About the Development of a Process for the On-Purpose Production of

Propene out of Ethene via a Sequence of Dimerization, Isomerization and

Cross-Metathesis

Über die Entwicklung eines Prozesses für die selektive Umsetzung von Ethen zu

Propen über eine Reaktionskaskade bestehend aus einer Dimerisierung,

Isomerisierung und Kreuz-Metathese

Der Technischen Fakultät der

Universität Erlangen-Nürnberg

zur Erlangung des Grades

DOKTOR-INGENIEUR

vorgelegt von

Dipl.-Ing. Judith Scholz

Erlangen 2013

Als Dissertation genehmigt von der Technischen Fakultät

der Universität Erlangen-Nürnberg

Tag der Einreichung: 14.09.2012

Tag der Promotion: 12.04.2013

Dekan: Prof. Dr.-Ing. habil. Marion Merklein

Berichterstatter: Prof. Dr. rer. nat. Peter Wasserscheid

Prof. Dr.-Ing. Andreas Jess

Die Ergebnisse der vorliegenden Doktorarbeit entstanden von August 2008 bis Februar 2012 am

Lehrstuhl für Chemische Reaktionstechnik der Friedrich-Alexander-Universität Erlangen-

Nürnberg.

Danksagung

Gerne denke ich an die Zeit meiner Promotion zurück. Viele Menschen haben zu dem Gelingen

meiner Arbeit beigetragen und diese Episode zu etwas Besonderem gemacht. Daher möchte ich

an dieser Stelle die Gelegenheit nutzen, mich für die tollen dreieinhalb Jahre zu bedanken.

Dabei geht mein größter Dank an meinen Doktorvater Professor Dr. Peter Wasserscheid. Trotz

teilweise herausfordernder Aufgabenstellungen habe ich aufgrund Deiner Motivationskunst nie

die Hoffnung und den Spaß an meiner Arbeit verloren. Die Möglichkeiten, die Du im Rahmen

einer Promotion bietest, sei es in Form von technischer Ausstattung, Konferenzreisen im In- und

Ausland oder auch in Form von Unternehmungen mit dem Lehrstuhl wie Skifahren,

Bogenschießen, Rafting und Wandern halte ich für außergewöhnlich. Ich möchte Dir dafür

herzlich danken.

Herrn Professor Dr. Andreas Jess danke ich für die Übernahme des Zweitgutachtens und für die

interessanten Gespräche in Pruggern und auf Konferenzreisen. Darüber hinaus geht mein Dank

an die weiteren Mitglieder des Prüfungskollegiums, Herrn Professor Dr. Wilhelm Schwieger und

Herrn Professor Dr. Jörg Libuda. Letzteren danke ich zudem für die Übernahme des Mentoring

im Rahmen der EAM Graduate School.

Für die Finanzierung von industrieller Seite möchte ich mich bei der Süd-Chemie AG bedanken.

Zudem danke ich Dr. Normen Szesni und Dr. Roman Bobka für die unkomplizierte

Zusammenarbeit, ertragreichen Diskussionen und die Bereitstellung von Trägermaterialien.

Der Deutschen Forschungsgemeinschaft (DFG), die in ihrer Exzellenzinitiative den

Exzellenzcluster „Engineering of Advanced Materials“ unterstützt, danke ich ebenfalls für die

finanzielle Unterstützung und die Möglichkeit im Rahmen eines solchen Clusters forschen zu

können. Insbesondere allen Mitgliedern der Research Area D „Catalytic Materials“ bin ich für die

gewinnbringende Zusammenarbeit dankbar. Dr. Wolfgang Hieringer danke ich für die

Durchführung der DFT-Berechnungen, welche einen großen Teil zur Aufklärung des „Metathese-

Mysteriums“ beigetragen haben. Bei Xinjiao Wang bedanke ich mich für die Synthese

verschiedenster Nickelkomplexe. Dr. Carsten Schür, Monika Schenk und Kristina Muck danke

ich herzlichst für alle organisatorischen Dinge rund ums Thema Cluster und Graduate School.

Dr. Marco Haumann danke ich für die stets offene Bürotür, die hilfreichen Tipps rund ums

Thema SILP, die interessanten Gespräche zu jeder Gelegenheit und natürlich für das

Korrekturlesen meiner Arbeit. Bei Herrn Dr. Nicola Taccardi und Markus Berger möchte ich

mich für die Synthese von Katalysatorkomplexen und Ionischen Flüssigkeiten, die Hilfestellung

zu NMR-Analysen und die Beratung in chemischen Belangen bedanken. Auch Dr. Andreas

Bösmann und Dr. Friderike Agel bin ich für ihre fachliche Unterstützung äußerst dankbar. Dr.

Peter Schulz bin ich dankbar für seine Geduld und Beratung rund ums Thema Analytik und für

das Korrekturlesen meiner Arbeit.

Für die schnelle und kompetente Hilfe bei den Auf-, Um-, und Abbauarbeiten von Anlagen

möchte ich mich herzlich bei Michael Schmacks, Achim Mahnke, Julian Karl und Gerhard

Dommer bedanken. Kalle, Hendryk und Herrn Fischer danke ich für das Lösen sämtlicher

computertechnischer Probleme. Frau Menuet und Frau Singer danke ich für die zuverlässige

Unterstützung in allen organisatorischen Belangen.

Carolin Meyer danke ich für sämtliche Korrekturlesearbeiten, für das reibungslose Teilen eines

Abzuges, für die morgendlichen Kaffeerunden und für alle Erlebnisse die wir mit und ohne

Silvaner und unseren Friends hatten. Meinen ehemaligen Arena-Kollegen Caspar, Eva, Joni,

Jakitukki, Caro, Matz, Swetlana, Kerstin, Markus und Jens danke ich herzlichst für das super

angenehme, lustige und meist konstruktive Arbeitsklima. Insgesamt möchte ich mich bei allen

ehemaligen und jetzigen Mitarbeitern des CRT für die tolle Zeit am Lehrstuhl bedanken.

Großer Dank geht zudem an Anne, Ferdinand, Clara, Steffi, Denise, Lisa, Simon und Mammut –

ohne eure Hilfe im Labor hätte ich weit weniger Ergebnisse erlangt.

Zu guter Letzt bedanke ich mich bei allen meinen Freunden und meiner Familie, die mir jederzeit

zur Seite standen und ohne die diese Arbeit niemals zustande gekommen wäre.

Teile dieser Arbeit wurden bereits in der folgenden Fachzeitschrift veröffentlicht:

J. Scholz, S. Loekman, N. Szesni, W. Hieringer, A. Görling, M. Haumann, P. Wasserscheid,

Advanced Synthesis and Catalysis 2011, 353, 2701 – 2707.

Teile dieser Arbeit wurden bereits als Tagungsbeitrag veröffentlicht:

J. Scholz, M. Haumann, P. Wasserscheid: Investigations on the deactivation behavior of Grubbs-

type olefin metathesis catalysts. Vortrag, 8th European Congress of Chemical Engineering

together with ProcessNet Annual Meeting, Berlin, Deutschland, 2011.

J. Scholz, W. Hieringer, M. Haumann, P. Wasserscheid: Investigations on the activity profile of

immobilized Grubbs metathesis catalysts in a Supported Ionic Liquid Phase (SILP) system.

Poster, 4th

Congress on Ionic Liquids, Washington DC, USA, 2011.

J. Scholz, W. Hieringer, M. Haumann, P. Wasserscheid: Reversible deactivation of Grubbs

metathesis catalysts. Poster, 22nd

North American Catalysis Society Meeting, Detroit, USA,

2011.

J. Scholz, M. Haumann, P. Wasserscheid: Ethene induced deactivation of Grubbs metathesis

catalysts. Poster, 44. Jahrestreffen Deutscher Katalytiker mit Jahrestreffen Reaktionstechnik,

Weimar, Deutschland, 2011.

J. Scholz, M. Haumann, P. Wasserscheid: Immobilization of Grubbs metathesis catalysts. Poster,

3rd

EuCheMS Chemistry Congress, Nürnberg, Deutschland, 2010.

J. Scholz, M. Haumann, P. Wasserscheid: Olefin metathesis with supported ruthenium catalysts.

Poster, EUCHEM Conference on Molten Salts and Ionic Liquids, Bamberg, Deutschland, 2010.

J. Scholz, M. Haumann, P. Wasserscheid: Olefin metathesis with supported ruthenium catalysts.

Poster, 43. Jahrestreffen Deutscher Katalytiker, Weimar, Deutschland, 2010.

J. Scholz, M. Haumann, P. Wasserscheid: Alken-Metathese mit Hilfe geträgerter Ruthenium-

Katalysatoren. Jahrestreffen Reaktionstechnik, Würzburg, Deutschland, 2010.

J. Scholz, M. Haumann, P. Wasserscheid: Fast Optimization and Kinetic Measurements of

Supported Ionic Liquid Phase (SILP) Catalysts in a Parallel Reactor Set-Up. ISHHC XIV

International Symposium on Relations between Homogeneous and Heterogeneous Catalysis,

Stockholm, Sweden, 2009.

