Piers126 (Figure 2.1) showed almost no activation barrier as would be expected since there is no ligand loss needed to engage in productive metathesis steps.
Figure 2.1 Piers cationic fast-activating 14e- metathesis catalyst.126
Mechanistic studies in the gas phase by Adlhart and Chen aimed at elucidating the nature of the rate determining step for olefin metathesis.34c In this study a positively charged phosphine ligand allows for the generation of 14e- species 1 during electro-spray ionization in a modified TSQ 700 ESI-MS instrument (Scheme 2.2). Exposure of this species to different olefins in the gas phase produced complex 2 which was then mass selected and reacted with either 1-butene or norbornene in a subsequent reaction chamber to produce complex 3 or 4 respectively.
24
2. Ruthenium catalyzed A–ROMP
Scheme 2.2 Mechanistic studies in the gas phase by Adlhart and Chen.34c
Adlhart and Chen found an accelerating effect on the reaction of 2 with 1-butene when an electron-withdrawing substituent was added to the parent benzylidene. They also found evidence of an inverse isotope effect for the forward and the backward reaction (reaction of 3 with a styrene) of this nearly thermoneutral process. On the contrary, no evidence was found of a substituent or isotopic effect for the exothermic reaction of 2 with norbornene. At that moment these findings led to contradictory conclusions. On a theoretical potential energy surface for the reaction of 2 with 1-butene or norbornene (Scheme 2.3) two assumptions can be made. If the metallacyclobutane (MCB) structure is considered an intermediate (left) the energetic profile from olefin coordination (B to C and B' to C') to MCB formation (C to D and C' to D') are in essence the same for both 1- butene and norbornene since the ring-strain in norbornene is only released after cycloreversion. The exothermicity of the ring opening for norbornene stabilizes the product D' and at the same time lowers the energy of the corresponding cycloreversion transition state (D' to E') in accordance to the Hammond postulate (dashed line in the left potential energy surface). In this scenario both a substituent and isotopic effect should be measurable and of similar magnitude not only for the pseudodegenerate reaction with 1-butene but also for the norbornene case. On the other hand, if the MCB structure would be a transition state (Scheme 2.3, right), the exothermicity of the ring opening would lower the energy of transition state D' diminishing or eliminating the substituent and isotopic effects for the exothermic reaction. The experimental evidence favored the latter case, whereas computations predicted an intermediate for the MCB structure.
25
Discovery, Development and Study of Carbenoid Mediated Reactions
Scheme 2.3 Metallacyclobutane (MCB) as intermediate or transition state.
The explanation to this apparent conflict was put forth by Adlhart and Chen two years later. Using a QM/MM model it was possible to model the complete system including the bulky tricyclohexyl phosphine.127 It was discovered that the rate limiting step on degenerate metathesis in first generation systems (such as the ones studied by Adlhart and Chen) was a 60 ° phosphine rotation around its three-fold symmetry axis at the metallacyclobutane intermediate (Scheme 2.4). On the other hand, the stabilization of the intermediates after the cycloreversion step in exothermic reactions, i.e. reaction with strained olefins like norbornene, allows the ligand rotation to take place at lower energies at a later stage on the reaction coordinate. Such rotation is not necessary in second generation systems due to the two-fold symmetry of the NHC ligand. Phosphine rotation at intermediate D to yield D'' avoids the destabilizing steric interaction present in E but not in E''. More importantly, it was then recognized that the carbene moiety switches sides after each productive cycle. Although unimportant for traditional systems where the phosphine ligand is freely rotating, this switch is the cornerstone upon which the fully chemoselective alternating ROMP catalysts were designed (vide infra).
26
2. Ruthenium catalyzed A–ROMP
Scheme 2.4. Rate-limiting ligand rotation at metallacyclobutane structure.
Direct observation of metallacyclobutanes at low temperature in ruthenium-catalyzed metathesis was afterwards reported by Piers128 and Grubbs.129
As mentioned earlier, Grubbs' experiments in solution phase revealed that contrary to what was commonly accepted, second generation catalysts showed slower rates for initial phosphine dissociation. Therefore, their superior activity must come from an intrinsic higher reactivity of the 14e- intermediate. Evidence in the solution phase was obtained by Grubbs after reacting a second generation catalyst with ethyl vinyl ether; this class of substrates form an inactive alkylidene after one metathesis step.125b Grubbs found no rate dependence on the catalyst concentration over a wide range of concentrations, indicating a case of saturation kinetics. On the other hand, first generation systems were highly dependent on catalyst concentration. However, the intrinsic reactivity of these systems cannot be fully separated from the initial activation step in solution phase. Adlhart and Chen were able to study the metathesis reaction after the activation step by reacting tagged ruthenium catalysts in the gas phase with norbornene.130 Under the same conditions, the second generation catalyst added up to 4 monomer units, whereas the first generation catalyst reacted only once. Calculations gave a rationale to this behavior (figure 2.2).124 From this potential energy surface it can be concluded that the energy required to dissociate the phosphine ligand from the starting 16e- complex A' to form 14e- intermediate B' is higher in second generation (26.1 kcal mol-1) than in the first generation system (21.1 kcal mol-1). These results are in accordance with the solution-phase studies by Grubbs (27.0 and 23.6 kcal mol-1 for second and first generation systems respectively). More dramatic though is the difference in
27
Discovery, Development and Study of Carbenoid Mediated Reactions relative energy for the transition state connecting the -complex C' and the metallacyclobutane intermediate D' (energy of TSC'D'). For the second generation catalyst there is effectively no barrier, making the initial ligand dissociation rate determining. In contrast, first generation catalyst shows a barrier of 5.8 kcal mol-1 for the formation of the metallacyclobutane intermediate. In this case, TSC'D' is rate determining, taking into consideration that the phosphine dissociation event will most likely take place only once over the course of a catalytic reaction. The ratio between the barriers for propagation (C' to D') and pre-catalyst formation (C' to A') in a system is known as catalyst commitment. More committed catalysts show low barriers for productive steps and disfavor the formation of inactive forms. In this case the second generation Grubbs system shows a much higher commitment compared to first generation systems.
Figure 2.2 Calculated potential energy surfaces for the reaction of first (black) and second (grey) generation Grubbs catalysts with norbornene. Benzylidene moieties were simplified as methylidene, labels correspond to those in scheme 2.3 (Adlhart and Chen).124
The anionic ligands on the ruthenium center also play a crucial role in determining the relative activity of the catalyst. Straub calculated the influence of different anions on the active and inactive conformations of both the carbene moiety and the coordinated olefin.131 He found that in agreement with Grubbs’ experimentally observed trends,33b electron poor anions stabilize active conformations and thus lead to more active catalysts, for instance chloride and mesilate are better choices than electron rich alkoxide or thiolate anions. Straub found as well that strongly -donor ligands, e.g. the NHC carbene in second generation systems, should be positioned trans to the
28
2. Ruthenium catalyzed A–ROMP coordinated olefin in order to maximize the rate of metallacyclobutane intermediate formation and cycloreversion. A derivation of this concept was later found by Torker and Hoveyda132 to be of great importance for the high stereoselectivity shown by their dithiocatecholate–containing catalysts.60 In their case, strongly donating cis thiolate ligands force the metallacyclobutane to form cis to the NHC ligand, but trans to one of the thiolates, instead of the more normally found trans to NHC configuration. In this case the heterocyclic carbene enforces a big steric barrier that forces all substituents away from it, yielding highly Z-selective olefin metathesis.
2.1.2. Rationally designed A-ROMP catalysts
The mechanistic studies by Adlhart and Chen, discussed in the preceding sections, revealed that the rate-limiting transition state in a pseudo-degenerate metathesis reaction catalyzed by first generation Grubbs catalysts is the phosphine rotation in the metallacyclobutane intermediate. As a result, it was recognized that the carbene moiety flips between the two “sides” of the catalyst. Phosphine rotation in classical first generation systems ensures degeneracy between these two states. However, if rotation is inhibited, e.g. by chelation, and different substituents are placed on the phosphine, a catalyst stereogenic at both the phosphine and the ruthenium would be obtained. Inversion at ruthenium as a result from the swing of the carbene moiety after every metathesis step would switch between two different diastereoisomeric states, which by definition, are not degenerate. This concept was exploited by Bornand and Chen in the design of a catalyst capable of distinguishing two different olefins in order to produce alternating co-polymers.133
Bornand and Chen explained the observed chemoselectivity with the mechanism shown in scheme 2.4. In the first half of the cycle (A→B→C→D) the carbene moiety switches from a low energy diasteromeric state A, in which the interaction between the small phosphine substituent and the carbene is small, to a high energy state D, in which the big phosphine substituent exerts an increased steric interaction. The metathesis reaction only proceeds if the strain release provided by the monomer is enough to compensate for this energy increase. On the other hand, switching back from D to A is thermodynamically favored and can be performed by any olefin. When the copolymerization is performed using an excess of the low strain olefin an alternated polymerization sequence is favored. Although catalyst 5 (figure 2.3) showed the desired selectivity, the presence of one free tricyclohexyl phosphine promoted the formation of
29
Discovery, Development and Study of Carbenoid Mediated Reactions the starting first generation Grubbs system, which in turn lowered the selectivity of the copolymerization.
Scheme 2.4 Mechanism for the alternating ROMP of norbornene and cyclooctene catalyzed by asymmetric, rotation-locked first generation Grubbs catalysts (Bornand and Chen).133
2.1.3. Optimization of the catalysts for chemo and stereoselectivity
Subsequent optimization of the catalysts by Torker, Bornand and Chen allowed the preparation of a fully chemoselective, more robust system for the A-ROMP of norbornene and cyclooctene57b, 134 (figure 2.3). The use of a Hoveyda-type benzylidine avoided the formation of classic first generation systems and increased catalyst’s stability. Increasing the steric demand of one of the substituents on the phosphine produced more chemoselective catalysts, which in turn allowed the reaction to be performed at lower monomer ratios. For instance catalyst 6a produces a 72% content of alternating linkages with a norbornene to cyclooctene ratio of 1/200 whereas catalyst 6d achieves 97% of alternation using a monomer ratio of only 1/20.
30
2. Ruthenium catalyzed A–ROMP
Figure 2.3 Proof of concept catalyst 5. Optimization of the catalysts towards full alternation (6a to 6e) and chloride replacement with bulky sulfonates to induce Z selectivity (6f to 6j). TMP= 1,1,2,2-tetramethylpropyl.
Replacement of the chloride anion with bulky sulfonates increased the selectivity towards the formation of Z-arranged olefins without affecting the chemoselectivity.57a The effect on the stereoselectivity was explained on the basis of similar steric arguments, see scheme 2.5. At the metallacyclobutane intermediate the polymer chain and the olefin substituents can be arranged in opposite directions, leading to an E-olefin, or in the same direction to form a Z-olefin. The steric repulsion experienced in the metallacyclobutane structure favors the formation of E-olefins as evidenced by the 90% E-olefin content when using catalyst 6e. When bulky sulfonates are introduced, an additional, orthogonal steric interaction forces the arrangement of the polymer chain and olefin to form an all- cis metallacyclobutane intermediate, producing a Z-double bond. The balance between this two opposing interactions in the catalyst ultimately determines the extent of stereoselectivity of the polymerization.
31
Discovery, Development and Study of Carbenoid Mediated Reactions
Scheme 2.5 Steric interactions governing stereoselectivity on Ru–catalyzed metathesis.
Increasing the steric demand of the sulfonates increased the content of Z-arranged double bonds in the copolymer from 13% using 6f to 51% for 6i, when using in situ prepared catalysts. Further increase of the steric bulk of the sulfonate caused a dramatic reduction in the stereoselectivity. Catalyst 6j induces only 19% Z-selectivity under the same conditions. When isolated catalysts were tested in copolymerization it became evident that anion dissociation was playing a detrimental role. Z-selectivity of catalyst 6i went down to 36% (from 51%) when an analytically pure sample was used. The increased steric bulk of the sulfonates, while beneficial for the stereoselectivity, weakens the oxygen-ruthenium bond making the more demanding sulfonate anions also more prone to dissociation.
It was then predicted by Jovic ́ and Chen that the more strongly donating thiosulfonates would bind better to the ruthenium center and show increased effect in the stereoselectivity135 (figure 2.4).
32
2. Ruthenium catalyzed A–ROMP
Figure 2.4 Ruthenium catalysts for A-ROMP bearing thiosulfonate (left) and thiolate (right) ligands (Jovic ́ and Chen).135
The thiosulfonate ligands showed an increased stereoselectivity achieving up to 41% Z- olefin content using isolated catalyst 8f. However, together with the increase in the steric bulk of the thiosulfonate there was a significant loss in activity. While catalyst 8b produces a modest 18% of Z-arranged double bonds, and 82% copolymer, 8f only produces 4% copolymer. It was proposed that an excess of both steric demand and electronic donating capability of the thiosulfonate complexes renders the resulting complexes much less active than their sulfonate analogs. In an attempt to bring the steric influence closer to the metal center, thiolate ligands were used. The resulting catalysts however showed basically no control on the stereoselectivity and reduced activities. It is likely that the extremely crowded space around the ruthenium atom forces the ligand to dissociate before any metathesis can take place. Being very good electron donors, thiolates are strongly bound to the metal, disfavoring ligand dissociation.
Carboxylate ligands, as well as other oxyanions, were also employed in an attempt to steer the stereoselectivity towards Z-olefins.136 It was noted however that an unexpected drop in chemoselectivity was consistent among the tested catalysts. Nitro complex 12a being the worst, yielding only 77% alternating linkages in the copolymer compared with its chloride analog 6e at 97%. The exhibited low chemoselectivity could be explained after crystal structures of the complexes showed that the oxyanions effectively chelate the metal center. The additional coordination reduces the energetic difference between the two possible diasteromeric carbene states, therefore restricting the catalyst’s ability to distinguish between the two substrates. 33
Discovery, Development and Study of Carbenoid Mediated Reactions
Figure 2.5 Catalysts for A-ROMP bearing oxyanions136 (left). Representation of catalyst 12a showing nitrate chelation.
The planar geometry of the carboxylate anions places their steric demand away from the ruthenium center as shown by X–ray structures. Consequently, increasing the steric bulk of the carboxylate anion (catalysts 12a to 12f) led to no appreciable effect in the stereoselectivity of the catalysts, showing that the family of oxyanions is not a viable option to obtain active, yet more selective, catalysts.
It was clear after these attempts that a different approach would be needed in order to achieve a breakthrough in catalyst stereoselectivity. The use of thiosulfonates proved that more strongly bound anions do indeed increase the stereoselectivity of the system, bringing the steric barrier closer to the ruthenium center. However, their increased electron donating capability reduced the activity of the resulting catalysts. Thiolates on the other hand, appear to be far too close, inhibiting olefin coordination without prior ligand dissociation. There is however still another way to influence the bond strength of a ligand, and that is by modifying the donating properties of the ligand trans to it. The - donating capability of this ligand mainly affects the energy of the ground state, i.e. trans influence. On the other hand, the kinetic trans effect results from the –accepting character of the opposing ligand.137 If the catalyst framework could be modified in a way to include a weaker donor and more –basic anchoring group, it could be expected that the resulting catalysts would be both more active, according to the electronic effects of the ligands, and bind the sulfonate ligands more strongly, leading to more stereoselective A-ROMP.
34
2. Ruthenium catalyzed A–ROMP 2.2. Phosphinosulfonate system
Originally conceived of as supporting ligands for the palladium–catalyzed copolymerization of ethylene and polar compounds138 triarylphosphines with one ortho– sulfonate group139 were thought to form the active catalysts in situ by chelation when mixed with appropriate palladium sources. Later, the group of Stefan Mecking reported the first crystal structure of such a complex showing that the ligand indeed forms a palladacycle by binding with both the phosphorus and an oxygen atom to the metal center.140 The intense research on the palladium-catalyzed insertion polymerization of olefins has led to the development of several new polymers.141 Our desire to explore less electron donating groups as anchoring groups for our ruthenium catalysts framework prompted us to investigate the application of phosphine sulfonate ligands in the A- ROMP of olefins.
2.2.1. Synthesis of a ruthenium metathesis catalyst bearing an asymmetric phosphino sulfonate ligand
Scheme 2.6 Synthesis of an asymmetric phosphino sulfonate ligand, its sodium salt and the corresponding ruthenium Hoveyda-type benzylidene complex. The asymmetric phosphino sulfonate ligand 13a was easily synthesized by reaction of the appropriate chlorophosphine with ortholithiated benzenesulfonic acid in a slight modification of the procedure reported by Hearley et al.142 (scheme 2.6). The sodium salt 14 was prepared by direct deprotonation with sodium hydroxide.
Treatment of complex 1536 with sodium salt 14 in dichloromethane at 80 °C produces a dark green solution of intermediate 16 (scheme 2.6). The structure shown for this intermediate is proposed based on the characteristics found in the NMR spectra shown in figure 2.6 and the crystal structure of an analogous complex synthesized at a later
35
Discovery, Development and Study of Carbenoid Mediated Reactions stage, see section 2.2.8. The benzylidene proton ( 18.71 ppm) shows two distinct couplings with phosphorus nuclei exhibiting JH,P values of 8.48 and 15.75 Hz. Both the chemical shift and the multiplicity of the signal for the secondary proton in the isopropoxy unit of the benzylidene unit suggest rupture of the chelate in the Hoveyda- type carbene ( 4.79, heptuplet, JH,H = 6.06 Hz). For comparison, the chemical shift of the free isopropoxy unit is 4.68 ppm,36 and is 5.28 ppm in precursor 15, where it also shows a higher multiplicity due to its coupling with the phosphorus nucleus. In the 31P spectrum, two different phosphines with a small (JP,P = 27 Hz) coupling are observed, such small coupling is very indicative of a relative cis arrangement. For example, trans bisphosphine
134 complex 5 shows a much higher coupling constant (JP,P = 196 Hz).
Figure 2.6 NMR spectra of intermediate 16. Proton (left) and phosphorus (right).
Unfortunately, intermediate 16 cannot be isolated. Crystallization attempts showed that 16 releases triphenylphosphine slowly to form complex 17 toghether with decomposition products. Treatment of 16 with an excess CuCl, as phosphine scavenger, produces the brown complex 17 cleanly. After purification by column chromatography under strict exclusion of air and moisture, pure 17 can be obtained. X-ray quality single crystals allowed the determination of the molecular structure of 17 (figure 2.7).
2.2.2. Structural characterization of complex 17
Comparison of the molecular structure of complex 17 with the phosphino–phenolate complex 6a reveals a series of distinct structural features resulting from the use of a sulfonate as anchoring group. Most evident is the occupation of the coordination site
36
2. Ruthenium catalyzed A–ROMP trans to the benzylidene carbon by one of the oxygen atoms of the sulfonate group, having a Ru1–O7 distance of only 2.3957(14) Å. It is generally accepted that this coordination site remains vacant on most active ruthenium metathesis catalysts, nonetheless, weak agostic interactions on this site have led to the observation of unexpected structures.134 Two main geometric differences arise from the bridging over two atoms of the sulfonate group in comparison to the single–atom bridge in the phenolate case: 1) the non–chelating phosphine substituents (i.e. tert-butyl and phenyl) are forced closer to the chloride anion as evidenced by the significant increase in the C9– P2–Ru1 angle from 100.46(11) ° in 6a to 111.19(6) ° in 17. Concomitantly, the angles C14–P2– Ru1 (6a 118.55(13) ° vs 17 117.55(6) °) and C18–P2–Ru1 (6a 114.36(12) ° vs 17 112.53(7) °) are reduced. As a consequence of the increased steric demand the chloride atom is forced away from the phosphine and closer to the iso-propoxy unit of the Hoveyda-type benzylidene. The angle P2–Ru1–Cl4 in 6a is widened from 98.31(3) ° to 102.394(18) ° in complex 17. 2) The phosphine is twisted around the Ru–P bond placing its non–chelating substituents in an eclipsed rather than staggered configuration with respect to the other ruthenium ligands. The torsion angle C24–Ru1–P2–C9 is increased from 92.60(16) ° in 6a to 127.81(9) ° in 17.