Table of contents

1 Introduction 2

2 General part 6

2.1 Propene 6

2.1.1 Significance 6

2.1.2 On-purpose routes for propene production 7

2.2 Dimerization and oligomerization of ethene 12

2.2.1 Catalytic systems 13

2.2.2 Nickel hydride mechanism 16

2.2.3 Cationic nickel complexes for selective ethene dimerization and butene

isomerization 17

2.3 Olefin metathesis 21

2.3.1 Different kinds of metathesis reactions 21

2.3.2 Catalytic systems 23

2.3.3 Metallacyclobutane mechanism 25

2.3.4 Ruthenium based Grubbs-type catalysts 27

2.3.5 Proposed deactivation mechanism of Grubbs-type catalysts 30

2.3.6 Thermodynamic aspects of olefin metathesis 35

2.4 Immobilization of homogeneous catalysts 36

2.4.1 Overview of immobilization concepts 36

2.4.2 Immobilized nickel dimerization and oligomerization catalysts 40

2.4.3 Immobilized ruthenium Grubbs-type metathesis catalysts 44

2.5 Objectives of this work 52

3 Experimental 56

3.1 General working technique 56

3.2 Chemicals 56

3.2.1 Substrates 56

3.2.2 Homogeneous catalysts 57

3.2.3 Ionic liquids 58

3.2.4 Support materials 60

3.3 SILP catalyst preparation 61

3.4 Continuous gas-phase experiments 62

3.4.1 Tenfold screening-rig 62

3.4.2 Continuous test-rig 65

3.5 DFT calculations 70

4 Results and discussion 73

4.1 Conversion of ethene to 2-butene via dimerization and isomerization 73

4.1.1 Reproduction of catalytic performance with literature-known SILP system 73

4.1.2 Optimization of the composition of the SILP system 76

4.1.3 Investigations on the possible reasons for deactivation 87

4.1.4 Approaches for catalyst stabilization 97

4.1.5 Optimized SILP system 113

4.2 Cross-Metathesis 116

4.2.1 Summary of all beneficial trends reported in literature for immobilized

ruthenium metathesis catalysts 116

4.2.2 Systematic investigations on the influence of ionic liquid on catalyst stability 118

4.2.3 Influence of ethene on catalyst stability 122

5 Summary / Zusammenfassung 140

5.1 Summary / Abstract 140

5.2 Zusammenfassung / Kurzfassung 145

6 Appendix 153

6.1 Detailed synthesis conditions and NMR-characterization of ionic liquids 153

6.2 Flow sheets of rigs for continuous gas-phase experiments 156

6.3 Further experimental results 158

6.4 Calculations 159

6.5 List of abbreviations and symbols 160

6.6 List of Schemes 165

6.7 List of Figures 167

6.8 List of Tables 171

7 References 174

Chapter 1

INTRODUCTION

2 Introduction

1 Introduction

As a consequence of a rapidly developing polypropylene market, the demand of propene has

continuously increased within the last years with a forecast of even further growth in the future.

Since in conventional production processes, such as steam cracking or fluid catalytic cracking,

propene is only recovered as a by-product besides the actual primary product ethene, these routes

are expected to be unable to meet future market requirements anymore with regard to sufficient

and economical propene supply.[1]

In seeking to establish production routes that are unbound to

the demand of ethene, so-called on-purpose processes for the selective recovery of propene gain

increasingly in importance. Even though well-known processes, like propane dehydrogenation or

cross-metathesis of ethene and 2-butene, fulfill this prerequisite, they are offset by restrictedly

available and costly feedstock including propane and 2-butene respectively.[2]

Boosting the

propene output of a steam cracker by converting ethene into propene represents an attractive

option to circumvent these aforementioned drawbacks. Thereby, the supply of propene can easily

be adapted to the market requirements without the need for further substrates. However, currently

known catalytic systems for a direct conversion are limited by low activity, moderate propene

selectivity and/or severe deactivation issues.[3-11]

In contrast, a successive transformation of

ethene to propene via a cascade of reactions, including dimerization, isomerization as well as

cross-metathesis, in which each step is catalyzed by a particularly optimized system, enables the

realization of a process with highest activity and selectivity.

State of knowledge in the field of ethene oligomerization allows the breakdown of this sequential

conversion to only two required process steps, since appropriate catalytic systems are not only

able to perform selective dimerization of ethene to 1-butene, but also provide highly inherent

isomerization activity and thus render an immediate, subsequent reaction from 1-butene to 2-

butene possible. Among numerous well-known oligomerization systems, especially

homogeneous, cationic nickel complexes with a P^O-chelating ligand turned out to be excellent

catalysts for selective dimerization of ethene and subsequent isomerization to 2-butene.[12]

This

characteristic behavior is ascribed to their positive charge resulting in an enhanced

electrophilicity along with an increased affinity toward olefins, so that the essential re-insertion

of 1-butene for the isomerization to 2-butene is facilitated.

Introduction 3

In the second part of the reaction, non-converted ethene reacts with the as-produced 2-butene in a

cross-metathesis under the formation of propene. A wide array of catalytic systems with cross-

metathetical ability is known, including heterogeneous systems based on tungsten, rhenium or

molybdenum as well as homogeneous transition metal catalysts composing of titanium, tantalum,

tungsten, molybdenum or ruthenium. However, due to their structural robustness combined with

a pronounced functional group tolerance, and their high activity under very mild reaction

conditions, ruthenium-based Grubbs-type catalysts prevailed over plenty of other metathesis

catalyst, especially in the academic research area. Nonetheless, the application of Grubbs

catalysts on the commodity industrial scale has not been demonstrated because of their lack of

stability in continuously operated processes. Quite a few possibilities being responsible for this

instable behavior are discussed in literature. Proposed deactivation pathways vary with the kind

of applied catalyst, the used substrates and the adjusted reaction conditions. In this context, the

detrimental effect of ethene on Grubbs metathesis catalysts is mentioned several times though has

not been explicitly evidenced yet.[13-15]

For employing homogeneous catalysts in continuously operated gas-phase processes, such as the

cascade reaction of ethene to propene, an effective immobilization strategy is required to

overcome substantial drawbacks of transition metal catalysis in terms of challenging product

separation and catalyst recycle. In this regard, the immobilization via the so-called Supported

Ionic Liquid Phase (SILP) concept emerged as a highly efficient technique to heterogenize

homogeneous catalysts while maintaining their attractive features like high activity and

selectivity obtained under very mild reaction conditions.[16-17]

In SILP systems, a thin film of an

ionic liquid (IL), containing a therein dissolved molecularly defined catalyst, is dispersed on the

surface of a highly porous support material. Thus, the as-obtained SILP catalyst appears

macroscopically as a solid and consequently can be handled like a heterogeneous system while

maintaining beneficial properties of the homogeneous catalyst.

The present thesis includes the design of a stepwise ethene-to-propene-conversion route by the

purposive development of catalysts for each sequential reaction. Thereby the main focus is

directed on the optimization of literature-known SILP-immobilized homogeneous catalysts

systems that are able to operate continuously under very mild reaction conditions. For the first

part of the reaction, the selective conversion of ethene to 2-butene via an ethene dimerization and

1-butene isomerization, cationic nickel complexes, that were already successfully applied by

4 Introduction

Melcher[12]

, are subject to scrutiny. Due to their lack of stability during continuous operation,

predominantly the elucidation of the cause of deactivation and its circumvention are matter of

examination. In the case of cross-metathesis of ethene and 2-butene, Loekman’s results[15]

,

obtained from extensive studies in the field of SILP-catalyzed propene cross-metathesis, are

taken as basis. Since Loekman also struggled with repetitively occurring deactivation problems of

Grubbs catalysts in a continuous operation mode, clarification of the reason for this activity loss

and the role of ethene therein cover the main objective of this part of the work.

Chapter 2

GENERAL PART

6 General part

2 General part

2.1 Propene

The development of a continuous process for the production of propene out of ethene via a

cascade reaction represented the objective target of this work. In this section, the relevance of

propene as intermediate on the commodity scale is highlighted. Existing processes for propene

production are described, including both, conventional routes and selective processes. By

depicting the major drawbacks of these production routes, the necessity for the development of

alternative technologies based on generally available feedstock is pointed out.

2.1.1 Significance

Propene is the second most important commodity in the chemical industry after ethene. In the

year 2010, both, the global production and consumption accounted for 77 million metric tons.

The demand of propene is expected to grow even more with an average of around 5.0 % per year

from 2010 to 2015 followed by a slower annual growth of 3.7 % from 2015 to 2020. Mainly, this

increasing demand can be assigned to a rapidly developing polypropylene market, adding up to

more than 60 % of the total world consumption of propene.[1]

Aside, further product spectrum

based on propene ranges from acrylonitrile, propylene oxide, cumene, oxo alcohols to

isopropanol. Despite its important role in the chemical industry, propene is still recovered almost

entirely as a byproduct besides the main product ethene or gasoline in steam cracking or

petroleum refinery units, respectively. In other words, the availability of propene is determined

primarily by the demand of other products. Since the need for ethene is forecasted to stagnate or

even to decline, an imbalance between supply and demand of propene is suspected.[2]

To

guarantee an ongoing sufficient propene supply, different approaches are pursued for obtaining

processes with a higher propene yield. On the one hand, efforts are made to improve propene

selectivity in conventional processes. In steam cracking plants using liquid feedstock, for

example, a trend toward less severe cracking conditions has been observed which results in a

boosted propene yield.[18]

Nevertheless, since ethene remains the primary product, so-called on-

purpose routes for the selective production of propene gain in importance making propene

production independent of the demand for gasoline and ethylene, on the other hand. Up to now,

only 6 – 8 % of the manufactured propene worldwide is an on-purpose product mainly stemming

General part 7

from processes such as propane dehydrogenation or olefin metathesis reaction. In the following

chapter, the most prevalent on-purpose routes for propene production are described in detail.