Another particular characteristic in the structure of complex 17 is the exceptionally long Ru–OiPr(benzylidene) bond (Ru1–O31 2.3261(14) Å). For comparison, distances ranging between 2.238 and 2.281 Å have been reported for several catalysts bearing an isopropoxy unit.143 Comparable distances have been only been observed when the alkoxy unit is modified with the more sterically demanding cyclohexyl group. For example, catalyst 6a has a Ru–O bond distance of 2.310(2) Å. A long ruthenium–oxygen bond is normally associated with a faster catalyst initiation.
The chemical shift of the carbenic proton on the benzylidene moiety is highly influenced by the electronic environment arround the ruthenium center. The chemical shift of the benzylidene proton in complex 17 is shifted downfield with respect to its phenolate analog 6a ( 17.06 vs 15.14 ppm) as a result of the reduced electronic donation from the anchoring sulfonate to the metal. As expected, the Cl–Ru bond is shortened, to some extent, from 2.3397(10) in 6a to 2.3387(5) Å in complex 17.
37
Discovery, Development and Study of Carbenoid Mediated Reactions Figure 2.7 Crystal structure of complex 17. ORTEP Plot, ellipsoids at 50% probability. Hydrogen atoms have been omitted for clarity. Selected bond distances (Å) and angles (°): Ru1–O5 2.1060(14), Ru1–O7 2.3957(14), Ru1–O31 2.3261(14), Ru1–P2 2.2593(2), Ru1–Cl4 2.3387(5), Ru1–C24 1.846(2), C24–Ru1–P2 92.75(6), C9–P2–Ru1 111.19(6), C14–P2–Ru1 117.55(6), C18–P2–Ru1 112.53(7), P2–Ru1–Cl4 102.394(18), C24–Ru1– P2–C9 127.81(9)
2.2.3. Alternating ROMP catalyzed by complex 17
Complex 17 was tested on the copolymerization of norbornene (NBE) and cis-cyclooctene (COE) in dichloromethane at various temperatures and olefin ratios. For comparison purposes complex 6a was also used (Table 2.1).
Table 2.1 Copolymerization of NBE and COE using complexes 17 and 6a.a
Entry Comp. mol ratio T t Yield Alternating Poly Poly Z COE/NBE (°C) (min) (%)b linkages NBE COE olefin (%)c (%)c (%)c (%)d 1 17 50 25 15 113.4 79.9 2.9 17.2 45.7 2 17 20 25 15 100.8 79.9 7.6 12.4 43.7 3 17 20 0 15 94.0 88.6 6.4 5.0 45.2 4 17 20 -13 15 51.1 90.5 8.2 1.4 45.2 5 17 20 -23 60 49.7 88.4 9.8 1.8 44.6 6 6a 50 25 30 45.2 46.7 52.5 0.7 30.7 7 6a 20 25 15 49.9 34.0 66.0 0.0 29.9 8 6a 20 0 30 43.1 20.8 78.5 0.8 25.6 a Performed normally with ca. 150 mg of NBE and the corresponding amount of COE in a total volume of 20 mL completed with dichloromethane. Catalyst to NBE ratio in all experiments was 1:1000. b The yields calculated as: Yield=((mass of polymer/(molar mass NBE + molar mass of COE)/(n mol of NBE))*100. Yields over 100% are consequence COE polymerization after NBE depletion. c Determined by 13C NMR integration, see figure 2.8. d Measured by integration on the 1H NMR spectra. Comparing the yields obtained at the same conditions, it is evident that complex 17 catalyzes the copolymerization of NBE and COE more efficiently than complex 6a. At a NBE/COE molar ratio of 1:50, complex 17 produces 113.5% polymer after just 15 minutes
38
2. Ruthenium catalyzed A–ROMP whereas 6a only reaches 45.2% after 30 minutes (table 2.1 entries 1 and 6). At this olefin ratio the chemoselectivity of 17 is almost twice that of 6a, reaching 79.9 % compared to 46.7 %. It must be noted that complex 17 is active enough to homopolymerize excess COE after alternating copolymerization has depleted the NBE monomer, giving rise to higher than 100% yields and an apparent reduced chemoselectivity. On the other hand, complex 6a affords a polymer showing alternating pairs and poly-NBE, as a result of a lower activity and a more modest selectivity. Complex 6a has been shown to reach its best chemoselectivity (76 %) at 0 °C when reacted with a 1:200 mixture of NBE and COE.134
Upon reduction of the olefin ratio to 1:20 complex 17 retains its overall chemoselectivity as a result of an increase of poly-NBE content together with a reduction of the amount of poly-COE (entry 2). Complex 6a shows a significantly reduced chemoselectivity of 34% (entry 7) when used in the same conditions. Lowering the reaction temperature to 0 °C has the beneficial effect of reducing the amount of homopolymers formed when using 17, affording the best overall result, achieving 94.0% yield with 88.6% alternating linkages where only 6.4 and 5.0 % of homo–NBE and homo–COE were present respectively (entry 3). A polymer with a high content of poly-NBE was produced by 6a under these conditions (entry 8). Further reduction in the temperature did not improve the results and led to reduced yields even at prolonged reaction times (entries 4 and 5). Figure 2.8 represents graphically the effects discussed above.
A surprising characteristic of the polymers obtained when catalyst 17 was used is theier high content of Z-olefin. In comparison, complex 6a produced only 30.7 % Z-olefin in the best case (entry 6) compared to 45.7 % for 17 (entry 1). The observed stereoselectivity is not only higher than that shown by 6a but also higher than that of complex 6c, a catalyst exhibiting essentially the same chemoselectivity on this copolymerization (89 %).57b Only complex 6i, the most stereoselective A-ROMP catalyst reported, produces a polymer with a higher Z-olefin content of 51%.57a
Both the observed increase in chemo– and stereoselectivity in complex 17 can be related to the geometric parameters observed in its molecular structure. The expanded bridging between the aryl substituent on the phosphine and the ruthenium center provided by the sulfonate group forces the other two substituents closer to the metal center, most likely increasing the energetic difference between the two diasteromeric carbene states. The repositioning of the substituents also forces the chloride anion away from the axial phosphine, increasing the steric demand just above the region where the decisive
39
Discovery, Development and Study of Carbenoid Mediated Reactions metallacyclobutane structure is formed. In accordance to the model proposed by Torker and Chen for the control of the stereoselectivity,57a an increase in the content of Z-olefin in the polymer is expected from the anion placement.
Figure 2.8 13C NMR spectra of the copolymers obtained using complexes 17 (left) and 6a (right).
2.2.4. Calculation of the catalyst commitment
In terms of activity, both complexes 17 and 6a are expected to activate with a relatively similar rates due to their comparable Ru–O(CH(CR)2Ar) bond distances. Nonetheless, complex 17 shows an overall higher activity, producing higher yields and even homopolymerizing cyclooctene, a monomer that is essentially unreactive when 6a is used. This difference must then arise from an intrinsically more reactive propagating species. As mentioned before, Grubbs125 and later Chen130 explained the difference in intrinsic reactivity between first and second generation catalysts in terms of catalyst commitment.144 On a potential energy surface low barriers for metallacyclobutane formation and high barriers for precatalyst formation, i.e. phosphine coordination, are characteristic of highly committed catalysts. In view of the marked difference in activity displayed by the two catalysts compared, a theoretical evaluation of their commitment was performed. For this purpose, density functional theory (DFT) was used to calculate the relative Gibbs–corrected energy (table 2.2) of the first 5 critical points of the reaction profile describing the energy change for going from precatalyst (A) to 14e- methylidene (B), olefin –complex (C) followed by the transition state (D) leading to the
40
2. Ruthenium catalyzed A–ROMP metallacyclobutane intermediate (E) (figure 2.9). The reaction profiles correspond to the metathesis of ethylene with a methylidene analog of complex 17 and 6a. Due to the asymmetric nature of the phosphine in these complexes, there are two possible starting geometries for the calculations, one where the carbene moiety is positioned under the phenyl substituent, and a second where it is located under the tert-butyl group. Both starting structures should lead to different energetic profiles. For the calculation of the commitment, the path starting with the lower energy structure, methylidene under the phenyl group, was chosen, since the resulting profile should be endothermic in character and therefore performance–limiting. In both the molecular structure of complex 17 and 6a the benzylidene moiety is found under the phenyl group, confirming the thermodynamic preference for this state. For comparison purposes, the energetic profile was also calculated for the first and second generation Grubbs catalysts (figure 2.9).
Figure 2.9 Calculated Gibbs–corrected energy surface for the first steps in the metathesis of ethylene with several ruthenium catalysts. Calculated at the level M06L/6-31G**(non metals)+SDD(ruthenium). Structures show the key intermediates, mass balances are not shown but have been taken into account. All curves are referenced to structure C.
41
Discovery, Development and Study of Carbenoid Mediated Reactions Table 2.2 Calculated Gibbs–corrected energies (at 273.15 K) for the structures figure 2.9
Entry Complex Gibbs free energy of Catalyst Ligand structures in figure 2.9 commitment dissociation (kcal/mol) B/D energy B-A A B D E (kcal/mol) 1 17 -18.0 3.0 3.1 -3.3 0.96 21.0 2 6a -15.0 4.1 10.3 0.2 0.39 19.1 1st gen. 3 -23.3 1.6 2.9 -5.9 0.54 24.9 Grubbs 2nd gen. 4 -22.3 5.0 1.2 -6.5 4.19 27.3 Grubbs Graphically presented in figure 2.9, the calculated Gibbs–corrected energy values at 298.15 K (table 2.2) for the metathesis of ethylene show clearly that the phosphino sulfonate ligand provides a more favorable ratio between the energies for metallacyclobutane formation (productive step) and ligand coordination (unproductive) than the phosphino phenolate system (catalyst commitment: 0.96 for 17 vs. 0.39 for 6a, entries 1 and 2). This result is in line with Straub’s calculations which predict a stabilization of the active carbene conformations when electron poor anionic ligands are used.131 The ligand dissociation energy calculated for complex 17 (19.1 kcal/mol) is very close to that of complex 6a (21.0 kcal/mol), and in very good agreement with the predictions based entirely on the Ru–O(CR2)(benzylidene) bond distances found in the molecular structures of these complexes. For comparison, the profiles for first and second generation Grubbs catalysts were also calculated (entries 3 and 4). The calculated catalyst commitment for these complexes (0.54 and 4.19 for first and second generation respectively) corresponds with the markedly higher intrinsic activity of the second generation systems. Moreover, the ligand dissociation energy for the first generation catalysts is calculated to be lower than for the second generation, 24.9 kcal/mol and 27.3 kcal/mol respectively, in accordance with the experimentally determined values.125 Although general agreement between this model and the experimentally determined rates is achieved, one must exercise caution to draw conclusion from the comparison of absolute energetic values. The main goal of the theoretical calculations here presented is to gain insight into the intrinsic reactivity differences between the complexes studied, and only the trends are accurate enough to be considered. If absolute values of activation parameters are desired, this level of calculation is by no means replacement for the solution phase kinetic measurements.