2.1.2 On-purpose routes for propene production

2.1.2.1 Propane dehydrogenation (PDH)

The majority of on-purpose produced propene is obtained by propane dehydrogenation (PDH).

Since PDH is an endothermic equilibrium reaction, maximum yields of propene are achieved at

higher reaction temperature and lower pressure. However, elevated temperature causes cracking

of propane and consequently coke formation on the catalyst. To reach a maximum of propene

selectivity, commercially available PDH technologies are based on various approaches that differ

in operation mode, dehydrogenation catalyst and methods of catalyst regeneration. The UOP

OleflexTM

technology is a continuous adiabatic process, operated at a slight positive pressure,

which is separated into three different sections: reactor section, product recovery, catalyst

regeneration. The catalyst, based on platinum, circulates through the reactor section, consisting of

four radial-flow reactors. In a separate regeneration vessel coke is removed from the surface of

the catalyst by combustion in air before the catalyst is returned to the first dehydrogenation

reactor. The overall selectivity for propene is claimed to be 89 – 91 %. Similar selectivities can

be reached in the Catofin process provided by Lummus Technology. Here, PDH is conducted in

an adiabatic mode, operated under a slight vacuum, with a catalyst, consisting of activated

alumina pellets impregnated with chromium. The continuous process is characterized by a cyclic

reactor operation in which multiple reactors go through a controlled sequence of reaction and in

situ catalyst regeneration. In contrast, to shift the equilibrium of the PDH reaction on the product

side, ThyssenKrupp Uhde applies a significantly different concept in the Uhde STAR (steam

active reforming) process. By adding oxygen to the system, H2O is formed, which in turn keeps

the overall pressure positive and simultaneously reduces the partial pressure of propene and

hydrogen. Thus, the equilibrium is shifted toward increased conversion reaching a propene yield

of 80 %. In addition, the process is run under isothermal conditions. The STAR catalyst is based

on a zinc and calcium aluminate support that is impregnated with various metals and exhibits

good stability toward steam and oxygen due to its basic nature. Deactivation of the catalyst

occurs by coke deposition and results in an offline-catalyst regeneration by combustion.[2]

8 General part

Although all presented dehydrogenation technologies show high propene yields up to 91 %, the

main criticism of this route is the need for a long-term and low-cost supply of propane. If this is

not the case, these energy-intensive processes are not economically viable, due to the relatively

high capital costs.

2.1.2.2 Cross-metathesis of ethene and 2-butene

Much effort has been made to increase the propene output from steam crackers or fluid catalytic

cracking units. Here, the integration of the cross-metathesis of ethene and 2-butene offers an

option to boost the propene yield. ABB Lummus provides the Olefin Conversion Technology

(OCT) which was originally developed by Phillips Petroleum Company in 1972 for the

conversion of propene into ethene and 2-butene due to a different situation on the olefin market at

that time. In the OCT process, a variety of mixed C4-feeds can be utilized since the metathesis is

preceded by an isomerization reaction. A MgO catalyst converts 1-butene to 2-butene which is

then consumed in the following cross-metathesis catalyzed by WO3/SiO2 at T > 260 °C and 30-

35 bar in a fixed-bed reactor. Butanes, which can be present in the applied feedstock, pass

through the system as inerts. Since the product yield of a metathesis reaction is limited by the

position of the thermodynamic equilibrium, a recycle of unconverted ethene is necessary to

increase the overall conversion. During the process, a small amount of coke is formed, so that the

beds are periodically regenerated using nitrogen-diluted air. When integrated in a steam cracker,

OCT can increase the propene-to-ethene ratio above one. The conversion of ethene is around

60 % per pass with a propene selectivity of over 95 %. The Institut Francais du Pétrole together

with the Chinese Petroleum Corporation have developed a process for the selective production of

propene, called Meta-4-process. In contrast to OCT, the reaction is carried out in the liquid phase

at 35 °C and 60 bar using a Re2O7/Al2O3 catalyst. A 2-butene conversion of 63 % is reported

which corresponds to equilibrium conversion with a propene selectivity of over 98 %. Until now,

the process has been piloted but not yet commercialized, mainly due to the cost of the catalyst

and the requirement of a high purity feed.[19]

The metathesis reaction represents an attractive option for increasing the propene output in a

steam cracker. However, similar to PDH, metathesis units require large C4-streams which are not

necessarily present in a steam cracker. Therefore, a process which converts ethene directly to

propene without the addition of other hydrocarbons is highly desirable. These so-called ethene-

to-propene (ETP) processes are introduced in chapter 2.1.2.3.

General part 9

2.1.2.3 Ethene-to-propene (ETP) process

The advantages of a direct conversion of ethene to propene are, on the one hand, the

independence of additional hydrocarbon feeds, and, on the other hand, the flexibility to adapt the

propene and ethene output to the respective market when integrated in a steam cracking plant.

There are no processes commercialized up to now, mainly because of the lack of efficient catalyst

systems. However, high research activity based on diverse approaches can be observed on this

topic. The investigations differ not only in the kind of applied catalyst but also in the kind of

reaction mechanism causing the desired conversion. The group of Baba[3, 20]

reported on zeolitic

materials that catalyze the reaction via a carbenium-ion mechanism (see Scheme 1).

Scheme 1: Reaction scheme of the conversion of ethene to propene via a carbenium mechanism.[3]

The reaction includes oligomerization/polymerization of lower olefins and subsequent

decomposition to yield propene provoked by the shape selectivity of the zeolite pores. According

to this mechanism, the fission of a carbon-carbon bond in the -position of the hexyl carbenium

ion (2) and/or the 4-methyl-2-pentyl carbenium ion (4) plays a decisive role. In the latter reaction,

for the formation of (4), the tertiary carbenium ion of (3) is converted to the secondary carbenium

ion of (4). The authors presumed that the formation of (2) via (1) proceeded more easily than (4)

via (3). The performance of the zeolites was mostly determined by the acid strength of the acidic

protons and the pore size of the catalyst. The silicoaluminophosphate molecular sieve SAPO-34

showed the best results in the conversion of ethene to propene. With an intermediate acid strength

and a pore size comparable to the kinetic parameter of propene, it afforded propene with a

10 General part

selectivity of 73.3 % and 71.2 % ethene conversion at 450 °C. Li et al.[4]

confirmed the

effectiveness of SAPO-34 in the conversion of ethene to propene, but only obtained lower

activities under similar reaction conditions. Moreover, they observed a rapid deactivation of

SAPO-34 with prolonging time-on-stream. Lin and co-workers[21]

tested eleven kinds of

molecular sieves for the direct conversion of ethene to propene. Among the applied catalysts, H-

ZSM-5 exhibited the highest activity with an ethene conversion of 58 % and a propene selectivity

of 42 % at 450 °C. Exchanging H+ and varying the Si/Al ratio of the zeolite revealed the

necessity of Brønsted acid sites for obtaining catalytic activity. In situ Fourier Transformed

Infrared (FT-IR) spectroscopic studies proved that C2H4 molecules mainly react with the

Brønsted acid sites on H-ZSM-5 surfaces and undergo oligomerization. Furthermore, the authors

assumed that the oligomeric species were cracked subsequently and resulted in the formation of

C3H6.

Independent of shape selectivity are multifunctional catalysts which perform the conversion of

ethene to propene via a sequence of dimerization, isomerization and metathesis reactions.

Catalytic activity for this reaction sequence has been reported on supported molybdenum[5-6]

and

tungsten oxides[7]

, but the conversion was that low as to be observed only in a closed

recirculation system. Iwamoto and co-workers[8-10, 22]

prepared a catalyst consisting of nickel ions

loaded on the mesoporous acidic silica MCM-41 by template-ion exchange method[23]

and tested

it in a continuous flow system at 400 °C and atmospheric pressure for ethene conversion in the

presence of water. An ethene conversion of 55 % with selectivities to propene and butenes of

54 % and 35 %, respectively, could be achieved. No long-term stability studies were conducted.

Due to a known instability of MCM-41 in steam[24]

, Lin et al.[21]

questioned the long-term

stability of this catalytic systems under the required reaction conditions. For the conversion of

ethene to propene with Ni-MCM-41, the authors suggested a mechanism as depicted in Figure 1.

According to this mechanism, the dimerization of ethene to 1-butene is catalyzed by nickel ions.

The catalytic activity of nickel ions for dimerization is known for more than 50 years and widely

studied.[25]

The resulting 1-butene is then isomerized to 2-butene at the acid sites of Ni-MCM-41.

The conversion of 1-butene to 2-butene is a typical acid-catalyzed reaction, and was confirmed

on MCM-41 before.[26]

Eventually, 2-butene reacts with unconverted ethene in a cross-metathesis

reaction to propene.

General part 11

Figure 1: Proposed reaction mechanism for the conversion of ethene to propene on nickel ion

loaded on MCM-41.[8]

The metathesis activity is attributed to the nickel ions, although, until then, no reports explicitly

claimed nickel containing catalysts as catalytically active in metathesis reactions. The assumption

of metathesis activity of nickel ions was referred to observations made in the retro-metathesis

reaction, namely the cross-metathesis of propene. When propene was contacted with Ni-MCM-

41, equimolar amounts of ethene and 2-butenes were formed, whereas formation of by-products

was insignificantly small. Consequentially, the authors concluded that, instead of an

oligomerization combined with homolytical decomposition, metathesis reactions were taking

place. The employment of the parent MCM-41 in this reaction did not show any catalytic activity.

The proposed mechanism was confirmed by Lehmann et al.[27]

, who also applied Ni-MCM-41,

prepared by equilibrium adsorption of different nickel precursors, in the direct conversion of

ethene to propene. The analysis of the product spectra at different temperatures, residence times

and inlet compositions revealed reaction kinetics consistent with a sequence of ethene

dimerization, 1-butene isomerization and metathesis of 2-butene and ethene. The maximal ethene

conversion of 36 % could be observed at 400 °C while propene selectivity reached 45 %.

Basset and co-workers[11]

reported on a direct ETP conversion over a tungsten hydride supported

on -alumina, namely W(H)3/Al2O3. By passing ethene over the catalyst in a continuous flow-

reactor at 150 °C, ethene conversion at the initial stage reached 40 % and decreased to 7 % after

120 hours of reaction, while propene selectivity increased rapidly up to 95 % and was then kept

12 General part

almost unchanged. A turn over number (TON) of 1,120 was achieved after 120 hours. Initiation

of the reaction occurred by addition of ethene under the formation of a surface tungsten-ethyl-

ethylidene species [W](CH2CH3)(=CHCH3).

Scheme 2: Proposed mechanism for the direct conversion of ethene into propene on a tungsten hydride catalyst.[11]

According to numerous examples, olefin metathesis is expected to be catalyzed by this tungsten-

ethyl-ethylidene species (see Scheme 2 a).[28]

The same tungsten species is also involved in the

ethene dimerization[29]

corresponding to the classical Cossee-Arlmann mechanism[30-31]

which

proceeds via a double ethene insertion into the W-H bond of the tungsten ethylidene hydride (see

Scheme 2 b). The latter species also catalyzes the isomerization of 1-butene into 2-butene (see

Scheme 2 c). Since all reactions take place at the same site, the catalyst behaves like a

trifunctional single-site catalyst.

All presented catalyst systems for the direct conversion of ethene to propene are associated with

different kind of drawbacks, such as severe deactivation, low conversion and/or moderate

propene selectivities, making a commercialization in close future rather improbable. In contrast

to an ETP process catalyzed with a multifunctional catalyst, an ETP route consisting of a

sequence of dimerization, isomerization and metathesis, in which each reaction is catalyzed by a

particularly optimized catalyst system is conceivable. Thus, a process with highest activity and

selectivity toward propene can be realized.

2.2 Dimerization and oligomerization of ethene

This work focuses on the realization of a cascade reaction for the conversion of ethene to propene

via a dimerization, isomerization and a metathesis step. In the ideal case, a selective dimerization

General part 13

system with isomerization activity represents the first part of this reaction sequence. In the

following chapter, first, general ethene oligomerization systems are introduced. Based on these

systems, the relevant mechanistic and steric properties that are necessary for obtaining catalysts

for the selective conversion of ethene to 2-butene are then explained and exemplified by

respective catalysts.

2.2.1 Catalytic systems

Because of the great significance of linear -olefins, the research activity concerning the

development of new catalysts for ethene oligomerization is intensive. As a result, numerous

catalytic systems of high-performance oligomerization catalysts, based on a wide array of early

and late transition metals, have been reported.[32-38]

For industrial oligomerization processes,

especially nickel-based complexes are commonly used as active and selective homogeneous

catalysts. To take advantage of an easy separation of catalyst and oligomers, considerable

research effort has also been directed toward the development of heterogeneous oligomerization

catalysts. Efficient oligomerization of C3-C4 olefins can be realized with solid acid catalysts as

zeolites (MOGD process)[39]

and phosphoric acid impregnated on kieselguhr (Captoly process)[40]

via a mechanism involving carbenium ions. However, ethene reactivity in acid-catalyzed

oligomerization is very low, consequently, catalysts containing transition metals such as nickel,

rhodium and palladium, are required for this reaction. For nickel-based systems, representing the

most investigated catalysts due to their elaborated activity, it is generally accepted that low valent

nickel ions represent the active centers in the oligomerization and this activity is positively

influenced by the acid sites.[41-47]

Different heterogeneous oligomerization catalysts have been

reported comprising nickel oxides on silica[48-49]

, silica-alumina[50]

, titanium dioxide[51]

or

zirconium dioxide[52]

modified with sulfate or tungstate ions, as well as nickel containing zeolites

which cover the major part of all published heterogeneous systems[41, 53-54]

. A major drawback of

most of these systems is a severe deactivation during catalysis caused by intracrystalline diffusion

limitations which results in a rapid enrichment of polymeric waxes in the micropores. An

approach to circumvent this blocking process is the employment of materials with larger pores,

such as nickel on sulfated alumina[55]

, nickel on amorphous silica-alumina[56]

and nickel-

exchanged mesostructured silica alumina[45, 57]

. The latter systems showed activities up to

63.2 g gcat-1

h-1

at 150 °C and thus, exceeded activities of previously reported Ni-exchanged

amorphous silica-alumina systems. However, no information about long-term stabilities of the

14 General part

investigated catalysts was given.[56]

In contrast, oligomerization experiments with nickel

exchanged amorphous silica-alumina conducted in a slurry reactor in a continuous mode revealed

indeed a long-term stability of more than 900 hours. Conversions of 90 % were maintained

throughout the catalytic run and oligomers of up to C16 were obtained.[56]

Even though great progresses have been attained in the field of heterogeneously catalyzed

oligomerization reaction, heterogeneous systems still cannot compete with homogenous catalysts

in matters of activity and tunable selectivity. Hence, current industrial processes mainly employ

homogeneous catalysts consisting of either alkylaluminum compounds[58-59]

or a combination of

alkylaluminum compounds and transition metal complexes[60-61]

. With an appropriate choice of

metal-ligand and, if necessary, metal-activator combination, the desired fraction of oligomers can

be obtained selectively under mild reaction conditions.

Homogeneous oligomerization catalysts can be divided basically into multi-component systems

(Ziegler-type catalysts) and defined single-component systems. Ziegler-type catalysts are

prepared from a transition metal complex which is activated by a co-catalyst, usually an

aluminum compound. Depending on whether a reduction agent or an oxidation agent is used for

the generation of the active compound, the preparation route is called a reduction or an oxidation

route, respectively. In both cases the formation of the active complex takes place in situ. For

financial reasons, the reduction route is commonly preferred in industry. The most prominent

process based on the reduction route is the Shell Higher Olefin Process (SHOP) for the

production of -olefins.[62-64]

Herein, the synthesis of the catalyst occurs in situ out of a cheap

divalent nickel salt, e.g. NiCl2, a reduction agent, e.g. NaBH4 and a bidentate P^O chelating

ligand. Single-component systems are isolated complexes with a defined composition. In contrast

to multi-component systems, they are catalytically active without the addition of a co-catalyst,

whereas the applied complexes not necessarily represent the active species. These precursor

complexes can be easily characterized and reproduced.

Furthermore, oligomerization systems can be classified according to their employed transition

metal and ligands. Transition metals used for oligomerization reactions include nickel, palladium,

cobalt, vanadium, rhodium, titanium and zirconium, among others. Concerning the differentiation

by ligands, complexes containing monodentate as well as bidentate ligands have been reported

for ethene oligomerization. Monodentate ligands possess only one bond to the central metal atom,

while bidentate ligands coordinate with two atoms to the metal. Bidentate ligands form circular

General part 15

structures together with the metal center, which are denoted as chelate complexes. In contrast to

monodentate ligands, such chelate systems show higher stabilities. In the oligomerization with

nickel catalysts, a wide array of different chelating ligands have been employed, such as P^P,

P^O, O^O, N^O and S^O ligands.

For the application of an immobilized selective dimerization catalyst in a continuous gas-phase

process, as relevant for this work, single-component systems are preferred, since the addition of

an activator is associated with drawbacks related to handling issues. Single-component systems

which have been successfully applied in oligomerization reactions are mainly based on nickel

with P^O ligands.[65-77]

Generally, these complexes consist of a chelate part (three electron

chelate ligand), an organo part and nickel as central atom (see Figure 2).[78]

Figure 2: General structure of a precursor

complex.[79]

The role of the organo part comprises the stabilization of the catalyst complex before the reaction

and the charge equalization[80]

, whereas the chelate part, consisting of a donor (D) and an

acceptor (A) center, significantly influences the catalytic activity and selectivity. All active

systems with a three electron chelate ligand have a square-planar complex structure which is

regarded as essential for catalytic activity. Moreover, for obtaining catalytic activity, one ligand

or even both ligands of the organo part must be able to dissociate easily from the complex to

provide a coordination site for the substrate molecule.[81]

Single-component systems, can be

differentiated in neutral and cationic complexes. Because of their high dimerization as well as

isomerization activity under very mild reaction conditions observed in continuously operated gas-

phase reactions[12]

, the structural aspects of cationic nickel complexes will be discussed in detail

in chapter 2.2.3.