42
2. Ruthenium catalyzed A–ROMP 2.2.5. Symmetrically substituted phosphino sulfonate complex
As seen in the molecular structure of complex 17, the sulfonate anchoring group binds the ruthenium atom with two of its oxygen atoms. This chelating coordination introduces an additional strain not found on the phosphino phenolate system. As discussed previously, Adlhart and Chen found that the carbene moiety must switch sides after each productive metathesis step on ruthenium catalysts.127 In the case of the phosphino sulfonate system, after one metathesis step the carbene should end up in the coordination position initially occupied by the one of the sulfonate’s oxygen atoms (Scheme 2.7, transition from 18 to 19). If the rupture of the chelation introduces an additional energetic barrier the anchoring group itself would influence the catalyst’s chemoselectivity. On the other hand, if an isomerization pathway exists to regenerate the chelation on the side opposing the final carbene location no influence on the selectivity would be expected (scheme 2.7 structure 20).
Scheme 2.7 Inversion of sulfonate chelation and its effect on the energy of the two possible diasteromeric states of the carbene in phosphino sulfonate systems (left). Symmetrically substituted phosphino sulfonate complex 21 (right).
In order to test whether the sulfonate group influences the energy of the two possible diasteromeric states of the carbene the symmetrically substituted complex 21 was synthesized i and tested for the copolymerization of NBE and COE. Under the same conditions used for complex 17 (NBE/COE ratio 1:20, room temperature) a 54.4% yield of polymer is obtained. This polymer is composed mainly of polynorbornene, some polycyclooctene and a small fraction of alternating copolymer, see figure 2.10.
Although no formation of copolymer is reported for a symmetrical variant of complex 5133 this complex is also not a good catalyst for the homopolymerization of cyclooctene, contrary to what was found for the phosphino sulfonate complex 17. In light of this difference, the small amount of alternating copolymer produced by 21 is likely the result
i For details see experimental part, a molecular structure is available in the appendices. 43
Discovery, Development and Study of Carbenoid Mediated Reactions of a statistically possible insertion of cyclooctene in an otherwise typical polynorbornene chain. The composition of the polymer obtained using catalyst 21 strongly suggests that the chelate ring inverts during the metathesis reaction producing energetically equivalent carbene states, therefore not affecting the chemoselectivity induced by the catalyst.
Figure 2.10 13C NMR of the polymer obtained using complex 21.
2.2.6. Chloride replacement with bulky sulfonate anions
Following the methodology used by Torker and Chen57a a series of complexes was prepared in which the chloride anion of the base complex 17 was replaced by arylsulfonates of increasing steric demand.
Figure 2.11 Phosphino sulfonate complexes bearing arylsulfonate anions.
Single crystals of suitable quality for molecular structure determination by X-ray diffraction were obtained for complexes 22a, 22b and 22c, these structures are shown in figures 2.12 to 2.14. In table 2.3 selected geometric parameters for those structures are compared with the available structures of phosphino phenolate complexes 6f and 6i.57a
44
2. Ruthenium catalyzed A–ROMP As seen in the case of complex 17, the reduced electron donation provided by the sulfonate anchoring group causes a bond shortening of the anion trans to it. In the case of complexes 22a to 22c the arylsulfonates exhibit shorter Ru1–O6 bond lengths, the longest being 2.058(7) Å in 22a, compared to their phenolate analogs where the shortest measured Ru1–O6 distance is 2.070(5) Å. Although subtle in appearance, minor geometric differences have been already shown to yield catalysts capable of producing polymers with very different characteristics (vide supra). Contrary to what is observed in the phosphino phenolate complexes, addition of ortho substituents on the aryl sulfonate causes a shortening of the Ru1–O6 bond from 2.058(7) Å in 22a to 2.041(4) in 22b or 2.040(9) in 22c. In addition, a significant increase in the bond length Ru1–O6 is observed. Ligands 7b and 7c induce a larger trans influence than the simple tosylate, as a result of the increased electron density provided by the ortho alkyl groups. The lengthening of the Ru1–O6 could be therefore explained on the basis of the electronic character of the aryl sulfonates used. Complex 6i also exhibits an increase in this bond length compared to 6f, although the corresponding anion is not closer to the metal center. It appears as if the Ru1–O6 bond length differences in the phosphino phenolate systems arise mainly because of the steric requirement of the aryl sulfonates rather than their electronic character.
Table 2.3 Selected bond lengths (Å) and angles (°) from the molecular structures of complexes 22a, 22b, 22c, 6f and 6i.
Entry 1 2 3 4 5
Complex 22a 22b 22c 6f 6i
Ru1–O6 2.107(6) 2.122(4) 2.127(8) 1.975(5) 1.986(2)
Ru1–O13 2.058(7) 2.041(4) 2.040(9) 2.070(5) 2.076(2)
Ru1–O5 2.441(7) 2.387(4) 2.478 - -
Ru1–O7 2.285(7) 2.294(4) 2.294(8) 2.242(6) 2.250(2)
Ru1–P9–C10 112.6(3) 111.36(19) 111.1(4) 99.6(3) 100.32(11)
Ru1–O13–S4 131.0(4) 132.7(2) 128.2(5) 136.0(4) 128.91(13)
C8–Ru1–P9–C10 122.0(4) 124.7(2) 109.2(5) 89.3(4) 94.4
45
Discovery, Development and Study of Carbenoid Mediated Reactions Figure 2.12 Crystal structure of complex 22a. ORTEP plot, ellipsoids at 30% probability. Hydrogen atoms have been omitted for clarity. For selected bond lengths and angles see table 2.3, entry 1.
Figure 2.13 Crystal structure of complex 22b. ORTEP plot, ellipsoids at 30% probability. Hydrogen atoms have been omitted for clarity. For selected bond lengths and angles see table 2.3, entry 2.
46
2. Ruthenium catalyzed A–ROMP Figure 2.14 Crystal structure of complex 22c. ORTEP plot, ellipsoids at 30% probability. Hydrogen atoms have been omitted for clarity. For selected bond lengths and angles see table 2.3, entry 3.
Both the series of complexes 22a to 22c, and 6f and 6i have shorter Ru1–07 distances compared to the corresponding distances in the chloride parent complexes 17 and 6a. It could be predicted that the presence of aryl sulfonates as ligands in this type of complexes may cause a reduction in the initiation rate of the catalysts. One of the main differences between the phosphino sulfonate and phenolate systems is the ability of the former to force the phosphine’s non–chelating substituents (i.e. tert-butyl and phenyl) closer to the metal center. The indicative Ru1–P9–C10 angle is first increased in complex 22a (112.6(2) ° vs 11.19(6) ° in 17) but then it decreases along the series in 22b and 22c. As a result of this structural changes an effect on the chemoselectivity of the complexes is expected.
Table 2.4 Chemical shifts for the benzylidene proton on phosphino sulfonate and phenolate57a complexes.
Entry Complex benzylidene proton chemical shift 1H NMR ( in ppm) CD2Cl2 C6D6 1 17 17.14 17.13 2 22a 17.63 17.86 3 22b 17.70 17.84 4 22c 17.70 17.82 5 22d 17.65 17.75
47
Discovery, Development and Study of Carbenoid Mediated Reactions
6 6a 15.65 N.A. 7 6f 16.29 N.A. 8 6i 16.34 N.A.
The chemical shifts for the benzylidene proton in CD2Cl2 and C6D6 on complexes 22a to 22d (table 2.4 entries 2 to 5) are shifted downfield with respect to complex 17 (entry 1) as expected from a more electron poor metal center. However, in CD2Cl2, as observed in the phosphino phenolate case, the presence of ortho alkyl groups causes a further downfield shift opposing what would be expected from the increased electron density in ligands 7b to 7d and particularly contrasting to the bond lengths observed in the solid state. In
C6D6, a less polar solvent, the increase in electron donating capability of the ligands is reflected in a progressive upfield shift. It has been observed that the use of a less polar solvent benefits the stereoselectivity of the catalyst most likely by disfavoring ligand dissociation.145 The marked difference in trend for the benzylidene chemical shift in
CD2Cl2 and C6D6 strongly supports this hypothesis, i.e. it suggests different states of association in the two solvents.
2.2.7. Polymerization experiments with sulfonate complexes
Complexes 22a to 22d were tested for the copolymerization of norbornene and cyclooctene in dichloromethane solution at an olefin ratio of 1:20. 1H and 13C NMR spectra of the obtained copolymers are shown in figure 2.15. Upon replacement of the chloride (complex 17) for tosylate (complex 22a) the content of Z-olefins increases from 43.7 to 47.2 % (table 2.5, entries 1 and 2). Further increase in the steric demand of the sulfonate anions produces an additional increase in Z-olefin content of the polymer, reaching 49.3, 58.8 and 57.1 % for complexes 22b, 22c and 22d respectively (entries 3 to 5). Although the stereoselectivity observed for complex 22c is the highest ever achieved in the alternating ROMP of norbornene and cyclooctene, analysis of the carbon NMR spectra of the polymers and the obtained yields reveal that it was accomplished at the expense of the catalyst’s chemoselectivity and activity.
48
2. Ruthenium catalyzed A–ROMP
Figure 2.15 1H and 13C NMR of the copolymers obtained using the all the prepared phosphino sulfonate complexes.