16 General part

2.2.2 Nickel hydride mechanism

The generally accepted oligomerization mechanism for single-component systems based on

nickel with a P^O chelating ligand follows a step growth process starting from a coordinatively

unsaturated nickel-hydride species that results from the precursor complex via dissociation of a

ligand of the organo part (see Scheme 3).[82]

Scheme 3: Postulated nickel-hydride mechanism for the oligomerization of ethene catalyzed by

a P^O chelated nickel-hydride species.[79]

The so-obtained nickel-hydride complex is assumed to be the actual active species. After -

coordination of ethene, a migratory insertion in the Ni-H-bond takes place under formation of a

nickel-alkyl species. During the insertion of the olefin in the Ni-H-bond, again, a free

coordination site is created, which can be occupied by another olefin. That way, further olefins

can be integrated in the alkyl chain resulting in chain growth. It is generally believed that chain

termination occurs via -hydride-elimination regenerating the nickel-hydride species. From such

a chain-growth mechanism, olefins are obtained in a Schulz-Flory distribution[83-84]

, which can be

described by the -value. The -value is defined as the ratio of termination rate to propagation

General part 17

rate and depends on numerous factors, such as electronic and steric properties of the ligand,

reaction temperature and substrate concentration. During the catalytic cycle, a permanent

coordination of the chelate ligand is necessary to achieve selective formation of the respective

olefins.[80]

Until today, it is not fully elucidated whether the active species is a nickel-hydride or a

nickel-alkyl complex. In situ-NMR-measurements revealed the existence of nickel-hydrides.[82]

Whereas in case of high concentrations and pressures of alkenes the presence of a metal-alkyl

species seems to be more probable according to DFT calculations.[85]

Thus, instead of a -

hydride-elimination for the regeneration of the metal-hydride, further alkyl coordination occurs in

that case, followed by a hydride transfer.

Particularly with regard to the objective of this work, the mechanism of the subsequent

isomerization of 1-butene to yield 2-butene is depicted in Scheme 4.

Scheme 4: Postulated isomerization reaction of 1-butene to 2-butene by a nickel-hydride species.[86]

For the formation of 2-butene, a reinsertion of the previously formed 1-butene is necessary. The

generated secondary alkyl species results in 2-butene via a -elimination.[86]

Consequently, in a

catalytic system, that produces selectively 2-butene, the elimination of the dimer must be

essentially faster than the insertion of another ethene molecule. Hence, the -value must be very

high. Moreover, the catalyst must show the ability to reinsert the formed 1-butene in high

amounts to obtain 2-butene.

2.2.3 Cationic nickel complexes for selective ethene dimerization and butene isomerization

In contrast to neutral nickel oligomerization catalysts, cationic nickel complexes possess

decisively different catalytic properties concerning the resulting product distributions in ethene

oligomerization reactions. Whereas neutral complexes usually produce higher olefins in a relative

wide Schulz-Flory distribution, short olefins are mainly obtained with cationic catalysts.[87-88]

Responsible for this deviating behavior is the positive charge which results in an increased

electrophilicity toward olefins, so that the coordination of -donor and substrate is facilitated.

18 General part

This model of coordination can be described according to Chatt-Dewar-Duncanson (see Figure

3).[89]

Figure 3: Schematic representation of relevant orbital interactions during

the coordination of an olefin to a transition metal.[89]

The coordination of the olefin takes place via the -orbital of the olefin which is occupied with

two electrons. This orbital overlaps with the dx2-y2-orbital of the metal fragment, while the

electron density of the olefin passes over to the metal. At the same time, the electron density of

the dxy-orbital of the metal is changed over from the -bond to the *-bond of the olefin. Through

the interactions of both orbitals, the C=C-bond is destabilized, resulting in the activation of the

olefin. Due to the reduced electron density at the metal center, the electron affinity and

simultaneously the affinity to eliminate -positioned hydrogen from a growing alkyl chain as

hydride increases. Hence, the -elimination is energetically favored over chain growth, thus, the

degree of oligomerization is reduced. However, the high electrophilicity of cationic nickel

complexes not only results in the coordination of ethene molecules but in the coordination of

formed product molecules, as well. By coordination of these higher olefins, an isomerization to

internal olefins can occur which is a prerequisite for the selective formation of 2-butene.

This reduced chemoselectivity for ethene, especially for cationic nickel complexes with

monodentate and weakly coordinating ligands, was confirmed by Gehrke et al.[90]

, who observed

a high amount of 2-butenes in the oligomerization of ethene with [(3-allyl)NiL2][PF6]

(L=P(OR)3, ½ cod or SbPh3).

General part 19

The recoordination of a formed product molecule may not only result in an isomerization of that

olefin, but also may lead to an insertion of that higher olefin in growing chains. By that, the

branching degree of the product olefins increases, whereas the linearity decreases. Consequently,

for the selective production of 2-butenes, a compromise between high isomerization activity and

low insertion in the growing alkyl chain of the product olefin must be achieved. On the one hand,

process engineering approaches can be applied, such as the limitation of ethene conversion or the

elevation of ethene partial pressure. On the other hand, the modification of the catalyst by ligand

tailoring offers another possibility to control the selectivity. Strong donor ligands enhance the

electron density at the metal, so that the affinity to bind higher olefins is reduced. However, to

obtain an adequate isomerization activity for internal olefins, particularly 2-butene, 1-butene must

be coordinated. By the introduction of a sterical demanding ligand, skeletal isomerization can be

avoided whereas isomerization to internal olefins is preserved. Multidentate ligands, such as P^O

ligands, provoke a modification in the sterical and electronic properties compared to monodentate

ligands. By tuning their bulky structure and their increased rigidity, the isomerization activity of

the catalyst can be adjusted not only by electronic but also by sterical factors.[89]

Matt et al.[77]

reported firstly on the synthesis of cationic organyl nickel complexes containing

single chelating P^O and P^N ligands. Starting with neutral, pentaphenyl substituted

cyclopentadienyl nickel complexes, the respective cationic complex was achieved by protonation.

However, no catalytic results were presented. Whereas the preparation of two other complexes,

namely [methyl-2-(diphenylphosphino)-benzoate-2P,O](

3-methallyl)nickel(II) tetrafluoro-

borate and [methyl-2-(diphenylphosphino)-nicotinate-2P,O](

-methallyl)nickel(II) tetra-

fluoroborate, resulted in catalytically active catalysts. During the oligomerization of ethene, a

mixture of linear and branched olefins was obtained with reasonable activity under milder

conditions in comparison to their neutral analogues.[76, 91]

Ecke[80]

and Schulz[92]

synthesized a variety of further cationic nickel complexes with P^O-

chelating ligands. The investigated methallyl nickel complexes, predominantly with

phosphinocarboxylates as ligands, displayed high activities already at mild temperatures and low

ethene pressures. A comparison between catalytic results of neutral and analogous cationic

complexes showed that the latter exhibited a dramatic increase in activity with a lower selectivity

to linear and higher -olefins. This behavior was ascribed to a loss of chelate coordination of the

20 General part

P^O ligand under catalytic conditions, which is necessary for a high selectivity toward linear -

olefins (see Scheme 5).

Scheme 5: Concept of hemilability with P^O-chelating ligands.[79]

Instead, due to the hemilability, the functionalized phosphine acted like a monodentate phosphine

ligand. Hemilability of the bidentate ligand under catalytic conditions was also postulated for

further cationic methallyl nickel complexes, investigated by Lankes[93]

and Ecke[80]

, because of

the low selectivity to linear -olefins.

In search of defined cationic nickel complexes with selectivities toward higher linear olefins,

Brassat prepared methallyl nickel complexes with alkylene bridged biphosphine monoxides (see

Scheme 6).[73, 79]

From the spectroscopic data of the complexes, a strong oxygen-nickel

interaction could be deduced, and a chelate coordination of the ligand in methylene chloride

solution in the absence of further donor ligands was derived. Brassat assumed that the absence of

hemilability resulted in the obtained impressive selectivities toward linear olefins in ethene

oligomerization experiments. Nevertheless, the low -selectivity of the complexes confirmed an

efficient isomerization of -olefins to internal olefins. The catalytic results could be rationalized

by the assumption that the chelate coordination remained intact throughout the course of

catalysis. Thus, ethene did not replace the phosphoryl group but it was coordinating and inserting

into the organyl group without the cleavage of the nickel-oxygen bond (see Scheme 6).

Scheme 6: Proposed coordination mode of ethene to nickel in a methallyl nickel complex with

biphosphine monoxide ligand.[73]

This assumption was supported by the fact that the prepared cationic complexes followed the

same trend as neutral complexes with anionic chelate P^O-ligands. The activity as well as the

maximum chain length of the product olefins as a parameter for oligomerization grade increased

General part 21

with decreasing chelate ring size of the complex. Based on the theory that only square-planar

nickel-hydride complexes are active in oligomerization, the chain length of the formed olefin

depends on the chelate ring size, because larger chelate rings prefer tetrahedral configuration. The

closer the coordination geometry to square-planar coordination is approached, the higher the

oligomerization grade. Aside from biphosphine monoxides, Brassat employed diphenylphosphino

ether as ligands in the nickel catalyzed oligomerization. The complexes showed good activities

for the formation of mainly lower olefins with a maximum carbon number of eight as products.

Similar to results with monodentate phosphine ligands, the reduced linearity within the hexene

fraction indicated the formation of higher olefins via cross linking. Brassat assumed that the

phosphino ether acted as hemilabile ligand under catalytic conditions and ethene was coordinated

via the replacement of the ether function.

2.3 Olefin metathesis

In the second part of the ETP cascade process, the produced 2-butene is converted together with

ethene in a cross-metathesis reaction to propene. In this chapter, the fundamentals of olefin

metathesis as well as appropriate catalytic systems and their mechanistic properties are described.

2.3.1 Different kinds of metathesis reactions

The etymology of the word “metathesis” comes from the Greek and means

transposition. Although the first observation concerning metathesis of propene at high

temperature was already reported in 1931, and the first catalyzed metathesis reactions were found

in the 1950’s, it was not until the year 1967, that Calderon et al.[94]

applied the name metathesis

for this reaction for the first time.[95]

To that time, it was firstly recognized that the ring-opening

metathesis polymerization (ROMP) observed by DuPont and Natta and the disproportionation of

acyclic olefins observed by Banks and Baily were similar reactions.[96]

Nowadays, olefin

metathesis is defined as the metal-catalyzed redistribution of carbon-carbon double bonds. As

shown in Scheme 7, this transformation includes a variety of reactions, which offer possible

routes to the production of unsaturated molecules that are often challenging or even impossible to

obtain by any other means.

22 General part

Scheme 7: Variety of metathesis reactions.[97]

Metathetical conversions can be categorized according to the used starting material and the

outcome of the reaction. A very popular kind of metathesis reaction among organic chemists is

the ring-closing metathesis (RCM). Herein, a diolefin is intramolecularly coupled under the

formation of large ring system and the release of the corresponding aliphatic olefin. The analog

reverse reaction to the RCM is the ring-opening metathesis (ROM) in which the position of the

thermodynamic equilibrium depends on the size and ring strain of the products.[98]

The metathesis

reaction can additionally be combined with a polymerization. Especially in the beginning of

catalyst development, the main interest was focused on this reaction. This polymerization is

possible for diunsaturated olefins (acyclic diene metathesis, ADMET) as well as for cyclic olefins

(ring-opening metathesis polymerization, ROMP) and enables the simple and selective

production of functionalized polymers.[99]

Relevant to this work is the cross-metathesis (CM) in

which two aliphatic olefins are converted to another two product olefins. This reaction is

industrially employed in several processes, like the SHOP process or the OCT process.[100]

General part 23

2.3.2 Catalytic systems

The history of olefin metathesis starts in the mid-1950s. Until the 1980s, actually all metathesis

reactions were accomplished with poorly defined, multicomponent homogeneous or

heterogeneous catalytic systems. These systems were composed of transition metal salts

combined with alkylating agents or deposited on solid supports. Some prominent examples

include WO3/SiO2, Re2O7/Al2O3, MoO3/SiO2, WCl6/Bu4Sn, and WOCl4/EtAlCl2. Because of

their relative low costs and simple preparation route, some of these systems still have an

important place in commercial applications of olefin metathesis. Currently, the largest-scale

industrial employment of this kind of metathesis catalyst is in the SHOP process (see chapter

2.2.1) for the oligomerization of ethene to give long-chain -olefins. Undesired short-chain

olefins (< C10) and higher alkenes (> C20) are converted in a cross-metathesis unit to C10 – C14

internal alkenes after an isomerization step by means of an alumina-supported molybdenum

catalyst. Some more commercialized metathesis processes comprehend the Neohexene process

(Phillips)[101]

, the OCT process (ABB Lummus) (see chapter 2.1.2.2) and the Polynorbene

process (CdF Chimie)[19]

, amongst others. Nevertheless, the utility of these catalysts is limited by

harsh reaction conditions and strong Lewis acids which are required for the activation, making

them incompatible with most functional groups. Motivated by these drawbacks, strong efforts on

the elucidation of the mechanism and development of well-defined catalysts with a high

functional group tolerance have been made and led finally to the discovery of the first single-

component homogeneous catalysts for olefin metathesis during the late 1970s and early 1980s.

Among these new complexes were pentacarbonyl(diphenylmethylene) tungsten[102]

,

bis(cyclopentadienyl) titanocyclobutanes[103]

, tris(aryloxide) tantalacyclobutanes[104]

and various

dihaloalkoxide-alkylidene complexes of tungsten[105-106]

. In comparison to heterogeneous

systems, these complexes exhibited better initiation and higher activity under milder reaction

conditions. Particularly the molybdenum imido alkylidene complex (N-2,6-Pri2-

C6H3)(OCMe(CF3)2)2Mo=CHMe2Ph belonged to the catalysts which became widely used mainly

because of its impressive activity. It allowed to react with both terminal and internal olefins and

to catalyze the ROMP of low-strain monomers, as well as the ring-closing of sterically

demanding and electron-poor substrates.[107-109]

In the course of time, eventually, transition metal

complexes based on molybdenum and ruthenium prevailed over metathesis catalysts consisting of

titanium, tungsten and tantalum. Modern homogeneous catalysts are prepared mainly on the basis

24 General part

of molybdenum (Schrock-type catalysts) and ruthenium (Grubbs-type catalysts). Schrock-

carbenes (see Scheme 8) show a very high reactivity toward olefins, however, limitations arise

due to their high oxophilicity of the metal centers rendering them extremely sensitive to oxygen

and moisture.

Scheme 8: Example of a Schrock-carbene

metathesis catalyst.[110]

Handling of these complexes requires an absolute inert atmosphere and rigorously purified, dried

and degassed solvents and reactants. Additionally, Schrock-carbenes based on early transition

metals like molybdenum are more fundamentally restricted in terms of moderate to poor

functional group tolerance diminishing the number of potential substrates.[111]

In general, the key

to improve functional group tolerance in olefin metathesis is the development of a catalyst that

reacts preferentially with olefins in the presence of heteroatomic functionalities. Studies on the

relationship of structure and reactivity of molecular defined single-component systems revealed a

more selective reactivity with olefins as the metal centers were varied from the left to right and

bottom to top on the periodic table.[112]

With titanium and tungsten farthest to the left, they are

most reactive to olefinate ketones and esters. Molybdenum based catalysts are more reactive to

olefins, although they react with aldehydes and other polar or protic groups, as well. For

ruthenium, which is located farthest to the right in the periodic table, a strong reaction preference

toward carbon-carbon double bonds over most other species results, making it exceptionally

stable toward alcohols, aldehydes, amides, carboxylic acids, water and oxygen. For this reason,

from the early 1990s, Grubbs has started to focus on the development of carbenes based on

ruthenium, the so-called Grubbs-type catalysts. Because of their structural robustness, Grubbs-

type catalysts have been exclusively used in this work.[97]

An overview of their characteristics is

given in chapter 2.3.4.

General part 25

2.3.3 Metallacyclobutane mechanism

Several mechanistic hypotheses were postulated during the early period of metathesis

exploration.[113-115]

Even though these mechanistic proposals explained the exchange process of

the reactions, they did not match the results of some metathesis experiments. In the year 1971,

Chauvin and Hérrison[116]

came up with a proposition of mechanism (see Scheme 9) that was

consistent with experimental observations and which is generally accepted until now.

Scheme 9: Mechanism of metathesis according to Chauvin.[116]

According to this mechanism, the activation of the catalyst precursor results in a metal-alkylidene

which represents the actual active species. In a subsequent [2+2]-cycloaddition, the metal-

alkylidene complex reacts with an olefin under the formation of a metallacyclobutane

intermediate. In the next step, this intermediate decomposes in a [2+2]-cycloreversion into the

first olefinic product and a new metal-alkylidene complex. The latter contains the metal with its

ligand and an alkylidene from the substrate olefin. This metal-alkylidene reacts with a new olefin

molecule, yielding another metallacyclobutane intermediate. On decomposition in forward

direction, this species results again in an olefinic product and metal-alkylidene which re-enters

the catalytic cycle. Thus, the metathesis cycle can be considered as a series of cycloaddition and

26 General part

–reversion reactions in which each step involves an alkylidene-alkene exchange. Almost all steps

during this catalytic cycle are equilibrium reactions. Hence, thermodynamic aspects of olefin

metathesis have to be considered in the determination of maximum possible conversions and are

accordingly described in chapter 2.3.6.

Experimental evidence for this mechanism was brought by Chauvin[116-117]

, Grubbs[118-119]

,

Katz[120-121]

, Schrock[122-124]

and others[125-129]

. Even though there is consensus on the principle

steps during the catalytic cycle, the detailed course of the reaction depends on the chosen catalyst.