The most stereoselective complex 22c also shows a modest activity, producing only 38.8% copolymer over a period of 60 minutes. At the same time, chemoselectivity drops significantly to only a 44.5% of alternating dyads. It is then not surprising that the bulky arylsulfonates reduce the overall activity of the complexes, but it is not straightforward to explain why these catalysts become less chemoselective. When strongly bound thiosulfonates were used in the classical phenolate framework (vide supra) a drop in activity was observed and explained in terms of an electronic deactivation and steric shielding imposed by the ligands on the catalysts. However, no detrimental effect on the chemoselectivity was observed. In the case explored here, the catalyst’s steric shielding could account for the lower activity of the complexes bearing larger ligands, but not the drop in chemoselectivity. From the molecular structures of complexes 22a, 22b and 22c it was noted that the larger ligands force the phosphine non–chelating substituents away from the metal center (decreasing Ru1–P9–C10 angle, see table 2.3). Reducing this angle also reduces the influence of these phosphine substituents on the two diasteromeric carbene states, which ultimately determine the catalyst’s ability to energetically distinguish between the two monomers. The use of a sulfonate as an anchoring group also gives the backbone an increased flexibility not present, in the phenolate case, allowing the anionic ligand to influence not only the stereoselectivity in a direct way but also the chemoselectivity indirectly.
49
Discovery, Development and Study of Carbenoid Mediated Reactions Table 2.5 Copolymerization of NBE and COE using complexes 17, 22a, 22b, 22c and 22d.a
Entry Complex T t Yield Alternating Poly Poly Z (°C) (min) (%)b linkages NBE COE olefin (%)c (%)c (%)c (%)d 1 17 25 15 100.8 79.9 7.6 12.4 43.7 2 22a 25 30 76.9 80.5 18.3 1.2 47.2 3 22b 25 60 74.0 78.3 19.5 2.2 49.3 4 22c 25 60 38.8 44.5 55.5 0.0 58.8 5 22d 25 60 21.5 40.3 59.7 0.0 57.1 a Performed normally with ca. 150 mg of NBE and the corresponding amount of COE in a total volume of 20 mL completed with dichloromethane. Catalyst to NBE ratio in all experiments was 1:1000. b The yields calculated as: Yield=((mass of polymer/(molar mass NBE + molar mass of COE)/(n mol of NBE))*100. Yields over 100% are consequence COE polymerization after NBE depletion. c Determined by 13C NMR integration. d Measured by integration on the 1H NMR spectra. 2.2.8. Attempts to enhance the chemoselectivity of the phosphino sulfonate systems.
The marked drop in activity of the complexes bearing the most sterically demanding sulfonates suggest that the phosphino sulfonate framework is at the limit where a further increase in the steric demand would just lead to inactive complexes. Nonetheless, if, as in the case of the phosphino phenolate system, the phosphine substituents could be optimized to achieve perfect alternation, possibly a better compromise between activity, chemoselectivity and stereoselectivity would be achieved.
With this goal in mind, the synthesis of two complexes analogous to 17 was attempted (figure 2.16). The asymmetrically substituted phosphines (13c and 13d, see experimental part) necessary for the preparation of complexes 23 and 25 are easily obtained following a procedure analogous to the one for the synthesis of phosphine 13a. Reaction of 13c with 15, followed by phosphine removal with CuCl affords complex 23. The molecular structure of complex 23 could be determined by X-ray diffraction of a single crystal, see section 6.2. The structures of complexes 23 and 17 resemble one another with only minor, but still significant differences. As a result of the presence of the adamantyl substituent the phosphorus–ruthenium bond gets elongated from 2.2593(2) Å in 17 to 2.2711(10) in 23. As a consequence, the angle Ru1–P2–C13 in 23 is reduced to 109.57(13) ° from 111.19(6) ° in 17. As could be expected from the increased distance between the phosphine substituents and the metal center, the chloride anion can be placed further away from the chelating isopropoxy unit. The angle Cl4–Ru1–P2 is widened from 102.394(18) ° in 17 to 109.57(13) ° in 23. Perhaps more interesting is the reduction in the torsion angle C36– Ru1–P2–C13 which only reaches 109.57(13) ° in 23 compared to 127.81(9) ° in 17. The
50
2. Ruthenium catalyzed A–ROMP increased steric demand of the adamantyl moiety together with the flexibility of the sulfonate anchor allows the ligand to flex in a way such that to reduce the steric interaction it places the phosphine away from the metallic center.
Figure 2.16 Complexes with phosphines bearing adamantyl (23) and neopentyl (24+25) substituents.
In contrast with complex 23, complex 25 could not be isolated in a pure from. After reaction of the phosphine precursor 13d with 15 only a mixture of the intermediate 24 and the desired complex 25 could be obtained. Addition of excess of CuCl was not effective in removing PPh3. During purification attempts it was noted that in solution the predominant form was that of 25 plus free PPh3, however, upon evaporation of the solvent 24 was formed. The solid state structure of 24 can be seen in figure 2.17. As expected from the weak donor character of the sulfonate group, the ligand trans to the sulfonate, in this case tryphenylphosphine, is strongly bound. The ruthenium– phosphorus(PPh3) bond is actually shorter than that of the bidentate phosphine (2.2916(8) vs 2.3818(8) Å respectively), although the longer distance could also be attributable to the more donating chloride trans to the bidentate phosphine. The presence of tryphenylphosphine makes difficult the comparison of the geometric parameters of the intermediate with complexes 17 and 23. Nonetheless, the expected configuration of the chelating phosphine around the ruthenium atom is similar to that of the analogous complexes. Although the neopentyl substituent is considered sterically more demanding than t-butyl or adamantyl, the lack of two substitutions at the carbon provides room for the PPh3 to remain bound. Such effects were not observed in the original phosphino phenolate complexes.
51
Discovery, Development and Study of Carbenoid Mediated Reactions
Figure 2.17 Molecular structure of intermediate 24. Ortep plot, ellipsoids at 30% probability. Hydrogens are omitted for clarity. Selected bond distances (Å) and angles (°). Ru1–O2 2.377(2), Ru1–O3 2.242(2), Ru1–P3 2.2916(8), Ru1–Cl1 2.3831(8), Ru1–C101 1.873(3), Ru1–P2 2.3818(8), Ru1–P2–C50 110.58 (11), C101–Ru1–P2–C50 113.87(13).
Complex 23 as well as the mixture of 24 and 25 were tested in the copolymerization of NBE and COE under the standard conditions. Complex 23 is even more active than 17, producing 116.5% of copolymer (table 2.6, entry 2). The composition of the polymer reveals that homopolymerization of COE is the cause for the increased yield. There is a reduction on the poly-NBE content of the polymer, which indicates an increase in the specific chemoselectivity of the complex, later spoiled partially by the poly-COE component. The stereoselectivity is somewhat reduced, down to 42.6% from 43.7 % in 17, as could be expected from the reduced steric demand on top of the chloride, that has more room to position itself away from the metallacyclobutane formed during polymerization. The mixture of 24 and 25 shows a reduced activity, 80.6% polymer produced (entry 3), under the same conditions. The observed chemoselectivity is mostly on par with that of 17 with an increase of poly-NBE. Most interesting is the reduction in stereoselectivity, 31.2 % Z vs 43.7% for 17. Most likely, the same effect that allows the PPh3 to stay in place allows the chloride to be pushed away from the reaction center, reducing its influence on the relative orientation of the substituents at the metallacyclobutane intermediate. A reduction of reaction temperature diminished to some extent the formation of poly-COE at the expense of some reactivity (entry 4).
52
2. Ruthenium catalyzed A–ROMP Table 2.6 Copolymerization of NBE and COE using complexes 23 and a mixture of 24 and 25.a The result obtained with complex 17 is shown for comparison.
Entry Complex T t Yield Alternating Poly Poly Z (°C) (min) (%)b linkages NBE COE olefin (%)c (%)c (%)c (%)d 1 17 25 15 100.8 79.9 7.6 12.4 43.7 2 23 25 15 116.5 75.4 4.6 20.0 42.6 3 24+25 25 15 90.2 80.6 13.0 6.4 31.2 4 24+25 0 15 81.9 82.1 16.0 2.0 32.2 a Performed normally with ca. 150 mg of NBE and the corresponding amount of COE in a total volume of 20 mL completed with dichloromethane. Catalyst to NBE ratio in all experiments was 1:1000. b The yields calculated as: Yield=((mass of polymer/(molar mass NBE + molar mass of COE)/(n mol of NBE))*100. Yields over 100% are consequence COE polymerization after NBE depletion. c Determined by 13C NMR integration, see figure. d Measured by integration on the 1H NMR spectra. It is possible that the most selective substituent in the case of the phosphino phenolate system (1,1,2,2-tetramethylpropyl = TMP) would have also contributed to a better chemoselectivity in the new framework. However, the subsequent substitution of the halide by bulky sulfonates would have most likely led to rather inactive complexes due to overcrowding of the catalyst. In this regard, the sulfonate as anchoring group accomplished its task, allowing the bulky arylsulfonate to be strongly bound to the ruthenium center. Nonetheless, this structurally beneficial effect caused a negative effect on the general activity of the prepared complexes. Contemporary approaches for the control of stereoselectivity in olefin metathesis promise an alternative that could be exploited to make fully chemoselective, fully stereoselective catalysts. Those approaches will be discussed in the following section.
2.3. Discussion
The first reports of alternating ROMP date back to 2002. The first strategies did not involve any modification of the catalyst itself. Coughlin obtained a high degree of alternation in special cases when the balance between the steric demand and ring strain of the two monomers was correctly adjusted.146 Grubbs prepared alternating polymers by insertion of a diacrylate into an already formed homopolymer chain.147 Following similar monomer based interactions, the group of Sanda produced alternating copolymers of an amino acid-derived norbornene unit in a system where acid-base interactions led to a preferential sequence of monomer consumption.148 More recently, Sampson reported a new case where two monomers were found to copolymerize in a sequential manner, despite the fact that neither of them would generate a
53
Discovery, Development and Study of Carbenoid Mediated Reactions homopolymer.149 Apart from our own system, catalyst controlled alternating ROMP has only been reported, by Blechert and Buchmeiser using unsymmetrical, chiral or achiral monodentate NHC carbenes in ruthenium catalysts.150 In this case, the high degree of alternation was explained by an enhanced cyclooctene insertion rate on a norbornene derived terminus and vice versa. The reason behind this selectivity was proposed to be purely steric. After a norbornene insertion the catalyst would become too sterically crowded to allow an additional norbornene molecule to insert, favoring the cyclooctene insertion, after which the more reactive norbornene would outcompete the low strain olefin producing an alternating copolymer. This type of sequence control, where the last inserted monomer determines the next, is called chain-end control. A key requirement for this proposed mechanism is the free rotation of the NHC ligand around the Ru–C bond. Rotation makes distinction between the two carbene states on each ‘side’ of the catalyst impossible and therefore rules out a diasteromeric site control mechanism (vide infra). However, if rotation of the NHC ligand were to be prevented by the steric bulk of the growing polymer chain during catalytic conditions, differentiation between the two sides of the catalyst would become possible and the mechanism of alternation would probably resemble the one operative in the phosphino–phenolate and sulfonate catalysts. In fact, thermally switchable latent ruthenium catalysts exhibiting a cis– dichloro arrangement effectively prevent the rotation of the NHC (IMes-H2) ligand at room temperature; rapid rotation around the Ru–C(carbene) bond is only observed at elevated temperatures.151
Scheme 2.8 Explanation proposed by Blechert and Buchmeiser for their alternating copolymerization of NBE and COE.150
Contrary to Blechert and Buchmeiser’s our catalyst framework allows the distinction between the two possible diasteromeric carbene states formed after each productive step. The sequence control in which the catalyst determines the next monomer to be
54
2. Ruthenium catalyzed A–ROMP inserted is known as diasteromeric site control. Our proposed mechanism bears close resemblance to that of the stereospecific polymerization of propylene catalyzed by chiral metallocenes.152 In ethylene-bridged ansa zirconocenes the relative orientation of the growing polymer chain and the incoming monomer is a result of the steric influence of the ligand.
The replacement of the phenolate by a sulfonate group as anchoring group in the described catalytic systems seems to not have modified the controlling mechanism of the copolymerization. Even though the sulfonate group clearly shows a chelating interaction with the metal center, as seen in the molecular structures of complexes 17, 22a, 22b and 22c, the same does not negatively affect the differentiating ability of the catalysts. Symmetrically substituted complex 21 shows no preference for the alternating copolymerization, strongly supporting the diasteromeric site control as the operating mechanism in this new catalyst framework.
Figure 2.16 Traditional (left) vs contemporary (right) approach to stereoselectivity on ruthenium catalyzed olefin metathesis.
The pursuit of stereoselectivity on metathesis catalysts has also received attention in the past. As discussed in the introduction, the more common approaches involve the utilization of bulky anionic ligands for ruthenium as well as molybdenum and tungsten complexes. These complexes place the bulky anions cis to the strongest donor ligand, i.e. the phosphine, NHC or imido ligand, in order to maximize the steric influence on the metallacyclobutane intermediate. The complexes described in this chapter follow the same guidelines in order to influence the Z-olefin content on the produced copolymer. More recently, a series of complexes where the chloride anions have been replaced by dithiocatecholates have shown extremely high selectivity towards the Z-arranged olefin.60 In those complexes the donating capabilities of the thiolate anions combined with their forced cis arrangement favors the metallacyclobutane formation cis to the NHC ligand, contrary to traditional systems.
55
Discovery, Development and Study of Carbenoid Mediated Reactions 2.4. Conclusions
The reduced electron-donating capability of a sulfonate group as compared to a phenolate group prompted us to use it as an anchoring group in our alternating ROMP catalyst framework. It was predicted that the reduced trans influence of a sulfonate chelating group would reduce the extent of ligand dissociation which limites the achievable Z-selectivity of catalysts bearing bulky sulfonates. The base complex 17 shows an enhanced chemo– and stereoselectivity, both of which could be explained in terms of the geometric characteristics imposed by the sulfonate as anchoring group. In this regard comparison of its molecular structure with that of previously reported phosphino phenolate complexes was crucial to determine which parameters are directly responsible for the resulting polymer properties.
Complex 17 was also found to be significantly more active than its phenolate analog 6a. The difference in intrinsic activity was explained in terms of a more favorable ratio between the energies for product and pre-catalyst formation, as revealed by DFT calculations. In this sense, phosphino sulfonate catalysts exhibit a higher commitment than the original phenolate catalysts, just as second generation Grubbs systems are better propagating than the first generation.
Replacement of the chloride anion with sulfonate ligands of increasing steric demand yielded complexes that showed increased stereoselectivity, reaching a maximum of 58.8% Z–olefin content. Unfortunately, the increased steric demand rendered the catalysts less active, most likely by steric overcrowding, and less chemoselective, by forcing the phosphine away from the metal center. The latter effect is also a result of the increased flexibility of the ligand backbone, which is not present in the original phenolate ligands.
The molecular structures of catalysts 22a, 22b and 22c indicate that their aryl sulfonate ligands are in fact more strongly bound than in the case of phosphino phenolate complexes, supporting our original hypothesis that weaker donor ligands in trans position promote stronger binding of the bulky sulfonates.
Even though these systems represent a new step towards fully stereoselective catalysts, it also seems that the use of more complex, more flexible backbones also brings drawbacks like the interconnection of chemoselectivity and stereoselectivity, which were perfectly orthogonal in the original catalysts. These unforeseen consequences remind us of the complexity and difficulty of designing highly active and selective catalysts. 56
57
58
3. Lithiomethyl trimethylammonium salts as methylene donors
3.1. Introduction
3.1.1. Yildes: bonding and reactivity
Ever since the pioneering work by Nobel laureate George Wittig in the 1950s, ylides have become one of the most important tools in synthetic organic chemistry.153 Ylides are in essence carbanionic nucleophiles, but at that, a particular set of them. Best described by A. W. Johnson “an ylid can be defined as a substance in which a carbanion is attached directly to a heteroatom carrying a high degree of positive charge”.154 Ylides can be represented with two extreme resonance structures (scheme 3.1). On the left, a zwitterion or ‘ylide’, and on the right one without formal charges, an ‘ylene’.
Scheme 3.1 Ylide and ylene representations of (Ph)3PCH2, a common olefination reagent.
The precise contribution to the actual structure from these two extremes depends on how strong the stabilizing interaction between the onium group (positively charged heteroatom) and the carbanion is. Cheng studied the acidity constants of various ylides in DMSO and found that the stabilizing capability of the onium fragment followed the order N < P < S.155 In this work, the stabilization mechanism was explained by the combination of two main components: 1) a simple electrostatic interaction between the positive and negative charges, which in turn depends on the C-X bond distance and the onium group polarizability, and 2) a resonance stabilization. The nature of the resonance stabilization in ylides, particularly the well-studied phosphorus ylides, has been source of intense debate. Involvement of the phosphorus 3d orbitals in a d–p bonding scheme (valence shell expansion) was proposed in early works.156 More recently, accurate calculations have shown that the phosphorus d orbitals act as polarization functions but are not directly involved in bonding.157 The currently accepted model involves a combination of negative hyperconjugation and polarization effects.158 The details of this stabilization will be discussed in comparison to the nitrogen ylides in section 3.2.2. The body of evidence nonetheless points towards more stabilized, more ylidene-like structures for the S and P ylides,159 whereas the N ylides, uncapable of resonance stabilization, exhibit a structure more in line with the ylide representation.160
59
Discovery, Development and Study of Carbenoid Mediated Reactions Ylides can undergo three main types of reactions: olefination, cyclization to three– membered rings and rearrangement.103, 161 The differences in reactivity are dictated by the central heteroatom and a) its influence on the nucleophilicity of the adjacent carbanion and b) the leaving group ability of the onium fragment.162 The stabilization provided by the onium fragment affects directly the nucleophilicity of the carbon atom, however, the barriers of addition to different electrophiles are generally very low and therefore are usually not rate determining. On the contrary, departure of the onium fragment constitutes the highest barrier and determines the outcome of the reaction. The order of leaving group ability has been calculated to follow the order O > S > N > P. 162 In fact, the calculated energy for phosphine displacement is so high that the alternative formation of oxaphosphetanes becomes competitive, ultimately leading to the extrusion of phosphine oxide and the formation of an olefin, i.e. the useful Wittig reaction.163 For the sulfur and nitrogen ylides similar reactivities have been so far described, albeit under different conditions. The ease by which sulfides are displaced from the product of addition of sulfur ylides together with their great stability, have certainly helped the wide spread adoption of the Corey-Chaykovsky reaction104 for the formation of three– membered rings.
3.1.2. Nitrogen ylides
Nitrogen ylides have been mostly used in relation with their participation in rearrangements, of which Stevens164 and Sommelet-Hauser165 are the most prominent examples (scheme 3.2).
Scheme 3.2 Stevens (left) and Sommelet-Hauser (right) rearrangements.
The rare use of nitrogen ylides, in comparison with their sulfur and phosphorus analogs, is in sharp contrast with the fact that N-ylides were, in fact, discovered first. The first report of proton removal from a pyridinium salt dates back to 1935. Krohnke and coworkers described the deprotonation of N-phenacylpyridinium bromide,166 marking the start of their research on pyridinium ylides and related substances.
60
3. Lithiomethyl trimethylammonium salts as methylene donors
Scheme 3.3 Preparation of an pyridinium ylide by Krohnke in 1935.166
After 12 years, in 1947, Wittig and Wetterling, during an attempt to prepare pentavalent nitrogen compounds, reported the deprotonation of tetramethylammonium bromide with phenyllithium in ethereal suspension.167 The product of which was rationalized as the LiBr adduct of trimethylammoniummethylide (scheme 3.4).
Scheme 3.4 Wittig and Wetterling synthesis of lithium coordinated ammonium ylide.167 ‘N-Ylide’ 27 is shown as originally proposed by Wittig.
The reactivity shown by 27 was that expected of a typical nucleophile (scheme 3.5). It produced tetramethylammonium hydroxide when reacted with water, the product of addition 32 in presence of benzophenone (30), and ethyltrimethylammonium iodide (33) in reaction with methyl iodide. However, the almost complete insolubility of both the starting ammonium salt and the ‘ylide’ product in ethereal solvents forced extremely long reaction times (deprotonation and subsequent reaction) and reduced the general applicability of this reagent.
Scheme 3.5 Reactivity of nitrogen ‘ylide’ 27 with diverse electrophiles.