Consequently, the mechanism of olefin metathesis by ruthenium carbene complexes has been

subjected to experimental[130-139]

and computational scrutiny[140-148]

. Fine mechanistic studies of

complexes with the general structure L2X2Ru=CHR revealed a distorted square pyramidal

geometry with the alkylidene in the axial position and the trans phosphines and halides in the

equatorial plane.[149-150]

However, there is still an ongoing debate on the nature of the

metallacyclobutane which is either an intermediate or transition state[151]

, the rate-limiting steps

which can be either phosphane dissociation, metallacyclobutane formation, or cycloreversion and

particularly on the stereochemical orientation of the ligands. By now, there is a general

agreement that the reaction initiates with the reversible dissociation of the phosphine ligand,

leading to the formation of the unsaturated 14-electron complex 2 (see Scheme 10).

Scheme 10: Initial steps of the olefin metathesis mechanism according to the dissociative pathway for ruthenium-

based catalysts.[136]

Chen and co-workers[151]

confirmed this by identification of this species by gas-phase mass

spectrometry. The olefin coordination can then occur according to two possible pathways. In one

General part 27

possible pathway A (bottom-face pathway), the phosphine dissociates, while the alkylidene

rotates in order to generate the 16-electron intermediate 3a, in which the olefin remains in cis

position to the alkylidene. The formed intermediate is subjected to metallacyclobutane formation

cis to the bound phosphine, followed by cleavage to release the metathesis products. The

alternative pathway B (side-on pathway) includes, on the contrary, phosphine dissociation and

arrangement of the olefin trans to the remaining phosphine (see Scheme 10). Beyond that, it is

also conceivable, that the binding of the olefin occurs in an unpreferential manner through a

mixture of intermediates 3a and 3b.

2.3.4 Ruthenium based Grubbs-type catalysts

Although the catalytic activity of ruthenium salts in olefin metathesis was known already in the

1960s[152-153]

, it was not until the late 1980s, that the potential of ruthenium catalysts for ROMP

applications was reexamined[154]

. When performed in organic solvents, Grubbs found out that

RuCl3(hydrate) catalyzed ROMP, but the polymerization was preceded by long initiation periods.

In contrast, when water was used as solvent, the initiation period could be reduced to less than 30

minutes. Upon screening of other ruthenium complexes, Ru(H2O)6(tos)2 (tos = p-toluene

sulfonate) exhibited even shorter initiation times, in the range of a few minutes.[155-156]

Shortly

afterwards, Grubbs could show the formation of a ruthenium alkylidene intermediate in the

course of the same reaction, which he assumed to be the active species, whereas the initiation

process remained unclear.[157-158]

First attempts to generate ruthenium alkylidene species included

the addition of ethyl diazoacetate, representing the carbene source, to various ruthenium

precursors. Though, a decisive step forward was accomplished by Grubbs, when he applied the

methodology for the synthesis of tungsten alkylidenes, in which 3,3-disubstituted cyclopropenes

are used as carbene precursor, to the synthesis of ruthenium catalyst.[157-159]

Thereby, he managed

to synthesize the first molecularly well-defined ruthenium carbene complex that promoted the

ROMP of highly strained olefins. To expand the activity to the ROMP of low-strain monomers

and the metathesis of acyclic olefins, a systematic modification of the ligand environment

followed. Surprisingly, quite contrary to early transition metal catalysts for metathesis[160-161]

, the

activity of the systems increased with larger and more basic phosphines. The complex

[RuCl2(PCy3)2(=CH-CH=CPh2)] (see Scheme 11, complex 5) catalyzes the ROMP of low-strain

olefins, such as cyclopentene, and became the first ruthenium alkylidene complex which is active

in the metathesis of acyclic olefins.[162]

Shortly afterwards, the new catalyst

28 General part

[Ru(=CHPh)Cl2(PCy3)2] (see Scheme 11, complex 6), whose structure is closely related to the

vinylidene one published before, appeared and was commercialized as the first-generation

Grubbs catalyst.[163]

In this catalyst, the bulky and strongly electron-donating PCy3 ligand

combined with the benzylidene moiety result in a complex with high stability to air and

compatibility with a large array of functional groups.[164]

The activity of Grubbs-type catalysts of the first generation strongly depends on the nature of the

X- and L-type ligands. As mentioned before, catalyst activity increases with larger and more

electron-donating phosphines (L-type ligand), whereas it decreases with larger and more

electron-donating halides (X-type ligand). This behavior can be explained with the suggested

mechanism (see chapter 2.3.3). The phosphine ligand supports the -donation to the metal center,

which promotes formation of the mono-phosphine-olefin complex by facilitating phosphine

dissociation and by stabilizing the vacant trans site in complex 3.[165-166]

Moreover, the-

donation helps to stabilize the 14-electron metallacyclobutane species. Thus, the catalytic activity

is directly linked to the electron-donating ability of the phosphine ligands. Besides, the steric bulk

of the ligand also represents an influencing factor in the dissociation of the phosphine by

destabilizing the biphosphine olefin complex. In contrary, concerning the halide ligand, the

increase in size and electron-donating ability are correlated with lower catalytic activity. Since

the incoming olefin may initially bind trans to the halide, a more electron-donating halide should

weaken the ruthenium-olefin bond and disfavor olefin coordination. The nature of the alkylidene

moiety mainly contributes to catalyst initiation and life-time. Alkyl-substituted alkylidenes

generally show more effective initiation than a methylidene complex.[167]

An even more rapid

initiation can be achieved with ester-substituted alkylidenes, but these complexes appear to be

less stable than the alkyl derivatives.[168]

Studies on the stability of various catalysts disclosed the

existence of a different decomposition pathway depending on the structure of the alkylidene

moiety. Detailed information on deactivation mechanisms is given in chapter 2.3.5.

The identification of the highly active mono(phosphine) intermediate during the catalytic cycle

led to a step-by-step improvement of catalytic systems by using this species as design motif. The

introduction of a relatively stable cyclic bis-amino carbene ligand in place of one phosphine

ligand resulted finally in the 2nd

Grubbs catalysts [RuCl2[C(N(mesityl)CH2)2](PCy3)(=CHPh)]

(see Scheme 11, complex 7).[97]

Its catalytic activity was successively proposed within a few

months by the groups of Nolan[169]

, Grubbs[170-172]

and Herrmann[173]

. Superior performances that

General part 29

previously were only possible with the most active early transition metal systems were achieved.

Grubbs-type catalysts of the second generation are able to convert low-strain monomers as well

as sterically hindered substrates containing trisubstituted olefins in a ROMP. Additionally,

sterically demanding dienes can be transformed into tri- and tetrasubstituted olefins by RCM.[170-

171] Moreover, catalyst 3 produced the first example of CM to yield a trisubstituted olefin

[174], as

well as CM and RCM reactions in which one partner is directly functionalized with a deactivating

group, such as acrylate or siloxane[175]

. The high activity of 2nd

generation Grubbs catalysts can

be attributed to the introduction of the N-heterocyclic carbene (NHC) ligand, which is

significantly larger and more electron-donating than trialkylphosphines. Presently, this now

commercially available catalyst is the most used system for efficient CM, not only due to its

superior activity but also due to its high thermal stability. Besides, among many others,

Hoveyda[176]

, Hofmann[177-179]

, Grela[180]

and Blechert[181-184]

reported on other related, very

active, stable and functional-group tolerant systems. Derived from the first and second generation

of Grubbs catalysts, Hoveyda developed complexes bearing a chelating benzylidene ether ligand

(see Scheme 11, complexes 8 and 9) which are now commercially available under the name

Hoveyda-Grubbs catalysts.

It was claimed, that the initiation step involves dissociation of the benzylidene ether and that

following the olefin metathesis reaction, the bidentate ligand returns back to the ruthenium as a

Scheme 11: Series of ruthenium-based olefin metathesis catalysts.

30 General part

benzylidene ligand. This so-called “boomerang effect” or “release-return mechanism” certainly

belongs to the most interesting properties of these ligands. Based on this effect, tagged

benzylidenes have been applied for the immobilization of metathesis catalysts. Since the tagged

ligand “returns” at the end of the reaction, the original recyclable complex can be restored.

Several strategies for the separation of the complex from the product mixture have been reported

and are presented in chapter 2.4.3. However, recent studies involving fluorescence or 19

F-NMR

investigations cast doubt on this boomerang mechanism.[185]

Variations on Hoveyda-Grubbs catalysts have been published by Grela and Blechert who

examined the influence of functional groups at the benzylidene ligand. Hoveyda-Grubbs

catalysts, which generally display comparable activities to those of 2nd

generation Grubbs

catalysts, are especially valuable for challenging cases of metathesis of polysubstituted olefins

and selective cross-metathesis in which homo-coupling needs to be avoided.[176, 186-188]

Although Grubbs-type catalysts have mainly been applied in the academic research area, olefin

metathesis with ruthenium-based catalysts has several attractive features from an industrial

perspective. The extremely high activity leads to the necessity of only small amounts of active

metal resulting in high turnover frequencies. Furthermore, they are able to operate under mild

conditions, such as low temperature and low pressure, which positively influence the production

of propene in the cross-metathesis of 2-butene and ethene with respect to the thermodynamic of

the reaction (see chapter 2.3.6).

2.3.5 Proposed deactivation mechanism of Grubbs-type catalysts

In contrast to their successful use in organic synthesis, the application of 1st and 2

nd generation

Grubbs-type catalysts on the commodity industrial scale has not been successfully demonstrated

up to now. One severe drawback of ruthenium-based metathesis catalysts being responsible for

this is their lack of stability in continuously operated processes. Evidently, the proposed

decomposition pathways vary with the kind of applied ruthenium catalyst, the used substrates and

the adjusted reaction conditions.

The development of ruthenium-based catalyst containing NHC ligands led to a considerable

improvement concerning thermal stability and functional group tolerance compared to

bis(phosphine)-based 1st generation Grubbs catalyst. Nonetheless, Grubbs and co-workers

[189]

synthesized a ruthenium-based catalyst consisting of a NHC ligand with a phenyl group as

General part 31

opposed to mesityl groups which showed a strongly reduced life-time compared to 1st generation

Grubbs catalyst. This catalyst (see Scheme 12) seems to be more vulnerable to decomposition by

C-H activation than its mesityl containing analog.

Scheme 12: Proposed decomposition route for Grubbs-type catalyst with phenyl containing NHC ligand.[189]

The proposed deactivation mechanism, as illustrated in Scheme 12, starts with the initiation step

in ruthenium-catalyzed metathesis, the phosphine dissociation. A ruthenium hydride species is

formed by the oxidative addition of an ortho C-H bond of an N-phenyl group of N,N’-

diphenylbenzimidazol-2-ylidene (biph) ligand to the ruthenium center. The resulting hydride then

inserts at the -carbon atom of the benzylidene ligand. A reductive elimination step between the

metalated phenyl carbon atom of biph and the -carbon of the benzylidene follows and

eventually, C-H insertion together with the PCy3-mediated elimination of HCl leads to the final

decomposition product. Not all of the suggested intermediate complexes could be observed by

spectroscopic methods, most probably because of their short life-times. Theoretical studies of

Mathew et al.[190]

and Poater and co-worker[191]

revealed that the proposed intermediates are

indeed energetically feasible. The flexibility of the phenyl group on the NHC ligand seems to

play the most important role in the initiation and propagation of the reaction.

During their study of diastereoselective ring rearrangement metathesis reactions, Blechert et

al.[192]

focused on the development of bulky ruthenium carbene complexes to increase the

diastereoselective interaction between the olefin moiety and the catalytically active species. By

32 General part

connecting an N-aryl substituent with the NHC through a C2 unit in a Hoveyda-Grubbs-type

catalyst of 2nd

generation, they obtained a complex with a much stronger steric influence on the

ruthenium alkylidene moiety. This steric hindrance in ortho position of the arene ligand gave rise

to intramolecular C-H insertion, leading to metathesis-inactive ruthenium complexes. The

decomposition only occurred in the presence of oxygen reflecting that handling in an inert

atmosphere is essential to remain catalytic activity.

Particularly, thermal decomposition of 1st and 2

nd generation Grubbs catalysts account for

limitation of catalyst turnover numbers, while decomposed ruthenium complexes may contribute

to detrimental side reactions such as olefin isomerization. Since ruthenium methylidenes serve as

critical intermediates in most metathesis reactions but also represent one of the least stable

isolable species, Ulmann and Grubbs[193]

investigated the thermal decomposition behavior of

these species to obtain a thorough mechanistic understanding for designing more stable catalyst

systems.

Scheme 13: Phosphine dissociation and attack mechanism according to Grubbs.[194]

From these studies it was concluded that methylidene decomposition is primarily first order,

whereas the exact nature of the inorganic decomposition products was not identified. The authors

presumed that a unimolecular decomposition pathway, involving incorporation of methylidene

hydrogens, was favoured for bisphosphine methylidene complexes, while a bimolecular

decomposition route seemed to be more probable for monophosphine methylidenes. This

bimolecular decomposition (see Scheme 13) occurs mainly by the nucleophilic attack of the

dissociated PR3 group on the methylidene carbon after initiation to form a binuclear ruthenium

hydride complex and a methylphosphonium salt.[194-195]

General part 33

All the aforementioned deactivation pathways are relevant for catalytic systems in absence of a

substrate. However, it has been demonstrated that interaction of the substrate olefin with Ru-

carbene species opens up new decomposition pathways. The group of Cole-Hamilton[196]

observed a correlation between deactivation of supported 2nd

generation Hoveyda-Grubbs-type

catalyst and the applied substrate. In case of a continuous flow cross-metathesis of methyl oleate

with 2-octene under compressed CO2, no loss of activity was detected within six hours. On the

contrary, the self-metathesis of 1-octene led to a deactivation of the catalyst within shortest time.

Generally, on the basis of other studies using a variety of different substrates, the authors

concluded that as long as no terminal double bonds are present in the reaction system, the catalyst

proceeds with a high activity and almost constant stability. This assumption is supported by

reports in which the degradation by dimerization of ruthenium methylidenes in presence of

terminal alkenes has been described before.[194, 197]

Van Rensburg et al.[13-14]

described a substrate-induced decomposition of 1st and 2

nd generation

Grubbs catalysts using ethene as model substrate. Based on theoretical and experimental findings,

they proposed a deactivation route (see Scheme 14) according to which the respective ruthenium

methylidene species decomposes in the presence of ethene via a ruthenium allyl species formed

by -hydrogen abstraction from the corresponding ruthenium cyclobutane intermediate.

Scheme 14: Substrate-induced decomposition route according to van Rensburg.[14]

A subsequent reductive elimination results then in the formation of propene as major olefinic

compound. Since van Rensburg and co-workers were not able to characterize the major

phosphine decomposition products, Grubbs et al.[194]

reexamined this reaction and found

methyltriphenylphosphonium chloride as one of the major products of degradation. From this

evidence, they concluded that the previously mentioned phosphine attack on the methylidene

carbon is also responsible for the decomposition of the methylidene complexes in presence of

ethene.

The detrimental irreversible effect of ethene on the stability of Grubbs-type catalysts has been

postulated by several other groups, as well. Lim et al.[198]

immobilized a 2nd

generation Hoveyda-

34 General part

Grubbs catalyst by covalent linking to mesoporous silica and tested the activity and recyclability

for the RCM of various dienes in a batch reactor. Conducted in different solvents, the catalyst

exhibited a sharp decrease in conversion in both, toluene and methylene chloride (DCM) with a

slower decrease of the activity in DCM. The authors assumed that the solvent effect may be

attributed to higher solubility of ethene in toluene. To check whether the product ethene was

responsible for the deactivation behavior, a circulating flow reactor was devised to facilitate the

removal of in situ generated ethene from the reactor. During the course of circulation, gaseous

ethene was expelled simultaneously by a degasser and open reservoir to minimize the

deactivation of the catalyst. The circulating flow reactor resulted in an increased TOF, while the

catalytic activity was retained over more than eleven hours showing that the formed ethene has a

decisive impact on catalyst stability. Plenio et al.[199]

synthesized mass-tagged 2nd

generation

Hoveyda-Grubbs catalysts which they applied in RCM of diethyl diallylmalonate in a continuous

membrane reactor. Reaching a maximum of approximately 37 % after about 120 min, the

conversion dropped down to a final value of about 6 % after 500 min. The reason for this shape

of the conversion curve was not clear, but the authors assumed that the increase of ethene

concentration in the reaction volume may likely be associated with the slowdown of the catalytic

conversion. Lysenko and co-workers[200]

used a continuous flow reactor to study the effect of

ethene on the stability of 1st generation Grubbs catalyst. They compared the activity of two

systems which have been pre-exposed to ethene and cis-2-butene, respectively, in the cross-

metathesis of cis-2-butene and ethene. A direct correlation between ethene pretreatment and

reduced initial activity could be observed, whereas the cis-2-butene pretreated system did not

show an activity loss at the beginning. Similar experiments have been carried out by Loekman[15]

who investigated the influence of pure ethene prior and in-between the continuous cross-

metathesis of propene with a Supported Ionic Liquid Phase (SILP) (see chapter 2.4.1.3) 2nd

generation Hoveyda-Grubbs-type catalyst. The obtained results were then compared to the

influence of nitrogen prior and in-between the reaction. With an increased ethene exposure prior

to the reaction, a greater drop of the initial conversion was detected, whereas the treatment with

nitrogen gave no change in initial activity compared to an untreated system. Moreover, the

change of propene to ethene during the reaction caused further and stronger deactivation,

indicating again a detrimental effect of ethene on the stability of Grubbs-type catalysts.

General part 35

2.3.6 Thermodynamic aspects of olefin metathesis

Since almost all steps of the metathesis cycle are reversible, the maximum possible conversion of

olefin metathesis is limited by the position of the thermodynamic equilibrium. To determine this

conversion in dependency of the reaction temperature, the calculation of the equilibrium constant

is necessary.

For the case of cross-metathesis of ethene and 2-butene yielding propene, Loekman[15]

calculated

the equilibrium constants and the corresponding maximum conversions depending on the reaction

temperature. Figure 4 clearly illustrates that for the slightly exothermic cross-metathesis of ethene

and 2-butene, the equilibrium constant decreases with increasing temperature with the course of

the equilibrium conversion following the same trend.

Figure 4: Thermodynamic equilibrium constant and maximum equilibrium conversion

related to ethene depending on the reaction temperature for the cross-metathesis of

ethene and 2-butene.[15]

Thus, to obtain a maximum possible propene yield, low temperatures are required. In contrast to