Studies by Wittig and Polster confirmed the importance of lithium coordination by studying the decay behavior of 27.168 Deprotonation of 26 with phenyllithium in ether over a period of 90 hours produced 20% of triethylamine and 12% of polymethylene, the
61
Discovery, Development and Study of Carbenoid Mediated Reactions latter supposedly formed by polymerization of the resulting free methylene formed in the decomposition of the ylide into the amine and the carbene. In presence of more coordinating solvents like dimethylether and ethylene glycol, the polymethylene yield increased to 74%. This increase was related to the higher ability of these solvents to coordinate the lithium cation removing it from the ‘ylide’ 27 promoting its decay. Although proposed as the key intermediate in this polymerization, the formation of free methylene under this conditions is highly unlikely.
Scheme 3.6 Wittig’s proposed decay pathway of 27 in presence of coordinating solvents.168
The possibility of free carbenes being formed from the decomposition of 27 was explored in relation with their cyclopropanating ability. In 1960, Franzen and Wittig reported the cyclopropanation of cyclohexene with 27 in variable but low yields (between 5 and 18%).169
Scheme 3.7 Franzen and Wittig’s reported cyclopropanation of cyclohexene.169
Later attempts to reproduce this reaction by Krauss and Wittig were however unsuccessful.170 They could nonetheless cyclopropanate cyclohexene with the ‘ylides’ obtained by deprotonation of alkoxy–substituted ammonium salts. A mixture of endo and exo isomers resulted as expected from the asymmetry of the expected methylene.
The study of the chemistry of ‘ylide’ 27 and other non-stabilized ylides by Wittig171 and others172 continued over the years, however the results pointed to a reactivity parallel to the one of traditional carbanionic nucleophiles. No special “ylide” behavior was found in contrast to what was discovered for the sulfur and phosphorus analogs.
62
3. Lithiomethyl trimethylammonium salts as methylene donors
Scheme 3.8 Cyclopropanation of cyclohexene with ‘N-ylides’ derived from alkoxy- substituted ammonium salts.170
Only very recently, nitrogen ylides stabilized with electron withdrawing groups at the carbanionic atom have been considered for the synthesis of three–membered rings.173 Initial reports by Jończyk and coworkers of the stoichiometric use of cyano-stablized ylides for the synthesis of epoxides, aziridines and cyclopropanes with electrophilic olefins174 were soon followed by the discovery by the group of Gaunt that nitrogen ylides could be used in catalytic conditions for the cyclopropanation of Michael acceptors.175 These reports marked the start of a revitalized interest in the nitrogen ylides as nucleophilic substrates, providing an alternative to the well stablished sulfur ylide methodologies.
3.2. Lithiomethyl trimethylammonium reagents as methylene donors
Even though great advances haven been made on the synthetic use of nitrogen ylides, the structure and more detailed reactivity of the simplest nitrogen ‘ylide’ 27 remain unclear. Earlier attempts to study 27 have been limited to Wittig’s own work and O’Hair’s gas–phase studies.176 Our interest in simple nitrogen ylides as potential methylene donors motivated us to gain further understanding of these reagents in solution.
3.2.1. Soluble tetramethylammonium salts
The acidity constant of the tetramethylammonium ion has been estimated
177 178 experimentally and computationally to be close to 42 pKa(DMSO) units. Therefore, common organolithium compounds like phenyl and butyl lithium (the conjugated acid
179 of CH3Li, CH4 has a pKa(DMSO) ≈ 56 ) should be efficient bases for proton removal from an ammonium salt like 26. Nonetheless, extremely long deprotonation times, of up to two days, have been characteristic in previous reports. It was rationalized that these very long reaction times are consequence of the almost negligible solubility of the parent ammonium salt rather than a low intrinsic acidity.
63
Discovery, Development and Study of Carbenoid Mediated Reactions Replacement of the halide in tetramethylammonium iodide 43 with the weakly
F coordinating anion tetrakis(3,5-bis(trifluoromethyl)phenyl)borate (BAr ) produced NMe4 BArF 45, which was found to be fully soluble in ether and THF (Scheme 3.9, top). Reaction of 45 in THF at -78 °C with 1 equivalent of nBuLi in a closed vessel generated the nitrogen ‘ylide’ 47 in only 5 minutes (scheme 3.9 bottom). Addition of a second equivalent of nBuLi formed the ‘diylide’ 48, however this latter species is basic enough to deprotonate the BArF anion regenerating 47.
F Scheme 3.9 Top: Synthesis of NMe4 BAr 45. Bottom: Reaction of 45 with nBuLi to generate nitrogen ‘ylide’ 47 and ‘diylide’ 48.
For the first time, the structure of the simplest nitrogen ‘ylide’ 47 could be assesed by NMR spectroscopy. 1H, 1H-6Li HMBC and 1H-15N HMCB experiments were performed to confirm the structure of 47 in solution (figure 3.1). ‘Ylide’ 47 shows two distinct signals on the 1H NMR spectrum at 2.27 and 2.95 ppm with an integral ratio of 2 to 9. The upfield shift observed for both the three methyl and one methylidene groups is expected as removing one proton increases the electron density of the molecule compared with the parent tetramethylammonium ion. Comparison of the chemical shifts of other similar products of deprotonation with those of ‘ylide’ 47 is shown in table 3.1.
Heteronuclear multiple-bond correlation experiments allowed the connectivity on 47 to be established. As seen in the 1H,6Li HMBC spectrum (figure 3.1), there is only one correlation between lithium and the deprotonated methyl group ( 2.27 ppm) in 47. On the other hand, the 1H,15N HMBC shows two correlations, one with the three methyl groups ( 2.95 ppm) left in the ammonium ion, and a second one with the methylene group correlating with lithium. These results strongly support a structure where a
64
3. Lithiomethyl trimethylammonium salts as methylene donors specific methyl group is deprotonated generating a carbanion stabilized by bonding with lithium, reminiscent of the structure of neopentyllithium 55.
Figure 3.1 1H, 1H,6Li HMBC and 1H,15N HMCB NMR spectra of 47.
Table 3.1 Chemical shifts for some deprotonated species and their parent compounds
entry Parent /ppm Derivate (CH3) /ppm (CH2) /ppm
1 2.47180 1.38181 -0.7181
50 (DCCl3) 51 (N/A)
2 3.86182 3.13183 1.7183 (DMSO) (Dioxane) 52 53
3 0.902184 1.13185 -0.67185 54 (CCl4) 55 (Benzene)
4 3.17 2.96 2.27 46 (THF) 47 (THF)
In order to test experimentally whether lithium is essential for the stability of 47, like in the case of organolithium compounds and unlike P and S ylides, the stability of 47 in a closed NMR tube at -30 °C in THF-d8 was monitored in the absence and in the presence of 12-crown-4, a strong lithium-coordinating agent. As it can be seen in figure 3.2, in a closed NMR tube 47 is rather stable, showing no major changes in concentration over a period of 70 minutes. In the blank experiment addition of a portion of solvent under
65
Discovery, Development and Study of Carbenoid Mediated Reactions argon counterflow causes some degradation of 47 but once the NMR tube is closed and cooled down the remaining 47 shows the same stability as before. On the contrary, addition of 12-crown-4 causes a higher fraction of 47 to decay and increases the rate of decomposition irrespective of the closure or the cooling of the NMR tube.
0.050
0.045
0.040 [47]blank = -1.94E-05t + 2.98E-02
(M) 0.035 R² = 9.87E-01
47 0.030
0.025
0.020
0.015 [47]12-crown-4 = -5.75E-05t + 2.54E-02 Concentration 0.010 R² = 9.91E-01
0.005 12‐crown‐4 experiment blank experiment
0.000 0 50 100 150 200 t (min)
Figure 3.2 Stability of 47 in THF solution at -30 °C in absence (white circles) and in presence of 12-crown-4 (black circles).
Based on the evidence obtained so far, it seems clear that the lithium cation is essential for the stability of 47, in contrast to what is observed for P and S ylides, which can be isolated “salt free”.183, 186 The former designation of 47 as “trimethylammonium methylide” wrongly describes the nature of this species, therefore the term “lithiomethyl trimethylammonium”, which relates to the actual structure of 47 in solution, is recommended over the former name.
3.2.2. Bonding in the lithiomethyl trimethylammonium ion
In order to gain a better understanding of the bonding situation in 47, and how it compares with its closest P analog, trimethylphosphonium methylide (50), a series of DFT calculations were performed.187
The calculated thermochemistry for the isodesmic188 reaction shown in table 3.2, reveals that the lithium cation stabilizes structure 47b by 8.1 kcal/mol more than 51b in vacuo, and by 3.6 kcal/mol in solution.
66
3. Lithiomethyl trimethylammonium salts as methylene donors Table 3.2 Stabilization of ylides by lithium cation
Environment Reaction Energy (kcal/mol)
In vacuo -8.1a
Solution -3.6b a M06L/aug-cc-pVTZ yero point corrected energies. b M06L/aug-cc-pVTZ/SMD(THF) Gibbs-corrected energies (298.15 K) with two explicit THF molecules. Moreover, extended transition state analysis of the structures shown above187 reveal that the number of stabilizing orbital interactions and steric repulsion remain relatively constant irrespective of whether lithium is present or not in the nitrogen ‘ylide’ (structures 47a and 47b). On the other hand, the presence of lithium in 51b reduces the amount of stabilizing orbital interactions and increases the steric repulsions compared to 51a, resulting in a net destabilizing effect. More illustrative perhaps is the effect of these interactions on the corresponding optimized geometries (figure 3.3).
In the lithium-free nitrogen ylide 47a the methylide moiety adopts a tetrahedral geometry, whereas in the P-ylide 51a the geometry is closer to trigonal planar (X-C-H angle widens from 114.1 ° in 47a to 148.3 ° in 51a). Lithium-coordinated ylides 47b and 51b exhibit both a tetrahedral geometry in their lithiomethyl moieties. Similar structural differences have been calculated for oxygen and sulfur ylides.189 Another differentiating characteristic can be observed in the heteroatom-carbon bond distances on the two ylides. In lithum-free phosphorus ylide 51a the P-methylene bond is shortened and the P- Me(1) bond is elongated compared to the P-Me bonds in tetramethyphosphonium ion 50a. In the lithium-coordinated structure 51b these bonds are affected in the same manner albeit to a smaller degree. On the other hand, both lithium-free (47a) and lithium-coordinated (47b) nitrogen ylides have slightly longer N-CH2(Li) and N-Me(1) bonds than the parent tetramethylammonium cation (46a).
67
Discovery, Development and Study of Carbenoid Mediated Reactions
+ Figure 3.3 Selected bond lengths and angles for a) N(CH3)4 (46a), b) N(CH3)3CH2 (47a), c) + + + N(CH3)3CH2Li (47b), d) P(CH3)4 (50a), e) P(CH3)3CH2 (51a) and f) P(CH3)3CH2Li (51b).
In summary, the removal of the lithium cation from 47b causes a localized electron density to accumulate at the methylene carbon of 47a, whereas in the phosphorus case, after removal of lithium from 51b the excess electron density is delocalized over the rest of the molecule in 51a. This redistribution is possible due to 1) an enhanced p,*(P-C) negative hyperconjugation; 2) enhanced electron dispersion towards P due to the lower polarization of P-C bonds as compared to N-C bonds; and 3) enhanced overlap of the methylene orbitals with the diffuse 3p orbitals of phosphorus. Both the experimental and the theoretical evidence agree that lithium coordination is essential for the stabilization of lithiomethyl trimethylammonium reagent 47b.
3.2.3. Reactivity of lithiomethyl trimethylammonium triflate with aldehydes, ketones and imines
Having, for the first time, a soluble version of the simplest nitrogen ‘ylide’ at hand, our next goal was set to test its reactivity towards the most representative electrophiles, however a more convenient source of a soluble ‘ylide’ was desirable. During our studies it was discovered that other anions could also enhance the solubility of the precursor salts or the resulting lithiomethyl species in THF (figure 3.4). Even though tetramethylammonium salts of pivalate (56) and triflate (57) anions show only marginal solubility in THF, their corresponding ‘ylides’ are fully soluble. In particular, from the salts prepared, tetramethylammonium triflate (57) could be easily prepared and purified, showed no hygroscopic character and was easy to handle (free flowing solid).
68
3. Lithiomethyl trimethylammonium salts as methylene donors
Figure 3.4 Precursors of soluble nitrogen ‘ylides’ in THF: tetramethylammonium BArF (45), pivalate (56), triflate (57) and triflimide (58).
The ‘ylide’ derived by deprotonation of 57 was tested in methylenation of aldehydes, ketones, imines and epoxides. The corresponding products were obtained in reasonable to excellent yields (table 3.3)
Table 3.3 Methylenation of aldehydes, ketones, imines and epoxidesa
Entry Substrate R1 R2 Product Yieldb
1 59a Ph Ph 60a 67%
2 59b Ph nPr 60b 73%
3 59c nBu nBu 60c 84%
4 59d cPr p-(OMe)Ph 60d 68%c
5 61a Ph Ph 62a 46%
6 61b Ph tBu 62b 73%
7 63 Ph - 64 58%
8 60a Ph Ph 65 70%
a For details see experimental part. b Isolated yields after column chromatography. c Crude Yield.
69
Discovery, Development and Study of Carbenoid Mediated Reactions Using 57 bis-aryl (table 3.3, entry 1), alkyl-aryl (entries 2 and 4), and bis-alkyl (entry 3) ketones can be transformed into their corresponding epoxides efficiently. The mechanism for this methylenation most likely involves the consecutive 1,2-addition of lithiomethyl trimethylammonium triflate (66) to the ketones 59 and intramolecular cyclization with concomitant release of trymethylamine (34), see scheme 3.10.
Scheme 3.10 Proposed mechanism for the methylenation of carbonyl compounds.
The formation of aziridines from imines (entries 5 and 6) proceeds reasonably well under the same reaction conditions, however, the reaction with the less electron rich aldehyde 63 (entry 7) requires heating to force closure. Methylenation of epoxide 60a (entry 8) yields allylic alcohol 65 in reasonable yield.
In all of these cases the reactivity of the soluble nitrogen ‘ylide’ matches exactly the one of sulfur ylides in the Corey-Chaykovsky reaction. However, the nature of the soluble nitrogen ‘ylides’ was proven to resemble more closely that of an organolithium reagent than the traditional P and S ylides. Organolithium reagents are more powerful nucleophiles, being capable of addition to a wider range of substrates, among them certain olefins, e.g. styrenes. If reagent 66 could perform addition to styrenes, it would open the possibility to a very facile synthesis of cyclopropanes; relevant molecules for which an alternative method of preparation is still very desirable, vide supra.
3.2.4. Reactivity of lithiomethyltrimethylammonium triflate with styrenes
Carbolithiathion, the nucleophilic addition of organolithium reagents to styrenes, has been extensively studied since its discovery190 for diverse applications,191 including the formation of heterocycles192 and the stereoselective formation of cyclopropane derivatives.193 However, the formation of cyclopropanes via carbolithiation was limited to substrates incorporating a leaving group. Potentially, a bigger substrate scope for the cyclopropanation could be obtained when nucleophilic reagents possessing a leaving group, like the lithiomethyl reagent 66, are used.
70
3. Lithiomethyl trimethylammonium salts as methylene donors
Scheme 3.11 Top: Previously reported cyclopropanation of styrenes using carbolithiation.193 Bottom: Cyclopropanation of styrenes and stilbenes with 66.
Table 3.4 Effect of the solubilizing anion on the cyclopropanation of styrene.a
Entry Precursor Anion Yield of 71ab Remaining 69ab
1 45 -BArF 24% 76%
- 2 56 OOC(CH3)3 30% 70%
- 3 57 OSO2CF3 60% 40%
- 4 58 N(SO2CF3)2 19% 81%
a 0.3 mmol scale. b Determined by GC-FID, see experimental part for details.
Lithiated ammonium salts with all the available counterions (figure 3.4) were tested for the cyclopropanation of styrene in THF. Deprotonation of the tetramethylammonium salts at 0 °C for 30 min, followed by addition of styrene (69a), produced cyclopropylbenzene (71a) in all cases. THF-soluble salts tetramethylammonium BArF (45) and triflimide (56) afforded 70a in 24% and 19% yield respectively (Table 3.4, entries 1 and 2). Ammonium salts with triflate (57) and pivalate (58) and anions were also evaluated. Higher yields of 60% and 30% were obtained when 57 and 58 were used respectively (entries 3 and 4). The strong dependence of the reaction yield on the anion might correlate to the ability of the anion to coordinate the lithium cation, i.e. this coordination could influence the rate of ring closure from intermediate 70 (vide supra).
71
Discovery, Development and Study of Carbenoid Mediated Reactions Table 3.5 Cyclopropanation of olefins at 1.2 mmol scale with lithiomethyl trimethylammonium triflate.
Entry Substrate R1 R2 Product Yielda
1 69a Ph H 71a 71%
2 69b 4-CH3Ph H 71b 88%
3 69c 4-C(CH3)3Ph H 71c 80%
4 69d 4-OCH3Ph H 71d 88%
5 69e 3-OCH3Ph H 71e 77%
6 69f 4-FPh H 71f 17%b
c 7 69g 4-NO2Ph H 71g 0%
8 69h Ph E-Ph trans-71h 92%
9 69i Ph Z-Ph trans-71h 98%
10 69j 4-OCH3Ph E-Ph trans-71j 73%
11 39 cyclohexene 40 0%d
a Isolated yields after purification unless otherwise stated, see experimental part for details. b Estimated from GC data. c Exclusively polymeric products were obtained. d Unreacted starting material was recovered.
At a 1.2 mmol scale styrene could be methylenated by 66 to produce phenylcyclopropane 71a in 71% isolated yield (Table 3.5, entry 1). In a similar fashion, p-methyl substituted (69b) and p-tBu substituted (69c) styrenes afforded their corresponding cyclopropanes 71b and 71c in 88% and 80% yields, respectively (entries 2 and 3). Styrenes bearing methoxy substituents in the para-position (69d) as electron-donating group (= -0.268), or in the meta-position (69e) as electron-withdrawing group ( = +0.115), can be methylenated to afford cyclopropanes 71d and 71e in 88% and 77% yield (entries 4 and 5). Substitution with electron withdrawing groups in the para position leads either to a lower yield (p-F, entry
6) or to no traces of cyclopropane (p-NO2, entry 7). In both of these cases the formation of polymers was observed, for p-NO2 styrene 69g the polymer was the only product. Methylenation of stilbenes 69h, 69i and 69j afforded trans-1,2-bissubstituted
72
3. Lithiomethyl trimethylammonium salts as methylene donors cyclopropanes194 71h and 71j exclusively, in 92, 98 and 73% yields respectively (entries 8 to 10). Kinetic experiments have shown that Z-4-methoxystilbene racemises to E-4- methoxystilbene before carbolithiation occurs in the presence of 66. Finally, reaction of 66 with the aliphatic olefin cyclohexene (39), the original substrate in Franzen and Wittig’s report,169 did not give the corresponding cyclopropane; instead, unreacted starting material could be recovered after workup (entry 11). This is in agreement with the required forcing reaction conditions reported in literature for the nucleophilic addition of carbanions to nonactivated olefins.195
0.03 Cyclopropane 71c formation -2 A0=2.3(±0.1)x10 M -3 -1 0.03 k1=1.1(±0.1)x10 s
0.02
0.02
0.01 Concentration [M] 'N-C Ylide‘ 66 consumption -2 A0=2.8(±0.2)x10 M -4 -1 0.01 k1=7.5(±1.0)x10 s
0.00 0 500 1000 1500 2000 2500 3000 3500 4000 time (s)
Figure 3.5 Typical plot of the absolute concentrations of 66 (white circles) and 71c (black circles) over time. Also shown are the average fitting parameters obtained after three different measurements.
The irreversible nature of carbolithiations, and the irreversible formation of trimethylamine gas in the second step of our mechanism, allows the straightforward kinetic study of this reaction. Monitoring the reaction of 66 with excess of styrene 69c using a combination of GC and ESI-MS analysis (see experimental part) showed first order kinetics for the consumption of 66 and the formation of cyclopropane 71c. The pre- exponential factors and kinetic constants found for both curves are statistically very similar (figure 3.5). Assuming the model of two consecutive, irreversible reactions (A → B → C) a kinetic model can be deduced.196 In case the first reaction is much slower than the second, i.e. k1 << k2 (scheme 3.11, bottom), the analytic solution simplifies to: