Catalysis, Volume 22 (RSC Specialist Periodical Reports)

Catalysis Volume 22 A Specialist Periodical Report Catalysis Volume 22 A Review of Recent Literature Editors James J...

Author: James J. Spiveya | Kerry M. Dooleya

243 downloads 1349 Views 9MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

Catalysis Volume 22

A Specialist Periodical Report

Catalysis Volume 22 A Review of Recent Literature Editors James J. Spivey, Louisiana State University, USA Kerry M. Dooley, Louisiana State University, USA Authors Nicolas Bion, University of Poitiers, France Cristina Della Pina, University of Milan, Italy Daniel Duprez, University of Poitiers, France Florence Epron, University of Poitiers, France Ermelinda Falletta, University of Milan, Italy Jo´zsef L. Margitfalvi, Institute of Surface Chemistry and Catalysis, Budapest, Hungary F. C. Meunier, University of Caen, France Michele Rossi, University of Milan, Italy Johannes W. Schwank, University of Michigan, USA K. Seshan, University of Twente, The Netherlands Andrew R. Tadd, University of Michigan, USA Emı´lia Ta´las, Institute of Surface Chemistry and Catalysis, Budapest, Hungary

If you buy this title on standing order, you will be given FREE access to the chapters online. Please contact [email protected] with proof of purchase to arrange access to be set up. Thank you.

ISBN 978-1-84755-951-7 DOI 10.1039/9781847559630 ISSN 0140-0568 A catalogue record for this book is available from the British Library & The Royal Society of Chemistry 2010 All rights reserved Apart from fair dealing for the purposes of research or private study for non-commercial purposes, or criticism or review as permitted under the terms of the UK Copyright, Designs and Patents Act, 1988 and the Copyright and Related Rights Regulations 2003, this publication may not be reproduced, stored or transmitted, in any form or by any means, without the prior permission in writing of The Royal Society of Chemistry, or in the case of reprographic reproduction only in accordance with the terms of the licences issued by the Copyright Licensing Agency in the UK, or in accordance with the terms of the licences issued by the appropriate Reproduction Rights Organization outside the UK. Enquiries concerning reproduction outside the terms stated here should be sent to The Royal Society of Chemistry at the address printed on this page. Published by The Royal Society of Chemistry, Thomas Graham House, Science Park, Milton Road, Cambridge CB4 0WF, UK Registered Charity Number 207890 For further information see our web site at www.rsc.org

Preface James J. Spiveya and Kerry M. Dooleya DOI: 10.1039/9781847559630-FP005

Recent research activity in catalysis has centered on energy related processes, and the synthesis of higher value compounds using biomass as well as conventional reactants. The work reported here addresses these two important areas. First, Nicolas Bion, Florence Epron, and Daniel Duprez (Universite´ de Poitiers, France) provide a review of bioethanol reforming, particularly as compared to conventional hydrocarbon reforming. Reforming of bioderived ethanol is an essential element in an overall process to deliver hydrogen from renewable resources. The authors show important diﬀerences between reforming of an oxygenate such as ethanol and reforming of conventional hydrocarbons—e.g., deactivation. Johannes Schwank and Andrew Tadd (Univ. Michigan) examine a closely related reaction—catalytic reforming of liquid hydrocarbons, particularly for application to solid oxide fuel cells, which are being developed commercially for use as auxiliary power systems. They consider steam reforming, catalytic partial oxidation, and autothermal reforming, each of which has diﬀerent challenges in terms of heat transfer, kinetics, and catalyst deactivation. Another energy-related subject is the use of spectroscopic methods to study reactions of interest in environmental control systems. Speciﬁcally, Fred Meunier (CNRS, France) reviews spectrokinetic methods to study reactions such as NOx reduction using IR spectroscopy. He also shows how spectroscopy can be applied to the water-gas-shift reaction, and important reaction in the catalysis of fuel reforming for hydrogen production. He shows how DRIFTS can be combined with isotopic analysis to study the dynamics of the active catalyst surface with a combination of these two methods. K. Seshan (Univ. Twente, Netherlands) reports on oxidative conversion of low molecular weight alkanes to the corresponding oleﬁns. These oleﬁns are important feedstocks for the chemical industry. This particular review focuses solely on oxidative conversion of these alkanes to oleﬁns, which has advantages in terms of thermodynamics and kinetics compared to alternative processes, primarily catalytic or steam cracking. Asymmetric hydrogenation of activated ketones is a particularly demanding reaction, producing high-value, enantiopure products for the specialty chemical industry. Jozef Margitfalvi and Emilia Ta´las (Institute of Surface Chemistry and Catalysis, Budapest) review in particular heterogeneous catalysts for these reactions, although homogeneous catalysts are perhaps more widely used at present. However, homogeneous metal-based catalysts are expensive due to the need for chiral ligands. This chapter shows

a

Gordon A. and Mary Cain Dept. Chemical Engineering, Louisiana State University, Baton Rouge, LA 70803

Catalysis, 2010, 22, v–vi | v

c

The Royal Society of Chemistry 2010

recent progress in development of heterogeneous catalysts, such as encapsulation of a chiral metal complex in micropores. Finally, Cristina Pina, Ermelinda Falleta, and Michele Rossi (Universita` di Milano, Italy) provides a review of gold catalysis. This focuses on reactions such as selective oxidation of alcohols and carbohydrates, and selective oxidation of hydrocarbons. Within this chapter, a more general summary of gold catalysis in the synthesis of advanced materials (e.g., those based on PAN) is provided. We greatly appreciate the eﬀorts of the authors who have contributed to this volume. We thank the Royal Society of Chemistry for their support of this series. Comments are welcome.

vi | Catalysis, 2010, 22, v–vi

CONTENTS Cover Image provided courtesy of computational science company Accelrys (www.accelrys.com). An electron density isosurface mapped with the electrostatic potential for an organometallic molecule. This shows the charge distribution across the surface of the molecule with the red area showing the positive charge associated with the central metal atom. Research carried out using Accelrys Materials Studioss.

Preface James J. Spivey and Kerry M. Dooley

v

Bioethanol reforming for H2 production. A comparison with hydrocarbon reforming Nicolas Bion, Florence Epron and Daniel Duprez 1. Introduction 2. The steam reforming of hydrocarbons 3. Steam reforming of ethanol 4. Utilisation of crude bioethanol 5. Conclusions and recommendations References

1

Catalytic reforming of liquid hydrocarbons for on-board solid oxide fuel cell auxiliary power units Johannes W. Schwank and Andrew R. Tadd 1. Introduction 2. Fuel properties and SOFC fuel requirements 3. Catalysts for reforming of liquid hydrocarbons

1 2 23 37 46 48

56

56 58 61

Catalysis, 2010, 22, vii–ix | vii

c

The Royal Society of Chemistry 2010

4. Fuel reforming methods 5. Deactivation of reforming catalysts 6. On-board reforming of fuels for SOFC APU applications 7. Systems engineering aspects of on-board fuel processing 8. Conclusions Acknowledgments References

62 69 76 81 84 85 85

Coupling kinetic and spectroscopic methods for the investigation of environmentally important reactions

94

F. C. Meunier 1. Introduction 2. The bases of spectrokinetic analyses 3. Investigation of the selective reduction of NOx with propene over Ag/Al2O3 4. Spectrokinetic operando investigation of catalytic reactions 5. Overall conclusions References

105 116 117

Oxidative conversion of lower alkanes to oleﬁns

119

K. Seshan 1. Introduction 2. Oxidative conversion of alkanes to oleﬁns over oxide catalysts with redox properties 3. Oxidative conversion of alkanes over oxide catalysts with no formal ‘‘redox’’ properties 4. Catalytic alkane oxidation at ambient conditions using cold plasma C–C, C–H scission vs C–C bond coupling Acknowledgements References

94 95 96

119 122 126 131 139 139

Asymmetric hydrogenation of activated ketones

144

Jo´zsef L. Margitfalvi and Emı´lia Ta´las 1. Introduction 2. Cinchona alkaloids 3. Alkaloids used in Oritos’s reaction 4. Methods and approaches used 5. Speciﬁcity of Orito’s reaction

144 151 160 164 178

viii | Catalysis, 2010, 22, vii–ix

6. Spectroscopic investigations 7. Theoretical calculations 8. Reaction mechanisms and related calculations 9. Conclusions Abbreviations used References

220 235 243 258 261 262

Gold catalysis in organic synthesis and material science

279

Cristina Della Pina, Ermelinda Falletta and Michele Rossi 1. Introduction 2. Gold catalysis in organic synthesis 3. Selective oxidation of carbohydrates 4. Selective oxidation of hydrocarbons 5. Gold catalysis in material science 6. Conclusions References

279 280 292 296 300 313 314

Catalysis, 2010, 22, vii–ix | ix

Bioethanol reforming for H2 production. A comparison with hydrocarbon reforming Nicolas Bion,a Florence Eprona and Daniel Dupreza DOI: 10.1039/9781847559630-00001

Hydrogen is essentially produced by steam reforming (SR) of hydrocarbon fractions (natural gas, naphtha, . . .) on an industrial scale. Replacing fossil fuels by biofuels for H2 production has attracted much attention with an increased interest for bioethanol steam reforming. Kinetics and mechanisms of hydrocarbon-SR and alcohol-SR present some similarities but also some very important diﬀerences due to alcohol reactivity much more complex than that of hydrocarbons. The scope of this report is to compare the two processes in terms of reaction mechanisms. Attention will also be paid to the case of crude bioethanol. 1.

Introduction

Whereas hydrogen is the most abundant element of the Universe, it is relatively rare on Earth (0.9 atom % in the outer shell of our planet).1 Virtually, it does not exist as dihydrogen: it is associated with oxygen in water, with carbon in fossil hydrocarbons, both with oxygen and carbon in bioresources (carbohydrates, cellulosic and lignocellulosic matter, lignin, . . .) and more rarely with other elements. Water is by far the main source of hydrogen on Earth (Table 1). The stock of hydrogen available in fresh waters (lakes and rivers) is then of 1.3 1013 tons while the total ressources in hydrogen in oceans ans seas amount to 1.5 1017 tons. Comparatively, the ressources in hydrogen available in fossil fuels are modest (Table 2). Assuming a mean H/C atomic ratio of 1.66 in crude oil,6 of 3.8 in natural gas7 and of 0.8 in coal,8 the stock of hydrogen in fossil fuels would not exceed 111 109 T (23 in crude oil, 30 in natural gas and 58 GT in coal reserves), i.e. two orders of magnitude less than the amount of hydrogen contained in fresh waters. Unfortunately, water is a stable molecule needing a high energy input to recover hydrogen as H2. This energy may be provided by (i) a chemical source by oxidizing the carbon of an hydrocarbon into CO and CO2 (steam reforming); (ii) by electricity (water electrolysis) or (iii) by photons (water splitting). Steam reforming is by far the main process for H2 Table 1 Total volumes of water available on Earth and equivalent amount of H22,3 Oceans and Seas Deep continental waters Fresh waters (easily accessible) Lakes and Rivers Rivers

1 350 000 000 km3 36 000 000 km3

1.5 1017 T H2 4 1015 T H2

110 000 km3 1700 km3

1.3 1013 T H2 1.9 1011 T H2

a

University of Poitiers & CNRS. LACCO, Laboratory of Catalysis in Organic Chemistry, 40Av. Recteur Pineau, 86022 Poitiers Cedex, France

Catalysis, 2010, 22, 1–55 | 1 c

The Royal Society of Chemistry 2010

Table 2 World proven reserves and consumptions of fossil fuels in 20084,5

Oil Gas Coal

World proven reserves (109 TOE)

Annual consumption (109 TOE)

Number of years of reserve

181 172 610

4.0 2.7 3.8

45 65 160

1 TOE (Ton Oil Equivalent)=7.33 barrels=1000 m3 of gas=1.5 T of coal

production9,10 and only this way of hydrogen production will be examined in this Chapter. On an environmental point of view, steam reforming is not a green process since all (or almost all) the carbon of the hydrocarbons is transformed into carbon dioxide. To avoid this drawback, fossil fuels may be replaced by biofuels. Carbon dioxide is still produced but it may be recycled to new biomolecules by photosynthesis. The annual production of biomass in the World would be comprised between 150 and 420 109 metric tons.11–13 The mean hydrogen content in biomass being comprised between 5 and 7 wt-%,13 the stock of hydrogen in this renewable matter would be close to 11 109 T/year. In other words, ten years of biomass production would be suﬃcient to recover all the hydrogen content of fossil fuels. However, a great part of this biomass is composed of wood, diﬃcult to transform into valuable products. For that reason, only products derived from cellulosic and hemicellulosic biomass have been considered for hydrogen production. Since ten years, intensive researches have been devoted to the steam reforming of bioethanol which is a fuel well-adapted to the production of hydrogen.14 This Chapter deals for a great part with this process, with a special attention paid to the use of crude bioethanol. In a ﬁrst part, however, the steam reforming of hydrocarbons (aromatics and alkanes) will be reviewed as the model of many mechanistic investigations. This will allow one to compare the steam reforming of hydrocarbons with the steam reforming of ethanol, an alcohol leading to more complex kinetic schemes. 2.

The steam reforming of hydrocarbons

2.1

Thermodynamics

For methane, four main reactions can occur:15–18 1. The steam reforming reaction leading to CO and H2 CH4 þ H2 O ! CO þ 3H2

DH0298 ¼ þ206 kJ mol1

ð1Þ

2. The steam reforming reaction leading to CO2 and H2 CH4 þ 2H2 O ! CO2 þ 4H2

DH0298 ¼ þ165 kJ mol1

ð2Þ

3. The water gas shift reaction (WGS) CO þ H2 O ! CO2 þ H2

DH0298 ¼ 41 kJ mol1

ð3Þ

4. The coking reaction CH4 ! C þ 2H2 2 | Catalysis, 2010, 22, 1–55

DH0298 ¼ þ75 kJ mol1

ð4Þ

For higher hydrocarbons, a similar set of reactions may be written; for instance the reactions of n-heptane and of toluene leading to CO and H2 become: C7 H16 þ 7H2 O ! 7CO þ 15H2

DH0298 ¼ þ1107 kJ mol1

ð5Þ

C7 H8 þ 7H2 O ! 7CO þ 11H2

DH0298 ¼ þ869 kJ mol1

ð6Þ

Except for the WGS reaction, all the reactions involved in the hydrocarbon steam reforming are strongly endothermic with an increase of the number of molecules. They are thus favored at high temperatures and low pressures. The temperature eﬀect is illustrated in Fig. 1 which shows the change with T of the gas composition at equilibrium in the methane steam reforming (initial conditions H2O/CH4=1, P=1 bar). Calculations were carried out by minimizing the sum of the Gibbs free energies of formation of all the compounds (reactants and products) while keeping constant the number of moles of each element (here C, H and O). Details of the procedure are given in Perry’s Handbook.19 Thermodynamic data (molar Gibbs free energy of each compound) are taken from Stull et al.20 Maximal H2 production is observed around 700 1C. Above 700 1C, the H2 mol% does no longer increase because of the preferential formation of CO at high temperature. This is coherent with the WGS equilibrium: reaction 3 being exothermic, CO2 is favored at low temperature while the reverse reaction (RWGS) yielding CO is favored at high temperature. At 900 1C, total conversion of methane can be achieved yielding quasi-exclusively a syngas with the composition given in equation 1 (75% H2 þ 25% CO). 80% 70% H2 60%

Mole %

50% CH4 40% H2O 30%

CO

20% CO2 10% 0% 100

300

500

700

900

T (°C) Fig. 1 Equilibrium composition of the methane steam reforming (without C formation). Initial state: H2O/CH4 molar ratio of 1, P=1 bar

Catalysis, 2010, 22, 1–55 | 3

Table 3 Equilibrium gas compositions and number of moles of gas formed (nG) in the steam reforming of methane, n-heptane and toluene at 700 1C (P=1 bar)

Methane n-Heptane Toluene

Wet % Dry % Wet % Dry % Wet % Dry %

H2O

H2

CO

CO2

CH4

nG

5.46 – 4.91

65.62 69.41 58.04 61.03 50.52 52.70

18.69 19.77 25.51 26.83 32.90 34.32

2.39 2.53 3.32 3.49 4.15 4.33

7.84 8.30 8.23 8.65 8.29 8.65

3.46

4.14 –

18.89 15.44

Initial conditions: H2O/CH4=1; H2O/n-C7=H2O/Tol=7

The equilibrium gas compositions in methane, n-heptane and toluene steam reforming at 700 1C are compared in Table 3. The initial state is a steam/ hydrocarbon mixture with a molar ratio corresponding to the stoichiometry of equations 1, 5, 6 written with H2 and CO as products of steam reforming. Methane and steam conversions are very high but not total even at 700 1C. The vol.% of hydrogen expected in dry gases amounts to 70%. The formation of methane and CO is favored in the steam reforming of C7 compounds at 700 1C, which tends to decrease the hydrogen content in dry gases. In every cases, the CO-to-CO2 molar ratio is remarkably constant (around 7.7) whathever the starting hydrocarbon. The conversion of C7 hydrocarbons (not reported in Table 3) is total whatever the temperature. Thermodynamics predicts that H2 production from C2 þ hydrocarbons is controlled by methane formation. If, kinetically, methane formation can be avoided, the equilibrium gas composition is signiﬁcantly changed. An example is shown in Fig. 2 for toluene steam reforming without CH4 formation. The complete conversion of toluene only occurs around 550–600 1C and the maximal H2 formation is observed at 500 1C, much below the corresponding value when methane can be formed. Interestingly, benzene formation can then be observed with a maximum around 400 1C. If methane formation may be avoided or suppressed, toluene dealkylation can occur in the 400–500 1C range of temperature: this is the so-called toluene steam dealkylation (TSDA) which has been largely studied in the past as a way to produce benzene from toluene21–28 (and more generally from alkylbenzenes)29,30 without H2 consumption. 2.2

Kinetics and mechanisms

Methane being a very stable molecule, the steam reforming of natural gas should be carried out at high temperatures (around 600–700 1C) and it can be expected that methane activation is a critical step of the reaction (see Section 2.2.3). Heavier hydrocarbons are more reactive and there are many indications in the literature that water activation may be the rate determining step in the steam reforming of these compounds, specially at lower temperatures (400–600 1C). 2.2.1

Aromatics

2.2.1.1 Reaction scheme and catalytic activity. Toluene will be chosen as model hydrocarbon illustrating this class of compounds. As mentioned in 4 | Catalysis, 2010, 22, 1–55

80% 70% H2O

H2

60%

Mole %

50% CO

40% CO2

30% TOL 20% 10% BENZ 0% 100

200

300

400

500

600

700

800

T (°C) Fig. 2 Equilibrium composition of the toluene steam reforming at 1 bar without C and methane formation (initial conditions H2O/TOL=7).

Section 2.1, the toluene steam reforming may lead to dealkylation (7) in the 400–500 1C temperature range where methane formation is kinetically unfavored (8, 9). C7 H8 þ H2 O ! C6 H6 þ CO þ 2H2 C7 H8 þ H2 ! C6 H6 þ CH4 C7 H8 þ 10H2 ! 7CH4

DH0298 ¼ 164 kJ mol1

DH0298 ¼ 42 kJ mol1

DH0298 ¼ 574 kJ mol1

ð7Þ ð8Þ ð9Þ

Equations 8, 9 clearly show that methane formation decreases the hydrogen yield. If the reactions leading to benzene, CO, CO2 and CH4 are considered (i.e. 3, 5, 7–9), the following relationship between the product yields could be established:30 7YH2 ¼ 3YB þ 11YCO þ 18YCO2 10YCH4

ð10Þ

Intrinsic activity and selectivity to benzene of Group 8-9-10 metals (ex Group VIII) supported on g-Al2O3 (210 m2 g 1) are reported in Table 4. All the catalysts deactivate with time-on-stream. It was shown that the selectivities vary very little when the catalysts are deactivating and that the ﬂow rate of dry gases (H2 þ CO þ CO2 þ CH4) is proportional to the toluene conversion. This property allows one to extrapolate the running activity to zero-time. Turnover frequencies given in Table 4 are the intrinsic activities determined by this technique. The deactivation is due to a carbon deposit on the catalyst (metal and support). It was proved that toluene steam dealkylation is a relatively ‘‘structure’’ insensitive reaction: the Catalysis, 2010, 22, 1–55 | 5

Table 4 Intrinsic activity and selectivity to benzene of Group 8-9-10 metals supported on g-Al2O3 (reaction conditions: T=440 1C; P=1 bar; H2O/TOL molar ratio of 6). From ref. 27 Metal/g-Al2O3

Turnover frequency (h 1)

Relative activity (based on Rh=100)

Initial selectivity to benzene (SB %)

0.6% Rh 0.6%Pd 1.1%Pt 5%Ni 5%Co 0.6%Ru 1.15%Ir

470 134 88 77 67 64 60

100 29 19 17 15 14 13

81 97 98 59 51 53 88

carbon deposit does not change the turnover frequency per free metal atom which remains very close to the turnover frequency extrapolated at zero-time. Free metal surface area in coked catalysts were measured according to a procedure detailed in refs.31,32 Initial selectivity to benzene is the selectivity extrapolated at zero conversion. An interesting feature of the toluene steam dealkylation reaction is the increase of SB with conversion. For instance, rhodium selectivity reaches 86% at 10% conversion and 92% around 40% conversion. This is due to a complex behaviour of the catalyst including two phenomena: (i) though the dealkylation is structure insentive, the total steam reforming of toluene (6) is not and is strongly poisoned by coke deposits; (ii) CO is an inhibitor of both dealkylation and total reforming but the former reaction is much less aﬀected than the latter one.22,33 The relative activities reported in Table 4 show that the steam dealkylation reaction is not very sensitive to the nature of metal: there is less than one order of magnitude diﬀerence between the most active metal (Rh) and the less active one (Ir). The metal ranking found by Duprez et al.27 is close to that reported by Grenoble24 in similar conditions (440 1C, alumina support of 175 m2 g 1), except the waterto-toluene ratio (3.25 in the Grenoble’s study). The only diﬀerence lies in the position of Ru found more active by Grenoble. Kim showed that promoting alumina by vanadium oxide did not change signiﬁcantly the metal ranking.34 2.2.1.2 Support eﬀects and reaction mechanism on rhodium catalysts. Contrasting with the low sensitivity to nature of metal, the steam dealkylation reaction appeared as very sensitive to nature of support. This is illustrated in Table 5 with the rhodium as active metal. There are two to three orders of magnitude diﬀerence between the activity of rhodium catalysts supported on chromia or aluminochromia catalysts35 and those supported on silica or carbon.25,27 The data of Table 5 show that the differences are not due to metal dispersion eﬀects. Catalyst characterization carried out after test also conﬁrmed that there was no sintering during the kinetic measurements aﬀecting more certain supports. A bifunctional mechanism was proposed by Grenoble26 and Duprez et al.27 to explain this dramatic support sensitivity. In this mechanism schematized on Fig. 3,

6 | Catalysis, 2010, 22, 1–55

Table 5 Support eﬀect in toluene steam dealkylation at 440–480 1C (base 100 for well dispersed Rh on g-Al2O3) Support

Surface area (m2 g 1)

Rh dispersion (%)

Relative activity of rhodium (per metal site)

Reference

Cr2O3 g-Al2O3 a-Cr2O3 Al2O3-SiO2 TiO2 Al2O3-SiO2 SiO2 SiO2 SiO2 C unsupported

40 210 10 300 10 110 330 260 300 950 5

20 90–100 90–100 50 30 40 40 90–100 70 40 1.2

300–400 100 50 40–50 20 15 20 20 3 1 0.1

35 23,25,27 36 30 27 37 27 37 25 25 25

H2O

CH3

Surface migration OH

CXHY

OH

metal support

Fig. 3 Bifunctional mechanism of toluene steam dealkylation

the hydrocarbon molecule would be adsorbed and activated on metal sites while the water molecule would be activated on support sites. The molecule of toluene undergoes a dissociative chemisorption on a metal site M leading to a molecule of benzene and an alkyl fragment (11). This reaction should require two metal sites. However, benzene being less strongly adsorbed than toluene on metals, it is immediately desorbed once formed. C6 H5 CH3 þ M ! CHx M þ C6 H6 þ

2x H2 2

ð11Þ

Water is activated on support sites (S–O–S) according to equation 12: H2 O þ SOS ! 2SOH

ð12Þ

The ﬁnal step is a transfer of OH groups to metal particles where they react to form carbon oxides and hydrogen: x CHxM þ 2SOH ! CO þ H2 þ SOS þ M 2

ð13Þ

Catalysis, 2010, 22, 1–55 | 7

The global reaction in this catalytic cycle is the dealkylation to benzene, CO and H2. It is implicitely assumed that CO2 is formed by water gas shift. The reaction rate (per gram of catalyst) derived from equation 13 is: r ¼ kyCHx x2OH

ð14Þ

where yCHx is the surface coverage of CHx fragments on metal sites while xOH is the OH group coverage on support sites. As step (13) implies a transfer of OH groups through the metal/support interface, it is assumed that k=KI0, I0 being the length of the particle perimeter per gram of catalyst. Surface coverages are deduced from equations 11, 12 and Langmuir-type hyperbolic expressions for equilibium adsorption of molecules A can be approximated to power-law expressions using the following equation: KA PA ¼ aðKA PA Þn 1 þ KA PA

ð15Þ

Inserting power-law expressions for the surface coverages in equation 14 leads to the following rate equation:27,38 TOF ¼

r mð1nÞ ¼ CðS20 l0 Þ1n PnT PW M0

ð16Þ

where M0 and S0 are the numbers of metal sites and of support sites, respectively, per gram of catalyst; PT and PW are the partial pressures of toluene and water. Knowing the weight loading (xm %) and the dispersion D0 (%) of metal, the speciﬁc perimeter I0 can be calculated by: I0 ¼ bD20 xm with b ¼ 1:6 107

rA2mol M

ð17Þ ðcubic particlesÞ

ð18Þ

r being the metal density (g m 3), Amol the molar surface of metal (m2 mol 1) and M, the atomic weight (g mol 1). For Rh, r=12.45 106 g m 3, Amol=47633 m2 mol 1 (equidistribution of low index faces) and M=102.9 g mol 1, which gives b=4.28 105 m g 1. It is worth noting that, for a catalyst commonly used in steam reforming (0.6%Rh, mean dispersion of 50%), I0 amounts to 6.4 108 m g 1, a length greater than the EarthMoon distance. This justiﬁes the great impact the reactions at metal/support interfaces may have in Catalysis. Equations 16, 17 shows that TOF values should be proportional to I01 n. Grenoble reported kinetic orders of 0.08 and of 0.41 with respect to toluene and to water.25 Most authors found orders with respect to toluene comprised between 0.03 and 0.25 while those with respect to water are generally close to 0.5.30 One may thus expect TOF values to be proportional to I0. Table 6 reports the results obtained on a series of Rh/Al2O3 catalysts. There is a complex variation of the intrinsic activity with the metal loading and the dispersion (almost a factor 20 between the most active catalyst and the less active one). Applying the bifunctional model with n=0 reduces the variation to a factor 4 while a very good ﬁt is obtained 8 | Catalysis, 2010, 22, 1–55

Table 6 Veriﬁcation of the bifunctional mechanism in toluene steam dealkylation (440 1C; PT=0.145 atm and PW=0.855 atm). From ref. 32 Rh loading (xm % Rh)

Dispersion (D0 %)

TOF (h 1)

0.031 0.063 0.18 0.58 0.60 4.82 4.87 10.1 10.3

100 100 95 92 96 45 61 13.7 32

60 160 180 430 460 650 1070 200 700

TOF ðD20 xm Þ1n

n=0

n=0.3

0.194 0.254 0.111 0.088 0.083 0.067 0.059 0.106 0.066

1.08 1.76 1.02 1.12 1.10 1.05 1.12 1.02 1.07

Table 7 Support eﬀects of diﬀerent metals in toluene steam dealkylation (reaction conditions: 440 1C, partial pressures: 0.145 bar of toluene and 0.855 bar of water) Metal

TOF (M/Al2O3) h 1

TOF (M/SiO2) h 1

TOF ratio (Al2O3/SiO2)

1.1%Pt 0.6% Rh 0.6%Pd 1.15%Ir 5%Ni 5%Co 0.6%Ru

88 470 134 60 77 67 64

15 91 54 45 69 64 62

5.87 5.16 2.48 1.33 1.11 1.05 1.03

for n=0.3 (factor 1.7 and less than 1.1 for nine of the ten samples). To sum up, the kinetic model based on the bifunctional mechanism represents adequately the changes of activity of Rh/Al2O3 catalysts in toluene steam dealkylation. 2.2.1.3 Other metals. The support eﬀect evidenced for Rh catalysts was investigated for other metals by comparing their activity on alumina and silica27 (Table 7). There are two groups of metals: those showing a strong support eﬀect in steam dealkylation (Pt, Rh, Pd and to a lesser extent, Ir) and those having the same activity on alumina and silica (Ni; Co and Ru). In an investigation of multifuel reforming (including toluene), Wang and Gorte conﬁrmed the dramatic eﬀect of the support in these reactions.39 For instance, Pd/CeO2 was about 7 to 10 times more active than Pd/Al2O3 in the steam reforming of toluene between 400 and 500 1C. As a rule, ceria or ceria-based oxides were found to be very good supports for metals in several steam reactions (steam reforming and water-gas shift).40–43 The mechanism of these reactions on ceria-based catalysts is likely to diﬀer from that proposed on alumina catalysts in equations 11–13. Ceria is a reducible support Catalysis, 2010, 22, 1–55 | 9

able to dissociate the water molecule and to produce directly hydrogen in a redox process. As proposed by Wang and Gorte,39 hydrocarbon activation would be very similar to equation 11 of the OH-migration mechanism while equation 12, 13 should be replaced by the following steps: H2 O þ Ce2 O3 ! 2CeO2 þ H2

ð19Þ

CHx M þ 2CeO2 ! COM þ Ce2 O3 þ x2 H2

ð20Þ

The possible occurrence of step 19 has been demonstrated by direct decomposition of water over reduced ceria.44–46 The writing of equations 19, 20 is certainly oversimpliﬁed since Ce2O3 very likely cannot be formed under the conditions of steam reforming. There exist numerous sub-oxides CeO2 x (with 0oxo0.5) whose existence is more probable in these conditions.47 Metals of class 2 (Ni, Co, Ru) are not support-sensitive. These metals have the lowest Mn þ /M1 electrochemical potential and can dissociate the water molecule in steam reforming conditions. For these metals, a monofunctional mechanism with all steps occuring on the metal was proposed.27 equations 12, 13 are replaced by equations 21, 22 respectively H2 O þ M ¼ HO M þ 12 H2

ðequilibrium constant KW Þ

CHx M þ HO M ! CO þ x2 H2 þ 2M

ðrate constant kÞ

ð21Þ ð22Þ

This mechanism led to the following rate equation: TOF ¼

kT PT kT PT 1 1þK kKM0

ð23Þ

kT being the rate constant of the dissociative adsorption of toluene (11) and K is a constant depending on water dissociation equilibium (21): K¼

KW PH2O 1=2

PH2

ð24Þ

2.2.1.4 Isotopic exchange studies. 16O/18O and H/D isotopic exchange studies were performed to measure the rate of diﬀusion of OH groups involved in the bifunctional mechanism of steam reforming.48–51 The principle of the measurement is depicted in Fig. 4. Exchange experiments are typically performed with pure 18O2 at t=0. Two conditions should be fulﬁlled: (i) adsorption/desorption of O2 on the metal particles (step 1) should be very fast. This is veriﬁed by measuring the rate of isotopic equilibration between 18O2 and 16O2 on the metal; (ii) the rate of direct exchange between 18O2 and the support (step 5) should be negligible. This is veriﬁed by measuring the rate of exchange on bare supports. Similar experiments were carried out with deuterium to measure the rate of hydrogen migration. A simple kinetic model is suﬃcient to calculate the coeﬃcient of surface diﬀusion DS for supports having moderate O or H mobility.50 When surface mobility is very fast, more complex kinetic models should be used.52,53 Several important conclusions could be drawn from these studies. 10 | Catalysis, 2010, 22, 1–55

18O16O 18O

Surface mobility

2

16O

18O 2

2 16O

18

O 18O

Metal

18

O

16O

Support

Fig. 4 18O2 exchange with the 16O of the support via the metal particles. Step 1: adsorption/ desorption of O2 on metal particles; Step 2: O transfer from metal to support; Step 3: surface migration; Step 4: place exchange of 18O with 16O Step 5: direct exchange gas/support.

Table 8 Relative mobility of oxygen and hydrogen on oxide-supported Rh catalysts (Rh is used here as a porthole for O and H diﬀusion on oxide surfaces). From ref. 48,51 Oxygen at 400 1C CeO2 MgO ZrO2 CeO2/Al2O3 Al2O3 SiO2

Hydrogen at 75 1C 2300 50 28 18 10 0.1

CeO2 MgO CeO2/Al2O3 Al2O3 ZrO2 SiO2

80 22 16 10 2.3 Very low

1. While hydrogen is very mobile on most supports below 100 1C, oxygen surface diﬀusion is a relatively slow process and requires temperatures in the 300–500 1C range to be measurable. 2. However, activation energy for oxygen surface diﬀusion (50–100 kJ mol 1) is signiﬁcantly higher than that for hydrogen (10–20 kJ mol 1). As a consequence, oxygen and hydrogen migrations occur at similar rates around 5001C. 3. Rhodium is the best metal for oxygen activation.54 Adsorption/desorption of O2 on this metal is very fast and is not very sensitive to the metal particle size: it is slightly faster on small Rh clusters than on big particles. By contrast, step 1 is not so fast on Pt. However, this step is very sensitive to the metal particle size and, contrary to the case of Rh, it is much faster on big Pt particles. Oxygen equilibration is very slow on Pd while Ru and Ir could be good candidate for replacing Rh under certains conditions.55 4. The relative values of coeﬃcients of surface diﬀusion for oxygen and hydrogen on selected oxides are given in Table 8. Oxygen and hydrogen diﬀusing at the same rate in the 400–500 1C range of temperature, it is very likely that there is a collective migration of OH groups at these temperatures on most oxides. A crucial point is that OHs are virtually immobile on silica while the coeﬃcient of diﬀusion would be two orders of magnitude higher on alumina. This result is in agreement with the bifunctional mechanism with the slow step being the OH group migration Catalysis, 2010, 22, 1–55 | 11

from support to metal. This explains why silica is a poor support for steam reforming reactions. Unfortunately, carbon supports could not be investigated by isotopic exchange. There are some indication in the literature that O species are quite mobile on carbon: for instance Lim et al. found that NO decomposition was favored on Pt/C by continuous removal of O species from the metal by the support itself.56,57 The most probable explanation for the low activity of carbon-supported metal catalysts might be the high hydrophobicity of carbon surface which cannot easily activate the water molecule. 2.2.1.5 Recent studies on fuel and tar reforming. In the last ten years, much attention was paid to steam reforming reactions for hydrogen production from gasoline, heavy oils, biomass and tars. Toluene was chosen as model aromatic hydrocarbon in many studies with the objective to reform totally the molecule and obviously not to produce benzene. However, steam dealkykation could be observed when the steam reforming of aromaticreach fractions was used to produce hydrogen. Fuel reforming. Hydrogen production for on-board applications was investigated by Springmann et al.58 at 600–800 1C, 2–5 bar over rhodium catalysts deposited on metallic monoliths. The reaction led essentially to a syngas (30% H2 þ 12% CO) with small amounts of benzene, methane and CO2. A detailed analysis of benzene reactivity supports the interpretation that toluene is ﬁrst dealkylated to benzene which is then gasiﬁed. Due to signiﬁcant formation of coke, the catalyst is not stable in toluene steam reforming. Stable hydrogen production (with less carbon deposit) was obtained in an autothermal ATR process (H2O:O2:TOL=14:3:1). Steam reforming or ATR of iso-octane, hexene and gasoline were also investigated on the same catalysts. These hydrocarbons were slightly more reactive than toluene with higher amounts of CO2 in the syngas at 675 1C (3.5% CO2 instead of 1% with toluene). Similar results were obtained by Qi et al. who studied gasoline autothermal reforming at 650–800 1C over a complex Rh catalyst (0.3%Rh/ 3%MgO/20%CeO2-ZrO2) washcoated on a ceramic monolith.59 The authors showed that alkanes led to minor amounts of methane together with the syngas while aromatics (e.g. toluene) rather led to methane-free syngas. Qi et al. also pointed out the role of sulfur in promoting coke formation as well as the detrimental eﬀect of washcoating the catalyst on a ceramic monolith of low thermal conductivity. This sulfur eﬀect on coke formation is not in agreement with previous results of Duprez et al. who showed that sulfur might hinder coke formation in toluene steam dealkylation.31–32,60 The apparent discrepancy is likely coming from the diﬀerences of temperatures (440 1C in steam dealkylation and 650–800 1C in ATR of gasoline). Sulfur poisoning is a critical problem in steam reforming processes. SR is operating on reduced catalysts under conditions where the sulfur organic compounds are themselves steam reformed into hydrogen, carbon oxides and dihydrogen sulﬁde. Several studies were carried out to attempt solving this drawback. Azad et al. have developped several catalyst formulation for the steam reforming of jet fuels in the presence of large amounts of sulfur

12 | Catalysis, 2010, 22, 1–55

(up to 1000 ppm).61–63 The best support tested by Azad et al. was a ternary oxide composed of 10 mol-% Gd2O3, 25 mol-% ZrO2 and 65 mol-% CeO2 doped or not by Y2O3 or CuO. Active metals were Rh,61,63 Pd62 or combination of both metals61 tested in the steam reforming of toluene at 825 1C (with or without 50 ppmS as thiophene). Ceria was shown to play a signiﬁcant role in the catalyst thioresistance. Two reactions can occur (25, 26) which help mitigate sulfur-led poisoning and long-term deactivation: Reduction reaction: CeO2 ðsÞ þ ð2nÞH2 ðgÞ ¼ CeOn ðsÞ þ ð2nÞH2 OðgÞ

ð25Þ

Sulﬁdation reaction: 2CeOn ðsÞ þ H2 S þ ð2n3ÞH2 ðgÞ ¼ Ce2 O2 SðsÞ þ 2ðn1ÞH2 OðgÞ

ð26Þ

The speciﬁc role of ceria in sulfur resistance of Rh catalysts for jet fuel reforming applications was conﬁrmed by Strohm et al. who also showed that Ni added to Rh reinforced the thioresistance of a Rh-CeO2-Al2O3 catalyst.64 Rhodium-based catalysts oﬀered a better resistance to deactivation than Pd-ones and the addition of Cu signiﬁcantly improved the thioresistance of the GdZrCeOx support. This property may be paralleled with the exceptional OSC (oxygen storage capacity) of Rh-CuCeOx catalysts, showing the great mobility of oxygen (and probably sulfur) in these materials.65 To increase the hydrogen yield, combination of steam reforming (SR), autothermal reforming (ATR) and water gas shift is often proposed in integrated processes.66,67 Gasiﬁcation of a fuel composed of methylcyclohexane and toluene was studied by Wang et al. at 530 1C.68 Combining SR and WGS on a multifunctional catalyst (Ni-Re/Al2O3) allows to increase the CO2/CO þ CO2 ratio up to 92% (at 16% conversion). This is coherent with the excellent performances of Ni and Re in WGS as reported by Grenoble et al.69 These authors showed that the WGS reaction was very sensitive both to nature of metal and to that of support (Table 9). Table 9 Turnover frequency of alumina-supported metals and support eﬀect in water-gas shift reaction at 300 1C (24 vol-% CO þ 32 vol-% H2O). From ref. 69 Support eﬀect Metal/Al2O3

TOF (h 1)

Catalyst

Relative activity

Cu Re Co Ru Ni Pt, Os Au Fe, Pd Rh Ir

43 400 1380 890 695 370 225 130 50 31 12

Pt/Al2O3 Pt/SiO2 Pt/C

90 9 1

Rh/Al2O3 Rh/SiO2

13 1

Catalysis, 2010, 22, 1–55 | 13

As WGS is an exothermic reaction, it is favored at low temperature. For this reason, Wang et al. found better perfomances in a classical two-bed process with ATR (instead of SR) at relatively high temperature and WGS at low temperature. A Ni/Ce-ZSM5 was developped for this application. The use of Ce as dopant could be anticipated, ceria being a very good promoter of the WGS reaction.40,41,70,71 Table 9 shows that Rh is a bad WGS catalyst while it is considered as the best SR metal. Pt is a better WGS catalyst than Rh but it is less active in steam reforming. Interestingly, Pd has often intermediary properties between those of Pt and Rh in several reactions with the following ranking: Rh W Pd W Pt in SR and Pt W Pd W Rh in WGS. For these reasons, the most advanced catalyst formulation for multifuel reforming applications include generally both Rh or Pd and a WGS component (most often a ceria-based materials). Wang and Gorte showed that their Pd/CeO2 catalyst was active in the reforming of many hydrocarbons: methane, ethane, n-butane, n-hexane, 2,4-dimethylhexane, n-octane, cyclohexane, benzene, and toluene.39 Nilsson et al. investigated the ATR of hydrocarbons, alcohols, gasoline and E85 over a catalyst composed of Rh supported on Ce/La-doped g-Al2O3 and deposited on cordierite monoliths.72 Provided that the fuel is correctly vaporized under the form of very ﬁne droplets (the critical point of the process), good hydrogen yields and selectivities may be obtained by ATR of complex mixtures such as Diesel gasoils or gasolines. It should however be mentionned that the swedish fuels used in this study contained low amounts of sulfur and that the aromatics content of the diesel was comparatively lower than in other countries. These factors allowed the catalyst to be more resistant to sulfur poisoning and to coke formation than with more common fuels. Steam reactions (HC’s steam reforming or WGS) are key-reactions in Three-Way Catalysis (TWC) for converting pollutants in rich conditions when O2 concentration in gas phase falls below stoichiometry.41,42,73,74 However, with aromatic gasolines, there is a risk to form benzene by steam dealkylation. In the temperature window of 550–730 1C, Bruelmann et al. observed the net formation of benzene and toluene in the gases issued from a Pd-Rh-CeZrOx-Al2O3 TWC in real conditions.75 The gasoline used in this study contained 41.2% (wt.%) monoaromatics, including 3.4% benzene, 16.6% monoalkylated, 15.8% dialkylated, 5.2% trialkylated, and 0.3% tetraalkylated benzenes, respectively. Whereas all the C8 þ alkylated compounds decreased in the post-catalyst exhaust gases, it was shown that the concentration of benzene and toluene increased. Bruelmann et al. studied in detail the conversion of 12 alkylbenzenes representing 75% of all possible benzene precursors in the post-catalyst exhaust gases. These alkylbenzenes were individually spiked in the pre-catalyst chamber to follow their conversion by chemical ionization mass spectrometry (CI-MS). Dealkylation reactions of ortho-substituted dialkylbenzenes show a clear preference towards benzene formation, whereas meta- and para-isomers mainly form toluene at 630 1C with some shift of the product distribution towards benzene at higher temperature (680 1C). Steam dealkylation is by far the most important reaction leading to benzene formation even though some hydrodealkylation cannot be discarded. 14 | Catalysis, 2010, 22, 1–55

Steam reforming can be applied in the Exhaust Gas Recirculation (EGR) loop of cars to increase the hydrogen content in the gases reinjected into the combustion chamber.76 The combustion of the gasoline with a gas enriched in H2 leads to a signiﬁcant decrease of pollutants and to a better engine yield. Tar reforming. Gasiﬁcation process of biomass leads to a variety of products, specially tars which should be steam reformed to increase H2 yields.77 Again, benzene, toluene or oxygenated aromatics such as phenol or cresols are often chosen as model compounds. In the 1970–1980’s, considerable attention was paid to the treatment of tars by CO/H2O mixtures, hydrogen being in situ generated by the water-gas shift reaction.78,79 In these studies, the main objective was to hydrogenolyze C–C bonds to produce lighter hydrocarbons without paying attention to H2 production. Steam or oxy-steam gasiﬁcation of tars was studied in the 90’s over diﬀerent dolomites with the objective to favor syngas formation.80–82 These materials were proved to oﬀer a relatively good gasiﬁcation activity while minimizing the catalyst cost. They may be a good support for Ni to increase H2 yield while minimizing the coking rate.83 However, their poor mechanical stability made them useless in ﬂuidized reactors. They were progressively replaced by olivine, a Mg-Si-Fe ternary oxide very resistant to attrition.84 Olivine is a natural mineral whose mean composition is 49 wt-% MgO, 42 wt-% SiO2, 8 wt-% Fe2O3 and about 1% Al2O3 and CaO, with a BET area of 4–5 m2 g 1. It was used alone in biomass gasiﬁcation or as a support of metals (mainly Ni) to get more active catalysts.85 A detailed investigation of toluene steam reforming over Ni/olivine was performed by Swierczynski et al.86,87 The reaction temperature was varied in the 550–850 1C range and the H2O-to-toluene molar ratio was maintained between 7.5 and 24 with a standard value of 16 for most kinetic studies. Total conversion of toluene is reached at 650 1C over Ni/olivine while the reaction on the bare support starts at 750 1C. Reaction selectivities are reported in Table 10 for standard tests at 850 1C. The hydrogen yield is 82% on Ni catalyst and only 28% on olivine (a 100% yield corresponds to 18 moles of H2 produced per mole of toluene injected). The thermodynamic limitation at high temperature which tends to favor CO formation (only 11 moles H2 per mole of toluene according to eq. 6, i.e. a H2 yield of 61%) is compensated by the high water/ toluene ratio. Moreover, on Ni/olivine, no hydrocarbon is detected which proves that methane, benzene and polyaromatic hydrocarbons (still formed on olivine) are totally converted.

Table 10 Comparison of catalytic perfomances of olivine and Ni/olivine in toluene steam reforming at 850 1C (H2O/TOL=16). From ref. 87 Selectivities %a Catalyst

CO

CO2

CH4

Benzene

Polyaromatics

H2 yield (%)b

Olivine Ni/olivine

69 66

5 31

2 0

6 0

14 0

28 82

a deﬁned as the % of C in the product per C in reacted toluene. H2 produced per mole of toluene injected.

b

based on 100% for 18 moles

Catalysis, 2010, 22, 1–55 | 15

A detailed characterization of the active sites of Ni/olivine was performed by Swierczynski et al. by means of Mo¨ssbauer spectroscopy and several other techniques (XRD, SEM, H2-TPR).88 They showed that the good activity of Ni/olivine was linked to the formation of a NiO-MgO solid solution at the olivine surface during the oxidation and to Ni alloying by Fe during the reduction. Both phenomena contribute to make the catalyst more resistant to sintering and to coke formation. To increase activity in tar gasiﬁcation, Zhang et al. proposed to dope Ni/olivine with ceria.89 The best results for the steam gasiﬁcation of benzene and toluene were obtained between 700 and 830 1C on a 3% NiO-1%CeO2/olivine. Ceria allows to increase benzene and toluene conversion and to form more H2 and CO2 and less CO and methane (Table 11). Other supports of Ni or Co were used for the steam reforming of model tar compounds: ceria-zirconia,90 Al-La coprecipitate in the presence of Ni and Co91 or Ni-Mg-Al-based commercial steam reforming catalysts.92 These studies were carried out at relatively high temperatures. Lamacz et al. investigated the toluene steam reforming over Ni or Co/Ce0.7Zr0.3O2 catalysts (100–130 m2 g 1) from 400 to 900 1C.90 They found however a total conversion of toluene at 600 1C and above. Bona et al. restricted their study to a temperature of 650 1C focusing their investigation on the optimization of La/Ni and Co/Ni ratios.91 Temperatures from 700 to 875 1C were chosen by Coll et al.92 to study the steam reforming of ﬁve aromatic compounds: benzene, toluene, naphthalene, anthracene and pyrene representative of a biomass gasiﬁcation tar whose composition is given in Table 12. Benzene and toluene are by far the most reactive of the compounds tested, with conversions in the order of 1 gHC gcat 1 min 1 at a H2O/C ratio around 4. On the other hand, pyrene (H2O/C=12.6) and naphthalene (H2O/C=4.2) are the least reactive compounds, with conversions varying from 0.01 to 0.025 gHC gcat 1 min 1, depending on temperature. As a rule,

Table 11 Comparison of catalytic performances of 3%NiO/olivine (A) and 3%NiO-1%CeO2/ olivine (B) in steam reforming of benzene and toluene at 750 1C. Reaction conditions: S/C ratio of 5; space velocity: 862 h 1. From ref. 89 Catalyst Benzene steam reforming

Toluene steam reforming

Conv. % H2 % CO % CO2 % CH4 % Conv. % H2 % CO % CO2 % CH4 % A B

40.5 60.5

61.3 63.6

32.5 23.8

6.3 12.6

0.01 0.01

45.9 64.8

61.2 63.6

28.9 22.5

Table 12 Typical composition of a biomass gasiﬁcation tar. From ref.

10.2 14.6

0.16 0.11

92

Compound

wt-%

Compound

wt-%

Benzene Toluene Other alkylbenzenes Naphthalene Alkylnaphthalenes

37.9 14.3 13.9 9.6 7.8

Three-ring aromatics Four-ring aromatics Phenolic compounds Heterocyclic compounds Other

3.6 0.8 4.6 6.5 1.0

16 | Catalysis, 2010, 22, 1–55

Table 13 Experimental limit of H2O/C ratio for carbon formation at the lower temperature studied. From ref. 92 Compound

T (1C)

Limit H2O/C ratio

Toluene Naphthalene Anthracene Pyrene

725 795 790 790

2.5 3.7 6.6 8.4

the gas composition in these studies90–92 follows the thermodynamic tendency: decrease of CO2 and CH4 and increase of CO when the temperature is increased. Catalyst deactivation is one of the major problems encountered in tar gasiﬁcation. Bain et al. investigated tar reforming with a special attention to benzene, toluene and light alkane transformation. They modeled the kinetics of reforming and in parallel, the kinetics of deactivation of their Ni-alkali-Al2O3 catalyst at ﬁve temperatures from 775 to 875 1C.93 Firstorder rate equations were found to represent both the kinetics of reforming and that of deactivation. Large diﬀerences in activation energies for reforming and deactivation of total tar gasiﬁcation on one hand and of individual hydrocarbon gasiﬁcation (benzene, ethane, methane, . . .) on the other hand were observed. This shows that tar gasiﬁcation is a complex process and that modeling gasiﬁcation of some hydrocarbons present in these tars may not represent the total gasiﬁcation kinetics. The most critical parameter for coke formation is the H2O/C ratio. Interestingly, Coll et al.92 gave the limit values of this parameter to prevent massive coke formation when diﬀerent aromatic hydrocarbons are gasiﬁed (Table 13). 2.2.2 Non-methanic alkanes. Less attention was paid to the steam reforming of C2 þ alkanes. Kikuchi et al. reported a detailed investigation of the reforming of n-heptane at 5501C over Rh catalysts.94 N-heptane may be converted by two reactions: (i) gasiﬁcation by steam reforming and (ii) aromatisation into toluene. Toluene itself may be gasiﬁed by steam; however, if the reaction is carried out at relatively low space time, it is possible to determine the initial selectivities to gasiﬁcation and to aromatisation. Interestingly, it can be seen on Fig. 5 that the intrinsic rates of steam reforming ﬁt well with the kinetic equation developped for the toluene steam dealkylation (16, 17). In Fig. 5, the ratio TOF/D20xm is remarkably constant when the metal loading in the catalysts (xm) was varied from 0.1 to 5% with metal dispersion decreasing from 100% (0.1%Rh) to 45% (5%Rh). Aromatisation is likely to occur on metal sites exclusively: no correlation can be found between TOF of this reaction and the parameter representative of a bifunctional metal/oxide reaction. The role of the support was conﬁrmed by Muraki and Fujitani95 who proposed the following rate equation for n-heptane reforming over Rh/ MgAl2O4 catalysts: Kc Pc KW PW r¼k 1 þ Kc Pc 1 þ KW PW

ð27Þ

Catalysis, 2010, 22, 1–55 | 17

25

TOF/A

20 Steam reforming

15 10 5

Aromatisation 0 0

1

2

3

4

5

6

%Rh Fig. 5 Reaction of steam with n-heptane at 550 1C on Rh/Al2O3 catalysts: comparison of steam reforming and aromatisation reactions

where subscript C refers to n-C7 (adsorbed on metal sites) and W to water (adsorbed on support sites). Wang and Gorte investigated the steam reforming of diﬀerent fuels over a Pd/CeO2 catalyst (see Section 2.2.1.5 fuel reforming).39 The same catalyst was proved to be more active and more stable than Pd/Al2O3 in the steam reforming of n-butane with a remarkable CO2 selectivity even for a relatively low H2O/C ratio.96 The speciﬁc role of ceria in steam reforming was linked to its redox properties, which led Wang and Gorte to extent to higher hydrocarbons the cycles (27)–(30) already proposed for ceria in the steam reforming of methane.97 CH4 þ s ! CHx;ads þ ð4 xÞHads

ð28Þ

H2 O þ Ce2 O3 ! 2CeO2 þ H2

ð29Þ

2Hads ! H2 þ s

ð30Þ

CHx;ads þ 2CeO2 ! CO þ ðx=2ÞH2 þ Ce2 O3 þ s

ð31Þ

where s represents a metal site. Steam is a cause of severe metal sintering in SR or ATR processes.98 The good performances of ceria as a support of steam reforming could be improved by doping the support with other oxides such as Gd2O3: very stable performances (for 160 h) were observed in iso-butane steam reforming on a Pt-Ce0.8Gd0.2O1.9 catalyst.99 Coke resistance of steam reforming catalysts is a key-point for aromatic processing and also for light alkane gasiﬁcation. A remarkable improvement of coke resistance of Ni catalysts in propane steam reforming was reported by Takenaka et al.100 Three Ni catalysts (Ni/MgO, Ni/Al2O3 and Ni/SiO2) were prepared by conventional impregnation of pre-formed supports while a special catalyst was prepared by a microemulsion route including both the Ni precursor, TEOS and hydrazine. The resulting material prepared by microemulsion (named coat-Ni) consists in Ni particules of homogeneous size coated with an external layer of silica (Fig. 6). 18 | Catalysis, 2010, 22, 1–55

Fig. 6 TEM images of 10 wt-%Ni/SiO2 (a) and coat-Ni (10 wt-%) (b). Reprinted with permission from ref. 100.

The strong interaction of Ni metal particles with silica, in the silica-coated Ni catalysts, prevents the sintering of Ni metal and carbon deposition during the steam reforming of propane: the rate of coke formation at 600 1C was decreased by a factor 2 to 3 on coat-Ni with propane conversion approaching 100%. Alumina was often used as support of Ni in alkane steam reforming. Ni, Fe and Co are metals able to form carbides, precursors of carbon ﬁlaments, when they react in a hydrocarbon atmosphere.101,102 Zhang et al. succeeded in preventing sintering and coke formation by using a one-step sol-gel preparation.103 Nickel nitrate and aluminium tri-sec butoxide dissolved in ethanol were used as precursors. SG catalysts were compared to conventional catalysts prepared by impregnation. SG samples are characterized by smaller Ni particles strongly bound to alumina surface. The formation of carbon ﬁlaments requiring a detachment of Ni particles from the surface104–106 is signiﬁcantly slowed down in SG catalysts. 2.2.3 Methane steam reforming (MSR). Natural gas reforming is the main source of hydrogen in the World. Annual production od H2 is estimated to 400 billions m3, North America being the most important producer (230 billions m3).107 About 60% come from narural gas reforming and the major part of rest from petroleum reﬁnery or naphtha reforming. For this reason, methane steam reforming has been the subject of hundred of papers and many reviews or monographies (see for instance thoses of Rostrup-Nielsen,15,16 Ross108 and Inui).109 Only the studies dealing with mechanistic considerations will be reviewed here. The central question in this section is: does methane reforming obey a speciﬁc mechanism, diﬀerent of that proposed for heavier hydrocarbons? 2.2.3.1 MSR mechanism with CH4 activation as rate determining step. Bodrov and Apel’baum were among the ﬁrst authors to show similarities between the mechanisms of steam and dry reforming of methane (32).110 They suggested that CO2 was shifted to CO and water which could be the main (or unique) reactant of methane gasiﬁcation. CH4 þ CO2 ! 2CO þ 2H2

ð32Þ

Catalysis, 2010, 22, 1–55 | 19

Table 14 Turnover frequency (molec. site 1 s 1) of MgO-supported metals in steam and dry reforming of methane. From Rostrup-Nielsen and Bak Hansen111

Catalyst

Metal area (m2 g 1)

H2O-CH4 reforming 550 1C

CO2-CH4 reforming 550 1C

H2O ref/CO2 ref ratio

CO2-H2 reverse shift 500 1C

1.4% Ni 1.4% Ru 1.1% Rh 1.2% Pd 0.9% Ir 0.9%Pt

1.1 3.0 2.2 1.4 1.3 1.0

2.2 8.9 8.1 1.6 4.5 2.0

1.9 2.9 1.9 0.18 0.44 0.36

1.2 3.1 4.3 8.9 10.2 5.6

5.3 8.7 5.4 8.0 8.6 7.2

In 1993, Rostrup-Nielsen and Bak Hansen showed that methane steam reforming (MSR, 1) and methane dry reforming (MDR, 32) had relatively close rates on most metal catalysts (Ni, Ru, Rh, Pd, Ir, Pt) supported on magnesia.111 Turnover frequencies were compared at 550 1C in the following conditions at the reactor inlet: 18.5 vol-% CH4, 74 vol-% CO2 or H2O and 7.5 vol-%H2 (Table 14). The CO2-H2 reverse shift reaction was also investigated at 500 1C in similar conditions without methane in the inlet gases. Except for Ni, the data of Table 14 seem not to support the conclusion of Rostrup-Nielsen and Bak Hansen since the steam reforming reaction would be 3 to 10 times faster than dry reforming on the other metals. RostrupNielsen and Bak Hansen explain this apparent discrepancy by the CO inhibition of both reactions. As dry reforming produces much more CO, the reaction would be more aﬀected by this inhibition. This explanation was supported by a clear relationship between the CO inhibition factor and the adsorption heat of CO on metal. Moreover, all metals being very active in RWGS reaction (Table 14, last column), the tendency to form CO in dry reforming would be the same for all the catalysts. The work of RostrupNielsen and Bak Hansen was qualitatively conﬁrmed by Qin et al.112,113 who showed that steam and dry reforming had the same types of reaction intermediates. In 2004, Wei and Iglesia published a series or papers on the crucial role of the CH4 activation step both in methane steam reforming and in methane dry reforming (MDR) on Ni,114 Rh,115 Ir,116,117 Ru,118 and Pt.119,120 The main conclusions of these studies at 600 1C are: For all the metals, the reactions (MSR and MDR) are of ﬁrst-order in CH4 and virtually of zero-order in H2O and CO2. For all these metal, the turnover frequency of the steam reforming is very close to that of dry reforming, suggesting that neither H2O activation nor CO2 activation intervenes in the rate determining steps of methane conversion. These observations were corroborated by isotopic exchange studies including the use of CD4, 13CO and D2. The comparison of CH4 and CD4 reforming reveals a kinetic isotopic eﬀect k(C-H)/k(C-D) close to 1.5 for all the metals. Moreover, the rates of CH4/CD4 cross-exchanges are signiﬁcantly smaller than those of reforming reactions. All these features suggest

20 | Catalysis, 2010, 22, 1–55

that the steps including C–H breaking in methane are irreversible and are the slow steps of the mechanism for both MSR and MDR. Reactions with CH4/CO2/D2 and CH4/CO2/13CO mixtures show that H/D distributions in H-containing products (including water) and 12C/13C distributions in COx are very close to equilibrium. These results conﬁrm that both water and CO2 activation are fast processes and as a rule, the water gas shift reaction is equilibrated on all the metals at 600 1C. Wei and Iglesia also found a strong particle size eﬀect: turnover freqencies in methane reforming generally increased when decreasing the cluster size of metal. This was ascribed to the creation of a high number of low-coordination metal sites on small particles favoring the C–H cleavage of the methane molecule. Finally, they showed that there was a moderate support eﬀect if any: for instance, TOF are very close when Rh or Pt are supported on Al2O3, ZrO2 or CeO2-ZrO2.115,119 The crucial role of CH4 activation in methane steam reforming on Rh/aAl2O3 was conﬁrmed by Tavazzi et al.121 and Donazzi et al.122,123 These kinetic studies were carried out in an annular reactor avoiding any mass and heat transfer artifact. Maestri et al. used the experimental data of Donazzi to establish a hierarchy between the diﬀerent kinetic models of methane conversions (oxidation, steam reforming, dry reforming) by a microkinetic approach including water-gas shift and reverse water gas shift reactions.124 The main steps involved in methane conversions would be: Methane activation steps (methane pyrolysis) CH4 þ 2 ! CH3 þ H

ð33Þ

CH3 þ ! CH2 þ H

ð34Þ

CH2 þ ! CH þ H

ð35Þ

CH þ ! C þ H

ð36Þ

Water or CO2 activation steps H2 O þ 2 ! OH þ H

ð37Þ

CO2 þ 2 ! CO þ O

ð38Þ

CO2 þ H þ ! CO þ OH

ð39Þ

Carbon oxidation steps C þ OH ! CO þ H

ð40Þ

C þ O ! CO

ð41Þ

CO þ OH ! CO2 þ H

ð42Þ

This kinetic scheme is completed by CO, CO2 desorption and H recombination and desorption steps. From the microkinetic data, Maestri

Catalysis, 2010, 22, 1–55 | 21

et al. concluded that CH3* decomposition (34) would be the slow step of methane activation and very likely the rate-determining step of the whole process. They also showed that all the steps including OH* are more probable than those including O*: for instance CO2 would be activated via step 39 and not via step 38. The most abundant surface species are CO* and H*. It is shown that CO coverage is signiﬁcantly higher in dry reforming than in steam reforming while the reverse is expected for H coverage. This strengthens the hypothesis of a CO inhibition aﬀecting more dry reforming than the steam reaction. However, the impact of CO2 in methane conversion remains questionnable. Donazzi et al. concluded that CO2 would not directly intervene in the kinetic of dry reforming which would be a combination of reverse water gas shift and steam reforming.123 Xu and Saeys conﬁrmed that carbon species would be the most important intermediates in MSR on Ni(111).125 Subsurface carbon signiﬁcantly increases the methane dissociation barrier, thus decreasing the rate of MSR reaction, while boron promoter has the reverse eﬀect.126 2.2.3.2 MSR mechanisms with rate-determining steps including O species. The works of Donazzi et al. and of Maestri et al. carried out on Rh catalysts were tentatively extended to diﬀerent metal/zirconia catalysts by Jones et al.127 A complete picture of the reactivity was established at 500 1C by combining thermodynamic and kinetic models. The reaction is structuresensitive with turnover frequencies increasing with dispersion, suggesting that the reaction is dominated by step and corner sites. For a given dispersion (40%) the following trend was observed: Ru E Rh W Ni E Ir E Pt E Pd while thermodynamic and DFT calculations led to the following ranking: Ru W Rh W Ni W Ir W Pt E Pd, in excellent agreement with experimental results. It was conﬁrmed that the critical steps having the highest energy barriers on most metals were CH4 dissociation (33–36) and CO formation (41). Contrary to the conclusions of the previous studies, CO formation may be the rate determining step on most metals at 500 1C. However, there is clear tendency that CH4 dissociation becomes predominent at higher temperatures, in agreement with the conclusions of Wei and Iglesia. A similar picture is proposed by Blaylock et al. who showed that CH* would be the most important C-containing intermediate over Ni(111) while CHO* and CHOH* cannot be excluded even though the formation of these intermediates would be strongly dependent on the reaction conditions.128 Earlier studies in the 1970–1980’s have already proposed formaldehyde as an intermediate in methane steam reforming according to the reactions:108,129 CHx þ H2 O þ x ! CH2þx O þ x ! CH2 O þ xH

ð43Þ

CH2 O ! CO þ H2 þ

ð44Þ

Formaldehyde was detected in methane steam reforming over Ni/Al2O3 above 400 1C with a maximum around 600 1C, which proves that it is not only a key-intermediate at low temperature.129 22 | Catalysis, 2010, 22, 1–55

The role of oxygen species has been evoked in many studies using ceria as a support. Craciun et al. have investigated the steam reforming of methane over ceria-supported Pd, Rh and Pt catalysts between 350 and 550 1C.130 They found that all metal supported on ceria exhibited virtually the same activity, 4 to 5 orders of magnitude higher than the activity of the silicasupported catalyst. A similar result was observed over Pd-CeO2-Al2O3 on which MSR reaction rates is two orders of magnitude higher than on PdAl2O3.131 Oxygen species from the ceria support seem to play a major role in this catalysis.

3.

Steam reforming of ethanol

Alcohols are very reactive molecules whose decomposition over catalyst surfaces (or in gas phase) is much faster than with hydrocarbons. In the presence of steam, the reaction stoichiometry of alkanol steam reforming (ASR) is: Cn H2nþ1 OH þ ðn 1ÞH2 O ! nCO þ 2nH2

ð45Þ

Coupled with the WGS reaction, ASR may give carbon dioxide and hydrogen: Cn H2nþ1 OH þ ð2n 1ÞH2 O ! nCO2 þ 3nH2

ð46Þ

The same reaction with alkanes (HSR) is: Cn H2nþ2 þ nH2 O ! nCO þ ð2n þ 1ÞH2

ð47Þ

In the alkane series, the C1 compound (methane) is the most stable one. Contrary to hydrocarbons, the C1 alkanol (methanol) is the most reactive alcohol. It decomposes spontaneously at relatively low temperatures without water in the reacting gases (n=1 in 45). For this reason, methanol is considered as a ‘‘liquid’’ syngas, much easier to transport than the syngas itself. The Gibbs free energy of reaction (45) per carbon atom (i.e. per mole of CO þ 2H2 produced) is: DG0T ¼ n1 ½DG0CO;T ðn 1ÞDG0H2 O;T DG0A;T for alkanols

ð48Þ

while that of reaction (47) will be: DG0T ¼ n1 ½DG0CO;T nDG0H2 O;T DG0HC;T for alkanes

ð49Þ

A comparison of the Gibbs free energy of the steam reforming reaction at 25 1C (298K) and 427 1C (700 K) is reported in Table 15 for n=1 to 4. The steam reforming reaction is more facile on alcohols than on corresponding alkanes. Data of Table 15 also conﬁrmed that the thermodynamic tendency is in the following order: C1 W C2 W C3 W C4 for alcohols and in the reverse order for alkanes. However, the global reaction scheme is much more complex than the simple reactions giving CO and H2. Replacing CO by CO2 does not change Catalysis, 2010, 22, 1–55 | 23

Table 15 Gibbs free energy (kJ mol 1) of the steam reforming reaction at 25 1C and 427 1C for C1-C4 alkanols and C1-C4 alkanes. DG are given per mole of carbon Alkanol

DG0298

DG0700

Alkane

DG0298

DG0700

Methanol Ethanol Propan-1-ol Butan-1-ol

25.21 61.00 69.40 71.80

69.78 47.76 26.42 24.52

Methane Ethane Propane Butane

142.07 107.72 99.09 95.56

37.42 12.29 3.12 0.66

the thermodynamic tendency but, as seen later, the steam reforming reactions can also produce methane, which modiﬁes signiﬁcantly the thermodynamic equilibrium, CH4 being the compound the most diﬃcult to reform. 3.1

Thermodynamic of ethanol steam reforming (ESR)

Ethanol-steam mixtures can give rise to numerous reactions, the most important being: 1. The steam reforming leading to CO and H2: DH0298 ¼ þ255 kJ mol1

C2 H5 OH þ H2 O ! 2CO þ 4H2

ð50Þ

2. The steam reforming leading to CO2 and H2: C2 H5 OH þ 3H2 O ! 2CO2 þ 6H2

DH0298 ¼ þ173 kJ mol1

ð51Þ

DH0298 ¼ 157 kJ mol1

ð52Þ

3. The hydrogenolysis to methane C2 H5 OH þ 2H2 ! 2CH4 þ H2 O

4. The ethanol dehydration to ethylene DH0298 ¼ þ45 kJ mol1

C2 H5 OHþ ! C2 H4 þ H2 O

ð53Þ

5. The dehydrogenation to acetaldehyde DH0298 ¼ þ68 kJ mol1

C2 H5 OHþ ! CH3 CHO þ H2

ð54Þ

6. The cracking to methane, CO and H2 DH0298 ¼ þ49 kJ mol1

C2 H5 OH ! CH4 þ CO þ H2

ð55Þ

7. The cracking to methane and CO2: 1 3 C2 H5 OH ! CO2 þ CH4 2 2

DH0298 ¼ 74 kJ mol1

ð56Þ

8. The cracking to carbon, CO and H2: C2 H5 OH ! C þ CO þ 3H2

DH0298 ¼ þ124 kJ mol1

ð57Þ

9. The cracking to carbon, water and H2: C2 H5 OH ! 2C þ H2 O þ 2H2

DH0298 ¼ 7 kJ mol1

ð58Þ

10. The cracking to carbon, methane and water: C2 H5 OH ! CH4 þ C þ H2 O 24 | Catalysis, 2010, 22, 1–55

DH0298 ¼ 82 kJ mol1

ð59Þ

80% 70% H2 60% CH4

Mole %

50% 40%

H2O

CO

30% 20%

CO2

10% 0% 100

300

500

700

900

T (°C) Fig. 7 Equilibrium composition of gases in the reaction of ethanol with steam (without carbon formation). Initial state: H2O/EtOH molar ratio=1, P=1 bar.

The equilibrium composition of the gases with a water/ethanol inlet stoichiometry of 1 (corresponding to reaction 50) is shown in Fig. 7. Except carbon, all compounds present in equations 50–59 are included in the thermodynamic calculations but acetaldehyde and ethylene are never favored. Moreover, thermodynamics predicts that ethanol should be totally converted in the whole range of temperature. At low temperatures (100– 300 1C), the cracking into methane and carbon dioxide (56) is thermodynamically favored. Hydrogen and CO contents progressively increase with temperature; at 900 1C, the steam reforming to H2 and CO (50) is the only reaction to occur with a H2-to-CO molar ratio of 2. All the reactions leading to methane are exothermic and are thus favored at low temperature. As for heavy hydrocarbons, methane formation can severely limit the hydrogen yield. If methane was not formed in ESR, the equilibrium composition would change drastically (Fig. 8A). Ethanol is then not totally converted below 300 1C and hydrogen is formed as of the lowest temperatures (below 100 1C). The reaction favored at low temperature is the steam reforming to CO2 and H2. Carbon dioxide is very brieﬂy observed and disappears by 350–400 1C. Above this temperature, the steam reforming to CO and H2 would be the unique reaction observed in ESR. Finally, the formation of carbon may also change the reaction thermodynamics. Equilibrium compositions were calculated with possible formation of all the compounds of equations 50–59, including carbon (Fig. 8B). Carbon and methane can be formed at low temperatures with a molar ratio of 1 suggesting that the reaction favored at low temperature is the ethanol cracking to methane, carbon and water (59). An interesting fact is that methane decreases more rapidly than in absence of carbon (compare Fig. 7 Catalysis, 2010, 22, 1–55 | 25

80%

80% H2

70%

70%

50% EtOH H2O

40% 30%

Mole %

Mole %

H2

60%

60%

CO

50%

H2O

40% 30%

CO

C

20%

20%

10%

10%

CO2 0% 100 300

500

700

900

T (°C)

(A)

0% 100

CH4

300

CO2 500

700

900

T (°C)

(B)

Fig. 8 Equilibrium composition of gases in the reaction of ethanol with steam. Initial state: H2O/EtOH molar ratio=1, P=1 bar. A: without methane or C formation; B: with methane and carbon formation

and 8B) while the carbon itself can be produced over a large range of temperature. One should keep in mind that the initial state (H2O/EtOH=1) strongly favors carbon formation. Most of experimental works were performed with a H2O/EtOH ratio of 3 or higher. In this case, equilibrium gas composition are closer to that of Fig. 7 even if carbon can ever be formed. Equilibrium compositions were also calculated for diﬀerent H2O/EtOH molar ratio R and at higher pressures P. The results are reported in Table 16 for two temperatures: 600 and 700 1C. H2 yield (YH) is deﬁned as the number of moles of hydrogen produced per mole of ethanol entered: YH ¼ n G

%H2wet 100

ð60Þ

Table 16 Equilibrium gas compositions, total number of moles of gas (nG) and H2 yield (YH) in the steam reforming of ethanol at 600 and 700 1C for diﬀerent values of R=H2O/EtOH and diﬀerent values of the total pressure P (bar) T (1C)

P (bar)

R

600

1

1

600

1

2

600

1

5

600

5

1

700

1

1

700

1

2

700

1

5

700

5

1

Wet % Dry % Wet % Dry % Wet % Dry % Wet % Dry % Wet % Dry % Wet % Dry % Wet % Dry % Wet % Dry %

26 | Catalysis, 2010, 22, 1–55

H2O

H2

CO

CO2

CH4

nG

YH

10.75 – 18.29 – 33.89 – 18.87 – 4.76 – 11.59 – 30.83 – 12.18 –

41.81 46.85 46.15 56.48 44.81 67.80 25.55 31.49 56.40 59.21 58.19 65.82 49.05 70.91 40.80 46.47

15.89 17.81 11.62 14.22 6.18 9.35 7.71 9.51 27.07 28.42 20.93 23.67 10.08 14.57 18.16 20.68

10.39 11.64 11.71 14.33 11.89 17.98 14.49 17.86 3.51 3.68 6.40 7.24 9.73 14.07 8.33 9.49

21.15 23.69 12.22 14.96 3.21 4.86 33.37 41.12 8.26 8.68 2.88 3.26 0.31 0.45 20.51 23.36

4.22

1.76

5.62

2.59

9.39

4.21

3.60

0.92

5.15

2.90

6.62

3.85

9.94

4.86

4.25

1.73

The main conclusions are: H2 and CO2 contents in dry gases increase with R while CO and CH4 contents decrease. Though a maximum of the %H2 in wet gases could be observed, the H2 yield always increases with R and tends to 6, the maximal value of R corresponding to the stoichiometry of equation 51. In the meanwhile, the amount of unreacted water increases with R, which implies that the energy balance passes through an optimum when R is increased (the better H2 yield is compensated by the energy needed for gasifying a higher amount of water). Increasing the total pressure has a dramatic eﬀect on the H2 yield which decreases by a factor 1.7–1.9 when the reaction is performed at 5 bar instead of 1 bar. This is essentally due to a strong increase of the methane yield at the expense of the CO yield. Increasing P has a moderate, positive eﬀect on the CO2 yield. Of course, diluting the reactants (H2O þ EtOH) in an inert gas has exactly the reverse eﬀect: increase of YH and correlative decrease of YCH4. In conclusion, in those studies where the ethanol steam reforming is carried out at a high R value and/or with dilute reactants, the catalyst performances are artiﬁcially improved and the results should be considered with circumspection. When C1 compounds are possibly formed, thermodynamics does never favor dehydration or dehydrogenation of ethanol. And yet, there are many indications in the literature that these reactions on alcohols can occur at relatively low temperature on acid, basic or metal sites. To know the thermodynamic tendency of each reaction, calculations were made in absence of C1 compounds. Dehydrogenation (ethanol/acetaldehyde/H2) was ﬁrst considered, then dehydration (ethanol/ethylene/water) and ﬁnally the two reactions together. The results are shown in Fig. 9. Dehydrogenation (Fig. 9A) is favored at higher temperatures than is dehydration (Fig. 9B): while total conversion of ethanol can be reached by dehydration at less than 200 1C, dehydrogenation requires almost 400 1C. It should be noted that acetaldehyde and hydrogen, on one hand and ethylene and water on the other hand are produced in same amount: the mole percentage of each product is 50% at full conversion of ethanol (Fig. 9A and 9B). When the two reactions are possible, dehydration is strongly favored with a maximal production of ethylene around 200 1C while that of acetaldehyde increases only slowly with temperature (Fig. 9C). As this will be discussed later, dehydration occurs mainly on acid sites while dehydrogenation is catalyzed preferably on basic or metal sites. To counterbalance the thermodynamic trend, it should be necessary to eliminate virtually all the acid sites of the catalyst if one wants avoiding ethylene formation for kinetic or mechanistic reasons. Ethanol may also give rise to many other products: diethyl ether, acetic acid, acetone, n-butanol.132,133 The richness of possibility of ethanol decomposition over acid, basic and metal sites makes that numerous surface species can be detected according to the nature of metal and supports.

Catalysis, 2010, 22, 1–55 | 27

100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%

100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%

Mol %

Mol %

Ethanol

Acetaldehyde/H2

0

200

400

600

800

T (°C)

(A)

Ethanol Ethylene/H2O

0

200

400

600

800

T (°C)

(B)

100% 90% 80%

Ethanol

Mol %

70% 60%

Ethylene

50% 40% 30% 20%

Acetaldehyde

10% 0% 0 (C)

200

400

600

800

T (°C)

Fig. 9 Equilibrium composition in ethanol dehydrogenation (A), in ethanol dehydration (B) and in both reactions (C).

3.2

Noble metals

3.2.1 Rhodium catalysts. Rhodium catalysts were early recognized as very active in ethanol steam reforming.134,135 Cavallaro, Freni and the Group of Messina were among the ﬁrst authors to investigate ethanol steam reforming over Rh/Al2O3 catalysts.136–139 The reaction was carried out at temperatures between 50 and 650 1C with a standard H2O/EtOH ratio of 4.2–8.4136,137 with or without O2 addition for autothermal process.138,139 These Authors concluded that the mechanism starts with ethanol dehydrogenation and/or dehydration followed by the gasiﬁcation of acetaldehyde or ethylene intermediates. Ethylene would be formed on acidic sites of alumina while all other steps (including dehydrogenation and gasiﬁcation) would be catalyzed by the metal. Catalyst coking seems linked to ethylene formation and can be largely suppressed by O2 addition.138,139 Interestingly, it was proved that methane is a primary product whose selectivity decreases with contact time (Table 17). A comparison of Rh with other metal catalysts was performed by Aupretre et al.,140 Breen et al.141 and Liguras et al.142 The catalysts and the conditions used in these studies are reported in Table 18. Catalysts performances are given in the table as H2 yield or ethanol conversion. The total conversion of ethanol being very high in their conditions, Breen et al. 28 | Catalysis, 2010, 22, 1–55

Table 17 Ethanol conversion, C1-product selectivity and hydrogen yield YH in ethanol steam reforming at 650 1C (R=4.2 with 9.6% EtOH, 80.4% H2O and 10% N2). Catalyst: 5%-wt Rh/ Al2O3 (Rh crystallite size: 8-9 nm). From ref. 138 C1-selectivity (%)

Contact time (s)

Ethanol conversion %

CO2

CO

CH4

YH (mol H2/mol EtOH)

0.120 0.048 0.033 0.020

100.00 100.00 80.59 55.59

69.26 55.60 49.35 47.04

27.41 32.00 33.69 35.37

3.33 12.40 14.00 17.59

5.17 4.34 3.43 2.30

Table 18 Comparison of metal catalyst performances in ethanol steam reforming. Unless otherwise indicated, all metals are deposited over alumina supports Aupretre et al.140

Breen et al.141

Liguras et al.142

700 1C, R=3 No dilution

400–750 1C, R=3. Dilution in N2. H2O/EtOH/N2=3/1/7.6

600–850 1C, R=3 No dilution

Catalyst

H2 yield (g h 1 g 1cat) Catalyst

1%Rh 1%Pt 0.75%Pd 0.67%Ru 9.7%Ni 9.1%Cu 1% Rh/CeZrOx 9.7%Ni/CeZrOx

2.3 0.6 1.1 0.3 3.1 0.4 5.1 4.4

1%Rh 5%Ni 0.5%Pd 1%Pt

CCOX (Useful conversion) at 700 1C (%)

Catalyst (% dispersion Conversion in parenthese) at 750 1C (%)

90 15 30 15

0.5%Rh (62) 1%Rh (45) 2%Rh (50) 1%Ru (14) 3%Ru (22) 5%Ru (21) 1%Pd (39) 1%Pt (98)

52 79 91 18 69 70 25 32

deﬁned a ‘‘useful’’ conversion CCOX in steam reforming corresponding to the partial conversion of ethanol in H2 via equations 50, 51: CCOX % ¼

1 ½CO þ CO2 out 100 2 ½Ethanolin

ð61Þ

The performances of the alumina-supported metals can be ranked as follows: NiWRh c PdWPtWCuWRu, according to Aupretre et al.140 Rh c PdWNiEPt, according to Breen et al.141 Rh c PtWPdWRu, according to Liguras et al.142 As a rule, Rh appears as the most active metal but interesting performance can be achieve over highly loaded Ni140 or Ru142 catalysts. The product distribution and the stability of Rh catalysts can also depend on the metal precursor used for the catalyst preparation.143,144 A comparison of MgO-supported metals was also carried out by Frusteri et al. at 650 1C.145 The results reported in Table 19 conﬁrm the very good Catalysis, 2010, 22, 1–55 | 29

Table 19 Comparison of MgO-supported Rh, Ni, Pd and Co catalysts in ethanol steam reforming. T=650 1C, R=4.2 (H2O/EtOH/N2=4.2/1/3), GHSV: 300000 h 1. EtOH conversions are between 7 and 20% C-product distribution (vol-%)

Catalyst

BET area (m2 g 1)

Metal dispersion %

CO

CO2

CH4

CH3CHO

TOF (s 1)

3%Rh 3%Pd 21%Ni 21%Co

2.1 2.0 6.3 7.6

16.0 12.8 14.0 15.6

19.5 31.8 16.5 18.8

67.1 32.8 58.2 44.4

8.1 21.7 5.7 5.6

5.3 14.5 19.6 31.2

12.1 5.5 2.1 3.3

performance of Rh on this support. On the basis of 100 for Rh, the following ranking was obtained: Rh,100 W Pd,45 W Co,27 W Ni,17, not so far from the relative activity of these metals in toluene steam dealkylation (See Table 7). Aupretre et al.140 also showed that CeZrOx mixed oxides could be excellent supports for Rh and Ni, increasing signiﬁcantly the H2 yield. Diagne et al. investigated the hydrogen production by ethanol reforming over Rh catalysts supported on CeO2, ZrO2 and various CeZrOx oxides (Ce/Zr=4, 2 or 1).146,147 The reaction was carried out between 300 and 500 1C with a high H2O/EtOH ratio and a large dilution in Ar (H2O/EtOH/ Ar=8/1/35). These conditions strongly favor H2 yields and are not representative of real conditions. At 450 1C, YH is close to 5.7 mol H2/mol EtOH for pure zirconia and CeZrOx support at 50% Ce. Beyond 50% Ce, YH slightly decreases with %Ce to slow down to 5.0 for pure ceria. Interestingly, the oxide basicity was characterized by CO2 chemisorption and the determination of the adsorption constant KCO2 at 0 1C. It was shown that ceria was the most basic oxide (KCO2=0.7 kPa 1) and that KCO2 decreased linearly with the Zr/Ce ratio down to 0.25 kPa 1 for pure zirconia. Diagne et al. concluded that the hydrogen yield is not favored by a high basicity of support. Rh/CeZrOx catalysts were also studied by De Rogatis et al.148 and the group of Wang at Richland.149–151 In spite of diﬀerent experimental conditions (R=5 and dilution in Ar for De Rogatis et al.; R=4, no dilution for Wang et al.), the conclusions of these studies were rather similar: While alumina or Al-spinel supports favor ethanol dehydration to ethylene,149,151 CeZrOx supports strongly favor the acetaldehyde route to COx and H2.148,149 It is not clear, however, if the CeZrOx support promotes ethanol dehydrogenation or inhibits ethanol dehydration. Once acetaldehyde is formed, CeZrOx favors its gasiﬁcation into COx and H2 instead of its decomposition into methane and CO.149 Oxygen storage capacity (and mobility) on CeZrOx would play a crucial role in the reaction mechanism, O species promoting the fast oxidation of the CH3 group of acetaldehyde, thus hindering methane formation. CeZrOx also promotes the water-gas shift reaction and favors the ﬁnal conversion of CO to CO2.148 Owing to its redox properties, CeZrOx would also favor the direct decomposition of water into hydrogen (and not only into OH groups).148,151 30 | Catalysis, 2010, 22, 1–55

This is in line with direct evidence of water dissociation into hydrogen over reduced Rh/CeO2.45 CeZrOx gives more stable catalysts by reinforcing their resistance to coking.150 This is linked to the inhibition of the dehydration route to ethylene which is a strong coke precursor and to the oxygen storage capacity promoting CHx oxidation and surface cleaning along the steam reforming process. Aupretre et al. have studied the ethanol steam reforming at 700 1C under 1 or 11bar and R=4 over 0.2%Rh/MgxNi1 xAl2O4 catalysts.152 Metal accessibility were close to 40% for all the catalysts. Tests under 11 bar were performed in view of a coupling of the reformer with a separation membrane for H2 puriﬁcation. The acid-base properties of the supports were carefully characterized by CO2 chemisorption and dimethyl-3,3-but-1-ene isomerization as well as FTIR of lutidine and DRIFT. The morphology of the support, with a Mg-deﬁcient spinel layer (thickness of about 8–9 nm) intimately covering all of the alumina grains (around 40 nm in size), can explain the neutralization of most acidic sites. Some of them are created during the metal precursor impregnation. The highest H2 yield at 1 bar was obtained with the Rh/Mg0.75Ni0.25A2O4 catalyst (4.68 mol H2/mol EtOH and 5.58 mol/mol with WGS) while the best performances under 11 bar were obtained with the nickel-free catalyst (2.74 mol H2/mol EtOH and 3.25 with WGS). Acid-base properties of alumina can be tailored by adding some oxide promoters. Recently, Can et al. have investigated the eﬀect of Sc, Y, La, Er and Gd addition to a Rh/Al2O3 in ethanol steam reforming (675 1C, 2 bar, R=4).153 The best performances were observed with the most basic catalysts presenting a new type of acid site less able to form coke precursor. Most of the previous studies report that ethanol conversion in gas phase is far to be negligible and it seems that several products of gas phase conversion are coke precursors. For instance, Aupretre et al. showed that a signiﬁcant amount of ethylene was produced when the reactor was loaded with SiC only (the ‘‘inert’’ material used as diluant of the catalyst).152 At 700 1C, 1 bar, ethanol conversion was 71% while it increased with the pressure to reach 98% at 11 bar (residence time eﬀect). The formation of undesirable by-products in gas phase can be avoided by performing the reaction of steam reforming or autothermal reforming in ultra-short residence time reactors (o10 ms). Deluga et al. reported very interesting results on this type of reactor with a catayst consisting in a Rh/CeO2 layer deposited on a ceramic foam.154 Surface species and gas phase products in the steam reforming of ethanol on TiO2 and Rh/TiO2 were studied by Rasko´ et al. by coupling FTIR and TPD-MS.155 Ethanol dissociation forms ethoxides C2H5O at ambient temperature. Dehydrogenation leads to acetaldehyde whose formation is accelerated in the presence of rhodium and steam. The same study was extended to other metals and to other supports (alumina, ceria, . . .).156 At low temperature, ethanol is molecularly adsorbed on oxides while it forms ethoxides and proton on supported noble metals. Heating these adsorbed species above 150 1C progressively leads to acetate species (on the support) or to their decomposition into hydrogen and carbon oxides (at metal/ Catalysis, 2010, 22, 1–55 | 31

support interface). Certain acetate species are so strongly adsorbed that they may deactivate the catalyst. Dehydration to ethylene is observed only on alumina and virtually not on ceria. Surface species issued from ethanol adsorption at room temperature and their decomposition by HT-treatment were investigated by Idriss over ceria-supported Rh, Pd, Pt and Au monometallics as well as over Pt-Rh, Rh-Au, Rh-Pd and Pt-Pd bimetallics.157 O–H bond dissociation leading to ethoxy species can be observed on Rh, Pt and Pd. In the meanwhile, and only on Rh, a C–H bond dissociation of the CH3 group leading to oxometallate species is also observed. As a result, Pd and Pt may produce signiﬁcant amounts of acetaldehyde while the C–C bond dissociation, leading to CO and methane, is favored on Rh. From all the studies carried out over Rh catalysts, a global picture of the reaction mechanism as illustrated in Fig. 10 can be proposed.

H2O

CH3CH2OH

CH3CHO

H2, CO, CO2, CH4 C2H4

CH3CH2OH

OH OH Metal Basic site

Acid site Support

Fig. 10 Mechanism of ethanol steam reforming. Ethanol is either dehydrogenated into acetaldehyde over basic (or metal site) or dehydrated into ethylene over acid sites. Acetaldehyde is then decomposed in CO þ CH4 þ CO2. Methane can be reformed at elevated temperature while the CO/CO2 mixture is shifted to more or less CO2 according to the temperature and the R ratio. Ethylene can be directly reformed into syngas or hydrogenolyzed into methane. This is also a coke precursor whose formation is preferably to be avoided.

In this mechanism, methane is formed by acetaldehyde (or ethanol) cracking and is then steam reformed in a further step. It would be a primary product of ESR. To support this mechanism, it may be remarked that the actual methane concentration in gas phase is sometimes higher than the equilibrium concentration. Another route to methane, not evoked in equations 50–58 is the hydrogenation of CO or CO2. This way of methane formation was investigated by Birot et al.158 over Rh/CeZrOx ESR catalysts. They showed that methane can be produced by CO hydrogenation but not by CO2 hydrogenation whose rate is negligible in ESR conditions. 3.2.2 Other metals. Though group 10 noble metals were compared to rhodium in many studies (y 3.2.1), they were sometimes investigated alone, generally with the objective to prove their eﬃciency when deposited on adequate supports. Platinum. One of the ﬁrst study on ethanol steam reforming was reported by Rampe et al.159 Catalysts were Pt, Ru, Pd and Ni or bimetallics Ni-Pt and Ni-Pd on alumina. Their performances were evaluated at R=2, 3 or 4, P=2, 5 or 9 bar and T=600, 700 or 800 1C. The main objective of this study was to demonstrate the feasibility of an ethanol reformer of 3kW. 32 | Catalysis, 2010, 22, 1–55

Jacobs et al. have investigated the Pt/CeO2 system in ethanol steam reforming at 200–550 1C in diluted medium (H2O/EtOH/N2 þ H2/= 33/1/29).160 Adding hydrogen to the inlet feedstream allows to maintain ceria in a reduced state. Jacobs et al. showed that ceria played a signiﬁcant role in the reaction mechanism: ethanol would be adsorbed on defect sites of ceria forming ethoxy species which, in turn, are transformed into acetate species. One of the role of the metal would be to hydrogenate surface species into carbon dioxide and methane. In this mechanism, CO2 is the primary COx product. Unfortunately, as Pt/CeO2 is a good WGS/RWGS catalyst, a part of the CO2 is shifted to CO with a loss of hydrogen. Other studies were performed by the Group of Davis at Lexington and the Group of Noronha at Rio de Janeiro over Pt/CeZrOx catalysts.161,162 A wide range of by-products were identiﬁed. The catalyst is mainly deactivated by carbon which creates a barrier at the metal/support interface, thus inhibiting acetate decomposition and hydrogenation. Co-feeding oxygen with water strongly reduces coke formation and increases catalyst stability at the expenses of H2 yield.162 Platinum was also used to improve Ni stability by decreasing coke formation in Ni-Pt bimetallic catalysts.163 Ruthenium. Ru/alumina catalysts (in the form of pellets or wash-coated on cordierite monolith or on ceramic foams) were investigated by Liguras et al.164 On such structured catalysts, the control of temperature along the bed (or monolith) axis is fair (T may vary from 900 to 600 1C). However, Ru appears to be a good candidate for the steam reforming or even autothermal reforming of ethanol. Contrary to the conclusions of the previous studies on Pt, CO would be a primary product of the reaction and subsequently transformed into CO2 and H2 by WGS. As Ru is a very good catalyst for this reaction, CO and CO2 concentration would be close to equilibrium. Palladium. Ethanol steam reforming on Pd catalysts was studied by Goula et al. over Pd/Al2O3165 and by Casanovas et al. over Pd/ZnO.166 At low temperature, Goula et al.165 observe an important formation of CO and methane corresponding to ethanol (or acetaldehyde) decomposition. Increasing the temperature favors the formation of CO2 and H2. Above 500 1C, the formation of CO increases again in accordance with the WGS equilibrium. Casanovas et al. have tested their 5%Pd/ZnO catalyst between 250 and 4501C with a H2O/EtOH ratio of 13.166 They report very good activity at medium temperature (300–400 1C) with, however, a signiﬁcant formation of acetaldehyde. The major problem they encountered was the formation of PdZn alloy at the highest temperature. A two-layer ﬁxed bed reactor was proposed by Galvita et al.167 for ethanol reforming. Ethanol is ﬁrst transformed by dehydrogenation and cracking on a Pd/C catalyst in a gas mixture composed of methane, COx and hydrogen (see equations 54– 56, section 3.1). Methane is then reformed at high temperature to carbon oxides and hydrogen on a conventional Ni catalyst. Iridium. Ethanol steam reforming was investigated over Ir/CeO2 by Zhang et al.168 Cai et al.169,170 In spite of a very low H2O/EtOH ratio (R=1.8), these authors showed that their catalyst was stable over 60 h in ESR or autothermal ESR. It seems that the combination of Ir metal and ceria can prevent coke formation even at low steam concentration. Catalysis, 2010, 22, 1–55 | 33

Zhang, Cai et al. explained this result by the high oxygen storage properties of the Ir/CeO2 system and the exceptional stability of Ir in these conditions. These results are coherent with the OSC and 18O/16O exchange properties measured on similar Ir/CeO2 catalysts by Bedrane et al.55 As on most ceria-supported catalysts, Ir/CeO2 produces a signiﬁcant amount of acetaldehyde at low temperature (below 450 1C). In addition, acetone was also observed up to 500 1C by secondary reaction of acetaldehyde with water (62): 2CH3 CHO þ H2 O ! CH3 COCH3 þ CO2 þ 2H2

ð62Þ

Ethanol conversion is total above 4501C and the product distribution is close to equilibrium above 550 1C except for methane and CO2 (15% CO, 20% CO2, 7% CH4 and 58% H2 at 600 1C, to be compared to equilibrium values: 15% CO, 14% CO2, 16% CH4 and 55% H2). 3.3

Non-noble metal catalysts

3.3.1 Nickel catalysts. Nickel-based catalysts are widely used in industry for many chemical reactions. Because of its high performances and its low cost, nickel is also one of the most studied metals for ethanol steam reforming. Nickel-containing catalysts are reported to have a high activity for ethanol steam reforming. Fatsikostas and Verykios171 studied the ethanol steam reforming over nickel catalysts supported on g-Al2O3, La2O3, and La2O3/g-Al2O3. They demonstrated that Ni promoted reforming of ethanol and acetaldehyde as well as the water–gas shift and methanation reactions. At temperature below 300 1C, pure nickel causes bond breaking of ethanol in the following order: O–H, –CH2–, –C–C–and CH3, the key reaction being ethanol dehydrogenation. At elevated temperature, the overall reaction network of ethanol steam reforming results in the formation of H2, CO, CO2, and traces of CH4. Nickel is also known to present a high hydrogenation activity and it may help to combine hydrogen atoms on the catalyst surface to form molecular hydrogen.135,172 However, nickel presents a limited WGS activity, limiting the selectivity towards H2 and CO2. Consequently, the support will also play a role in the catalytic performances. The ethanol conversion as well as the selectivity to hydrogen of nickel catalysts on various support reported in the literature for temperatures higher than 550 1C are given in Table 20. The results presented in Table 20 show that whatever the experimental conditions and the nickel content, the majority of the supports lead to high selectivity to hydrogen. In fact, one of the major problem to overcome with nickel catalysts is to avoid the catalyst deactivation due to metal particle sintering as well as to coke deposition. To improve the stability of the nickel catalyst, several ways were investigated by modifying either the nature of the support or the metallic phase. Thus, various supports were studied, such as Al2O3, SiO2, MgO, MgAl2O4, La2O3, ZnO, CeO2, CeO2–ZrO2, CexTi1 xO2 or perovskite-type oxides (LaAlO3, SrTiO3 and BaTiO3).135,140,173–180,188,189 The better performances were obtained with the most basic supports favoring the ethanol dehydrogenation and inhibiting ethanol dehydration leading to ethylene, which is coke precursor. The addition of dopants to the support 34 | Catalysis, 2010, 22, 1–55

Table 20 Initial ethanol conversion and initial selectivity to hydrogen obtained on various Ni supported catalysts and diﬀerent reactions conditions (temperature, water to ethanol ratio R, in the presence or not of inert gas), at atmospheric pressure Ni content (wt.%)

Support

R

Inert gas

T

Ethanol Selectivity conversion (%) to H2 (%) Ref.

10 15 17

g-Al2O3 g-Al2O3 g-Al2O3

4 6 3

600 1C 600 1C 750 1C

100 100 100

75 87 93

185 186 187

17 17 17

La2O3 La2O3 La2O3

3 3 3

600 1C 700 1C 750 1C

93 100 100

87 95 90

188 188 187

15 10 10

La2O3-Al2O3 TiO2 Ce0.5Ti0.5O2

6 4 3

600 1C 600 1C 600 1C

100 100 100

87 86 58

186 185 189

30 30 10 17

Ce0.74Zr0.26O2 Ce0.74Zr0.26O2 ZnO MgO

8 8 4 3

600 1C 650 1C 600 1C 750 1C

98 100 95 100

88 93 80 79

178 178 185 187

17

YSZ

3

– – He, 62.5 vol.% – – He, 62.5 vol.% – – N2, 80 vol.% N2 N2 – He, 62.5 vol.% He, 62.5 vol.%

750 1C

100

92

187

can also improve the activity, the selectivity and the stability of the catalysts. For example, alkali are very often added to the support to neutralize the acid sites, inhibiting in this way the dehydration reaction leading to the formation of oleﬁns responsible for coke production.181–184 It was also shown that the addition of potassium to the support improves the stability of Ni/MgO catalysts by limiting the sintering of nickel particles without any eﬀect on coke formation rate.181 Ce, Co, Cu, Mg, Zn, Fe, Cr190–193 were also added to a Ni/g-Al2O3 catalysts for auto-thermal steam reforming. Iron is known to be an active metal for the water-gas-shift reaction thus increasing the hydrogen yield, and to relieve the carbon deposition. Moreover, iron has electronic properties and a ion radius similar to that of nickel and then it can easily replace nickel in nickel-containing phase. The iron-promoted nickel-based catalysts showed high selectivity to hydrogen as well as high resitance to both sintering and oxidation in oxidative atmosphere. The improved durability, compared to the iron-free nickel catalyst was attributed to the presence of NiAl2O4-FeAl2O4 mixed crystals.193 Copper-based catalysts are known to favor the ethanol dehydrogenation route and copper is also more active in methane steam reforming. Associated to nickel, copper increases the water-gas shift activity as well as the stability of nickel particles.194,195 Biswas and Kunzru evidenced178 that the addition of copper to Ni/CeO2-ZrO2 catalyst not only increased the steam reforming and WGS activities but also enhanced acetaldehyde decomposition and reforming. Klouz et al.190 and Fierro et al.191 also shown that the addition of Cu to Ni-based catalysts enhanced the selectivity Catalysis, 2010, 22, 1–55 | 35

towards hydrogen and improved its stability. This was related to the ability of copper of modifying the aﬃnity of nickel particles for carbon, thus inhibiting coke formation. The interaction between the Cu–Ni phase and the support plays also an important role in the complex reaction network taking place during the ethanol steam reforming. Calles et al.,196 prepared Cu-Ni bimetallic catalysts on Ce and La modiﬁed SBA15 support. They observed that on the Ce-modiﬁed support, the formation of large CeO2 particles decreased the metal-support interaction thus increasing the metal particle size, leading to a low hydrogen selectivity. On the contrary, a high metalsupport interaction, as observed with high contents of La2O3 in the modiﬁed support, leads to a better metallic dispersion as well as a lower coke formation. Another way of improving the stability of Ni-based catalysts consists of adding small amount of noble metals such as platinum197,198 or palladium198 The addition of promoters caused a decrease in the NiO reduction temperature. Moreover, the bimetallic catalysts showed a higher ethanol conversion and higher hydrogen yield than the monometallic one, whatever the nature and concentration of the noble metal. Rhodium was also added to to Ni/CeZr.199 In that case, the presence of rhodium had a negative eﬀect, the presence of rhodium favoring the rejection of nickel oxide from the support. The reduction of the Rh–Ni/CeZr mixed oxide during the catalytic test decreased the nickel-support interaction, thus favoring the carbon ﬁlament production and diminishing the oxidizing properties of the support. 3.3.2 Cobalt-based catalysts. Cobalt has also been the object of many papers on ethanol steam reforming.179,200–209 It was demonstrated that the support is of major importance for the catalytic performances in ethanol steam reforming in the presence of Co-supported catalysts. Among the supports studied, the alumina support was reported to yield the highest selectivity towards hydrogen200,207 but the basic supports, such as MgO lead to more stable catalyst, resistant to coke formation.207 The metal-oxide interaction appeared to be also essential for the stability.201 Thus, to favor Co-support interactions, ﬂuorite-type Ce-Zr-Co oxide was prepared and evaluated in real bioethanol steam reforming at 540 1C. At this temperature, the catalyst deactivated, due to the formation of carbonaceous products at the oxide surface and also of ﬁlaments. This result was explained by the reduction of cobalt cations of the mixed oxide, weakening the metal support interaction, thus favoring carbon deposition.210 The addition of Rh to this CeZrCo catalyst increased the catalyst stability by avoiding the accumulation of carbon species on the surface.211 As recently developped by Urasaki et al.,179 the Co catalysts supported on perovskites such as SrTiO3 and LaAlO3 showed higher catalytic activities and much longer–term stabilities compared to conventional Co/MgO and Co/gAl2O3 catalysts. It was inferred that the lattice oxygen in perovskites played a positive role in inhibiting the coke. 3.3.3 Copper-based catalysts. Copper is a good dehydrogenation catalyst212 and is known to present a high WGS activity. A copper-based catalyst, Cu/ZnO/Al2O3 is used for industrial hydrogen production by methanol steam 36 | Catalysis, 2010, 22, 1–55

reforming. It was demonstated that this catalyst is also active for ethanol steam reforming, yielding CO, CO2 and H2 as the main products at low temperature, above 360 1C.213 Recently,214 nickel, cobalt and/or manganese were added to a commercial Cu/ZnO/Al2O3 and the resulting catalysts were tested in ethanol steam reforming. It was demonstrated that, below 480 1C, the presence of modiﬁers leads to a decrease in the methane formation and an increase in the hydrogen yield and selectivity. At this temperature, the formation of organic by-products, excepted methane is strongly decreased. 4.

Utilisation of crude bioethanol

Ethanol has been used as an automotive biofuel ever since the introduction of the combustion engine. From 1935 to 1945, ethanol was used pure, or mixed with petrol before to be replaced by gasoline or diesel with the development of the petrochemical industry. The crisis in the 1970s revived the interest of ethanol with the running of the Brazilian National Alcohol Production Progam (PROALCOOL), in 1975. However except for this decade the price of the oil remained low making the use of other alternative energy resource hardly acceptable. It was necessary to wait for the beginning of the 21st century with the realization of two simultaneous concerns, the decreasing of the resources of the fossil fuels and the global warming to really consider that an alternative energy became essential. One way of reducing environmental eﬀects and the dependance on fossil fuels is to use renewable bioethanol as direct fuel for transportation. Another way is to use hydrogen as an energy carrier to support sustainable energy development. Hydrogen can be used in a fuel cell to generate electricity with high eﬃciency. It is extremely clean as the only by-product is water. In order to support sustainable hydrogen economy, it is crucial to produce ethanol and hydrogen cleanly and renewably. 4.1

Bioethanol production and composition

Bioethanol is an ethanol produced renewably by fermentation by yeast of biomass materials, such as sugar cane, sugar beet, wheat, potatoes, corns, and other sugar- and starch-rich plants.215 Such crops which have a food use are often referred as ‘‘ﬁrst generation’’ biomass crops. Sugar and starch crops have been utilized for bioethanol production due to the relative ease at which their constituent reducing sugar unit can be separated in water (hydrolysis) and subsequently fermented. It should be noted that that fermentation is not a ‘‘green’’ process since one C6 unit of sugar gives only two moles of ethanol, two carbons being lost as CO2. Brazil and United States produce large quantities of bioethanol from sugar cane and corn respectively. In the European Union, where France and Spain are the largest producers, the ethanol production is more modest and comes from the fermentation of sugar beet or wheat in a large part. As ethanol is produced from biomass, its environmental performances in terms of greenhouse gas (GHG) emissions and energy saving are strongly related to the raw material. As an example, depending on the crop, the ethanol yield could vary drastically. Table 21 gives a comparison of the performances obtained in terms of ethanol yields from main ‘‘ﬁrst generation’’ feedstocks. Catalysis, 2010, 22, 1–55 | 37

Table 21 Some examples of ethanol yields from various crops (from 216 and 217) Crop

Ethanol yield (hl/ha)

Sugar cane Sugar beet Wheat Corn

70 21.7–47.4 11.6–20.1 18.2–32.8

These data corresponding to the cultivation eﬃciency have a direct impact on the climate beneﬁts. Other factors inﬂuence the GHG emissions : (i) the fuel and fertilizer used in the ethanol plant; (ii) the eﬃciency with which by-product are dealt with; (iii) the type of land for cultivation.218 According to the raw material, to the agricultural yield and to the transformation process eﬃciency, the energy balance (energy used/energy produced) has been shown to vary from 0.3 to 1.6 when bioethanol is used as biofuel.217 Because of the existence of the high diversity of ways to produce ethanol, the divergences about the method to be used for calculating the level of GHG saving cannot be avoided. Diﬀerent studies can bring to entirely opposite conclusions. For Searchinger et al. who assume that a production of biofuel requires new cultivation of cropland by ‘‘displacement eﬀect’’, the production of bioethanol is considered as a threat for the environment.219 From other authors it seems clear that, under most production scenari, the net greenhouse gas eﬀect of bioethanol is positive.220 This last conclusion is usually accepted even if the data calculted in term of reduction of GHG emissions vary. ADEME reported a detailed comparison of the two ﬁrst works reported in the Table 22 in order to analyse the reasons of the divergences. One of the conclusions was that as far as normalized results are concerned the values become comparable.

Table 22 Reduction of GHG emissions for bioethanol from diﬀerent European feedstock as compared with fossil fuel emissions. From ref. 221 Source

Bioethanol from sugar crops (%)

Bioethanol from grain (%)

Ref.

Concawe/Eucar/JRC ADEME PWC

37–44 75 40–60

6 to þ 43 75 40–70

222 223 224

An other vehement debate is caused by the use of ‘‘ﬁrst generation’’ biomass: the ethical issues of using food for fuels.225 For instance, corn and sugarcane, are used to produce food and sugar as consumer goods. The recent rise in the price of various crops such as the wheat has been attributed sometimes to the greatly increased demand for biofuels in recent years by many developed nations. The use of ‘‘second-generation’’ lignocellulosic biomass, emerged to be the best option available currently. This secondgeneration bioethanol only requires inexpensive cellulosic biomass as 38 | Catalysis, 2010, 22, 1–55

feedstock, which is plentiful and easily obtainable throughout the world (wood, grasses or the non-edible parts of plants). However because of its more complex molecular structures, the production of ethanol from lignocellulosics is more diﬃcult than from sugar cane or stach-rich materials. Lignocellulose consists of three main components: cellulose, hemicellulose and lignin, the ﬁrst two being composed of chains of sugar molecules. The percentage of each component diﬀer with the nature of the feedstock (Table 23). The diﬀerence between ﬁrst- and second-generation bioethanol production is that, in the latter case, an extra step is required to hydrolyse lignocellulose biomass.226,227

Table 23 Composition of the lignocellulosic biomass from diﬀerent feedstock. From ref. 228 Biomass

Lignin (%)

Cellulose (%)

Hemicellulose (%)

Soft wood Hard wood Wheat straw

27–30 20–25 15–20

35–42 40–50 30–43

20–30 20–25 20–27

The ﬁnal step, the distillation process, remains the same for both generations of bioethanol. Important factors to consider when evaluating the use of bioethanol are fuel or feedstock is the overall economy of processes involved as well as the energy-eﬃcency from biomass to wheel. Fuel grade bioethanol needs to be water-free, thus the production requires distillation beyond the azeotrope point, and this is one of the major production costs of fuel-ethanol.229 On the contrary, ESR process requires, as mentioned previously, a signiﬁcant amount of water (50, 51), which makes the expensive distillation superﬂuous. Only a simple ﬂash distillation to some 50–70% is necessary thereby considerably lowering the production costs of bioethanol. However the use of the diverse feedstocks as well as the ﬂash distillation can pose diﬀerent challenges in a subsequent ESR catalyst, due to the variety of contaminants present in the diﬀerent bioethanol fractions. The composition of two qualities of bioethanol produced in France from sugar beet is presented in Table 24. In the crude bioethanol fraction, the main impurities are alcohols accounting for 87% of the impurities contained in crude ethanol, the most important being propan-1-ol (27%) and methyl-3 butanol-1 (27%). One can also note the presence of esters, aldehydes, acetic acid and nitrogen-containing bases. The following paragraph of this review will report a study of the eﬀect of impurities present in raw bioethanol, such as esters, aldehydes, higher alcohols or acetic acid, on the stability of the several catalysts during bioethanol steam reforming. As far as ‘‘ﬁrst generation’’ bioethanol is concerned, the major part of the contaminants will be always heavy alcohols with a variation in the composition that depends on the crop (Table 25). To our knowledge, only one group reported the composition of diverse fractions (Table 26) obtained from a ‘‘second generation’’ bioethanol and investigated the eﬀect of its utilization as a feedstock in the ESR process.231 The biomass used is the wheat straw which was mixed with enzymes and Catalysis, 2010, 22, 1–55 | 39

Table 24 Composition of rectiﬁed and crude bioethanol produced from sugar beet. Analysis performed by the alcohol distillers union (n.d. : non detected) Crop

Units

Rectiﬁed Alcohol

Raw Alcohol

Alcohol percentage

%vol. (@ 20 1C)

96.3

92.9

0.8 20 o1 o1 42.5 o0.5 o0.5 o1 n.d. o0.5 n.d. n.d. o0.5 o1

0.4 39 123 108 94 o10 581 304 o10 o10 273 582 855 1746

Total acidity (Acetic acid) Dry extract Esters Aldehydes Methanol Butan-2-ol Propan-1-ol Methyl-2 Propanol-1 Propen-2 ol-1 Butan-1-ol Methyl-2 butanol-1 Total higher alcohols Total sulphur Volatile nitrogenated bases

3

gm gm3 gm3 g m3 g m3 gm3 g m3 g m3 gm3 g m3 gm3 gm3 gm3 gm3

Table 25 Constituents met in diﬀerent bioethanol charges from several type of crops (% vol.). From ref. 230 Crops Constituents

Sugar beet

Cereal

Potatoes

Methanol n-propan-1-ol Sec-butan-1-ol Isobutan-1-ol n-butan-1-ol 2-methylbutan-1-ol 3-methylbutan-1-ol

5.1 31.7 – 16.6 – 14.9 31.7

– 9.1 – 19.2 0.2 19.0 52.4

– 16.4 – 15.9 1.2 13.6 52.9

yeast and fermented for 5 days which resulted in an ethanol concentration of around 5–6% (fraction 1 in Table 26). Fraction 2 and 3 were obtained from diﬀerent parts of the distillation Inbicon Process.232 Because the fraction 1 caramelized when evaporized, this fraction was not directly used and the authors needed to further distill in order to remove the diﬀerent sugars. Looking at the major contaminants of the fractions 2 and 3 we ﬁnd again similarities with the composition of the raw bioethanol obtained from sugar beet. The consequences of these feedstocks on the catalytic activity in ESR is reported hereafter. 4.2

Crude bioethanol steam reforming

Until now crude bioethanol has been very rarely used as ethanol source for ESR reaction, whereas it is a solution to produce hydrogen in a costeﬀective manner. The majority of the studies reported in the literature on the production of hydrogen by ethanol steam reforming were performed 40 | Catalysis, 2010, 22, 1–55

Table 26 Constituents met in diﬀerent fractions of bioethanol from wheat straw fermentation. From ref. 231

Major constituents

Other constituents (traces)

Fraction 1

Vol% Fraction 2

Mol% Fraction 3

Mol%

Ethanol Xylose

5.7 1.2

18 0.7

Ethanol Ethyl acetate

44 0.5

Glucose

1.1

0.3

1,1-Diethoxyethane 0.2

Lactate

0.5

Acetate Glycerol

0.5 0.4

Ethyl acetate Acetic acid

Ethanol 3-Methyl-1butanol 2-Methyl-1propanol 2-Methyl-1butanol Propanol Cyclopentanone Furfural 4-Hexen-1-ol 1,1-Diethoxyethane

0.3 0.1 0.05 Propanol 2-Methyl-1propanol 3-Methyl-1butanol 2-Methyl-1butanol Cyclopentanone

using pure ethanol. The steam reforming of crude ethanol diﬀers from that of pure ethanol by the fact that the impurities present in the crude ethanol feed may inﬂuence the hydrogen yield and the catalyst stability. Two types of studies on the eﬀect of the use of crude bioethanol have been performed by either using model ethanol and water mixtures containing one impurity as identiﬁed in crude bioethanol or using directly a real crude bioethanol feed. Very few studies report the use of crude ethanol for hydrogen production by steam reforming. Akande et al.233 used directly the mixture provided by Pound Maker Agventures (Canada) and obtained by fermentation of high starch feed wheat. This crude bioethanol contained, for a most part, water (86 vol.%) and ethanol (12 vol.%), and impurities such as lactic acid (1 vol.%), glycerol (1 vol.%) and traces of maltose. Crude bioethanol steam reforming was directly performed on the crude ethanol sample at atmospheric pressure and 400 1C, at a rather low weight hourly space velocity (WHSV) of 1.68 h 1. Ni/Al2O3 catalysts containing 10 to 25 wt.% of Ni and prepared by various techniques (coprecipitation, precipitation and impregnation) were used for the experiments. Whatever the nickel content and the preparation method, the initial ethanol conversion was high, between roughly 55 and 95%, and then decreased with time on stream (TOS) to stabilize at about 200 min. The same catalyst behaviour was obtained during the steam reforming of crude bioethanol from sugar beet, in the presence of a Rh/MgAl2O4 catalyst.234 In both cases, the deactivation observed at the onset of the reaction was attributed to coking. In another study, Vargas et al.235 observed also a deactivation, but only after 11h of time-on-stream. They used a real bioethanol obtained by sugar cane molasses fermentation followed a simple distillation at 60 1C. Catalysis, 2010, 22, 1–55 | 41

The solution was mainly made of water and ethanol (ethanol/water= 1/5.95 mol/mol) and contained only traces of methanol, n-propanol, n-butanol and 3-methylbutanol (isoamylic alcohol) as impurities. The reaction was also performed at atmospheric pressure, 540 1C and the real bioethanol mixture was diluted in a Ar/N2 gas mixture. The catalyst was a Ce-Zr-Co ﬂuorite-type oxide. The results obtained showed that the catalytic behavior is similar when using a water/pure ethanol mixture or a real bioethanol feed. The only diﬀerence concerned the slightly higher hydrogen yields obtained from bioethanol due to the steam reforming of the higher alcohols. Nevertheless, whatever the origin of the ethanol used for the reaction, the ethanol conversion was 100% at the onset of the reaction, remained stable during the ﬁrst 11h of TOS and then declined continuously to reach roughly 45–50% after 25 h TOS. Consequently, the hydrogen yield, which was higher than 0.3 g H2 h 1 gcat 1 at the beginning of the reaction, falled down 0.1 H2 h 1 gcat 1 after 25 h. The apparent stability observed during the ﬁrst hours of TOS may be explained by the large amount of catalyst used for this reaction, in excess compared to the one needed to reach the complete ethanol conversion. Possibly during the ﬁrst hours of TOS a part of the catalyst did not work. In this study, the deactivation was also explained by the authors by the formation of carbon during the reaction. More recently, Rass-Hansen et al.231 used a technical bioethanol produced from wheat straw. Two fractions of technical alcohol were tested in ethanol steam reforming. These fractions were obtained from diﬀerent parts of the distillation process. They contained contaminants such as ethyl acetate , 1,1-diethoxyethane, higher alcohols and cyclopentanone, the percentage of each type of impurity depending on the fraction considered. The catalysts used for this study were Ni and Ru catalysts supported on MgAl2O4 spinel. The authors observed that whatever the reaction temperature and the catalyst, the rate of carbon formation is generally slightly higher for the technical bioethanol than for a pure ethanol/water mixture. The rate of carbon formation is also the highest at the onset of the reaction. They also noticed that the presence of higher alcohols is likely to contribute to the fast deactivation of catalyst. In conclusion, whatever the origin of the bioethanol, a deactivation of the catalyst was observed during the steam reforming reaction that was attributed to the formation of carbon deposit. 4.3

Ethanol steam reforming in the presence of various impurities

As previously detailed, the main impurities present in crude raw bioethanol (92.9 vol.%) obtained from sugar beet are higher alcohols accounting for 87% of the impurities, the most important being propan-1-ol (27%) and methyl-3 butanol-1 (27%), and also esters, aldehydes, acetic acid and nitrogen-containing bases.236 In order to determine which type of impurity is responsible for the deactivation observed during the steam reforming of crude bioethanol, the impact of various impurities on this reaction was studied. For that purpose, ‘‘model’’ raw ethanol feeds were prepared by addition of 1 mol% of one impurity in ethanol. Three series of impurities 42 | Catalysis, 2010, 22, 1–55

have been studied, namely (i) molecules with four carbon atoms and different functions (butanal, diethylether, butanol, ethylacetate) (ii) molecules with acidic and basic properties (acetic acid and diethylamine) and (iii) linear or branched alcohols (methanol, propan-1-ol, butan-1-ol, pentan-1ol, isopropanol, 2-methylpropan-1-ol, 3-methylbutan-1-ol). The results obtained in the presence or not of various impurities after 8h of time on stream on a 1 wt.%Rh/MgAl2O4 catalyst with a molar water to ethanol ratio R of 4, a weight hourly space velocity of 19.5 h 1 are summarized in Table 27. These reaction conditions were chosen in order to have less than 100% of ethanol conversion (XEtOH %) to discriminate the possible eﬀects of the compounds present in bioethanol on the catalyst stability. 4.3.1 Acid and basic impurities. Table 27 shows that the presence of diethylamine favors the ethanol conversion, and slightly increases the hydrogen yield. This promoting eﬀect of diethylamine has been explained by a competition of this basic molecule with the alcohol molecules for the acidic sites.236 Diethylamine being adsorbed preferentially on the acidic sites of the support, the dehydration of ethanol on these sites is thus inhibited. The promoting eﬀect of diethylamine on ethanol conversion was explained by a modiﬁcation of the metal electronic properties resulting from an electron transfer of the free nitrogen doublet toward the metal.236

Table 27 Performances in steam reforming of ethanol with or without 1% of impurity using Rh(1%)/MgAl2O4/Al2O3 catalyst after 8 h of time on stream (T=675 1C, P=2 bar, R=4) Distribution of products (mol molEtOH 1) Impurity

H2

CO

CO2

CH4

C2H4

C2H6

CH3CHO

H2O

XEtOH (%)

Cokea

– Diethylamine Acetic acid Butanal Butanol Ethyl acetate MeOH n-C3H7OH n-C4H9OH n-C5H11OH i-C3H7OH i-C4H9OH i-C5H11OH

2.35 2.53 2.08 2.6 1.42 0.97 2.48 1.52 1.42 1.23 1.22 1.05 0.89

0.57 0.67 0.49 0.67 0.35 0.31 0.57 0.47 0.35 0.33 0.38 0.31 0.28

0.34 0.37 0.31 0.39 0.19 0.09 0.37 0.19 0.19 0.14 0.13 0.10 0.09

0.3 0.38 0.26 0.35 0.26 0.29 0.25 0.38 0.26 0.29 0.33 0.29 0.25

0.07 0.06 0.07 0.06 0.09 0.13 0.06 0.12 0.09 0.11 0.13 0.11 0.10

0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.05 0.04 0.04 0.04 0.04 0.04

0.03 0.02 0.04 0.02 0.04 0.05 0.03 0.04 0.04 0.05 0.04 0.05 0.05

3.5 3.45 3.6 3.41 3.95 4.06 3.06 3.47 3.63 3.46 3.76 3.61 3.78

78 88 72 86 64 57 78 74 65 57 70 56 53

29.5 64.0 68.0 93.3 77.4 67.0 74.0 80.0 77.4 62.0 101.0 62.5 75.0

a

In mgC.gcat 1 obtained from TPO experiments on spent catalyst after reaction.

Contrary to what is observed with diethylamine, the presence of acetic acid leads to a decrease of both ethanol conversion and hydrogen yield compared to what is obtained without impurity. Acetic acid may promote ethylene formation by increasing the acidity of the support surface and then favors the catalyst deactivation by coke deposition. The yield in ethylene observed in the presence of acetic acid is similar to that obtained in the absence of this impurity, but it is at a lower ethanol conversion, which Catalysis, 2010, 22, 1–55 | 43

means that the selectivity in ethylene is higher in the presence of acetic acid. The catalyst deactivation observed in the presence of acetic acid may also be explained by the formation of intermediate acetate species that decompose into H2, CO2 and adsorbed carbon on the surface. Then, whatever the intermediate species invoked (ethylene or acetate) the deactivation of the catalyst observed in the presence of acetic acid was explained by the formation of coke, favored in the presence of this acidic molecule.236 4.3.2 Various types of impurities with four carbon atoms. Various type of impurities identiﬁed in raw bioethanol were studied (aldehyde, higher alcohol and ester) with the same amount of carbon atoms, i.e. butan-1-al, butan-1-ol and ethyl acetate were also studied. Table 27 shows that in the presence of butanal, the ethanol conversion and the hydrogen yield are increased compared to the reference test (without impurity), as it was observed in the presence of diethylamine. On the contrary, the presence of butanol and ethylacetate strongly deactivates the catalyst, but the most deactivating impurity is diethylether since the ethanol conversion is of 64 and 57% in the presence of the alcohol and the ester, respectively. In the presence of these impurities, the yields of ﬁnal products (H2, CO, CO2, CH4) are strongly decreased, whereas the yields of intermediate products, especially ethylene and acetaldehyde are higher than that of the reference test. Then, it can be inferred that intermediate products are rapidly produced by ethanol dehydration or dehydrogenation on the catalyst but they are more slowly transformed. The presence of high amounts of ethylene may explain the catalyst deactivation by formation of carbonaceous products. The deactivation observed in the presence of butanol may also be linked to the production of butene, which is also a coke precursor, by butanol dehydration. The deactivation by the ester may be due to the hydrolysis of ethylacetate on the acidic sites of the support, yielding ethanol and acetic acid, and then to the presence of acetic acid. But, the deactivation observed is much more important than that observed in the presence of acetic acid. The deactivation may be also explained by a competitive adsorption, since it has been reported in the literature that ethylacetate is more strongly adsorbed on alumina than ethanol.237 4.3.3 Eﬀect of the alcohols. In the presence of methanol (Table 27), the ethanol conversion is not modiﬁed but the hydrogen yield is slightly increased, compared to the reference test. Methanol is easily converted by steam reforming, thus producing also hydrogen, and then its presence does not aﬀect the ethanol conversion. In the presence of the higher alcohols (more than three carbon atoms), linear or branched, the ethanol conversion and the hydrogen yield decrease when the amount of carbon atoms in the molecule is increased. This eﬀect is more pronounced in the presence of branched alcohols compared to the linear ones. It can be seen from the value presented in Table 27 that the yield in ethylene is much more important in the presence of these alcohols than with ethanol alone, thus leading to a more important amount of coke. It has been also demonstrated by studying the steam reforming of these higher alcohols234 that they are dehydrated to the corresponding oleﬁn. The oleﬁns may be then polymerized to yield coke, the coke extent increasing with the amount of carbon 44 | Catalysis, 2010, 22, 1–55

atoms in the oleﬁn. The more important deactivating eﬀect of the branched alcohols may be explained by the formation of more stable carbocations thus facilitating the oleﬁn production.210 In conclusion, except methanol, diethylamine and butanal, all the impurities identiﬁed in crude bioethanol from sugar beet lead to the deactivation of the Rh/MgAl2O4 catalyst, mainly by coke deposition. 4.4

Catalysts for crude bioethanol steam reforming

As, whatever the experimental conditions and the catalysts, the catalyst deactivation observed during the steam reforming of crude bioethanol is due to coke formation,233–235 it is of major importance to reconsider the catalyst formulation, by modifying the support and then metallic phase, in order to ﬁnd a stable catalyst able to convert crude bioethanol by steam reforming. In order to improve the catalyst stability in the presence of crude bioethanol, various rare earth elements (Sc, Y, La, Er and Gd) were added to an alumina support. Indeed, these species are known to improve the catalytic performance in steam reforming and the stability of the catalysts, by decreasing the support acidity, thus disfavoring the oleﬁn formation, and by increasing its basicity, necessary to activate water. These supports were used for depositing 1wt.% of rhodium. The characterization of this catalyst series has been described in detail by Can et al.153 Table 28 reports the results obtained during the pure ethanol steam reforming in the presence of these catalysts, compared to that of a Rh/MgAl2O4 one. All modiﬁed samples clearly show higher ethanol conversions (XEtOH), hydrogen yields and lower intermediate product yields, especially ethylene, compared to the Rh/MgAl2O4 catalyst. The highest performances were obtained with the Rh/Y2O3-Al2O3 (Table 28). Nevertheless, whatever the catalyst, the methane yield is much higher than the value at the thermodynamic equilibrium. This result may be explained if methane is considered as an intermediate product. Then, a way to improve the yield in hydrogen would be to decrease the yield in methane. For that purpose, the metallic phase was also modiﬁed by adding a second metal.236 Pd, Pt (1 wt.%) and Ni (6 wt.%) precursor salts were thus Table 28 Performances in steam reforming of pure ethanol, using Rh(1%) catalyst supported on MgAl2O4 or on alumina modiﬁed by addition of rare earth elements, after 8 h of time on stream (T=675 1C, P=2 bar, R=4) Yield (mol mol 1) Catalyst Impurity

H2

CO

CO2

CH4

C2H4

C2H6

CH3CHO

H2O

XEtOH (%)

Thermodynamic equilibrium Rh/MgAl2O4 Rh/Sc2O3-Al2O3 Rh/Y2O3-Al2O3 Rh/La2O3-Al2O3 Rh/Er2O3-Al2O3 Rh/Gd2O3-Al2O3

3.77

0.54

0.71

0.10

0.00

0.00

0.00

3.03

–

2.35 3.30 3.43 3.34 3.09 3.29

0.57 0.71 0.67 0.62 0.69 0.67

0.34 0.63 0.65 0.69 0.59 0.66

0.3 0.53 0.50 0.54 0.57 0.54

0.07 0.01 0.01 0.01 0.02 0.01

0.04 0.02 0.01 0.01 0.02 0.02

0.03 0.01 0.01 0.01 0.01 0.00

3.50 2.56 2.50 2.50 2.64 2.54

78.0 99.4 98.7 98.6 98.7 98.7

Catalysis, 2010, 22, 1–55 | 45

Table 29 Performances in steam reforming of ethanol with 1% of 2-methylpropan-1-ol as impurity, using Rh(1%)-X catalysts (X=Pt, Pd or Ni) catalysts supported on Y2O3-Al2O3, after 8 h of time on stream (T=675 1C, P=2 bar, R=4) Yield (mol mol 1) Metal phase

H2

CO

CO2

CH4

C2H4

C2H6

CH3CHO

H2O

XEtOH (%)

Rh Rh-Pt Rh-Pd Rh-Ni

3.43 3.53 3.64 3.84

0.67 0.69 0.79 0.75

0.65 0.77 0.69 0.76

0.50 0.57 0.49 0.46

0.01 0.02 0.01 0.01

0.01 0.02 0.02 0.01

0.01 0.01 0.01 0.01

2.50 2.04 2.24 2.10

98.7 93.4 98.6 98.2

coimpregnated with the Rh precursor onto the Y2O3-Al2O3 support. The results presented in Table 29 show that the presence of nickel or palladium allows one to increase the hydrogen yield, whereas platinum has no beneﬁcial eﬀect. Nevertheless, the promoting eﬀect of Ni is more pronounced than that of palladium. The better performances of the Rh-Ni catalysts may be explained either by a better water-gas shift activity or by a better methane steam reforming activity, both reactions leading to the formation of hydrogen. The yield in CO is slightly increased in the presence of Rh-Pd and Rh-Ni catalysts compared to that obtained in the presence of Rh (from 0.7 to 0.79 and 0.75 respectively), but the methane yield is much more decreased (from 0.55 to 0.49 and 0.46 respectively). It can be inferred from these results that the presence of Pd or Ni decreases slightly the activity of the catalysts for the water gas shift reaction but increases in a major extent the activity in methane steam reforming. As this reaction yields 3 molecules of hydrogen per molecule of methane converted, the highest hydrogen yield obtained with the Rh-Ni catalyst may be explained by its higher activity in methane steam reforming compared to the Rh catalysts. The Rh-Ni/Y2O3-Al2O3 was tested in the presence of crude bioethanol from sugar beet and its stability was compared to that of the Rh/MgAl2O4 catalyst in the same conditions. Fig. 11 presents the hydrogen yield vs. time on stream for these two catalysts. After 24 h of time-on-stream, the hydrogen yield is very high (3.49 mol/molethanol) with the Rh-Ni/ Y2O3Al2O3 catalyst, and the conversion is only slightly decreased from 100% at the onset of the reaction to 97% after 24 h of time-on-stream (97% of conversion compared to 100% at the beginning of the reaction). On the contrary, in the same reaction conditions, the Rh/MgAl2O4 catalyst is strongly deactivated by coke deposition, especially during the ﬁrst 2 h of reaction. 5.

Conclusions and recommendations

Although catalytic steam reforming (SR) is widely used for hydrogen production, it is not a green process since all the carbon contained in the feedstock is transformed into CO2. Replacing fossil fuels (mainly natural gas today) by biofuels may circumvent this drawback. Bioethanol is an excellent candidate in the perspective of a hydrogen-based economy. 46 | Catalysis, 2010, 22, 1–55

5

Hydrogen yield (mol/mol)

4.5 Rh-Ni/Y2O3-Al2O3

4 3.5 3 2.5 2 Rh/MgAl2O4

1.5 1 0.5 0 0

4

8

12

16

20

24

Time (h) Fig. 11 Hydrogen yield as a function of time-on-stream during the steam reforming of crude bioethanol in the presence of the Rh/MgAl2O4 and Rh-Ni/Y2O3-Al2O3 Rh-Ni/Y-Al2O3 catalysts.

On a kinetics and catalysis point of view, there are some similarities between the steam reforming of hydrocarbons and that of alcohols: Rhodium is the most active metal for the reaction. However, the steam reforming being not very sensitive to the nature of metal, other metals (Ni, Co, . . .) can also be used. Thermodynamically, the steam reforming requires relatively high temperatures and low pressures. When the reaction is carried out at moderate temperatures (400– 500 1C; case of aromatic steam dealkylation), the support may play a dominant role, specially on Rh. At higher temperatures (550 1C and above), and specially for methane steam reforming, hydrocarbon activation is the determining step of reaction. But there are also some remarkable diﬀerences between hydrocarbons and alcohols. Alcohols currently lead to complex kinetic schemes including dehydration, dehydrogenation, cracking,. These reactions may be much more rapid than the steam reforming itself, which makes that alcohol conversion is not fully representative of the SR rate and of the rate of H2 production. Alcohols are more reactive than the corresponding hydrocarbons in steam reforming but the maximal H2 yield is lower (for instance, it is of 6 moles H2 per mole of ethanol and 7 moles H2 per mole of ethane). Several grades of bioethanol can be processed to produce hydrogen but high puriﬁcation costs should be avoided. Most studies were devoted to the steam reforming of pure ethanol. The present tendency is to produce biofuels of second generation from lignocellulosic biomass or even from lignine. It is recommended to pay more attention to the use of raw bioethanol Catalysis, 2010, 22, 1–55 | 47

that could be produced by such processes and to develop catalysts more resistant to deactivation.

References 1 N. N. Greenwood and A. Earnshaw, in Chemistry of the Elements, Pergamon Press (1984). 2 E. Kossinna, in Handbuch der Geophysik, ed. B. Gutenberg, Vol. 2, Borntra¨ger, Berlin, 1933. 3 A. R. Wyatt, Nature, 1984, 311, 370. 4 EIA report 2008 http://www.eia.doe.gov/emeu/international/reserves.html 5 World Coal Institute, coal facts at http://www.worldcoal.org/pages/content/ index.asp?PageID=188 6 J. Woods, J. Kung, D. Kingston, L. Kotlyar, B. Sparks and T. McCracken, Oil & Gas Sci. Technol.-Rev. IFP, 2008, 63, 151. 7 R. Tilagone, S. Venturi and G. Monnier, Oil & Gas Sci. Technol.-Rev. IFP, 2006, 61, 155. 8 J. R. Smith and J. W. Smith, Intern. J. Coal Geol., 2007, 70, 79. 9 T. Rostrup-Nielsen, Catal. Today, 2005, 106, 293. 10 J. D. Holladay, J. Hu, D. L. King and Y. Wang, Catal. Today, 2009, 139, 244. 11 D. J. Cuﬀ and William J. Young, U.S. Energy Atlas, Free Press/McMillan Publishing Co. New York, 1980. 12 M. Parikka, Biomass and Bioenergy, 2004, 27, 613. 13 P. McKendry, Bioresource Technol., 2002, 83, 37–47. 14 V. Subramani and C. Song, in Catalysis, The Royal Society of Chemistry, London, 2007, Vol. 20, p. 65. 15 J. R. Rostrup-Nielsen, in Steam Reforming Catalysts, Teknisk Forlag, Copenhagen, 1972. 16 J. R. Rostrup-Nielsen, Catalytic Steam Reforming in Catalysis, Science & Technology, Springer-Verlag, Berlin,Vol. 5, 1984. 17 J. R. Rostrup-Nielsen, Phys. Chem. Chem. Phys, 2001, 3, 283. 18 F. Joensen and J. R. Rostrup-Nielsen, J. Power Sources, 2002, 105, 195. 19 K.-H. Li, H. C. Van Ness and M. M. Abbott, Reaction Kinetics, Reactor Design and Thermodynamics (Section 4) in Perry’s Chemical Engineering Handbook, Mc Graw Hill, 1984. 20 D. R. Stull, E. F. Westrum Jr. and G. C. Sinke, The Chemical Thermodynamics of Organic Compounds, 2nd ed., R. E. Krieger Publ., 1987. 21 G. L. Rabinovich, G. N. Maslyanskii and L. M. Treiger, Kinet. Katal., 1971, 12, 1567. 22 L. M. Treiger, G. L. Rabinovich and G. N. Maslyanskii, Kinet. Katal., 1973, 14, 1582. 23 G. L. Rabinovich and V. N. Mozhaiko, Neftekhimiya, 1975, 15, 373. 24 S. Mori and M. Uchiyama, J. Catal., 1976, 42, 323. 25 D. C. Grenoble, J. Catal., 1978, 51, 203. 26 D. C. Grenoble, J. Catal., 1978, 51, 212. 27 D. Duprez, P. Pereira, A. Miloudi and R. Maurel, J. Catal., 1982, 75, 151. 28 P. Beltrame, G. Casalone, L. Forni, I. Ferino, B. Marongiu and S. Torrazza, Ind. Eng. Chem. Proc. Des. Dev., 1984, 23, 176. 29 D. Duprez, A. Miloudi, G. Delahay and R. Maurel, J. Catal., 1984, 90, 292. 30 D. Duprez, Appl. Catal. A, 1992, 82, 111. 48 | Catalysis, 2010, 22, 1–55

31 D. Duprez, P. Pereira, J. Barbier and R. Maurel, React. Kinet. Catal. Lett., 1980, 13, 217. 32 D. Duprez, M. Mendez and J. Little, Appl. Catal., 1986, 27, 145. 33 D. Duprez, R. Maurel, A. Miloudi and P. Pereira, Nouv. J. Chim., 1982, 6, 163. 34 C. J. Kim, J. Catal., 1978, 52, 169. 35 D. Duprez, A. Miloudi and L. Tournayan, Appl. Catal., 1985, 14, 333. 36 L. Forni, I. Ferino, B. Marongiu and S. Torrazza, J. Catal., 1984, 85, 169. 37 K. Koechloeﬂ, in Proc. 6th Intern. Cong. Catalysis, London 1976, The Chemical Society, London, 1977, p. 1122. 38 D. Duprez, Catal. Today, 2006, 112, 17. 39 X. Wang and R. J. Gorte, Appl. Catal. A, 2002, 224, 209. 40 B. Harrison, A. F. Diwell and C. Halett, Plat. Metal Rev., 1988, 32, 73. 41 J. Barbier Jr. and D. Duprez, Appl. Catal. B, 1993, 3, 61. 42 J. Barbier Jr. and D. Duprez, Appl. Catal. B, 1994, 4, 105. 43 S. Hilaire, X. Wang, T. Luo, R. J. Gorte and J. Wagner, Appl. Catal. A, 2004, 258, 271. 44 K. Otsuka, M. Hatano and A. Morikawa, J. Catal., 1983, 79, 493. 45 F. Sadi, D. Duprez, F. Gerard and A. Miloudi, J. Catal., 2003, 213, 226. 46 X. Zhao, S. Ma, J. Hrbek and J. A. Rodriguez, Surf. Sci., 2007, 601, 2445. 47 A. Trovarelli, in Catalysis by Ceria and Related Materials, Catalysis Science Series, Vol. 2, Imperial College Press, 2002. 48 D. Martin and D. Duprez, J. Phys. Chem., 1996, 100, 9429. 49 D. Martin and D. Duprez, J. Phys. Chem. B, 1997, 101, 4428. 50 Y. Madier, C. Descorme, A.-M. Le Govic and D. Duprez, J. Phys. Chem. B, 1999, 103, 10999. 51 D. Duprez, in In ‘‘Isotopes in Catalysis’’ ed. J. Hargreaves, D. Jackson and G. Webb, Imperial College Press. Catalytic Science Series, 2006, Vol. 4, Chapter 6, pp. 133–181. 52 A. Holmgren, B. Andersson and D. Duprez, J. Catal., 1999, 182, 441. 53 A. Galdikas, C. Descorme and D. Duprez, Solid State Ionics, 2004, 166, 147. 54 C. Descorme and D. Duprez, Appl. Catal. A, 2000, 202, 231. 55 S. Bedrane, C. Descorme and D. Duprez, Appl. Catal. A, 2005, 289, 90. 56 K. J. Lim, D. G. Lo¨ﬄer and M. Boudart, J. Catal., 1986, 100, 158. 57 K. J. Lim and M. Boudart, J. Catal., 1987, 105, 263. 58 S. Springmann, G. Friedrich, M. Himmen, M. Sommer and G. Eigenberger, Appl. Catal. A, 2002, 235, 101. 59 A. Qi, S. Wang, C. Ni and D. Wu, Int. J. Hydrogen Energy, 2007, 32, 981. 60 D. Duprez, M. Mendez and G. Delahay, Nouv. J. Chim., 1985, 9, 545. 61 A.- M. Azad and M. J. Duran, Appl. Catal. A, 2007, 330, 77. 62 A.- M. Azad, M. J. Duran, A. K. Mc Coy and M. A. Abraham, Appl. Catal. A, 2007, 332, 225. 63 A. K. Mc Coy, M. J. Duran, A.-M. Azad, S. Chattopadhyay and M. A. Abraham, Energy and Fuels, 2007, 21, 3513. 64 J. J. Strohm, J. Zheng and C. Song, J. Catal., 2006, 238, 309. 65 S. Kacimi, J. Barbier Jr., R. Taha and D. Duprez, Catal. Lett., 1993, 22, 343. 66 S. Ahmed and M. Krumpelt, Intern. J. Hydrogen Energy, 2001, 26, 291. 67 A. Heinzel, B. Vogel and P. Hu¨bner, J. Power Sources, 2002, 105, 202. 68 L. S. Wang, K. Murata and M. Inaba, Ind. Eng. Chem. Res., 2004, 43, 3228. Catalysis, 2010, 22, 1–55 | 49

69 D. C. Grenoble, M. M. Estadt and D. F. Ollis, J. Catal., 1981, 67, 90. 70 T. Bunluesin, E. S. Putna and R. J. Gorte, Catal. Lett., 1996, 41, 1. 71 T. Bunluesin, R. J. Gorte and G. W. Graham, Appl. Catal. B, 1998, 15, 107. 72 M. Nilsson, X. Karatsas, B. Lindstro¨m and L. J. Petersson, Chem. Eng. J., 2008, 142, 309. 73 D. Duprez, J. Chim. Phys., 1995, 92, 1952. 74 T. Maillet, J. Barbier and D. Duprez, Appl. Catal. B., 1996, 9, 251. 75 S. Bruehlmann, P. Novak, P. Lienemann, M. Trottmann, U. Gfeller, C. N. Zwicky, B. Bommer, H. Huber, M. Wolfensberger and N. V. Heeb, Appl. Catal. B, 2007, 70, 276. 76 Y. Jamal, T. Wagner and M. L. Wyszinski, Int. J. Hydrogen Energy, 1996, 21, 507. 77 C. Li and K. Suzuki, Renewable & Sustainable Energy Reviews, 2009, 13, 594. 78 H. R. Appell, E. C. Moroni and R. D. Miller, Energy Sources, 1977, 3, 163. 79 Y. Takemura, H. Itoh and K. Ouchi, Ind. Eng. Chem. Fundam., 1981, 20, 94. 80 A. Orio, J. Corella and I. Narvaez, Ind. Eng. Chem. Res., 1997, 36, 3800. 81 A. Olivares, M. P. Aznar, M. A. Caballero, P. Gil, E. France´s and J. Corella, Ind. Eng. Chem. Res., 1997, 36, 5220. 82 P. A. Simell, E. K. Hirvensalo, V. T. Smolander and A. O. I. Krause, Ind. Eng. Chem. Res., 1999, 38, 1250. 83 J. Srinakruang, K. Sato, T. Vitidsant and K. Fujimoto, Catal. Comm., 2005, 6, 437. 84 S. Rapagna, N. Jand, A. Kiennemann and P. U. Foscolo, Biomass Bioenergy, 2000, 19, 187. 85 S. Rapagna, H. Provendier, C. Petit, A. Kiennemann and P. U. Foscolo, Biomass Bioenergy, 2002, 22, 377. 86 D. Swierczynski, S. Libs, C. Courson and A. Kiennemann, Appl. Catal. B, 2007, 74, 211. 87 D. Swierczynski, C. Courson and A. Kiennemann, Chem. Eng. Process., 2008, 47, 508. 88 D. Swierczynski, C. Courson, L. Bedel, A. Kiennemann and J. Guille, Chem. Mater., 2006, 18, 4025. 89 R. Zhang, Y. Wang and R. C. Brown, Energy Conversion & Management, 2007, 48, 68. 90 A. Lamacz, A. Krzton, A. Musi and P. Da Costa, Catal. Lett., 2009, 128, 40. 91 S. Bona, P. Guille´n, J. G. Alcalde, L. Garcı´ a and R. Bilbao, Chem. Eng. J., 2008, 137, 587. 92 R. Coll, J. Salvado´, X. Farriol and D. Montane´, Fuel Process. Technol., 2001, 74, 19. 93 R. L. Bain, D. C. Dayton, D. L. Carpenter, S. R. Czernik, C. J. Feik, R. J. French, K. A. Magrini-Bair and S. D. Phillips, Ind. Eng. Chem. Res., 2005, 44, 7945. 94 E. Kikuchi, K. Ito; T. Ino and Y. Morita, J. Catal., 1977, 46, 382. 95 H. Muraki and Y. Fujitani, Appl. Catal., 1989, 47, 75. 96 X. Wang and R. J. Gorte, Catal. Lett., 2001, 73, 15. 97 R. Craciun, B. Shereck and R. J. Gorte, Catal. Lett., 1998, 51, 149. 98 B. Silberova, H. J. Venvik, J. C. Walmsley and A. Holmen, Catal. Today, 2005, 100, 457. 99 C. K. Acharya, A. M. Lane and T. R. Krause, Catal. Lett., 2006, 106, 41. 100 S. Takenaka, Y. Orita, H. Umebayashi, H. Matsune and M. Kishida, Appl. Catal. A, 2008, 351, 189.

50 | Catalysis, 2010, 22, 1–55

101 R. T. K. Baker, M. A. Barber, P. S. Harris, F. S. Feates and R. J. Waite, J. Catal., 1972, 26, 51. 102 R. T. K. Baker and J. J. Chludzinski, J. Catal., 1980, 64, 464. 103 L. Zhang, X. Wang, B. Tan and U. S. Ozkan, J. Mol. Cat. A., 2009, 297, 26. 104 N. M. Rodriguez, J. Mater. Res., 1993, 8, 3233. 105 D. Duprez, M. C. Demicheli, P. Marecot, J. Barbier, O. A. Feretti and E. N. Ponzi, J. Catal., 1990, 124, 324. 106 M. C. Demicheli, D. Duprez, J. Barbier, O. A. Feretti and E. N. Ponzi, J. Catal., 1994, 145, 437. 107 Hydrogen Data Book, Hydrogen Analysis Research Center of the US Department of Energy, data at http://hydrogen.pnl.gov/cocoon/morf/hydrogen/ article/103 108 J. R. H. Ross, in Surface and Defect Properties of Solids, Vol. 4, The Royal Society of Chemistry, 1975, pp. 34–67. 109 T. Inui, in Catalysis, Vol. 16, The Royal Society of Chemistry, 2002, p. 133. 110 I. M. Bodrov and L. O. Apel’baum, Kinet. Katal., 1967, 8, 379. 111 J. R. Rostrup-Nielsen and J.-H. Bak Hansen, J. Catal., 1993, 144, 38. 112 D. Qin and J. Lapszewicz, Catal. Today, 1994, 21, 551. 113 D. Qin, J. Lapszewicz and X. Jiang, J. Catal., 1996, 159, 140. 114 J. M. Wei and E. Iglesia, J. Catal., 2004, 224, 370. 115 J. M. Wei and E. Iglesia, J. Catal., 2004, 225, 116. 116 J. M. Wei and E. Iglesia, Phys. Chem. Chem. Phys., 2004, 6, 3754. 117 J. M. Wei and E. Iglesia, Ang. Chem. Int. Ed., 2004, 43, 3685. 118 J. M. Wei and E. Iglesia, J. Phys. Chem. B, 2004, 108, 7253. 119 J. M. Wei and E. Iglesia, J. Phys. Chem. B, 2004, 108, 4094. 120 J. M. Ramallo-Lo´pez, F. G. Requejo, A. F. Craievich, J. Wei, M. AvalosBorja and E. Iglesia, J. Mol. Catal. A, 2005, 228, 299. 121 I. Tavazzi, A. Beretta, G. Groppi and P. Forzatti, J. Catal., 2006, 241, 1. 122 A. Donazzi, A. Beretta, G. Groppi and P. Forzatti, J. Catal., 2008, 255, 241. 123 A. Donazzi, A. Beretta, G. Groppi and P. Forzatti, J. Catal., 2008, 255, 259. 124 M. Maestri, D. G. Vlachos, A. Beretta, G. Groppi and E. Tronconi, J. Catal., 2008, 259, 211. 125 J. Xu and M. Saeys, J. Phys. Chem. C., 2009, 113, 4099. 126 J. Xu, L. W. Chen, K. F. Tan, A. Borgna and M. Saeys, J. Catal., 2009, 261, 158. 127 G. Jones, J. G. Jakobsen, S. S. Sim, J. Kleis, M. P. Andersson, J. Rossmeisl, F. Abild-Pedersen, T. Bligaard, S. Helveg, B. Hinnemann, J. R. Rostrup-Nielsen, I. Chorkendorﬀ, J. Sehested and J. K. Norskov, J. Catal., 2008, 259, 147. 128 D. W. Blaylock, T. Ogura, W. H. Green and G. J. O. Beran, J. Phys. Chem. C., 2009, 113, 4898. 129 P. G. Menon, J. C. Deken and G. F. Froment, J. Catal., 1985, 95, 313. 130 R. Craciun, B. Shereck and R. J. Gorte, Catal. Lett., 1998, 51, 149. 131 R. Craciun, W. Daniell and H. Kno¨zinger, Appl. Catal. A, 2002, 230, 153. 132 H. Kno¨zinger, K. Kochloeﬂ and W. Meye, J. Catal., 1973, 28, 69. 133 J. I. DiCosimo, V. K. Diez, M. Xu, E. Iglesia and C. R. Apesteguia, J. Catal., 1998, 178, 499. 134 V. Fierro, O. Akdim and C. Mirodatos, Green Chem., 2003, 5, 20. 135 P. D. Vaidya and A. E. Rodrigues, Chem. Eng. J., 2006, 117, 39. 136 S. Cavallaro, Energy and Fuels, 2000, 14, 1195. 137 S. Freni, J. Power Sources, 2001, 94, 14. 138 S. Cavallaro, V. Chiodo, A. Vita and S. Freni, J. Power Sources, 2003, 123, 10. 139 S. Cavallaro, V. Chiodo, S. Freni, N. Mondello and F. Frusteri, Appl. Catal. A, 2003, 249, 119. Catalysis, 2010, 22, 1–55 | 51

140 F. Aupretre, C. Descorme and D. Duprez, Catal. Comm., 2002, 3, 263. 141 J. P. Breen, R. Burch and H. M. Coleman, Appl. Catal. B, 2002, 39, 65. 142 D. K. Liguras, D. I. Kondarides and X. E. Verykios, Appl. Catal. B, 2003, 43, 345. 143 F. Aupretre, C. Descorme and D. Duprez, Stud. Surf. Sci. Catal., 2003, 145, 303. 144 F. Aupretre, C. Descorme and D. Duprez, Topics Catal., 2004, 30/31, 405. 145 F. Frusteri, S. Freni, L. Spadaro, V. Chiodo, G. Bonura, S. Donato and S. Cavallaro, Catal. Comm., 2004, 5, 611. 146 C. Diagne, H. Idriss and A. Kiennemann, Catal. Comm., 2002, 3, 565. 147 C. Diagne, H. Idriss, K. Pearson, M. A. Gomez-Garcia and A. Kiennemann, Comptes Rendus Chimie., 2004, 7, 617. 148 L. De Rogatis, T. Montini, M. F. Casula and P. Fornasiero, J. Alloys and Compounds, 2008, 451, 516. 149 H. S. Roh, Y. Wang, D. L. King, A. Platon and Y. H. Chin, Catal. Lett., 2006, 108, 15. 150 A. Platon, H. S. Roh, D. L. King and Y. Wang, Topics Catal., 2007, 46, 374. 151 H. S. Roh, D. L. King and Y. Wang, Topics Catal., 2008, 49, 32. 152 F. Aupretre, C. Descorme, D. Duprez, D. Casanave and D. Uzio, J. Catal., 2005, 233, 464. 153 F. Can, A. Le Valant, N. Bion, F. Epron and D. Duprez, J. Phys. Chem. C, 2008, 112, 14145. 154 G. A. Deluga, J. R. Salge, L. D. Schmidt and X. E. Verykios, Science, 2004, 303, 993. 155 J. Rasko´, A. Hancz and A. Erdohelyi, Appl. Catal. A, 2004, 269, 13. 156 A. Erdohelyi, J. Rasko´, T. Kecske´s, M. To´th, M. Do¨mo¨k and K. Baa´n, Catal. Today, 2006, 116, 367. 157 H. Idriss, Platinum Metals Rev., 2004, 48, 105. 158 A. Birot, F. Epron, C. Descorme and D. Duprez, Appl. Catal. B, 2008, 79, 17. 159 T. Rampe, P. Hu¨bner, B. Vogel and A. Heinkel, Proc. 1st World Conference Biomass for Energy & Industry, Sevilla, Spain, June 2000, James & James, Science Publishers, 2001, p. 1889. 160 G. Jacobs, R. A. Keogh and B. H. Davis, J. Catal., 2007, 245, 326. 161 S. M. de Lima, I. O. da Cruz, G. Jacobs, B. H. Davis, L. V. Mattos and F. B. Noronha, J. Catal., 2008, 257, 356. 162 S. M. de Lima, A. M. Silva, U. M. Graham, G. Jacobs, B. H. Davis, L. V. Mattos and F. B. Noronha, Appl. Catal.A, 2009, 352, 95. 163 E. O¨ru¨cu¨, F. Go¨kaliler, A. E. Aksoylu and Z. I. O¨nsan, Catal. Lett., 2008, 120, 198. 164 D. K. Liguras, K. Goundani and X. E. Verykios, Int. J. Hydrogen Energy, 2004, 29, 419. 165 M. A. Goula, S. K. Kontou and P. E. Tsiakaras, Appl. Catal. B, 2004, 49, 135. 166 A. Casanovas, J. Llorca, N. Homs, J. L. G. Fierro and P. Ramirez de la Piscina, J. Molec. Catal. A, 2006, 250, 44. 167 V. V. Galvita, G. L. Semin, V. D. Belyaev, V. A. Semikolenov, P. Tsiakaras and V. A. Sobyanin, Appl. Catal. A, 2001, 220, 123. 168 B. Zhang, X. Tiang, Y. Li, Y. Xu and W. Shen, Int. J. Hydrogen Energy, 2007, 32, 2367. 169 W. Cai, B. Zhang, Y. Li, Y. Xu and W. Shen, Catal. Comm., 2007, 8, 1588. 170 W. Cai, F. Wang, E. Zhan, A. C. Van Veen, C. Mirodatos and W. Shen, J. Catal., 2008, 257, 96. 52 | Catalysis, 2010, 22, 1–55

171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199

A. N. Fatsikostas and X. E. Verykios, J. Catal., 2004, 225, 439. J. Kugai, S. Velu and C. Song, Catal. Lett., 2005, 101, 255. Y. Yang, J. Ma and F. Wu, Int. J. Hydrogen Energy, 2006, 31, 877. J. Sun, X. P. Qiu, F. Wu and W. T. Zhu, Int. J. Hydrogen Energy, 2005, 30, 437. F. Frusteri, S. Freni, V. Chiodo, L. Spadaro, O. Di Blasi, G. Bonura and S. Cavallaro, Appl. Catal. A: Gen., 2004, 270, 1. F. Frusteri, S. Freni, V. Chiodo, S. Donato, G. Bonura and S. Cavallaro, Int. J. Hydrogen Energy, 2006, 31, 2193. Y. Yang, J. X. Ma and F. Wu, Int. J. Hydrogen Energy, 2006, 31, 877. P. Biswas and D. Kunzru, Catal. Lett., 2007, 118, 36. K. Urasaki, K. Tokunaga, Y. Sekine, M. Matsukata and E. Kikuchi, Catal. Commun., 2008, 9, 600. L. J. I. Coleman, W. Epling, R. R. Hudgins and E. Croiset, Appl. Catal. A, 2009, 363, 52. F. Frusteri, S. Freni, V. Chiodo, L. Spadaro, G. Bonura and S. Cavallaro, J. Power Sources, 2004, 132, 139. F. Marino, G. Baronetti, M. Jobbagy and M. Laborde, Appl. Catal. A, 2003, 238, 41. F. Marino, M. Boveri, G. Baronetti and M. Laborde, Int. J. Hydrogen Energy, 2001, 26, 665. A. J. Vizcaino, P. Arena, G. Baronetti, A. Carcero, J. A. Calles, M. A. Laborde and N. Amadeo, Int. J. Hydrogen Energy, 2008, 33, 3489. A. Denis, W. Grzegorczyk, W. Gac and A. Machocki, Catal. Today, 2008, 137, 453. J. W. C. Liberatori, R. U. Ribeiro, D. Zanchet, F. B. Noronha and J. M. C. Bueno, Appl. Catal. A, 2007, 327, 197. A. N. Fatsikostas, D. I. Kondarides and X. E. Verykios, Catal. Today, 2002, 75, 145. A. N. Fatsikostas, D. I. Kondarides and X. E. Verykios, Chem. Comm., 2001, 851. J. L. Ye, Y. Q. Wang, Y. Liu and H. Wang, Int. J. Hydrogen Energy, 2008, 33, 6602. V. Klouz, V. Fierro, P. Denton, H. Katz, J. P. Lisse, S. Bouvot-Mauduit and C. Mirodatos, J. Power Sources, 2002, 105, 26. V. Fierro, O. Akdim, H. Provendier and C. Mirodatos, J. Power Sources, 2005, 145, 659. M. H. Youn, J. G. Seo, P. Kim, J. J. Kim, H. I. Lee and I. K. Song, J. Power Sources, 2006, 162, 1270. L. Huang, J. Xie, R. Chen, D. Chu, W. Chu and A. T. Hsu, Int. J. Hydrogen Energy, 2008, 33, 7448. J. H. Lee, E. G. Lee, O. S. Joo and K. D. Jung, Appl. Catal. A, 2004, 269, 1. T. J. Huang, T. C. Yu and S. Y. Jhao, Ind. Eng. Chem. Res., 2006, 45, 150. J. A. Calles, A. Carrero and A. J. Vizcaı´ no, Microporous and Mesoporous Materials, 2009, 119, 200. F. Soyal-Baltacioglu, A. E. Aksoylu and Z. I. O¨nsan, Catal. Today, 2008, 138, 183. L. P. R. Profeti, J. A. C. Dias, J. M. Assaf and E. M. Assaf, J. Power Sources, 2009, 190, 525. F. Romero-Sarria, J. C. Vargas, A.-C. Roger and A. Kiennemann, Catal. Today, 2008, 133, 149. Catalysis, 2010, 22, 1–55 | 53

200 F. Haga, T. Nakajima, H. Miya and S. Mishima, Catal. Lett., 1997, 48, 223. 201 J. Llorca, P. Ramirez de la Piscina, J. A. Dalmon, J. Sales and N. Homs, Appl. Catal. B, 2003, 43, 355. 202 M. S. Batista, R. K. S. Santos, E. M. Assaf, J. M. Assaf and E. A. Ticianelli, J. Power Sources, 2003, 124, 99. 203 J. Llorca, N. Homs, J. Sales, J. L. G. Fierro and P. Ramirez de la Piscina, J. Catal., 2004, 222, 470. 204 J. Llorca, N. Homs and P. Ramirez de la Piscina, J. Catal., 2004, 227, 556. 205 M. S. Batista, R. K. S. Santos, E. M. Assaf, J. M. Assaf and E. A. Ticianelli, J. Power Sources, 2004, 134, 27. 206 A. Kaddouri and C. Mazzocchia, Catal. Commun., 2004, 5, 339. 207 S. Cavallaro, N. Mondello and S. Freni, J. Power Sources, 2001, 102, 198. 208 R. U. Ribeiro, J. W. C. Liberatori, H. Winnishofer, J. M. C. Bueno and D. Zanchet, Appl. Catal. B, 2009, 91, 670. 209 S. S.-Y. Lin, H. Daimon and S. Y. Ha, Appl. Catal. A, 2009, 366, 252. 210 J. C. Vargas, S. Libs, A.-C. Roger and A. Kiennemann, Catal. Today, 2005, 107–108, 417. 211 M. Virginie, M. Araque, A.-C. Roger, J. C. Vargas and A. Kiennemann, Catal. Today, 2008, 138, 21. 212 B. A. Raich and H. C. Foley, Ind. Eng. Chem. Res., 1998, 37, 3888. 213 S. Cavallero and S. Freni, Int. J. Hydrogen Energy, 1996, 21, 465. 214 W. Grzegorczyk, A. Denis, W. Gac, T. Ioannides and A. Machocki, Catal. Lett., 2009, 128, 443. 215 M. Galbe and G. A. Zacchi, Appl. Microbiol. Biotechnol., 2002, 59, 618. 216 D. Budny, in Brazil Institute Special Report, P. Sotero Ed., April 2007, issue no 3. 217 N. Jeuland, X. Montagne and X. Gautrot, in Which Fuels For Low CO2 Engines? P. Duret, X Montagne Eds. and Editions Technip, Paris, 2004, 7. 218 P. Bo¨rjesson, Appl. Energy, 2009, 86, 589. 219 T. Searchinger, R. Heimlich, R. A. Houghton, F. Dong, A. Elobeid, J. Fabiosa, S. Tokgoz, D. Hayes and T.-H. Yu, Science, 2008, 319, 1238. 220 F. Randelli, Reg. Environ. Change, 2008, doi: 10.1007/s10113-008-0055-7. 221 Communication from the commission of European Communities. An EU Strategy for Biofuels, Brussels, SEC(2006) 142. 222 CONCAVE, EUCAR, JRC. Well-to-Wheels Analysis of Future Automotive Fuels and Powertrains in European Context. Well-to-Wheels Report. Version 2a, December 2005. 223 ADEME/DIREM. Bilans e´nerge´tiques et gaz a` eﬀet de serre des ﬁlie`res de production de biocarburants, November 2002. 224 Price Waterhouse Cooper. Biofuels and other renewable fuels for transport. A study commissioned by the Federal Public Service of Public Health Food Chain Safety and Environment, Brussels, Belgium, 2005. 225 A. E. Farrell, R. J. Plevin, B. T. Turner, A. D. Jones, M. O’Hare and D. M. Kammen, Science, 2006, 311, 506. 226 K. T. Tan, K. T. Lee and A. R. Mohamed, Energy Policy, 2008, 36, 3360. 227 D. J. Hayes, Catal. Today, 2009, 145, 138. 228 J. F. Gruson, F. Monot, http://webetab.ac-bordeaux.fr/Etablissement/ AFallieres/forumDD/biocarburants_seconde_generation.pdf 229 G. M. Bohlmann, Ind. Biotechnol., 2006, 2(1), 14. 230 N. Kosaric and F. Vardar-Sukan, in The Biotechnology of Ethanol. Classical and Future Applications, ed. M. Roehr, Editions, Wiley-VCH, Weinheim, 2001. 54 | Catalysis, 2010, 22, 1–55

231 J. Rass-Hansen, R. Johansson, M. Moller and C. H. Christensen, Int. J. Hydrogen Energy, 2008, 33, 4547. 232 J. Larsen, M. Ø. Petersen, L. Thirup, H. W. Li and F. K. Iversen, Chem. Eng. Technol., 2008, 31, 1. 233 A. J. Akande, R. O. Idem and A. K. Dalai, Appl. Catal. A, 2005, 287, 159. 234 A. Le Valant, PhD thesis, University of Poitiers (France), 2008. 235 J. C. Vargas, S. Libs, A.-C. Roger and A. Kiennemann, Catal. Today, 2005, 107, 417. 236 A. Le Valant, A. Garron, N. Bion, F. Epron and D. Duprez, Catal. Today, 2008, 138, 169. 237 D. Wang, D. Montane´ and E. Chornet, Appl. Catal. A, 1996, 143, 245.

Catalysis, 2010, 22, 1–55 | 55

Catalytic reforming of liquid hydrocarbons for on-board solid oxide fuel cell auxiliary power units Johannes W. Schwanka and Andrew R. Tadda DOI: 10.1039/9781847559630-00056

Catalytic reforming of liquid transportation fuels and their hydrocarbon components via steam reforming, catalytic partial oxidation, and autothermal reforming is reviewed. The review focuses on fuel reforming to generate hydrogen-rich syngas for on-board applications, with emphasis on solid-oxide fuel cell (SOFC) based auxiliary power units (APU). After a brief overview of fuel properties of gasoline, diesel and kerosene-based jetfuels such as JP-8, reforming methods including steam reforming, catalytic partial oxidation and autothermal reforming are discussed. Strategies to deal with catalyst deactivation caused by carbon deposition and sulfur poisoning are delineated. The review also addresses the special engineering challenges associated with the development of compact, on-board fuel reformers for gasoline, diesel, and jet fuels.

1.

Introduction

Motivated by requirements for better fuel economy and lower emissions, the automobile industry has been working on various concepts for electriﬁcation of drive trains. While the immediate future may see the large-scale deployment of hybrid and plug-in hybrid electric vehicles, the development of hydrogen-powered proton exchange membrane (PEM) fuel cell vehicles is being pursued as an alternative option for eventually replacing the internal combustion engine. With growing interest in fuel cells as primary propulsion systems for automobiles,1 a concerted research and development eﬀort was mounted towards on-board reforming of gasoline.2 PEM fuel cells require high-purity H2. Given the challenges of limited on-board hydrogen storage capacity and the lack of hydrogen refueling infrastructure, on-board processing of gasoline into pure H2 has been considered as a possible solution.3 Most of the research projects focused on partial oxidation or autothermal reforming processes, as on-board applications required fast start-up and good transient response.4 Large-scale conventional steam reforming technology developed for stationary applications cannot easily be transferred to automotive applications, due to size and weight limitations, and the limited availability of steam on board a vehicle. To generate the amount of hydrogen required for a PEM fuel cell vehicle would have taken a reformer system too large to ﬁt into a typical passenger car. Therefore, major eﬀorts had to be directed towards downsizing of fuel reformer systems and the development of novel catalytic reactor conﬁgurations, including microstructured and plate reformers.5–15 a

Transportation Energy Center, Department of Chemical Engineering, University of Michigan, 3014 H.H. Dow Building, 2300 Hayward Road, Ann Arbor MI 48109-2136

56 | Catalysis, 2010, 22, 56–93 c

The Royal Society of Chemistry 2010

The technical challenges in developing compact on-board fuel processors are quite formidable, especially for applications requiring the fuel processing of logistic fuels such as JP-8 and diesel. Catalytic fuel reforming generates a hydrogen-rich gas stream from hydrocarbons.16,17 Catalysts for onboard fuel reforming must meet a large number of requirements. First, they should be able to catalyze reforming reactions of hydrocarbons and their thermal decomposition products, and be able to function under transient conditions, permitting multiple startups and shutdowns without requiring elaborate regeneration schemes. Catalysts should be resistant to coking, and they must be thermally stable without sintering at the high temperatures encountered under typical fuel processing conditions. Furthermore, they must be mechanically strong and withstand the vibrations on moving vehicles. Ideally, they should also be tolerant to sulfur, even if low-sulfur fuels are being used. Finally, catalysts must be able to cope with the variability in the chemical composition of fuels and fuel additives. Despite considerable progress towards technical targets set by the U.S. Department of Energy, the complexity and technical diﬃculty of generating pure H2 from gasoline on-board of vehicles prompted the U.S. Department of Energy in 2004 to make a no-go decision, and funding of research eﬀorts for on-board fuel reforming of gasoline into PEM-fuel cell grade H2 dried up.18 Instead, the research and development eﬀort was redirected towards centralized hydrogen generation and on-board storage of H2. There is, however, another fuel cell application that could greatly beneﬁt from on-board reforming of liquid transportation fuels, namely auxiliary power units (APU) containing solid oxide fuel cells. Heavy-duty trucks, military vehicles, and recreational vehicles have signiﬁcant needs for electric power even when the vehicle is not moving.19–23 For example, heavy-duty trucks tend to idle their diesel engines for many hours at a time, burning fuel at very low eﬃciency, to provide heating and cooling and generate power for sleeper compartment accessories such as televisions, microwave ovens, and refrigerators. According to recent studies, idling diesel engines burn nearly a billion gallons of diesel fuel every year in the United States alone, and shifting to SOFC-APU systems could result in signiﬁcant fuel savings and lower emissions.24,25 Depending on the degree of thermal integration, on-board SOFC APUs can be more eﬃcient than conventional combustionengine based electricity generators and provide the additional beneﬁts of lower emissions and lower noise levels. The major impediment to the commercial deployment of SOFC-based APU systems is the challenge to convert liquid transportation fuels such as diesel and JP-8 military fuels onboard into hydrogen-rich syngas suitable for SOFCs. Several books on the subject of fuel cell technology contain chapters providing an introduction into the basic principles of fuel processing.26–29 There have also been several review papers published on the subject of fuel processing for transportation and other portable power applications.30–34 Song provided an overview of the fuel processing options for lowtemperature and high-temperature fuel cells.35 Qi et al. reviewed the technological progress in the development of integrated fuel processors, with focus on process optimization and process intensiﬁcation technologies such as Catalysis, 2010, 22, 56–93 | 57

engineered catalysts, heat integration, and in situ product puriﬁcation.36 Recently, two German reviews were published, dealing with the reforming of liquid fuels with special regard to the properties of diesel fuel and light fuel oil.37,38 Semelsberger et al. analyzed various reforming processes from a thermodynamic point of view.39 An extensive review of catalytic reforming of liquid hydrocarbon fuels for fuel cell applications appeared in 2006,40 focusing on reforming of diesel, gasoline, and representative model compounds. This earlier review40 provided an excellent treatment of the relevant thermodynamics, especially with regard to formation of carbon and coke, and also summarized kinetics and rate laws. Studies of heavy liquid hydrocarbon reforming started to appear during the 1980s,41 and there has been considerable interest in smaller scale fuel reforming using monolithic structures for distributed power generation in both stationary and mobile applications.42 Considerable progress has been made towards reforming of C8 þ hydrocarbons that are typically encountered in gasoline, diesel, or jet fuel.43–52 However, there are still many open questions as to how speciﬁc types of molecules aﬀect the reforming behavior of commercial fuels. Research publications on the subject of fuel processing of liquid fuels fall into two general categories: the ﬁrst category includes fundamental studies of catalytic conversion of individual molecular components of a fuel or mixtures of model compounds serving as surrogates for fuels. The second category contains papers that demonstrate a catalyst’s ability to reform gasoline, diesel, or jet fuel at laboratory scales to demonstrate the feasibility of various fuel processing strategies. Here, we are focusing exclusively on reviewing on-board catalytic fuel processing of hydrocarbon-based liquid transportation fuels with emphasis on SOFC-based APU applications. This review will not deal with other reforming methods such as plasma-assisted reforming,53 or supercritical reforming that have been covered in earlier reviews.40 This review also does not cover the extensive literature on reforming of methane, propane, butane,54,55 or alcohols such as methanol56 and bioethanol.57 2.

Fuel properties and SOFC fuel requirements

PEM fuel cells function only with very pure H2 that contains at most ppm levels of CO. Consequently, on-board fuel processing of gasoline or diesel for a PEM fuel cell requires an elaborate set of catalytic reactors operating in series, including a desulfurizer, a reformer, high and low temperature water gas shift reactors, and a selective oxidation reactor, methanation reactor, or palladium membrane to remove the remaining traces of CO. This poses a major challenge to operate all these reactors in a synchronized, load-following mode. SOFCs do not require such an elaborate fuelprocessing scheme. The only requirement is to convert the liquid hydrocarbons into hydrogen-rich syngas and to protect the fuel cell from sulfur contamination. This, for all practical purposes, requires only a reformer and desulfurizer. Furthermore, APUs for heavy duty vehicles tend to operate under more or less steady-state load for long periods of time, and there is less demand for dynamic load following. 58 | Catalysis, 2010, 22, 56–93

To gain a better understanding of the challenges involved in converting gasoline, diesel, or jet fuels into H2-rich syngas, a brief overview of properties and compositions of these fuels may be helpful. Gasoline, diesel, and commercial jet fuels consist of hundreds of compounds. The exact fuel compositions may be adjusted depending on the season and geography. Gasoline is a complex fuel and contains many impurities that can create problems in catalytic fuel processing. Its catalytic conversion to syngas requires high temperatures in excess of 930 K. Commercial jet fuels JET-A and JET-A1 that are primarily used in the US are kerosene-based fuels. The military uses specially formulated kerosene jet fuels, JP-4, JP-5, JP-8, and JP-100. These fuels have been hydrotreated and contain several additives. Jet fuels are considered kerosene-type fuels because fuel vaporization takes place mainly within a temperature range of 473–623 K. This range is much narrower compared to diesel, to meet the speciﬁcations for turbine engines. The high temperature boiling ranges of diesel and jet fuels make laboratoryscale fuel reforming studies quite diﬃcult. Diesel and jet fuels contain three major types of hydrocarbons: linear and branched paraﬃns, cycloalkanes, and aromatics. The ring structures in cycloalkanes and aromatics are generally believed to be more diﬃcult to reform than paraﬃns. Diesel has typical carbon numbers of 10–16, while JP-8 can be represented by an average carbon number of 11, excluding sulfur compounds and heteroatom containing additives, and gasoline by a carbon number of 8. Often, studies of fuel reforming of complex fuels are carried out using simpler surrogate model compounds and their mixtures.58 A good fuel surrogate should capture the major characteristics of the fuel, but make the analysis simpler. For example, JP-8 can be modeled with mixtures of n-dodecane, methylcyclohexane, and butylbenzene. For diesel, hexadecane could be used as surrogate as it provides a match of the C16 carbon number, but it turns out that the combustion properties of diesel are closer to dodecane, which is therefore often used as a surrogate. Incidentally, dodecane is also a suitable surrogate for paraﬃns in JP-8. Violi et al. developed a surrogate blend of six pure hydrocarbons to simulate the distillation and compositional characteristics of JP-8.59 Although heteroatoms such as sulfur and copper are present in only small concentrations, they can have a signiﬁcant eﬀect on a fuel’s reforming chemistry and on the performance of reforming catalysts. Jet fuels such as JP-8 can contain 300 ppm to 5000 ppm of sulfur and 100 ppb copper. Nitrogen containing compounds in the fuels seem to have less inﬂuence on reforming catalysts. Jet fuels contain a number of additives, and the type and concentration of additives represents the main diﬀerence between Jet-A and JP-8.60 For example, the military speciﬁcation MIL-DTL-83133 for JP-8 requires the addition of compounds that can act as static electricity dissipators, corrosion inhibitors, lubricity enhancers, fuel system icing inhibitors, and antioxidants. Jet fuels tend to be quite susceptible to oxidation, since they may have undergone hydrotreating to remove mercaptans. The role of antioxidants is to prevent reactions of dissolved oxygen with fuel molecules that could lead to the formation of peroxides, gums, and particulates. Typical Catalysis, 2010, 22, 56–93 | 59

antioxidants are sterically hindered phenols, for example 2,6-ditertiary butyl-4-methylphenol.

The purpose of adding static dissipators is to increase the electrical conductivity of the fuel, thereby preventing the build-up of hazardous static charges. The currently used Statdis 450, manufactured by Octel, contains dinonylnaphthalene sulfonic acid along with other organic solvents and two additional ingredients consisting of sulfur and nitrogen-containing polymers.61 Corrosion inhibiting and lubricity improving additives protect the fuel distribution system from corrosion and lubricate metal surfaces in the turbine engine. These are compounds that strongly interact with metal surfaces, and are added at typically 20 mg/L concentrations. There are many diﬀerent trade names for these additives, for example QPL-25017-7 shown below.

Fuel system icing inhibitors consist of di-ethyleneglycol monomethylether (di-EGME). CH3 OCH2 CH2 OCH2 CH2 OH This additive, present at 0.1–0.15 vol%, prevents water that is dissolved in the fuel from freezing at low temperatures. The handling of fuels containing di-EGME in a fuel processor is diﬃcult because di-EGME can precipitate out and form a gelatinous phase when water and fuel are co-fed in liquid form. Therefore, it is of critical importance to completely vaporize both the water and the fuel prior to feeding them into the fuel processor.

The purpose of the metal deactivating additive N,N 0 -disalicylidene-1,2propane diamine, added typically at 10 mg/L is to improve the thermal stability of fuels by deactivating trace contaminations of metals. 60 | Catalysis, 2010, 22, 56–93

One of the major issues in fuel processing of liquid fuels is the requirement to vaporize the fuel prior to feeding it into the reformer. Heavy hydrocarbons typically present in diesel have boiling points higher than 620–670 K, and at these temperatures, some fuels begin to pyrolyze. Some reforming methods, as we will see later, require the co-feeding of steam, and this can cause complications with miscibility and gelation of some of the fuel components. 3.

Catalysts for reforming of liquid hydrocarbons

The catalysts used for reforming of liquid hydrocarbons are usually precious metal such as rhodium or nickel supported on oxide supports. There seems to be a consensus that the reforming of hydrocarbons in liquid transportation fuels is possible, but diﬃcult because of problems associated with carbon formation and sulfur poisoning. Consequently, the literature shows a strong interest in catalytic supports with oxygen storage capacity, particularly mixed ceria-oxides, as a means to reduce coke formation. Mobile oxygen in the support might play an important role in the reforming of the liquid hydrocarbons, possibly by making the catalyst more resistant to coking by facilitating the oxidation of surface carbon species.62,63 In addition to typical supported metal catalysts, doped or substituted metaloxide catalysts have also been applied to liquid fuel reforming. Of note are pyrochlore catalysts and hexaaluminate catalysts.67 The majority of reforming catalysts reported in the literature have been prepared by conventional techniques. General descriptions of catalyst synthesis may be found in two recent books.68,69 Typically, a catalyst support is chosen and the active metal component is added as a solution. Removal of the solvent by drying leads to crystals of the metal precursor distributed through the pore structure of the support. These precursor crystals are converted to metal or metal oxide particles usually by heating under hydrogen or air. The usual goal is to distribute the metal as widely as possible, achieving very small average metal particle sizes so that the greatest possible portion of metal atoms are in the surface of the particles, available for reaction. A number of variables can aﬀect the metal dispersion, including but not limited to the precursor, solvent, support, and drying method. Catalyst supports may be obtained from suppliers or prepared using various techniques. Many mixed oxide supports are prepared by co-precipitation or a sol-gel approach. Both approaches begin with solutions of the desired metals. In co-precipitation a base or other precipitating agent is added or evolved in solution to cause precipitation of the metal hydroxides or other insoluble forms, such as carbonates. After recovery by ﬁltration and washing, the solids are calcined to convert to the desired oxides. Supports such as Al2O3 and SiO2 are easily obtained from major suppliers and are less often prepared by investigators. Most often catalysts are tested in packed beds of small particles, but ultimately for mobile applications they must be prepared in some structured form. A washcoat on planar surfaces, honeycomb-type monoliths, or other support is usually required to provide low pressure drop and prevent catalyst attrition. Catalysis, 2010, 22, 56–93 | 61

There are a large number of catalyst characterization techniques available.70–72 Among the most important physical characteristics of reforming catalysts are the total surface area and the metal dispersion. High speciﬁc surface areas provide space to widely distribute metal particles, and high metal dispersions yield eﬃcient use of the active component, particularly important when using noble metals. Often dispersions as high as 25–50% may be obtained when using noble metals.73–75 Nickel-based catalysts are most often reported to have lower dispersions.74,101,126,127 Some investigators, however, have reported obtaining high dispersions when working with nickel.76 Catalyst surface areas are usually measured by N2 physisorption at cryogenic temperatures following the BET method. Metal particle sizes are often estimated by titrating the metal with a chemisorbing species such as H2 or CO, which adsorb selectively on the metal. Metal particle sizes may also be estimated using SEM or XRD, provided that the particles are large enough to yield suﬃciently well deﬁned peaks. Operation under reforming conditions often leads to decreases in the catalyst surface area.73,77–79,138,141 Surface area losses can be caused by high operating temperatures or phase transitions in unstable support materials. Blocking of pores by coke can also lead to lower measured surface areas if the carbonaceous species are not removed prior to characterization. The metal dispersion often decreases during reforming.73,141 High temperatures tend to cause sintering of the metal particles, leading to lowered dispersions. Components of the feed can also lead to reduction in available metal surface area; for example operation with S-containing feeds may lead to lower measured metal dispersions. It is important to note that surface bound species may block chemisorption sites without changing the actual metal particle sizes. 4.

Fuel reforming methods

In the next section, three methods for fuel processing are reviewed, namely steam reforming (SR), catalytic partial oxidation (CPOX), and autothermal reforming (ATR). 4.1

Steam reforming

Steam reforming dates back to the late 1800s and early 1900s where Mond and Langer pioneered the use of nickel or cobalt on pumice as catalysts,80 and Mittasch and Schneider demonstrated methane steam reforming over magnesia supported nickel.81 Today, steam reforming is practiced on a large industrial scale for generating hydrogen-rich syngas from natural gas. Typical steam reforming catalysts contain anywhere from 4–30 wt% nickel supported on refractory oxides. More expensive cobalt and noble metal catalyst have also been used. An extensive review of catalytic steam reforming was written by Rostrup-Nielsen.82 Steam reforming is typically conducted at operating temperatures of 950 to 1200 K under 101–2500 kPa of pressure and steam to carbon ratios of 2/1 to 3/1. Excess steam is beneﬁcial, as it drives the reaction to completion, and it also decreases the deposition of carbonaceous species, ‘‘coke’’, on the catalyst surface. The 62 | Catalysis, 2010, 22, 56–93

endothermic nature of the steam reforming reaction requires high operation temperatures to achieve good conversion. Since steam reforming reactors require indirect heating, the process does not lend itself well to applications where rapid startup and response to transients is necessary. Steam reforming of hydrocarbons involves a number of consecutive and parallel reactions.83 The most important reactions can be described by the following overall equations: Cn Hm þ nH2 O ! nCO þ ðn þ 1=2mÞH2

ð1Þ

Cn Hm þ 2nH2 O ! nCO2 þ ð2n þ 1=2mÞH2

ð2Þ

The latter overall equation includes the formation of CO2, which is formed in the exothermic water gas shift reaction: CO þ H2 O , CO2 þ H2

ð3Þ

In addition to water gas shift, CO and hydrogen can also undergo methanation: CO þ 3H2 , CH4 þ H2 O

ð4Þ

Furthermore, carbon deposition on the catalyst can occur by the following reactions: 2CO , C þ CO2

ð5Þ

Cn Hm , nC þ m=2H2

ð6Þ

Carbon deposition via reactions (5) and (6) and dehydrogenation of larger hydrocarbons would be detrimental to the maintenance of catalyst activity, but these reactions can in principle be avoided by feeding water in excess to operate under high steam/carbon ratios. There have been many studies aimed at minimizing the deposition of carbon during steam reforming.84–89 The need for high steam/carbon ratios makes on-board steam reforming impractical, because of the limited supply of water on board of a vehicle. While many contributions to the steam reforming literature focus on steam reforming of methane or natural gas, there is some literature dealing with the steam reforming of heavier hydrocarbons. For example, Gorte and coworkers have investigated steam reforming of benzene, toluene, cyclohexane and n-octane, molecules that are relevant as surrogates for gasoline reforming.90 This study, conducted on ceria-supported Pd catalysts, demonstrated that paraﬃns are more amenable to steam reforming than aromatic compounds. Benzene in particular had a much higher activation barrier than paraﬃnic molecules. An extensive review of steam reforming of aromatic compounds over Al2O3 and SiO2 supported group VIII metals, published by Duprez, delineated trends that can provide guidance for fuel processing of kerosene-based liquid fuels.91 As a normalized measure of catalytic activity, Duprez compared the turnover frequencies for benzene, alkylbenzenes, alkylnaphthalenes, and heteroatom-containing aromatics. Rh proved to be four to ﬁve times more active than Pt or Ni for overall Catalysis, 2010, 22, 56–93 | 63

steam reforming, encompassing the summation of dealkylation, dehydrogenation, and total gasiﬁcation reactions. For fuel processing aimed at generating hydrogen-rich gas for fuel cells, total gasiﬁcation is the most important reaction. In terms of total gasiﬁcation activity, Rh proved to be the most active catalyst, closely followed by Ni. Pt, however, was less active by an order of magnitude. The activity of Rh appeared to be strongly affected by the nature of the support, while Pt and Ni seemed to be less sensitive. Rh/Al2O3 had much higher activity than Rh/SiO2. There has been some work on Ru-based steam reforming catalysts. For example, Suzuki et al. demonstrated that hydrodesulfurized kerosene could be reformed over a Ru/CeO2–Al2O3 catalyst at 1073 K and a steam to carbon ratio of 3.5, and 100% conversion of hydrodesulfurized kerosene with high H2 yield could be maintained for 8000 h on stream.92 However, under their reaction conditions, signiﬁcant amounts of methane, approximately 5%, were formed. Hu et al. used a proprietary Pd/ZnO catalyst for steam reforming of iso-octane, synthetic diesel, desulfurized JP-8, and JP-8.9 However, it turned out that not only the unmodiﬁed sulfur containing JP-8 fuel, but also desulfurized JP-8 deactivated the catalyst within a few hours on stream. In summary, it can be stated that steam reforming of hydrocarbons typical for kerosene-type fuels is possible under relatively high steam/carbon ratios and at high temperatures as long as catalyst poisons, especially sulfur, are removed. In general, aromatic hydrocarbons are more diﬃcult to reform than paraﬃns. It needs to be noted that steam reforming, due to its slow dynamic response and large water supply requirements, is not a very attractive option for on-board fuel processing systems. 4.2

Catalytic partial oxidation (CPOX)

Partial oxidation of hydrocarbons in presence of substoichiometric amounts of oxygen is an eﬃcient method for generating CO and H2.93 Partial oxidation of hydrocarbons is a highly exothermic process and raises the unconverted reactants and products to very high temperatures. The noncatalytic, homogeneous partial oxidation occurs at temperatures of about 1600–1700 K, but in presence of suitable catalysts, the temperature can be signiﬁcantly lowered to about 1150 K. The incomplete combustion of hydrocarbons leading to formation of CO and H2 is attractive for on-board reforming because it does not require water to reform the fuel, and the reaction has rapid light-oﬀ, providing fast start-up and good transient response. The overall reaction is depicted below: Cn Hm þ 1=2nO2 þ 1:88nN2 ! nCO þ 1=2mH2 þ 1:88nN2

ð7Þ

CPOX reactions are highly exothermic with fast kinetics, and may become mass transfer limited. Nevertheless, the high reaction rates and the concomitant rapid release of heat facilitate rapid start-up of the fuel processor, an important issue for on-board applications. A major concern is the susceptibility of catalysts to coking, as the feed to the reactor does not 64 | Catalysis, 2010, 22, 56–93

contain any steam, which is known to inhibit coke formation. The absence of steam in the feed leads to lower H2 yield compared to steam reforming. It is very important to carefully control the oxygen to carbon ratio (O/C), as even small variations in the O/C ratio can lead to signiﬁcant changes in catalyst bed temperature, product yields, and coking. The sensitivity to O/C ratios and the highly exothermic nature of the process pose a major challenge for controlling CPOX reactors, and there has been major emphasis on developing coke resistant CPOX catalysts.94 There is an extensive body of literature dealing with partial oxidation of methane, but there have also been many studies regarding CPOX of liquid hydrocarbons and transportation fuels.95 Early developments of POX reactors focused on the use of methanol, which thanks to its lower heat of reaction permits a relatively simple reactor design.96 An example of this is the HotSpotTM reactor for methanol fuel processing developed by Johnson Matthey.97 POX technology for gasoline or diesel reforming is very challenging, mainly in terms of temperature control. On the fundamental research side, the Schmidt group at the University of Minnesota has investigated the partial oxidation of C1–C16 hydrocarbons and low sulfur diesel into syngas with Rh catalysts on monoliths at high space velocities under very short contact times.50,98,99 They also explored the contributions of homogeneous and heterogeneous reactions in the catalytic partial oxidation of n-octane, iso-octane, and their mixtures on Rh-coated a-alumina foams.100 They found that the syngas product selectivities were independent of the structure of the reacting fuel, but not the oleﬁn product distribution. Iso-octane produced mostly propylene and isobutylene, while n-octane gave mostly ethylene and propylene. Pengpanich et al. studied the partial oxidation of iso-octane over Ni–Sn/ Ce0.75Zr0.25O2 catalysts.101 It was found that the addition of small amounts of Sn (o0.5 wt%) lowered the activity for iso-octane POX only slightly, while causing a signiﬁcant decrease in the extent of carbon deposition. These results were interpreted in terms of Sn species partially covering the Ni surface, thereby limiting the number of Ni atom surface ensembles available for C–C bond formation and coke build-up, while leaving the active sites for partial oxidation more or less intact. Since the reaction environment in a CPOX reactor is rather complex, involving highly exothermic combustion and partial combustion reactions, followed by endothermic steam reforming using H2O generated in the combustion zone and slightly exothermic water gas shift that becomes signiﬁcant in the cooler regions of the reformer, it would be important to experimentally measure products, intermediates, and temperatures as a function of time and position in the reformer. The boundaries between the diﬀerent reaction zones in the reformer are most likely dynamic, and signiﬁcant temperature and concentration gradients will be present due to variations in the steam/C and O/C ratio. To probe these spatial variations in reforming reactors and to gain insight into the complex interplay of reactions and intermediates, analytical techniques such as Spatially Resolved Capillary Inlet Mass Spectrometry (SpaciMS) have been used by Galen Fisher at Delphi Research Labs in collaboration with a group at ORNL led by William Partridge and Jae-Soon Choi.102 This approach permits mapping of Catalysis, 2010, 22, 56–93 | 65

Fig. 1 Experimental setup for spatially resolved analysis (SpaciMS) of compositions and temperatures in a catalytic fuel reformer (reproduced with permission of Galen B. Fisher).102

composition and temperature proﬁles within the working reactor, in both axial as well as radial directions. The SpaciMS method utilizes capillaries inserted into monolith channels, and the gas collected at various positions in the monolith is fed to a mass spectrometer for analysis. Temperatures are measured by thermocouples. A schematic of the experimental setup is shown in Fig. 1. This method was originally pioneered for time-resolved measurements of emission transients103 and in situ measurement of reactions occurring in lean-NOx trap catalysts.104–107 While diesel lean NOx traps operate at lower temperatures in the range of 473–773 K, reformers reach much higher temperatures creating a challenge for the thermal and mechanical stability of the capillary probes. At ORNL, temporal resolutions of 7–1 Hz have been obtained for capillaries with lengths of 0.4 and 2.5 m, respectively. The capillaries can be moved via a translation device within the monolith channels, thereby probing gas compositions as function of position. To map temperature proﬁles, either thermocouples or phosphor-tipped non-conductive optical ﬁbers can be used, with diﬀerent types of phosphors oﬀering the ability to monitor diﬀerent temperature regimes.108 The SpaciMS analysis has been used to investigate the CPOX reaction of light hydrocarbons, methane, and propane. Fig. 2 shows typical results obtained for propane reforming under O/C of about 1 and at space velocity of 30 000 h 1.109 In the oxidation zone, which is about 0.6 mm wide, syngas is generated and both combustion and then reforming appear to occur, and CO2, H2O, and temperature peaks are observed. This zone is followed by a 0.6 mm wide C3H8 depletion zone, where steam and dry reforming occurs, and CH4 generation is observed. Steam and dry reforming dominate the downstream section of the monolith, and water gas shift equilibrium is established. A similar approach was taken by the Schmidt group110 to obtain species and temperature proﬁles for catalytic partial oxidation of methane on Rh-coated a-Al2O3 foams.111,112 Their experimental set-up included a moveable thin quartz capillary for sampling of gases via mass spectrometry and a thermocouple. Applying such spatially resolved analysis methods to liquid fuel reforming would be very useful. 66 | Catalysis, 2010, 22, 56–93

Fig. 2 Reaction zones in propane reforming measured by SpaciMS (reproduced with permission of Galen B. Fisher).109

4.3

Autothermal reforming (ATR)

Autothermal reforming, a combination of steam reforming and partial oxidation, may be more suitable for on-board reforming than steam reforming because of its better response to transient operation. In autothermal reforming, a portion of the fuel is oxidized by air, and the heat released in the process is used to drive the endothermic steam reforming reactions in the downstream sections of the reactor. The overall heat duty of the reactor can be adjusted by using varying O/C and steam/C ratios. The introduction of superheated steam leads to the onset of steam reforming and water gas shift reactions, causing the downstream section of the catalyst bed to cool. The overall stoichiometry of the autothermal reforming process can be described by equation (8): Cn Hm Oy þ xðO2 þ 3:76N2 Þ þ ð2n 2x yÞH2 O

ð8Þ

¼ nCO2þ ð2n 2x y þ m=2ÞH2 þ 3:76xN2

Since the autothermal reforming process uses both water and oxygen in the feed stream and produces CO2 and H2, it might intuitively appear to be a poor choice for a fuel processor because of CO2 dilution of the SOFC anode feed stream. However, in reality the reaction does not proceed to completion and there is a signiﬁcant concentration of CO present in the reactor eﬄuent, according to the water gas shift equilibrium equation (3). There is a large body of literature on autothermal reforming of methane and light hydrocarbons.113 However, for heavier liquid hydrocarbons, it has proven diﬃcult to establish reaction mechanisms and determine rate laws. One of the major challenges is that the reaction is highly non-isothermal, with fast partial oxidation reactions rapidly raising the temperature at the catalyst bed entrance, followed by gradual decrease in temperature in the Catalysis, 2010, 22, 56–93 | 67

Fig. 3 Reaction zones in autothermal reforming reactor.

downstream sections of the catalyst bed. In addition, there is the possibility of homogeneous and heterogeneous thermal cracking reactions of hydrocarbons. A schematic of these reaction zones in shown in Fig. 3. The photograph of a nickel-containing monolith shows clearly the demarcation between the partial oxidation and steam reforming zones. The generation of hydrogen through autothermal reforming of iso-octane with lowering of CO concentrations through water-gas shift and preferential oxidation was studied by several research groups, for example Moon et al.114 and Thompson and coworkers.115 A comparison of the iso-octane reforming performance of Ni/CeZrO2 catalysts in packed beds versus microreactors showed the importance of properly managing heat transfer eﬀects, an issue that is very critical in microchannel reactors.116 A group at Argonne National Laboratories has worked on reforming liquid hydrocarbons with a Pt catalyst supported on cerium oxide and gadolinium oxide.117 A temperature of 1123 K was required to achieve satisfactory hydrogen yields and bring the methane concentration to low levels. To describe the reaction kinetics relevant for autothermal reforming of iso-octane, Pacheco et al.118 developed a mathematical model using the Aspen Plus process simulator, and using the experimental results obtained by the Argonne team. Palm et al. reformed a mixture of C13–C19 hydrocarbons in presence of precious metal based catalyst, and they achieved high hydrogen and low methane yields.119 To capture the characteristics of naphthenes, aromatics, and sulfur in diesel fuel, which are thought to be responsible for coking and catalyst poisoning, the eﬀect of model compounds 1,2,3,4-tetrahydronaphthalene, decahydronaphthalene and 1-benzothiophene on ATR conversion was also investigated. It was found that decahydronaphthalene did not have a detrimental eﬀect on conversion, but 1,2,3,4-tetrahydronaphthalene gave a decrease in conversion. It was hypothesized that the aromatic character of the 1,2,3,4-tetrahydronaphthalene made it more diﬃcult to reform, thus lowering the conversion. The addition of 1-benzothiophene, at 30 ppm sulfur, sharply decreased the conversion and also altered the temperature proﬁle of the reactor. The changes in temperature proﬁle suggest that the sulfur poisons primarily the endothermic steam reforming sites of the catalyst. Fig. 4 shows a schematic autothermal reforming reaction network for dodecane over ceria-zirconia supported Ni catalysts, taking into account homogeneous and heterogeneous reaction pathways.120 These proposed routes do not explicitly show the contributions of water gas shift, but the contribution of water gas shift is implied by the product distribution. The primary route to syngas is CPOX of dodecane, with steam reforming acting 68 | Catalysis, 2010, 22, 56–93

Fig. 4 Schematic of autothermal reforming reaction network for dodecane, taking into account homogeneous and heterogeneous reaction pathways (adapted from120).

primarily on conversion of C1–C4 hydrocarbons produced in various homogeneous and heterogeneous cracking reactions. A recent study reported the use of perovskite oxides containing rare earth elements for autothermal reforming of isooctane in a ﬁxed bed microreactor.121 It turned out that the binary oxides LaNiO3 and LaCoO3 were active and gave high yields of H2, but suﬀered from structural stability problems, as the oxides decomposed under the reducing conditions of the reformer. Other binary oxides such as LaCrO3, LaFeO3, and LaMnO3 were stable but less active. Similar to steam reforming, aromatic compounds are more diﬃcult to process by ATR than ring and chain compounds. The eﬀect of fuel additives on catalyst activity and durability remain largely unknown. The review of the literature on the processing of JP-8 into syngas suggests that sulfur poisoning is the dominant challenge with coking being the second most important concern.

5. 5.1

Deactivation of reforming catalysts Carbon deposition

A major technical obstacle to reforming of liquid hydrocarbon fuels is carbon deposition and coking on the catalyst. During reforming, hydrocarbons are broken down on catalytic surfaces, their constituent fragments react with steam or oxygen, and CO, CO2, and H2 are released from the catalyst surface. Accumulation of carbon can lead to catalyst deactivation, reactor fouling and pressure drop problems, and in some cases even structural failure of catalysts. However, the detailed mechanisms for carbon Catalysis, 2010, 22, 56–93 | 69

deposition and coke formation remain open to debate, and a clear understanding does not currently exist how the chemical structure of fuels inﬂuences coking behavior. One school of thought is that coke forms by the polymerization of large hydrocarbons into arrays of polycyclics. It is believed that aromatics are detrimental to reforming because the compounds act as precursors for the formation of the polycyclic networks. On nickel catalysts, an additional route for carbon deposition needs to be considered, namely the growth of carbon ﬁlaments. In general, nickel-based catalysts are more susceptible to carbon accumulation than are those based on precious metals. In his general review, Sehested has summarized the development of diﬀerent types of carbon structures formed from higher hydrocarbons during steam reforming on nickel catalysts.122 Of signiﬁcant importance is the mechanism of carbon ﬁlament formation, which can lead to macroscopic catalyst failure beyond simple deactivation. It has been proposed that the formation of ﬁlamentous carbon involves the diﬀusion of carbon atoms formed via decomposition of adsorbed hydrocarbons through Ni particles, forming a nickel carbide intermediate phase.123 One approach to increase the coking resistance of Ni catalysts is the use of supports with mobile oxygen, for example mixed oxides of ceria and zirconia, but even with this support, some carbon deposition mainly in form of ﬁlaments was observed.124,125 Cordierite monoliths loaded with Ni used in the reforming of dodecane showed signiﬁcant carbon deposition following reaction.126 At high nickel loadings (W7 wt%) the carbon deposition was so severe that the monoliths disintegrated during the reaction. TPO of the recovered monoliths showed that carbon deposited during POX, SR, and ATR had diﬀerent oxidation temperatures, indicating diﬀerences in composition or morphology. When Ni/Ce0.75Zr0.25O2 was loaded on the monoliths, carbon deposition was greatly reduced. Carbon deposition was also observed during ATR of isooctane over Ni/Ce0.75Zr0.25O2 catalysts of varying nickel loadings.127 Here the amount of carbon was found to increase with increasing nickel loading, which was associated with increasing average nickel particle size. SEM analysis of the spent catalysts showed carbon ﬁlaments of various sizes, as well as what appeared to be amorphous carbon coating the particles. Chen et al. carried out a detailed spatially-resolved analysis of carbon deposition in monoliths coated with Ni/ceria-zirconia or just Ni alone under ATR, CPOX, and SR conditions.128 They mapped the amount and types of carbon deposited in diﬀerent sections of the monoliths with a combination of scanning electron microscopy, energy-dispersive X-ray spectroscopy (EDS), and temperature-programmed oxidation. They found signiﬁcant diﬀerences in the amounts and location of carbon, depending on the type of reforming reaction, reﬂecting axial variations in temperature and oxygen concentration. Fig. 5 shows a typical SEM image of ﬁlamentous carbon deposited in the downstream sections of a monolith catalyst. An alternative approach to impart better resistance to carbon deposition and coking is to modify the nickel surface. Sn has been successfully used to decrease carbon deposition in several catalytic processes, for example 70 | Catalysis, 2010, 22, 56–93

Fig. 5 Scanning electron micrograph of carbon ﬁlaments formed in pores of the downstream section of a Ni/CZO monolith catalyst after steam reforming of dodecane.128

aromatization and dehydrogenation of paraﬃns. Motivated by the beneﬁcial eﬀect of Sn in these reactions, the eﬀect of Sn was explored for Nicatalyzed steam reforming, partial oxidation,129,130 and dry reforming.131 The beneﬁcial eﬀect of adding small amounts of tin to nickel was also observed on reducible oxides, such as ceria-zirconia.124,125 Nikolla et al. showed by combined DFT and experimental studies that the carbon surface chemistry of nickel catalysts can be very eﬀectively controlled by alloying the nickel surface with tin.132 Long-term activity maintenance of a nickel catalyst is governed by the prevention of carboncarbon bond formation leading to coke while selectively facilitating the formation of C–O bonds. On Ni surfaces, there is no diﬀerentiation in the activation barriers for C–C versus C–O bond formation, while introducing Sn into the Ni surface leads to a situation where the overall rate of carbon oxidation is much greater than the rate of C–C bond formation. It is fortunate that in the limit of small Sn concentrations, the formation of Sn/Ni surface alloys is favored over Sn/Ni bulk alloys or pure Sn and Ni phases.133 The electronic structures of monometallic Ni and Sn/Ni surface alloy catalysts supported on yttria stabilized zirconia, a ceramic used in solid oxide fuel cell anodes were measured using a variety of experimental probes, and the results supported theoretical models indicating a relationship between catalytic activity and the position of the center of the electronic d band.134 Contrary to the common belief that the change in the electronic structure of metals in an alloy is caused by charge transfer among the alloy components, this work showed that the interaction of Sn and Ni leads to shared electronic states. Consequently, the ﬁlling of the Ni d band remains unchanged (Fig. 6). This has important consequences for the catalytic performance of Sn/Ni alloy catalysts. The Sn-induced decrease of the average energy of d electrons decreases the binding strength of carbon species. This lowers the surface concentration of carbon-containing reaction Catalysis, 2010, 22, 56–93 | 71

Fig. 6 Illustration of the interactions between an adsorbate and the d bands of Ni on a monometallic Ni surface and on a Sn/Ni surface alloy.134

intermediates during reforming. The result is a signiﬁcant decrease in the driving force for coke deposition. While more resistant to carbon deposition than nickel, precious metal catalysts do exhibit varying levels of carbon accumulation during reforming. Shekhawat et al. investigated POX of tetradecane over Pt/Al2O3, Pt/ZrCeO2, and Pd/ZrCeO2 catalysts and found 0.85, 0.69, and 0.21 gC/g catalyst after reaction, respectively.48 Carbon deposition levels appeared to be correlated with the presence of unsaturated intermediates. Addition of 5% 1-methylnaphthalene to the feed increased carbon deposition for all the catalysts. Although signiﬁcant carbon was found in the spent catalysts, no signiﬁcant impact on performance was reported. As with several nickel-based investigations, TPO of the spent catalysts revealed very different oxidation temperatures for the carbon species deposited on the catalyst. Dreyer et al. observed carbon deposition during POX and ATR of decane and hexadecane.135 Although the carbon was not strictly quantiﬁed, carbon burn-oﬀ temperature proﬁles indicated there was more carbon deposited under pure POX conditions than ATR. No eﬀect on the catalytic performance due to carbon was noted. The presence of heavy polyaromatic compounds in Diesel creates a tendency for deposition of large amounts of carbon. This can become a serious issue especially under autothermal conditions where the downstream sections of the catalyst bed may experience a decrease in temperature due to endothermic steam reforming reactions. To prevent carbon formation, the reactor temperature must be kept at temperatures above 1070 K. One strategy to decrease carbon formation has been to introduce additives such as tin, to use rare earth oxide supports,136 and to avoid conditions that would lead to the formation of carbides.88 After reviewing carbon deposition data across many investigations it may be concluded that carbon deposition remains a concern for both nickel- and precious-metal based catalysts. It is important to note, however, that the mere presence of carbon on a spent catalyst is not necessarily indicative of carbon poisoning. Surface carbon is a necessary intermediate of reforming and may be expected to be present on all reforming catalysts; it is the quantity and form of carbon that lead to performance issues. 72 | Catalysis, 2010, 22, 56–93

5.2

Deactivation by sulfur

There are various strategies that can be pursued to prevent deactivation of catalysts during reforming of sulfur-containing fuels. The ﬁrst strategy aims at development of catalysts that can tolerate signiﬁcant levels of sulfur without experiencing long-term deterioration of performance.35,137 This is challenging, as sulfur strongly binds to the active sites on the catalyst and prevents access by other reactants. In addition to simple site blocking, sulfur atoms can have detrimental electronic eﬀects on active metals, which can eﬀectively poison surrounding sites. Mitigation of sulfur poisoning requires operation of the catalysts at high temperatures, and may still require the removal of sulfur compounds from the fuel reformer eﬄuent to protect the SOFC anode catalyst from sulfur poisoning. Signiﬁcant research eﬀorts, however, have been directed towards developing hydrocarbon-reforming catalysts capable of tolerating moderate to high concentrations of sulfur. Several factors in catalyst formulation and operation can be used to enhance sulfur tolerance. The most straightforward is the use of a noble metal such as Pt, Pd, or Rh as the active component. Noble metals exhibit lower susceptibility to poisoning by sulfur than other transition metals. Within the noble metals, some appear better than others. Azad’s group at the University of Toledo has reported that 1 wt% Rh or Rh-Pd supported on Gd0.1Ce0.9O1.9 or Zr0.25Ce0.75O2 was stable for steam reforming toluene in presence of 50 ppm thiophene.138 Catalysts of the same formulation where Pd was substituted for Rh showed greater deactivation.139 Schmidt’s group has reported partial oxidation of methane with 28 ppm CH3SH (0.5 wt% Rh/CeO2/Al2O3)140 and JP-8 (5 wt% Rh/Al2O3).135 Catalytic performance was lowered but stable after addition of sulfur to the feed, with a decrease in hydrogen yield observed. Shekhawat et al. directly compared Rh/ZrO2–CeO2, Pt/ZrO2–CeO2, and Pt/Al2O3.48 The Rh/ ZrO2–CeO2 catalyst was the only one to show stable operation under 1000 ppmw S content in the fuel. They also demonstrated that the Rh catalyst can substantially recover activity when sulfur is removed from the feed. While Rh appears to be superior, Pt has also been shown to act as a stable reforming component under some conditions. Lu et al. have reported that 1.5 wt% Pt/ Gd0.1Ce0.9O1.9 was stable for steam reforming of isooctane with a sulfur level of 300 ppmw.75 Some deactivation was observed at sulfur loadings of 500 ppmw, but hydrocarbon conversion was maintained above 90% for 100 hours. The use of reducible metal oxide supports has been shown to be superior to conventional supports such as Al2O3. Cheekatamarla and Lane have compared various bimetallic catalysts supported on both Al2O3 and CeO2 for reforming of JP-8.141 They found that the catalyst activity was higher for all metal combinations when using CeO2 as a support. Mixed oxides including TiO2 and CeO2 are a central claim in BASF’s patent application for a sulfur tolerant natural gas steam reforming catalyst.142 A group at NETL, in conjunction with Spivey, has studied pyrochlores as catalysts for liquid fuel processing. Catalytically active metals may be substituted into the pyrochlore structure, which has excellent thermal properties of interest for high temperature catalytic reactions. La1.5Sr0.5Ru0.05Zr1.95O7 y Catalysis, 2010, 22, 56–93 | 73

(LSRuZ) was compared to Ru/g–Al2O3 for the catalytic partial oxidation of n-tetradecane and n-tetradecane/1-methylnaphthalene/dibenzothiophene mixtures (50 ppmw).64 Both catalysts showed reduction features assignable to Ru during temperature programmed reduction, although the reduction in the pyrochlore took place at higher temperature. The Ru dispersion was 2.3% for the pyrochlore and 27% for the Ru/g–Al2O3 catalyst. The pryochlore catalyst was signiﬁcantly more resistant to deactivation during S-feeding, maintaing H2 and CO yields in excess of 70% with only small increases in CO2 and CH4 production. The Ru/g–Al2O3 catalyst showed catastrophic loss of H2 and CO yields with larger increases in CO2 and CH4 production. The supported Ru catalyst also gave a much higher oleﬁn yield during S-feeding than the pyrochlore. Upon removal of S from the feed, neither catalyst completely recovered its prior activity, but the pyrchlore recovered to a much higher H2 and CO yield than the alumina-supported Ru catalyst. The aluminasupported Ru catalyst also showed much higher C deposition. Rh-substituted La-Zr pyrochlores also showed better S tolerance than comparable supported Rh catalyst.65 During catalytic partial oxidation of n-tetradecane with addition of 1000 ppmw dibenzothiophene, the pyrochlore (La1.50Sr0.50Rh0.10Zr1.90O6.70) showed a drop in H2 and CO yield from 90% to 70–75% on S addition. The original H2 and CO yields were almost totally recovered 200 minutes after removing S from the feed. In contrast, the supported Rh catalyst showed a drop in H2 and CO from 75–80% to less than 40%, with continuing deactivation. Neither H2 nor CO yield recovered to above 50% upon removing S from the feed stream. Oleﬁn production during S-feeding was lower on the pyrochlore than on the supported Rh catalyst. The level of activity recovery was negatively correlated to the amount of C found deposited on the catalyst following reaction. The amounts of C deposited on the catalysts were higher following reaction with S than during partial oxidation with n-tetradecane only.66 Hexaaluminates are oxide materials with a spinel structure and refractory properties. The cations may be chosen from among the transition metals. Some oxygen sites within the framework are more accessible than others, and this has led to their study for oxidation reactions. Hexaaluminates of the formula ANi0.4Al11.6O19 d (A=La, Sr, Ba) have been studied for the catalytic partial oxidation of n-tetradecane.67 The Ba and Sr substituted catalysts showed a b-alumina structure, while the La substituted material exhibited a magnetoplumbite structure. Catalyst surface areas ranged from 14 to 22 m2/g, and nickel dispersions (as measured by H2 chemisorption) were quite low, as expected since Ni is incorporated in the bulk structure. Partial oxidation of n-tetradecane was tested at an O/C ratio of 1.2, 850 1C, and a GHSV of 50 L g 1 h 1. The La-substitued catalyst showed a peak in H2 and CO production, followed by slight decreases then steady performance. This was attributed to carbon deposition deactivating the most active nickel sites. Both the Ba and Sr-substituted catalysts showed no loss in H2 and CO production. All catalysts produced some CO2 and some light hydrocarbons. The tolerance of the catalysts to S was tested by introduction of 50 ppmw dibenzothiophene to the fuel stream during reaction. The La-substituted catalyst showed greater loss of H2 and CO production than 74 | Catalysis, 2010, 22, 56–93

the Sr-substituted catalyst, although it did recover some activity after dibenzothiophene was removed whereas the Sr-substituted catalyst did not. Additives such as promoters or sacriﬁcial components have also been used to improve sulfur tolerance. Azad et al. examined the eﬀect of adding Y2O3 and CuO to Rh-based steam reforming catalysts. Y2O3 had a beneﬁcial eﬀect on performance, apparently by increasing or stabilizing the Rh dispersion.139 CuO was added as a sacriﬁcial component to react with sulfur, thereby removing sulfur from the active metal sites.138 While this improved the hydrogen yield obtained during reforming with sulfur, in the absence of sulfur more coking was observed as compared to the catalyst without CuO. Similarly, BASF’s methane steam reforming catalyst also includes a transition metal component to capture sulfur during reforming, which is released during a subsequent regeneration step.143 Dinka has reported that addition of 2 wt% K to a La0.6Ce0.4Fe0.8Ni0.2O3 perovskite catalyst increased sulfur tolerance, allowing an increase from 50 to 225 ppm S in autothermal reforming of JP-8 with no change in performance.144 The eﬀect of 2 wt% K addition was similar to the eﬀect of 1 wt% Ru. Mawdsley and Krause reported that introducing Cr as stabilizing element into LaNiO3 improved the sulfur tolerance.121 Modiﬁcation of Ni/Sr/ZrO2 catalysts with Re or La was also found to improve the sulfur tolerance of the catalyst during autothermal reforming of hydrocarbon fuels.145 Molybdenum carbide catalysts have also shown some tolerance to sulfur poisoning in a study of steam reforming of tri-methylpentane.146 Many sulfur tolerant catalysts have been studied at modest space velocities, allowing for greater contact time with the catalyst region and opportunity for reaction. Reactions can be carried to high conversion in very short contact times when no sulfur is present in the feed. Schmidt’s group has published extensively on successful CPOX using precious metal catalysts of hydrocarbons from methane to tetradecane at contact times of a few milliseconds.98,147 Schwank’s group at the University of Michigan has demonstrated ATR of isooctane, dodecane, and isooctane at gas hourly space velocities of 200 000 hr 1.148,149 Of the sulfur tolerant catalysts surveyed here, the space velocities range from 8000 hr 1 to 62 000 hr 1. These lower space velocities allow catalysts with lower activities, a result of partial sulfur poisoning, to achieve the required hydrocarbon conversion. The organic sulfur compounds present in liquid fuels can react with catalytic surface sites to form stable metal sulﬁdes, thus causing severe deactivation. One possible solution is to remove sulfur compounds from fuels via selective adsorption on adsorbents such as zeolites that can be regenerated.150,151 Yang and co-workers developed a p-complexation method for selective sorption of sulfur compounds by copper and palladium halide sorbents and CuY zeolites.152–156 They also investigated the use of carbon-based sorbents.157 For on-board applications, the need for regeneration of the sorbents introduces an additional degree of complexity, and in some cases there may also be some issues with eﬀective removal of sterically hindered sulfur compounds in presence of aromatic hydrocarbons. Given the diﬃculties in fuel processing of heavy hydrocarbon fuels, a third strategy may be pursued, namely catalytic hydrodesulfurization of the fuel prior to sending it into the reformer. The heavier organosulfur Catalysis, 2010, 22, 56–93 | 75

compounds in the fuel can be converted into H2S in presence of a suitable catalyst, and the H2S is then absorbed in a sulfur absorption bed, for example zinc oxide. This method is widely used on an industrial scale,158 but since hydrodesulfurization reactors require high pressure H2, it may not be practical for on-board deployment. A group from PNNL has recently described a very interesting alternative.159 Their process uses an integrated steam reformer to generate hydrogen for hydrodesulfurization and a microchannel distillation unit upstream of the hydrodesulfurizer. This approach makes it possible to process a lighter feed fraction instead of the unmodiﬁed JP-8 fuel. The light fraction from the microchannel distillation unit contains smaller concentrations of refractory sulfur compounds, thereby facilitating the hydrodesulfurization. The U.S. Navy has also pursued a fuel processing system capable of handling JP-8 type fuels that utilizes a sulfur-tolerant autothermal reformer (ATR).160 6.

On-board reforming of fuels for SOFC APU applications

Conceptually, three diﬀerent strategies can be used in fuel reforming for SOFC APUs. The ﬁrst strategy is external reforming, where the catalytic conversion of liquid fuels into syngas takes place in a separate catalytic reactor. The H2-rich syngas product is then fed to the anode compartments of the fuel cell stack. From a reaction engineering perspective, external reforming is the simplest method for reforming of complex liquid transportation fuels such as diesel and gasoline, but suﬀers from low overall eﬃciency and high system cost. The second strategy is indirect internal reforming. This method is very similar to external reforming, but in this case the catalytic reformer is designed in such a way that it is in direct thermal contact with the anode compartment of the solid oxide fuel cell. This method was initially applied to molten carbonate fuel cells.161 Having the reforming reactor in thermal contact with the high-temperature fuel cell facilitates better thermal integration of the heat released during the electrochemical reactions on the anode with the heat requirements for vaporizing steam and fuel, and can be utilized to provide additional heat for endothermic steam reforming reactions in the fuel reformer. The practical implementation of this concept has to deal with the problem of thermal mismatch between the relatively fast endothermic steam reforming reactions and the much slower exothermic electrochemical reactions in the fuel cell. This thermal mismatch can lead to cold spots in sections of the reformer, and in a worst case scenario could cause fractures of the system.162 The third possible strategy is direct internal reforming, which is perhaps the ideal heat integration strategy. Due to high SOFC operating temperatures, internal reforming of methane or light hydrocarbons such as propane is possible. Internal reforming is attractive because of the SOFCs ability to utilize CO along with H2. Carrying the reforming reaction out directly in the anode compartment of the fuel cell provides the most eﬃcient way to transfer heat from the electrochemical reactions to the catalytic sites where the reforming reactions take place. The direct reforming of liquid fuels is very challenging, but there have been some reports of successful direct 76 | Catalysis, 2010, 22, 56–93

reforming of iso-octane on carbon-resistant Sn/Ni alloy anodes that were designed with guidance from DFT calculations.163 Barnett and Zhan carried out internal partial oxidation of iso-octane on SOFC anodes, but had to place an additional Ru-CeO2 layer between the fuel stream and the anode to obtain stable operation without anode coking.164 6.1

Gasoline

In view of the extensive infrastructure for gasoline distribution, there have been substantial eﬀorts towards the development of gasoline reformers in industry, involving companies such as Hydrogen Burner Technology and Arthur D. Little,165 which later partnered with the Italian company De Nora to start Nuvera. A. D. Little developed a system that included a POX reactor as primary fuel processor and water gas shift and preferential partial oxidation (PROX) reactors to generate PEM-fuel cell grade pure H2. Gasoline fuel processor development also was carried out at ExxonMobil in collaboration with General Motors. Shell worked on the development of gasoline CPOX reactors for Daimler Chrysler in partnership with Ballard and UTC. Delphi Corporation has been one of the leading developers of on-board SOFC APU technology with integrated fuel processor and SOFC stack,166 in partnership with BMW, Battelle, Global Thermal Electric of Canada, TotalFinaElf, and Los Alamos National Laboratory. In 2001, a ‘‘proof of concept’’ gasoline powered SOFC APU was demonstrated on a BMW 7-series sedan.167 The fuel processor for this system involved a catalytic partial oxidation (CPOX) reformer containing alumina or zirconia based catalyst formulations.168 This fuel processor shown in Fig. 7 does not include provisions for removal of sulfur, requiring that the catalysts must have adequate sulfur tolerance. As strategy for decreasing the size of the reformer, planar

Fig. 7 Delphi Corporation’s gasoline catalytic partial oxidation reformer.168

Catalysis, 2010, 22, 56–93 | 77

geometries have been considered that facilitate heat integration with an energy recovery unit. On a more fundamental basis, researchers at Argonne National Laboratory investigated the eﬀect of the major constituents of gasoline, fuel additives, and impurities on fuel processor performance.169 They found that at high space velocities and/or low reforming temperatures antioxidant additives in gasoline decreased the hydrogen yield in the reformate. 6.2

Diesel

Diesel fuel has a relatively high hydrogen content, making it an attractive fuel for on-board reforming.170 H2-rich reformate gas can be generated from diesel fuel not only through reforming, but also though direct hydrocarbon decomposition. A group at Argonne National Lab has utilized simpliﬁed mixtures of fuel components to understand how factors such as H2O:C and O2:C ratios, temperature, and fuel composition aﬀect the reactions in diesel reforming.171 They indentiﬁed intermediates in the oxidation and coke-forming reactions. While steam reforming provides in principle the highest H2 yield, the endothermic nature of the process and the need to supply large amounts of steam makes this process unfavorable for on-board, transient operation. The exothermic catalytic partial oxidation (CPOX) lends itself much better to transient operation. Diesel CPOX reactors can be operated without catalyst, and after start-up in air, the reactors operate at very high temperatures in excess of 1500 K. By adding catalysts, the reactor temperature can be signiﬁcantly lowered, making such systems attractive for on-board vehicle applications. The major components of diesel fuel, n-decane and nhexadecane, can be converted to syngas with high selectivity.172 CPOX, thanks to its exothermic nature, requires only very short contact times in the order of milliseconds.173 However, catalytic diesel reforming is quite challenging due to the propensity of diesel to pyrolyze, and coke formation and sulfur deactivation of the catalysts are major issues that need to be carefully managed. Due to the large number of hydrocarbon components in the fuel, a multitude of reactions are involved, making it very diﬃcult to unravel mechanistic details.88,174 For diesel reformers, typical catalysts contain either precious metals (Pt, Rh, Ru) or less expensive base metals (Ni) that are supported on carefully engineered oxide supports. For example, Pt/ceria and bimetallic Pt-Pd ceria catalysts175,176 and Pt catalysts supported on Al2O3–CeO2 or Al2O3–La2O3177 have been used for autothermal reforming of synthetic diesel. As Spivey has pointed out in a recent ACS meeting, the tendency of reforming catalysts to deactivate by coke deposition and sulfur poisoning calls for innovative approaches to novel catalytic materials and reactor designs, guided by computational catalysis methods.178 In a study of autothermal diesel fuel reforming over Ru-doped lanthanum chromite and aluminite catalysts, adequate fuel mixing prior to feeding diesel, steam and air into the autothermal reformer was identiﬁed as a critical issue.179 There has been recent patent activity regarding thermoneutral reforming processes for conversion of fuels including light naphtha, heavy naphtha, 78 | Catalysis, 2010, 22, 56–93

kerosene, or diesel in absence of any external heat source.180 These patents claim the use of multicomponent catalysts containing Ni, Ce2O3 La2O3, Pt, ZrO2, Pt, Rh, and Re for reforming of feedstocks containing o 200 pm sulfur without coke formation. Alternate strategies to circumvent carbon formation have been employed. Low and intermediate temperatures can favor the formation of C, as can low local O/C ratios. Mundschau and coworkers have reported the use of a membrane reactor comprising YSZ walls.181 Fuel is fed inside the membrane and air passes through the membrane walls to mix with the fuel above the catalyst bed. The air passing through the walls increases the local O/C ratio in the cool regions of the reactor, suppressing the driving force for C formation. The catalyst employed inside the reactor is a La0.5Sr0.5CoO3–d perovskite, operated at 1223 K to suppress carbon formation in the catalyst region. The perovskites catalyzed both total and partial oxidation of commercial diesel, depending on the O/C ratio employed.

6.3

Jet fuel

As already discussed above, the reforming of kerosene-based jet fuels such as JP-8 is diﬃcult due to the presence of heavy hydrocarbons in these fuels. Among heavy hydrocarbons, paraﬃns and cycloalkanes are relatively easy to convert while the aromatics are known to be the most diﬃcult ones to process. The challenges in reforming heavy aromatics stem both from their lower reactivities as well as from their higher propensities to form coke under typical reforming conditions. Sung and Ibaretta developed a model for reforming of a kerosene surrogate over a wide range of conditions with varying feed temperatures, operating pressures, steam/carbon ratio and O/C ratio.182 Based on calculations using ﬁnite gas-phase chemistry, the concluded that short residence times and partial oxidation with minimal water addition gave the most efﬁcient reforming performance. Gould et al. examined the performance of a nickel-ceria-zirconia catalyst for autothermal reforming of n-dodecane, tetralin, and their mixture, as representative compounds for alkanes and bicyclic compounds in jet fuel.149 It was found that the mixture of tetralin and n-dodecane did not react as expected for a linear combination of the two types of molecules. Instead, the reforming behavior was dominated by reforming characteristics of pure tetralin. This observation appeared to be counterintuitive, as one would expect that the aromatic character of tetralin would make it more diﬃcult to reform. It is well documented that compared to alkanes, aromatic molecules have higher activation energies for steam reforming.183 The simple explanation for the tetralin-dominated reforming behavior of the mixture was that the tetralin-containing mixture led to higher temperature proﬁles in the reactor compared to n-dodecane, due to the higher adiabatic equilibrium temperature obtained with tetralin. As potential low-cost coking-resistant catalysts for reforming of JP-8 fuel surrogate, Ce- and Ni-substituted LaFeO3 perovskites have been used.184 The improved coking resistance was attributed to improved oxygen ion Catalysis, 2010, 22, 56–93 | 79

mobility imparting higher activity for carbon oxidation on the catalyst surface. An alternative strategy for avoiding the coking problem is to remove the heavy hydrocarbons, especially the heavy aromatics from the feedstocks before reforming. In reﬁneries, aromatics can easily be separated from petroleum fractions through solvent extraction or adsorption. Clearly, such strategies are not applicable for compact, on-board fuel processors. The unit operations required for the regeneration of solvent or adsorbent would introduce far too much complexity. Unfortunately, the boiling points of the various hydrocarbon species in these types of fuels are too close to each other for eﬀective separation via distillation. Therefore, alternative approaches have been considered for selective removal of heavy polynuclear aromatics prior to reforming. One such approach involves the catalytic cracking of JP-8, followed by separation of light cracked gases from heavies before reforming, thereby eliminating non-volatile aromatic species.186 Catalytic cracking can convert heavier hydrocarbons to C1–C3 compounds. Since cracking reactions are generally associated with carbon deposition and catalyst deactivation, pre-reforming strategies relying on cracking might require frequent catalyst regeneration. As an alternative, reactive separation of heavy aromatics appears possible, taking advantage of the diﬀerences in the relative reactivities of various hydrocarbons. Prereforming has been practiced widely in industry as a preliminary step to convert distillate fuels for the production of synthesis gas.185 Naphtha feed is processed ﬁrst in a prereformer to produce an equilibrium mixture of methane, hydrogen, CO and CO2 that is then subsequently sent to the main reformer where the conversion to synthesis gas takes place. In this approach, reactive hydrocarbons would be converted into lighter components such as methane, hydrogen and oxides of carbon, while keeping the less reactive heavy aromatics intact. The product mixture will then be cooled down to a temperature where the unconverted heavy aromatics condense as liquid. The methane-rich gaseous stream would be fed into the reformer to produce synthesis gas, while the condensed heavy aromatics would be used as fuel to generate heat. To deal with non-volatile residues that can constitute up to 1.5 vol% in jet fuels, and to decrease the danger of coking due to heavy hydrocarbons, catalytic cracking of the fuel with zeolite catalysts and manganese/alumina catalysts and subsequent separation of the light cracked gas from the non-volatile aromatic species has been investigated.186 The Air Force Research Laboratory (AFRL) has developed a JP-8 fuel processor capable of generating 3kWe of SOFC grade feed reformate.187 Their fuel processor uses a combination of partial fuel vaporization and catalytic cracking to pre-reform the fuel. Their strategy for JP-8 processing is twofold. First, the fuel is partially vaporized so that the bulk of the sulfur containing species remain in the liquid phase, which can be fed to a combustor that heats a steam reformer. Second, the remaining organosulfur compounds are converted into H2S by catalytic cracking, which can subsequently be adsorbed on ZnO. Additionally, the cracking unit converts the larger hydrocarbons, which have a propensity to coke, into smaller noncoking species. 80 | Catalysis, 2010, 22, 56–93

McDermott Technology Inc has developed a fuel processor for Navy ship service capable of converting military logistic fuel (NATO F-76) into a H2 rich gas.188 In conjunction with Siemens Westinghouse and Phillips Petroleum they developed a 250 kWe low-sulfur diesel powered SOFC.189 Their design incorporates an ATR with a sulfur removal unit operation downstream of the ATR to protect the SOFC. They have operated a pilot scale processor at 10–30 kWe on low sulfur diesel (7 ppm) for 110 hours with 99% conversion. A recent study of autothermal reforming of kerosene in a microreformer system used Pt supported on Gd-doped CeO2, but this system was designed for stationary, residential applications, rather than for on-board deployment.190 The reformer was coupled with a downstream ZnO bed to remove sulfur, to protect the SOFC. 7.

Systems engineering aspects of on-board fuel processing

No discussion of on-board fuel reforming can be complete without addressing system and engineering considerations. Issues of reactant delivery, heat integration, and others must be dealt with to achieve on-board reforming of liquid fuels. This balance of plant introduces parasitic losses on the system, reducing the amount of power that can be eﬀectively extracted from the fuel. High operating pressures or long processing trains, which introduce signiﬁcant pressure drops across the system, will lead to increased energy costs for reactant delivery. Separate heat exchange steps will increase the system weight and volume, adversely aﬀecting power density of the full system. The primary driver for system-related considerations is the choice of reforming scheme. While SR oﬀers high hydrogen yields without nitrogen dilution, it requires a supply of water, a high-duty vaporization step, and high rates of heat transfer into the reforming reactor from a separate heat source. ATR also requires water, but the feed ratio to the fuel will be much lower than SR. POX operation requires only air as a co-reactant for the fuel, but may be prone to carbon deposition and higher operating temperatures than SR and ATR. For mobile applications, POX or ATR are likely the most practical reactions schemes. The literature regarding design of reformers and systems integration is relatively sparse. Studies of reforming trains for PEM fuel cells, which include water gas shift and preferential oxidation reactors, are available.191 There has also been an appreciable research eﬀort directed towards membrane reactors, particularly for hydrogen separation using Pd-based membranes, which can accommodate the PEM’s requirement for high-purity hydrogen.192–194 As SOFCs are able to utilize CO as well as hydrogen, the membrane approach is unnecessary. In contrast to large-scale stationary reformer operation,195 on-board systems have to deal with several additional major technical challenges, including frequent transient operation during start-up and shutdown, operation under high space velocities, and ability to function under harsh thermal conditions and mechanical vibrations. Conventional packed-bed reactor technology does not meet these constraints, motivating the move Catalysis, 2010, 22, 56–93 | 81

towards wash-coated monoliths containing either noble or base metals. Such monolith designs have the advantage of lower pressure drops, and they are more responsive to fast transients.196–198 Delphi Corporation teamed up with PACCAR Incorporated and Volvo Trucks North America to deﬁne what system level requirements must be met for diesel fuelled SOFC based auxiliary power units on commercial trucks.199 Battery power was used to bring the SOFC up to operating temperature. Once the reformer was operating in partial oxidation mode and supplied hydrogen-rich syngas to the SOFC, it became possible to recycle anode tail gas, providing additional steam thereby moving from CPOX mode towards autothermal reforming. A potential hazard encountered during warm-up was that at temperatures below 773 K, H2 leaking from the reformer or fuel cell stack could collect in the system and ignite as the temperature increases. To deal with this challenge, sensors had to be added to detect leakage of H2 and CO, and the sensor signal would as necessary trigger appropriate control action to shut the system down. As additional safety precaution, special eﬀorts were made to tightly seal the components to minimize leakage and maximize containment. In principle, water required for ATR can be recovered from the SOFC anode exhaust or tail gas burner, as schematically shown in Fig. 8. If condensation is used as a recovery step, re-vaporization will be required. If water is kept in the gas phase by recycling uncondensed SOFC exhaust, there will be a practical limit to the steam/C ratio that may be obtained as CO2 and N2 will be recycled along with the steam (Fig. 9).200 A group from Los Alamos National Laboratory found that changing the recycle rate from 20% to 30% resulted in a much slower temperature rise in the autothermal catalyst bed and required operation under higher O/C ratios to compensate for the diﬀerences in the catalyst bed temperature

Fig. 8 Schematic of APU with SOFC anode tail gas recycle.

82 | Catalysis, 2010, 22, 56–93

Fig. 9

Steam/carbon ratio as function of anode tail gas recycle ratio.200

proﬁles.201 An ASPEN simulation showed how the condensed exhaust gases could dramatically alter the anode feed.202 The patent literature covering low and high temperature fuel cells, reformers, and combinations of fuel cells and reformers is rich. Over the past decade signiﬁcant eﬀorts have been made to advance fuel cells closer to commercialization. A great portion of the eﬀort, however, has been directed at PEM fuel cells and reforming systems designed to support them. General Motors has disclosed work on heat management in a PEM-supporting reformer system wherein water vaporization is used to control the PrOX reactor temperature and fuel cell exhaust is combusted to increase heat recovery, integration with a water-cooled high-temperature PEM stack, and heat-exchange networks to enable rapid startup.203–205 They have worked on design considerations for decreasing the volume and mass of the reformer train, and managing steam within the fuel processor/fuel cell.206,207 Volvo has also published some work on fuel cell systems, but as with GM it is aimed at motive power and low temperature fuel cells.208 Ford Motor Company made a signiﬁcant number of patent applications regarding fuel cells, but apparently only one that is inclusive of an on-board reformer.209 The company’s focus appeared to be on the fuel cell stack, hydrogen separation membranes, and the use of metal-hydrides as a load leveling and startup reservoir of hydrogen. Delphi has perhaps the strongest patent literature record of fuel cell system integration development. Their work has been geared towards SOFCs as the electrical generation device. Work has covered the range of process considerations including but not limited to startup strategies,210,211 SOFC tail gas recycling,212,213 heat transfer,214 waste heat recovery,215 and temperature control.216 In work which is more applicable to SOFC-APUs, General Motors has investigated solutions to reformer startup using electrical preheat to initiate light-oﬀ.217 This pre-heated reformer is nominally intended for use within Catalysis, 2010, 22, 56–93 | 83

the PEM-supporting fuel processor train disclosed in other General Motors patents. Besides supplying hydrogen rich gas for fuel cell applications, on-board reformers can also be used for hydrogen-assisted engine operation, enhanced rapid 3-way catalyst light-oﬀ in emission control systems, and for providing reductant for enhanced diesel aftertreatment and selective catalytic reduction (SCR) of NOx. In SCR systems, NOx is converted into N2 with reducing agents like ammonia, urea, or hydrocarbons, for example: 3NO2 þ 4NH3 ! 3:5N2 þ 6H2 O

ð9Þ

2NO þ 2NH3 þ 1=2O2 ! 2N2 þ 3H2 O

ð10Þ

10NO þ C3 H8 ! 5N2 þ 3CO2 þ 4H2 O

ð11Þ

2NO þ CH4 þ O2 ! N2 þ 2H2 O þ CO2

ð12Þ

While ammonia and urea, which can be thermally decomposed into ammonia, are very eﬀective for selective catalytic reduction of NOx, they are cumbersome to use, as they require separate tanks to be installed on the vehicle. Furthermore, there is the danger that unconverted ammonia slips through the catalytic converter. An alternative strategy is to rely on hydrocarbons from the fuel as reducing agents.218 Compared to ammonia, hydrocarbons are less eﬀective NOx reducing agents, but adding small amounts of hydrogen can drastically improve the rates of reduction.219 An on-board fuel reformer could provide not only hydrogen, but also CO and light hydrocarbons to facilitate NOx reduction to N2. Not surprisingly, supported Pt catalysts are among the most active catalysts for NOx reduction with H2, but their N2 selectivity is not very good.220 Other group VIII noble metals, such as Pd, are also active and selective at moderate temperatures.221 While CO by itself can be used as a reductant under oxidizing conditions, it appears to be less eﬀective compared to hydrocarbons or hydrogen, but on a Pd/Al2O3 catalyst good NOx conversion and selectivity was achieved with a syngas mixture of CO and H2.222 Another promising catalyst system is Ag/Al2O3.223–228 On Ag catalysts, hydrogen addition to ammonia229 or hydrocarbons230,231 has proven to be very beneﬁcial for NOx conversion and N2 selectivity, and it signiﬁcantly lowers the light-oﬀ temperature of the catalyst.219,232,233 8.

Conclusions

While there has been considerable eﬀort devoted to the development of compact fuel processors for liquid fuels, there are still many challenges remaining that need to be addressed. From a fundamental science standpoint, more detailed reaction kinetics and rate laws need to be determined, not only for pure hydrocarbons, but also for mixtures of hydrocarbons where synergistic eﬀects might be at work, and ultimately for actual fuels. The extension of spatially resolved analysis of axial and radial composition and temperature proﬁles in monolith reactors from light gases to liquid 84 | Catalysis, 2010, 22, 56–93

hydrocarbons and actual fuels appears promising. Guided by densityfunctional theory, the design and synthesis of advanced carbon- and sulfur tolerant catalyst formulations comes within reach. From a reaction-engineering standpoint, major issues that remain to be addressed are adequate performance under transient operation, ability to mitigate coking and sulfur poisoning of catalysts, thermal integration and eﬃciency, and overall system integration within the weight and space constraints of vehicles. There is also a need to demonstrate long term maintenance of catalytic activity under typical on-board conditions with repeated start-up and shutdown cycles. Start-up sequences for auxiliary power units appear to be closely guarded secrets in the industrial community. There is a need for more extensive system level auxiliary power unit research to gain a better understanding of how fuel processors can be best integrated with the fuel cell stacks and the balance of plant. Focusing on optimization of individual components is unlikely to lead to optimized overall system performance.

Acknowledgments The authors would like to acknowledge the ﬁnancial support provided by the U.S. Army Tank-Automotive Research, Development & Engineering Center under Cooperative Agreement Number W56HZV-05-2-0001, and details on the SpaciMS method provided by Galen B. Fisher.

References 1 F. R. Kalhammer, P.R. Prokopius, V. P. Roan and G. E. Voecks, Status and Prospects of Fuel Cells as Automobile Engines – A Report of the Fuel Cell Technical Advisory panel, July 1998, pp. III-27. 2 S. G. Chalk, J. Milliken, J. F. Miller and S. R. Venkateswaran, J. Power Sources, 1998, 71, 26. 3 B. J. Bowers, J. L. Zhao, M. Ruﬀo, R. Khan, D. Dattatraya, N. Dishman, J.-C. Beziat and F. Boudjemaa, Int. J. Hydrogen Energy, 2007, 32, 1437. 4 Y. Chen, H. Xu, Y. Wang, X. Jin and G. Xiong, Fuel Processing Technology, 2006, 87, 971. 5 A. Y. Tonkovich, J. L. Zilka, M. J. LaMont, Y. Wang and R. S. Wegeng, Chem. Eng. Sci., 1999, 54, 2947. 6 R. S. Wegeng, L. R. Pedersen, W. E. TeGrotenhuis and G. A. Whyatt, Fuel Cells Bull., 2001, 28, 8. 7 V. Hessel, V. Cominos, G. Kolb, H. Lo¨we and R. Zapf, Chemical Engineering Transactions, 2004, 4, 373. 8 J. Schwank, A. Tadd and B. Gould, Chemical Engineering Transactions, 2004, 4, 75. 9 J. Hu, Y. Wang, D. VanderWiel, C. Chin, D. Palo, R. Rozmiarek, R. Dagle, J. Cao, J. Holladay and E. Baker, Chem. Engin. J., 2003, 93, 55. 10 J. D. Holladay, E. O. Jones, M. Phelps and J. Hu, J. Power Sources, 2002, 108, 21. 11 P. Pfeifer, M. Fichtner, K. Schubert, M. A. Liauw and G. Emig, Microreaction Technology: Proceedings of 3rd International Conference on Microreactor Technology, Springer Verlag, Berlin, 2000, 372. Catalysis, 2010, 22, 56–93 | 85

12 P. M. Irving, W. Lloyd Allen, T. Healey and W. J. Thomson, Proceedings of 5th International Conference on Microreactor Technology, Springer Verlag, Berlin, 2001, pp. 286. 13 S. V. Karnik, M. K. Hatalis and M. V. Kothare, Proceedings of 5th International Conference on Microreactor Technology, Springer Verlag, Berlin, 2001, pp. 295. 14 P. Reuse and A. Renken, Proceedings of 6th International Conference on Microreactor Technology, New Orleans 2002, AIChE Pub. No. 164, 73. 15 H. Ehwald, G. Ku¨rschner, Smejkal, Europa-Cat-VI, Poster A3.163, 2003. 16 J. N. Armor, Appl. Catal. A, 1999, 176, 159. 17 J. R. Rostrup-Nielsen and T. Rostrup-Nielsen, Cattech, 2002, 6, 150. 18 On-board Fuel Processing Go/No-Go Decision DoE Decision Team Committee Report, Department of Energy, August 2004; http://www1.eere. energy.gov/hydrogenandfuelcells/pdfs/committee_report.pdf 19 H. H. Dobbs, E. T. Kallio, J. M. Pechacek, SAE Paper 2001-01-2792. 20 M. Venturi, A. Martin, SAE Paper 2001-01-2716. 21 M. Venturi, E. Kallio, S. Smith, J. Baker, P. Dhand, SAE Paper 2003-01-2166. 22 J. Zizelman, J. Botti, J. Tachtler, W. Strobl, SAE Paper 2000-01-C070. 23 Z. Filippi, L. Louca, A. Stefanopoulou, J. Pukrushpan, B. Kittirungsi and H. Peng, SAE Paper 2004-01-1477. 24 F. Stodolsky, L. Gaines and A. Vyas, Technology Options to Reduce Truck Idling, Argonne National Laboratory, Argonne, IL, 2001. 25 S. Jain, H. Y. Chen and J. Schwank, J. Power Sources, 2006, 160, 474. 26 J. Larminie and A. Dicks, Fuel Cell Systems Explained, John Wiley & Sons, 2003. 27 Fuel Cell Technology Handbook, G. Hoogers (Ed.), CRC Press 2003. 28 J. H. Hirschenhofer, D. B. Stauﬀer, R. R. Engleman, M. G. Klett, Fuel Cell Handbook, 4th ed., DOE/FETC-99/1076, US Department of Energy, Federal Energy Technology Center, Morgantown, WV, November 1998. 29 R. O’Hare, S.-W. Cha, W. Colella and F. B. Prinz, ‘‘Fuel Cell Fundamentals’’, John Wiley & Sons Inc, New York, 2006. 30 Y. Jamal and M. L. Wyszynski, Int. J. Hydrogen Energy, 1994, 19, 557. 31 M. Prigent, Revue De L Institut Francais Du Petrole, 1997, 52, 349. 32 M. Krumpelt, T. R. Krause, J. D. Carter, J. P. Kopasz and S. Ahmed, Catal. Today, 2002, 77, 3. 33 A. F. Ghenciu, Curr. Opin. Solid State Mater. Sci., 2002, 6, 389. 34 P. K. Cheekatamarla and C. M. Finnerty, J. Power Sources, 2006, 160, 490. 35 C. S. Song, Catal. Today, 2002, 77, 17. 36 A. D. Qi, B. Peppley and K. Karan, Fuel Processing Technology, 2007, 88, 3. 37 L. Hartmann, Ch. Mengel, K. Lucka, H. Koehne, VDI-Berichte 2003, 1750 (Verbrennungen), 461–468, VDI Verlag GmbH. 38 H. Koehne, S. Wenzel, K. Lucka, M. Konrad, and S. Koehne, VDI-Berichte 2008, 2036 (Brennstoﬀzelle), 365–375, VDI Verlag GmbH. 39 T. A. Semelsberger, L. F. Brown, R. L. Borup and M. A. Inbody, Int. J. Hydrogen Energy, 2004, 29, 1047. 40 D. Shekhawat, D. A. Berry, T. H. Garner and J. J. Spivey, Catalyis, 2006, 19, 184. 41 M. Flytzani-Stephanopoulos and G. E. Voecks, Energy Progress, 1981, 1, 53. 42 T. Giroux, S. Hwang, Y. Liu, W. Ruettinger and L. Shore, Appl. Catal. B, 2005, 56, 95. 43 J. Hu, Y. Wang, D. VanderWiel, C. Chin, D. Palo, R. Rozmiarek, R. Dagle, J. Cao, J. Holladay and E. Baker, J. Chem. Eng., 2002, 4049, 1. 44 Q. Ming, T. Healey, L. Allen and P. Irving, Catal. Today, 2002, 77, 51. 86 | Catalysis, 2010, 22, 56–93

45 P. K. Cheekatamarla and A. M. Lane, J. Power Sources, 2006, 153, 157. 46 R. K. Kaila and A. O. I. Krause, Int. J. Hydrogen Energy, 2006, 31, 1934. 47 J. Pasel, P. Cremer, B. Wegner, R. Peters and D. Stolten, J. Power Sources, 2004, 126, 112. 48 D. Shekhawat, T. H. Gardner, D. A. Berry, M. Salazar, D. J. Haynes and J. J. Spivey, Appl. Catal. A, 2006, 311, 8. 49 R. L. Borup, M. A. Inbody, T. A. Semelsberger, J. I. Tafoya and D. R. Guidry, Catal. Today, 2005, 99, 263. 50 J. J. Krummenacher, K. N. West and L. D. Schmidt, J. Catal., 2003, 215, 332. 51 C. Palm, P. Cremer, R. Peters and D. Stolten, J. Power Sources, 2002, 106, 231. 52 B. Lenz and T. Aicher, J. Power Sources, 2005, 149, 44. 53 G. Petitpas, J. D. Rollier, A. Darmon, J. Gonzalez-Aguilar, R. Metkemeijer and L. Fulcheri, Int. J. Hydrogen Technology, 2007, 32, 2848. 54 M. Andersen, O. Lytken, J. Engbek, G. Nielsen, N. Schumacher, M. Johansson and I. Chorkendorﬀ, Catal. Today, 2005, 100, 191. 55 S. Ayabe, H. Omoto, T. Utaka, R. Kikuchi and K. Eguchi, Appl. Catal. A: General, 2002, 63, 1. 56 B. A. Peppley, J. C. Amphlett, L. M. Kearns and R. F. Mann, Appl. Catal. A: General, 1999, 179, 31. 57 M. Ni, D. Y. C. Leung and M.K. H. Leung, Int. J. Hydrogen Technology, 2007, 32, 3238. 58 T. Edwards and L. Q. Maurice, Journal of Propulsion and Power, 2002, 17, 461. 59 A. Violi, S. Yan, E. G. Eddings, A. Saroﬁm, S. Granata, T. Faravelli and E. Ranzi, Comb. Sci. and Tech., 2002, 174, 399. 60 Handbook of Aviation Fuel Properties, Coordinating Research Council Inc., Report No. 530, (1983). 61 http://chemtrails.cc/docs/Stadis450_antistatic additive MSDS.pdf 62 F. B. Passos, E. R. de Oliveira, L. V. Mattos and F. B. Noronha, Catal. Today, 2005, 101, 23. 63 N. Laosiripojana and S. Assabumrungrat, Appl. Catal. A: General, 2005, 290, 200. 64 D. J. Haynes, A. Campos, D. A. Berry, D. Shekhawat, A.Roy, and J. J. Spivey, Catal. Today, (2009), doi:10.1016/j.cattod.2009.03.025. 65 D. J. Haynes, D. A. Berry, D. Shekhawat and J. J. Spivey, Catal. Today, 2009, 145, 121. 66 D. J. Haynes, D. A. Berry, D. Shekhawat and J. J. Spivey, Catal. Today, 2008, 136, 206. 67 T. H. Gardner, D. Shekhawat, D. A. Berry, M. W. Smith, M. Salazar and E. Kugler, Appl. Catal. A: General, 2007, 323, 1. 68 Fundamentals of Industrial Catalytic Processes, 2nd ed., C. H. Bartholomew and R. J. Farrauto, Wiley-Interscience, New York 2005. 69 Catalyst Preparation: Science and Engineering, J. R. Regalbuto, ed., Taylor and Francis/CRC Press, 2006. 70 J. Haber, J. H. Block and B. Delmon, Pure & Appl. Chem., 1995, 67, 1257. 71 Catalyst Characterization: Physical Techniques for Solid Materials, B. Imelik and J. C. Vedrine, Plenum Press, New York, 1994. 72 Analytical Methods in Fine Particle Technology, C. Orr , P. A. Webb, R. W. Camp and J. P. Oliver, Micromeritics Instrument Corp. 1997. 73 M. Ferrandon, J. Mawdsley and T. Krause, Appl. Catal. A: General, 2008, 342, 69. 74 J. J. Strohm, J. Zheng and C. Song, J. Catal., 2006, 238, 309. Catalysis, 2010, 22, 56–93 | 87

75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93

94 95 96 97 98 99 100 101 102 103

104 105 106 107 108

Y. Lu, J. Chen, Y. Liu, Q. Xue and M. He, J. Catal., 2008, 254, 39. H. H. Ibrahim, P. Kumar and R. O. Idem, Energy and Fuels, 2007, 21, 570. A. S. Mukasyan and P. Dinka, J. Power Sources, 2007, 167, 472. A. Qi, S. Wang, G. Fu and D. Wu, D. Appl. Catal. A: General, 2005, 293, 71. A. C. McCoy, M. J. Duran, A.-M. Azad, S. Chattopadhyay and M. A. Abraham, Energy Fuels, 2007, 21, 3513. L. Mond and C. Langer, British Patent 12608 (1888). A. Mittasch and C. Schneider, U.S. Patent 1,128,804 (1915). J. R. Rostrup-Neilsen, Catalysis: Science and Technology, vol. 5, Chap. 1, Catalytic Steam Reforming, Springer-Verlag, 1984. J. Rostrup-Nielsen, J. Catal., 1973, 31, 173. J. Rostrup-Nielsen and D. L. Trimm, J. Catal., 1977, 48, 155. J. R. Rostrup-Nielsen, NATO ASI Ser. Ser. E: Appl. Sci., 1982, 54, 127. D. L. Trimm, Catal. Rev.-Sci. Eng., 1977, 16, 155. T. Borowiecki, A. Machocki and J. Ryczkowski, Stud. Surf. Sci. Catal., 1994, 88, 537. D. L. Trimm, Catal. Today, 1999, 49, 3. E. Nikolla, J. Schwank, and S. Linic, J. Catal. 2009, doi:10.1016/j.jcat.2009. 02.006 (in press). X. Wang and R. J. Gorte, Appl. Catal., A: General, 2002, 224, 209. D. Duprez, Appl. Catal. A: General, 1992, 82, 111. T. Suzuki, H. Iwanami and T. Yoshinari, Int. J. Hydrogen Energy, 2000, 25, 119. J. P. Kopasz, R. Wilkenhoener, S. Ahmed, J. D. Carter and M. Krumpelt, Preprints of Symposia- American Chemical Society, Division of Fuel Chemistry, 1999, 44(4), 899. K. A. Avci, Z. I. Onsan and D. L. Trimm, Appl. Catal. A, 2001, 216, 243. W. Yanhui and W. Diyong, Int. J. Hydrogen Energy, 2001, 26, 795. R. Kumar, S. Ahmed and M. Krumpelt, Proceedings of the US Fuel Cell Seminar, Kissimee, Florida, November 1996, 17. N. Edwards, S. R. Ellis, J. C. Frost, S. E. Golunski, A. N. J. van Keulen, N. G. Lindewald and J. G. Reinking, J. Power Sources, 1998, 71, 123. L. D. Schmidt, E. J. Klein, C. A. Leclerc, J. J. Krummenacher and K. N. West, Chem. Engin. Science, 2003, 58, 1037. L. D. Schmidt, E. J. Klein, R. P. O’Connor, and S. Tummala, Abstracts of Papers of the American Chemical Society, 2001, 222 (FUEL Part 1) 111. G. J. Panuccio, K. A. Williams and L. D. Schmidt, Chem. Engin. Sci., 2006, 61, 4207. S. Pengpanich, V. Meeyoo, T. Rirksomboon and J. Schwank, Catal. Today, 2008, 136, 214. Galen B. Fisher, personal communication. W. P. Partridge, J. M. E. Storey, S. A. Lewis, R. W. Smithwick, G. L. DeVault, M. J. Cunningham, N. W. Currier, T. M. Yonushonis, SAE Technical Paper, 2000-01-2952. J.-S. Choi, W. P. Partridge and C. S. Daw, Appl. Catal. A, 2005, 293, 24. J.-S. Choi, W. P. Partridge, W. S. Epling, N. W. Currier and T. M. Yonushonis, Catal. Today, 2006, 114, 102. B. H. West, S. P. Huﬀ, J. E. Parks, S. A. Lewis, J.-S. Choi, W. P. Partridge and J. M. Storey, SAE Technical Paper 2004-01-3023. J. Parks, S. Huﬀ, J. Pihl, J.-S. Choi and B. West, SAE Technical Paper 200501-3876. S. W. Allison and G. T. Gillies, Rev. Sci. Instrum., 1997, 68, 2615.

88 | Catalysis, 2010, 22, 56–93

109 G.B. Fisher, E. Ta, J. Weissman, K. Rahmoeller, K. Wadumesthrige, C. DiMaggio and J. Kirwan, Presentation at Danish Technical University, May 27, 2008. 110 R. Horn, N. J. Degenstein, K. A. Williams and L. D. Schmidt, Catal. Lett., 2006, 110, 169. 111 R. Horn, K. A. Williams, N. J. Degenstein and L. D. Schmidt, J. Catal., 2006, 242, 92. 112 R. Horn, K. A. Williams, N. J. Degenstein and L. D. Schmidt, Chem. Engin. Sci., 2007, 62, 1298. 113 W. Wang, S. M. Stagg-Williams, F. B. Noronha, L. V. Mattos and F. B. Passos, Abstracts of Papers, 227th ACS National Meeting, Anaheim, CA, March 28–April 1, 2004, FUEL-054. 114 D. J. Moon, K. Sreekumar, S. D. Lee and H. S. Kim, Appl. Catal. A: Gen., 2001, 215, 1. 115 C. H. Kim, A. R. Tadd, S. Srinivas, G. Y. Kim, E. Gulari, J. Ni, J. Schwank, and L. T. Thompson, ‘‘Micro-channel Fuel Processor for Portable Fuel Cells’’, AIChE 2005 Annual Meeting, Cincinnati, OH, Oct. 30–Nov. 4, 2005. 116 A. R. Tadd, B. D. Gould and J. W. Schwank, Catal. Today, 2005, 110, 68. 117 D. Liu, T. D. Kaun, H. Liao and S. Ahmed, Int. J. Hydrogen Energy, 2004, 29, 1035. 118 M. Pacheco, J. Sira and J. Kopasz, Appl. Catal. A: General, 2003, 250, 161. 119 C. Palm, C. Cremer, R. Peters and D. Stolten, J. Power Sources, 2002, 106, 231. 120 B. D. Gould, X. Chen and J. Schwank, J. Catal., 2008, 334, 277. 121 J. R. Mawdsley and T. T. Krause, Appl. Catal. A: General, 2008, 334, 311. 122 J. Sehested, Catal. Today, 2006, 111, 103. 123 M. Perez-Cabero, E. Romeo, C. Royo, A. Monzo´n, A. Guerrero-Ruı´ z and I. Rodriguez-Ramos, J. Catal., 2004, 224, 197. 124 S. Pengpanich, V. Meeyoo and T. Rirksomboon, Catal. Today, 2004, 93–95, 95. 125 S. Pengpanich, V. Meeyoo, T. Rirksomboon and J. Schwank, Appl. Catal. A, 2006, 302, 133. 126 B. D. Gould, X. Chen and J. Schwank, Appl. Catal. A General, 2008, 334, 277. 127 X. Chen, A. R. Tadd and J. W. Schwank, J. Catal., 2007, 251, 374. 128 X. Chen, B. D. Gould and J. W. Schwank, Appl. Catal. A: General, 2009, 356, 137. 129 N. N. Nichio, M. L. Casella, G. F. Santori, E. N. Ponzi and O. A. Ferretti, Catal. Today, 2000, 62, 231. 130 M. J. Ferreira, N. N. Nichio and O. A. Ferretti, J. Mol. Catal. A, 2003, 202, 197. 131 Z. Hou. O.Yokota, T. Tanaka and Y. Yashima, Appl. Surf. Sci., 2004, 233, 58. 132 E. Nikolla, A. Holewinski, J. Schwank and S. Linic, J. Am. Chem. Soc., 2006, 128, 11354. 133 E. Nikolla, J. Schwank and S. Linic, J. Catal., 2007, 250, 85. 134 E. Nikolla, J. Schwank and S. Linic, J. Am. Chem. Soc., 2009, 131, 2747. 135 B. J. Dreyer, I. C. Lee, J. J. Krummenacher and L. D. Schmidt, Appl. Catal. A: General, 2006, 307(2), 184. 136 Z. A. Quan, Y. N. Qin and C. Liu, Applied Catalysis, 1991, 70, 1. 137 L. S. Wang, K. Murata and M. Inaba, Appl. Catal. A: General, 2004, 257, 43. 138 A.-M. Azad and M. J. Duran, Appl. Catal. A: General, 2007, 330, 77. 139 A.-M. Azad, M. J. Duran, A. K. McCoy and M. A. Abraham, Appl. Catal. A: General, 2007, 332(2), 225. Catalysis, 2010, 22, 56–93 | 89

140 A. Bitsch-Larsen, N. J. Degenstein and L. D. Schmidt, Appl. Catal. B: Environmental, 2008, 78, (3-4), 364. 141 P. K Cheekatamarla and A. M. Lane, J. Power Sources, 2006, 154, 223–231. 142 T. Giroux, E. Waterman, R. Farrauto, US Patent Application 2008/0041766. 143 T. Giroux, E. Waterman, R. J. Farrauto, US Patent Application 2008/ 0011646. 144 P. Dinka and A. S. Mukasyan, J. Power Sources, 2007, 167(2), 472. 145 K. Murata, M. Saito, M. Inaba and I. Takahara, Appl. Catal. B: Environmental, 2007, 70, 509. 146 P. K. Cheekatamarla and W. J. Thomson, Appl. Catal. A: General, 2005, 287, 176. 147 N. J. Degenstein, R. Subramanian and L. D. Schmidt, Appl. Catal. A: General, 2006, 305, 146. 148 B. D. Gould, X. Chen and J. Schwank, J. Catal., 2007, 250, 209. 149 B. D. Gould, A. R. Tadd and J. Schwank, J. Power Sources, 2007, 164, 344. 150 R. T. Yang, Catal. Rev., 2004, 46, 111. 151 S. Velu, X. Ma and C. Song, Ind. Eng. Chem. Res., 2003, 42, 5293. 152 A. J. Hernandez-Maldonado, G. S. Qi and R. T. Yang, Appl. Catal. B, 2005, 61, 212. 153 A. J. Hernandez-Maldonado, H. H. Yang, G. S. Qi and R. T. Yang, J. Chinese Inst. Chem. Eng. (James Wei Festschrift Issue), 2006, 37, 1. 154 Y. H. Wang, F. H. Yang, R. T. Yang, J. M. Heinzel and A. D. Dickens, Ind. Eng. Chem. Res., 2006, 45, 7649. 155 A. Jayaraman, F. H. Yang and R. T. Yang, Energy & Fuels, 2006, 20, 909. 156 Y. W. Li, F. H. Yang, G. S. Qi and R. T. Yang, Catal. Today, 2006, 116, 512. 157 Y. H. Wang and R. T. Yang, Langmuir, 2007, 23, 3825. 158 H. Topsøe, B. S. Clausen and F. Massoth, Hydrotreating Catalysis, in: Catalysis Science and Technology, J. R. Anderson and M. Boudart, (Eds.), Vol. 11, Springer, 1996. 159 X. Huang, D. A. King, F. Zheng, V. S. Stenkamp, W. E. TeGrotenhuis, B. Q. Roberts and D. L. King, Catal. Today, 2008. 160 S. Roychoudhury, M. Lyubovsky, D. Walsh, D. Chu and E. Kallio, J. Power Sources, 2006, 160(1), 510. 161 S. Freni and G. Maggio, International Journal of Energy Research, 1997, 21, 253. 162 P. Aguiar, D. Chadwick and L. Kershenbaum, Chemical Engineering Science, 2002, 57, 1665. 163 E. Nikolla, J. W. Schwank and S. Linic, Catalysis Today, 2008, 136, 243. 164 S. A. Barnett and Z. Zhan, J. Power Sources, 2006, 155, 353. 165 W. T. Teagan, J. Bentley and B. Barnett J. Power Sources, 1998, 71, 80. 166 P. Lamp, J. Tachtler, O. Finkenwirth, S. Mukerjee and S. Shaﬀer, Fuel Cells, 2003, 3(3), 146. 167 J. Zizelman, S. Shaﬀer, S. Mukerjee, Solid Oxide Fuel Cell Auxiliary Power Unit – A Development Update, SAE Technical Paper Series, 2002-01-0411, (2002). 168 S. Shaﬀer, 2004 SECA Review Meeting, Boston, MA, URL: www.netl.doe. gov/publications/proceedings/04/seca-wrkshp/Delphi%20-%20Shaﬀer.pdf. 169 J. P. Kopasz, D. Applegate, L. Miller, and C. Rossignol, ‘‘Hydrogen, Fuel Cells, and Infrastructure Technologies’’, 2003 Progress Report. http:// www1.eere.energy.gov/hydrogenandfuelcells/pdfs/ivd13_kopasz.pdf 170 C. Pereira, J. M. Bae, S. Ahmed and M. Krumplet, Liquid Fuel reformer development: ATR of diesel fuel, in: US Department of Energy, 2000, Hydrogen Program Technical Review, San Ramon, CA. 90 | Catalysis, 2010, 22, 56–93

171 J. P. Kopasz, D. Applegate, L. Miller, H. K. Liao and S. Ahmed, Int. J. Hydrogen Energy, 2005, 30, 1243. 172 R. Subramanian, G. J. Panuccio, J. J. Krummenacher, I. C. Lee and L. D. Schmidt, Chem. Engin. Science, 2004, 59, 5501. 173 R. P. O’Connor, E. J. Klein and L. D. Schmidt, Catal. Lett., 2000, 70, 99. 174 S. Springmann, G. Friedrich, M. Himmen, M. Sommer and G. Eigenberger, Appl. Catal. A: General, 2002, 235, 101. 175 P. K. Cheekatamarla and A. M. Lane, J. Power Sources, 2005, 152, 256. 176 P. K. Cheekatamarla and A. M. Lane, Int. J. Hydrogen Energy, 2005, 30, 1277. 177 M. C. Alvarez-Galvan, R. M. Navarro, F. Rosa, Y. Briceno, M. A. Ridao and J.L. G. Fierro, Fuel, 2008, 87, 2502. 178 J. J. Spivey, Abstracts of Papers, 232nd ACS National Meeting, San Francisco, CA, Sept. 10–14, 2006, FUEL-136 179 D.-J. Liu, M. Krumpelt, H.-T. Chien and S.-H. Shen, J. Mat. Engineering and Performance, 2006, 15, 442. 180 T. Inui, B. O. Dabbousi, S. Ahmed, F. I. Al-Muhaish,and M.A.B. Siddiqui, Application WO 2007-US26384 20071226 and US 2006-.645970 20061226. 181 M. V. Mundschau, C. G. Burk and D. A. Gribble, Jr., Catal. Today, 2008, 136, 190. 182 C.-J. Sung and A. F. Ibaretta, Int. J. Hydrogen Energy, 2006, 31, 1066. 183 X. Wang and R. J. Gorte, Appl. Catal. A, 2002, 224, 209. 184 A. Varma, P. Erri and P. Dinka, Chem. Engin. Sci., 2006, 61, 5328. 185 F. Joensen and J. R. Rostrup-Nielsen, J. Power Sources, 2002, 105, 195. 186 T. J. Campbell, A. H. Shaaban, F. H. Holcomb, R. Salavani and M. J. Binder, J. Power Sources, 2004, 129, 81. 187 R. Salavani, A. H. Shaaban, T. Campbell, M. Sawyer, R. Trotta, R. Hallett, Logistics Fuel Processor Development, AFRL-ML-TY-TR-2004-4506, (2004). 188 Annual Meeting Archive American Institute of Chemical Engineers, Indianapolis, IN, Nov. 3–8, 2002, 2775–2785. 189 G. A. Farthing, R. H. Pahl, S. E. Veyo, Diesel–Powered Solid Oxide Fuel Cell, URL: http://www.sofco-efs.com/tppdfs/02-24.pdf 190 S. Yoon, J. Bae, S. Kim and Y.-S. Yoo, J. Power Sources, 2009, 192, 360. 191 J. B. Pearlman, A. Bhargav, E. B. Shields, G. S. Jackson and P. L. Hearn, J. Power Sources, 2008, 185, 1056. 192 A. S. Damle, J. Power Sources, 2009, 186, 167. 193 A. S Damle, J. Power Sources, 2008, 180, 516. 194 M. J. Varady, L. McLeod, J. M. Meacham, F. L. Degertekin and A. G. Fedorov, J. Micromechanics and Microengineering, 2007, 17, S257–S264. 195 E. H. Sitt, NATO Science Series, II: Mathematics Physics and Chemistry, 2005, 191, 185. 196 D. A. Hickman and L. D. Schmidt, Science, 1993, 259, 343. 197 T. Giroux, Prepr. Pap.-Am. Chem. Soc., Div. Fuel Chem., 2005, 50, 463. 198 C. A. Leclerc, J. M. Redenius and L. D. Schmidt, Catal. Lett., 2002, 79, 39. 199 G. M. Simopoulos, DE-FC36-04GO14319 Project Report, May 24, 2005. 200 B. D. Gould and J. Schwank, ‘‘Autothermal Reforming of Jet Fuel for SOFC APUs’’, Paper 460f, AIChE 2005 Annual Meeting, Cincinnati, OH, Oct. 30–Nov. 4, 2005. 201 R. L. Borup, M. A. Inbody, J. I. Tafoya, D. R. Guidry and W. J. Parkinson, 205th Meeting of the Electrochemical Society, 2004, Abs.670. 202 F. Baratto, U. M. Diwekar and D. Manca, J. Power Sources, 2005, 139, 205. Catalysis, 2010, 22, 56–93 | 91

203 S. D. Burch, S. G. Goebel, and W. H. Pettit, W. H. Fuel Processor Thermal Management System. WO 03/060043 A1, 2002. 204 S. D. Burch, and J. C. Fagley, High-eﬃciency Fuel Processor via Steam Integration From a Water-Cooled Stack. US 2005/0019626 A1, 2005. 205 S. G. Goebel, W. H. Pettit, S. D. Burch, P. T. Yu, Y. Zhang and M. D. Cartwright, Integrated Fuel Processor for Rapid Start and Operational Control. US 2005/0091922 A1, 2005. 206 S. D. Burch and W. H. Pettit, W. H. Methods for Reducing Mass and Volume of a Fuel Processor. WO 03/092108 A1, 2003. 207 D. H. Keskula, B. J. Clingerman, S. D. Burch and P. T. Yu, System and Method for Controlling Steam in a Fuel Cell System with a Fuel Processor. WO 03/094277, 2003. 208 O. Duebel, A. Koenig, P. Ekdunge, P. Alin, J. G. Reinkingh, and R. Mallant, Fuel Cell System and Method for Generating Electrical Energy using a Fuel Cell System. US 7,160,638 B1, 2007. 209 B. D. James, Fuel Reformer System for a Fuel Cell. US 6,329,901 B1, 2001. 210 M. J. Grieve, K. J. Haltiner, P. Hendler and J. Noetzel, Fuel Reformer System. US 2007//089368, 2007. 211 G. B. Fisher, and J. E. Kirwan, J. E. Fast Light-oﬀ Catalytic Reformer. US 2005132650, 2005. 212 S. M. Kelly, G. E. Geiger, J. A. Macbain, S. R. Shaﬀer, Anode tail Gas Recycle Cooler and Re-Heater for a Solid Oxide Fuel Cell Stack Assembly. EP 1 724 869 A1, 2006. 213 B. A. Fischer, Method and Apparatus for Fueling a Solid Oxide Fuel Cell Stack. EP 1 998 398 A2, 2008. 214 J. F. O’Brien and F. V. Oddi, Integrated reformer and heat exchanger. EP1 462 159 A2, 2004. 215 S. M. Kelly, K. J. Haltiner, M. T. Faville and D.R. Schumann, Solid-oxide Fuel Cell System having an Integrated Reformer and Waste Energy Recovery System. EP 1 447 874, 2004. 216 M. J. Grieve, Temperature/reaction management system for fuel reformer systems. EP 1 231 664, 2002. 217 S. D. Burch, W. H. Pettit, S. G. Goebel, Fuel Processor Primary Reactor and Combustor Startup via Elecrically-heated Catalyst. US2004154222A1, 2004. 218 W. Held, A. Ko¨nig, T. Richter, and L. Puppe, SAE Tech. Paper 1990 Ser. No. 900496. 219 S. Satokawa, Chem. Lett., 2000, 294. 220 R. Burch and M. D. Coleman, Appl. Catal. B: Environmental, 1999, 23, 115. 221 G. S. Qi, R. T. Yang and F. C. Rinaldi, J. Catal., 2006, 237, 381. 222 Y. W. Lee and E. Gulari, Catal. Comm., 2004, 5, 499. 223 K. A. Bethke and H. H. Kung, J. Catal., 1997, 172, 93. 224 T. Nakatsuji, R. Yasukawa, K. Tabata, K. Ueda and M. Niwa, Appl. Catal. B: Environ., 1998, 17, 333. 225 J. Shibata, K. Shimizu, S. Satokawa, A. Satsuma and T. Hattori, Phys. Chem. Chem. Phys., 2003, 5, 2154. 226 R. Burch, Catal. Rev., 2004, 46, 271. 227 J. Shabata, K. Shimidzu, Y. Takada, A. Shichi, H. Yoshida, S. Satokawa, A. Satsuma and T. Hattori, J. Catal., 2004, 227, 367. 228 P. Sazama, L. Capek, H. Drobna´, Z. Sobalı´ k, J. Dedecek, K. Arve and B. Wichterlova´, J. Catal., 2005, 232, 302.

92 | Catalysis, 2010, 22, 56–93

229 M. Richter, U. Bentrup, R. Eckelt, M. Schneider, M. M. Pohl and R. Fricke, Appl. Catal. B, 2004, 51, 261. 230 T. Miyadera, Appl. Catal. B, 1993, 2, 199. 231 T. Miyadera and K. Yoshida, Chem Lett., 1993, 1483. 232 K. Era¨nen, F. Klingstedt, K. Arve, L. E. Lindfors and D. Y. Murzin, J. Catal., 2004, 227, 328. 233 S. Satokawa, J. Shibata, K. Shimizu, A. Satsuma and T. Hattori, Appl. Catal. B: Environmental, 2003, 42, 179.

Catalysis, 2010, 22, 56–93 | 93

Coupling kinetic and spectroscopic methods for the investigation of environmentally important reactions F. C. Meuniera DOI: 10.1039/9781847559630-00094

1.

Introduction

The improvement of the activity and selectivity of catalysts is a perpetual objective for researchers in catalysis and can rely on a number of approaches. While high-throughput combinatorial methods are raising a lot interest and ﬁnding some success in heterogeneous catalysis,1 the understanding of reaction mechanism through detailed kinetic and spectroscopic studies is another proven approach to support catalyst development. The purpose of this report is to present some examples as to how insights into catalyst structure and/or reaction mechanisms can be obtained from combining in situ/ operando spectroscopic data and kinetics (i.e. ‘‘spectrokinetics’’). The examples treated here are (mostly taken from the previous work of the author) related to the production of hydrogen via the water-gas shift reaction (WGSR) and the selective reduction of NOx, which are both of current interest with respect to environmental issues. The most recent work put the emphasis on using so-called ‘‘operando’’ conditions, in which the spectroscopic data were recorded while the reaction rate was simultaneously measured over the catalyst using a single bed reactor. One of the aims of this report was to highlight some of the beneﬁts of combining kinetic and spectroscopic analyses, but also some of the shortcomings. The ﬁrst example discusses the seminal work of Tamaru and co-workers2 dating back to the 1960’s. The decomposition rate of surface formates were compared to the rate of CO2 production during WGSR obtained over the same sample in a diﬀerent experiment. The second example describes how a combination of kinetic and spectroscopic data (here recorded on two separate apparatuses) helped understanding, in parts, a very complex reaction, which is the selective catalytic reduction of NO with propene. The subtle role of NO oxidation on the reaction pathway of NO reduction over AgAl2O3-based catalysts is discussed. The diﬀerent mechanisms taking place depending on the Ag loading are highlighted. The formation of oxidised NOx(ads) species, important reaction intermediates, was evidenced by in situ DRIFTS and thermogravimetric measurements. The isocyanates observed by DRIFTS were also proposed as a crucial surface intermediate. The third and ﬁnal example documents the use of operando DRIFTS with steady-state isotopic kinetic analysis (SSITKA), using a mass spectrometer (MS) to follow mass transients, to investigate catalytic reactions. The water-gas shift (WGS) reaction over noble metals supported on ceriumcontaining oxides is presented in details. The DRIFTS-MS-SSITKA technique proved invaluable in determining the true role of formates ‘‘seen by a

Laboratoire Catalyse et Spectrochimie, ENSICAEN, Universite´ de Caen, CNRS, 6 Bd Mare´chal Juin, 14050, Caen, France

94 | Catalysis, 2010, 22, 94–118 c

The Royal Society of Chemistry 2010

IR’’, which turned out to be minor reaction intermediates (i.e. essentially spectators). Many advantages of coupling kinetic and spectroscopic measurements are discussed, e.g. cell validation as a kinetic reactor, role of the observed surface species, oxidation state of the working catalyst, spectator versus potential reaction intermediate.

2.

The bases of spectrokinetic analyses

Tamaru and co-workers reported investigative work in heterogeneous catalysis combining spectroscopic and kinetic data dating back to the 1960’s.2–5 The corresponding studies represented some of the ﬁrst attempts to relate the concentration and reactivity of surface species to the rate of the reaction measured over the same catalyst. The water-gas shift reaction, CO þ H2OCO2 þ H2, over base metal oxides was one of the reaction investigated. The type of experiments was highly challenging at the time, bearing in mind the technological limitation of the equipment (e.g. dispersive IR, since FTIR only become widespread much later)6 and supply of high purity gases (e.g. CO was sometimes obtained from the decomposition of sodium formate by sulphuric acid and using a liquid nitrogen trap).4 Custom-made cells were designed and used for the spectroscopic and kinetic measurements. It must be stressed that the system used was actually made of a dual cell, each of those containing a catalytic bed. The ﬁrst bed was used for the transmission IR analysis of a single wafer (e.g. with a mass of 300 mg) and the second bed contained a much larger mass of catalyst (e.g. 11 g) to ensure a measurable conversion.4 The utilisation of a dual bed implies a nonnegligible risk that each bed experienced diﬀerent experimental such as temperature and concentration gradients. The decomposition rate of surface formates were compared to the rate of CO2 production during the water-gas shift reaction (WGSR) obtained over a MgO sample at 280 1C. The WGSR was measured under a feed of CO and water in a recirculation mode. In a diﬀerent experiment, the rate of formate decomposition was obtained by following the decay of the formate bands for various initial surface coverages. The quantitative relation between IR band intensity and formate concentration was obtained via calibration curve realised using reference samples derived from adsorption of know amounts of formic acid on the catalyst. The values of WGSR rate and formate decomposition were suﬃciently similar (Table 1) so that the formates seen by IR could be conclusively proposed as the main reaction intermediate. Table 1 Comparison of the rate of formate decomposition to CO2 þ H2 over MgO at 280 1C and the corresponding WGSR rate. Both set of data were obtained for the same surface coverage of formate [reproduced from reference 4, copyright RSC] Formate fractional surface coverage

Rate of formate decomposition to CO2 þ H2 (mm3 g 1 h 1)

Rate of the water-gas shift reaction (mm3 g 1 h 1)

0.06 0.07 0.08

17 25 37

11 23 31

Catalysis, 2010, 22, 94–118 | 95

This type of study, based on a transient involving a change in the chemical potential of one or more of the chemical elements present, assumes that the reactivity of the surface species is the same under reaction condition and under concentration changes. This is clearly not always the case7 and will be discussed in more details in section four of the present report. A large number of spectrokinetic studies were also carried out in the 1970’s in the former USSR by Mathyshak and co-workers.8 A recent review gathering many examples of this work has been published in Catalysis Today.9 In essence, these authors varied many experimental parameters (in particular reactants concentrations) for the reaction of interest and measured the consumption rate of the reactant(s), formation rates of the product(s) and the surface coverage of adsorbates observable by IR. A microkinetic model was then developed and the experimental and simulated variations of rates and surface coverages were compared to ascertain the model.9 The procedure, leading to a possible reaction mechanism, appears to be experiment and time-intensive. Unfortunately, the methods used to carry out spectral decomposition are often unclear, while this point is often the bottle neck when complex spectra are considered. Other diﬃculties associated with the techniques regard the determination of molar absorption coeﬃcients and, sometimes, the use of chemical transients. In summary, the full microkinetic analysis combined with spectroscopy is an elegant method but clearly requires a signiﬁcant amount of work and still some assumptions. Many teams around the World have since used spectrokinetic methods derived from these seminal investigations, in particular by using isotopic transients (e.g. for studying syn-gas conversion,10–14 CO2-reforming of methane15,16 and nitrogen oxides decomposition,17,18) but it is not the scope of the present article to provide an exhaustive review of this ﬁeld. 3. Investigation of the selective reduction of NOx with propene over Ag/Al2O3 The example treated in this section regards a combination of kinetic, thermodynamic, spectroscopic and gravimetric data that was very useful in our investigation of the selective catalytic reduction of NOx with hydrocarbons, a highly complex reaction. The data corresponding to diﬀerent experimental techniques were recorded on separate apparatuses, each having its own reactor. Therefore, the experimental parameters (e.g. temperature proﬁle, gas ﬂow pattern, catalyst bed geometry and dead-volume, impurities) were possibly not strictly identical for each type of measurement. Nonetheless, the trends observed between the results obtained from the various techniques were consistent and worthwhile qualitative conclusions could be drawn. A commercial cell from Spectra-Techs was used to carry out the diﬀuse reﬂectance FTIR (DRIFTS) measurements. The bottom part of the catalyst bed, which was not probed by the IR beam, was possibly by-passed by the reaction mixture, due to the high pressure drop of the original reactor frit holding the sample.19 The catalytic data were obtained 96 | Catalysis, 2010, 22, 94–118

using a tubular quartz plug ﬂow reactor, as the amount of catalyst that could be placed in the DRIFTS cell (typically 20 mg) was not suﬃcient to induce a signiﬁcant conversion. The thermogravimetric data were recorded on an IGA microbalance. More experimental details can be found elsewhere.20,21 3.1 Selective catalytic reduction (SCR) over alumina and silver-based catalysts Many base oxides/metals (e.g., Al2O3, TiO2, ZrO2, and these oxides promoted by, e.g., Co, Ni, Cu, Fe, Sn, Ga, In, Ag) are active catalysts for the selective reduction of NOx (NO and NO2) with hydrocarbons (HC-SCR).22 Note that under typical lean-burn conditions, the promoting metals are almost exclusively in an oxidised state. Catalysts based on Ag supported on g-alumina received a particular attention, as these materials are among the most active and selective for this reaction. The activity of alumina and that of the same support promoted by a low (i.e. 1.2 wt.%) and high (i.e. 10 wt.%) loading of silver are shown in Fig. 1. The alumina was active and selective for the formation of N2 at higher temperatures (W400 1C), under these experimental conditions. It is interesting to note that some N2O and especially high concentrations of NO2 were observed over the alumina after complete propene conversion, i.e. above 565 1C. On the contrary, some NH3 could be observed but only before complete propene conversion. The 1.2% Ag/g-Al2O3 yielded similar conversions to N2 as those obtained over the alumina but at lower temperatures. Low concentrations of N2O, NO2 and NH3 were also obtained in this case. The activity of the high loading silver catalyst was signiﬁcantly diﬀerent from that of the g-Al2O3 and the 1.2% Ag/g-Al2O3. Complete combustion of the reductant was achieved at 350 1C over the 10% Ag/g-Al2O3, in contrast to the temperatures of 100% combustion of the reductant over 1.2% Ag/g-Al2O3 and g-Al2O3 of 500 1C and 565 1C, respectively. In addition, the N2 yield remained signiﬁcantly lower than that of N2O (obtained at the lower temperatures) and NO2 (obtained at the higher temperatures). Over the 10% Ag/g-Al2O3 and at the higher temperatures, the conversion to NO2 was limited by the thermodynamics of the reaction (see dotted line in Fig. 1, NO2 yield): NO þ 12O2 , NO2

ð1Þ

One of the striking features of the catalytic data reported in Fig. 1 was the sharp increase in the NO2 yield obtained over the alumina as soon as complete conversion of propene was achieved, at ca. 565 1C. The values of NO2 yield obtained at these temperatures were signiﬁcantly higher than that allowed by the thermodynamics of the reaction represented by equation (1). It has to be stressed that identical plots were obtained independently of using increasing or decreasing temperature proﬁles and no change in the yield of NO2 was observed after several hours on stream. A NO2 yield in slight excess of the thermodynamic limit was also observed over the 1.2% Ag/g-Al2O3. This surprising observation was not due to any quantiﬁcation Catalysis, 2010, 22, 94–118 | 97

100 NO conversion /%

C3H6 conversion /%

100 80 60 40 20 0 150

350

N2 Yield /%

NO2 Yield /%

10

350

20 350

550

40

20

0 150

550

350

550

350

550

30 N2O Yield /%

6 NH3 Yield /%

40

60

20

4

2

0 150

60

0 150

550

30

0 150

80

350 Temperature /°C

550

20

10

0 150

Temperature /°C

Fig. 1 C3H6-SCR of NO over g-Al2O3 (), 1.2% Ag/ g -Al2O3 ( ) and 10% Ag/g -Al2O3 ( ) catalysts as a function of temperature. Feed: 500 ppm NO þ 500 ppm C3H6 þ 2.5% O2 /He, W/F ¼ 0.06 g s cm 3 (GHSVB50 000 h 1). The dotted line in the plot giving the NO2 yield represents the thermodynamic limit associated with the reaction NO þ 1/2O23NO2. (Reprinted from reference 20.)

error, as indicated by the fact that the NO2 yield value measured in the case of the 10% Ag/g-Al2O3 was exactly equal to that expected by the thermodynamics (Fig. 1). Such high NO2(g)/NO(g) ratios largely exceeding the value expected from thermodynamics were also observed during the course of the SCR reaction over other excellent SCR catalysts based on cobalt/alumina.23 These observations clearly indicates that the main route to NO2(g) over these catalysts is not the direct oxidation of NO by O2 as described by the equation (1), contrary to what had been suggested by many authors.24 Hamada et al. had proposed that NO2 was an intermediate, based on the facts (i) that the reaction starting from this molecule was much faster than that starting from NO and (ii) that the activity measured in the presence of NO2 did not require the presence of the promoter. 98 | Catalysis, 2010, 22, 94–118

20 Thermal Equilibrium 10% Ag/Al2O3 15 NO2 Yield /%

Al2O3 1.2% Ag/Al2O3 10

5

0 150

250

350

450

550

650

Temperature /°C

Fig. 2 Conversion of NO to NO2 over g-Al2O3 ( ), 1.2% Ag/g-Al2O3 ( ) and 10% Ag/g-Al2O3 ( ) catalysts as a function of temperature. Feed: 0.05% NO þ 5% O2 in Ar, W/F ¼ 0.06 g s cm 3. (Reprinted from reference 20.)

3.2

Oxidation of NO(g) to NO2(g) and to NOx(ads)

The NO(g) to NO2(g) oxidation ability of our samples was measured and the best SCR catalysts (i.e. those leading to N2, alumina and 1.2% Ag-promoted alumina) displayed a very poor activity at the optimum temperatures for the SCR reaction (Fig. 2). The activity of the 1.2% Agalumina was essentially identical to that of the support, while the high loading sample (10 wt.% Ag) was very active for NO2 formation. IR and thermogravimetric data were also collected in order to unravel the role of the Ag promoter. The in situ DRIFTS spectra reported in Fig. 3 qualitatively describe the growth of various surface nitrate species (bands at ca. 1560, 1305 and 1255 cm 1) when the samples were exposed to a NO þ O2 stream. Both silver-promoted materials appeared to exhibit a faster uptake rate of NOx(ads) than that observed over the plain alumina. This trend was conﬁrmed quantitatively using the mass uptake measured with the microbalance (Fig. 3). It is clear that at least one of the main role of the 1.2% silver promoter was to favour the oxidation of NO(g) to strongly bound NOx(ads) species (and not to NO2(g), as shown above). 3.3

Proposed reaction mechanisms of the SCR over Ag/Al2O3

The marked diﬀerences between the activity of, on the one hand, the 10 wt.% Ag sample and, on the other hand, the 1.2% Ag and plain alumina (see Figs. 1, 2 and 3) suggests that diﬀerent reaction mechanisms may occur on these catalysts. A NO decomposition-type mechanism was suggested to occur on the high loading material, which probably consisted of large metallic silver particles deposited on the alumina (Fig. 4). To this respect, the activity pattern of this sample resembles that of platinum group metals. In the case of the low loading Ag/Al2O3, dispersed Ag þ cations and/or small Catalysis, 2010, 22, 94–118 | 99

Absorbance

0.3 a 1550

1700

1600

1235

1500

1400 1300

1200

1550 1305

b

1255

1700

1600

1500

1400 1300

1560

Relative weight uptake / 10−3

6

1200

1300

c

10% Ag/Al2O3 4

1.2% Ag/Al2O3

2 Al2O3

0 1700

1600

1500

1400 1300

1200

0

Wavenumber /cm-1

1 2 3 4 Time under the NO/O2/He stream /min

5

Fig. 3 Right: in situ DRIFTS analysis of the formation of ad-NOx species at 4001C over (a) g-Al2O3, (b) 1.2% Ag/g-Al2O3 and (c) 10% Ag/g-Al2O3. For each catalyst, time on stream was 15 min (lower spectrum), 60 min (middle spectrum) and 180 min (upper spectrum). Left. Thermogravimetric analysis of the formation of ad-NOx species at 400 1C over g-Al2O3 (x), 1.2% Ag/g-Al2O3 ( ) and 10% Ag/g-Al2O3 ( ) as a function of time. Feed: 0.05% NO þ 2.5%O2/He. (Reprinted from reference 20.)

NO / O2

N2O + N2

C3H6

NO + O2 C3H6

H2O COx

O2

Ag °

N2

NO2

O O O N N O CxHy NOxAg+

R-ONO R-NO2

R-NCO

R-NH2

NH3

alumina Fig. 4 Schematic representation of the two main reaction pathways taking place on Ag/g-Al2O3. (Reprinted from reference 20.)

100 | Catalysis, 2010, 22, 94–118

electropositive silver clusters in strong interaction with the alumina prevailed. The exact nature of the reaction mechanism(s) occurring over alumina and low loading Ag/Al2O3 is not yet fully understood, but N2 is probably formed via a series of parallel and consecutive reactions involving numerous intermediates over both the silver and alumina phases. Oxidised species of nitrogen (e.g., inorganic nitrates) are thought to react with reduced forms of this element (e.g., isocyanate, ammonia, formed via organo-nitrogen compounds) to produce N2 (Fig. 4). The typical volcanoshape of the N2 yields plots is ascribed to the competitive reactions between NO and O2 for the reductant. The unselective combustion of the reducing agent with O2 becomes much faster than the SCR at higher temperatures and diminishes the number of reductant molecules available for the SCR. The temperature of maximum N2 yield often corresponds to ca. 90% conversion of the reductant. The reaction scheme proposed for the low loading sample is supported by the evidence described thereafter. The formation of organonitrogen compounds was observed over some SCR catalysts.25 The formation and decomposition of such compounds would rationalise the high concentrations of NO2 observed. The example given below equations (2–4) is arbitrarily based on nitromethane, simply because thermodynamic data are readily available for this molecule. The organonitrogen compound would be formed along with CO (or CO2) from the reaction of propene, O2 and NO equation (2). The subsequent oxidation of this organo-NOx species would yield NO2 equation (3): C3 H6 þ 32O2 þ 2NO ! 2CH3 NO2 þ CO

ð2Þ

2CH3 NO2 þ 72O2 ! 2NO2 þ 2CO2 þ 3H2 O

ð3Þ

Overall, the equations above combine to give: C3 H6 þ 5O2 þ 2NO ! 2NO2 þ 2CO2 þ CO þ 3H2 O

ð4Þ

While it is likely that the actual mechanism occurring on the low loading Ag materials involves more than one organonitrogen species according to a more complex reaction scheme, these equations are important in the sense that the reactions are strongly exergonic over the temperature range investigated here. The standard Gibbs free energy of reaction at 813 K associated with eqn. 4.3.2 and 4.3.3, are DrG1 ¼ 298 kJ mol 1 and 734 kJ mol 1, respectively. The global reaction equation (4) is therefore also strongly favoured and provides a rational explanation for the high NO2/NO ratio observed during the experiments (Fig. 1). 3.4

Reactivity of organonitrogen species over Ag/Al2O3

Organonitrogen species can be readily formed non-catalytically by reaction of hydrocarbon, dioxygen and nitric oxide in the liquid or gas phase.26 The decomposition products of organo-nitrogen species yield similar products to those observed during the SCR reactions (e.g., cyanide, isocyanates), supporting their role as intermediates. NH3 can be obtained from nitromethane through the tautomerisation to the corresponding oxime followed by dehydration to a nitrile N-oxide equation (5) which isomerise to an Catalysis, 2010, 22, 94–118 | 101

1554 Abs 0.1

1306 1258

(b)

1598 2228

1393 1376 2902 3390

2094

3003

(a)

4000

1447

3600

3200

2800

2400

2000

1600

1200

ν / cm−1 Fig. 5 DRIFT spectra at 573 K in argon of Al2O3 following pre-adsorption of (a) nitromethane and (b) tert-butyl nitrite at room temperature. (Reprinted from reference 28, RSC.)

isocyanate before yielding a primary amine and NH3 by hydrolysis equation (6, H2O/CO2 are not reported). Over alumina, the possibility of forming NH3 from reaction of organo-nitrile N-oxides species was conﬁrmed;27 the organonitrile N-oxide were formed from organo-nitroso compounds, via enol and cyanide formation (eqn. 7, only the N-containing fragment is shown). CH2 NO2 ! CHQNOðOHÞ ! CQNQO þ H2 O

ð5Þ

CQNQO ! NQCQO ! NH2 ! NH3

ð6Þ

CH2 NO ! CHQNðOHÞ ! CRN ! CQNQO

ð7Þ

We have shown that the nature of the organonitrogen compounds greatly inﬂuenced the nature of the products formed, as two main reactivity 102 | Catalysis, 2010, 22, 94–118

patterns were observed during the oxidation over the 1.2% Ag/alumina.28,29 Nitro-type molecules led to the formation of reduced nitrogen products, including ammonia and hydrogen cyanide, with isocyanates being a main surface species (Fig. 5). In contrast, nitrito-type compounds led to NO2 and surface nitrates and nitrites. The latter type of compounds is therefore an obvious candidate for the source of the NO2 formed under SCR conditions. 3.5

Proposed SCR reaction mechanisms over oxide-based catalysts

The global reaction scheme proposed in the case of our 1.2% Ag/Al2O3 can probably be extended to other oxide-based SCR catalysts (Fig. 6).22,23 The role of dioxygen is intricate and paradoxal, as this oxidiser strongly favours the reduction of NO over most catalytic formulations. Most authors acknowledge that the two main functions of O2 are the oxidation of NO and of the reductant to form the various reaction intermediates. As far as the hydrocarbon is concerned, one of the important initial steps proposed is its oxidation to strongly bound oxidised species such as acetates. The acetate species or other adsorbed oxidised hydrocarbon species (e.g. acrylates)30 are then believed to react with the surface nitrates (or possibly with gas-phase NOx) to yield organo-nitrogen species, the exact nature of these species remaining unclear. The formation of the organonitrogen species is likely to be the rate-determining step of the reaction, as

NO (g) + O2 (g) + CxHy (g) adNOx Inorganic NOx (ads) (several species, in particular, monodentate nitrate)

Organo-nitrogen, e.g., R-NO2(g or ads)

R-CN R-NCO R-NH2 NH3

CxHyOz (ads) (several species, in particular acetate)

R-ONO (g or ads)

Minor route NO2 (g)

N2 (g) Fig. 6 Schematic representation of the two main reaction pathways taking place on oxidebased catalysts for the selective reduction of NO with hydrocarbons. (Reprinted from reference 23.)

Catalysis, 2010, 22, 94–118 | 103

these can only be observed during transient experiments such as temperature-programmed surface reaction monitored by in situ IR. Like some other researchers, we have therefore proposed that the coupling of nitrogen atoms to form N2 could occur via the reaction between the oxidised (e.g., NO(g), nitrate) and reduced (e.g, –NCO, NH3) forms of nitrogen. This observation stresses that the reaction mechanism is very complex since NO will react through a series of parallel pathways to form numerous intermediates. The relevance and rate of each step of the scheme represented in Fig. 6 depends on the nature of the reductant, the catalyst and experimental conditions. The overall rate-determining step and the surface concentrations of each species will vary accordingly. For instance, the rate at which acetates formed was shown to be dependent on the chain length of the alkane.31 As a result, the relative surface coverage of acetates and nitrate species can vary as a function of type of feed, catalyst and experimental conditions used. In the case of the C3H6–SCR of NO over Al2O3, the rate of nitrate formation is slower than their rate of consumption and as a result acetates predominate (Fig. 7, bottom spectrum). When Ag is added to the Al2O3, the oxidation of NO to ad-NOx species is promoted and surface nitrate species now predominate (Fig. 7, upper spectrum). Other experimental parameters such as temperature, water vapour pressure will aﬀect the weight of the reactions as reported in Fig. 6. The possible participation of homogeneous reactions must also be considered, especially at the higher temperatures and bearing in mind that NO and NO2 are radicals.32 Yet, it appears the majority of data reported on oxides/base metals are consistent with this scheme.

1555 NOx (ads)

Absorbance

1460

a.u. 1% Ag /γ-Al2O3 1600

H γ-Al2O3

2910 3005

3800

3400

3000

1400

C CO2

2600

2200

O

O

1800

1460

1380

1400

Wavenumbers /cm−1 Fig. 7 In situ DRIFTS spectra of the surface species formed over Al2O3 and 1.2% Ag/Al2O3 during the SCR of NO with propene. Feed: 500 ppm NO þ 500 ppm C3H6 þ 2.5% O2. (Reprinted from reference 20.)

104 | Catalysis, 2010, 22, 94–118

3.6 Conclusions on the study of HC-SCR of NO by in situ IR and combined kinetics The work reported in this section shows that a combination of kinetic, thermodynamic, spectroscopic and gravimetric data proved useful in determining some of the main aspects of the hydrocarbon selective catalytic reduction of NO over oxide-based catalysts. One of the role of the Ag promoter at low loadings was to favour the formation of surface nitrate/ nitrite species, but not directly NO2(g). The formation of NO2(g) occurred via a complex reaction scheme involving the oxidation of organonitrogen species, probably of the nitrito-type. A second major outcome of our work was to evidence the occurrence of another distinct reactions mechanism at higher silver loadings, similar to that taking place of platinum, based on a decomposition-type mechanism. While no unambiguous conclusion could be given on the true nature as reaction intermediates of the surface species observed by in situ IR, the observed species nonetheless helped building a realistic model of this complex reaction pathway. 4.

Spectrokinetic operando investigation of catalytic reactions

As described in the previous section, the collection and comparison of kinetic and spectroscopic data can be useful in gaining some understanding of the mechanism of a catalytic reaction. However, data pertaining to different techniques are usually collected on separate apparatuses, each having its own reactor. The simultaneous collection of various spectroscopic data in a single reactor is currently receiving much attention as a means to overcome the possibility of diﬀerences in the actual experimental conditions prevailing in separate reactors.33,34 In order to identify more focussed analytical techniques a new expression, i.e. ‘‘operando’’, was put forward. The term ‘‘operando spectroscopy’’ refers to spectroscopic measurements of catalysts under working conditions with simultaneous on-line product analysis. This term was used in the literature starting from 200235,36 with the aim to distinguish work in which on-line activity measurement was performed alongside spectroscopic measurements (i.e. operando) from those in which only spectroscopic data were recorded (i.e. in situ). The on-line analysis of the reactor eﬄuent is useful in many ways. Firstly, it allows collecting kinetic data that are directly related to the spectroscopic data simultaneously measured, which is particularly useful when carrying out isotopic transients. Secondly, it ensures that the activity data obtained in the operando reactor are consistent with those observed in a conventional ‘‘ideal’’ reactor. These conditions required the (often challenging) development of reaction cells operating in a kinetically relevant mode, able to withstand extreme conditions, while still allowing the electromagnetic radiation to escape from the reactor. In spite of the technical diﬃculties, the success of operando techniques is such that the number of teams switching their studies towards this methodology is increasing.37 The investigations reported in this section will describe how a DRIFTS cell can be checked for kinetic relevance and the modiﬁcation that can be made to correct any ﬂaw. Quantitative aspects of DRIFTS work will also be Catalysis, 2010, 22, 94–118 | 105

addressed. Isotopic transient techniques (i.e. SSITKA) will be used to determine the actual role of surface species seen by DRIFTS. The water-gas shift reaction will be investigated in details and some data related to CO hydrogenation will also be presented. 4.1

Development of a kinetically relevant DRIFTS cell reactor

Diﬀuse reﬂectance FT-IR spectroscopy (DRIFTS38,39) is increasingly being used as a means to investigate the reactivity of surface species under reaction conditions, but it is usually considered only as a qualitative technique. However, it was demonstrated that DRIFTS spectroscopy can be an accurate quantitative tool for operando studies, providing that an appropriate analytical transformation of the diﬀused intensity is used (i.e. in most cases the pseudo-absorbance rather than the Kubelka-Munk function40) and that a calibration curve relating band intensity to adsorbate concentration is available.41 We also showed that an appropriately modiﬁed DRIFTS cell reactor19 led to reaction rates identical to those measured in a linear quartz tube plug ﬂow reactor.41 DRIFTS reactors are particularly suited to operando investigations since the catalyst powder can be used as such, whereas FTIR-transmission techniques require pressing wafers, which can lead to mass-transport limitations and catalyst modiﬁcations. The environmental DRIFTS chambers proposed by Spectra-Tech have been widely used to carry out in situ and operando analyses. It has been known for many years42 that these cells present some bypass of the catalyst bed. As a result, a large proportion of the incoming gas directly goes to the outlet, without passing through the catalyst bed located on top of the ceramic crucible. This problem is made worse by the fact that the porous frit used to support the catalyst presents a very high pressure drop. More, in the case of the high-pressure cell, the ZnSe dome can be very close to the rim of the crucible limiting its accessibility. As a result, the volume delimited by the catalyst bed and the void above it can become a dead-zone, with very slow diﬀusion pathway to the circulating gas passing through the cell. The extent of bed bypass can be assessed (i) by comparing the catalytic activity measured in the DRIFTS reactor to that measured in a traditional tubular reactor and (ii) measuring the time needed to purge the bed area, monitoring an IR-sensitive tracer.19 The DRIFTS cell that was used throughout this section was the hightemperature model, in which ﬂat ZnSe windows are several millimeters apart from the crucible, therefore not limiting the access to the catalyst bed. The bed bypass when using a ﬂowrate of 100 ml min 1, which was the ﬂowrate value typically used during our operando experiments, was estimated by measuring the extent of CO conversion during oxidation with an excess of O2 over a Pt-based catalyst. A complete conversion of CO to CO2 would be expected above the light-oﬀ temperature, when no bypass is taking place. The oxidation data showed that almost 80 % of the feed bypassed the catalyst bed in the case of the original cell, as the CO conversion leveled oﬀ at ca. 20% after the light-oﬀ.19 The cell was modiﬁed by replacing the original crucible with a custommade ceramic reactor. The crucible presented no signiﬁcant pressure drop as 106 | Catalysis, 2010, 22, 94–118

a metallic mesh (mesh size 45 micron) was used to support the catalyst instead of the ceramic porous frit of the original cell. The catalyst powder was sieved using the same mesh as that located in the crucible and only the fraction that did not cross over the mesh was used for the DRIFTS analysis. No study was carried out to investigate the eﬀect of the particle size on the intensity of the IR signal obtained. The gap between the crucible stem and the metallic base was sealed by wrapping the lower part of the crucible stem with Teﬂon-tape before inserting into the base. The Teﬂon tape was in close contact with the thermostated cell metal base, the temperature of which always remained moderate (typically less that 120 1C even when the crucible itself was heated up to 400 1C). Therefore, the PTFE tape did not show any signiﬁcant level of degradation under standard conditions of use. A level of CO combustion higher than 98.5% was achieved above the temperature light-oﬀ on the modiﬁed system, indicating that the bed bypass was negligible. A further validation of the kinetic relevance of our modiﬁed DRIFTS cell come from the fact that the catalytic activity measured during the WGS reaction using the modiﬁed cell was shown to be equal to that measured in a conventional tubular plug-ﬂow reactor.41 Note that no conversion of reactant was observed when the crucible was heated up to the reaction temperature in the absence of the catalyst. The DRIFTS and mass spectrometry (MS) data collected during our DRIFTS-MS experiments actually relate to diﬀerent regions of the reactor. The DRIFTS data are collected at the front (and top) part of the bed, while the MS data are collected at the exit of the cell, i.e. after the bed (Fig. 8a). If the evolution of the MS signals of reactions products during an isotopic switch is to be compared to the evolution of the DRIFTS intensities of surface intermediates, then the gas phase all along the catalyst bed should be CO + H2O

ca. 0.2 mm 2 mm

H2 + CO2 adsorbates

(b)

DRIFTS signal

Pt /CeO2

CO2 + H2

MS signal

Relative MS or DRIFTS intensity

(a)

1 Kr

0.8

13CO

0.6

(IR) 2

x (MS)

0.4 0.2 0 0

100

200

300

400

500

Time after isotopic switch /s

Fig. 8 (a) Schematic representation of the catalyst bed and the zone probed by the DRIFTS beam. The MS data are collected at the exit of the DRIFTS cell, i.e. after the bed. (b) Relative intensity of the Kr tracer (solid line), 13CO2 measured by mass spectrometry () and 13CO2 measured by the IR signal of the DRIFTS cell (&) following a 12CO13CO isotopic switch under steady-state WGS conditions over a 2% Pt/CeO2. T ¼ 473 K, feed: 1% 13CO þ 10% H2O in 2% Kr/Ar. The feed was 1% 12CO þ 10% H2O in Ar before the switch. (Reprinted from reference 19.)

Catalysis, 2010, 22, 94–118 | 107

homogeneous (to avoid chromatographic eﬀects). This is to ensure that the variation of the gas-phase composition at the front of the bed, which directly relates to the surface species observed by the DRIFTS, is equivalent to that measured by MS after the bed. It is therefore important to use diﬀerential conditions throughout the bed by keeping a conversion lower than ca. 15%. In this way, the concentration of surface species should be essentially constant throughout the catalyst bed. The fact that only a fraction of the bed volume is analysed by DRIFTS still allow doing quantitative analysis knowing the full bed mass, providing calibration curves are available based on the very same sample impregnated by known concentrations of adsorbates (see Section 4.5). The DRIFTS signal of the surface species of interest measured under reaction condition and that measured over the calibrated samples are related to a same analysis volume (or mass) and knowing the concentration (in mol/g) in the sample allows deriving the full amount of sorbate present (in mol) using the actual sample mass. It is clear that using a diﬀerent sample for calibration purposes (e.g. with a diﬀerent particle size, surface area or composition) would lead to a diﬀerent calibration curve, as shown in reference 45, and does not allow quantitative insights. Bed packing must also be carried out in a reproducible manner, which is usually the case when using the same operator. In an example treated here based on water-gas-shift data, we are fortunate that the reaction product CO2 can be observed both by using the DRIFTS signal and by the MS. Note that the bands measured by DRIFTS are related to the surface species present in the upper volume of the catalyst bed (typically at less than 200 microns depth for a total bed length of ca. 2 mm), while the signal of CO2(g) is both coming from the same volume and the free gas volume above the bed. An isotopic exchange of CO2 was followed during a SSITKA-DRIFTSMS experiment over the Pt/CeO2 catalyst (see subsequent sections for the description of the method). The CO conversion was 10% in these conditions, ensuring diﬀerential conditions. The evolution of the 13CO2 signals following the switch to the 13CO-containing feed is given in Fig. 8b. The MS signal of the Kr tracer is also reported for the sake of completeness. The Kr proﬁle (Fig. 8b) (and that of the reactant CO, not shown here) was essentially a step-function in comparison with the CO2 signals, indicating that the variation of the 13CO2 concentration was not limited by mass transport (i.e. by the supply of labeled gas). Note that this is not the case in the work reported by Jacobs and Davis,43 in which the isotopic exchange of surface and gaseous compounds is limited by a very slow gradual introduction of the labeled gas (as the switching valve is located before the CO mass ﬂow controller). The disadvantage of the technique used by Jacobs and Davis is that all processes with a time constant lower than that of the supply of the labeled compounds (several minutes) will not be resolved, while the time constant of our method with this modiﬁed cell is ca. 7 s. It is clear that the relative variations of the DRIFTS and MS signals associated with 13CO2 (Fig. 8b) were identical, displaying an almost exponential increase with a 50%-exchange time of about 55 s. These data unambiguously show that the signal of the reaction product CO2 measured 108 | Catalysis, 2010, 22, 94–118

by the MS after the catalyst bed perfectly corresponded to that measured at the bed entrance, which was measured using the DRIFTS signal. Therefore, the gas-phase proﬁle in the thin (ca. 2 mm) catalyst bed was homogeneous, in the reaction conditions used here. This fact justiﬁes the comparison of the curves obtained by DRIFTS for the surface species at the top of the bed and the curves obtained for the products of reaction by MS (or any other analytical techniques) after the reactor, at least as long as diﬀerential conditions apply. 4.2

Quantitative DRIFT analysis of adsorbates concentrations

There is a widespread misconception that the IR signal measured in the diﬀuse reﬂectance mode of species adsorbed at surfaces must always be reported as Kubelka-Munk units, which are given by the equation below (eqn. 8), in which RN ¼ I/Io is the reﬂectance of the sample, that is the ratio of the intensity diﬀused (noted I) to that incident (noted Io): FðR1 Þ ¼

K ð1 R1 Þ2 ¼ 2R1 S

ð8Þ

In fact, a diﬀerent expression (eqn. 9) has been proposed to account for the concentration of the adsorbates,44 in which R is the reﬂectance measured in the presence of the adsorbate: Mathyshak KrylovðRÞ ¼ ½c ¼ a

ðR1 RÞ 1 R1 R1 R

ð9Þ

Based on these equations, we have shown40 that the relative absorbance, i.e. ¼ log R 0 with R 0 ¼ relative reﬂectance ¼ R/RN, is a more linear function of the surface concentration than the Kubelka-Munk function in the range of relative reﬂectance R 0 comprised between 100 and 60%. This range of relative reﬂectance is pertinent to most DRIFT studies of surface species, for which the signal loss due to the absorption of the adsorbed species is weak. The use of the absorbance function also overcomes the problem associated with baseline drifts during measurements.40 It is only in the case of low values of reﬂectance that using the Kubelka-Munk transform may be more appropriate. Further evidence that units of absorbance are appropriate when investigating surface species by DRIFTS is given by the data reported in Fig. 9, which relate to the decomposition of 12C-containing surface formates species during the water-gas shift reaction over a ceria-based catalyst (vide infra). The intensity of the formate bands was expressed in absorbance units. An almost perfect exponential decay was observed (Fig. 9a), as expected in the case of ﬁrst order processes, since the corresponding semilogarithmic plot yielded a straight line (Fig. 9b). As a consequence, the absorbance units were always used during our investigations whenever any quantitative work was carried out. Another a common misconception is that DRIFTS work cannot be quantitative, contrary to the case of transmission IR data. On the contrary, we were able to draw calibration curves to accurately quantify the concentration of formates and use these curves to quantify the concentration (in mol/g) during our operando work.41,45 Catalysis, 2010, 22, 94–118 | 109

(b) Time after isotopic switch /s

1 0.8 0.6 0.4 0.2 0 0

500

1000

1500

Time after isotopic switch /s

Log10 of the proportion of 12Ccontaining surface formates

Proportion of 12C-containing surface formates

(a)

0

0

500

1000

1500

-1

-2

Fig. 9 (a) Relative DRIFTS signals measured in absorbance units associated with the 12Ccontaining formate species following a 12CO–13CO isotopic switch under steady-state WGS conditions over a 2% Pt/CeO2. T ¼ 473 K, feed: 1% 13CO þ 10% H2O in 2% Kr/Ar. The sample was initially at steady-state under 1% 12CO þ 10% H2O in Ar. (b) Logarithmic plot of the data reported in (a). The dotted line is a straight line used as an eye guide. (Reprinted from reference 19.)

4.3

The operando SSITKA-DRIFTS-MS method

Spectroscopic studies are more powerful when combined with isotopic transient methods (SSITKA46,47), which allow operating at the chemical steady-state. The operando DRIFTS-SSITKA method described here relies on using a single catalytic bed, which allows the characterisation by DRIFT spectroscopy of the surface of the very same catalyst particles that are responsible for the catalytic activity measured at the exit of the cell by gaschromatography or mass-spectrometry.48 The group at Queen’s University Belfast was the ﬁrst to report data coupling in situ DRIFTS with activity measurement using a mass spectrometer (MS) during a steady-state isotopic exchange kinetic analysis (SSITKA) using a single bed reactor. This methodology is similar to that developed earlier for transmission FTIR by Chuang et al.14,49 Note that mass transport limitations may occur with wafers and that the temperature control can also be diﬃcult (as is for DRIFTS cells at high temperatures). These techniques derived from the socalled ‘‘isotopic jump’’ technique of Tamaru et al.,5 which relied on a twobed IR cell (see Section 2 of this report). The principle of the DRIFTS-MS-SSITKA method that was developed for the operando investigation of catalytic reactions is schematically represented in Fig. 10. The use of SSITKA method allows us assessing the chemical reactivity of surface species with respect to the formation of a reaction product under chemical steady-state conditions. This technique consists in replacing one of the reactants (e.g. 12CO) by an isotopomer (e.g. 13 CO) during reaction and following simultaneously the exchange of the labelled reaction product (here 13CO2) by mass-spectrometry and the surface species (e.g. 13 or 12C-containing carbonyl and formates) by DRIFTS. The DRIFTS bands of the surface species typically shift in wavenumbers during the analysis, and various integration methods can be used to quantify accurately the band heights or areas.19 A surface species observable by IR can only be a main reaction intermediate if it exchanges at least as fast as the reaction product in the 110 | Catalysis, 2010, 22, 94–118

Fig. 10 Schematical representation of the DRIFTS-MS-SSITKA technique for the operando investigation of catalytic reactions. This technique is based on the utilisation of a single reactor (i.e. the DRIFTS reactor) and allows following the kinetic of exchange of both gas phase products P (by MS) and surface intermediates (by DRIFT) during a SSITKA-type experiment. The surface species noted I represent a true reaction intermediate, while the surface species S is a ‘‘spectator’’ (better called a minor intermediate). I* and S* are the corresponding labeled surface species.

gas-phase (e.g., species I in Fig. 10). A signiﬁcantly slower exchange or no exchange at all indicates a minor reaction intermediate and a spectator species, respectively (e.g., species S in Fig. 10). Using a single bed and differential conditions ensures that any observed variations of surface species concentrations can be related to that of products in the gas-phase. We have reported the ﬁrst use of this setup in an investigation of the reverse watergas shift (RWGS) reaction:48 RWGS : CO2 þ H2 ) CO þ H2 O

ð10Þ

The importance of using chemical steady-state conditions for the study of the reverse water-gas shift reaction was highlighted during our investigation comparing the rate of removal of surface species during a purge in an inert gas and during an isotope exchange.50,51 Our data clearly showed that the reactivity of surface species, in particular that of carbonate species, was markedly diﬀerent under these two gas streams (Fig. 11). We believe that the diﬀerence of reactivity observed was related to the diﬀerence in the oxidising/reducing nature of the feed, which altered ceria surface oxidation state, which in turn modiﬁed the strength of the adsorbate bonding to the ceria surface.52 Another example of the importance of using chemical steady-state was recently reported by Mims and co-workers in the case of methanol synthesis over Cu/SiO2 catalysts.53 The authors measured the rate of decomposition of the formates formed at the surface of Cu supported on silica during methanol synthesis from CO/H2 mixtures. The formate decomposition rate Catalysis, 2010, 22, 94–118 | 111

Relative DRIFTS intensity

1 0.8 0.6 0.4 0.2 0 0

5

10

15

20 Time /min

25

30

35

40

73

Fig. 11 Relative intensity of the IR bands of the formate ( , ), carbonyl (’,&) and carbonate ( , ) species as a function of time on stream under Ar (solid symbols) and under RWGS stream containing 13CO2 (open symbols). The sample was at steady-state state in 1% 12 CO2 þ 4% H2 and T ¼ 498 K before switching to either Ar or the 13CO2-containing feed. (Reprinted from reference 50.)

was up to 3-fold lower in the case of a CO-free feed, as compared to the full CO/H2 mixture. Note that in this particular example,53 formates ‘‘seen by IR’’ were conclusively shown to be the main reaction intermediate. The lower reactivity observed under not chemical steady-state was assigned to the loss of co-adsorbates eﬀects. These two examples strongly emphasise that not using chemical steady-state conditions may lead to ﬂawed kinetic investigations.

4.4

Operando SSITKA-DRIFTS-MS study of the RWGS

An example of the application of our DRIFTS-MS-SSITKA technique is given below for the RWGS reaction.48 Fig. 12 shows the typical DRIFT spectra of formate, carbonyl and carbonate species during the isotope exchange 12CO2 to 13CO2 of the reactant. The replacement of 12C with a heavier isotope led to a red-shift of most of the wavenumber of the vibration of the surface species of interest. Note that we were able to integrate in an unequivocal manner the concentrations of the formate, carbonyl and carbonate species. The corresponding exchange data of the surface species are reported in Fig. 13, which showed that carbonates and carbonyl species were typically half-exchanged in about 60 s. On the contrary the exchange time of formate species was much longer, i.e. about 11 min. The corresponding MS data showing the exchange of the gas-phase product CO is also shown in Fig. 13. It is clear that CO was exchanged at a timescale similar to that of the carbonyl and carbonates, and therefore these species are potentially main reaction intermediates, contrary to the formates. Formates should be named as minor intermediates, rather than spectator species, since those still exchanges and probably led to some CO, albeit at a much lower rate than those formed via the other surface intermediates. 112 | Catalysis, 2010, 22, 94–118

2841 cm-1 0.05 log 1/R 2947 cm-1 a b c d e f 3000

2916 cm-1

2825 cm-1

2900

2800

2700

Wavenumber / cm-1 2057 cm-1 0.1 log 1/R 1977 cm-1 a b c d e f

2010 cm-1 2100

2000

1904 cm-1 1900

Wavenumber / cm

0.1 log 1/R

1800

-1

851 cm-1

866 cm-1

a b c d e f

831 cm-1 862 cm-1

900

880

860

840

820

800

Wavenumber / cm-1 Fig. 12 Typical DRIFTS data of a DRIFTS-MS-SSITKA experiment during the RWGS over Pt-CeO2: exchange of (top) formate (middle) carbonyl and (bottom) carbonate species at various times after an isotope exchange 12CO2 to 13CO2. Feed: 1% CO2 þ 4% H2. (Reprinted from reference 41.)

4.5

Operando SSITKA-DRIFTS-MS study of the WGS

An example of DRIFTS-SSITKA-MS data relating to the water-gas shift (WGS, 11) reaction over a 2% Pt-CeO2 is given in Fig. 14.54 WGS : CO þ H2 O ) CO2 þ H2

ð11Þ

The comparison of the exchange curves of the reaction product (here CO2) and formate was more intriguing that in the case discussed above. The Catalysis, 2010, 22, 94–118 | 113

Relative IR / MS intensity (a.u.)

1 Formate 0.8 Carbonyl 0.6

Carbonate 13CO

0.4 0.2 0 0

5

10

15

20

25

30

Time (min) Fig. 13 Comparison of DRIFTS and MS data during a SSITKA experiment relating to the RWGS over Pt-CeO2: the carbonate and carbonyl species are exchanged at a time scale similar to the reaction product CO and are potentially reaction intermediates. On the contrary, formates are clearly not main kinetic intermediates. (Reprinted from reference 48.)

Relative IR and MS signals

CO2, 220 °C

CO2, 160 °C

1 0.8 0.6 0.4

Formate, 160 °C

0.2

Formate, 220 °C

0 0

5

10 Time / min

15

20

Fig. 14 Comparison of the relative exchange of the gas-phase CO and CO2 and surface formate species during an isotopic exchange over the Pt/CeO2 at various temperatures. Feed: 2% 13 CO, 7% H2O in Ar. The sample was initially at state-state under the corresponding nonlabeled feed. (Reprinted from reference 54.)

formate exchange was signiﬁcantly slower than that of CO2 at 160 1C, suggesting that formates were unimportant reaction intermediates at these temperatures (Fig. 14). However, the exchange of these two species was essentially identical at 220 1C, suggesting that formates could potentially be a main reaction intermediate under these conditions. The relevance of the formates seen by DRIFTS in the formation of CO2 was ascertained by a quantitative comparison of the speciﬁc rate of CO2 formation (measured by GC analysis of the DRIFTS cell eﬄuents, which is more precise than that obtained by MS) and the speciﬁc rate of formate decomposition. The latter was calculated as the product of the formate 114 | Catalysis, 2010, 22, 94–118

concentration by the pseudo-ﬁrst-order rate constant (noted k) of formate decomposition:45 Rate of formate decomposition ¼ k ½formate 100% The value of k was determined from the formate exchange curve following the isotopic switch, which showed a ﬁrst order decay (i.e. exponential curve) (see Fig. 14). The formate concentration was determined under reaction conditions via the calibration curves described elsewhere.41,45 The term 100% is introduced to indicate that we arbitrarily assumed that the formate totally decomposed to CO2 þ H2 and not at all to CO þ H2O. Therefore, the decomposition rate of the formate species yielded the upper limit of the rate of formate decomposition to CO2, since it is likely that formates were not only decomposing to give CO2, but instead gave a mixture of CO and CO2. Even assuming that formates decomposed solely to CO2, the rate of formate decomposition for highly active Pt/CeO2 WGS catalysts accounted for less than 10% of the total rate of CO2 formation in the range of temperatures and experimental conditions used here, respectively (Fig. 15). Therefore, the formates seen by DRIFTS are minor reaction intermediates over this very active WGS catalyst. These observations stresses that the similarity of the time constant of formates and CO2 is not a suﬃcient condition to guarantee that formates seen by DRIFTS are main reaction intermediates. In some cases, formates seen by IR have been also considered essentially as ‘‘buﬀer’’ surface species;55 whether those participate or not to the main reaction pathway is unclear. In any case, while our work cannot indicate what the main WGS reaction pathway is, it certainly stresses that observing formates displaying some sort of reactivity is not a suﬃcient criterion to elect those as important reaction intermediates as proposed in many instances.56,57 Of course, our conclusions are based on the samples that we investigated under our experimental conditions; therefore any extrapolation outside this system is hazardous. DFT-based work carried out both on ceria-supported noble metals and copper seems to favour the role of

Rate /10-6 mol s-1g-1

12 10 8 6

CO2 formation Formate decomposition

4 2 0 160 °C

180 °C

220 °C

Reaction temperature Fig. 15 Rate of CO2 production and rate of formate decomposition over the Pt/CeO2 at three diﬀerent temperatures under 2% CO þ 7% H2O in Ar. (Reprinted from reference 45).

Catalysis, 2010, 22, 94–118 | 115

carboxyl species (HO–CO) as main reaction intermediate and exclude any signiﬁcant role of formates.58,59 The surface concentration of the carboxyl intermediates was estimated to be more than a thousand-fold lower than that of the ‘‘buﬀer’’ formates, suggesting that a direct observation of these short-lived species may be impossible by the current IR investigation techniques.58

4.6

Conclusions on the spectrokinetic investigation of catalytic reactions

A reliable study of a catalytic system using spectroscopic tools must to take into account the following points: i. It is crucial to check the kinetic relevance of the spectroscopic cell to be used for operando work. Appropriately modiﬁed DRIFTS reaction cells can behave as true kinetic reactor. ii. DRIFTS work can be performed in a fully quantitative manner. In most cases, the absorbance units are more appropriate than Kubelka-Munk units. iii. Chemical steady-state must be used to determine the true operando reactivity of surface species, particularly for catalysts whose composition are aﬀected by the reaction conditions. iv. The DRIFTS-SSITKA-MS is a truly operando technique and proved useful in unravelling the (minor) role of formate species seen by DRIFTS for the water gas shift reaction or CO hydrogenation over our samples. The nature of the true reaction intermediates remains unknown over these materials. 5.

Overall conclusions

The examples reported here show that combining spectroscopic and kinetic studies can greatly beneﬁt to the understanding of heterogeneous catalytic reaction. The example dealing with the selective reduction of NOx showed that reaction steps and/or elusive reaction intermediates could be postulated by the observation of side-products and the reactivity of model compounds. The use of fully quantitative methods also revealed that ‘‘slowly’’ reacting surface species (i.e. spectators) can sometimes be easily mistaken for true reaction intermediates. The conclusions derived from in situ/operando spectroscopic work should therefore always be carefully thought through and the corresponding limits of validity clearly deﬁned. We must be aware that far too often species observed by in situ/operando spectroscopic analysis are spectators or minor reaction intermediates (belonging to a slow parallel pathway), and it is careless to jump to the conclusion that these species are important, simply because those can be seen and even react. We hope that a more sensible view will now prevail, that is ‘‘because it can be seen, it is very possibly irrelevant to the main reaction pathway’’ and a proper quantitative spectrokinetic analysis is then carried out to relate the speciﬁc rate of decomposition of these surface species to the speciﬁc rate of product formation. 116 | Catalysis, 2010, 22, 94–118

References 1 P. Cong, A. Dehestani, R. Doolen, D. M. Giaquinta, S. Guan, V. Markov, D. Poojary, K. Self, H. Turner and W. H. Weinberg, Proc. Natl. Acad. Sci. U S A, 1999, 96, 11077. 2 K. Tamaru, Adv. Catalysis, 1964, 15, 65. 3 K. Tamaru, J. Chem. Soc. Japan, 1966, 87, 1007. 4 A. Ueno, T. Onishi and K. Tamaru, Trans. Farad. Soc., 1970, 66, 756. 5 K. Tamaru, Dynamic Heterogeneous Catalysis, Academic Press, N.Y., S.F., 1978. 6 A. Bell, Springer Ser. Chem. Phys., 1984, 34, 23. 7 D. Tibiletti , A. Goguet , F.C. Meunier, J. P. Breen, R. Burch, Chem. Commun. 2004, 1636. 8 V. A. Matyshak, A. A. Ukharskii and A. A. Kadushin, Zhumal Prikl. Spektmskopii, 1971, 24, 179. 9 V. A. Matyshak and O. V. Krylov, Catal. Today, 1995, 25, 1. 10 D. M. Stockwell, J. S. Chung and C. O. Bennett, J. Catal., 1988, 112, 135. 11 A. M. Efstathiou, T. Chaﬁk, D. Bianchi and C. O. Bennett, J. Catal., 1994, 147, 24. 12 A. M. Efstathiou, T. Chaﬁk, D. Bianchi and C. O. Bennett, J. Catal., 1994, 148, 224. 13 M. Agnelli, H. M. Swaan, D. Marquez-Alvarez, G. A. Martin and C. Mirodatos, C. J. Catal., 1998, 175, 117. 14 M. W. Balakos, S. S. C. Chuang and G. Srinivas, J. Catal., 1993, 140, 281. 15 Y. Schuurman and C. Mirodatos, Appl. Catal. A, 1997, 151, 305. 16 V. C. H. Kroll, H. M. Swaan, S. Lacombe and C. Mirodatos, J. Catal., 1997, 164, 387. 17 T. Chaﬁk, A. M. Efstathiou and X. E. Verykios, J. Phys. Chem. B, 1997, 101, 7968. 18 C. N. Costa and A. M. Efstathiou, J. Phys. Chem. B, 2004, 108, 2620. 19 F. C. Meunier, A. Goguet, S. Shekhtman, D. Rooney and H. Daly, Appl. Catal. A, 2008, 340, 196. 20 F. C. Meunier, J. P. Breen, V. Zuzaniuk, M. Olsson and J. R. H. Ross, J. Catal., 1999, 187, 493. 21 F. C. Meunier, J. P. Breen, J. R. H. Ross, Chem. Commun., 1999, 259. 22 R. Burch, J. P. Breen and F. C. Meunier, Appl. Catal. B, 2002, 39, 283. 23 F. C. Meunier, V. Zuzaniuk, J. P. Breen, M. Olsson and J. R. H. Ross, Catal. Today, 2000, 59, 287. 24 H. Hamada, Y. Kintaichi, M. Inaba, M. Tabata, T. Yoshinari and H. Tsuchida, Catal. Today, 1996, 29, 53. 25 T. Nanba, A. Obuchi A, H. Izumi, Y. Sugiura, J. Y. Xu, J. Uchisawa, S. Kushiyama, Chem. Commun., 2001, 173. 26 R. H. H. Smits and Y. Iwasawa, Appl. Catal. B., 1995, 6, L201. 27 A. Obuchi, C. Wogerbauer, R. Koppel and A. Baiker, Appl. Catal. B, 1998, 19, 9. 28 V. Zuzaniuk, F. C. Meunier, J. R. H. Ross, Chem. Commun., 1999, 815. 29 V. Zuzaniuk, F. C. Meunier and J. R. H. Ross, J. Catal., 2001, 202, 340. 30 A. Martı´ nez-Arias, M. Ferna´ndez-Garcı´ a, A. Iglesias-Juez, J. A. Anderson, J. C. Conesa and J. Soria, Appl. Catal. B, 2000, 28, 29. 31 K. Shimizu, J. Shibata, A. Satsuma and T. Hattori, PCCP, 2001, 3, 880. 32 K. Arve, E. A. Popov, F. Klingstedt, K. Eranen, L. E. Lindfors, J. Eloranta and D. Y. Murzin, Catal. Today, 2008, 100, 229. 33 A. Brucker, Chem. Commun., 2005, 1761.

Catalysis, 2010, 22, 94–118 | 117

34 S. J. Tinnemans, J. G. Mesu JG, K. Kervinen, T. Visser, T. A. Nijhuis, A. M. Beale, D. E. Keller, A. M. J. van der Eerden and B. M. Weckhuysen, Catal. Today, 2006, 113, 3. 35 B. M. Weckhuysen, Chem. Commun., 2002, 97. 36 M. O. Guerrero-Perez, Banares, M. A. Ban˜ares, Chem. Commun., 2002, 1292. 37 F. C. Meunier and M. Daturi, Catal. Today, 2006, 113, 1. 38 V. Dal Santo, L. C. Dossi, A. Fusi, R. Psaro, C. Mondelli and S. Recchia, Talanta, 2005, 66, 674. 39 M. M. Schubert, T. P. Haring, G. Brath, H. A. Gastiger and R. J. Behm, Appl. Spectrosc., 2001, 55, 1537. 40 J. Sirita, S. Phanichphant and F. C. Meunier, Anal. Chem., 2007, 79, 3912. 41 F. C. Meunier, D. Reid, A. Goguet, S. Shekhtman, C. Hardacre, R. Burch, W. Deng and M. Flytzani-Stephanopoulos, J. Catal., 2007, 247, 277. 42 C. Mirodatos, Personal communication, 1994. 43 G. Jacobs and B. H. Davis, Appl. Catal. A: General, 2005, 284, 31. 44 V. Matyshak and O. Krylov, Catal. Today, 1995, 25, 1. 45 F. C. Meunier, A. Goguet, C. Hardacre, R. Burch and D. Thompsett, J. Catal, 2007, 252, 18. 46 S. L. Shannon and J. G. Goodwin, Chem. Rev., 1995, 95, 677. 47 Y. Yang, R. S. Disselkamp, J. Szanyi, C. H. F. Peden CHF, C. T. Campbell and J. G. Goodwin, Rev. Sci. Inst., 2006, 77, 094104. 48 A. Goguet, D. Tibiletti, F. C. Meunier, J. P. Breen and R. Burch, J. Phys. Chem. B, 2004, 108, 20240. 49 R. W. Stevens and S. S. C. Chuang, J. Phys. Chem. B, 2004, 108, 696. 50 D. Tibiletti, A. Goguet, D. Reid, F. C. Meunier and R. Burch, Catal. Today, 2006, 113, 94. 51 D. Tibiletti , A. Goguet , F. C. Meunier, J. P. Breen, R. Burch, Chem. Commun., 2004, 1636. 52 F. C. Meunier, D. Tibiletti, A. Goguet, D. Reid and R. Burch, Applied Catalysis A, 2005, 289, 104. 53 Y. Yang, C. A. Mims, R. S. Disselkamp, C. H. F. Peden and C. T. Campbell, Topics Catal., 2009, 52, 1440. 54 F. C. Meunier, D. Tibiletti, A. Goguet, S. Shekhtman, C. Hardacre and R. Burch, Catalysis Today, 2007, 126, 143. 55 R. Leppelt, B. Schumacher, V. Plzak, M. Kinne and R. J. Behm, J. Catal., 2006, 244, 137. 56 T. Shido and Y. Iwasawa, J. Catal., 1992, 136, 493. 57 G. Jacobs and B. H. Davis, Appl. Catal. A: General, 2007, 333, 192. 58 A. A. Gokhale, J. A. Dumesic and M. Mavrikakis, J. Am. Chem. Soc., 2008, 130(4), 1402. 59 J. A. Rodriguez, P. Liu, J. Hrbek, J. Evans and M. Perez, Ang. Chem. Int. Ed., 2007, 46, 1329.

118 | Catalysis, 2010, 22, 94–118

Oxidative conversion of lower alkanes to olefins K. Seshana DOI: 10.1039/9781847559630-00119

1.

Introduction

In the current scenario of decreasing oil reserves, light alkanes oﬀer choice as feedstock for chemicals, fuels and energy. They are available as Natural Gas (NG, methane, ethane) and Liqueﬁed Petroleum Gas (LPG, propane, butanes) in the earth’s crust. Gas reserves are expected to last for a number of years and will thus become important feedstocks, compared to crude oil, for reﬁneries and petrochemical plants for years to come.1–3 Light oleﬁns (ethene, propene, butenes) are the current building blocks for the synthesis of: (i) bulk chemicals, e.g., ethylene oxide, acrolein (ii) polymers, e.g., polyethylene, -propylene or -butenes or (iii) fuels such as diesel, gasoline, e.g., by butene/butane alkylation. Direct synthesis of these bulk products from alkanes is still not extensively commercialized maleic anhydride from butane being one of the few example.4 Alkane conversion to oleﬁns is therefore a key step in the manufacture of bulk chemicals and fuels. As the title suggests, this review is restricted in scope to oxidative alkane conversions to lower oleﬁn, which are intermediates to fuels e.g., gasoline, and chemicals e.g. oxygenates such as acrolien, acetone, maleic anhydride etc, via various chemical routes as practiced in a reﬁnery/petrochemical complex. Direct/single step conversion of alkanes to such chemicals is equally interesting from the scientiﬁc and economic point of view. There has been tremendous amount of attention in the academic and industrial laboratories for the development of eﬃcient catalysts for such routes. This topic is beyond the scope of the current review, and readers are directed to reviews on these topics. Some key areas that have dominated in the last years are (i) direct oxidation of methane to methanol or formaldehyde,5–7 ethane to acetic acid/acetaldehyde8–10 (iii) propane to acrolien,11,12 acetone,13–15 propylene oxide,16 (iv) butane to maleic anhydride.4,17 = The present industrial capacity for lower oleﬁns (C= 2 –C4 ) is expected to be insuﬃcient, as the demand for these important components in the modern reﬁnery/petrochemical industry grows.18 These light oleﬁns are currently produced using fossil oil, e.g., from catalytic or steam cracking of naphtha and associated gas, or from ﬂuid catalytic cracking of vacuum gas oil. While these two routes are well developed, increasing the capacity of these processes is only possible to some extent, as changing regulations limit the use of byproducts (notably aromatic molecules) in fuel. In this context, methane as a feedstock has drawn wide attention. Tremendous eﬀorts both at academic and industrial laboratories in the 80’s and 90’s to convert methane to ethylene via catalytic oxidative coupling did not lead to commercial success.19,20 The research failed a

Catalytic Processes & Materials, Faculty of Science & Technology, University of Twente, #ME-361, PO Box 217, 7500 AE, Enschede, The Netherlands

Catalysis, 2010, 22, 119–143 | 119 c

The Royal Society of Chemistry 2010

to develop catalysts that resulted in appreciable ethylene yields the maximum yields reported were o30%,21 and this was the major stumbling block. Oxidative coupling of methane basically requires two elementary steps, viz., (i) C–H bond cleavage in methane to produce CHd3 or CHd2 radicals, and (ii) coupling of these radicals to produce ethane or ethylene, respectively. The former is an endothermic reaction and considering the strength of the C–H bond (232 kJ/Mol) in methane, requires very high temperatures (W750 1C).22 Unfortunately, the C–C coupling is an exothermic reaction favored at lower temperatures. There is thus a fundamental barrier to attempts to convert methane via oxidative coupling, and the failure to achieve appreciable ethane or ethylene yields is not surprising. Methane activation at lower temperatures where the exothermic coupling is more favorable could be a more promising method, as will be shown later in this chapter.18 The conversion of propane and butanes is more straightforward, as (e.g.) steam cracking is well established and is commercially practiced for the production of C2 to C4 oleﬁns. Steam cracking follows a radical chemistry route, the carbon radicals (primary or secondary) formed initially via C–H bond cleavage result in smaller primary radicals after subsequent b-cleavage (see Fig. 1). As radicals do not undergo isomerisation, every further b-cleavage of the primary radicals formed results in C2 product. Steam cracking therefore maximizes ethylene yields, and this route becomes less attractive when propylene demand grows faster than that of ethylene.23 Catalytic cracking over zeolitic and other solid acid catalysts, on the other hand, follows an ionic mechanism (Fig. 2) involving carbo-cations (carbeþ nium ions mostly, CnH2n þ 1). Carbo-cations do undergo isomerisation easily, since hydride (H ) transfer is facile over the ionic zeolite lattice. Isomerisation leads to the more favorable secondary carbo-cations, which on b-cleavage lead to oleﬁns with three or more carbon atoms depending on the carbon chain length in the reactant hydrocarbon. Catalytic cracking is thus an option for higher oleﬁns, however, most of the capacity in Fluid Catalytic Cracking units is used for fuels, e.g. gasoline production. Dehydrogenation of alkanes to oleﬁns is an excellent option for three reasons: (i) alkanes are cheap feedstocks (ii) it has the advantage that it generates oleﬁns, e.g., C3 H8 ! C3 H6 þ H2

DH ¼ 117 kJ mol1

with the same carbon number as the alkanes and (iii) byproduct hydrogen is in extremely high demand. However, dehydrogenation reaction technology +

Fig. 1

+ + Fig. 2

120 | Catalysis, 2010, 22, 119–143

has some major disadvantages, as the yields are limited by thermodynamic equilibrium, and there is a strong tendency to coking and consequently catalyst deactivation can be severe, leading to short lifetimes.24–28 Catalyst lifetimes are indeed small in practical commercial applications, e.g., eight hours to a few seconds (continuous catalyst regeneration) are reported. The existing processes for the dehydrogenation of light paraﬃns such as OLEFLEX (UOP, Pt/Al2O3 catalyst),25 CATOFIN (ABB and Lummus Crest, Cr catalyst),26 STAR (Phillips Petroleum Company, Pt based catalyst),27 and FDB-4 (Snamprogetti-Yarsintez, Chromium oxide)28 typically consist of catalyst regeneration (i.e., carbon burn-oﬀ) in combination with heat integration. These processes appeared in the early 80’s but have made only limited breakthrough commercially. Continuously increasing global demand for light alkenes has, therefore, spurred substantial interest in the development of alternative routes involving light alkanes as feedstocks. The incentive for alternative processes is predicted to grow in the future as the increase in the availability of light alkenes from reﬁneries is expected to be quite limited.29 Therefore oxidative conversion of alkanes to oleﬁns presents an available option. This is achieved via selective combustion of the hydrogen formed in the conversion of alkanes to oleﬁns, C3 H8 þ 1=2O2 ! C3 H8 þ H2 O

DH ¼ 126 kJ mol1

and the reaction is termed oxidative dehydrogenation.30 The major advantages of oxidative dehydrogenation over conventional dehydrogenation is that it: (i) overcomes the thermodynamic equilibrium limitations on oleﬁn yield faced in the direct catalytic dehydrogenation (ii) minimizes the coke formation and the related catalyst deactivation during conversion due to the presence of oxygen and (iii) avoids the need for heat input, as the reaction is exothermic and can be run adiabatically and at lower temperatures.30,31 Oxidative conversion (coupling) of methane was extensively studied, as mentioned earlier, however oleﬁn yields have not been impressive enough for commercial applications. Oxidative dehydrogenation of ethane has also been studied by more or less the same researchers and over similar catalysts also without success.32 Currently, there is tremendous interest in the oxidative conversion of propane and butane. Oxidative dehydrogenation of alkanes to oleﬁns is therefore still at a developmental stage despite the enormous amount of research and development activities carried out by academic/industrial laboratories. No commercial process is operative at the moment. Development of catalysts that minimize combustion and maximize oleﬁn yields is still the bottleneck.33 This review addresses the eﬀorts over the last few years, by us speciﬁcally and by others, toward characterizing the catalyst materials and discerning their relation to the kinetic and mechanistic sequences involved in the conversion of alkanes to oleﬁns in the presence of oxygen. This review is spread into three general areas. Many catalysts reported are essentially oxides,34 and they operate via a Mars van Krevelen ‘redox’ type mechanism and generate oleﬁns with the same carbon number. These are discussed ﬁrst. Recent studies on oxidic catalysts with no formal ‘redox’ properties have shown tremendous improvement in oleﬁn yields. In this situation, in Catalysis, 2010, 22, 119–143 | 121

addition to the C–H bond scission via oxidative dehydrogenation, C–C bond cleavage is also observed and this results in a mixture of oleﬁns with diﬀerent carbon numbers. This process is termed oxidative cracking and discussed next. Finally, recent eﬀorts in alkane activation using cold plasma at ambient conditions in micro reactors shows that oxidative C–C bond coupling is possible at these conditions. This leads to molecular weight buildup, i.e., the formation of oleﬁns with a higher C-number than the starting alkanes, and promises exciting new chemistry and applications. This is discussed last. A few thoughts on futuristic scenarios complete the manuscript. 2. Oxidative conversion of alkanes to oleﬁns over oxide catalysts with redox properties Among the alkanes, methane conversion to oleﬁns has been the most studied process. Most of these developments took place in the 80’s and 90’s. All sorts of oxidic catalysts were attempted, with alkali and alkaline earth oxides the most studied.1,2 In the case of methane, attempts were focused in the catalyst/oxygen assisted cleavage of C–H bonds to yield CH2, CH3 radicals. Coupling of these radicals was expected to yield ethylene or ethane, respectively, and the process was termed oxidative coupling. In spite of the enormous amounts of research and development, C2 yields were limited (o30 mol%), breakthroughs did not appear and eﬀorts stopped in the 90’s. The major barrier to conversion was that selective C–H bond activation in methane required very high temperatures (700–800 C), even in the presence of a catalyst and oxygen. At these high temperatures, radical coupling, which is essentially exothermic, was not a facile process and hence C2 yields never were appreciable. We do not attempt to go into these details and readers are directed to excellent reviews from M. Baerns, J.R.H. Ross1 and Lunsford.2 Later in this manuscript, we will show that if the activation of C–H bonds in methane can be achieved at lower temperatures, C–C coupling reactions can be facilitated.18 Ethane is a relatively cheap feedstock and is mostly associated with methane in the natural gas stream. As in the case of methane, ethane conversion focused on (oxy)dehydrogenation to ethylene. Most of the research was carried in the same laboratories that were studying oxidative coupling of methane. The catalysts studied were therefore similar to those attempted for methane.32 Again, no signiﬁcant breakthroughs were made. Results from Union Carbide on V-Nb-Mo based catalysts showed initial promise (very high ethylene selectivities were reported), but did not lead to commercial success.35,36 Conversion to syngas (CO þ H2) by oxidation with water (steam reforming), carbon dioxide (dry reforming) or oxygen (partial oxidation) remains still the main route for the valorization of these alkanes, especially methane. The ever increasing demand for higher oleﬁns such as propene and butene, their limited availability from steam/catalytic cracking operations and the cheapness/abundance of propane and butane (LPG) has meant that 122 | Catalysis, 2010, 22, 119–143

development of catalysts for oxidative dehydrogenation of these feed stocks is still an attractive option. Most of this work was focused on vanadia- or molybdena-based oxide systems which undergo redox changes during the conversion. Acidic (Al2O3, SiO2) and basic (MgO, CaO) oxides have been used as supports. Basic supports such as MgO have been the most beneﬁcial, because they minimize re-adsorption of product oleﬁn and thereby minimize combustion. Relatively high yields were reported for Mo-Ni-O catalysts,37 (30% conversion of propane with 60% selectivity at 600 1C) and V-Mg-O catalyst34 (33% conversion of propane with 42% selectivity, also at 600 1C). However, limited oleﬁn selectivity at higher alkane conversions is linked to strong oleﬁn adsorption (on cationic Lewis acid sites) and the subsequent oxidation to carbon oxides. This is clearly visible in the results shown in Fig. 3.31 This behavior is typical for almost all the catalysts studied. At higher propane conversions, sequential oxidation of the propylene product to carbon oxides becomes signiﬁcant. Among the vanadia based catalysts reported, use of niobia as a support has been shown to improve propylene selectivity31 up to 80 mol%. Niobia is a non acidic support it is reported to lose all its Bronsted and Lewis acidity when calcined above 350 1C.38 The absence of acidity clearly reduces product oleﬁn re-adsorption, and minimizes its sequential combustion. Such catalysts give higher oleﬁn selectivities, but the presence of residual acidity, which cannot be completely removed, is still detrimental. However, even on these catalysts, higher propane conversions lead to loss of oleﬁn selectivity, as can be seen from Fig. 4. Higher temperatures give better yields, but the secondary combustion of oleﬁn cannot be avoided. Lattice oxygen in these oxides participates in the oxidation (oxidative abstraction of hydrogen) of the alkane. It is suggested by Kung et al.34 that the C–H bond is split homolytically over the vanadia catalysts, creating propyl radicals. Kung et al. further elucidated the role of homogeneous radical reactions in oxidative dehydrogenation of propane over V-Mg-O catalysts, revealing the contribution of each component (homogeneous and 6

Yield [mol%]

425 °C

Propene

4 CO2 2 CO 0 0

2.5

5

7.5

10

Contact time [s] Fig. 3 Oxidative dehydrogenation of propane over V2O5/MgO catalyst, inﬂuence of contact time (conversion) on yields.

Catalysis, 2010, 22, 119–143 | 123

8

Propene yield [mol %]

425°C

V-Nb-O monolayer

V-Nb-O bulk

4

Vanadia Niobia 0

0

1

5

15

20

Propane conversion [mol %] Fig. 4

Conversion–yield plots for a series of vanadia based catalysts at 425 1C.

Intrinsic activity per site (Arbitrary Units)

4 3 2 1 0 0

10

20

30

40

50

60

Surface V concentration (mol %) Fig. 5 The intrinsic activity per V site as a function of the surface V site concentration determined by LEIS. The intrinsic activity per site is determined by dividing the propane conversion by the surface V concentration and is normalized to 1 for the lowest activity.

heterogeneous) in the overall process.39 The hydroxyls thus formed are released to the gas phase as water, reducing the catalyst. Alternatively, the lattice oxygen also takes part in the combustion of the alkane. Re-oxidation of the catalyst by gas phase oxygen completed the catalytic cycle and regenerates the catalyst. The nature and concentration of surface active sites determine propane conversion and product selectivity (oleﬁn vs. COx). Low Energy Ion Scattering is a useful technique to determine surface concentrations of oxide species. Compared to XPS, which probes a few atom layers, LEIS allows measurement at the outermost layer. Hence it is an optimal method to determine activity correlations. Fig. 5 shows a correlation between intrinsic activity (TOF, s 1) for propane conversion as a function of surface vanadia concentration for a series of catalysts determined by LEIS.40 An increase in 124 | Catalysis, 2010, 22, 119–143

Fig. 6 Width of the static coordination.

51

V chemical shielding anisotropy as a function of the vanadium

intrinsic activity (activity per site) as a function of vanadium concentration suggests that diﬀerent vanadia sites of varying catalytic activity must be present at the surface. 51 V solid-state NMR allows one to discriminate various vanadia species41 based on chemical shift anisotropies (see Fig. 6). Our measurements showed that three types of vanadia species were present i.e., vanadium in an isolated tetrahedral oxygen environment, in corner sharing tertrahedra, and in distorted octahedra. The number of octahedral sites increased with increasing vanadium concentration. Thus when the vanadium concentration is changed the types of species on the surface and their concentrations change. Larger vanadia clusters (Q1, Oct) have linking oxygen bonds between them which are susceptible to redox changes under the oxidative dehydrogenation conditions. Vanadium in diﬀerent oxygen environments and their diﬀerent redox natures would be the reason for the changes in intrinsic activities when catalysts with diﬀerent vanadia loadings are considered. Oxides typically contain a variety of oxygen species, e.g., vanadia exhibits both V–O–V bonds holding the vanadia octahedra together or dangling VQO bonds. These bonds can be easily characterized by IR spectroscopy. There is debate over the role of these oxygens in the selective oxidation and combustion of propane. It has been claimed that VQO bonds lead to combustion while V–O–V bonds are involved in selective oxidation to oleﬁns.42 This is based on the observation that V2O5, which exhibits VQO bonds, mostly gives combustion. Supporting V2O5 on MgO results in the formation of magnesium orthovanadates, which do not have VQO bonds. These catalysts are more selective to propylene formation and combustion is minimized. Alternatively, the oxygen present in the V–O–V bond has been characterized as responsible for oleﬁn formation as well as combustion. However, oxide based catalysts which show facile redox properties tend to catalyze sequential combustion of the oleﬁn formed. A detailed survey by Cavani and Triﬁro showed,37 the limits achieved oleﬁn yields were almost always below 30 mol%. This situation has changed very little in the last few Catalysis, 2010, 22, 119–143 | 125

years, and the search for an eﬃcient catalyst to convert C3–C4 alkanes to oleﬁns is still on. In order to eﬃciently convert alkanes to oleﬁns, the critical issue for catalyst development is to minimize oleﬁn sorption and further oxidation. This issue is discussed in the next section. 3. Oxidative conversion of alkanes over oxide catalysts with no formal ‘‘redox’’ properties Despite the tremendous amount of work reported in the literature,19,20,33,34,37–43 a commercial breakthrough for oxidative conversion of alkanes to oleﬁns is still lacking. As discussed in the previous section, one of the major diﬃculties in developing eﬃcient catalysts has been the fact that the product oleﬁn is more prone to combustion than the starting alkane.45 Thus higher oleﬁn selectivities can be obtained only at low conversions, and oleﬁn yields decrease at higher conversions due to sequential oxidation of the oleﬁn.31 Oleﬁn re-adsorption and subsequent combustion on the reducible oxide catalyst is indeed claimed as the reason for the lack of exciting progress in the ﬁeld of oxidative conversion/dehydrogenation of alkanes.23,31–33,45 One way to overcome this problem may be to minimize the oleﬁn sorption on the oxide catalyst, even modifying it to favor desorption. Oleﬁns are nucleophilic, electron rich, and thus an oxidic surface with basic (nonacidic) properties would be a proper choice.23 Basic alkali and alkaline earth oxides have been attempted as catalysts for the oxidative coupling of methane to ethylene9,43 and the oxidative dehydrogenation of ethane to ethylene.32,44 One of the most studied catalysts is Li/MgO. Work in the last years shows that it is also a promising catalysts for the oxidative conversion of propane and butane.23 As expected (Fig. 7), the Li/MgO catalyst shows oleﬁn selectivity for the oxidation of propane that is almost independent of conversion. This is diﬀerent from the case of redox-type catalysts reported in the previous section. It can be seen from 10

mol%

8

olefin selectivity

6 585°C 4 yield 2 0 0

30

60

conversion (%) Fig. 7 Oxidative dehydrogenation of propane selectivity to oleﬁns function of propane conversion for 1 wt% Li/MgO catalyst, results obtained varying space velocity.

126 | Catalysis, 2010, 22, 119–143

50

Total olefins

Yield (mol%)

600°C COx 25

Propene 0 0

25

50

75

100

Conversion (mol%) Fig. 8 Oxidative dehydrogenation of propane over Li-MgO based catalysts.

Fig. 8 that the oleﬁn yields vs. conversion are almost linear and no secondary conversion of oleﬁns to COx is observed. This also implies that high oleﬁn yields (50 mol%) can be achieved, as seen from Fig. 8. As reported often in the case of methane coupling, the catalytic activity of Li-promoted MgO is determined by surface [O ] species, and their existence in MgO was mainly shown using the electron paramagnetic resonance (EPR) technique.43,46,47 Lunsford suggested that the [O ] species was created by the substitution of Li þ for Mg2 þ ions to allow charge balance, and was stabilized in the MgO lattice as [Li þ O ] centers.48 Remarkably, [O ] are reported to be very stable at high temperatures and can exist in the crystal lattice of metal oxides even in the absence of oxygen in the gas phase.49 The similar ionic radii of Li þ (rLi þ =0.76 A˚) and Mg2 þ (rMg2 þ =0.72 A˚) allows easy accommodation of Li þ in the lattice of MgO.50 Replacement of Mg2 þ by Li þ creates lattice defects, i.e., oxygen vacancies (positive holes) (Scheme below). The proposed active site [Li þ O ] is produced by a hole adjacent to a Li þ site trapping an oxygen atom.51,52 2Li0 Ox þ V d þ 1= O ! 2Li0 O d þ Ox Mg

O

O

2

2

Mg

O

O

Scheme: Proposed mode of formation of the [Li þ O ] (shown in KrogerVink notation above) active site in Li/MgO catalysts. A hole trapped at the O2 is adjacent to Li þ sites.51,52 Unlike conventional Li/MgO catalysts prepared by impregnation, sol gel methods used recently give high surface area Li/MgO, where such defect sites are enhanced and lead to improved catalyst activity.53 Evidence for the presence of [Li þ O ] defect sites in these catalysts also comes from IR spectroscopy measurements.54 In the case of MgO-based þ 2 materials, surface sites of low coordination, i.e., Mg2LC OLC pairs can behave as strong acid-base pairs. The weak adsorption of CO (hence the low Catalysis, 2010, 22, 119–143 | 127

C=O C=O

Mg2+5C

Mg2+5C

Mg2+4C

Mg2+4C C=O

Absorbance (a.u.)

C=O

Mg2+3C

a b c

2205

2185

2165

2145

2125

2105

Wavenumber cm-1 Fig. 9 IR spectra of CO adsorption at -193 1C over (a) 1 wt% Li/MgO, (b) 3 wt% Li/MgO and (c) 5 wt% Li/MgO catalysts prepared using sol-gel method.

temperatures required to observe them, at 193 1C) on Lewis acid sites þ produces IR bands at frequencies higher than the stretching such as Mg2LC frequency of the free CO molecule in the gas phase,55 and thus provides a tool for probing these surface sites. IR spectra of CO adsorbed on Li/MgO catalysts prepared by sol gel methods, at 193 1C, (Fig. 9) show evidence for the presence of such sites. In the case of 1 and 3 wt% Li/MgO, the spectra show three bands at approximately 2164–2168, 2152–2155 and 2146–2148 cm 1. These bands were also present in the spectra of CO adsorbed on MgO-sg and are assigned to CO adsorbed on Mg24Cþ single sites, Mg25Cþ single sites, and anchored to Mg24Cþ and Mg25Cþ at the step site, respectively. For the sample with the highest lithium loading (5 wt%), additional weak bands at 2200 and 2184 cm 1 were also present. The band at 2200 cm 1 is attributed to a CO molecule interacting with a Mg23Cþ site.56 Based on DFT calculations, the adsorption at 2184 cm 1 can be attributed to the addition of a second CO molecule to the CO-Mg23Cþ adduct, resulting in the formation of dicarbonyl species.56,57 In addition, changes in the relative distribution of all the components present in the spectra were observed. The appearance of sites of lower coordination of lithium atoms due to a decrease of the particle size can be certainly excluded, since the addition of lithium causes a substantial decrease in surface area and increase in particle size. Thus, sites of low coordination such as Mg23Cþ , which require oxygen vacancies near them, indicate incorporation of lithium in the lattice structure of MgO, creating the oxygen vacancies.

128 | Catalysis, 2010, 22, 119–143

Extensive work from the groups of Lunsford et al., and Ross et al., Seshan et al., have shown, in the case of Li/MgO catalysts, that the ﬁrst step in the oxidative conversion of methane involves the homolytic scission of C–H bonds forming surface –OH groups and alkyl radicals (1):19,23,43,46,47 ½OðSÞ þ CH4 ! ½OHðSÞ þ CH3d

ð1Þ

The resulting radicals are released from the catalyst surface and subsequently initiate gas-phase chain propagation reactions to yield products.23 In the case of vanadia-type redox catalysts, the only products obtained in the case of propane or butane were the corresponding oleﬁns (i.e., propene or butene), and carbon oxides. The product distribution obtained during the oxidative dehydrogenation of propane over a Li/MgO catalyst is given in Table 1. The presence of C1–C2 hydrocarbons in the products indicates that Table 1 Oxidative dehydrogenation of propane over Li/MgO catalysts at 550 C, propane conversion 15% Component

CO þ CO2

Methane

C2 þ C= 2

C= 3

Selectivity (mol%)

15

10

30

45

both C–H scission (oxidative dehydrogenation) and C–C bond splitting (oxidative cracking) occur over Li/MgO. In order to explain these observations, reaction sequences similar to that of oxidative coupling of methane over Li/MgO catalysts have been proposed (Fig. 10). These sequences involve initiation at the catalyst surface followed by gas phase homogeneous reactions.43 Leveles et al.23 discussed the performance of the catalyst tested in relation to the reaction mechanism for propane ODH and reported that propane activation occurs via a C–H bond splitting

CH2O

O2 C3H8

H2 O

2 HO·

·CH3

COx C2H4

CH4 ½nC3H7· H2 H2O2

½iC3H7·

C3H8 H·

O2

C3H6

HO2· Fig. 10 Proposed reaction mechanism for gas phase propyl radicals.

Catalysis, 2010, 22, 119–143 | 129

which is also the rate determining step. Propane activation takes place on the [Li þ O ] active site by homolytic hydrogen abstraction, forming [Li þ OH ] and propyl radicals equation (1). There is general evidence that the two different primary radicals formed after heterogeneous activation, n- and isopropyl radicals, are released from the catalyst surface to the gas phase, and radical chain reactions lead to the ﬁnal products. For the propyl radicals formed, two diﬀerent decomposition routes have been proposed for the gas phase: (i) the isopropyl radicals can undergo scission of C–H bond at the aposition and decompose into propylene and H an radical and (ii) the n-propyl radical undergoes C–C cleavage in the b-position, forming a methyl radical and ethylene (Fig. 10). Thus, the cleavage of a C–H bond (dehydrogenation) versus cleavage of a C–C bond (cracking) is the processes controlling the selectivity to, respectively, propylene and ethylene. At a relatively low T (o600 1C), and at a low conversion level of propane, the contribution of the gas phase reactions is less, and a propylene to ethylene ratio higher than 1 was observed.23 At higher temperatures (Z600 1C), a decrease in the rate of dehydrogenation and an increase in cracking was recorded. Surprisingly, for identical conversions, increasing the number of defect [Li þ O ] sites improved selectivity to oleﬁns and higher propylene to ethylene ratio in the products was observed. Indeed, as Kondratengo et al. reported, in the case of propane ODH over vanadium oxides systems increasing the density of active sites aﬀected the oleﬁn distribution.58 In particular, they suggested that a high density of oxidizing sites is essential for further ODH of the propyl radicals to propylene, and therefore suppression of the concurrent cracking reaction pathway to ethylene. Similarly, in the case of Li/MgO-sg catalysts with a high density of active sites (per volume of catalytic bed), heterogeneous H-atom abstraction from C3Hd7 -radicals yielding propylene eq. (2) can be more eﬃcient than reaction of C3Hd7 -radicals in the gas phase and this can aﬀect the propylene to ethylene ratio. þ ½Liþ O þ C3 H! 7 ½Li OH þ C3 H6

ð2Þ

Propane activation on [Li þ O ] active centers, leading to the propyl radical, is a single site interaction. The chance that the propyl radicals may further react with a second [Li þ O ] active site leading to propylene (multiple site interactions) strongly depends on site density. The ﬁnal step43,59 is the regeneration of the active site, which involves electron transfer to the anion vacancy and the dissociative chemisorption of oxygen, respectively eqn. (3) and (4): ½Liþ O2 þ ½Liþ Va ! ½Liþ O þ ½Liþ V a

ð3Þ

þ ½Liþ V a þ O2 ! ½Li O þ O

ð4Þ

However, certain features in the Ito-Lunsford mechanism appear to be unlikely, especially at lower temperatures. In particular, removal of oxygen from the lattice is not facile and this would be the rate-limiting step of the catalytic cycle rather than hydrogen abstraction.59 Moreover, migration of a proton would require: (i) substantial energy to overcome the electrostatic barrier60 and (ii) the proximity of [Li þ O ] centers.61 130 | Catalysis, 2010, 22, 119–143

H2O ½ O2 H

H

O

O

Fig. 11 Schematic drawing: mechanism of regeneration of the active sites as suggested by Sinev.35

Alternatively, Sinev62,63 proposed a new mechanism for regeneration of the active sites that does not require the removal of lattice oxygen and thus the formation of oxygen vacancies. In fact, the re-oxidation of the catalyst can proceed by the mechanism of oxidative dehydrogenation of surface OH groups, which requires the scission of strong O–H bonds. More speciﬁcally, the regeneration reactions proposed by Sinev (Fig. 11) are summarized as:64 O2 þ ½OH ! ½O þ HO2d þ

ð5Þ

HO2d ½OH ! ½O þ H2 O2

ð6Þ

H2 O2 ! 2OH

ð7Þ

OH þ ½OH ! ½O þ H2 O

ð8Þ

It is appropriate to stress that the overall reaction for regeneration is the same as in the mechanism proposed by Ito and Lunsford.59 In fact, it involves the participation of two surface [OH ] groups and the formation of water. However, it does not require removal of lattice oxygen. It is known that Li/MgO catalysts deactivate if no oxygen is present during reaction. Additionally, the deactivation was not accompanied by water formation. This implies surface deactivation without any lattice oxygen removal and, most likely, that formation of oxygen vacancies as intermediates does not take place. To conclude, under these deactivation conditions surface [OH ] groups are formed, and are stable in the absence of gas phase oxygen. Thus, Li/MgO catalysts do not show any reducibility at 550 1C. Remarkably, only during the interaction of oxygen with the catalysts pretreated in propane or hydrogen was the evolution of water was observed. This may suggest, as proposed by Sinev, that at 550 1C the re-oxidation of Li/MgO catalysts proceeds as some sort of oxidative dehydrogenation of surface hydroxyl groups (Fig. 15). Our observations suggest that only at higher temperatures (700 1C) i.e., at methane coupling conditions, the scheme proposed by Lunsford is operative. But, in the case of oxidative dehydrogenation of higher alkanes, which occur at lower temperatures (500 1C) the catalyst regeneration goes via traditional re-oxidation with a de-hydroxylation step involving the formation of oxygen vacancies. 4. Catalytic alkane oxidation at ambient conditions using cold plasma C–C, C–H scission vs C–C bond coupling Oxidative conversion of alkanes to oleﬁns, as shown in the last two sections, is reported to be initiated by homolytic splitting of a C–H bond resulting in Catalysis, 2010, 22, 119–143 | 131

radicals.23,39,40,43,46,47,53,65–67 In the case of catalysts with pronounced ‘‘redox’’ properties, e.g. supported vanadia catalysts such as V2O5/MgO,39 further conversions of the alkyl radical, adsorbed on the nucleophilic oxygen of the catalyst, occur on the oxide surface in a typical ‘‘Mars-van Krevelen’’ type redox mechanism. Oxygen from the gas phase regenerates the oxide, which is reduced during alkane oxidation.40 Accordingly, Kondratengo et al. have shown recently that in the case of vanadium oxide systems, a high density of oxidizing sites can favour H-atom abstraction from C3H7 radicals yielding propene.58 In the case of oxide catalysts with no formal redox properties, e.g. Li/MgO, the initial activation involves, again, homolytic C–H bond splitting forming alkyl radicals. Because the oxygen in these type oxides is less nucleophilic, the radicals are then released to the homogeneous gas phase where radical chain reactions determine the composition of oleﬁns formed.53,58 Thus alkane to oleﬁn conversion on such catalysts involves a route of heterogeneously-initiated homogeneous reactions. Results also indicate that total oxidation (combustion) reactions occur both in the gas phase and on the surface of the oxide.68 The contribution of such homogeneous gas phase routes in heterogeneous catalysis has been especially discussed in the last 20 years in the case of oxidative coupling of methane to higher hydrocarbons43 and oxidative dehydrogenation of light alkanes (ODH) to oleﬁns.23 By varying the post catalytic volume of the reactor and observing that this caused an increase in conversion,66 it was concluded that gas phase radical reactions make a major contribution to such processes. In addition, direct evidence for the presence of surface-generated gas-phase radicals has been provided by spectroscopic methods, in particular by techniques such as matrix isolation electron spin resonance (MI-ESR) and infrared spectroscopy (MI-IR), in tandem with a catalytic reactor.46 In the case of oxidative dehydrogenation of alkanes, using Li/MgO catalysts, an increase in oleﬁn selectivity could be achieved by increasing the number of active sites per volume of catalytic bed.68 We suggested recently, that the propyl radicals formed ½Li þ O þ C3 H7d ! ½Li þ OH þ C3 H6

ð9Þ

could undergo a second hydrogen abstraction at the active sites leading to propene,68 indicating that oleﬁn formation on the catalyst surface is possible. Generally, propane activation involving C–C or C–H bond scission37 requires higher temperatures (TW550 1C) even in the presence of a strong [Hd] abstractor such as [Li þ O ]. Lowering the temperature of the reaction can facilitate further radical conversions on the catalyst surface as in equation (1). Therefore the oleﬁn selectivity can be manipulated by developing more active catalysts. The use of cold plasma is one way to achieve C–H bond activation at ambient temperatures. Stable and cold gaseous plasma can be generated at room temperature inside a micro reactor by dielectric barrier discharge.70,71 The advantage of using a microreactor is that it allows generation of nonthermal plasma at higher pressures, i.e., without the need for vacuum. This 132 | Catalysis, 2010, 22, 119–143

plasma consists of energetic electrons which can activate alkane C–H bonds as a result of inelastic collisions.71 Additionally, performing the reaction in a micro scale system (channel dimensions 10–1000 mm) with intrinsically high surface to volume ratio provides extreme quenching conditions on the catalyst surface.72,73 Thus a micro plasma reactor containing catalyst can be used to convert light alkanes into more valuable molecules at room temperature and atmospheric pressure.74 Incorporation of a stable catalyst layer on the reactor wall is crucial in such a situation. In the case of catalytic reactions many attempts have been made to deposit catalysts on the walls of micro reactors, instead of introducing powder-type materials in the micro channels in packed-bed fashion, which leads to a high pressure drop.75 Figs. 12,13 show, respectively, the top and schematic cross sectional view of the microplasma reactor used by Trionfetti et al.68 It consists of a Pyrex rectangular chip of 50 mm length 15 mm width. Microchannels of 30 mm length 5 mm width and a channel depth of 500 mm were realized in the chip by means of chemical etching using aqueous HF. Sandwich thermal bonding of three Pyrex plates (one with, two without a channel) allowed fabrication of the micro reactor. Details of the processing scheme are given elsewhere.76 Gas inlet and outlet holes were created by powder blasting using Al2O3 particles. The two copper ribbon electrodes connect to an external power supply in order to generate plasma by dielectric barrier discharge (DBD) at atmospheric pressure.90,91 A power output between 2 and 25 W is easy to stabilize. The power absorbed by the plasma can be calculated from the corresponding V–Q Lissajous Figures.77 The plasma generated in the DBD conﬁguration consists of high energy electrons and is characterized by a large number of micro-discharge ﬁlaments (ionization of the medium by the electrons), each lasting few nanoseconds.78 These high energy electrons (in the range 3–4 eV) are able to activate hydrocarbons and oxygen at room temperature and atmospheric pressure.79 The short life time of the current spikes (ns) helps in minimizing local heating. Moreover, the

Fig. 12 Top view of the employed microplasma reactor made of Pyrex. The inlet and out let are indicated (A, B). The copper plate is also well visible (C). This is connected to a power supply using adhesive copper foils (D).

Catalysis, 2010, 22, 119–143 | 133

Dielectric-Pyrex

Surface

∼

Discharges

Catalyst layer additional dielectric Copper Electrode Fig. 13 Schematic drawing of the cross-sectional view of the employed microplasma reactor. The 3 Pyrex plates forming the microreactor are schematically represented.

small volume and the large surface to volume ratio of the micro reactor allow fast removal of the heat produced during oxidation of alkanes. The channels of the micro reactor were coated with Li/MgO catalyst layer by wetting the walls with sol via micro-pipetting, allowing the sol to gel and subsequently calcining at 500 1C to get the catalyst coating.68,69,80–83 Details of the characteristics of the reactor and catalyst system are extensively discussed elsewhere.68,80–83 Results obtained by us for propane oxidation highlight some very exciting chemistry.68 Fig. 14 shows the optical emission spectrum of C3H8–He plasma with 3 W power input. Electronic excitation of ‘CH’ corresponding to the A2D-X2P transition at 431.5 nm was used to determine the kinetic gas temperature in this emission region.77 The resolution of the spectrometer, calibrated using a UV lamp, was determined to be 0.7 nm (full width at half maximum). This is not enough to resolve the individual rotational lines of the Q, R and P branches of the CH band. However, the rotational temperature, which reﬂects the gas temperature inside the ﬁlamentary discharge, was calculated by comparing the CH band in Fig. 14 with those in simulated spectra at various temperatures using the LIFBASE software.84 The best ﬁt was obtained in the region of 25–75 1C. Remarkably, a thermocouple inserted inside the micro channel measured 75 1C at the highest power of 24 W supplied from the source. In the case of the 3W used in our experiments an average gas temperature of 25 50 1C was obtained. Thus, the spectra in Fig. 14 indicate that propane activation occurs at close to room temperature. The existence of CH and H bands in the spectra shown in Fig. 14 indicates decomposition of propane via both C–C and C–H bond scission. In agreement with this supposition, the product stream showed components typical of the oxidative cracking of propane as discussed earlier, i.e., COx, ethane, propene and acetylene. Most remarkably, in the case of the Plasma 134 | Catalysis, 2010, 22, 119–143

He

He

Intensity (a.u.)

CH A2Δ→X2Π He

Hα

C2(d3Π-a3Π)

500

400

600

700

Wavelength (nm) Fig. 14 Optical emission spectrum for a gas mixture of 10% propane in helium in the presence of plasma; 3W power was applied.

Selectivity (mol%)

30

20

10

0

C3H6

COx

C2H6

C2H4

C2H2

CH4

C4

>C4

Products Fig. 15 Selectivity to the main products for a plasma microreactor. Conditions: ﬂow rate 15 ml/min, feed composition 10% propane, 1% oxygen and balance helium; activation at room temperature. Formation of hydrogen was detected but not quantiﬁed.

Catalysis, 2010, 22, 119–143 | 135

microreactor a substantial amount (37 mol%) of products with higher molecular weights than propane, i.e., C4, C4þ , were also observed (Fig. 15). The presence of products containing four or more C atoms requires C–C bond formation under the conditions present in the microreactor. For Li/MgO catalysts, propane activation requires the presence of oxygen and higher temperatures (TW600 1C)).35,53,85 In these catalysts [Li þ O ] type defect centers are the active sites and their formation requires the higher temperatures. The [Li þ O ] site is claimed to have a high H-atom aﬃnity and, at relative high temperature, is able to abstract Hd from propane molecules forming n- and iso-propyl radicals as primary intermediates.85–87 These propyl radicals are formed depending on whether primary or secondary hydrogen is abstracted from propane. At the higher temperatures required for the reaction, endothermic decompositions are favored and the two types of radicals (n & iso) undergo diﬀerent unimolecular reaction routes in the homogeneous phase: iso-propyl radicals undergo C–H bond scission at the a-position, yielding propene and a hydrogen atom, while n-propyl radicals preferentially undergo b-scission of the C–C bond, forming a methyl radical and ethylene.88 The various products formed can be accounted for by a series of radical chain reactions in the gas phase, details of which were discussed earlier.89 In a plasma microreactor, C3H8 molecules are directly activated/converted via collisions with energized electrons. Activation produces radicals such as C3Hd7 due to cleavage of C–H bonds (2). These can initiate radical chain reactions. Thus, reaction (10) is strongly inﬂuenced by the number of charges transferred or accumulated on the dielectric surface,92 and therefore by the relative permittivity. C3 H8 þ e ! C3 H7d þ Hd þ e

ð10Þ

The reactivity of micro-plasma towards propane is improved when a layer of Li/MgO was present. This is due to the larger permittivity of oxide layer (9.7 C2/N*m2 for MgO) compared to Pyrex (4.8 C2/N*m2).93–95 This allows more impacts (and more inelastic collisions) giving rise to excitation of a higher number of propane molecules. In addition, propane activation can also occur via an indirect route, i.e., activation of gas phase oxygen by the plasma. It is also observed that in the presence of oxygen and plasma the propane conversions are higher. Among the atomic processes taking place in a non-thermal plasma, the electron impact dissociation of O2 to form charged and e þ O2 ! 2O þ e ! O þ O

ð11Þ

e þ O2 ! O2 þ e ! O þ O þ e

ð12Þ

neutral oxygen has been reported in literature and is described in reaction equations 3 and 4.96 The O species, present in the homogeneous phase, has been reported to cause C–H bond scission in alkanes e.g., methane,97 ethane.98 In the case of propane this will result in the formation of propyl and hydroxyl radicals as shown below: ½O þ C3 H8 ! ½OH þ C3 H7d 136 | Catalysis, 2010, 22, 119–143

ð13Þ

Further, C3H7 radicals react fast with O2 forming hydroperoxyl (HOd2 ) radicals, which can react with propane molecules to form H2O2. Decomposition of H2O2 results in hydroxyl radicals (OHd), which become the main chain propagators.47 [Li þ O ] sites are defect sites and their existence at lower temperatures has been investigated and conﬁrmed, using EPR spectroscopy, by Lunsford and others.48,59,100 Improved propylene yields suggest that defect [Li þ O ] sites may already be present at low temperature in the presence of plasma and oxygen, and that these sites abstract hydrogen from propyl radicals forming propylene.99 Alternatively, the presence of plasma can also help to create other defect sites which are able to selectively terminate radicals. Detailed studies are present in the literature describing the emission of UV light during dielectric barrier discharge (DBD).101 Nelson et al.102 and later E. Knozinger et al.103 reported, using EPR studies, that interaction between UV light and MgO particles can give rise to surface paramagnetic centers (trapped electrons, typically F-centers, [VdO]). Goodman et al. suggest that these oxygen vacancies, containing one or two electrons (F-type defects), are able to activate C–H bonds during methane oxidative coupling.104 Bond scission occurs and the Hd is trapped at the defect site. Thus, the presence of [VdO]-type defect sites caused by the plasma may allow Hd abstraction from propane and even from propyl radicals, and in the latter case enhanced selectivity to propene can be expected. Optical spectra recorded in presence of plasma (Fig. 18) show that CHtype radical species are formed. Dimerization of such species can result in ethyne, which is often detected in experiments with plasma and hydrocarbons.108 Hydrogen redistribution during this reaction, associated with the formation of dehydrogenated products such as ethyne, may account for the observation of methane. This argument regarding dimerisation and formation of ethyne105–107 can also be logically connected to the large amounts of C4 and C4þ products that we observe. The role of plasma is in the activation of propane and the formation of radicals at ambient temperatures. The formation of C4 and C4þ products from propane essentially requires C–C bond formation, which is an exothermic process and therefore favored at lower temperatures. Therefore it is not surprising to see C–C bond formation reactions under our conditions. In conventional ﬁxed bed reactors, propane activation occurs at higher temperatures (TW600 1C) in the presence of catalyst. These conditions favor the rupture of C–C bonds, and thus we see only products of cracking or combustion, i.e., with molecular weights lower than that of propane. Considering that similar amounts of C4 þ C4þ products are formed in the microreactor both with and without catalyst, it is suggested that the coupling of the radicals occurs predominantly in the homogeneous phase. Unlike propane, ethane and methane contain only primary carbon atoms. It would thus be interesting if C–H bond activation is possible in these molecules with plasma under oxidative catalytic conditions. Under identical conditions ethane is less active than propane, as expected. Table 2 shows typical product selectivities. Again, an appreciable amount (22 mol%) of C–C coupling products (C4 þ C3) were observed. Catalysis, 2010, 22, 119–143 | 137

Table 2 Oxidative conversion of ethane. Selectivity (mol%) to the products observed in a plasma microreactor. Conditions: 10% ethane and 1% oxygen in helium, ﬂow rate 15 ml/min, RT; 15% conversion level of ethane (mol%) COx

C3H8 þ C3H6

C2H4

C2H2

CH4

C4

13

12

23

15

27

10

Table 3 Oxidative conversion of methane. Selectivity (mol%) to the products observed in a plasma microreactor. Conditions: 10% methane and 1% oxygen in helium, ﬂow rate 15 ml/min, RT; 10% conversion level of methane (mol%) COx

C3H8 þ C3H6

C2H4 þ C2H6

C2H2

C4

30

11

49

8

2

C2 products were the most abundant species when methane was the feedstock (Table 3). Appreciable amounts of C3 (11%) and C4 (2%) products, demanding multiple C–C couplings, were also found. Thus activation by cleavage of the C–C and C–H bonds and C–C coupling at ambient temperatures are consistently observed for all three hydrocarbons studied, viz, propane, ethane and methane.109 Alkane activation occurs in the presence of plasma via collisions with energized electrons, e.g., C3H8 þ e -C3Hd7 þ Hd þ e .110 In the presence of helium this mechanism might become more eﬃcient because the mean free path of the electrons increases111 due to the small cross section of highly excited helium species created during electric discharge. These excited species of helium, can also possess higher energy, up to nearly 20 eV, enough to activate a hydrocarbon by collisions.112 Alkane activation can also occur via an indirect route, i.e., activation of gas phase oxygen by plasma. Among the atomic processes taking place in a non-thermal plasma, the electron impact dissociation of O2 to form charged and neutral oxygen has been reported in the literature and is described in equations 11 and 12.113,114 The O species, present in the homogeneous phase, have been reported to cause C–H bond scission in alkanes, e.g., methane,115 ethane.116 In the case of propane this will result in the formation of propyl and hydroxyl radicals as in equation 13. The reaction between C3Hd7 radicals with O2 results in hydroperoxyl (HOd2 ) radicals which react further with a propane molecule to form H2O2. Decomposition of the latter gives two hydroxyl radicals (OHd), which become the main chain propagators.23 All these reactions can initiate radical chain reactions, but the main diﬀerence is that these propagation and termination reactions now take place at ambient temperature. Eﬀorts to make C2 products by the oxidative coupling of methane, extensively studied over several years,117 had the inherent diﬃculty that methane activation required high temperatures and at the high temperatures C–C bond formation would be expected to be less favorable. Typically, formation of C–C bonds is an exothermic process and therefore favored at lower temperatures. Therefore, it is not surprising to see C–C bond 138 | Catalysis, 2010, 22, 119–143

formation reactions under the mild conditions present during plasma experiments. C–C bond coupling is also observed during oleﬁn metathesis type reactions, also at room temperature, but requiring a transition metal catalyst. Current scenarios for upgrading light hydrocarbons (increasing molecular weight) for, e.g., alkylation,118 metathesis119,120 or oligomerisation121,122 reactions (e.g., SHOP process123,124) involve at least one oleﬁn. However, in our studies direct homologation of alkanes is observed. This represents an interesting development for upgrading cheap low molecular weight alkanes to commercially useful fuels and/or feedstock materials for the chemical industry. Acknowledgements The author wishes to state his pleasure and thanks to students and colleagues who helped to carry out the work. In the hope of not missing anyone, they include J. R. H. Ross, R. H. H. Smits, J. G. van Ommen, H. M. Swaan, S. J. Korf, J. A. Lercher, I. Babich, S. Fuchs, L. Levels, L. Leﬀerts, A. Agiral, J. G. E. Gardeniers and C. Trionfetti. References 1 G. Centi, V. C. Corberan, S. Perathoner and P. Ruiz, Catal. Today, 2000, 61, 1. 2 J. H. Lunsford, Adv. Catal., 1987, 35, 139. 3 J. E. Lyons and G. W. Parshall, Catal. Today, 1994, 22, 313. 4 N. Song, Z. Xuan, J. K. Bartley, S. H. Taylor, D. Chadwick and G. J. Hutchings, Catal. Lett., 2006, 106, 127, and references therein. 5 N. D. Parkyns, C. I. Warburton and J. D. Wilson, Catal. Today, 1993, 18, 385. 6 R. H. Crabtree, Chem. Rev., 1995, 35, 987. 7 S. H. Taylor, J. S. J. Hargreaves, G. J. Hutchings, R. W. Joyner and C. W. Lembacher, Catal. Today, 1998, 42(3-9), 217. 8 L. Tessier, E. Bordes and M. Gubelmann-Bonneau, Catal. Today, 1995, 24, 335. 9 K. Ruth, R. Kieﬀer and R. Burch, J. Catal., 1998, 175, 16. 10 J. M. Galownia, A. P. Wight, A. Blanc, J. A. Labinger and M. E. Davis, J. Catal., 2005, 236, 356. 11 R. K. Grasselli, Topics in Catal., 2002, 21, 79. 12 C. Zhang and C. R. A. Catlow, J. Catal., 2008, 259, 17. 13 N. Mizuno, Catal. Surv. Japan., 2000, 4, 149. 14 G. Centi, V. C. Corberan, S. Perathoner and R. Ruiz, Catal. Today, 2000, 61, 1. 15 J. Xu, B. L. Mojet, J. G. van Ommen and L. Leﬀerts, J. Catal., 2005, 232, 411. 16 S. T. Oyama, K. Murata and M. Haruta, Shokubai (Catalysts & Catalysis), 2004, 46, 13. 17 M. T Sanane, G. J. Hutchings and J. C. Volta, J. Catal., 1995, 154, 253. 18 C. Trionfetti, A. Agiral, J. G. E. Gardeniers, L. Leﬀerts and K. Seshan, Chem. Phys. Chem., 2008, 9, 533. 19 M. Baerns and J. R. H. Ross, in ‘‘Perspectives in Catalysis’’, eds. J.M. Thomas, K.I. Zamarev, Blackwell, Oxford, 1992, p. 315. 20 Z. Zhang, X. E. Verykios and M. Baerns, Catal. Rev.-Sc. Engg., 1994, 36, 507. 21 M. V. Landau, M. L. Kaliya, M. Herskowitz, P. F. van den Oosterkamp and P. S. G. Bocque, ChemTech, 1996, 26, 24. Catalysis, 2010, 22, 119–143 | 139

22 S. J. Korf, R. A. Roos, J. A. Vreeman, J. W. H. C. Dersen, J. G. van Ommen and J. R. H. Ross, Catal. Today, 1990, 6, 417. 23 L. Leveles, K. Seshan, J. A. Lercher and L. Leﬀerts, J. Catal., 2003, 218, 296. 24 S. Gussow, D. C. Spence and E. A. White, Oil & Gas J, Dec 1980, 8, 96. 25 P. R. Cortelli, S. T. Bakas, M. F. Bentham, J. H. Gregor, C. R. Hamlin, L. F. Smith, ‘‘Oleﬂex process – The proven route to oleﬁns’’ UOP, Des Plaines, III, 1992. 26 D. Sanﬁlippo, ChemTech, August 1993, 35. 27 H. Bolt, H. Zimmerman, in M. Baerns, J. Whitelamp (eds), DGMK Proceedings, Kassel, 1993, p 175. 28 P.R. Sarathy, G.S. Suﬀridge, Hydrocarbon Processing, February 1993, 43 29 Chem. & Engg. News, April 1990, 21. 30 F. Cavani, N. Ballarini and A. Cericola, Catal. Today, 2007, 127, 113. 31 T. C. Watling, G. Deo, K. Seshan, I. E. Wachs and J. A. Lercher, Catal. Today, 1996, 28(1-2), 139. 32 H. M. Swaan, A. Toebes, K. Seshan, L. G. van Ommen and J. R. H. Ross, Catal. Today, 1992, 13, 201. 33 L. Leﬀerts, K. Seshan, B. Mojet and J. G. van Ommen, Catal. Today, 2005, 100, 63. 34 K. T. Nguyen and H. H. Kung, J. Catal., 1990, 122, 415. 35 E. M. Thorsteinsen, T. P. Wison, F. G. Yound and P. H. Kasai, J. Catal., 1978, 52, 116. 36 J. H. McCain, U.S. Patent 4524236, Union Carbide, 1984. 37 F. Cavani and F. Triﬁro, Catal. Today, 1995, 24, 307. 38 K. Iizuka, K. Ogasawara and K. Tanabe, Bull. Chem. Soc. Japan, 1983, 56, 2927. 39 M. A. Chaar, D. Patel, M. C. Kung and H. H. Kung, J. Catal., 1988, 109, 463. 40 R. H. H. Smits, K. Seshan, J. R. H. Ross, M. R. Anantharaman, L. C. A. van den Oetelaaar, J. H. J. M. Helwegen and H. H. Brongersma, J. Catal., 1995, 157, 585. 41 R. H. H. Smits, K. Seshan, J. R. H. Ross and A. P. M. Kentjens, J. Phys. Chem., 1995, 99, 9169. 42 K. Seshan, H. M. Swaan, R. H. H. Smits, J. G. van Ommen and J. R. H. Ross, Stud. Surf. Sc. & Catal., 1990, 55, 505. 43 D. J. Driscoll, K. Campbell and J. H. Lunsford, Adv. Catal., 1987, 35, 139. 44 E. Heracleous, A. F. Lee, K. Wilson and A. A. Lemonidou, J. Catal., 2005, 231, 159. 45 R. H. H. Smits, K. Seshan and J. R. H. Ross, J. Chem. Soc., Chem. Commun., 1991, 8, 558. 46 J. H. Lunsford, Langmuir, 1989, 5, 12. 47 J. X Wang and J. H Lunsford, J. Phys. Chem., 1986, 90, 5883. 48 K. Aika and J. H. Lunsford, J. Phys. Chem., 1978, 82(16), 1794. 49 M. . Wu, C. M. Truong, K. Coulter and D. W. J. Goodman, Vac. Sci. Technol., A, 1993, 11(4), 2174. 50 T. Berger, J. Schuh, M. Sterrer, O. Diwald and E. Knozinger, J. Catal., 2007, 247, 235. 51 P. J. Gellings and H. J. M. Bouwmeester, Catal. Today, 2000, 58, 1. 52 C. R. A. Catlow, R. A. Jackson and J. M. Thomas, J. Phys. Chem., 1990, 94, 7889.

140 | Catalysis, 2010, 22, 119–143

53 L. Leveles, K. Seshan, J. A. Lercher and L. Leﬀerts, J. Catal., 2003, 218, 306. 54 C. Trionfetti, I. V. Babich, K. Seshan and L. Leﬀerts, Langmuir, 2008, 24, 8220. 55 A. Palmenschikov, G. Morosi, A. Gamba, S. Coluccia, G. Martra and L. G. M. Pettersson, J. Phys. Chem., 2000, B104, 11497. 56 R. Soave and G. Pacchioni, Chem. Phys. Lett., 2000, 320, 345. 57 G. Martra, T. Cacciatori, L. Marchese, J. S. J. Hargreaves, I. M. Mellor, R. W. Joyner and S. Coluccia, Catal. Today, 2001, 70, 121. 58 E. V. Kondratenko and M. Y. Sinev, Appl. Catal., A, 2007, 325, 353. 59 T. Ito, J. X. Wang, C. H. Lin and J. H. Lunsford, J. Am. Chem. Soc., 1985, 107, 5062. 60 A. L. Shluger, R. W. Grimes, C. R. A. Catlow and N. Itoh, J. Phys. Cond. Matter, 1991, 3, 8027. 61 A. M. Ferrari and G. Pacchioni, J. Phys. Chem., 1995, 99, 17110. 62 M. Y. Sinev, V. Y. Bychkov, V. N. Korchak and O. V Krylov, Catal. Today, 1990, 6, 543. 63 M. Y. Sinev, J. Catal., 2003, 216, 468. 64 M. Y. Sinev and V. Y. Bychkov, Kinet. Katal., 1993, 34, 309. 65 V. P. Vislovskiy, T. E. Suleimanov, M. Y. Sinev, Y. P. Tulenin, L. Y. Margolis and V. Cortes Corberan, Catal. Today, 2000, 61, 287. 66 O. V. Buyevskaya and M. Baerns, Catal. Today, 1998, 42, 315. 67 I. Balint and K. Aika, J. Chem. Soc., Farday Trans., 1997, 93(9), 1797. 68 C. Trionfetti, A. Agiral, J. G. E. Gardeniers, L. Leﬀerts and K. Seshan, J. Phys. Chem. C, 2008, 112, 4267. 69 C. Trionfetti, I. V. Babich, K. Seshan and L. Leﬀerts, Applied Catal. A, 2006, 310, 105. 70 H. Kim, Plasma Processes and Polymers, 2004, 1, 91. 71 U. Kogelschatz, Plasma Chemistry and Plasma Processing, 2003, 23(1), 1. 72 A. de Mello and R. Wootton, Lab on a Chip, 2002, 2, 7N. 73 K. F. Jensen, Chemical Engineering Science, 2001, 56, 293. 74 B. Eliasson, C. Liu and U. Kogelschatz, Ind. Eng. Chem. Res., 2000, 39, 1221. 75 S. K. Ajmera, C. Delattre, M. A. Schmidt and K. F. Jensen, J. Catal., 2002, 209, 401. 76 R. M. Tiggelaar, F. Benito-Lopez, D. C. Hermes, H. Rathgen, R. J. M. Egberink, F. Mugele, D. N. Reinhoudt, A. van den Berg, W. Verboom and J. G. E. Gardeniers, Chemical Engineering J., 2007, 131, 163. 77 T. Nozaki, N. Muto, S. Kadio and K. Okazaki, Catal. Today, 2004, 89, 67. 78 U. Kogelschatz, Plasma Physics and Controlled Fusion, 2004, 46, B63. 79 M. Okumoto, H. H. Kim, K. Takashima, S. Katsura and A. Mizuno, IEE Transaction on Industry Applications, 2001, 37(6), 1618. 80 Y. Diao, W. P. Walavender, C. M. Sorensen, K. J. Klabunde and T. Ricker, Chem. Mater., 2002, 14, 362. 81 H. Chen, L. Bednarova, R. S. Besser and W. Y. Lee, Appl. Catalysis A, 2005, 286, 186. 82 A. Agiral, C. Trionfetti, K. Seshan, L. Leﬀerts, J. G. E. Gardeniers, ICPIG Meeting Abs, Czech Republic, July 2007. 83 A. Agiral, C. Trionfetti, K. Seshan, L. Leﬀerts, J. G. E. Gardeniers, ISPC18 Meeting Abs., Japan, August 2007. 84 J. Lugue, D. R. Crosley, 1999 Lifbase: Database and Spectral Simulation Program (ver.1.6) SRI International Report MP 99-009.

Catalysis, 2010, 22, 119–143 | 141

85 L. Leveles, S. Fuchs, K. Seshan, J. A. Lercher and L. Leﬀerts, Appl. Catal., A, 2002, 227, 287. 86 H. S. W. Zhang, J. Xiang, D. J. Driscoll and J. H. Lunsford, J. Catal., 1988, 112(2), 366. 87 D. Wang, M. Xu, C. Shi and J. H. Lunsford, Catal. Lett., 1993, 18, 323. 88 N. Yamauchi, A. Miyoshi, K. Kosaka, M. Koshi and N. J. Matsui, Phys. Chem., A, 1999, 103(15), 2723. 89 M. Y. Sinev, Res. Chem. Intermed., 2006, 32, 205. 90 R. Li, Y. Yamaguchi, S. Yin, Q. Tang and T. Sato, Solid State Ionics, 2004, 72, 235. 91 X. Xu, Thin Solid Films, 2001, 390, 237. 92 J. Pons, E. Moreau and G. J. Touchard, Phys. D: Appl. Phys., 2005, 38, 3635. 93 T. J. Lewis and A. J. Wright, Appl. Phys., 1970, 3, 1329. 94 R. Li, Q. Tang, S. Yin and T. Sato, Fuel Processing Technol., 2006, 87, 617. 95 A. P. Huang, P. K. Chu, H. Yan and M. K. Zhu, J. Vac. Sci. Technol., B, 2005, 23(2), 566. 96 P. C. Cosby, J. Chem. Phys., 1993, 98(12), 9560. 97 T. Nozaki, A. Hattori and K. Okazaki, Catal. Today, 2004, 98, 607. 98 X. Zhang, A. Zhu, X. Li and W. Gong, Catal. Today, 2004, 89, 97. 99 D. J. Driscoll, W. Martir, J. X. Wang and J. H. Lunsford, J. Am. Chem. Soc., 1985, 107, 58. 100 C. L. Bothe-Almquist, R. P. Ettireddy, A. Bobst and P. G. Smirniotis, J. Catal., 2000, 192, 174. 101 X. Lu and M. J. Laroussi, Appl. Phys., 2005, 98, 23301. 102 R. L. Nelson and A. J. Tench, J. Chem. Phys., 1964, 40(9), 2736. 103 M. Sterrer, O. Diwald, E. Knozinger, P. V. Sushko and A. L. Shluger, J. Phys. Chem., B, 2002, 106, 12478. 104 M. C. Wu, C. M. Truong, K. Coulter and D. W. Goodman, J. Vac. Sci. Technol., A, 1993, 11(4), 2174. 105 Y. Yang, Plasma Chemistry and Plasma Processing, 2003, 23(2), 327. 106 X. Zhang, A. Zhu, H. Li and W. Gong, Catal. Today, 2004, 89, 97. 107 S. S. Kim, H. Lee, B. K. Na and H. K. Song, Korean J. Chem. Eng., 2003, 20(5), 869. 108 S. Kado, K. Urasaki, Y. Sekine, K. Fujimoto, T. Nozaki and K. Okazaki, Fuel, 2003, 82. 109 T. Nozaki, Y. Unno, Y. Miyazaki and K. Okazaki, J. Phys. D: Appl. Phys., 2001, 34, 2504. 110 J. Pons, E. Moreau and G. Touchard, J. Phys. D: Appl. Phys., 2005, 38, 3635. 111 M. A. Tas, PhD thesis.University of Twente, 2007 112 D. W. Lee, J. H. Lee, B. H. Chun and K. Y. Lee, Plasma Chem. and Plasma Proc., 2003, 23(3), 519. 113 P. C. Cosby, J. Chem. Phys., 1993, 98(12), 9560. 114 B. Eliasson and U. Kogelschatz, J. Phys. B: At. Mol. Phys., 1986, 19, 1241. 115 T. Nozaki, A. Hattori and K. Okazaki, Catal. Today, 2004, 98, 607. 116 X. Zhang, A. Zhu, X. Li and W. Gong, Catal. Today, 2004, 89, 97. 117 J. H. Lunsford, Catal. Today, 2000, 63, 165. 118 L. F. Albright and K. E. Kranz, Ind. Eng. Chem. Res., 1992, 31(2), 475. 119 J. M. Van de Graaf, M. Zwiep, F. Kapteijn and J. A. Moulijn, Chem. Eng. Sci., 1999, 54(10), 1441. 142 | Catalysis, 2010, 22, 119–143

120 Z. Lysenko, B. R. Maughon, T. Mokhtar-Zadeh and M. L. Tulchinsky, J. Organomet. Chem., 2006, 691(24-25), 5197. 121 T. Koerts, P. A. Leclercq and R. A. Van Santen, J. Amer. Chem. Soc., 1992, 114(18), 7272. 122 G. Boskovic and K. J. Smith, Catal. Today, 1997, 37, 25. 123 W. Keim, Chem. Ing. Tech., 1984, 56, 850. 124 B. Reuben and H. Wittcoﬀ, J. Chem. Edu., 1988, 65(7), 605.

Catalysis, 2010, 22, 119–143 | 143

Asymmetric hydrogenation of activated ketones Jo´zsef L. Margitfalvia and Emı´lia Ta´lasa DOI: 10.1039/9781847559630-00144

In this contribution key features of the asymmetric hydrogenation of activated ketones over cinchona–-Pt catalyst system are reviewed. Both historical backgrounds and recent results are evaluated and discussed. The focus is laid on the peculiarities of these reactions, such as (i) methods and approaches used, (ii) substrate speciﬁcity, (iii) rate acceleration, (iv) the form of conversion-ee dependencies; (v) inversion of ee, (vi) nonlinear phenomena, (vii) origin of enantio-diﬀerentiation and (viii) character of modiﬁer – Pt, substrate-modiﬁer and substrate-modiﬁer-Pt interactions.

1. 1.1

Introduction General information

In order to reduce the environmental and health hazards the demand for optically active compounds in high enantiopurity is increasing in the ﬁeld of pharmaceutical, agrochemical and cosmetic products. Although the most common applications are bio-related, there is also a great interest in the area of materials science for chiral compounds, such as chiral polymers or chiral liquid crystals. For this reason the interest to produce chiral compounds in highly pure form is expanded over the past decades. There are diﬀerent approaches to obtain compounds in high enantiopurity. One of the most environmentally friendly methods is the use of chiral catalysts.1–3 There are diﬀerent approaches in chiral catalysis, such as homogeneous,4 heterogeneous,5 enzymatic,6 and ‘‘artiﬁcial catalytic antibodies’’.7 The common feature of all approaches is that relatively small amount of chiral auxilarities is required to produce chiral compounds in high enantiopurity. In this respect the term ‘‘multiplication of chirality’’8 is often quoted. When heterogeneous catalysts are applied, the chiral auxilarities used are often called as chiral modiﬁers. In the past decades, signiﬁcant progress has been achieved in homogeneous enantioselective catalysis, what is reﬂected by the Nobel Prize awarded in 2001 to Sharpless, Noyori, and Knowles. Variety of transition metal complexes containing unique chiral ligands have been developed to induce enantioselectivity by homogeneous catalysts. These chiral metal complexes were used in various catalytic reactions, such as hydrogenation,9 dihydroxilation,10 epoxidation,11 Diels-Alder reaction,12 C–C bond formation12,13 Michaels reaction,14 etc. These reactions are considered have great importance for the production of ﬁne chemicals and pharmaceutical products.15 a

Institute of Surface Chemistry and Catalysis, Chemical Research Centre, Hungarian Academy of Sciences, POB 17, 1525, Budapest, Hungary

144 | Catalysis, 2010, 22, 144–278 c

The Royal Society of Chemistry 2010

Homogeneous catalytic reactions are highly selective and require relatively small amount of catalyst, however the chiral ligands and the metal complexes used are relatively expensive. Homogeneous catalysts are sensitive both to oxygen and moisture, for this reason the handling of these compounds is quite troublesome. Additional serious problem is the catalyst recovery. As the remaining trace impurities of metals have a deﬁnite environmental and health hazard the development of complex methods for the removal of traces of transition metals increases signiﬁcantly the production costs in homogeneous catalytic reactions. The immobilization of metal complexes into inorganic or organic supports is one of the approaches to overcome the above problems as immobilized catalysts can be separated by ﬁltration or can be used in continuous-ﬂow reactors. Homogeneous catalysts can be immobilized both to inorganic16 and polymeric supports.17 Most of the cases the solid contains a ligand, what is considered as an anchoring site. In case of inorganic supports the reactivity of surface OH groups is used to immobilize a given type of ligand upon using the reactive trimethoxy (or trichloro) silane derivatives.18 However, in this type of immobilized homogeneous catalysts leaching of the metal-complex has often been observed. Among the new strategies to heterogenize transition metal complexes the encapsulation of a chiral metal complex in micropores19 and the use of tethered type metal complexes have to be mentioned.20 The tethered metal complex catalysts were recently developed by Augustine and coworkers.21–23 These catalysts showed high activity and high chemo- and/or enantioselectivity in various hydrogenation reactions. When metal-complexes (both homogeneous and immobilized one) are used the enantio-diﬀerentiation is controlled by the structure of the metal complex. In this respect the molecular character of the catalytic step has to be emphasized, although the exact form or structure of the [metal complex – substrate] adduct is often unknown. In these enantioselective catalytic reactions the chirality and the helicity of ligands has a great importance to control the enantio-diﬀerentiation step.24 This control can be either thermodynamic or kinetic in character. The distinction between these two modes of control is often very diﬃcult. Here we should like to refer to the classical contribution made by Halpern and coworkers.25 The most important class of ligands used in asymmetric reactions has a chiral organic backbone with tertiary phosphino, amino and alcoholic functional groups. Highly eﬀective chiral ligands are often bidentate; i.e. they have two coordinating sites for the coordination of the substrate molecule. In chiral homogeneous catalysis, due to the high substrate speciﬁcity high enantioselectivities can only be achieved for a deﬁned class of substrate molecules, i.e., the right combination of metal and ligands has to be found for each individual catalytic reaction.26 Even though enantioselective homogeneous catalysis is still a relatively young discipline, several enantioselective homogeneous catalytic processes have already been used on an industrial scale.27,28 As concerns the performance (enantioselectivity), the mechanistic insight and general understanding heterogeneous enantioselective catalysis is far Catalysis, 2010, 22, 144–278 | 145

behind to its homogeneous analogue. Only a relatively narrow range of prochiral compounds with CQO and CQC bonds can be hydrogenated with high enantioselectivity. However, due to the above-mentioned drawbacks in homogeneous catalysis, such as separation, reuse, and stability, the interest for heterogeneous asymmetric catalysis increases permanently. The ﬁrst publication related to asymmetric heterogeneous catalytic hydrogenation was published by Schwab in the early thirties.29 In the ﬁrst attempts intrinsically chiral solids, such as quartz has been applied.30 In other approaches chiral biopolymers or natural ﬁbers, such as silk ﬁbroin were used,31 but due to the low optical yield and severe reproducibility problems this approach has been almost completely discarded. The discovery of the Ni/tartrate system for the enantioselective hydrogenation of beta-diketones or their analogs was the ﬁrst real breakthrough in this ﬁeld. This area has been recently reviewed.32–35 Optical yields over 90% were obtained for various substrates using the Ni-NaBr/tartrate system.36 In general this catalytic system requires a pre-modiﬁcation of the parent nickel catalyst with tartaric acid prior to the reaction in order to form chirally modiﬁed sites involved in enantio-diﬀerentiation. Recently, the use of in situ modiﬁcation procedures for Ni/tartrate system has been described.37 In this catalytic system reactive chemisorption of tartaric acid to the nickel surface resulting in some leaching of the nickel to the liquid phase has been evidenced.38 Due to the interaction with tartaric acid the formation of imprinted sites on the nickel surface has been suggested. The enantio-differentiation takes place over this type of chirally modiﬁed sites. Since the discovery of the cinchona-platinum catalyst system by Orito’s group39–43 for the enantioselective hydrogenation of methyl pyruvate (MePy) or ethyl pyruvate (EtPy) the platinum-cinchona system has been successfully applied in the enantioselective hydrogenation of a variety of a-functionalized activated ketones, such as various a-ketoesters, ketopantolactone (KPL), a-ketoamides, a-diketones, a-keto acetals, a-ketoethers, triﬂuoromethyl ketones, and pyrrolidine-2,3,5-triones. Recently the method has been extended to other types of ketones.44 This type of asymmetric hydrogenation reaction is considered as the most intensively studied heterogeneous enantioselective hydrogenation reaction. Since the early eighties great amount of knowledge accumulated on this catalytic system. The topic has been reviewed by diﬀerent authors.15,33,35,45–54 The most important characteristic features of the platinum-cinchona system are as follows: Variety of activated ketones can be used as substrate; Under properly chosen conditions the cinchona alkaloids can induce high enantioselectivity (ee W97%);55,56 When a-keto esters are used the enantio-diﬀerentiation ability is lost if the basic nitrogen at the quinuclidine moiety of the cinchona alkaloid is blocked by alkylation;57 Upon using various substrates the addition of the cinchona alkaloid results in signiﬁcant rate acceleration;58 The amount of modiﬁer required to induce high enantioselectivity is in the range of 1 10 5 M, or in other words the substrate/modiﬁer ratio can 146 | Catalysis, 2010, 22, 144–278

exceed a value above 100.000 (in case of KPL the above ratio was 276.000);59 The replacement of the quinoline ring of the modiﬁer by phenyl or pyridyl resulted in complete loss of ee;60 Inversion of ee is observed under various experimental conditions. In the last twenty years step-by-step progress has been made in the process of understanding the peculiarities of the platinum-cinchona system. This progress covers the following main issues: (i) elucidation of both the rate acceleration and the origin of enantio-diﬀerentiation, (ii) clariﬁcation of the nature of species formed both in liquid phase and on Pt surface by using various spectroscopic methods, (iii) establishment of general and speciﬁc kinetic patterns, and (iv) theoretical calculations and related modeling. As far as several reviews have already been published on the enantioselective hydrogenation of activated ketones in the presence of cinchona-Pt catalyst in this review an attempt was done focusing on (i) methods and approaches used, (ii) recent achievements, and (iii) recalling historical events. Contrary to earlier reviews in this contribution the mechanistic considerations will be treated without any preconception. It means that it will be a priori not accepted that all phenomena involved in the rate acceleration and induction of enantio-diﬀerentiation can be related to events taking place exclusively over Pt surface.61 Consequently, in this review we shall also refer to the general aspects of chiral induction generated by cinchona alkaloids. Possible interactions in the liquid phase will also be discussed. Enantio-diﬀerentiation is a phenomenon characteristic mostly for organic reactions. There are various synthetic methods in organic chemistry to induce chirality. It has to be emphasized that cinchona alkaloids are wellknown chiral compounds used by organic chemists to bring about chiral induction.62 This issue will be brieﬂy discussed in Section 2.1 We shall try to demonstrate that the enantioselective hydrogenation of activated ketones is a very complex reaction. Depending on the conditions of catalyst pretreatment, the accomplishment of the hydrogenation reaction and the type of substrate and modiﬁer used the key interactions responsible for the transformation of the chiral information can takes place both at the Pt surface and in the liquid phase. Methods and approaches used in this area will also be discussed as these issues were not treated in previous reviews. 1.2

Orito’s followers

After Orito’s publications39,40–43 intensive research programs were started by diﬀerent research groups. First two groups in Switzerland, one at Ciba Geigi in Basel under supervision by Dr. H.U. Blaser,63 the other in Zurich at ETH with professor A. Baiker.64 Parallel to that professor P.B. Wells65 in Hulls (Great Britain) started a program related to the use cinchona alkaloids in heterogeneous catalytic asymmetric hydrogenation. Later on other groups in the USA (Dr. D. Blackmond at Merck,66 Professor R. Augustine67 in Seaton Hall), in Hungary (Prof. Tungler68 at Catalysis, 2010, 22, 144–278 | 147

Technical University in Budapest, Prof. J.L. Margitfalvi 69 at the Hungarian Academy of Sciences and Prof. M. Bartok70 at University of Szeged ), in Finnland ( professors T. Salmi and D. Murzin33,71at Abo Academy University) joined to this research area. Today there are around 15–20 independent research groups all around the word that are involved in the investigation of one of the aspects of Orito’s reaction. It is interesting to emphasize that the method developed by Orito was quickly modiﬁed as the ‘‘pre-modiﬁcation’’ approach was replaced by in situ modiﬁcation technique. In this respect the pioneering works were done by the two Swiss groups. Only one group has continued for a while to apply the ‘‘pre-modiﬁcation’’ approach: (the group in Hulls), however today this approach is almost forgotten. Further details about modiﬁcation procedures will be given in Section 4. In the ﬁrst approaches mostly Pt/Al2O3 catalysts and EtPy were used in order to elucidate the general kinetic patterns.58,73 Later on the focus was laid on (i) the elucidation of the reaction mechanisms,61,65,67,74,75 (ii) the use of new substrates,76,77 (iii) the application of new modiﬁers,60,78 (iv) new type of catalysts,79,80 (v) the formation of by-products,81 (vi) the role of impurities,82 (vii) modeling substrate modiﬁer interaction.83–85 Today sophisticated experimental and computational techniques, such as reaction calorimetry,86 the AFM,87 NMR,88 FTIR,89 in situ reﬂection-adsorption infrared spectroscopy (RAIRS),90 attenuated total reﬂection infrared spectroscopy (ATRIR),91 surface-enhanced Raman spectroscopy (SERS),92 circular dichroism,93 electrochemical methods,94 Near-edge X-ray absorption ﬁne structures (NEXAFS)95,96 studies, DFT calculations97,98 are used to get as much as possible information about these unique asymmetric hydrogenation reactions. Orito’s approach later on was extended to other type of activated ketones. The substrates were classiﬁed according to the observed rates and enantioselectivities. This classiﬁcation is given in Fig. 1. Further discussion of substrate speciﬁcity will be given in Section 5.3. From the above discussion it can be concluded that the enantioselective hydrogenation of activated ketones is the most detailed studied asymmetric hydrogenation reaction. However, despite of the extensive studies there are plenty of unanswered questions related to the origin of enantiodiﬀerentiation. High rate – high ee 1

2

3

Medium rate – medium ee 4

Low rate – low ee 11

8

9

12

14 5

6

13

10

15

7

(a)

(b)

(c)

Fig. 1 Classiﬁcation of substrates according to their ability to give high rate and high ee. (Reproduced from ref. 72 with permission.)

148 | Catalysis, 2010, 22, 144–278

1.3

Speciﬁcity of heterogeneous enantioselective catalysis

Heterogeneous catalysis is a complex ﬁeld in physical chemistry. However, the state-of-the-art current knowledge in this area requires additional knowledge in the ﬁeld of materials science, surface chemistry, surface science, surface analysis, computational chemistry, chemical engineering, etc. Heterogeneous enantioselective hydrogenation requires an additional knowledge, i.e. an education in the ﬁeld of organic reactions. As far as enantioselective reactions are a speciﬁc area of organic catalysis the knowledge in this area should also be very speciﬁc. It has to be emphasized that those who joined to this research area have diﬀerent scientiﬁc background and diﬀerent research interest. In most of the cases we are witnessing the prevalence of approaches and views reﬂecting the mode of thinking of a chemical engineer or a surface scientist. Although excellent reviews were written in the last ten years,46–52 those who have a solid background in organic chemistry can realize that both the structure and the conclusions of these reviews reﬂect the view of those who have a ‘‘surface science’’ or ‘‘chemical engineering’’ background. The investigation of heterogeneous catalytic reactions requires a complex knowledge and the use of various experimental techniques. Among these methods reaction kinetics and surface characterization has the greatest importance. Detailed reaction kinetic studies can provide suﬃcient information to suggest or create a proposed reaction mechanism. However, even if reaction kinetics can completely be described by the given system of diﬀerential equations the elucidation of the exact reaction mechanism cannot be guaranteed. This would require the comparison of several possible or potential kinetic equations derived from other possible reaction mechanisms. Unfortunately, such kind of comparison is only seldom is performed in studies related to the elucidation of reaction mechanisms of asymmetric hydrogenation of activated ketones. In the enantioselective hydrogenation of activated ketones the accomplishment of detailed reaction kinetic measurements is strongly hindered by the peculiarities of the reaction, such as (i) the ability of modiﬁers to induce enantio-diﬀerentiation at very low modiﬁer/substrate ratio, (ii) the high reactivity of the substrates resulting in the formation of various by-products, (iii) the loss of alkaloid during the reaction, and (iv) the catalyst poisoning. All these issues will be discussed latter in separate sub-paragraphs. Due to the formation of byproducts, the loss of alkaloids and catalyst poisoning the analysis of the full kinetic curve is almost impossible. Consequently, the use of initial (or maximum) reaction rates cannot provide a proper background for exact kinetic analysis or for the establishment of correct reaction mechanism. The use of in situ calorimetry is one of the most precise methods to obtain direct reaction rates,99 although this method is not common. The preliminary analysis of results obtained by in situ calorimetry indicates that the rate follows the Michaelis-Menten mechanism,100 what is characteristic to enzyme catalytic reactions.101 The fact, that only trace amount of cinchona alkaloid can result in ee values above 90%, strongly indicate that this catalytic system is extremely unique. In this respect the results obtained in the enantioselective Catalysis, 2010, 22, 144–278 | 149

hydrogenation of KPL has to be emphasized, where the substrate/modiﬁer ratio is 276 000:1.59 In other studies related to the enantioselective hydrogenation of EtPy the above ratio is in the rage of 50 000 to 166 000.102 Such a high value of ‘‘chiral ampliﬁcation’’ is characteristic only to enzymes. Cinchona alkaloids, due to their high extent of ‘‘chiral ampliﬁcation’’, their distinct substrate speciﬁcity and their ﬂexible structure, can be considered as a ‘‘mini-enzyme’’.93 This aspect is often forgotten by those working in this area; although in a recent study the similarity to enzymes has already been mentioned.103 The other key approach for the elucidation of reaction mechanism is the investigation of the formation of diﬀerent surface intermediates by diﬀerent surface analytical tools. These approaches will be described in details in Section 6. In this respect the key issue is to answer the following question: ‘‘can the given observed surface entity be involved in the reaction route or does it represent a dead-end in the given reaction scheme?’’ In the former case we talk about ‘‘actors’’, while in the letter case about ‘‘spectators’’. Unfortunately, this kind of questions is only seldom raised. It has to be emphasized that the diﬀerentiation between ‘‘actors’’ and ‘‘spectators’’ is a long dispute in the area of heterogeneous catalysis. Even in gas phase reactions taking place at atmospheric pressure neither the ‘‘in situ’’ nor the ‘‘operando’’ spectroscopy can always provide an unambiguous exact answer about the involvement of a given surface species in the reaction network. When we are dealing with heterogeneous hydrogenation reactions taking place in a three-phase system at high hydrogen pressure the accomplishment of in situ or operando spectroscopy is very diﬃcult. Catalysis scientists like very much to refer to sophisticated spectroscopic data and use these data in favor of their ideas with respect to the reaction mechanism. Even one of the authors of this contribution walked into this catch. In early eighties the direct route for the hydrogenation of acetylene to ethane was conﬁrmed by a sophisticated ‘‘double isotope labeling’’ technique.104 Based on surface science results we proposed that the formed ethylidyne species (RC–CH3) are responsible for the direct route of ethane formation. Of course, we did not detect these species; we just referred to one of recently published LEED results.105 Later on it turned out that the ethylidyne species are so stable that they cannot be removed from the Pd surface by hydrogen at room temperature.106 Of course, several similar misinterpretations can be found in the literature. It will be discussed latter on that in the hydrogenation of activated prochiral ketones the presence of hydrogen at the Pt surface is very crucial. It strongly determines the performance of the catalyst. The hydrogenation of ketones requires relatively large abundance of hydrogen at the Pt surface. Probably, it is the reason that these enantioselective hydrogenation reactions cannot be performed under conditions of transfer hydrogenation. Any disturbance in the amount of available hydrogen can result in signiﬁcant alteration is the performance of Pt-cinchona catalytic system. It is especially notable when the hydrogenation is performed in a continuousﬂow reactor under trickle bed condition.107 However, too much hydrogen at the Pt sites will result in the hydrogenation of the quinoline ring of cinchona alkaloids. This leads to the loss of 150 | Catalysis, 2010, 22, 144–278

ee at low concentration of cinchona alkaloids.108,109 Contrary to that hydrogen ‘‘starving’’ conditions increases the chance for undesired side reactions, such as oligomerization and polymerization. Consequently, if in situ measurements cannot be performed under optimum hydrogen coverage the chance to detect ‘‘spectators’’ is very high. For this reason all spectroscopic data presented so far should be treated with deﬁnite precaution. 2. 2.1

Cinchona alkaloids Characteristic features of cinchona alkaloids

Cinchona alkaloids are used in many ﬁelds of our everyday life. They are widely used in the pharmaceutical and chemical industry. Quinine, derived from the bark of Cinchona ledgeriana Moens ex Trimen is the oldest known natural antimalarial drug. Cinchona alkaloids as easy available chiral agents have great importance both in the academic research and in large scale industrial use. In this respect the classical separation of racemic naproxene can be mentioned.110 An explanation why cinchona alkaloids are universal molecules for so many purposes was given by Wynberg.62 This is shown in Fig. 2. Various parts of the molecule ﬁll the following functions: (a) hydrogen bond formation (interaction with metals); (b) basic amine; (c) bulk aliphatic hydrocarbon moiety; (d) ‘‘handle’’ to modify; (e,f) chiral pocket (epimers available; conformer formation); (g) bulk aromatic hydrocarbon, polarizable, p-p interaction; (h) ‘‘handle’’ to modify; steric and polar inﬂuence. In this section the use of cinchona alkaloids in chiral separation and chiral catalysis will be reviewed very brieﬂy. More detailed reviews can be found elsewhere.62,111,112 2.2

Use of cinchona alkaloids as chiral auxiliaries

2.2.1 Chiral separation. The ﬁrst example of resolution through formation of diastereomeric salts was made by Pasteur113 who used quinicine and cinchonicine, derivatives of quinine and cinchonidine, respectively. Since that time Cinchona alkaloids have been largely employed for the

Fig. 2 Multifunctional nature of quinine as a catalyst. (Reproduced from ref. 62 with permission, (Figure 19))

Catalysis, 2010, 22, 144–278 | 151

separation of various racemic mixtures.114 In the sixties quinine and some other cinchona alkaloids were used to prepare chiral sorbents.115 Alkaloids of this type were covalently bonded to a silica support via their oleﬁnic group. In this way several functional groups in a bulky chiral system provided multiple contact points with the racemate to be resolved. Up to now big variety of separation techniques using cinchona alkaloids has been reported. A two-dimensional liquid chromatography–mass spectrometry (LC–MS) system was developed for the separation of both diastereomers and enantiomers of peptides.116 Generally the presence of electron-deﬁcient aromatic N-acyl constituents and bulky, highly lipophilic side chains enhances enantioselective adsorption, reﬂecting the importance of intermolecular p-donor-acceptor and hydrophobic interaction with the chiral selector.117 Mixed ternary ion associate formation between xanthene dye, cinchona-alkaloid and quaternary ammonium ion has been applied to determinate the trace amount of quaternary ammonium salts in pharmaceuticals by UV spectrophotometry.118 2.2.2 Chiral catalyst. In the ﬁeld of chiral catalysis huge amount of work has been done and cinchona alkaloids have been involved the in the following areas: chiral basic and nucleophilic catalysts in organo-catalytic reactions, chiral ligands coordinating metals like osmium in homogeneous catalytic reactions, phase transfer catalysts in form of quaternary ammonium derivatives, chiral modiﬁers (templates) in heterogeneous catalytic asymmetric reactions. The cinchona alkaloid catalysed reaction of diethylzinc and aldehydes has lead to optically active alcohols having enantiomeric excess up to 92%.119 Cinchona alkaloids have been used both in solute form in liquid phase and in bounded form immobilized into polymer or oxide type supports. In organo-catalysis based on cinchonas large variety of substrates and types of the reactions has been reported. In 1954 Prelog and Wilhelm described the behaviour of diﬀerent cinchona alkaloids and some of their derivatives in the asymmetric cyanhydrin synthesis.120 A review of cinchona alkaloidcatalyzed reactions covering the period prior 1968 was given by Pracejus.121 Cyanohydrin reaction, the Michael reaction, the 1,4-thiol and thiolacetate additions, selenophenol addition reactions, epoxidation of electronpoor oleﬁns, formation of the phosphorus-carbon bond using chiral amine catalysis, 1,2-additions in the presence of cinchonas has been detailed by Wynberg in 1986.62 Highly enantioselective Reformatsky reaction of ketones has been accomplished using cinchona alkaloids as chiral templates.122 Cinchona alkaloid-derived chiral bifunctional thiourea organocatalysts were synthesized and applied in the Michael addition between nitromethane and chalcones with high ee and chemical yield.123 The osmiumtetraoxide catalyzed asymmetric dihydroxilation (AD) is a very important ﬁeld of cinchona utilization (see Fig. 3).10,124 Cinchona alkaloid backbone is ideally suited for providing high ligand acceleration as well as enantioselectivity in AD. It has been found that the 152 | Catalysis, 2010, 22, 144–278

Fig. 3 Role of structural elements of Cinchona ligands in osmium-catalyzed asymmetric dihydroxilation (AD) reactions. (Reproduced from ref. 10 with permission (Figure 4))

enantioselectivity is inﬂuenced mainly by the nature of O9 substituent of the cinchona alkaloid backbone. Three diﬀerent classes of ligands are very eﬀective for the dihydroxylation of almost any oleﬁn (PHAL-,125 PYR-126 and IND-class127). Large scale of substrates (monosubstituted, 1,1-disubstituted, 1,2-disubstituted, trisubstituted even tetrasubstituted oleﬁns, enol ethers, cyclic oleﬁns, amides, enones, sulfur-containing oleﬁnes, protected divinyl ketones, conjugated dienes, trienes etc) can be successfully dihydroxylated.10 Phase transfer catalysis (PTC) based on cinchona alkaloids128–153 is a continuously developing practical method for asymmetric synthesis because these catalysts are very selective. Enantioselective epoxidation of a,b-unsaturated ketones utilizing cinchona alkaloid-derived quaternary ammonium phase-transfer catalysts bearing an N-anthracenylmethyl function gave also appropriate results. Common to all successful applications of cinchona alkaloid derived phase transfer catalysts is that the reaction conditions have to be optimized, consequently structures of substrate, reagent, and catalyst must ‘‘ﬁt together’’: An attempt has been made to understand the role of diﬀerent structure units of cinchona derivatives in PTC. The N-anthracenylmethyl group introduced by Lygo139,143 and Corey141 has been good for an increase in selectivity, although 1-naphthylmethyl was almost as eﬀective.144 Phase transfer catalysts having diaryl substitution at the 3-and 4-positions of the N-benzyl group in cinchonidinium salts were prepared to check how substituted aryl groups aﬀect the asymmetric induction in the benzylation reaction as compared to those having ﬂat linear aryl systems like naphthylmethyl and anthracenylmethyl groups.149 Tremendous amount of work has aimed the preparation and investigation of supported cinchona alkaloids as catalyst.154–170 Polymer bound cinchona alkaloids have been employed for a number of heterogeneous catalytic reactions e.g. asymmetric Michael additions,154–156,171 asymmetric synthesis of a-amino acids,169 enantioselective a-chlorination of acid chlorides,170 asymmetric aminohydroxilation167 asymmetric dihydroxilation of alkenes.160,163 To recycle the alkaloid-OsO4 complex in asymmetric hydroxilation reaction Catalysis, 2010, 22, 144–278 | 153

Kim and Sharpless159 have synthesized four diﬀerent polymers, Pini et al. have examined copolymers of acrylonitrile and substituted quinidine and quinine and reported very low optical purity of diols (up to 45% ee).160 Lohray et al. prepared several copolymers of styrene and 4-phenylstyrene with 10% DHQD-4-vinylbenzoate aﬀording the most eﬀective catalyst for heterogeneous AD.161 A nice example of immobilized quinine used as a catalyst for enantioselective a-chlorination of acid chlorides was given in ref. 170. Large scale of diﬀerent type of polymers has been reported for immobilization of cinchonas.155,157,158,163,165,166 Silica gel supported cinchona alkaloids have been used also as catalysts in asymmetric dihydroxylation and aminohydroxylation.167 Norcinchol supported on silica via ethoxy-silane compound has been applied for enantioselective hydrogenation of a-keto esters with moderate success.47 For diﬀerent purposes diﬀerent cinchona alkaloids are suitable. In methanolysis of diﬀerent tricyclic anhydrides quinidine and quinine has given higher ee than cinchonidine or cinchonine.172 For asymmetric dihydroxylation also quinine (QN) and quinidine (QD) have been found as the most eﬀective ligand.10 In a few cases etheral type cinchona alkaloids173 are also successfully used as chiral template, e.g. b-isocupreidine has been used in the asymmetric Baylis-Hillmans reactions of aromatic imines with 1,1,1,3,3,3-hexaﬂuoroisopropyl acrylate giving (S)-enriched, N-protected-a-methylene-b-amino acid esters. In contrast to the corresponding aldehydes, imines have shown the opposite enantioselectivity.173 Based on the above short review it can be concluded that cinchona alkaloids and their derivatives due to their multifunctional structure and easy availability have been widely used for chiral induction in asymmetric syntheses as well as chiral separations for a century. When new problems in asymmetric techniques have to be solved the application of cinchonas often provides the proper solution again. All of these results clearly indicate that cinchona alkaloids have been used by organic chemists in various areas in order to induce chiral induction or chiral recognition. Consequently, the use of these alkaloids in Orito’s reactions is only one of the opportunities provided by the unique chemical properties of these natural compounds. Any attempt to describe the action of cinchona alkaloids exclusive to surface phenomenon seems to us a mistake. 2.3 Structure of cinchona alkaloids, conformational analysis, and NMR studies 2.3.1 General information. Conformational investigations of cinchona alkaloids based on computational or spectroscopic methods have been made with the aim of understanding of chiral discrimination process.88,174 The most frequently investigated cinchona alkaloids and cinchona derivatives are summarized in Fig. 4. Cinchona alkaloids consist of two rigid cyclic systems, a heteroaromatic quinoline ring and a saturated cyclic quinuclidine ring connected by two carbon-carbon single bonds. They have four asymmetric centers: C3, C4, C8, and C9. However, their conﬁgurations diﬀer from each other only at C9 and C8. 154 | Catalysis, 2010, 22, 144–278

Catalysis, 2010, 22, 144–278 | 155

H OCH3 OCH3

C2H5

C2H3

C2H5

C2H5

dihydrocinchonidine

epiquinine

benzoylquinine

(p-Cl)-benzoyl-

H

C2H3

methoxydihidro-

OCH3

H

Cl

pClBzO

OBz

H

OH

OH

OH

OH

R3

H

H

H

H

H

OH

H

H

H

H

R4

N

OCH3

C2H5

(p-Cl)-benzoyl-

quinidine

methoxydihydro-

C2H5

OCH3

OCH3

C2H5 (dimethylcarbamoyl)dihidroquinidine

OCH3

C2H5

acetildihydroquinidine

dihydroquinidine

OCH3

C2H3 C2H5

OCH3

H

OCH3

H

OCH3

R2

R4 9 R3

8

epidihydroquinidine

C2H5

C2H5

C2H3

C2H3

R1

N

4′

H

epiquinidine

dihidrocinchonine

dihydroquinidine

cinchonine

quinidine

R2

3

OCH3

OCONMe

OAc

PClBzO

H

H

OH

OH

OH

OH

R3

Fig. 4 Structure and conﬁguration of the cinchona alkaloids most frequently investigated (Reproduced from ref. 176 with permission)

cinchonidine

H

C2H3

deoxycinchonidine

OCH3

C2H3

chloroquinine

dihydroquinidine

OCH3

C2H5

OCH3

H

C2H3

dihydroquinine

R2

cinchonidine

8

OCH3

R1

H

N

C2H3

N

4′

9

R3 R4

3

quinine

R2

R1

R1

R4

H

H

H

H

OH

OH

H

H

H

H

2.3.2 Conformational analysis. Crystallographic structure of QN,177 QD,178 CD179 and cinchonine (CN) 180 is well described. In solution however, existence of several other conformers can be supposed. Conformation of quinine and quinidine was a key issue in diﬀerent studies62,120,121,179,180 and the C8-C9 and C4 0 -C9 bonds were considered most important in determining the overall conformation of these compounds. Hiemstra and Wynberg181 have proposed that the most stable conformation of quinine have the largest substituent-the quinuclidine ring-on one side of the quinoline ring, and hydrogen at C8 and the hydroxyl at C9 on the other side. Prelog120 and Meurling180 found this conformation to be the most favorable too. In their pioneer work, Dijkstra and coworkers have combined NMR study and molecular modeling approaches to elucidate the conformational properties of QN and QD.175,176,183 By use of the molecular modeling program CHEMX, the conformational freedom with respect to the C8-C9 and C9-C4 0 bond was investigated.175 Molecular mechanics studies showed that cinchona alkaloids can in principle adopt four diﬀerent conformations: two ‘‘open’’ one in which the quinuclidine nitrogen points away from the quinoline ring and two ‘‘closed ’’ one in which the quinuclidine nitrogen points toward the quinoline ring176(see Fig. 5). One of the calculated conformations of QD (open conformation (3) in Fig. 5) has almost the same geometry as the crystal structure.175 Diﬀerent dihydroquinidine (DHQD)

Fig. 5 The four minimum energy conformations of quinidine (Reproduced from ref. 176 with permission, Figure 2)

156 | Catalysis, 2010, 22, 144–278

derivatives as model substances behave in diﬀerent way. Acetyldihydroquinidine exhibits the closed conformation in CDCl3.183 Hydroxy cinchona alkaloids exist in open conformation (3) (see Fig. 5) at least in 90%, wherein some conformational freedom of the quinuclidine ring exists. Methoxy derivatives predominantly adopt the open conformation (3) and to a lesser amount the closed conformation (2) in CDCl3. However, in CD2C12 the closed conformation (2) is found in excess.183 Hydroxy cinchona alkaloids exist in open conformation (3) (see Fig. 5) at least in 90%, wherein some conformational freedom of the quinuclidine ring exists. Methoxy derivatives predominantly adopt the open conformation (3) and to a lesser amount the closed conformation (2) in CDCl3. However, in CD2C12 the closed conformation (2) is found in excess.183 The combination of variable temperature ( þ 20 to 100 1C) NMR and circular dichroism spectroscopy as well as molecular mechanics computations have shown that in ether solution dihydroquinidine existed in two conformations, the open conformation (3) (anti, open) (80–90%) and closed conformation (1) (syn, closed) (10–20%) separated by a barrier of 8.3 kcal/mol.184 The results of computations favoured to the closed conformation (1) in the gas phase and this discrepancy was explained by preferential solvatation of hydroxy group, which is sterically more available in anti conformation. Bulky substituents on the hydroxy group, such as in the p-chlorobenzoate ester have the same eﬀect.184 The conformation of cinchonidine in solution has been investigated by NMR techniques as well as with theoretical tools.88 Three conformers of cinchonidine (CD) are shown to be substantially populated at room temperature, closed conformation (1), closed conformation (2), and open conformation (3). The latter is the most stable in apolar solvents. The stability of the closed conformers relative to that of open conformer (3), however, increases with solvent polarity. In polar solvents the three conformers have similar energies. The relationship between relative energies and the dielectric constant of the solvent is not linear but resembles the form of an Onsager function.88 In o-dichlorobenzene or dimethyl sulfoxide solution the dihydroquinidine (DHQD) and (p-chlorobenzoyl)dihydroquinidine (p-ClBzDHQD) were found to exist as an equilibrating mixture of two main conformers, see Table 1.174 The relative amounts of these two conformers depend on concentration as well as on solvent and temperature. Changes in the ratio of the two conformers of DHQD can explain the observed changes of the enantioselectivity in the indene rearrangement when the solvent was changed from o-dichlorobenzene to dimethyl sulfoxide. Solute-alkaloid interactions are also able to inﬂuence the conformational behavior.183 In case of ester derivatives the energy diﬀerence between closed and open conformation is less and is probably of the same order of magnitude as the amount of stabilization caused by interactions with solutes, such as methanol or weak acids, or with strong electrophiles, such as osmium tetraoxide. In case of the methoxy derivatives the energy diﬀerence between closed conformation (2) and open conformation (3) has vanished. In nonCatalysis, 2010, 22, 144–278 | 157

Table 1 Populations of open (A) and closed (B) conformers of dihydroquinidine (DHQD) and (p-chlorobenzoyl) dihydroquinidine (p-ClBzDHQD) calculated from JH8 H9.a (Reproduced from ref. 174 with permission, Table 7) Base

Solution (25 1C)

J (A)a

J (A)b

J(obs)

P(A)b

P(B)b

DHQD DHQD DHQD DHQD DHQD DHQD DHQD DHQD p-ClBzDHQD p-ClBzDHQD

CDCl3, 0.2 M CDCl3, 0.02 M THF-d8, 0.02 M o-DCB-d4, 0.005 M o-DCB-d4, 0.02 M dioxane-d8, 0.02 M acetone-d6, 0.02 M DMSO-d8, 0.02 M CDCl3, 0.2 M o-DCB-d4, 0.02 M

2.6 2.6 2.6 2.6 2.6 2.6 2.6 2.6 2.5 2.5

8.29 8.29 8.29 8.29 8.29 8.29 8.29 8.29 8.73 8.73

4.2 5.0 4.2 4.8 5.0 5.0 6.3 7.2 7.4 7.8

0.72 0.58 0.72 0.61 0.58 0.58 0.35 0.19 0.21 0.15

0.28 0.42 0.28 0.39 0.42 0.42 0.65 0.81 0.79 0.85

Based on AM1 structures, J values calculated with substituent corrections by Gandour et al.185 Populations of open (A) and closed (B) conformer.

a b

coordinating solvents like CD2C12, the methoxy derivatives are still predominantly found in the closed conformation (2), but in the presence of any electrophile the equilibrium shifts towards the open conformation (3). Quinine and quinidine (and other hydroxy derivatives) by themselves already possess a distinct preference for the open conformation (3) and thus do not depend on extra stabilization caused by interactions with solute.183 Upon investigation of the above mentioned cinchona alkaloid catalyzed Michael addition of thiols to enones, it was found from the NOESY spectra of QD and QN in the presence of 4-methylbenzenethiol that the alkaloid conformations do not change on formation of an ion pair.175 When cinchona alkaloids are used as chiral bases, the main interaction with the substrate comes from protonation of the tertiary nitrogen in the quinuclidine ring and a subsequent formation of an ion pair between the protonated alkaloid and the deprotonated substrate molecule.183 When the alkaloids are used as chiral ligands, the main interaction is the formation of a dative bond between the tertiary nitrogen of the quinuclidine ring and the metal atom of the substrate molecule (osmium tetraoxide).183 Investigation of the temperature eﬀect has led the authors to come to an important ﬁnding. Low temperature experiments at 20 1C and 60 1C in CDCl3, did not alter the 1H NMR spectra, no line broadening has been observed, and averaged spectra were still recorded at 40 1C. Thus, even at these low temperatures, it was not possible to freeze out diﬀerent conformers. This was an indicative of a fast exchange between the diﬀerent conformations on the NMR time scale and thus of a low energy barrier.176 The syn-anti barrier was estimated ca 8 kcal/mol and the closed-open barrier only half of this.174 The 1H NMR relaxation method was applied to QD. The proposed conformation had the following dihedral angles: o(C11–C10–C3–C4)= 1501 and o(C4 0 –C9–C8–C7)=701. The conformation of side-chain o(C11–C10–C3–C4) was found to be diﬀerent from the one found for crystalline form by X-ray analysis.186 Potential energy surface (PES) for QD has been comprehensively investigated using the molecular mechanics 158 | Catalysis, 2010, 22, 144–278

Fig. 6 NOE interactions in quinuclidine moiety of the cinchona ethereal isomers. (Reproduced from ref. 188 with permission, Scheme 2)

(MM) and quantum mechanical semi-empirical AM1 and PM3 methods. Theoretical results were in agreement with the experimental NMR data, i.e., there are two conformations of the quinidine molecule in solution.187 Structures of etheral and D3,10 isomers of cinchona alkaloids were also determined by NMR and supported with molecular mechanics.188 NOE interactions in quinuclidine moiety of the cinchona ethereal isomers are shown in Fig. 6. Further structural information on cinchona derivatives will be given in Section 6.1. 2.3.3 Solute-solute interaction. Intermolecular interaction of the alkaloid molecules in solution can also be observed. Signiﬁcant diﬀerence between the NMR spectra of optically pure and racemic dihydroquinidines was found under the same conditions (0.35 M in CDCl3). The spectral diﬀerences were greatly reduced when CD3OD was used as solvent. The acetates of optically active and racemic dihydroquinine showed signiﬁcantly smaller diﬀerences than those observed with dihydroquinine. The authors have explained the observations by solute-solute interactions of the enantiomers.189 Osmometry was used to measure average molecular weight for quinine. Results of these experiments have indicated the presence of particles larger than monomeric quinine at 37 1C for a 16 mM solution in toluene. For concentrationso4 mM the quinine was almost completely monomeric.182 The coexistence of monomer and dimers of quinidine was established in quinine-chloroform solutions by investigating the temperature and concentration dependence of the NMR spectral parameters by combination of 2D NOESY and proton-selective relaxation rate measurements. Similar conformation of the alkaloid was found both in the dimer and monomer forms. It was shown that the quinuclidine ring is on one side of the quinoline ring and the CHOH moiety on the other, with the quinoline plane almost bisecting the angle between C-H8 and C-OH190 see Fig. 7. Catalysis, 2010, 22, 144–278 | 159

Fig. 7 Conformation of quinine dimer from NMR results (Reproduced from ref. 190 with permission, Figure 8)

Upon investigation the circular dichroism spectra of cinchona alkaloids, exciton type Cotton eﬀect at 230 nm band of free bases (0.4 mM) was found in CH2Cl2 or dioxane, but not in MeOH. This eﬀect was attributed to the week association of alkaloid molecules in non-polar solvents via N?HO hydrogen bonds.191 3.

Alkaloids used in Oritos’s reaction

Studies aimed at systematic variation of cinchona modiﬁers and their analogs have played a deﬁnite role in building up hypotheses for the mechanism of Orito’s reaction. The structural units of cinchona alkaloids have been discussed in previous section. There are diﬀerent reviews46,192,193 related to the analysis of modiﬁers used in asymmetric hydrogenation of activated ketones. For this reason in this section only the key issues will be brieﬂy mentioned. We shall apply the following classiﬁcation for chiral modiﬁers applied (i) ﬂexible cinchona alkaloids, (ii) ﬂexible cinchona derivatives, (iii) rigid cinchona derivatives, (iv) ﬂexible cinchona analogues, and (v) other type of chiral templates. 3.1

Flexible cinchona alkaloids and their derivatives

Chiral templates most often used in the heterogeneous catalytic asymmetric hydrogenation of activated ketones are natural cinchona alkaloids such as, 160 | Catalysis, 2010, 22, 144–278

Fig. 8 Structure of natural cinchonas used in Orito’s reactions.

CD, CN, QN and QD (see Fig. 8).33,40,51 CD is the most frequently investigated chiral template used in these reactions. Upon hydrogenation of pyruvates QN and CD (C8(S), C9(R)) result in (R)-lactate while QD and CN (C8(R), C9(S)) give (S)-lactate.33,40,194,195 In general CD is a better modiﬁer than CN. This diﬀerence is more pronounced in ethanol than in toluene, but in AcOH the diﬀerence is negligible. With the exception of epicinchona alkaloids196 and isocinchonines197 it is a general observation, that the conﬁguration of C8 or C8 and C9 atoms of the cinchona alkaloid determines the product distribution.57,192,198 Already in one of the ﬁrst studies it was evidenced that tertiary N in the quinuclidine moiety of cinchonas plays crucial role194,198,199 although in recent studies it was shown that in case of ketopantolactone200 even N-alkylated derivatives of CD can induce very slight enantioselection. Surprisingly the N-oxide derivative of CD has also resulted in enantioselection. A possible reason is that N-oxide can be reduced very fast under the reaction conditions and than acts like 10,11-dihydrocinchonidine (DHCD)57 which is the most easily forming derivative of CD.63 Not only the vinyl group of cinchona alkaloids can be hydrogenated, but its quinoline ring. This is an undesired side reaction leading to the substantial loss of enantioselectivity.57,195,199 It has been suggested that the decrease in the ee values upon using CD derivatives with partially hydrogenated quinoline be attributed to a weaker adsorption of the alkaloid to the Pt surface.192 The phenomenon can also be explained by the loss of the shielding eﬀect of the aromatic p-system required for chemical shielding via p–p interaction83 (see Section 8.3). In a detailed study18 diﬀerent cinchona analogs and 8 diﬀerent substrates were investigated. The results indicated that no ‘‘best’’ chiral template exists Catalysis, 2010, 22, 144–278 | 161

for all substrates.201 This ﬁnding indicates that interactions between the substrate and the chiral modiﬁer template depend on various factors.201 C9 substituted compounds represent an important group of ﬂexible cinchona alkaloids. 9–O-methyl-10,11-dihydrocinchonidine (MeODHCD) the most frequently used C9 derivative generally behaves slightly better than CD in the hydrogenation of a-ketoesters.194,199,202,203 Similar positive results were obtained upon using other substrates.44,199,204 However diketones, such as 1-phenyl-1,2-propanedione produces lower ee in the presence of MeODHCD than in the presence of CD.205 Detailed studies on the use of these alkaloid derivatives can be found elsewhere.202,206 With respect to the use of O-substituted derivatives the inversion of ee has to be mentioned. These results will be discussed in Section 5.5.4. The inversion of ee in the case of bulky O-substituted derivatives of CD relative to DHCD has been attributed to a tilted mode of adsorption of these chiral templates to the Pt surface206 (see Sections 6 and 8). 3.2

Rigid cinchona derivatives

C8(S) C9(R) type of cinchona alkaloids, such as CN, QD and cupreidine can form inner ether derivatives. These derivatives are called ‘‘rigid’’ as in these alkaloids the rotation around the bond C8–C9 as an axis is not possible. These alkaloids were used to demonstrate that the formation of closed conformation of alkaloids is not a prerequisite for the formation of substrate-modiﬁer complex suggested by the ‘‘shielding eﬀect’’ model.83 3.3

Flexible cinchona analogues

Synthetic analogues of alkaloids have all of the key elements of cinchona alkaloids, such as aromatic ring, chiral moiety, and basic nitrogen. Important feature of these new analogues is the presence of an aromatic group in the close neighbourhood of the stereogenic center. Diﬀerent types of enantiomerically pure primary and secondary aminoalcohols78 and amines have been tested as chiral templates in the hydrogenation of pyruvate esters,207,208 ketopantolactone,209 triﬂuoromethyl ketones,210 1,1,1-triﬂuoro-2,4-diketones, etc.44 3.3.1 Aminoalcohols. Series of enantiomerically pure 2-hydroxy-2-arylethylamines (see Fig. 9) has been prepared from the corresponding oleﬁns.78 Upon using compound A in the hydrogenation of EtPy ee values higher than 75% was achieved.207,208 The replacement of the naphtyl ring by an anthracenyl one resulted in further increase of ee the up to 87%.60,211 However 1-(9-triptycenyl)-2-(1-pyrrolidinyl) ethanol resulted in signiﬁcant decrease in both ee (o5%)60 and reaction rate. 3.3.2 Amines. Upon using (R)-1-(1-naphthyl) ethylamine as chiral template in the asymmetric hydrogenation of EtPy 82% ee has been achieved in AcOH. It has been shown that (R)-1-(1-naphthyl) ethylamine is only a precursor of the actual chiral template, which is a secondary amine (aminoester) formed in situ from (R)-1-(1-naphthyl) ethylamine and EtPy by condensation to the corresponding imine and subsequent reduction of 162 | Catalysis, 2010, 22, 144–278

Fig. 9 Preparation of enantiomerically pure 2-hydroxy-2-arylethylamines. (Reproduced from ref. 78 with permission)

the CQN bond.46,199,212 The conﬁguration at the stereogenic center a to the ester group has no eﬀect on the enantioselectivity.199 Substituent at the amino group of naphthylethylamine can inﬂuence the enantiodiﬀerentiation ability; in general, more bulky substituent at the N atom is detrimental to enantioselectivity in the hydrogenation reaction of EtPy.199 Further details can be found elsewhere.210,212 3.4

Other type of chiral templates

Other natural alkaloids and their derivatives were also applied as chiral templates in the asymmetric hydrogenation of activated ketones although the ee values obtained were much lower than over CD and its derivatives. Blaser and coworkers tested about 100 diﬀerent chiral auxiliaries, but they never found any meaningful enantioselectivity.213 Ephedrine has given low or moderate optical yields in the hydrogenation of a-ketoesters.214 Codeine, strychnine, and brucine have provided only 2–12% ee.215 Using triﬂuoromethylcyclohexyl ketone substrate brucine has not resulted in optical yield.216 ( )-Dihydro-apovincaminic acid ethyl ester has also been applied as chiral template for EtPy substrate (27–30% ee).68,217–219 Other vinca derivatives have also been tested but ( )-dihydro-apovincaminic acid ethyl ester has been found to be the most eﬀective one.68,220,221 Upon using other compounds as chiral templates in the hydrogenation of EtPy (S)-a,adiphenyl-2-pyrrolidinemethanol222 resulted in moderate ee (max 25%) depending on the type of the solvent. (S)-proline chiral auxilary has also been tested.223 During the hydrogenation of EtPy in the presence of (S)-proline resulted in the formation of N-alkylated proline while in case of isophorone substrate a diastereoselective oxazolidine type intermediate was formed in a condensation reaction. Hydrogenation reaction itself proved to be Catalysis, 2010, 22, 144–278 | 163

diastereoselective. Isophorone has produced ee up to 60% but (R)-ethyl lactate has been formed in very low optical purity (1–5% ee).223 (S)-proline derivatives, such as Z-(S)-proline 2-naphthyl ester, Z-(S)-proline 2-(2naphthyl)-ethyl ester, Z-(S)-proline 3-ethyl-indole ester and (Z)-(S)-proline3-ethyl-indolamide, (S)-proline-2-naphthylamide hydrochloride has also been tested as chiral templates of new type in case of EtPy.224 a,a-DiphenylL-prolinol chiral template resulted in 14% (S) in the hydrogenation of triﬂuoromethylcyclohexyl ketone.216 Dextrocarbinol base has induced no enantioselectivity in the same reaction.216 ‘‘Tro¨ger’s base’’ ((5R,11S)-( þ )2,8-dimethyl-6H,12H-5,11methanodibenzo[d,f][1,5]diazocin) as a chiral templates has given 65% ee using acetic acide solvent in the asymmetric hydrogenation of EtPy.225 (R)-( )-2-phenylglycinol has induced poor enantioselectivity in the hydrogenation of 1,1,1-triﬂuoro-2,4-diketones.44,217 Other compounds tested in the hydrogenation of EtPy as ((R)-( þ )-N-(amethylbenzyl) phtalic acid monoamide, (R)-( )-1-1-naphthyl) ethylisocyanate has given moderate ee (23%, 59% respectively).225 It has been shown that 1-naphthyl-1,2-ethanediol226 is an eﬀective chiral modiﬁer in the hydrogenation of KPL and ethyl-4,4,4-triﬂuoroacetoacetate. It is the ﬁrst eﬀective nonamine-type chiral template used in Orito’s reactions. 4. 4.1

Methods and approaches used General information

In this section methods and approaches applied in the enantioselective hydrogenation of activated ketones will be described. One of the characteristic features of this catalytic system is the need for catalyst pretreatment in hydrogen at 350–400 1C prior to the reaction. The omitting of pre-reduction step resulted in low rates and low enantioselectivities. The need for catalyst pretreatment has already been described by Orito’s group.40 In a later study it has been shown that the modiﬁcation of the Pt surface by the alkaloid requires pure Pt sites.63 Recently a new type of Pt/Al2O3 catalyst has been developed by Degussa (catASium F214) which can be used without pre-reduction.227 This catalyst gives high rates and high ee values when it is used as received. Some authors claimed that the aerobic treatment of the catalyst, i.e. the formation of chemisorbed oxygen on the Pt sites, is needed to improve both the reaction rate and the ee values.65,228,229 This issue will be discussed in Section 4.3. The use of ultrasound resulted in also an improved performance of supported Pt catalyst.70,230 The other important issue is the mode of introduction of the modiﬁer. In Orito’s approach pre-modiﬁcation has been used.40 The discovery of in situ modiﬁcation was the next important ﬁnding.63 Upon using in situ modiﬁcation the ‘‘ligand acceleration’’ phenomenon has been discovered.58 However, it has to be mentioned that rate acceleration (RA) was not observed for all substrates and all modiﬁers investigated. Based on this fact recently same groups questioned the validity of the rate acceleration phenomenon.232–234 This issue will be discussed in Section 5.5.1. 164 | Catalysis, 2010, 22, 144–278

Most of the authors are calculating either the initial rate or the maximum rate observed after a short induction period. Unfortunately, due to side reactions and the transformation of the alkaloid during the hydrogenation reaction (see Section 5.1) the analysis of whole kinetic curve is very troublesome, although there were attempts to do that.100 We should like to emphasize that in order to get a full picture about the peculiarities of these unique reactions reliable data with respect to both the reaction rates and the optical yields should be provided. One of the most serious problems is that in large number of recent publications the rates are not given at all.210,235–237 Only conversion or yields values measured at the end of the reaction are compared. There were couples of papers devoted to the problems of reproducibility and variation of the initial rates measured under identical conditions.47,61 It has been shown that initial rates depend not only on the purity, but the origin of the substrate as well as on the batch number.69,93 Systematic study of this eﬀect was done by the Ciba group. These results are shown in Table 2.195 Of course, the purity of the reactant and solvent has a great impact on the validity of kinetic results, consequently results obtained upon using unpuriﬁed substrates, especially ketoesters, has to be treated with great concern. Researchers with suﬃcient background in organic chemistry realized very early that the puriﬁcation of the substrates before the use is a must. The use of puriﬁed modiﬁer is also very important.238 There were strong disputes in the literature concerning to the use of unpuriﬁed or contaminated substrates.66,69,82,228 The lack of background in organic chemistry often resulted in strange and unreliable results. For instance, the reaction mechanism was investigated by diﬀerent groups in alcoholic solvent, despite the fact that the most investigated substrates, i.e., EtPy or MePy, react with linear alcohols in a side reaction catalyzed by tertiary amines. The

Table 2 Eﬀect of substrate origin and quality on initial rate (100 mol/g catalyst/min) and ee values in the hydrogenation of EtPy in the presence of CD under diﬀerent conditions (catalyst, solvent, pressure in bar). Bold numbers show the highest and lowest rate or ee values. (Reproduced from ref. 194 with permission) Undistilled J, T, 20 Origin

rate

Fluka91 3 Fluka92 4 Lancaster 7 ICN, Ohio 9 Sigma 9 Jansen 18 R.de Haen 5 Aldrich 50 TCI, 50

Distilled J, T, 20

E, T, 20

J, Ac, 20

E, Et, 20

J,T,100

Average

ee

rate ee

rate

ee

rate

ee

rate

ee

rate ee

Rate ee

63 78 71 77 76 80 73 84 82

12 44 14 15 24 24 30 70 48

10 50 14 05 20 18 21 36 62

69 80 77 79 80 83 79 85 83

12 64 26 38 46 46 46 76 78

83 88 87 89 87 90 87 91 90

25 90 36 40 40 50 50 70 68

74 74 77 78 78 80 78 84 70

56 96 48 64 114 102 148 132 164

20 58 24 30 42 43 50 72 78

69 80 78 79 80 83 81 85 83

82 83 85 86 87 89 87 90 80

73 81 79 81 81 84 81 87 81

J=JMC, E=Engelhard, T=toluene, E=ethanol, Ac=AcOH, TCI=TCI, Tokyo

Catalysis, 2010, 22, 144–278 | 165

formation semi-ketals will be discussed in Section 5.1. Despite all disputes and argues even these days it is possible to ﬁnd papers, where no words is said how the substrate was puriﬁed or what even is worst, there are publications where unpuriﬁed substrate has been used.233,239 These facts often resulted in misinterpretation of experimental results. These issues will be described in Sections 5.5.1 and 5.5.2. In all pioneering studies, i.e., in early nineties, the determination of optical yield was not an easy task, especially at low conversion. For this reason, the changes in the ee values with conversion were not investigated, consequently the anomalous monotonic increase type ee-conversion dependencies were not discussed till 1995 (see details in Section 5.5.2.). In the last 10 years sophisticated physical or physical-chemical methods, such as STM, ATRIR, AFS, Raman spectroscopy, etc. has been used in order to elucidate the reaction mechanism or the origin of RA and enantiodiﬀerentiation (ED) (see Section 6.6). The common problem related to these studies is that conditions of these measurements are far away from those used in real hydrogenation reaction, although some measurement methods, such as (ATR-IR) were performed under condition close to hydrogenation.240 A typical misused situation is when the chemisorption of CD was investigated by electrochemical methods in concentrated H2SO4. Based on these results it was concluded that the adsorption of CD on Pt (111) is irreversible.241 The problem with these results is that those who need some additional proof with respect to ‘‘surface induced’’ RA and ED like these results and refer to these false ﬁndings quite often. There is one more problem what can be formulated in the following way: How to distinguish between surface species what are involved in the catalytic step from those, what are formed on the surface of platinum, but are not involved in the catalytic act? The latter species are often called as ‘‘spectators’’ in a given catalytic reaction. In many cases the surface concentration of ‘‘spectators’’ can be much higher than that of the ‘‘actors’’. In this respect let us remind the reader for the classical problem in homogeneous catalysis discussed by Halpern.225 In his classical study it was demonstrated that in homogeneous catalytic enantioselective hydrogenation not the most stable [substrate-catalyst] complex is involved in the ED step, but the less stable one, what reacts with hydrogen much faster than the former. In connection to the above discussion the use of sophisticated surface techniques for the elucidation of the origin of ED has to be mentioned. None of these methods can fully guarantee that the observed surface species is really involved in the given step of enantioselective hydrogenation. Consequently, it is almost impossible to distinguish, whether an identiﬁed surface entity is an ‘‘actor’’ or just a ‘‘spectator’’. 4.2

Catalysts applied

4.2.1 Supported metal catalysts. In the enantioselective hydrogenation of activated ketones supported Pt is the most commonly used catalyst. Pt/C catalysts have been used by Orito in his original approach. Alumina supported Pt catalysts containing around 5 wt% metal are the most commonly 166 | Catalysis, 2010, 22, 144–278

used catalyst. Two industrial catalysts, E 4759 from Engelhard and JMC 94 from Johnson Matthey, have been widely used by diﬀerent research groups. The Pt dispersion of these catalysts is in the range of 0.2–0.3.57 E4759 has rather small pores and a low pore volume, while JMC 94 is a wide-pore catalyst with a large pore volume. There are reports related to the use of Pt colloids both as prepared79,100,242–245 or stabilized on a support.246 The use of other supported noble metals, such as Ir,247–249 Ru250,251 and Rh252–256 is considered as a curiosity, although recent results using rhodium is very promising.257 Supported iridium catalysts were used in the enantioselective hydrogenation of diketones in order to suppress the hydrogenation of the second carbonyl group.258 Palladium is not a suitable metal for the hydrogenation of keto carbonyl group. Pt supported on HNaAY259 and ZSM-5 zeolites,260 MCM-41261 mesoporous materials, clays262 and ion exchanges resins250 were also tested in the enantioselective hydrogenation of EtPy, however, the performance of these catalysts was lower than that of the alumina or silica supported Pt. It is interesting to mention that most of the Pt/C catalysts resulted in low ee values (below 35%) and very moderate reaction rates.263 There are reports on the use of carbon nanotubes as support.264 We consider that all high surface area materials are ineﬃcient supports for this reaction, due to their high adsorption power resulting in high modiﬁer concentration at the support and lowering the modiﬁer concentration in the liquid phase. In earlier studies it has been suggested that Pt dispersion has a decisive inﬂuence on both the activity and ee and it was suggested that in order to obtain high optical yields the dispersion should be r0.2.63 It has been suggested that an appropriate ﬂat Pt surface be required to accommodate the modiﬁer or the modiﬁer-substrate complex in order to get pronounced ED.265 Contrary to the above results and suggestions results upon using a Pt/SiO2 catalyst (EUROPT-1) relatively high ee values were also obtained, although the dispersion of Pt in this catalyst is around 0.6–0.7.228 Further results on Pt nanocolloids prepared,79,243–246 indicated also that there is no real need to have large ﬂat Pt surface to get high ee values.

4.2.2 Pt colloids. The common feature of Pt colloids is that they are stabilized by nitrogen and oxygen containing ligands. Under properly chosen experimental condition these Pt colloids show high activity and relatively high enantioselectivity.246 Pt colloids were also used in kinetic investigations.100 It was demonstrated that the RA could also be observed when Pt colloids were used. In this respect Pt colloids stabilized by cinchona alkaloids have the greatest interest. The concept of using chiral stabilizing agent for the preparation of Pt colloids has been applied by Bo¨nnemann.79 These colloids were used to hydrogenate EtPy. Upon using DHCD or CD as stabilizing agent the mean size of Pt colloids was in the range of 1.5–2.8 nm. It is interesting to note that upon using these colloids in the hydrogenation of EtPy ee values in the range of 75–80% were obtained. In a recent study Pt nanocolloids stabilized by cinchona alkaloids were used in enantioselective Catalysis, 2010, 22, 144–278 | 167

hydrogenation of EtPy in as received form or immobilized on various supports.266 It is interesting to mention that colloids prepared by Bo¨nnemann’s method required addition of free alkaloid to induce high rate and high ee values,. The addition of alkaloid to the solution increased both the reaction rate and ee values in the range of 80–85% were obtained in the AcOH þ MeOH mixture.244 The results indicated that two forms of the alkaloids can be distinguished: (i) the stabilizing form ((CD)st or (CN)st), and (ii) the excess form ((CD)ex or (CN)ex), i.e., the amount of alkaloids added into the liquid phase. Under condition of enantioselective hydrogenation these two forms are in dynamic equilibrium. The Pt colloid prepared upon using cinchonine (PtCN) was used to investigate the possible exchange between the two forms of the alkaloid. These results are presented in Fig. 10. In the ﬁrst experiment the PtCN colloid was used and the concentration of (CN)ex was 6.8 10 3 M. In this experiment the ee was independent of the conversion and leveled oﬀ at ee= 0.6. In the second experiment instead of (CN)ex (CD)ex was added and its concentration was also 6.8 10 3 M. The initial ee values (ee= 0.6) show that at low conversions the initial (CN)st form is involved in the events controlling the asymmetric induction. As the reaction proceeded the (CN)st form was exchanged by (CD)ex resulting in a decrease in the ee values. The ﬁnal ee value (ee=0) indicates that the above exchange is almost quantitative. This result indicates that there is an exchange between the two forms of the alkaloid. In the third experiment the (CD)ex was added prior to the treatment with ultrasound. In this experiment

1.0

0.0

0.2

0.4

0.6

0.8

1.0

0.6

ee

0.2

-0.2

-0.6

-1.0 Conversion Fig. 10 The ee – conversion dependencies obtained in the presence of PtCN varying the – experiment in the presence of (CN)ex; D – character of excess alkaloid ((CN)ex or (CD)ex). experiment in the presence of (CD)ex without ultrasound treatment, E, B – experiment in the presence of (CD)ex after treatment with ultrasound. The concentration of excess alkaloids is 6.8 10 3 M; nanocolloid=0.020 g [EtPy]0=0.6 M, pH2=5 bar, solvent: CH3COOH/MeOH (5/1), Tr=12 1C. (Reproduced from 244 with permission)

7

168 | Catalysis, 2010, 22, 144–278

the ee value was constant but was opposite in sign, i.e., during the ultrasonic treatment full exchange between the two forms of the alkaloid ((CN)st and (CD)ex) took place (see Fig. 10). The phenomenon appeared to be fully reproducible. According to computer modeling the above Pt nanocolloids have particle size in the range of 1.6–2.8 nm, i.e., the size of accessible Pt (111) surface is very small (3 4 or 5 5 Pt atoms). This small Pt colloid can accommodate the ‘‘shielded’’ modiﬁer-substrate complex, while the accommodation of the open modiﬁer-substrate complex would require much larger surface sites. 4.2.3 Characteristic features of supported Pt catalysts used. The results discussed so far indicated there is no need to have a preferred particle size as high ee values were obtained over catalyst having broad Pt dispersion range. However, in active and enantioselective catalysts the Pt sites should be relatively clean. Both the pre-treatment in hydrogen at 300–400 1C and the treatment in ultrasound can provide clean Pt surface. Another important issue is that the catalyst used has to be relatively inert related to in the hydrogenation of the quinoline ring of the alkaloid. Conditions for ring hydrogenation of modiﬁers were investigated in various studies.267–271 Although it has been shown that upon using Al2O3 support in AcOH Al oxonium ions and their adducts with the alkaloid has been detected272 the involvement of these species in the catalytic reaction is quite doubtful. The suggested ‘‘electrostatic acceleration’’273 needs further experimental proof. As a rule the support should be relatively inert. Highly acidic supports can induce acid catalyzed undesired side reactions. Both the high acidity and the high surface area of supports decrease the amount of alkaloid available for ED. It was shown that Cl containing alumina precursor and chlorine-containing platinum salts exhibit signiﬁcantly higher optical yield than similar catalysts prepared from chlorine free starting-materials.265 It has also been demonstrated that the modiﬁcation of alumina support by alkoxy-silanes decrease both the rate and ee values274 (see Section 5.6.4). 4.3

Catalyst pretreatment

In the ﬁrst publication by Orito’s group Pt/C catalyst was applied and the beneﬁcial eﬀect of preheating the catalyst in hydrogen at 300–400 1C prior to the modiﬁcation was emphasized.40,41 The selection of a proper pretreatment procedure for supported Pt catalysts is one of the basic issues. Several other pretreatment methods were applied and diﬀerent explanations were given for the favourable eﬀect of reductive, aerobic and ultrasonic treatments. Fig. 11 shows the general scheme for catalysts pretreatment.76 The common feature is the reduction of the catalyst used at relatively high temperature (300–400 1C). It is called reductive treatment. The catalyst can be cooled either in hydrogen or in an inert atmosphere. In oxidative treatment after the reductive treatment the catalyst is treated in air and cooled down in an inert atmosphere. Most of the authors agree that upon using supported Pt catalyst a reductive treatment is a must and special care has to be done to prevent contamination of reduced catalyst with Catalysis, 2010, 22, 144–278 | 169

Fig. 11 A general scheme for catalyst pretreatment. (Reproduced from ref. 76 with permission)

oxygen. However, there are groups using pre-reduced catalysts kept or stored in air.275 4.3.1 Prereduction in hydrogen. In one of the ﬁrst publications using in situ modiﬁcation of Pt/Al2O3 catalyst63,276 it was mentioned that the thermal treatment at 400 1C hydrogen has pronounced eﬀect both on the activity and the enantioselectivity. After thermal treatments in hydrogen at 400 1C, 15–20% higher ee values were obtained than over untreated catalysts. In a recent review195 the role of pretreatment was formulated as follows: (i) pretreatment cleans up the surface of the catalyst by removing oxygen as well as impurities; (ii) residual Pt salts are converted to metallic Pt; (iii) the average particle size of Pt increases, (iv) the morphology of Pt particles, i.e. the distribution of exposed face, edge and corner atoms is also altered favourably; (v) it promotes adsorbate-induced surface restructuring. Restructuring during pretreatment of Pt/alumina catalyst used in enantioselective hydrogenation of KPL was also studied.277 The inﬂuence of reductive and oxidative heat treatment on the enantioselectivity of chirally modiﬁed Pt/alumina has been reinvestigated. Enhancement in ee by 39–49% has been observed after treatment in hydrogen at 250–600 1C, as compared to untreated or pre-oxidized catalysts. The changes in ee after reductive and oxidative treatments are reversible, and always the ﬁnal treatment is decisive. A HRTEM study indicates that adsorbate-induced restructuring of Pt crystallites during hydrogen treatment at elevated temperature can play a role in the selectivity improvement, but the changes are superimposed by the strong structure-directing eﬀect of the alumina support. 170 | Catalysis, 2010, 22, 144–278

4.3.2 Inﬂuence of oxygen. The eﬀect of oxygen on the performance of cinchona – Pt catalyst system has been studied by diﬀerent groups under various conditions.65,229 In these studies diﬀerent solvents, diﬀerent type of supported Pt catalysts and diﬀerent experimental conditions were used. For this reason it is very diﬃcult to make any right conclusion or interpretation related to the given observation; is a particular ﬁnding a general phenomenon or an experimental artifact? In a recent review33 a generalized comment was given, namely the eﬀect of modiﬁcation atmosphere of Pt-CD catalysts aﬀects on the ee values. For instance under air higher ee values has been achieved in EtPy hydrogenation, whereas under anaerobic condition ee decreases drastically.229,278 In ref. 229 it was demonstrated that the addition of oxygen during the enantioselective hydrogenation of EtPy has a positive eﬀect both on the rate and the ee values. The observed eﬀect was attributed to restructuring of the surface of Pt in the presence of oxygen. There is only one remark with respect to these ﬁndings, i.e., the ﬁnal ee value (below 40%) is extremely low for the experimental conditions applied. When anaerobic and aerobic treatments of Pt/SiO2 catalyst was compared after anaerobic treatment decreased enantioselectivity and greatly reduced activity was observed using DHCD as modiﬁer in the hydrogenation of MePy.65 It has to be mentioned that this pretreatment was performed in ethanol. In further studies it was shown that during this aerobic treatment ethanol was oxidized over platinum catalyst into acetic acid279 and probably the formed AcOH was responsible for the increased performance. Similar prove has been obtained in our laboratory.281 Table 3 shows the results obtained in the enantioselective hydrogenation of triﬂuoroacetophenone (TFAP).76 These results clearly show that either treatment in an oxygen atmosphere or stirring in air resulted in a decrease in the enantioselectivity. Barto´k and coworkers have applied a reductive treatment prior to the use of catalyst, but the catalyst is stored in air before its ﬁnal use. It was shown that the increase of the storage time up to one week has no pronounced eﬀect on the performance of the catalyst.277 With respect to the role of oxygen it was also suggested that during this treatment PtO could be formed. During the hydrogenation reaction PtO is reduced to metallic Pt and water. It is not excluded that the presence of small amount of water can result in some improvement in the performance.76 In the enantioselective hydrogenation of MePy or butane-2,3-dione over Pt in the presence of CD the coadsorption of oxygen with the alkaloid resulted in a positive eﬀect Table 3 Inﬂuence of catalyst pretreatment in the hydrogenation of TFAP (90 mg 5 w% Pt/alumina, 1.28 g TFAP, 10 bar, 20 1C (Reproduced from ref. 76 with permission) No

Pretreatment

Solvent

(CD) mg

Time (min)

Conv. (%)

ee (%)

1 2 3 4 5

– reductive A reductive B reductive A oxidative

toluene toluene toluene 1,2-dichlorobenzene 1,2-dichlorobenzene

2 2 2 4 4

63 50 86 105 105

89 96 95 95 91

16 33 33 45 29

Catalysis, 2010, 22, 144–278 | 171

both on the rate and ee.278 It was suggested that the presence of a strong coadsorbate, such as oxygen, the surface was not poisoned by CD. In a recent study the promoting eﬀect of helium treatment was also mentioned.281 However, it was admitted that the eﬀect is due to the small oxygen impurity in the helium used.

4.3.3 Use of ultrasound and microwave heating. The eﬀect of ultrasound radiation was investigated in details by Barto´k’s group.55,70,282–284 The method appeared to be highly eﬃcient as the ultrasound radiation resulted in and increase both in the reaction rate and the enantioselectivity. Table 4 shows representative results using three diﬀerent substrates and three modiﬁers.55 The decrease in metal particle size was given as an explanation of improved performance.282 In another study TFAP was applied as a substrate and the use of sonication resulted in positive eﬀect.273 The results indicated also that both the frequency of ultrasound and the duration of sonication have a strong inﬂuence of the enantioselectivity.55 However, it was also pointed out that the presence of the modiﬁer is absolutely crucial during sonication. It was proposed that the ultrasonic irradiation created a more eﬀective surface modiﬁcation, resulting in the formation of surface sites required for optimum enantio-diﬀerentiation. Besides it an additional positive eﬀect of oxygen was also observed.273 It was suggested that ultrasonic irradiation helps the removal of the impurities from the Pt surface.195

Table 4 Sonochemical and silent enantioselective hydrogenation of a-ketoesters over 5% Pt/Al2O3 in acetic acid using diﬀerent cinchona modiﬁers under 10 bar hydrogen pressure. (Reproduced from ref. 55. with permission) Optical yield (ee %) Entry

Substrate

Modiﬁer

Catalyst

Major product

‘‘silent’’

‘‘sonication’’

1 2 3 4 5 6 7 8 6

1 1 1 2 2 2 3 3 3

4 5 6 4 5 6 4 5 6

E40665 E40665 E40665 E4759 E4759 E4759 E4759 E4759 E4759

R S R R S R R S R

85 78 93 88 34 60 79 50 83

97 83 98 92 57 68 92 92 96

Substrates: 1=EtPy; 2=PhGl, 3=Phenylethyl; Modiﬁers: 4=CD, 5=CN; 6=O-Me-CD.

172 | Catalysis, 2010, 22, 144–278

In enantioselective hydrogenation of 1-phenyl-1,2-propanedione over 5 wt% Pt/SF (silica ﬁber) catalyst a notable enhancement of reaction rate, ee and rs was observed under ultrasound compared to silent conditions.231 In mesitylene solvent four-fold increase in the reaction rate was observed under ultrasound compared to identical silent conditions, while in methyl acetate and in toluene the rate enhancement was only minor. Upon using Pt/SF catalyst it was suggested that surface smoothening and cleaning take place under ultrasound irradiation. However, no signiﬁcant diﬀerences in Pt particle size distribution between sonic and silent treated catalysts were observed by TEM. Summing up results related to the use of ultrasound its positive eﬀect has been ascribed to the following facts:55 (i) ‘‘through the decrease in metal particle size, the platinum dispersion becomes close to optimal using a catalyst of large metal particle size,’’ (ii) ‘‘the surface density of the modiﬁer increases as a result of insonation, providing more chiral sites for the hydrogenation and, in parallel, suppresses the background reaction, i.e. racemic hydrogenation’’. Enantioselective and racemic hydrogenation of EtPy over Pt/Al2O3 catalyst was investigated under microwave dielectric and conventional heating.285 A homemade laboratory microwave loop reactor was applied allowing diﬀerentiating between dielectric and conventional heating. The eﬀects of polar and non-polar solvents on enantioselective hydrogenation of EtPy were studied in toluene and ethyl alcohol. In case of toluene, which is microwave transparent, no signiﬁcant diﬀerences in the reaction rate and enantioselectivity were observed between dielectric and conventional heating. In case of EtOH, the reaction rate remained unaﬀected. However, the ee dramatically decreased from 60 to 40% under microwave heating. No signiﬁcant improvement of the reaction rate with an increasing microwave power input was observed. The authors suggested that this observation caused by the local superheating of the polar EtOH in the cavity, which is not possible in the non-polar toluene. Our explanation is diﬀerent; namely the decreased ee is due to the formation of semi-ketal from the substrate and the alcohol used as solvent.

4.4

Premodiﬁcation of the catalyst with the alkaloid

As it has already been discussed earlier that in Orito’s pioneering studies a premodiﬁcation procedure was used to introduce the chiral modiﬁer into the Pt/C catalyst pretreated in hydrogen at 400 1C.40 It has to be emphasized that the premodiﬁcation was performed at higher temperature than that of the hydrogenation reaction. During premodiﬁcation the catalyst and the cinchona alkaloid has been stirred in a given solvent for a relatively long period (24 hours). This premodiﬁcation procedure strongly resembled the modiﬁcation process used for Ni/tartaric acid catalysts developed earlier.286 The premodiﬁcation was followed by ﬁltration and mild washing and the premodiﬁed catalyst was introduced into the reactor containing the solvent and the substrate.

Catalysis, 2010, 22, 144–278 | 173

The Orito’s original approach was followed by P.B. Wells’ group.65,228,278 However, there were several drawbacks in the use of this method. Further results indicated if the ﬁltration and the washing is not carefully performed solvated CD can be left in the pores of the catalyst support, for this reason, as it was pointed out in one of the comments,69 the exact amount of alkaloid introduced was not really known. Probably it was the reason that in ref. 287 the authors admitted the ‘‘erratic variation of the initial rates’’. After successful introduction of the in situ methods57 and demonstration of the advantage of premixing technique64 the premodiﬁcation method was almost entirely forgotten. This approach is still used when the enantioselective hydrogenation is investigated in the gas phase.288 4.5

In situ modiﬁcation

4.5.1 General Information. In situ modiﬁcation of the catalysts prior to the hydrogenation of activated ketones can experimentally be performed in diﬀerent way. These various approaches provides diﬀerent surface coverage at t=0, i.e., prior to the start of the hydrogenation reaction. Let us consider that the Pt catalyst is pre-reduced in hydrogen around 400 1C and kept in an inert atmosphere prior to its use. The catalyst in this form is introduced into the reactor containing the following components: (i) solvent, substrate and modiﬁer (premixing method); (ii) solvent, substrate (injection of the modiﬁer); (iii) solvent, modiﬁer (injection of the substrate). It is easy to propose that the above three methods shall provide completely diﬀerent surface coverage at t=0, what can have diﬀerent inﬂuence both on the kinetics and ED. 4.5.2 Premixing technique. This method has been applied ﬁrst by H.U. Blaser’s group.63 In this technique all components of the reaction are premixed prior to the hydrogenation reaction. This method provides high coverage of substrate and a relatively low coverage of modiﬁer at the Pt site. This new approach resulted in the discovery of the ‘‘ligand acceleration’’ phenomena.58 The rate increase was very pronounced and was proportional to the amount of alkaloid used. This phenomenon will be discussed in Section 5.5.1. During premixing the substrate can decompose resulting in carbon monoxide, which is considered as a strong catalyst poison. The substrate interacts also with the modiﬁer and induces the formation of high-molecular weight byproducts (see Section 5.1). These by-products have a negative inﬂuence on the initial rate by their poisoning eﬀects. If the amount of hydrogen in the overall hydrogen pool is high (i.e., when the catalyst is cooled in the hydrogen atmosphere from the temperature of re-reduction) in this case partial hydrogenation of both the cinchona alkaloid and the substrate can also take place prior to the introduction of hydrogen. Consequently, upon using the premixing technique a new problem appeared, i.e. the reproducibility of the reaction rate. It was observed that the rate of reaction depended on the duration of premixing, i.e. the time required to close the high-pressure autoclave, purge the reactor with nitrogen and hydrogen, and pressurize the reactor and start stirring.64 Reliable and reproducible rate could only be obtained when the duration of premixing 174 | Catalysis, 2010, 22, 144–278

was kept constant.64 This problem is even more pronounced when alcohols have been used as solvents. 4.5.3 Injection technique Injection of the modiﬁer. This method excludes all undesired interactions between the substrate and modiﬁer. It provides high coverage of the substrate at the Pt site. However, if the catalyst is not cooled in an inert atmosphere racemic hydrogenation of the substrate can also take place before starting the reaction (i.e. before t=0). Neither the spontaneous oligomerization/condensation of the substrate or its decomposition into CO can be excluded in this case. Injection of the substrate. This method excludes also all undesired interactions between the substrate and modiﬁer. It provides a deﬁnite surface concentration of the modiﬁer. No oligomers or condensation products exist at t=0. However, partial hydrogenation of the modiﬁer cannot be excluded if the catalyst has been cooled in a hydrogen atmosphere. The injection method gives an opportunity to get reliable kinetic data provided a proper ratio between the batch and the injected volumes is chosen. In this way the inﬂuence of some of the undesired side reaction can be eliminated, providing more chance to get intrinsic kinetic data. In this approach either the modiﬁer or the substrate is injected by high-pressure hydrogen.69,93,195,289 4.6

Hydrogenation of a-keto esters in continuous-ﬂow reactors

It is obvious that the separation, handling and reuse of the heterogeneous catalysts become very eﬃcient when the ﬁxed-bed reactor is used; consequently it is very promising to introduce ﬁxed bed reactors with the aim to industrialize the asymmetric catalysis. However, up to the late nineties there were only scarce data related to the use of continuous-ﬂow reactors in asymmetric hydrogenation reactions. As far as only trace amount of modiﬁer is required to induce high ee an attempt was done to use a continuous ﬁxed-bed reactor for the enantioselective hydrogenation a KPL.290 This approach resulted in signiﬁcant process intensiﬁcation; consequently upon using a small tubular reactor (size of a pencil) more than 14 kg (R)-pantolactone per hour could be produced. Later on the approach was extended to use for other substrates.291–293 High reaction rates and high ee values were obtained by continuous feeding of minute amounts of chiral modiﬁer to the reactant stream. The ee values for KPL and EtPy without optimization was 83.4 and 89.9%, respectively. Transient measurements by stopping of the ﬂow of CD indicate that continuous feeding of the modiﬁer in ppm concentration is crucial. There was a short induction period prior reaching stable high ee values. Knitted Pt/SiO2 was used in enantioselective hydrogenation of 1-phenyl1,2-propanedione giving relatively high enantiomeric excesses.294 The knitted silica ﬁber catalyst gave encouraging results in the continuous ﬁxed bed operation with enantiomeric excesses comparable to those obtained in the batch reactor. EtPy was hydrogenated in a continuous-ﬂow ﬁxed-bed reactor and high ee value up to 89% was obtained at modiﬁer/substrate molar ratio of only Catalysis, 2010, 22, 144–278 | 175

307 ppm.295 In another study296 the enantioselective hydrogenation of ethyl2-oxo-4-phenylbutyrate (EOPB) on Pt/g-Al2O3 catalyst in the presence of CD was investigated in a ﬁxed-bed reactor with the aim to synthesize enantiomerically pure (R)-( þ )-EHPB, a building block for the synthesis of several commercially important A.C.E. inhibitors. The highest ee value around 69% was obtained in toluene at 6 MPa hydrogen pressure. Although stable conversion values were obtained in time on stream experiments, the ee values decreased in time. The transformation of isopropyl-4,4,4-triﬂuoroacetoacetate to the corresponding hydroxyester was studied in a ﬁxed bed reactor over Pt/Al2O3 in the presence of MeO-CD and triﬂuoroacetic acid.297 Around 0 1C and upon varying the pressure, the total liquid ﬂow rate and the feed composition ee values up to around 90% were achieved. However, the time on stream pattern showed quite notable activity decrease. Enantioselective hydrogenation of EtPy was performed in a continuousﬂow ﬁxed-bed reactor using supercritical carbon dioxide and supercritical ethane (scC2H6).299 In the latter solvent much higher catalytic activity was observed. Ethyl benzoylformate (EBF) was hydrogenated in a continuous-ﬂow ﬁxed-bed reactor over Pt/Al2O3 catalyst in the presence of CD and CN.299 Variety of chemical and physico-chemical methods was applied to pretreat or clean the chiral ﬁxed bed between multiple hydrogenation reactions. It was observed that after an enantioselective hydrogenation with CD as modiﬁer at 0 1C, the continuous-ﬂow reactor could be eﬀectively cleaned at 0 1C, and that a racemic unmodiﬁed hydrogenation could be performed thereafter. This implies the eﬀective desorption of chiral species from the surface during cleaning. Contrary to that cleaning of the reactor at 50 1C resulted in a reproducible unmodiﬁed enantioselective hydrogenation, with a marked inversion of enantiomeric excess. The inversion of ee will be discussed in Section 5.5.4. In a recent study enantioselective hydrogenation of EtPy was also performed in continuous-ﬂow reactor in the presence of CD over Pt/Al2O3.300 All these results indicate that the use of continuous-ﬂow reactors should be applicable to diﬀerent substrates based on the use of chiral modiﬁers and supported metal hydrogenation catalysts. This approach provides more eﬃcient screening method for potential chiral modiﬁers establishing the basis for future technical applications. Based on the use of continuous-ﬂow reactors the so called ‘‘chiral switch’’ methodology has been developed for the investigation of the relative adsorption strength or the competition of chiral modiﬁers on a metal surface.238,301 4.7

Reuse and deactivation of catalysts

The reuse of the catalyst has a great practical signiﬁcance. The reuse is strongly connected to the catalyst deactivation phenomena. Deactivation will also be discussed in Section 5.2. Due to catalyst deactivation for the reuse of supported Pt catalysts fresh modiﬁer has to be added before each hydrogenation cycle,302,303 or the modiﬁer is fed permanently in continuous manner108,290 to ensure good activity and high enantioselectivity. 176 | Catalysis, 2010, 22, 144–278

Typical setup for these experiments is as follows: after the ﬁrst reaction ‘‘the mixing was stopped, the reaction mixture was left to settle for 30 min, the liquid was removed, and solvent, EtPy and occasionally modiﬁer were added to the reactor. Hydrogenation was repeated as described above’’.268 An intelligent approach has been developed for the reuse of hydrogenation catalyst.304 Magnetic Pt/SiO2/Fe3O4 catalyst was prepared and successfully applied for the enantioselective hydrogenation of various activated ketones. This catalyst modiﬁed with CD showed catalytic performance (activity, enantioselectivity) in toluene comparable to the best-known Pt/ alumina catalyst. The new catalyst can be easily separated by an external magnetic ﬁeld and recycled several times with almost complete retention of activity and enantioselectivity. Figs. 12A and B show two types of reuse experiments. In Fig. 12A fresh modiﬁer has been added in each run, consequently the ee values are constant; however there is a signiﬁcant catalyst deactivation. However, as shown in Fig. 12B both the activity and ee decrease on reuse of the catalyst if no fresh modiﬁer is added to the reaction mixture at the beginning of each new run. It is known from other studies246,248,261 that almost constant ee values can be achieved in reuse experiments, where ‘‘fresh’’ modiﬁer is added in every reuse. Interesting observation was described in ref. 305. Stopping the enantioselective hydrogenation of EtPy at a conversion of approximately 70%, long-term stability of the catalyst can be achieved. During 10 cycles of hydrogenation, the activity and enantioselectivity of the repeatedly used catalyst remain constant at high values even without adding fresh modiﬁer at the beginning of each new run. These observations indicate that the loss of modiﬁer takes place only at high conversion, i.e., the presence of excess of substrate prevents the hydrogenation of the quinoline ring of the modiﬁer. These results are shown in Fig. 13.

Fig. 12 Repeated use of catalyst for the enantioselective hydrogenation of EtPy. A: 5 ml of toluene, [DHCD]=0.1 mmol/l, fresh DHCD was added each time to the reaction solution. B: 5 ml of toluene, [DHCD]=0.01 mmol/l, no DHCD was added for reuse of catalyst. (Reproduced from ref. 268 with permission)

Catalysis, 2010, 22, 144–278 | 177

Fig. 13 Activity and ee values of the modiﬁed catalyst used repeatedly for the stopping after complete conversion; right: stopping at 70% conversion) without addition of fresh modiﬁer. (Reproduced from ref. 305 with permission)

Diﬀerent reuse methods and inﬂuence of the solvent were investigated by Barto´k et al.268 In the course of enantioselective hydrogenation of EtPy in toluene, ee increased with catalyst reuse. The increase was in the range of 10–20%, however this increase was not observed in acetic acid. The authors considered that the phenomenon of ‘‘increase in ee on reuse’’ is an intrinsic feature of the catalyst system used, i.e. new chiral centers making higher ee possible are formed. It was suggested that during reuse due to the intensive interaction of a solid surface and the chiral modiﬁer reconstruction of the Pt surface takes place. This is in a good agreement with recent ﬁndings306 that during reuse due to the intensive interaction of a solid surface and the chiral modiﬁer reconstruction of the Pt surface takes place. The authors believe that ‘‘the surface atoms of the catalyst are continuously reorganized during the reaction as a result of adsorption/chemisorption steps. In this respect it was supposed that the presence of trace amounts of oxygen might also play an important role in the reaction studied. The lack of increase in ee on reuse in acetic acid may indicate that in this solvent the reaction mechanism is diﬀerent. The authors most important conclusions with respect to the reuse of catalyst are as follows: ‘‘(i) Pt/Al2O3 catalysts with a Pt-dispersion of 0.2–0.3 and a mean Pt particle size of 3–5 nm are the best; (ii) prior to use, the catalyst should be prereduced at 673K for 1–1.5 h in hydrogen ﬂow; (iii) it is necessary to add fresh modiﬁer for each reuse of the catalyst’’.268

5.

Speciﬁcity of Orito’s reaction

As it has already been discussed that enantioselective hydrogenation of activated ketones has several speciﬁcities. The main speciﬁcities are as follows: (i) side reactions, (ii) catalyst deactivation, (iii) solvent eﬀect, (iv) substrate speciﬁcity, (v) rate acceleration (enhancement), (vi) enantioselectivity–conversion (time) dependencies, (vii) non-linear phenomenon, and (viii) inversion of enantioselectivity. In order to understand all peculiarities of these unique reactions the speciﬁcities have to be discussed separately. 178 | Catalysis, 2010, 22, 144–278

5.1

Side reactions

The intrinsic reactivity of activated ketones is high. In this respect diﬀerent type of side reactions, such as (i) semi-ketal formation, (ii) oligomerization (condensation, polymerization), (iii) hydrolysis, (iv) transesteriﬁcation, (v) deuterium exchange, and (vi) decarbonylation, can be distinguished. Side reactions have a great inﬂuence both on the rate and enantioselectivity. The occurrence of side reactions depends on the mode of introduction of the reaction components. The by-product formation can take place both in racemic and enantioselective hydrogenation of activated ketones. In most of the side reactions the substrates are involved. With respect to the modiﬁer the hydrogenation of the quinoline ring has to be mentioned.271 The ﬁrst thorough information related to the side reactions and byproducts formation was given in ref. 307 The formation of semi-ketals can take place between the substrate and alcoholic solvent,279 the substrate and the reaction product,233 the substrate and CD.109 It has been demonstrated in ref. 279 that CD catalyses the formation of semi-ketals. The extent of semi-ketal formation increases in the following order: t-BuOHoi-PrOHo EtOHoMeOH. Semi-ketals can be hydrogenolized to the corresponding alcohols in a racemic reaction, i.e. this side reaction strongly decreases to overall ee of the enantioselective hydrogenation. Oligomerization and polymerization reactions have been discussed by diﬀerent authors.109,308,309 In these reactions both Pt and alumina sites can be involved. A dimer of the substrate is formed in the condensation reaction of the enol and keto forms of EtPy.233 The keto-enol transformation of MePy has been recently investigated by using thermal programmed desorption (TPD), STM and reﬂectance adsorbance infrared spectroscopy (RAIRS).310 It was shown that MePy undergoes CH bond scission at room temperature on clean Pt(111) leading to surface mediated enol formation and assembly into H-bonded superstructure. The latter was severely inhibited by addition of hydrogen (10 6 Torr). STM data show no evidence for an irreversible polymerization reaction. In a recent study311 it was suggested that base-acid sites on the g-Al2O3 surface are responsible for the aldol reaction of EtPy to yield b-hydroxyl ketone, which is subsequently dehydrated to generate CQC containing species (see Scheme 1). The formed condensation product can be involved in cyclization reactions as shown in Scheme 2. These cyclic products were considered as one of the key compounds poisoning the catalyst during the hydrogenation of EtPy.81 In ref. 311 side reactions with the involvement of Pt sites were also investigated. Scheme 3 shows these reactions and the role of hydrogen in their suppression. The fact that the aldol condensation of EtPy on Al2O3 can be suppressed by adsorbed acetic acid may be interpreted as that the acetic acid adsorbs and blocks some basic sites on alumina.309 An additional set of side reactions was discussed in ref. 312. It was suggested that these reactions might take place over the Pt sites resulting in strong poisoning eﬀect.312 Several adducts originating from the base-catalysed EtPy condensation were detected by ESI-MS method.313 Decarbonylation of both linear (EtPy, MePy)82,91 and cyclic a-ketoesters (KPL)314 was evidenced by using Catalysis, 2010, 22, 144–278 | 179

Scheme 1 Proposed mechanism for aldol condensation of EtPy catalyzed by the base-acid sites on the g-Al2O3 surface. (Reproduced from ref. 311 with permission)

O

O O HO

O HO

O

O O

O

OH

O

1b

1a -EtOH

O

O O O O O

O OH O

2a

O

2b

Scheme 2 Further transformation of adduct formed in the condensation reaction of the keto and the enol forms of EtPy. (Reproduced from ref. 81 with permission)

ATR-FTIR method. It has been shown that both hydrogen and CD suppresses the decomposition of EtPy taking place on Pt surface.91 The stabilization of MePy against decomposition has been suppressed by benzene315 and 1-(1-naphthyl)ethylamine.316 180 | Catalysis, 2010, 22, 144–278

Scheme 3 Proposed mechanism for the side reactions of EtPy on Pt/Al2O3 inhibited by H2. (Reproduced from ref. 311 with permission).

In a recent study it was shown that although adsorbed CD suppresses the decomposition of EtPy it cannot suppress condensation and hydrolysis of EtPy on g-alumina. Coexistence of CD and hydrogen is needed to suppress all side reaction of EtPy over Pt/g-Al2O3.311 Recently it has been suggested that the polymerization of EtPy takes place preferentially at steps sites of Pt and CD or other tertiary amines inhibits propagation of polymerization over Pt sites.233 It is known that most of the products of the above side reactions are considered as catalyst poison, consequently their presence signiﬁcantly alter the intrinsic kinetic patterns of these reactions. All these facts strongly indicate that experimental conditions have to be strictly standardized in order to minimize the eﬀect of by-products formed. The role of by-products to the rate acceleration phenomena will be discussed in Section 5.5.1. 5.2

Catalyst deactivation

In the enantioselective hydrogenation of activated ketones catalyst deactivation takes place both in batch and continuous-ﬂow reactors. The deactivation can be attributed to both of chemical and physical processes. Catalysis, 2010, 22, 144–278 | 181

These processes can be explained as follows: (i) poisoning, (ii) coking, (iii) sintering, (iv) restructuring and (v) phase transformation. Deactivation is inevitable, but it can be retarded or prevented and some of its consequences can be avoided with clever process design.231 One of the main reasons for catalyst deactivation can be ascribed to the formation of by-products as described in the previous section. Catalyst deactivation cannot be directly investigated in a batch reactor; however the analysis of the time on stream behavior in continuous-ﬂow reactors provides useful information with respect to catalyst deactivation. It has been shown that that irreversible deactivation of Pt catalysts by condensation and oligomerization polymerization can be related to hydrogen starvation.312 In a recent study311 it was found that only the coexistence of CD and H2 could thoroughly inhibit the side reactions of EtPy on Pt/g-Al2O3. The decrease of the catalytic activity due to the use of unpuriﬁed substrates has been discussed in diﬀerent studies.66,69,82,317 Consequently, puriﬁcation of the reactant seems to be a necessary prerequisite to avoid catalyst deactivation.318 With respect to catalyst deactivation the dispute related to the origin of rate acceleration has to be mentioned.234,295 Further discussion of this dispute will be given in Section 5.5.1. Catalyst deactivation has been observed in kinetic experiments performed in batch reactor.233,289 Deactivation was also evidenced in continuous-ﬂow regime using various experimental designs.291 It was also observed on heating the platinum catalyst in either hydrogen or helium at 350 1C for two hours and then using it without exposure to oxygen.229 General principles of catalyst deactivation both in batch and continuous ﬂow reactors were given in ref. 319. In this work enantioselective hydrogenation of 1-phenyl-1,2-propanedione was studied. An elegant way to promote catalyst durability, activity and selectivity is to apply on-line acoustic irradiation during the course of reaction.320 Ultrasound can retard catalyst deactivation by (catalyst) surface cleaning and exposing fresh, highly active surface as well as by the reduction of diﬀusion length in the catalyst pores by alteration of the surface of catalyst. Furthermore, strongly absorbed organic impurities that block active sites can also be removed by soniﬁcation. In ref. 231 the deactivation was studied both under conventional and microwave heating using EtPy as a substrate. No catalyst deactivation was observed in three consecutive experiments with re-used catalyst. Previously, it has been reported107,321 that during continuous hydrogenation of EtPy and 1-phenyl-1,2-propanedione; a notable catalyst deactivation takes place. Based on these results it can be concluded that in the enantioselective hydrogenation of activated ketones deactivation is an inevitable phenomenon. However, there are measures to decrease the extent of deactivation. These measures are as follows: (i) application of acetic acid as solvent, which deactivates the amine modiﬁer and the alumina support for aldol reaction;295 (ii) decreasing the modiﬁer/substrate ratio to reduce the rate of side reactions in solution; (iii) working at high surface hydrogen concentrations, that is, at high hydrogen pressure and in the absence of mass transport limitation; and 182 | Catalysis, 2010, 22, 144–278

(iv) carefully avoiding the contact of Pt with pyruvate in the absence of modiﬁer at low surface hydrogen concentrations, and (v) minimizing the contact of the substrate and the modiﬁer before entering them into the reactor. 5.3

Substrate speciﬁcity

It is known that various prochiral ketones can enantioselectively be hydrogenated over Pt-cinchona catalyst system. This issue has been discussed in various earlier reviews.33,46,52,195,322 Characteristic feature of these substrates is the presence of an activating group with strong electronwithdrawing properties (ester, carbonyl, acetal, amido and a,a,a-triﬂuoro group) in the a-position to the prochiral keto group.76 As most of the substrates applied so far were extensively discussed in earlier reviews in this section we shall give only a brief summary. The ﬁrst substrates successfully hydrogenated asymmetrically were linear and cyclic a-ketoesters,48,84,282,290,323 a-diketones,318,324–327 a-ketoacetals,328,329 aromatic a,a,a-triﬂuoro-ketones,284,330–333 and linear and cyclic a-ketoamides.336–338 The structure of these activated ketones is given in Fig. 14. Later on the studies were extended to non-aromatic a,a,a-triﬂuoroketones236 and substituted aromatic a,a,a-triﬂuoroketones,53,333,337 triﬂuoro substituted a, b-ketones,44,338 b-ketoesters.339 Pt-cinchona system has also been used in the hydrogenation of substituted deactivated aromatic ketones, such as 3,5-bis(triﬂuoromethyl) acetophenone.340 The types of substrates containing activated keto group and the highest ee values in the presence of optimum modiﬁer are shown in Table 5.50 Most of the substrates, with the exception of the amido derivatives resulted in ee values above 80%. Additional results related to the inﬂuence of substituents in a-keto esters were published in refs. 338,341. In these studies both R1 and R2 substituents (see Scheme 4) were systematically altered. The Barto´k’s group found that in AcOH when R1 was methyl, iso-propyl, terc.-butyl, phenyl and phenylethyl or the R2 group was methyl, ethyl and isopropyl the sense of the enantio-diﬀerentiation was not altered. The increase of the size of R1 and R2 resulted in slight decrease in the reaction rate and ee values. Additional results presented by Baiker’s group are summarized in Table 6.338 In this series a of experiments the variation of the bulkiness both at the keto and ester sides in nine diﬀerent a-ketoesters was investigated upon using CD and QN, as chiral modiﬁers. In toluene in the presence of CD good to high ee values (eemax=94%) were achieved. Consequently, in the presence of CD the enantio- diﬀerentiation is controlled by the ester group, notwithstanding of the steric bulkiness or electronic structure of the

O O O

O

O

O NH O

O

O

O O

O

R

CF3

O

Fig. 14 Structure of ﬁrst substrates successfully hydrogenated asymmetrically.

Catalysis, 2010, 22, 144–278 | 183

184 | Catalysis, 2010, 22, 144–278

7

6

5

4

3

2

1

No

O

N

O H

O

O

O butanedione

O

O

O

O

O

O Ethyl benzoylformate

O

Substrates

OH

Toluene EtOAc

CD

AcOH

Toluene

EtOH/H2O

CD

CD

CD

MeOCD

AcOH þ Toluene

AcOH

QD

CD

Solvent

Modiﬁer

143

40

300

296000

350

868

1540

Substrate-Modiﬁer ratioa

66

2100

160

1040

440

300

1640

Substrate-Pt ratiob

Table 5 Hydrogenation of various activated ketones over Pt/Al2O3-cinchona catalysts. (Reproduced from ref. 50 with permission)

94

63

58

e

91d

82

98

346

326

334

323

345

56

344

c

98

Ref.

ee (%)

Catalysis, 2010, 22, 144–278 | 185

CF3

O

O

O

O

CF3

O

O

OR

O

N

OR O

O

O

OEt

CD

MeOCD

MeOCD CD

DHCD

CD

Toluene-TFA

THF-TFA

AcOH AcOH

Toluene

Toluene

290

290

1050 130

720

320

180

180

1320 180

710

210

92

96f

97 97

86

91

a substrate/modiﬁer molar ratio. b substrate/Pt molar ratio. c reaction mixture (without substrate) ultrasonic treatment. d 8 1C. e kinetic resolution. cinchonidine, DHCD: 10,11-dihydrocinchonidine, MeOHCD: methoxy-HCD, MeOCD: methoxy-CD, THF: tetrahydrofurane, TFA: triﬂuoroacetic acid.

12

11

10

9

8

f

20 1C, CD:

332

50

330 328

347

335

Scheme 4 R1 and R2 substituents systematically altered in the hydrogenation of a-keto esters. (Reproduced from refs. 338 and 341 with permission)

alkyl and functionalized aryl group on the other side of the keto group. Other studies reveal also that ester, carboxyl, amido, carbonyl, acetal, and triﬂuoromethyl functions have similar directing eﬀects. Results presented in Table 6 show that structurally more demanding substrates show signiﬁcant diﬀerences between toluene and acetic acid. However, none of the mechanistic models developed for the enantioselective hydrogenation of activated ketones over Pt-cinchona catalyst system can explain the following anomalies found in Table 6: (i) higher ee values in toluene in the presence of CD for substrates (2), (4), (5) and (9); (ii) the strong increase of ee in acetic acid in the presence of QN for substrate (1); (iii) almost complete loss of ee in AcOH in the presence of QN for substrates (3) and (9). In the presence of CD the greatest drop both in the conversion and ee values was measured for the hydrogenation of 9 (see Table 6) when changing from toluene to AcOH. This observation was attributed by the authors ‘‘to the strong, competing adsorption of AcOH’’. It seems to us that it is a very plausible explanation. We consider this observation as an artifact, as in our independent experiment no similar observation was found.342 There are additional controversial data with respect to the use of AcOH. In ref. 338 based on a relatively early study343 it has been mentioned that the enantioselectivities in toluene and acetic acid are similar.53 This statement is not really correct as in various other studies73 it has been shown that in acetic acid both the rate and ee values are higher than in toluene. It has already been shown that the addition of a small amount of acetic acid either to toluene or ethanol has a very pronounced eﬀect.84 Derivatives of triﬂuoroacetophenone with diﬀerent substituents at the aromatic ring (CF3, N(Me)2 and Me) were hydrogenated over Pt/Al2O3 in the presence of CD, CD HCl and 9–O-methyl-CD.332 It was shown that electron-withdrawing substituents increased and electron-releasing one decreased the rate and enantioselectivity in these reactions, although steric eﬀects (with m- or p-substituents) were also substantial. Cinchona alkaloids were also used in the asymmetric hydrogenation of non-activated ketones. In this case the enantioselectivity is rather moderate as it was emphasized in a recent review.195 Upon using various nonactivated triﬂuoromethyl ketones, such as methyl-, adamantyl, and terc-butyl235 in the presence of CD low ee values were obtained (eemax=44% for adamantyl derivatives). Positive eﬀect of TFA was also demonstrated, while the use of AcOH as a solvent resulted in low yields and low ee values. In propanol inversion of the ee was observed. These results conﬁrmed again that in the hydrogenation of non-activated ketones high ee values couldn’t be expected. Unfortunately, the initial rates were not determined in this study. The authors claimed that their result ‘‘indicates that enantioselectivity is guided by the triﬂuoromethyl 186 | Catalysis, 2010, 22, 144–278

Table 6 Enantioselective hydrogenation of various a-ketoesters in toluene and acetic acid (Reproduced from ref. 438 with permission) Substrate

Reaction in toluene

Reaction in AcOH

CD ee(%) [conv.]

QN ee(%) [conv.]

CD ee(%) [conv.]

QN ee(%) [conv.]

80 (R) [100] 56 (R) [91]

21 (R) [100] 12 (R) [61]

88 (R) [100] 27 (R) [99]

92 (R) [100] 2 (R) [99]

86 (R) [100]

79 (R) [100]

86 (R) [100]

4 (R) [100]

95 (R) [94]

89 (R) [43]

70 (R) [73]

24 (R) [64]

–CH2CH3

92 (R) [100]

75 (R) [197]

80 (R) [100]

31 (R) [100]

–CH2CH3

87 (R) [100]

76 (R) [99]

86 (R) [100]

76 (R) [100]

–CH2CH3

66 (R) [94]

47 (R) [90]

72 (R) [98]

72 (R) [98]

–CH2CH3

94 (R) [95]

84 (R) [100]

94 (R) [94]

84 (R) [91]

–CH2CH3

86 (R) [100]

60 (R) [32]

46 (R) [31]

0 (R) [7]

O O

R1

R2

O 1

–CH2CH3

–CH3

2

–CH2CH3

H3C H3C

CH3 –CH2CH3

3

4

H3C H3C

5

6

F

7

F3C

CH3

F

CF3 8

O

O 9

substitution rather than by the relative bulkiness of the substituents at the two sites of the carbonyl group. 5.4

Solvent eﬀect

In enantioselective hydrogenations of activated ketones both the rates and the enantioselectivities are greatly inﬂuenced by the type of solvents used. Catalysis, 2010, 22, 144–278 | 187

Both protonic and aprotonic solvents have been applied. The solvent can inﬂuence the enantioselective hydrogenation in diﬀerent ways; it changes the solubility of the hydrogen99 and the substrate, the mass transport properties of the reaction mixture, the adsorption behavior of substrates and modiﬁer on the Pt active sites. Solvents have also a great inﬂuence on the conformation of alkaloid used.88,176,183 Furthermore, with less rigid substrates, e.g. alkyl pyruvates, the solvent polarity can also aﬀect the conformation of the substrate.49 Unfortunately, there are no general rules for the selection of an optimum solvent as both the rate and ee values were aﬀected not only by the solvent, but also by the substrate and the alkaloid applied. In the ﬁrst studies ethanol was the most common solvent used; however as it was shown latter the alcohols react with a-ketoester with the formation of semi-ketals (see Section 5.1). This eﬀect was also discussed in a recent study related to the enantioselective hydrogenation of ethyl-4,4,4-triﬂuoroacetoacetate348 in ethanol and propanol. All these results indicate that the use of alcohols in kinetic or mechanistic investigations65,247,348,349 should be avoided as it was emphasized in ref. 53. Upon using O-alkylated CD (R-O-CD) derivatives the use of AcOH or TFA is not recommended as in their presence hydrolysis of R-OCD can take place. AcOH was not a proper solvent for KPL.77 In one of the ﬁrst studies the reaction rates and enantioselectivities were compared in ethanol, toluene and acetic acid.73 These results unambiguously show the advantage of using AcOH. In another studies it was shown that upon hydrogenating EtPy AcOH is the best solvent as the highest ee values (ee=98%)55 and highest reaction rates were obtained in this solvent. The inﬂuence of the two most commonly solvents, i.e., toluene and acetic acid, on the rates and ee in enantioselective hydrogenation of EtPy are shown in Table 7. These data clearly show the superior inﬂuence of AcOH on the ee values. Not only higher ee values were obtained in AcOH, but also the amount of modiﬁer required to get high ee values is one order less in AcOH than in toluene. However, it is interesting to note that contrary to earlier observations at atmospheric hydrogen pressure the rate of hydrogenation in toluene is higher than in AcOH. In this respect it is worth for

Table 7 Hydrogenation of EtPy at atmospheric and at high pressure. Comparison of reaction rates and ee values in toluene and acetic acid solvents. (Reproduced from refs. 267, 350 with permission) Hydrogenation in AcOH at 100 bar350

Hydrogenation in AcOH at 1 bar267

Hydrogenation in toluene at 100 bar350

Hydrogenation in toluene at 1 bar267

Modiﬁer, rate, Mmol/l mmol/sec

rate, ee, % mmol/sec

rate, ee, % mmol/sec

rate, ee, % mmol/sec

ee, %

0.001 0.01 0.1 1

83 92 94 94

69 91 92 92

35 75 82 83

16 64 78 77

35 80 105 135

16 55 85 63

188 | Catalysis, 2010, 22, 144–278

8 20 48 25

35 57 96 71

Fig. 15 Enantioselective hydrogenation of KPL over Pt/Al2O3 in the presence of synthetic modiﬁer (R,R)-PNEA. The solvent was toluene with increasing amount of TFA. (Reproduced from ref. 209 with permission)

mentioning that in racemic hydrogenation of EtPy higher rates were measure in toluene than in AcOH.342 Other organic acids were also used to improve the enantioselectivity in Ptcinchona system. Fig. 15 shows the inﬂuence of added triﬂuoroacetic acid (TFA) on the enantioselectivity in the hydrogenation of KPL in the presence of synthetic modiﬁer (R,R)-PNEA. The results showed that excess TFA was needed to get maximum ee values. It was suggested that part of the excess TFA be required to neutralize basic sites of the Al2O3 support. Excellent correlation was found between the dielectric constant of the solvents used and both the enantiomeric excess and the population of conformer Open(3) as calculated by density functional theory in combination with a reaction ﬁeld model (POpen(3)).88 This dependence is shown in Fig. 16. Earlier results indicated that both reaction rates and ee values decreased with the polarity of the solvent. Figs. 17A–C show the dependence of ee on the empirical solvent parameter ENT in three diﬀerent systems. In the hydrogenation of EtPy (see Fig. 17A) good ee values are obtained in moderately apolar solvents, in which the reactant and modiﬁer dissolve.46 Interestingly, primary alcohols are also good solvent, although they react rapidly with the substrate with the formation of corresponding semi-ketals. The highest ee that time was 95% obtained in acetic acid, while the lowest one in water. The former result was attributed to the protonation of the quinuclidine nitrogen of CD by AcOH and suggesting the alteration of the reaction mechanism in AcOH,283 while the low activity in water can be attributed to the side reaction between EtPy and water, i.e. to the formation of corresponding vicinal diol, and the racemic hydrogenolysis of the diol formed. The latter reaction is responsible for the loss of ee. Fig. 17A shows that the solvent inﬂuence is notable, although the slope in this ﬁgure is relatively moderate. Contrary to that in Rh/Al2O3-b-ICN system used in the hydrogenation of KPL the above slope is higher indicating that the solvent has more pronounced inﬂuence on the enantioselectivity (see Fig. 17B).340 Catalysis, 2010, 22, 144–278 | 189

Fig. 16 Combined plot of the ee values obtained in the hydrogenation of KPL over Pt/Al2O3CD system (left axis) and the population of conformer Open(3) as calculated by DFT in combination with a reaction ﬁeld model (P Open(3), right axis) v.s. the dielectric constant of the solvent (axis scale is arbitrarily chosen). Solvents: 1 – cyclohexane, 2 – hexane, 3 – toluene, 4 – diethyl ether, 5 – tetrahydrofurane, 6 – acetic acid, 7 – ethanol, 8 – water, 9 – formamide. (Reproduced from ref. 88 with permission)

The third system behaves in a completely diﬀerent way. The correlation between ee and ENT is not really good, however it is more interesting that the character of dependence is opposite compared to the systems given in Figs. 17A and B. Nevertheless, the striking eﬀect of solvent properties on the ee values is obvious. The highest ee to (S)-3,5-di(triﬂuoromethyl) phenylethanol was obtained in the weakly polar solvent toluene and ethyl acetate. The ee decreased in polar solvents. In dimethylformamide, isopropanol, and ethanol the ee inverted from the (S) to the (R) enantiomer. This behavior indicates that in case of substituted acetophenones the reaction mechanism is strongly altered. There is another unusual behaviour of this type of substrates. In the enantioselective hydrogenation of 3,5-bis(triﬂuoromethyl)acetophenone the addition of triﬂuoroacetic acid (TFA) resulted in strong decrease in the reaction rate at TFA/CD=around 5 and full inversion of ee at TFA/ CDW50.340 Among the solvents AcOH and its triﬂourinated derivative (TFA) has their own peculiarities. The diﬀerence in the rates in AcOH and other solvents is well documented in one of the earlier results, where the addition of small amount of acetic acid strongly increased the overall performance of the reaction both in toluene and ethanol solvents84 as shown in Table 8. Results given in Table 8 reﬂect also the inﬂuence of semi-ketal formation in alcoholic solvents on the reaction rate and ee. The rate decreases in the following order: n-butanolWethanolWmethanol, i.e. it follows the reactivity trend of alcohols to form semi-ketals: n-butanoloethanolomethanol. Very pronounced solvent eﬀect was observed in the hydrogenation of 1,1,1-triﬂuoro-5,5-dimethyl-2,4-hexanedione using Pt/Al2O3 in the presence 190 | Catalysis, 2010, 22, 144–278

Fig. 17 Eﬀect of solvent on enantiodiﬀerentiation of in diﬀerent enantioselective hydrogenation reactions. Dependence of the ee value on empirical parameter EN T (reproduced from refs. 46,340,352 with permission) A: Pt/Al2O3-CD and DHCD in EtPy hydrogenation.46 Solvents: 1=cyclohexane, 2= toluene, 3=chlorobenzene, 4=THF, 5=dichloromethane, 6=propanol, 7=1-pentanol, 8=ethanol, 9=AcOH, 10=methanol, 11=water; B: Rh/Al2O3b-ICN in KPL hydrogenation.351 Solvents: 1=toluene, 2=t-BuMe ether, 3=THF, 4=dichloromethane, 5=DMF, 6=t-butanol, 7=acetonitrile, 8=2-propanol, 9=AcOH; C: Pt/Al2O3-CD in the hydrogenation of 3,5-bis(triﬂuoromethyl) acetophenone.340

Table 8 Solvent eﬀect and inﬂuence of acetic acid. (Reproduced from ref. 84 with permission) No

Solvent

k1, min 1

k2, min 1

Optical yieldc, %

1 2 3 4 5 6 7 8 9

Methanol ethanol n-butanol toluene (EtPy)a MCH (EtPy)a toluene (AcOEt)a MCH (AcOEt)a ethanol þ AcOHb toluene þ AcOHb

0.022 0.057 0.099 0.057 0.106 0.063 0.069 0.074 0.120

0.019 0.034 k2Wk1 k2Wk1 0.106 k2Wk1 0.060 0.048 0.120

62.2 72.0 75.0 86.3 78.1 84.0 75.3 91.4 93.1

Reaction conditions: T=23 1C, P=50 bar, [EtPy]0=1.0 M, [CD]=8.4 10 4 M, CD injection. a Ethyl pyruvate or ethyl acetate (1.5 cm3) is added to dissolve cinchonidine in these solvents (8.5 cm3). b The solvent is mixed with acetic acid; [AcOH]0=5.0 M. c Measured at 90–100% conversion.

of synthetic modiﬁer, pantoyl-naphthylethylamine.210 The results show a dramatic inﬂuence of solvents on the ee values. No correlation can be obtained between ee and relative permittivity (er,) or empirical solvent parameter (ENT). Surprisingly high ee values were obtained only in halogen Catalysis, 2010, 22, 144–278 | 191

containing solvents. Some of these solvents, for instance dichloromethane, are often called as ‘‘reactive solvent’’. A rapid loss of activity of Pt/alumina was found in this solvent and steady-state conditions could not be reached in the continuous-ﬂow reactor at 10 bar.295 In one of the recent publications352 it was mentioned the use of dichloromethane should be avoided in hydrogenation reactions. As it was emphasized ‘‘dehalogenation of this solvent on Pt, particularly at high hydrogen pressure, aﬀords HCl, which induces a new set of acid-catalyzed side reactions. It is a very correct remark what was addressed against the use of CH2Cl2 at high pressure by an English group.233 However, we consider also that the use of CH2Cl2 as a solvent by Baiker’s group in ATR-FTIR spectroscopic studies314 can also be criticized. These ATR-FTIR studies are considered as one of the crucial proves for the formation of protonated CD in the absence of AcOH. However, as it was mentioned in ref. 352 HCl can be formed from CH2Cl2. Consequently, the use of CH2Cl2 would result in the protonation of CD in the absence of AcOH and hydrogen. 5.5

Kinetic aspects

5.5.1 Rate acceleration. The rate enhancement of enantioselective hydrogenation of activated ketones has been observed by various research groups. The ﬁrst results were published by the Ciba Group.57,58,63 The phenomenon was also observed by others.65,67,289 In all of these studies using EtPy as a substrate the common observation was that the modiﬁed reaction is 20–100 times faster than the unmodiﬁed one. The rate enhancement was also described as ‘‘ligand acceleration’’58 based on the analogous observation in homogeneous catalysis.353 It was proposed that ‘‘a reaction is considered ligand accelerated if there is a slower unmodiﬁed (unselective) cycle and a faster modiﬁed (selective) cycle’’.58 This term has been used for many years, although its chemical meaning is quite doubtful or even misleading. Most of the authors use the term rate acceleration (RA) or rate enhancement (RE). Not only cinchona alkaloids, but also other tertiary amines, such as quinuclidine, triethylamine, etc. can induce RA63. This behaviour was evidenced in various solvents.73,354 The addition of small amount of acetic acid into ethanol or toluene resulted in even more pronounced RA84. This behaviour is very characteristic for a-keto esters (Etpy, MePy, KPL77) and was observed not only over Pt, but other metals such as Rh252,254 and Ir.355 In ref. 341 it was found that in a-ketoester the increase of the size of R1 and R2 groups resulted in slight alteration in the extent of RA. With respect to kinetics studies the reproducibility problems related to diﬀerent impurities has to be mentioned.73,84 Diﬀerent batches of ethyl pyruvate can give completely diﬀerent kinetic results.84,356 The impurities alter both the initial rates and the enantioselectivity. When unpuriﬁed EtPy is used in this case the rate of racemic hydrogenation is extremely low.233 For this reason, kinetic results using the given substrate without any puriﬁcation65,66,227,232 should be treated with great precaution. However, even in 192 | Catalysis, 2010, 22, 144–278

this case the addition of cinchona alkaloids as modiﬁers or even diﬀerent tertiary amines increases the rate signiﬁcantly. The other problem is that initial rates or related kinetic information are only seldom published. Most authors prefer to provide activity data in the form of conversion values measured after a given reaction time. It is especially notable in recent publications.235,351,357 Consequently, less and less information is available with respect to the RA phenomena. It has to be mentioned that RA has been observed only in liquid phase hydrogenations; in gas phase hydrogenation of MePy no RA was measured.288 Direct measurement of the reaction rates by using reaction calorimetry in a transient experiment provided an unambiguous evidence for the RA100. The results show that upon injection of CD the rate increase is instantaneous and the rate increase is in the range of 5–12 depending on the type of catalyst used. Similar pronounced RA was observed in other studies over 5% Pt/Al2O3 catalyst using EtPy upon injecting cinchona type modiﬁers or tertiary amines at higher pressures.289,307 With respect to the RA one important question can be raised: is any direct coupling between the reaction rate and the enantioselectivity. This coupling was ﬁrst described by Blaser et al. in their ‘‘ligand acceleration model’’.58 Although the above coupling was clearly presented recently more and more evidences have been accumulated that this coupling is not a prerequisite to obtain high enantioselectivities. The ﬁrst evidences against the direct coupling were obtained in our studies.289,358 In ref. 358 it was shown that the modiﬁcation of Pt/Al2O3 by Sn(C2H5)x moieties strongly alter the reaction rates, but their eﬀect on the ee is negligible. Based on the analysis of the form of conversion-selectivity dependencies in ref. 289 it was stated that, ‘‘at low concentrations of substrate and modiﬁer, contrary to instantaneous rate acceleration, the maximum ee values are obtained only after a certain time delay. The increase of the rate of enantioselective hydrogenation with respect to the racemic one was well documented in ref. 203 upon using four diﬀerent substrates (EtPy, EOG, EBF, PADA) as shown in Fig. 18. It should also be mentioned that there is a deﬁnite class of substrates that do not show any RA or even the rate of enantioselective reaction is slower than the rate of racemic one. In most of these substrates the prochiral keto group is not activated. These substrates are as follows: acetophenone,359 3,5-bis(triﬂuoromethyl)acetophone,340 alkylsubstituted triﬂuoromethyl ketones.359,360 Classiﬁcation of substrates according to their extent of RA and ED was given recently (see Fig. 1).72 In addition, the following experimental conditions are not favourable for RA: (i) MePy pyruvate in gas phase;290 (ii) EtPy in the presence of a-ICN;197,361 (iii) EBF in AcOH at 1 bar hydrogen;56 (iv) non-activated aromatic ketones359 and non-activated triﬂuoromethyl derivatives;363 (v) hydrogenation of EtPy in the presence of CD at very low substrate concentrations.234 The appearance and the disappearance of RA acceleration are well documented in a series of experiments using triﬂuoro acetophenone and cyclohexyl analog. In the case of the former substrate pronounced rate Catalysis, 2010, 22, 144–278 | 193

Fig. 18 Conversion–reaction time dependencies in the enantioselective hydrogenations of activated ketones (standard conditions, DHCD concentration 0.01 mmol l 1, 0.16 ml; EOG: diethyl 2-oxoglutarate, EBF: ethyl benzoylformate, PADA: pyruvaldehyde dimethyl acetal, () racemic hydrogenations, () enantioselective hydrogenations.) (Reproduced from ref. 203 with permission).

Fig. 19 The eﬀect of CD concentration on the enantioselective hydrogenation of triﬂuoro compounds. A: Substrate=triﬂuoroacetophenone; B: Substrate=triﬂuoromethylcyclohexyl ketone. (Reproduced from refs. 360,363 with permission)

acceleration was observed as shown in Fig. 19A. Contrary to that upon using triﬂuoro cyclohexyl ketone in the presence of modiﬁer the rate of reaction decreased as presented in Fig. 19B. Further insight on the origin of RA was obtained upon comparing the hydrogenation of acetophenone and triﬂuoromethyl derivatives of 194 | Catalysis, 2010, 22, 144–278

acetophenone over Pt/Al2O3-CD system.332,362 In these studies both the rate and the ee values strongly depended on the electronic and steric eﬀect of the substituents and on the hydrogen-bonding interactions between the quinuclidine N atom of the alkaloid and the carbonyl group of the substrate. In the hydrogenation of acetophenone and triﬂuoromethylacetophenone derivatives on CD-modiﬁed Pt/Al2O3, the conversion rates and enantioselectivities varied strongly with the nature of the aromatic substituents.332,362 Diﬀerent reactivities were attributed to the electronic (and steric) eﬀect of the substituents and to hydrogen-bonding interactions between the quinuclidine N atom of the alkaloid and the carbonyl group of the substrate. A linear correlation has been found between the logarithm of the reaction rate and the highest occupied molecular orbital and lowest unoccupied molecular orbital stabilization of the carbonyl compounds (DEorb), relative to the reference compound.359 The by-products are high-molecular weight compounds and are considered as strong catalyst poisons reducing the number of available Pt sites resulting in substantial decrease in the reaction rates. Consequently, in this case the RA is masked by a catalyst poisoning eﬀect. Recently the RA phenomenon has been questioned by two research groups.233,234,364 In ref. 232 it was concluded that ‘‘rate enhancement is now attributed to reaction occurring at a normal rate at an enhanced number of sites, not (as previously proposed) to a reaction occurring at an enhanced rate at a constant number of sites’’. The ﬁnal conclusion was that the ‘‘rate enhancement in the presence of an alkaloid modiﬁer is attributed to the inhibition of the pyruvate ester polymerization at the Pt surface’’. In another recent study365 it was emphasized that ‘‘the reaction rate was lower in all chirally modiﬁed reactions as compared to the racemic reaction in the absence of modiﬁer’’. We believe that in references cited above experimental conditions were not properly chosen as their ﬁndings strongly contradict to results observed earlier by several groups. In a recent paper352 the use of ‘‘reactive’’ solvent in ref. 232, such as dichloromethane, was strongly criticized. Recently, with respect to the RA phenomena an open dispute has been emerged in Journal of Catalysis.295,352,366 In a recent study continuous–ﬂow experiments were performed providing clear evidence that the rate acceleration exists and it was concluded that ‘‘it is not the suppression of catalyst deactivation by addition of chiral modiﬁer, because under appropriate conditions catalyst deactivation is negligible in pyruvate hydrogenation’’.295 This statement was strongly opposed in ref. 366. Those who favor the role of deactivation defended their view referring to their earlier results shown in Fig. 20.234 This ﬁgure shows that the rate acceleration appears only at high concentration of substrate, while at low concentration the rate of enantioselective hydrogenation is lower than that of the racemic one. In this respect it has to be mentioned that the determination of reaction rate at low substrate concentration is very plausible. We consider that the minor diﬀerences shown in Fig. 20 cannot be considered as a real prove for the lack of RA. We have to emphasize again that the decrease of the reaction rate in enantioselective hydrogenation of EtPy at low substrate concentration366 Catalysis, 2010, 22, 144–278 | 195

Fig. 20 Initial hydrogenation rates of enantioselective (D) and racemic hydrogenation (E) of EtPy and the enantiomeric excess (K). (reproduced from refs. 234 with permission)

0.8

0.8 conversion

1.0

conversion

A 1.0

0.6 0.4 0.2

0.6 0.4 0.2

0.0

0.0 0

50 100 time, min

150

0

50 100 time, min

150

Fig. 21 Inﬂuence of the time delay in CD injection during raceme hydrogenation. Kinetic curves of EtPy hydrogenation upon using puriﬁed substrate; [CD]=5 10 5 M, T=20 1C, PH2=50 bar, catalyst=5% Pt/Al2O3 (E 4759), 0.125 g; 7 – CD injection at 0 min; & – CD injection at 15 min; – CD injection at 30 min; } – CD injection at 90 min; () – no CD (racemic hydrogenation); A: [EtPy]0=1.0 M (puriﬁed by distillation prior to the use); B: [EtPy]0=1.0 M (‘‘distillation residue’’ containing dimer 1a in the amount of 20%). (Reproduced from ref. 367 with permission)

can be related to the decreased concentration of substrate-modiﬁer complexes formed under condition of catalytic hydrogenation. In this respect it is irrespective where the above complex has been formed in the liquid phase or at the Pt surface. The decrease of the concentration of intermediate complex can result in pronounced rate decrease; similar to the kinetic patterns observed in enzymatic kinetics. In our recent study transient experiments with injection of CD during racemic hydrogenation of EtPy were investigated using puriﬁed substrates and a ‘‘distillation residue’’. The ‘‘distillation residue’’ contained 20% of compound 1a (see Scheme 2 in Section 5). Fig. 21A shows that the increase in the time delay between the start of racemic hydrogenation and the injection of CD has no inﬂuence on the rate of enantioselective 196 | Catalysis, 2010, 22, 144–278

hydrogenation. It means that racemic products formed up to 10% conversion have no measurable inﬂuence on the reaction rate, consequently the size of free Pt surface available for enantioselective hydrogenation is not altered during racemic hydrogenation. When similar series of experiments were performed in ethanol in the above conversion range slight decrease in the initial rates was observed.289 Results given in Fig. 21B clearly show that upon using the ‘‘distillation residue’’ the rate of racemic hydrogenation decreases, the decrease in rate is around eleven-fold compare to puriﬁed EtPy. The rate decrease in the racemic hydrogenation is due to the strong poisoning eﬀect induced by compound 1a. The poisoning eﬀect can also be observed in enantioselective hydrogenation, its extent is around three-fold. The introduction of CD during racemic hydrogenation of ‘‘distillation residue’’ resulted in also instantaneous rate acceleration in all cases (see Fig. 21B). Contrary to results obtained in the previous series of experiments upon increasing the time delay from zero to 90 minutes the rate of enantioselective hydrogenation decreases. All these results unambiguously show that the statement given in ref. 232 ‘‘rate enhancement is now attributed to reaction occurring at a normal rate at an enhanced number of sites, not (as previously proposed) to a reaction occurring at an enhanced rate at a constant number of sites’’ cannot be valid. It is hard to suggest that the addition of 5 10 5 M modiﬁer will compete with 0.2 M high molecular weigh product and can remove their adsorbed forms instantaneously from the Pt surface. In an analogous series of experiments shown in Fig. 22, upon using methyl-benzoyl formate (MBF) substrate367 similar trend in the concentration dependences was obtained as in the case of EtPy (see Fig. 20). However, the rate of the enantioselective hydrogenation was higher than that of the racemic one in the whole concentration range. Results obtained in series of experiments using MBF shows that the RA eﬀect is maintained in a relatively broad concentration range. It is a good example for the appearance of RA for the class of substrate with decreased ability to form by-products. Finaly let us conclude that we completely

0.030

d[R+S]/dt, M/min1

0.025 0.020

y = 0.007Ln(x) + 0.0306 R2 = 0.9794

0.015 0.010

y = 0.0007Ln(x) + 0.0053 R2 = 0.9206

0.005 0.000 0

0.2

0.4

0.6

Concentration MBF, M Fig. 22 Initial hydrogenation rates of enantioselective (&) and racemic hydrogenation (}) of MBF. (Reproduced from ref. 367 with permission)

Catalysis, 2010, 22, 144–278 | 197

disagree with the new views advertised in refs. 233, 234. Both our earlier69,289,307 and recent results269 show that upon introduction of CD during racemic hydrogenation of EtPy the RA are instantaneous. This conclusion has been supported by recent results obtained in continuous ﬂow reactor.295,366 5.5.2 Enantioselectivity–conversion dependencies. In enantioselective hydrogenation of EtPy one of the most disputed kinetic pattern is the form of the enantioselectivity–conversion (time) dependencies (ECD). In the hydrogenation of EtPy monotonic increase (MI) type dependencies were obtained at low conversion in various studies83,84,268,289,369 as shown in Fig. 23. The MI type behaviour is often called as initial transient period (ITP).195

Fig. 23 Changes in ee observed during the course of EtPy hydrogenation. A: hydrogenation over a DHCD-Pt catalyst;369 B: hydrogenation over CD-Pt catalyst84 (2, ’) in toluene, CDinj. [Etpy]0=1.0 M, [CD]0=8.4 10 4 M, 3.4 10 2 M, respectively; all other experiments in ethanol, [Etpy]0=1.0 M, (B) – [CD]0 =6.8 10 6 M, (E) – [CD]0inj=3.4 10 5 M, () – [CD]0=0.4 10 4 M, (CD þ EtPy)inj, (&) – [CD]0=8.4 10 4 M, (EtPy)inj, (CD)premixed. (Reproduced from refs. 84 and 369 with permission)

In ﬁrst publications integrated ee values were calculated from actual concentration of (R) and (S) products according to the general formula ee ¼ ð½R ½SÞ=ð½R þ ½SÞ: Further on incremental ee values (eeincr=eecalc or Dee) were also used.374 It was calculated using the following formula: eeincr ¼ ½c2 ee2 c1 ee1 =½c2 c1 where c is the actual concentration of ethyl lactate and ee is the measured optical yield. The use of eeincr reﬂects the ee values in a given time interval. It is applied when two diﬀerent types of modiﬁers are added to the reaction mixture or when the loss of the modiﬁer during the enantioselective hydrogenation is very pronounced. In addition kinetic ee values can also be calculated from the corresponding reactions rates: eekin ¼ ð½rR ½rS Þ=ð½rR þ ½rS Þ:

198 | Catalysis, 2010, 22, 144–278

This kind of behaviour was also observed upon using other activated ketones, such as PADA,281 MBF281 and KPL.53 In earlier studies this behaviour was not discovered as no attempt was done to determine the ee values at low conversion. Other characteristic feature of ee-conversion (time) dependencies is the decrease of the ee values at high conversion.108 This decrease was attributed to the loss of alkaloid during the enantioselective hydrogenation. The addition of further amount of alkaloid during the hydrogenation experiment resulted in almost constant ee values even at high conversion.108 Results shown in Figs 23A and B have one common feature, i.e., the use of injection method for the introduction of reaction components. Fig. 23A shows the ee-time dependencies in EtPy hydrogenation in ethylacetate injecting the substrate into the mixture containing the catalyst and the modiﬁer. When CD was injected into the reactor using toluene or ethanol as a solvent MI type enantioselectivity-conversion dependencies were also observed provided the concentration of CD was less than 10 4 M (see Fig. 23B). In both solvents the appearance of MI character was independent whether CD or the substrate was injected. In all cases the increase part was very pronounced in the ﬁrst 15–25% of conversion. Later on similar behaviour was also described by two other groups in the hydrogenation of EtPy in ethanol using the premixing technique.66,82 There was a very tough dispute between these two groups as they had completely diﬀerent view on this new kinetic phenomenon.82,239 Further results clearly indicated281,350,371 that the appearance of MI type of enantioselectivity – conversion (time) dependencies strongly depends on the following experimental conditions: (i) concentration of CD, (ii) the mode of introduction of reaction components, (iii) the purity of substrates, (iv) the solvent used, and (v) conditions of catalyst pretreatment. Figs. 24A and B show the eecalc-conversion dependencies obtained in toluene upon using premixing and injection techniques, respectively.93

Fig. 24 Enantioselectivity (eecalc)-conversion dependencies during the hydrogenation of EtPy in toluene; & – [CD]0=1 10 4 M; E – [CD]0=1.2 10 5 M; T=23 1C, PH2=50 bar, [Etpy]0=1.0 M, catalyst: 5 wt% Pt/Al2O3 (Engelhard, E4759); A: premixing, B: injection. (Reproduced from ref. 93 with permission.)

Catalysis, 2010, 22, 144–278 | 199

The comparison of these two techniques shows that upon using premixing the MI type behaviour disappears completely and there is only a very slight increase or decrease of enantioselectivity with conversion. Contrary to that when the injection technique was used the MI type behaviour was observed and its character depended strongly on the initial concentration of CD. The lower the concentration of CD the more pronounced the MI character (see Fig. 24B). The MI character was completely maintained when eekin values were used instead of eecalc. It has to be emphasized that when similar experiments were performed in ethanol the MI character appeared upon using both premixing and injection methods.84 This behaviour was attributed to the formation of semi-ketal from the substrate and the solvent during the period of premixing. Similar experiments were performed in the hydrogenation of 3,5-bis(triﬂuoromethyl)-acetophenone over Pt/Al2O3 in the presence of CD.306 MI type dependence was observed under general experimental condition, while almost constant ee values were obtained after premixing the reaction mixture in nitrogen as shown in Fig. 25A.53 Consequently, identical observation was obtained as in the hydrogenation of EtPy.93 The results clearly indicate that both surface cleanness and interactions in the liquid phase have their contribution for the appearance of MI type ee–conversion (time) dependencies. Diﬀerent aspects of the appearance of initial transient period were discussed by Barto´k upon using various substrates, such as EtPy, PADA, MBF.281 One of the most interesting observations was the dependence of initial transient period on the pre-treatment conditions and the purity of the substrate as shown in Fig. 25B. After pretreatment in helium ee values are 15–20% higher than those without this pre-treatment. This behaviour was attributed to the of 5 ppm oxygen in helium. It was suggested that the oxygen can alter the Pt surface, what is more favourable for the interaction with DHCD or CD. A

B 100

He

ee (%)

80

H2

60 98% purity / He 98% purity / H2 99.9% purity / He 99.9% purity / H2

40

20 0

20

40

60

80

100

Conversion (%)

Fig. 25 Eﬀect of catalyst pretreatment on ee-time (conversion) dependencies. A: substrate=3,5-bis(triﬂuoromethyl)-acetophenone, catalyst=Pt/Al2O3, chiral modiﬁer=CD, solvent=toluene, pH2=1 bar, r.t. a – no catalyst pretreatment;53 b – catalyst is prereduced in H2 at 400 1C; c-catalyst is prereduced in H2 at 400 1C, and then, the reaction mixture was stirred under N2 for 1 h before H2 was introduced (premixing); B: Pretreatment of re-reduced catalyst in toluene with H2 or He in the enantioselective hydrogenation of EtPy. (Reproduced from ref. 281 with permission.)

200 | Catalysis, 2010, 22, 144–278

100 3r 80

2r

1r

ee (%)

3 1

2 60 [DHCD] (mmol/L) 1 1r 2 2r 3 3r

40

20 0

20

40

0.01 0.01 0.1 0.1 0.01 0.01

60

use reuse use reuse use reuse

80

Temp. (°C) -10 -10 -10 -10 -20 -20

100

Conversion (%) Fig. 26 Repeated use of catalyst in the enantioselective hydrogenation of EtPy: eﬀect of DHCD concentration and reaction temperature on the ee-conversion dependencies (fresh DHCD was added for reuse, r=reuse). (Reproduced from ref. 268 with permission)

The inﬂuence of the reuse of the catalyst on the initial transient period was investigated by the Barto´k’s group. Typical MI type dependencies are shown in Fig. 26.268 The results showed that the expression of MI character depended on the amount of modiﬁer, but it was less pronounced after reuse of the catalyst. However, in repeated use, i.e. after removal of the reaction mixture and addition of fresh toluene, EtPy, and modiﬁer resulted in 10–20% increase in the ee values. On the basis of these results it was concluded that the ‘‘the phenomenon of increase in ee on reuse is an intrinsic feature of the catalyst system used, i.e. new chiral centers making higher ee possible are formed’’.195 In this respect restructuring of the Pt surface was suggested based on analogous analysis of results given in refs. 55, 90, 277. It was also supposed that oxygen plays a deﬁnite role in this process. A sequential introduction of diﬀerent substrates was performed in which the hydrogenation of MePy was carried out following the initial hydrogenation of EtPy using Pt/Al2O3-cinchona catalyst. In these experiments, the MI type character was obtained in both hydrogenation experiments (see Fig. 27A). The character of ee-conversion dependencies was maintained after reversed order of introduction of substrates (see Fig. 27B). The observation that the initial transient eﬀect is still observed with the sequential hydrogenation of EtPy and MePy indicates that the phenomenon cannot be attributed to impurity eﬀects. Consequently, it is more probable that the reaction-driven equilibrium of the chiral environment play a role in the MI character of ee-conversion dependencies. In one of the recent studies three diﬀerent modiﬁers, such as CD, 9–Ophenyl-CD (PhOCD), 9–O-pyridil-CD (PyrOCD) were investigated in the hydrogenation of EtPy.372 Well-expressed MI type behaviour was obtained for all three modiﬁers. However, despite all the convincing results presented Catalysis, 2010, 22, 144–278 | 201

Fig. 27 Enantioselectivity-conversion dependencies in sequential hydrogenation experiments. a: EtPy hydrogenation to 100% conversion prior to addition of MePy; b: MePy hydrogenation to 100% conversion prior to addition of EtPy; ’ – EtPy conversion, K – e.e. in (R)-EtLa, – MePy conversion, E – e.e. in (R)-MeLa. T=20 1C, reaction pressure 30 bar H2 for EtPy hydrogenation, 50 bar H2 for MePy introduction. (Reproduced from ref. 371 with permission.)

7

in this section the following statement was done: ‘‘We assume that the slow removal of surface impurities is the major reason for this behavior’’ (i.e. the MI type ee-conversion dependencies). Consequently, there are groups who do not learn from results obtained by other groups and keeping their old views as a dogma. Finally we can summarize the general views with respect to the origin of the initial transient behaviour of ECD: (i) it is related to impurities or other experimental artifacts,82 (ii) it is due to surface induced alterations,60,82,239 (iii) intrinsic kinetics.75,195 5.5.3 Non-linear phenomenon. Non-linear eﬀects (NLE) in homogeneous asymmetric catalysis have been investigated for many years since the pioneering work of H.B. Kagan.373 First the phenomenon was attributed to the diastereomeric association inside or outside the catalytic cycle.374 Later on the approach was extended to the use of mixtures of diastereomeric ligands.375 Recently this approach was been extended to the Orito’s reaction. It was suggested that the nonlinear behavior be due to the deviation from the expected ideal behaviour assuming that the molar ratios of the modiﬁers in solution and on the metal surface are identical. Consequently, the nonlinear behavior of mixtures of two diastereoisomers or two completely diﬀerent chiral modiﬁers has been attributed mainly to their diﬀerent adsorption strength,287 however the contribution of the adsorption geometries on the metallic sites was also emphasized and new term ‘‘non-linear phenomenon’’ (NLP) has been introduced.269 Besides it was also suggested269 that modiﬁer–modiﬁer interactions may also be involved in the NLP, but no experimental evidence has been found yet. It was concluded that the investigation of NLP behavior of mixtures of two modiﬁers is a powerful tool in heterogeneous catalysis for characterizing the relative adsorption strength of modiﬁers under truly in situ conditions. However, in this respect the controversy between catalytic and spectroscopic investigations related to the evaluation of the relative adsorption 202 | Catalysis, 2010, 22, 144–278

strength of various chiral modiﬁers has to be mentioned.376–378 We have to admit that this controversy can be attributed to the absence of substrate and hydrogen in spectroscopic investigations. In a recent review based on the above results the following statement was done: ‘‘An essential conclusion from this study is that, although strong adsorption is a crucial requirement for an eﬃcient modiﬁer, there is no positive correlation between the adsorption strength (AS) and the enantioselectivity achieved with the modiﬁer alone’’.53 In a recent study it was emphasized that the NLP ‘‘can presumably be interpreted on the basis of diﬀerences in the structure of the substratemodiﬁer complexes formed and in the adsorption-desorption processes of the complexes, thus the NLP is not solely dependent on the adsorption of cinchona alkaloids, as suggested by earlier experimental data’’.379 The above view is very close to our one. In one of our study361 we turned back to the original idea given by Kagan, namely to the formation of diastereomeric associations between the substrate and diﬀerent competing chiral entities as modiﬁers in the liquid phase. Consequently, in our interpretation the nonlinear behaviour is due to diﬀerent enantio-diﬀerentiation ability of two modiﬁers acting simultaneously in the liquid phase resulting in diﬀerent substrate-modiﬁer complexes (associations). Of course the enantio-diﬀerentiation ability of two modiﬁers is further inﬂuenced by diﬀerent factors, such as the adsorption-desorption behaviour, the abundance and the reactivity of the formed associates. It has to be added that kinetically the diﬀerence between the two interpretations (diﬀerent adsorption strengths v.s. diﬀerences in the structure and stability of substrate-modiﬁer complexes) for the NLP cannot be done. Only careful analysis of the chemical and surface properties can provide some hints inside the origin of these observations. Diﬀerent experimental techniques were used to investigate the NLP of two alkaloids: (i) variation of the initial ratio of two modiﬁers measuring the ee values at the end of the reaction,287 (ii) applying a ﬁxed initial ratio of two modiﬁers and following the ee-conversion dependencies,361 (iii) using transient method in a batch reactor, where one of the modiﬁers is introduced at t=0, while the other one after a given time lap,269 (iv) using continuous ﬂow reactors and creating transient conditions by switching from one modiﬁer to another one.301 The deviation from the expected linear correlation was ﬁrst observed in the hydrogenation of EtPy in the presence of CD–CN and QN–QD mixtures.287 At mole fraction of 0.5 of these alkaloids the ee value was higher than zero indicating that the ED ability of CD and QN is higher than that of the CN and QD, respectively. Similar result was also obtained in other publications.60,379 In all of these studies the ﬁndings were attributed to the diﬀerences in the adsorption strength of the alkaloids. The most striking NLP behaviour was observed in the hydrogenation of KPL over Pt/Al2O3 in the presence of CD-PhOCD mixtures (see Fig. 28A).238 It is known that PhOCD gives (S)-pantolactone, whereas CD aﬀords (R)-pantolactone as major enantiomer. The addition of small amount of CD (XCDo0.05) to a reaction mixture containing PhOCD resulted in drastic change from (S)-pantolactone to (R)-pantolactone as the Catalysis, 2010, 22, 144–278 | 203

Fig. 28 Non-linear eﬀect in enantioselective hydrogenation reactions over Pt/Al2O3 catalyst. A: substrate=KPL, chiral modiﬁer=CD-PhOCD mixtures; schematic illustration of the adsorption of CD and PhOCD on an idealized ﬂat Pt surface;238 B: substrate=EtPy, solvent=toluene, chiral modiﬁer=mixtures of diﬀerent modiﬁers, the 2nd modiﬁer was added at 10–20% conversion of EtPy.380 (Reproduced from refs. 238 and 380 with permission.)

major product. This non-linear behavior is attributed to the much stronger adsorption of CD compared to PhOCD. The weaker strengths of adsorption of PhOCD were related to its tilted form of adsorption as shown in Fig. 28A. Systematic investigation of NLP was done by two groups.197,269,378–381 Results obtained in a series of experiments using CD, CN, QN and QD is shown in Fig. 28B.378 Based on these results the following order was established for the adsorption strength of cinchona alkaloids: CDW CNWQNWQD. The above order was also supported by other studies.269,378 Contrary to that RAIRS measurements of adsorbed alkaloids resulted in a diﬀerent order for the strength of adsorption: QN,QDWCDWCN.376 In a more recent RAIRS experiments the order in the adsorption equilibrium constants (Kads) the following sequence was established: CNWQDW CDWQN.377 Probably based on these results in a recent review the following statement was done: ‘‘An essential conclusion from this study is that, although strong adsorption is a crucial requirement for an eﬃcient modiﬁer, there is no positive correlation between the adsorption strength (AS) and the enantioselectivity achieved with the modiﬁer alone’’.53 Upon investigating the behaviour of O-alkylated derivatives of CD it was nicely demonstrated that the adsorption strength of this type of modiﬁers on Pt decreases in the following order: CDWMeOCDWEtOCDWPhOCDE TMSOCD.202 204 | Catalysis, 2010, 22, 144–278

Fig. 29 Appearance of chiral switch. A: Chiral switch induced by replacing CN with CD (ﬁlled circles) and vice versa (ﬁlled squares). The conversion is shown with open symbols (open squares are barely seen due to overlapping). The second modiﬁer reached the catalyst after about 45 min (vertical dashed line); B: Inﬂuence of the modiﬁer concentration on the dynamics of the chiral switch. Conditions: 0.226 mM CD, 0.226 mM QD concentration for 1:1 ratio of CD and QD (ﬁlled circles and squares) and 2.26 mM for the 10 fold amount of QD relative to CD (open triangles) 100% conversion. (Reproduced from ref. 295 with permission)

A continuous-ﬂow ﬁxed-bed reactor was applied in the enantioselective hydrogenation of EtPy on Pt/Al2O3 using the principle of ‘‘chiral switch’’.301 These time on stream experiment start with the introduction of one of the modiﬁers and a given moment this modiﬁer is switched to another one. The QD-CD and CD-QD switch is shown in the Fig. 29A and B.295 The results clearly show that the enantio-diﬀerentiation ability of CD is stronger than that of the QD. All of the above results strongly support the general view that NLP and the adsorption strength of the modiﬁers are coupled. However, there are experimental ﬁndings indicating that the origin of NLP has more complex basis. In this respect let us refer to a series of transient experiments performed in a batch reactor. Fig. 28B shows the inﬂuence of solvents when CD was added to the reaction mixture containing QN. In the opposite situation when QN was added to the reaction mixture containing CD only minor changes in the ee values were observed. The most interesting ﬁnding is that the ‘‘rate of replacement’’ of QN by CD shows strong solvent dependency. No real explanation was given for this ﬁnding, although the possibility for the involvement of solvent polarity and the formation of an alkaloid–acid ion pair has been mentioned. Our view is that these experimental ﬁndings indicate that not only the diﬀerence in the adsorption strength controls NLP. Even more striking results related to NLP were observed when the amount of modiﬁers was varied in the above experiments. These results are shown in Figs. 30 A–C.269 When CD was added to the reaction mixture containing QD the direction of the enantioselectivity was immediately altered. The time period required to reach the maximum Dee showed strong concentration dependence, what was attributed to the fast hydrogenation of the quinoline ring of QN in the ﬁrst 30 minutes at low concentration of modiﬁers. In the opposite situation, i.e. when QD was added to CD (see Fig. 30C) the decrease part was also explained by the fast hydrogenation of CD, however no acceptable explanation was given for the increase part. Catalysis, 2010, 22, 144–278 | 205

Fig. 30 Hydrogenation of EtPy over Pt/Al2O3; A: Solvent eﬀect on the exchange of QD by CD. Addition of one equivalent CD after a 30-min reaction time; B: Inﬂuence of modiﬁer concentration on the transient behaviour. Addition of one molar equivalent CD after a 30-min reaction carried out in the presence of QD; C: Addition of one molar equivalent QD after a 30min reaction carried out in the presence of CD. Standard conditions; acetic acid; amounts of the modiﬁers: 1.7, 0.17, and 0.017 mm. (Reproduced from ref. 269 with permission)

The mixing of two diﬀerent alkaloids was also applied to study the anomalous behaviour of both ICN. The results shown in Fig. 31A and B indicate that the addition of a-ICN to CD, CN and b-ICN has no inﬂuence on the enantio-diﬀerentiation ability of these alkaloids.361 Based on these results it was suggested that the origin of enantio-diﬀerentiation ability of a-ICN is diﬀerent than that of for CD, CN and b-ICN. Results shown in Fig. 31B indicate also that in the presence of CN the addition of a-ICN resulted in less loss of ee at high conversions. It is due to the suppression of the hydrogenation of the quinoline ring of CN in the presence of a-ICN. Consequently, a-ICN should be strongly adsorbed on the Pt surface. The ﬁrst attempt to compare the behaviour of two substrates (EtPy and KPL) in NLP under identical conditions using CN and QN was done in a recent study. The investigations were performed in two solvents (toluene and AcOH).379 Three diﬀerent methods were applied. Here we show results obtained in a batch reactor using conventional and transient experiments. According to results given in Fig. 32A, in the hydrogenation of EtPy the direction of enantio-selection changes almost linearly with the concentrations of the two modiﬁers. On the contrary, in the hydrogenation of KPL the direction of enantio-selection is aﬀected to a much higher extent by CN 206 | Catalysis, 2010, 22, 144–278

Fig. 31 Enantioselectivity-conversion dependencies in the presence of mixtures of a-ICN and ﬂexible alkaloids. T=20 1C, pH2=50 bar, injection method, 500 rpm, reaction time=90 min; A: ’ – 1.2 10 5 M CD; & – 1.2 10 5 M CD þ 1.2 10 5 M a-ICN; K – 1.2 10 5 M b– 1.2 ICD; – 1.2 10 5 M a-ICN þ 1.2 10 5 M b-ICN; B: – 1.2 10 5 M a-ICN; 10 5 M CN; D – 1.2 10 5 M CN þ 1.2 10 5 M a-ICN. (Reproduced from ref. 361 with permission)

7

Fig. 32 Comparing the behavior of EtPy (EP) and KPL in NLP. A: QN-CN modiﬁer mixture, total modiﬁers concentration: 0.1 mM, solvent: toluene/AcOH=9/1; B: QN-CN modiﬁers; C: CN-QN modiﬁers. In B and C concentration of each modiﬁer=0.05 mM, ﬁrst abbreviationmodiﬁer used ﬁrst, second abbreviation-modiﬁer added afterwards; solvent=toluene (T) and AcOH; modiﬁers. (Reproduced from ref. 379 with permission)

Catalysis, 2010, 22, 144–278 | 207

than by QN. This might indicate that in the presence of EtPy the adsorption strengths are identical, while in the presence of KPL the adsorption strength of CN is higher than that of QN. Based on results shown in Fig. 32B, i.e. when QN was used as a ﬁrst modiﬁer, the following conclusions can be drawn: (i) CN desorbs QN more readily in the hydrogenation of KPL than in the hydrogenation EtPy, (ii) in the case of EtPy, CN cannot fully desorb QN from the surface; (iii) in the hydrogenation of KPL CN can nearly fully desorb QN. Based on these ﬁndings the order of adsorption strength in these two substrates is diﬀerent, namely in EtPy CNBQN, while in KPL CNWQN. When CN was used as the ﬁrst chiral modiﬁer (see Fig. 32C), in the hydrogenation of EtPy CN cannot be desorbed by QN, while in the hydrogenation of KPL under identical conditions CN acted as if QN was not present at all. Consequently, the order of the adsorption strength of the two cinchonas is diﬀerent in these to substrates, in case of EtPy CNBQN, while in case of KPL CNWWQN. These ﬁndings are similar as those observed in ref. 361, where CD, CN and b-ICN acted as if a-ICN was not present at all. Such observation was also made in another earlier study in transient experiments using CN and QN.197 Summing up all investigations related to the elucidation of the origin of NLP we accept the conclusions given in a recent study that ‘‘the NLP depends not only on the chiral modiﬁer but also on the substrate to be hydrogenated. This observation can presumably be interpreted on the basis of diﬀerences in the structure of the substrate-modiﬁer complexes formed and in the adsorption-desorption processes of the complexes, thus the NLP is not solely dependent on the adsorption of cinchona alkaloids, as suggested by earlier experimental data.197 The statement is in full agreement with our view related to the importance of the formation of substrate-modiﬁer complexes. 5.5.4 Inversion of enantioselectivity. To ﬁnd relationship between the conﬁguration of chiral centers of the modiﬁer and the chirality of the product was one of the early tasks. It has been generally accepted that the conﬁguration of C8 or C8 and C9 atoms of the cinchona alkaloid molecule determines the product distribution.57,192 Changes in the sense of enantioselection were ﬁrst observed by Augustine et al. in 1993. Upon varying the DHCD/ catalyst ratio in the hydrogenation of EtPy over Pt/Al2O3 catalyst (S)-ethyl lactate formed at low modiﬁer concentrations and (R)-enantiomer at higher modiﬁer levels.67 However, the extent of inversion is within the experimental error. The other intriguing fact is that the ee values are extremely low. It is unprecedented that in this reaction the ee values are less than 20%. Analogous observation was found in gas phase hydrogenation of EtPy over Pt/SiO2 catalyst pre-modiﬁed with a series of C9 cinchona derivatives299 i.e., the sense of enantioselectivity has changed as a function of the modiﬁer concentration. The inversion of ee was found to be dependent on the nature of the substituent at C9.382 The appearance of inversion of enantioselectivity was observed due to the changes in the modiﬁer structure,192,202,205,238 variation of the solvent44,56,216,238,340 changes of the modiﬁer concentration206 and even changes of the substrate.210 Inversion has been reported 208 | Catalysis, 2010, 22, 144–278

Fig. 33 Hydrogenation of EtPy over Pt/Al2O3 in toluene, at pH2=1 bar. CD: cinchonidine, MeOCD: 9-O-methyl-cinchonidine, EtOCD: 9-O-ethyl-cinchonidine, TMSOCD: 9-O-trimethylsylil-cinchonidine, PhOCD: 9-O-phenyl-cinchonidine, XylOCD: 9-O-(3,5-dimethylphenyl)-cinchonidine, HFXylOCD: 9-O-[3,5-bis(triﬂuoromethyl) phenyl]-cinchonidine, NaphOCD: 9-O-naphthyl-cinchonidine. (Reproduced from ref. 202 with permission)

both on Pt56,202 and Rh254,256,351 catalysts. Summary of recent results was given in ref. 383. Certain C9 substituted derivatives of cinchonas such 9–Oalky,202 -aryl192,202,238 and -silyl202,206,238 derivatives of cinchonidine and Sharpless-ligands,192,206 furthermore b-isocinchonine56,197,384 (the rigid derivative of CN) have resulted in product with the opposite sense of ee than the underivatized alkaloid. Fig. 33 shows both the diminished enantioselectivity and its inversion with increasing bulkiness of the ether function of the modiﬁers.202 Upon using b-isocinchonine-Pt/Al2O3 catalyst system in the hydrogenation of EtPy ee decreases continuously and turn to opposite value with decreasing of pH (see Fig. 34). Investigation of unexpected inversion has given a new possibility for mechanistic studies. In the hydrogenation of ethyl-4,4,4-triﬂuoroacetoacetate over O-methylcinchonidine-Pt/Al2O3 catalyst system a signiﬁcant variation of ee value was observed with the conversion in the presence of even trace amounts of water or catalytic amounts of a strong acid.349,385 This issue has been discussed in Section 5.1. The explanation for the inversion of enantioselectivity is not completely clear at the molecular level. It is obvious that the inversion of ee can be related to increasing bulkiness of the substituent at C9 and the increased rigidity of the alkaloid molecule. However, these factors alone give not suﬃcient answer why O-pyridoxy derivative372 of CD does not lead to inversion in spite of the fact that O-phenyl and O-pyridyl moieties have almost identical van der Waals volumes. Further arguments are necessary to explain why a-ICN197 and a-isoquinine385 owning similar rigid structure as b-ICN shows no inversion. To understand the origin of inversion diﬀerent physical chemical methods have been applied. HPLC-MS and GC-MS measurements have shown that Catalysis, 2010, 22, 144–278 | 209

Fig. 34 Hydrogenation of EtPy to (R)- and (S)-ethyl lactate on b-isocinchonine modiﬁed platinum in toluene and AcOH mixtures. (Reproduced from ref. 56 with permission)

b-isocinchonine modiﬁer keeps its inner ether structure during the enantioselective hydrogenation.386 NMR measurements have proved that the hydrogenation of phenyl group in 9–O-phenyl-CD (PhOCD) is not responsible for inversion.202 From ATR-FTIR spectroscopic measurements and DFT calculations it has been concluded that the shape of the chiral space formed by the adsorption of PhOCD onto the metal is altered compared to that formed by CD238,387 (see Fig. 28). The phenyl group has a complex interaction with platinum; it can adsorb via its p system inﬂuencing the strength of adsorption of the modiﬁer. However, at the same time it can generate steric repulsion in the proximity of the chiral site.388 The presence of the phenyl group in (PhOCD) can also hurt the eﬃciency of the shielding eﬀect (see Section 8.3). Chiral pocket389 has been deﬁned as a physical space that is able to accommodate, via bonding and repulsive interactions, a pro-chiral adsorbate, and that is able to discriminate between its enantiomers. It was suggested that although CD and PhOCD display similar adsorption modes, the diﬀerent adsorption strengths and the change of the chiral pocket are suﬃcient to induce the inversion of enantioselectivity.387 The diﬀerent role of the bulky ether groups i.e. repulsion by the phenoxy and attraction by the 2-pyridoxy group explains the diﬀerent behaviour of these derivatives.372 Based on results obtained in the hydrogenation of EtPy197 and ketopantolactone in the presence of b-ICN381 in toluene interactions responsible for the inversion were proposed. The conformational rigidity of both the chiral modiﬁer and the reactant may inhibit the geometrical ﬁt of the three components (modiﬁer, reactant, and Pt), consequently the formation of the adsorbed intermediate responsible for enantio-selection is hindered. Beside the interaction between the nucleophilic N atom of the quinuclidine skeleton and the electrophilic C atom of the keto group of KPL or EtPy, H-bonded interaction veriﬁed by McBreen are also suggested.72 It is proposed that the sense of enantioselection is controlled by the conformation of the adsorbed reactant–chiral 210 | Catalysis, 2010, 22, 144–278

Fig. 35 Enantioselectivities of PPD hydrogenation in toluene over Pt/Al2O3 catalysts modiﬁed with: (K) – CD; ( ) CN; (’) – MeOCD; (&) – MeOCN; ( ) – PhOCD; (D) – PhOCN; (E) – TMSOCD; (}) – TMSOCN. (Reproduced from ref. 390 with permission.)

7

modiﬁer (1:1) complex which can be inﬂuenced by the solvent.384 It has been concluded that the adsorption mode and conformation of the modiﬁer during interaction with the substrate play the crucial importance in the change of the sense of enantio-selection.53 In a recent study the inversion of enantioselectivity was investigated in the hydrogenation of of 1-phenyl-1,2-propanedione (PPD) using diﬀerent O-ether derivatives of CD.391 The data conﬁrmed that the origin of inversion is related to depletion of the H-bonding interaction between the modiﬁer OH-group and the carbonyl group of the reactant rather than to a decreased population of the Open(3) conformation in the solutions of O-ether derivatives when compared with the solution behavior of the parent alkaloid. In another recent study diﬀerent O-ethers of CD and CN were used in both the enantioselective hydrogenation of PPD and the kinetic resolution of the 1-hydroxyketones formed over Pt catalysts390 Characteristic results for PPD are shown in Fig. 35. As emerges from these results all O-ethers showed inversion of enantioselectivity. Similar trend was also observed in the kinetic resolution of 1-hydroxyketones. These results are diﬀerent from that obtained in the enantioselective hydrogenation of EtPy, where inversion was observed only in case of large substituents. Another important ﬁnding is that in the presence of AcOH the above modiﬁers showed only very low enantio-diﬀerentiation ability (eeo5%). Inversion of enantioselectivity has also been observed by Garland et al. using a continuous-ﬂow three-phase reactor.299 5.6

Addition of other components

Several papers are devoted to the investigation of the inﬂuence of various additives on the behaviour of Pt/cinchona catalyst. These additives can be considered as modiﬁers either of the platinum or the support. Catalysis, 2010, 22, 144–278 | 211

5.6.1 Modiﬁcation of Pt by tin tetraethyl. In one of our earlier studies the Pt sites in a Pt/Al2O3 catalyst were modiﬁed by Sn(C2H5)4.358 This modiﬁcation is a two step anchoring type surface reaction.392 PtHa þ SnðC2 H5 Þ4 ! PtSnðC2 H5 Þð4xÞ þxC2 H6 PtSnðC2 H5 Þð4xÞ þ H2 ! PtSn þðn xÞC2 H6

ðPSCÞ

ð1Þ

ðSBAÞ

ð2Þ

In reaction (1) primary surface complexes (PSC) are formed. After tin anchoring the surface of Pt is covered by surface organometallic complexes with general formula of Pt-SnR3 or Pt-SnR2. These PSCs are stable at room temperature. Upon heating in a hydrogen atmosphere (see reaction (2)) they decompose with the formation of alkanes and stabilized bimetallic alloy (SBA) type surface species, i.e., supported Pt-Sn alloys are formed.393 Further details on this kind of modiﬁcation of Pt can be found elsewhere.394 Results in the enantioselective hydrogenation using tin modiﬁed catalysts are summarized in Table 9. As emerges from this Table the modiﬁcations resulted in both PSC and SBA forms of Pt-Sn/Al2O3 catalysts with diﬀerent Snanch/Pts ratios. The activity of modiﬁed catalyst was strongly altered by the amount and the type of surface species. However, over catalysts containing PSC (see Exps. 5, 13, 14, 15, 16 in Table 9) the ee values were almost constant, i.e. they were not aﬀected by modiﬁcation of Pt (ee=86–89%). Striking observation was that upon using PSC the hydrogenation reaction was completely blocked at relatively small Snanch/Pts ratios. This was attributed to the selective blocking of kink and corner sites responsible for

Table 9 Enantioselective hydrogenation using tin modiﬁed catalysts (Reproduced from ref. 358 with permission) Catalyst

Exp. No

Code No

Temperature of H2 treatment, 1C

1 2 3 4 5 6 7 8 9 10 11 10 11 13 14 15 16

Pt Pt Pt Pt PtSn-1 PtSn-1 PtSn-1 PtSn-1 PtSn-1 PtSn-2 PtSn-2 PtSn-3 PtSn-3 PtSn-4 PtSn-4 PtSn-5 PtSn-5

no 150 200 400 no 100 200 200 400 200 400 200 400 no no no no

212 | Catalysis, 2010, 22, 144–278

Sn/Pts, g

Rate of hydrogenation, mol (kgcat sec) 1

Optical yield, %

0.000 0.000 0.000 0.000 0.025 0.025 0.025 0.025 0.025 0.036 0.036 0.056 0.056 0.030 0.030 0.008 0.008

0.83 1.70 1.66 2.00 3.00 2.60 2.12 2.17 1.15 1.06 1.08 0.02 0.74 1.77 1.55 2.00 1.77

64 82 87 88 89 84 80 86 85 72 81 72 81 86 86 89 88

hydrogen activation by Sn(R)(4 x) moieties. The highest rates in the enantioselective hydrogenation were obtained upon using modiﬁed catalysts containing PSC. This ﬁnding was attributed to the suppression of the poisoning eﬀect induced by byproducts formed. There is one interesting remark: in a related study it was conﬁrmed that in this kind of surface modiﬁcation of Pt at low tin coverage tin prefers kink and corner sites.395 Based in this old results it can be concluded that the involvement of kink and corner sites of Pt in the ED step, as it has been suggested by diﬀerent authors,233,369,396 is highly questionable. In addition the above results can be considered as a ﬁrst real hint that the reaction rates and the ee values are not well correlated, i.e. relatively high ee values can be obtained even when the reaction rate is strongly suppressed. Unfortunately, this result was almost forgotten, as its conclusion did not ﬁt into the concept of ‘‘ligand acceleration model ’’. This model a priori suggests a deﬁnite relationship between reaction rate and enantioselectivity. 5.6.2 Addition of achiral amines. Based on the use of diﬀerent experimental methods182,189–191,397,398 it has been suggested that cinchona alkaloids can form dimers. As far as any dimer would decrease the virtual concentration of CD in the liquid phase attempts were done to use diﬀerent achiral tertiary amines (ATAs) with the aim to shift the equilibrium between the dimer and the monomer form of CD as shown in the following scheme: ½CD2 ! 2½CD

ð4Þ

½CD2 þ ATA ! ½CD ATA þ CD

ð5Þ

This concept has been tested in the enantioselective hydrogenation of EtPy and hexanedione.93,399–401 It is suggested that the modiﬁer in the form of a dimer is a spectator in the asymmetric hydrogenation reaction. Results in the presence of various ATAs at diﬀerent experimental conditions are summarized in Tables 10 and 11. These results clearly show the Table 10 Inﬂuence of diﬀerent added achiral tertiary amines on the enantioselective hydrogenation of EtPy in the presence of CD-Pt/Al2O3 catalyst system. (Reproduced from ref. 93 with permission) Achiral tertiary amines added

Concentration of achiral amines (M)

Rate constant, Rate constant, Enantioselectivity k2 (min 1) (eemax) k1 (min 1)

No TEA DABCO QND QND QNDc QNDc QNDc

– 1.2 10 5 1.2 10 5 1.2 10 5 1.2 10 5 1.2 10 5 6.0 10 5 1.2 10 4

0.0352 0.0407 0.0886 0.1289 0.1297 0.0832 0.1267 0.1219

0.0465 0.0676 0.1588 0.1645 0.1757 0.1346 n.m.d n.m.d

0.750 (0.714)b 0.841 (0.793)b 0.915b 0.898b 0.909b 0.926b 0.936b 0.946b

a Reaction conditions, solvent: toluene; reaction temperature: 23 1C; hydrogen pressure: 50 bar; [Etpy]0=1.0M, [CD]0=1.2 10 5 M, TEA – triethylamine; DABCO – 1,4-diazabicyclo[2.2.2]octane; QND – quinuclidine. bee values measured at the end of reaction. creactions carried out at 10 1C. d n.m.: not measurable.

Catalysis, 2010, 22, 144–278 | 213

Table 11 Inﬂuence of diﬀerent tertiary amines on the reaction rate and enantiomeric excess in the enantioselective hydrogenation of EtPy. (Reproduced from ref. 400 with permission) No 1 2 3 4 5 6 7 8 9 10 11 12

ATAs a

No No quinuclidinea quinuclidine Dabco MPD TEA Edcha Edipa 3-quinuclidinol Nob quinuclidineb

k1, min 1

k2, min 1

eemax

eeend

0.0045 0.0236 0.0077 0.0482 0.0486 0.0322 0.0300 0.0220 0.0243 0.0468 0.0409 0.0699

0.0068 0.0747 0.0109 0.0997 0.1267 0.0989 0.0905 0.0735 0.1280 0.0867 0.1268 0.1518

– 0.838 – 0.901 0.909 0.895 0.849 0.850 0.824 0.888 0.945 0.946

– 0.819 – 0.882 0.905 0.872 0.843 0.785 0.796 0.870 0.945 0.946

[EtPy]0=1.0 M, [CD]=1.2 10 5 M, ATA=6 10 5 M, T=20 1C, pH2=50 bar, solvent= toluene, coinjection of ATA, Dabco:1,4-diazabicyclo-[2.2.2]octane, MPD: 1-methylpiperidine, TEA: triethylamine, Edcha: N-ethyldicyclohexylamine, Edipa: N-ethyldiisopropylamine. k1, k2: ﬁrst order rate constants calculated from experimental points measured in the ﬁrst 10 minutes and between 25–60 minutes, respectively. a in the absence of cinchonidine. b solvent=1 M acetic acid in toluene.

strong eﬀect of ATAs in toluene, however no eﬀect has been observed in EtOH and AcOH. In the presence of quinuclidine unprecedented high ee values were obtained at low CD concentration (1.2 10 5). The ee value equal to 0.946 is close to those obtained in pure AcOH. Further evidence with respect to the involvement of alkaloids dimers in the ATA eﬀect was obtained upon comparison of CD with 9-methoxy-CD (CH3OCD) in the hydrogenation of EtPy. As far as at C-9 position the OH group is replaced by a methoxy one, (CH3OCD) cannot interact with ATA. It is the reason that no increase in the ee is observed in case of (CH3OCD).400 Results of kinetic studies were supported by results of circular dichroism spectroscopy. In table 12 the intensities of the Cotton shift around 237 nm are shown in the presence of various ATAs. This Cotton shift has been ascribed to the dimer form.192 In case of CH3OCD no similar Cotton shift is observed. The calculated De values well correlated with the ability of ATAs to increase the reaction rate and ee values. These results suggest that the ATA added into the solution of CD be involved in new type of solute–solute interaction. Summing up the ATA eﬀect the following conclusions can be drawn: (i) ATA eﬀect appears only at low concentration of CD; (ii) no ATA eﬀect in EtOH and AcOH; (iii) the ATA eﬀect depends on its concentration; (iv) ATAs containing bulky substituents show more pronounced eﬀect; (v) the OH group of CD is involved in the interaction with ATAs. 5.6.3 Addition of nitrogen containing aromatic and condensed aromatic compounds. In our recent study402 the inﬂuence of the addition of various nitrogen containing aromatic and condensed aromatic compounds was studied. The aim of these studies was testing of the validity of the 214 | Catalysis, 2010, 22, 144–278

Table 12 Eﬀect of ATA on the circular dichroism data of cinchonidine (Reproduced from ref. 400 with permission)

No

ATA added

Concentration of ATA ( 10 3 M)

1 2 3 4 5 6 7 8 9 10 11 12 13

no no quinuclidine quinuclidine quinuclidine quinuclidine quinuclidine Dabco Dabco Dabco TEA TEA TEA

– – 0.4 0.8 2.0 4.0 8.0 2 4 8 2 4 8

ATA-CD molar ratio

solvent

De (M 1cm 1)

– – 1 2 5 10 20 5 10 20 5 10 20

ethanol CH2Cl2 CH2Cl2 CH2Cl2 CH2Cl2 CH2Cl2 CH2Cl2 CH2Cl2 CH2Cl2 CH2Cl2 CH2Cl2 CH2Cl2 CH2Cl2

– 0.94 0.91 0.74 0.70 0.61 0.49 0.92 0.69 0.39 0.94 0.90 0.79

[CD]=4 10 1 M, T: 25 1C, cell length: 0.2 cm, time mode detection, wavelength: 237.6 nm, Dabco: 1,4-diazabicyclo-[2.2.2]octane, TEA: triethylamine.

Table 13 Eﬀect of Q on the reaction rate and enantioselectivity in the enantioselective hydrogenation of EtPy. (Reproduced from ref. 402 with permission) No

[CD], 10 5 M

[QN], 10 5 M

k1, min 1

k2, min 1

eemax

eeend

1 2 3 4 5a 6 7 8 9 10

0.6 0.6 0.9 0.9 1.2 1.2 1.2 1.2 6.0 6.0

no 6.0 no 6.0 no 1.2 6.0 12.0 no 6.0

0.027 0.040 0.032 0.045 0.034 0.054 0.056 0.059 0.065 0.060

0.015 0.069 0.058 0.090 0.073 0.114 0.118 0.123 0.187 0.140

0.573 0.832 0.719 0.867 0.830 0.880 0.874 0.880 0.894 0.898

0.235 0.699 0.575 0.813 0.798 0.847 0.860 0.869 0.894 0.898

[EtPy]0=1 M, treact=90 min, catalyst: 0.125 g, 5 wt% Pt/Al2O3, solvent: toluene, mode of introduction: Pr-I for Q followed by Inj-I of CD, conversionW99%. a average of ﬁve parallel experiments.

‘‘surface model.’’ The ‘‘surface model’’ assumes that CD adsorbs by its aromatic quinoline ring almost parallel to the Pt surface. The condensed ring system has been considered as the anchoring site (AS) of the modiﬁer.46 Based on this view it was suggested if condition of competitive adsorption between CD and condensed aromatic compounds can be established, the number of ‘‘chirally modiﬁed sites’’ should decrease resulting in deﬁnite loss of enantioselectivity. Table 13 shows the inﬂuence of added quinoline (Q) on the reaction kinetic and ee values. These results unambiguously show that the addition of quinoline increases both the rate and the ee values. The eﬀect of quinoline is very pronounced at low concentration of CD, while upon increasing the Catalysis, 2010, 22, 144–278 | 215

A

ee

1.0

B

C

1.0

1.0

0.8

0.8

0.8

0.6

0.6

0.6

0.4

0.4

0.4

0.2

0.2

0.2

0.0

0.0

0.0 0.0 0.2 0.4 0.6 0.8 1.0

0.0 0.2 0.4 0.6 0.8 1.0

0.0 0.2 0.4 0.6 0.8 1.0

Conversion

Conversion

Conversion

Fig. 36 Inﬂuence of CD concentration on the hydrogenation of EtPy in the presence of Q. [EtPy]0=1.0 M, PH2=50 bar, solvent: toluene, 500 rpm, catalyst: 0.125 g, 5 wt.% Pt/Al2O3 (Engelhard 4759), mode of introduction: Pr-I for Q followed by Inj-I of CD; A: [CD]0=0.6 10 5 M; B: [CD]0=1.2 10 5 M; C: [CD]0=6.0 10 5 M, E – no Q; & – 6 10 5 M Q. (Reproduced from ref. 402 with permission)

Table 14 Hydrogenation of EtPy over cinchona-Pt/SiO2 catalyst system in the presence of condensed aromatic compounds and aromatic nitrogen bases. (Reproduced from ref. 402 with permission) No

Additives

k1, min 1

k2, min 1

convend, %

eemax

eeend

eeend/eemax

1 2 3 4 5 6 7 8

– Acridine Quinoline Pyridine 4-Picoline Naphthalene Antracene Pyrene

0.031 0.045 0.048 0.078 0.042 0.037 0.033 0.039

0.053 0.047 0.078 0.069 0.059 0.044 0.049 0.048

98.8 98.4 99.0 99.4 99.5 97.7 98.3 98.4

0.666 0.690 0.686 0.585 0.575 0.625 0.620 0.619

0.343 0.434 0.468 0.446 0.447 0.323 0.323 0.337

0.515 0.629 0.682 0.762 0.780 0.517 0.521 0.544

[EtPy]0=1 M, [CD]=1.2 10 5 M, [Additive]=1 10 4 M, treact=90 min, catalyst: 0.07 g 2.7 wt% Pt/SiO2, solvent: toluene, mode of introduction: Pr-I for additives followed by Inj-I of CD.

concentration of CD the eﬀect disappears. Due to the presence of quinoline unusually high ee values were obtained in toluene at [CD]0=1.2 10 5 M. Figure 36A–C shows the ee-conversion dependencies at diﬀerent CD concentration. Characteristic feature of these dependencies is that at low CD concentration ee decreases at high conversion (see Fig. 36A). The loss of ee at high conversion is attributed to the loss of CD due to the hydrogenation of its quinoline ring. However, this decrease is strongly suppressed by the addition of quinoline; consequently the results indicate that quinoline replaces CD from the Pt surface. This replacement reduces the chance of CD to be hydrogenated by its quinoline ring. Table 14 shows the inﬂuence of various nitrogen containing and condensed aromatic compound on the rate and enantioselectivity. These results were obtained on highly dispersed Pt/SiO2 catalysts. Over this catalyst the ring hydrogenation of the quinoline ring was relatively fast. It is the reason that over this catalyst the ee decreases with conversion. It is reﬂected by the low value of eeend/eemax. The results show that none of the additives used (see Table 14) resulted in measurable rate decrease. However, substantial rate increase was observed 216 | Catalysis, 2010, 22, 144–278

in the presence of acridine, quinoline and pyridine. Two classes of condensed aromatic compounds can be diﬀerentiated: (i) nitrogen containing one resulting in increased eeend/eemax values and (ii) condensed aromatic compounds having no inﬂuence on the eeend/eemax values. Summing up these results it was concluded that in the enantioselective hydrogenation of EtPy achiral condensed aromatic N-bases, as additives, are able to increase both the enantioselectivity and the reaction rate. Based on the concept of ‘‘chirally modiﬁed sites’’ the co-presence of CD and quinoline over Pt sites can be considered as a competition. The consequence of this competition is the decrease of the number of sites involved in ED. Therefore, one would expect a decrease in the ee values. However, it is not the case, the ee increases when the CD/quinoline ratio is properly chosen and the CD concentration is low. The observed eﬀect appeared both on alumina and silica supported Pt catalysts and was found to be strongly concentration dependent. However, according to the ‘‘Surface model’’ the co-adsorption of these compounds should result in a decrease in the enantioselectivity, what is not observed in our study. Consequently, our results might indicate that the ‘‘surface model’’ needs some corrections. 5.6.4 Modiﬁcation of the support. There are only scarce data on the modiﬁcation of the support. In one of our studies the inﬂuence of the modiﬁcation of alumina support by alkyl silanes was investigated. After dehydroxilation of the support at 400 1C it was modiﬁed by diﬀerent alkyl silanes resulting in anchored –Si(CH3)3, or –Si(CH3)2C8H17 moieties274. The rate of this anchoring type surface reaction can be controlled by the concentration of the modiﬁer, the temperature of anchoring reactions and the length of the R group in the alkyl silanes. Catalysts modiﬁed in this way were used in the enantioselective hydrogenation of EtPy in the presence of CD. The above modiﬁcation was not beneﬁcial for the above reaction as the modiﬁed catalysts showed pronounced decrease in reaction rates and slight loss in enantioselectivities compared to the unmodiﬁed Pt/Al2O3. These results are shown in Figs. 37A and B. The catalytic performance of these modiﬁed catalysts was

Fig. 37 The eﬀect of the surface coverage of CH3(CH2)7Si(CH3)3 moieties on the reaction rate (A) and the enantioselectivity (B). (Reproduced from ref. 274 with permission)

Catalysis, 2010, 22, 144–278 | 217

Table 15 Inﬂuence of the catalytic performance of modiﬁed catalysts as a function of the temperature of pre-activation. (Reproduced from ref. 274 with permission) Experiment No.

Temperature of preactivation (1C)

Rate constant, k1 (min 1)

ee (%)

1 2 3 4 5 6

150 250 400 400, blanka 400b 400, parent

o0.001 0.012 0.020 0.044 0.054 0.057

58.3 77.2 84.7 85.0 84.3 86.3

Catalyst tested: No.10 (see Table 1 in Ref. 274); a Catalyst No. 1 (see Table 1 in Ref. 274). b Modiﬁed catalyst treated in air at 300 1C prior its preactivation in a hydrogen atmosphere. c Catalyst without modiﬁcation.

signiﬁcantly altered upon using various pre-activation procedures as shown in Table 15. The removal of the anchored –O–Si(CH3)2R moieties the original activity and enantioselectivity was restored completely. The catalytic performance of these modiﬁed catalysts was signiﬁcantly altered upon using various pre-activation procedures as shown in Table 15. The removal of the anchored –O–Si(CH3)2R moieties the original activity and enantioselectivity was restored completely. The behaviour of this type of modiﬁed catalysts was explained by the following phenomena: (i) partitioning or retention of CD or the [substrate-CD] complex by anchored –Si(CH3)2R moieties, and (ii) decreasing the mobility of CD or the [substrate-CD] complex in the boundary layer. We believe that the results obtained in the above study provided further indirect evidences that interactions in the liquid phase play a very important role. Results obtained in this study strongly indicate that in this enantioselective hydrogenation reaction the enantio-diﬀerentiation cannot be attributed exclusively to the interaction between the half-hydrogenated substrate and CD on the Pt surface. 5.6.5 Modiﬁcation of Pt by other components. These studies were performed by English groups. These groups use the classical aerobic Premodiﬁcation procedure (see Section 4.4), In the enantioselective hydrogenation of MePy or butane-2,3-dione in the presence of cinchonidine-modiﬁed platinum catalysts it was shown that at the catalyst preparation stage, the co-adsorption of the alkaloid with a strong co-adsorbate has a strong positive eﬀect.278 One of these coadsorbates was oxygen or air dissolved in reactant and solvent. In addition acetylene, methyl acetylene and butadiene appeared to be eﬀective co-adsorbates. It was suggested that in the absence of a strong co-adsorbate the surface is poisoned by cinchonidine. It has been shown that the modiﬁcation under methylacetylene provides reaction rates and ee values excess under standard conditions (10 bar, 293 K) that are comparable to, or higher than those obtained with normal aerobic modiﬁcation. The importance of surface morphology of small supported Pt particles was conﬁrmed in refs. 233, 403. In these studies Pt/C and Pt/SiO2 catalysts 218 | Catalysis, 2010, 22, 144–278

Table 16 Variation of activity (rmax), enantiomeric excess (ee) and HMMP yield in reactions over Pt/graphite, Pt–Bi/graphite and Pt–S/graphite modiﬁed by cinchonidine (CD) and quinuclidine (QND). (Reproduced from ref. 232 with permission) Surface

Alkaloid modiﬁer rmax (mmol h 1gcat 1) ee (%(R)) HMMP yield

Pt Pt Pt Pt Pt-Bi ((YBi)ch=0.35) Pt-Bi ((YBi)ch=0.35) Pt-Bi ((YBi)ch=0.35) Pt-S ((YS)ch=0.19) Pt-S ((YS)ch=0.19) Pt-S ((YS)ch=0.19)

None CD QND 1:1 CD:QND CD QND 1:1 CD:QND CD QND 1:1 CD:QND

–b 850 440 1205 1350 1710 4600 440 310 645

0 41 0 37 35 0 15 52 0 51

2c 100 40 – 49 36 – 109 – –

a Conditions: 65 mmol ethyl pyruvate, 0.17 mmol CD and/or 0.17 mmol QND, 0.25 g catalyst, 12.5 ml dichloromethane, 30 bar hydrogen, 293 K, 1000 rpm. b For this reaction rinitial=24 mmol h 1gcat 1. c ConversionW20%.

modiﬁed by bismuth and sulphur and were characterized by electrochemical methods.404 Their key ﬁndings are summarized in Table 16. The authors’ main results and conclusions were as follows: A cyclic voltametric analysis indicated that in poisoning by Bi king sites, while in poisoning by sulfur terrace sites are involved The reaction rates increased in bismuth modiﬁed catalysts; while decreased in sulfur modiﬁed one. The formation of polymeric residues were strongly reduced over Bi modiﬁed catalysts. Both the enantioselective hydrogenation and the formation of polymeric residues are formed on king sites. RE is now attributed to reaction occurring at a normal rate at an enhanced number of sites, not (as previously proposed) to a reaction occurring at an enhanced rate at a constant number of sites. It has to be stressed out that in this study EtPy was used without any puriﬁcation. It is the reason for extremely low rates in the racemic hydrogenation (see Table 16). This fact strongly questions both the results and the conclusions. There is an additional serious drawback, i.e., the lack of information on racemic hydrogenation over catalysts modiﬁed by Bi. In this respect it has to mention that the adsorption of Bi on the Pt can give diﬀerent species, including metallic and ionic one.405 We suggest the rate increase and the decrease of the ee over Bi modiﬁed catalysts is due to the acceleration eﬀect in the racemic hydrogenation by bismuth cations created over the Pt surface. In this respect we should like to revert to our results on tin modiﬁed catalysts (see Section 5.6.1). This method clearly indicated that the involvement of kink sites in the ee is excluded as tin is located on the kink site and ee was independent on the amount of tin anchored to the platinum, while the rate showed a strong dependence. Catalysis, 2010, 22, 144–278 | 219

6. 6.1

Spectroscopic investigations NMR

NMR techniques related to substrates, reaction products and chiral modiﬁers have widely been used in the investigation of enantioselective hydrogenation of activated ketones. It has also been applied for the determination of by-products, alternative reaction routes and intermediate complexes and attempts were also made to use NMR for elucidation of diﬀerent hypothetical reaction-mechanisms. To study the directing eﬀect of ester group338 or triﬂuoromethyl group235 of the substrate a series of new compound were prepared and identiﬁed by NMR. It was also applied as a tool to check the purity of substrate.82 In many cases NMR gives an opportunity for identiﬁcation of reaction products.203,210,235,331,336,338,406 It is known long before that cinchona alkaloids are extremely active chiral modiﬁers in various organic reactions. Mapping the role of diﬀerent structural elements of cinchonas in the enantio-selection several derivatives, especially C9 substituted cinchonas,202 inner ethers,407 N-alkyl derivatives314 have been prepared and checked by NMR,44,238,256,408 Upon identiﬁcation of the structure of ether derivatives of cinchonidine such as 9–O-phenyl-202,206 9–O-pyridyl-,372 9–O-sylil-cinchonidine202 etc. NMR was an indispensable tool. Beside of the aforementioned chiral modiﬁers, the structure of several cinchona analogues, i.e. amines and amino alcohols44,378 and aryl alcohols226 prepared for chiral template in the Orito’s reaction has also been conﬁrmed by NMR. Chiral modiﬁers itself very often suﬀer changes during the enantioselective hydrogenation. To follow the conversion of 9–O-pyrydil-cinchonidine372 and the saturation of naphthalene ring of 1-naphthyl ethylamine derivatives208,212 NMR was applied as well as to check the resistance of phenyl group in 9–O-phenyl-cinchonidine203 and the stability of methoxycinchonidine372 and isocinchonines.351 NMR analysis of the reaction mixture showed that 1-naphthyl ethylamine derivatives is quantitatively consumed during the hydrogenation reaction and converted to the secondary amine.212 It has been shown by NMR that quaternary ammonium derivatives of CD as new chiral modiﬁers remained stable during hydrogenation.200 Formation and structure of hexahydro-cinchonidines and hexahydro-cinchonines has been investigated by NMR.271,409 According to NMR analysis, at 36% saturation of the quinoline rings of CD in acetic acid the ratio of homoaromatic and heteroaromatic hydrogenation products was 2.5 to 1.269 Upon hydrogenation of b-triﬂuoro ketones the sense of enantioselectivity changes when the polarity of the solvent changes. The phenomenon was explained by the shift of keto-enol equilibrium conﬁrmed by NMR.44,385 In alcoholic solvent ethyl-4,4,4-triﬂuoroacetoacetate has shown a reaction route via semi-ketal.348 Semi-ketal was also detected by use of NMR in other cases.331,410 IR and NMR experiments have revealed that the enantioselective hydrogenation of EtPy in nonacidic solvents is complicated by the simultaneously occurring self-condensation (aldol reaction) of the reactant.82 In the hydrogenation of ketopantolactone GC and NMR results 220 | Catalysis, 2010, 22, 144–278

has shown that no by-product formation appears.76 The reaction pathways were followed by NMR in the enantioselective reduction of isatin derivatives over cinchonidine modiﬁed Pt/alumina.411 In the study of hydrogenation of 1,1,1-triﬂuoro-2,4-diketones strong acid-base interaction has been revealed by NMR. The high chemo- and enantioselectivities in the above reaction were attributed to the formation of an ion pair involving the protonated quinuclidine part of the chiral modiﬁer and the enolate form of the substrate.406 In the hydrogenation of KPL the formation of dimer was conﬁrmed by NMR.209,323 Recently a mechanistic model involving nucleophilic catalysis and zwitterionic adduct formation between a cinchona alkaloid and activated ketone has been suggested.412 Upon investigation of hydrogenation of ﬂouoroketones the formation of the ionic species in the solution was studied by 13C-NMR spectroscopy. Triﬂuoromethylcyclohexyl ketone was used as a strong electrophile agent and the complex was created by the addition of excess tertiary amine quinuclidine. The adduct formation was studied in two diﬀerent solvent systems such as deuterated chloroform and acetone but the formation of the zwitterionic species was observed only in acetone.216 13C-NMR conﬁrmed the existence of other adducts of the zwitterionic type.360 However an NMR study has indicated that the zwitterions model is probably based on erroneous interpretation of the experimental data; the NMR spectra that had been reported for zwitterion formation may arise from an aldol addition the a,a,atriﬂuoromethyl ketone and the solvent acetone, and the reaction is catalyzed by the tertiary amine used as a model for the chiral amine modiﬁer.413 Upon using NMR technique it was veriﬁed that the enol form of EtPy is not the reacting species, but under condition of enantioselective hydrogenation deuterium exchange takes place not only at the quinoline ring, but at C9 carbon atom, too.414 6.1.1 Conformation analysis of the modiﬁer. The study of the conformation of cinchona alkaloids investigated by NMR has been brieﬂy mentioned in Section 2.3. The question of which conformer of cinchonidine is involved in the enantio-diﬀerentiation step is regarded as a key issue. Baiker el al. have used ab initio calculations and NMR measurements to investigate the conformers of (dihydro)cinchonidine in diﬀerent solvent (such as benzene, toluene, ethyl ether, acetone, etc.).88 The existence of a given conformer has been rendered by nuclear Overhauser enhancement spectroscopy. NOESY experiments have suggested that Open (3) and Closed (1) and Closed (2) conformers appears. This observation is in qualitative agreement with their calculations. The dihedral angles for different conformers have been calculated by ab initio methods. The measured coupling constants (13JH8H9(exp)) and the dihedral angles by applying the Karplus equation415 have given possibility for calculation of coupling constants of diﬀerent conformers (13JH8H9(i)). The above method is limited to the determination of only two conformers from one coupling constants measured, however the dihedral angle has been found very similar for the two closed conformers of CD, so the population of Open (3) and the sum of populations of closed conformers has been possible to calculate. The results are represented in Table 17. Catalysis, 2010, 22, 144–278 | 221

Table 17 Vicinal 3JH8H9 coupling constants for cinchonidine and derived population of conformer open (3) in diﬀerent solvents. (Reproduced from ref. 88 with permission) Solvent

hT

3

Popen

benzene toluene ethylether tetrahydrofurane acetone dimethylfomamide dimethylsulfoxide water ethanol

2,28 2,34 4,3 7,6 20,7 36,7 40 78,5 24,3

5.0 4,1 4 4,7 6,4 7 7,5 7,2 3,5

0.58 0,7 0,71 0,62 0,4 0,33 0,27 0,3 0,77

JH8H9

(3)

PClosed 0.42 0,3 0,29 0,38 0,6 0,67 0,73 0,7 0,23

For open(3) 3JH8H9 is calculated as 1.7 Hz. For Closed (1) and Closed (2), respectively, JH8H9 is calculated as 9.6 and 4 Hz. In this determination of P closed a value of 9.6 was used.

a 3

According to the above approach the conformer Open (3) has been found the most stable one.88 Based on parallel solvent dependence of ee and population of Open (3) it has been suggested that Open (3) conformer plays crucial role in the asymmetric induction. In case of 9-deoxy-CD derivatives similar method has been used.416 The coupling constants of 2-phenyl-9-deoxy-10,11dihydrocinchonidine (13JH9H8a (5.0 Hz); 13JH9H8b (9.3 Hz)) is similar to that of 9-deoxy-10,11-dihydrocinchonidine (5.5 and 8.5 Hz) in CDCl3 indicating that the relative stability of conformers is also similar. NMR experiments and ab initio calculations revealed that conformer Closed (1) is stabilized relative to Open (3) when going from CD to 9-deoxy-10,11-dihydrocinchonidine. In a recent study the eﬀect of the protonation on the conformation of CD was investigated.417 It was shown that protonation strongly hinders the rotation around the C4 0 –C9 and C9–C8 bond. Structures and conformational behaviour of several cinchona alkaloid O-ethers in solution (NMR and DFT) were also investigated.391 It was demonstrated that the conformation found to be abundant in the liquid phase has no direct correlation with the enantioselectivity of the PPD hydrogenation reaction. The authors concluded that the driving force for production of one of the enantiomers in excess is due to the speciﬁc adsorption of the modiﬁer on the catalyst surface, a phenomenon that does not correlate with the population of the conformers in the liquid phase. 6.1.2 Substrate-modiﬁer interaction. In an early work NMR measurements have already shown an interaction between CD and EtPy in the liquid phase.84 It was shown that in CD3OD in the presence of 0.15 M MePy the characteristic doublet of CD at 5.65 ppm was shifted to 5.85 ppm and a new small singlet was observed at 6.0 ppm. Upon increasing the concentration of MePy to 0.6 or 1.0 M the doublet vanished and only the new singlet at 6.0 ppm was found. More noticeable shift of the C(9) proton, up to 6.3–6.4 ppm with a formation of a singlet was observed in neat CD3COOD or if small amount of CD3COOD was added into the solution of CD in C6D6. These NMR results suggested that the torsional angle between the hydrogen atoms at C(8) and C(9) carbon atom of CD has been changed resulting in a new conformer of CD. 222 | Catalysis, 2010, 22, 144–278

In Pt-catalysed hydrogenation of 1,1,1-triﬂuoro-2,4-diketones the combined catalytic, NMR and FTIR spectroscopic, and theoretical study revealed that high chemo- and enantioselectivities are attributed to the formation of an ion pair involving the protonated amine function of the chiral modiﬁer and the enolate form of the substrate.406 On the basis of NOE studies and theoretical calculations related to the hydrogenation of ketopantolactone in the presence of the (R,R) and (R,S) diastereomers of a new chiral modiﬁer, pantoyl-naphthylethylamine, diﬀerent properties of the above diastereomers were investigated, in particular the eﬀect of acid on the modiﬁer structure.209 The results indicated that in case of the (R,R)-diastereomer in apolar solvent, a loose, extended structure changes to a compact one via an additional intra-molecular hydrogen bond, resulting in a more suitable ‘‘chiral pocket’’ available for the reactant on the Pt surface. Standard 2D NMR spectroscopic methods and diﬀusion-ordered NMR spectroscopy combined with theoretical calculations has been used to verify the formation of supramolecular complexes between the pairs O-methylcinchonine–ketopantolactone (KPL) and b-isocinchonine–KPL.418 When O-methyl-cinchonine or b-isocinchonine and KPL were mixed in dry deuterobenzene solution, time-dependent chemical shift changes for the cinchonas and new signals for KPL bound to the modiﬁer have been detected. The spatial pattern of the chemical shift diﬀerences and the conformations of the modiﬁers determined by NOESY demonstrated that the substrate binding occurs at the quinuclidine N atom, H9, and the quinoline H5 0 region for O-methyl-cinchonine (H8 and H5 0 for b-isocinchonine. Based on diﬀusion measurements hydrodynamic radii has been estimated which has proved the co-diﬀusion of the cinchonas and KPL in a complex. The results have shown that not only 1:1, but also 2:1 cinchona-KPL complexes must be taken into account. NMR evidences has also been found for the correlation between the solution-state concentration of the nucleophilic 1:1 modiﬁer-substrate complex and the ee on enantioselective hydrogenation of KPL using Pt–b-isocinchonine chiral catalyst.418,419 6.2

Circular dichroism

Vibrational circular dichroism (VCD) is a useful tool to determine the absolute conﬁguration of the enantiomer produced in excess in an enantioselective reaction when reference data on the enantiomer are not available. The absolute conﬁgurations of the enantiomers can be obtained by comparing the theoretically calculated VCD spectrum of one enantiomer with the experimental VCD spectrum of the product of the asymmetric reaction. It is important to know that VCD signal is about three orders of magnitude less intensive than the corresponding signal in the ordinary transmission spectrum.235 This method has been successfully applied for the determination of product alcohols in the studies related to the directing eﬀect of triﬂuoromethyl group235 or ester group338 of the substrates. It was also used upon investigating CD modiﬁed Rh/alumina catalyst420 in the hydrogenation of various aromatic ketones possessing an a-hydroxy or a-methoxy group and, in case of the enantioselective reduction of isatin derivatives over CD modiﬁed Pt/alumina.411 Catalysis, 2010, 22, 144–278 | 223

Theoretical (DFT) and VCD spectroscopy study has been applied for the conformational analysis of the synthetic chiral modiﬁer 9-O-phenyl-CD.421 According to these results 9-O-phenyl-cinchonidine behaves similarly to CD and shows four main conformers, denoted as Closed (1), Closed (2), Open (3), and Open (4). A combined theoretical-experimental VCD spectroscopy approach has given possibility to increase the spectroscopic sensitivity toward changes in the distribution of conformers upon changing the solvent polarity. The VCD spectra conﬁrm that Open (3) is the most stable conformation in CCl4. Changing from CCl4 to CDCl3 the equilibrium between the conformers does not change signiﬁcantly. Upon increasing solvent polarity besides similar non-coordinating properties the fraction of Closed (2) species increases considerably. Relating the conformational results to the enantio-diﬀerentiation shown by this modiﬁer (9-O-phenyl-CD) in the platinum-catalysed asymmetric hydrogenation of KPL the inversion of the sense of enantio-diﬀerentiation observed cannot be traced to the conformational behaviour.421 HPLC and UV-vis/circular dichroism299 has been used to assess conversion and selectivity in chiral ﬁxed bed reactor for stereoselective heterogeneous catalysis.422 The UV-CD method and the HPLC-CD method have been used to simultaneously determine ee values and concentration of each enantiomer.423 Tungler and coworkers have described that the circular dichroism spectrum of dihydrovinpocetine changes upon addition of both isophorone as well as EtPy indicating interactions between these two substrates and the chiral modiﬁer.68 The similar method has been applied in case of (S)-proline based chiral modiﬁers.224 The circular dichroism spectra of CD in toluene has been found to change by addition of EtPy as shown in Fig. 38.93 Addition of EtPy to cinchonidine in chloroform has also resulted in changes in the circular dichroism spectra of cinchonidine, although these changes were less pronounced than those in toluene. The above results strongly indicate that there is an interaction between CD and EtPy in the liquid phase. The results of circular dichroism spectroscopy401 have provided further proof for dimer formation of CD in liquid phase. These results are related to the addition of ATAs discussed in Section 5.6.2.

Fig. 38 The circular dichroism spectra of cinchonidine in toluene and its change by addition of EtPy. (from ref. 93 with permission)

224 | Catalysis, 2010, 22, 144–278

Fig. 39 Circular dichroism spectra of cinchonidine in the presence of diﬀerent amount of quinuclidine [CD]0=1.2 10 5 M. (from ref. 93 with permission)

The Cotton shift of CD was measured at diﬀerent [CD]0/[QN]0 ratios. The corresponding circular dichroism spectra are shown in Fig. 39. The Cotton shift around 235 nm appeared to be very sensitive to the amount of quinuclidine added. This Cotton shift is related to the dimer form of CD.191,400 Analogous results using MeO-CD have shown that this alkaloid cannot form a dimer. 6.3

Characterization of the solid and the solid-liquid interface

6.3.1 Introduction. Various surface science techniques were used so far to investigate the interaction of substrates and modiﬁers or the substratemodiﬁer complex with the Pt surface. Most of these methods are using conditions (often high-vacuum) far from those applied in catalytic hydrogenations. For this reason, although some important details of the adsorption behaviour of CD and substrates have been revealed, the results have to be treated with certain precaution. Surface characterization methods applied so far are as follows: (i) nearedge absorption ﬁne structure spectroscopy (NEXAFS),91 (ii) X-ray photoelectron spectroscopy424,425(XPS), (iii) low-energy electron diﬀraction (LEED,425 (iv) scanning tunnelling microscopy (STM),425–427 (v) reﬂectionabsorption infrared spectroscopy (RAIRS),428,429 (vi) surface-enhanced Raman spectroscopy (SERS),92,430 (vii) attenuated total reﬂection infrared (ATR-IR) spectroscopy,301,431,432 and (viii) electrochemical polarization.241 It had to be emphasized that only ATR-IR spectroscopic method240,232 and its combination with modulation excitation spectroscopy (MES) in a ﬂow-through cell433 can be considered as appropriate methods approaching almost real in situ conditions. It has to be emphasized that above two techniques have the advantage to obtain information about adsorption processes at the solid-liquid interface. In this respect it is important to mention that exact vibrational assignments for adsorbed CD on Pt surface using combination of experimental vibrational spectroscopic measurements and ab initio computational methods were also reported.92,434,435 Recently a Catalysis, 2010, 22, 144–278 | 225

new method for in situ spectroscopic investigation of heterogeneous catalysts and reaction media at high pressure have been developed.436 6.3.2 Investigation of the substrates. The adsorption of EtPy on Pt(111) at low temperature was investigated by XP and UP spectroscopy.424 The results indicated that EtPy adsorbed strongly to the Pt. The ketone carbonyl is more strongly involved in the chemisorption bond than the carboxyl one. Further analysis showed that EtPy is predominantly adsorbed in a tilted rather than a completely ﬂat mode. The behaviour of EtPy during adsorption on alumina-supported platinum ﬁlms and on a commercial 5 wt% Pt/ Al2O3 catalyst has been studied in the absence and presence of coadsorbed CD.91 The in situ ATR-IR study at room temperature using hydrogen-saturated CH2Cl2 as solvent demonstrated that upon adsorption on the Pt EtPy decomposes with the formation of strongly adsorbed CO and other organic residues. The presence of CD (10 4 M) strongly decrease the rate of decomposition of EtPy. Upon using STM method self-condensation of MePy over Pt surface was observed.308 This reaction took place in the absence of cinchona modiﬁer at low hydrogen coverages. Based on this ﬁnding new set of side reactions with the involvement of MePy was proposed and conditions to avoid the byproduct formation was discussed (see Section 5.1). Side reactions of EtPy during enantioselective hydrogenation on Pt/ Al2O3 have been investigated using in situ ATR-IR and ex situ DRIFT.309 The studies revealed that EtPy can decomposed and polymerize (aldol condensation) under conditions of hydrogenation. These side reactions take place both on the Pt site and the Al2O3 support. Based on analysis of the RAIRS spectra of MePy it has been shown that on Pt(111) at room temperature MePy undergoes ‘‘surface mediated enol formation’’ leading to an assembly of H-bonded superstructures.310 The decrease of the temperature and the use of low background hydrogen pressure suppress these surface reactions. In a recent study adsorption and reaction of EtPy on Pt/g-Al2O3 was studied by IR spectroscopy.311 Several side reactions of EtPy were detected. These results were discussed in Section 5.1. The adsorption mode of MePy and EtPy was studied under ultra-high vacuum conditions on Pt single-crystal surfaces using X-ray and UV photoelectron spectroscopies (XPS and UPS),424 NEXAFS,96 and (RAIRS).437 The results indicated that alkyl pyruvates adsorbs via lone pair-metal interaction of both carbonyl groups, i.e., in cis conformation with their plane oriented normal or tilted with respect to the surface. At high coverage, a minority species was assigned to an Z1-trans conﬁguration.437 The coadsorption of hydrogen resulted in signiﬁcant inﬂuence on the adsorption of alkyl pyruvates by lowering the tilting angle of the adsorbed species438 and suppressing surface polymerization of the adsorbed enediolate species observed earleir.308 6.3.3 Investigation of the modiﬁers. Adsorbed forms of cinchona alkaloids display diﬀerent IR spectra from each other and from the solution form of the alkaloids. This fact makes vibrational spectroscopy a suitable 226 | Catalysis, 2010, 22, 144–278

method to investigate the adsorption of cinchona alkaloids on metal surfaces. Although this method seems to be very powerful it could not answer the key question, namely, which of these species interacts with the substrate in the enantio-diﬀerentiating step. The ﬁrst information that the mode of adsorption of CD on Pt depends on its coverage was obtained by using in situ ATR-IR spectroscopy.240 At low coverage the ﬂat mode, while at high one the tilted mode prevails. Further study revealed that abstraction of hydrogen form the quinoline ring can also take place resulting in a so called a-H abstracted form.431 Study on Pt/Al2O3 in the presence of an organic solvent and hydrogen revealed three diﬀerent adsorption modes of CD as shown in Fig. 40. Infrared spectroscopy (IR), Raman spectroscopy, surface-enhanced Raman scattering (SERS) and reﬂection–adsorption infrared spectra (RAIRS) studies428,431,434 ratify the results discussed above (Fig. 40). The adsorption of CD on Rh/Al2O3 has also been investigated using ATR-IR spectroscopy. The adsorption appears to be more complex than that observed on Pt and Pd. Strongly adsorbed ﬂat form was observed on Rh when adsorption was performed in the absence of dissolved hydrogen. This form is responsible for the fast hydrogenation of the quinoline ring and does not allow the detection of the ﬂat form in the presence of dissolved hydrogen.270 Contrary to Pt it has been discovered that on Rh hydrogenation of the heteroaromatic part of the quinoline ring takes place. Adsorbed CD in the ﬂat geometry is the intermediate of the hydrogenation reaction,

Fig. 40 Suggested adsorption mechanism of cinchonidine on Pt/Al2O3 at 283 K based on ATR experiments; y represents the surface coverage. Species 1: p-bonded, 2: a-H abstracted and 3: N lone pair bonded (tilted). (Reproduced from ref. 431 with permission)

Catalysis, 2010, 22, 144–278 | 227

Zaera and coworkers investigated the adsorption of CD at Pt using in situ RAIRS technique.428 Good correlation has been found between the surface concentration of ﬂat lying CD and the enantioselectivity in EtPy hydrogenation. In another study it was found that the oxygen present in most CD solutions from dissolved air blocks the surface toward any CD uptake, presumably via its dissociation to atomic surface oxygen atoms (and maybe by partial oxidation of the platinum surface), while Ar, N2, or CO2 has no inﬂuence on the adsorption of CD. Hydrogen plays a unique role, initially facilitating the uptake of CD.90 Non-linear eﬀects (see Section 5.5.3) has also been characterized by in situ ATR-IR spectroscopy comparing the behaviour of CD and PhOCD.435 It was shown that both alkaloids are adsorbed via the quinoline ring and that the spatial arrangement of the quinuclidine ring is crucial for the chiral recognition. The result helped to elucidate the role of the phenyl group played in the creation of the chiral space responsible for the inversion of ED. Surface-enhanced Raman spectroscopy has been applied to investigate the adsorption of CD on polycrystalline Pt.92 The eﬀects of liquid-phase concentration in ethanol and that of co-adsorbed hydrogen were studied. It was found that CD is strongly and irreversibly adsorbed through its quinoline ring via p-bonding. Stronger adsorption of DHCD compared with CD was also suggested. The room-temperature adsorption of four cinchona alkaloids and three reference quinoline-based compounds from CCl4 solutions onto a polycrystalline Pt surface was characterized by in situ RAIRS.377 The results are shown in Fig. 41. Data show Langmuir type adsorption kinetics. The

Fig. 41 Adsorption uptakes for all the quinoline-derived compounds from CCl4 solutions onto Pt as a function of concentration. (Reproduced from ref. 377 with permission)

228 | Catalysis, 2010, 22, 144–278

Table 18 Adsorption equilibrium constants (estimated from the data given in Fig. 41 and expressed as Kads 1) and solubilities in CCl4 for the quinoline-derived compounds. (Reproduced from ref. 377 with permission) Compounds

Kads 1 mM

Solubility in CCl4, mM

Kads 1/solubility

Quinoline lepidine 6-methoxyquinoline QN CD QD CN

30 11 6.5 0.65 0.5 0.25 0.1

inﬁnite inﬁnite inﬁnite 8.63 1.56 16.3 0.30

0.075 0.32 0.015 0.33

calculated adsorption equilibrium constants (Kads) are given in Table 18 and were found to follow the sequence CNWQDWCDWQNW6–methoxyquinolineWlepidineWquinoline. Some of this ordering can be explained by diﬀerences in solubility, but QD displays a much larger Kads than expected on the basis of its large relative solubility. Results indicated also that each alkaloid binds diﬀerently on Pt at saturation coverages. At low concentrations all alkaloids adsorb with their quinoline ring ﬂat on the surface and then tilt abruptly upon increasing coverages, but the switch-over takes place at signiﬁcantly diﬀerent solution concentrations in each case. CD tilts mainly along its quinoline long axis, whereas CN does it along the short one. CN has also larger degree of ring distortion. The most surprising result is the fact that CN shows a higher Kads than CD, QN, or QD. In this respect results obtained in an earlier study has to be mentioned. As shown in Fig. 42 when in sequential introduction of CCl4 solutions of CN, CD, and back to CN was applied CN was replaced by CD, while CN cannot replace CD.376 Similar conclusions can be obtained from other studies using the ‘‘chiral switch’’ technique.238,379 The diﬀerence in the adsorption mode of CD and CN was investigated in a recent study.439 The main message from these studies is that the solvent has to be taken into consideration in the formation of the above discussed adsorbed forms of alkaloids. The adsorption of 1-(1-naphthyl)ethylamine (NEA) on platinum surfaces has also been characterized by RAIRS and temperature-programmed desorption (TPD) both under ultra-high vacuum and in situ from liquid solutions.440 ATR-IR spectroscopy was also used to prove the ﬂexibility of the quinuclidine moiety resulting in surface quinuclidine bound CD. It was done by comparison of the ATR-IR spectra of CD and PhOCD adsorbed on Pt.98,435 The diﬀerence in the intensity of the signal at 1458 cm 1 (d(C–H) deformation modes of the quinuclidine skeleton) was attributed to the possible interaction of the quinuclidine moiety of CD with the Pt surface. Based on the comparison of the ATR-IR spectra of CD and CD hydrochloride adsorbed on Pt under similar condition the authors came to the conclusion that at the Pt the quinuclidine moiety of CD has identical structure as in the protonated quinuclidine of CD hydrochloride.98 This ﬁnding was considered as an additional evidence that CD can be protonated Catalysis, 2010, 22, 144–278 | 229

Fig. 42 In-situ RAIRS from experiments: sequential introduction of CCl4 solutions of CN, CD, and back to CN (from top to bottom). (Reproduced from ref. 429 with permission)

by chemisorbed hydrogen in aprotic solvent. However, the authors admit that ‘‘this conclusion is tempting, further studies are needed to conﬁrm its validity. Doubts may arise from the presence of HCl originating from CH2Cl2 solvent decomposition on Pt’’.327 We also have serious doubt with respect to these interpretations. ATR-IR method was also used to give some new information about the formation of quinuclidine bonded form of alkaloids. However, these results are quite dubious. In this respect in a recent study the following information was given:270 ‘‘Recent results indicate that the quinuclidine moiety is also involved in adsorption on Pt.98,441 At low coverage, the energetically favoured geometry exhibits the aromatic ring parallel to the surface and the quinuclidine moiety oriented toward the metal surface in a geometry that has been named surface quinuclidine bound (SQB).442 However, the careful analysis of references given above clearly indicates that there is no experimental evidence for the above statement. In ref. 441 upon using molecular dynamics simulation ‘‘CD was found to adsorb with the quinoline ring oriented largely parallel (ao61) to the surface. CD surface attachment was found to be through both p- bonding of the aromatic group and adsorption of the CQC double bond of the vinyl group’’. It was also mentioned that ‘‘we found that CD conformation at the surface was not only aﬀected by the ethanol solvent, but also by the cinchonidine– cinchonidine steric interactions and their competition for surface sites’’. However, no words were given related to the involvement of the 230 | Catalysis, 2010, 22, 144–278

quinuclidine moiety in the adsorption. In ref. 98 only computational results were given, while ref. 442 contains scientiﬁc speculations and not real evidences. Consequently, there is no exact experimental evidence for the formation of surface quinuclidine bound CD. The adsorption of CD and CN on Pt(111) and Pd(111) single crystals has been investigated by means of time-lapse STM in an ultra-high vacuum system.443 CD and CN showed similar adsorption modes and diﬀusion behaviour on Pt(111). The only exception is that the ﬂatly adsorbed CN molecules were signiﬁcantly more mobile than CD. NEXAFS and corresponding STEX calculations have been applied to investigate the orientation of DHCD on Pt(111) at 298 and 323 K.91 The results indicate that at 298 K the quinoline ring is almost parallel to the Pt surface but is tilted up from the surface by 60 101. However, the results show that at higher temperatures the alkaloid dissociates to quinoline. Various techniques, such as NEXAFS, XPS, STM, and temperature programmed reaction was applied to investigate at 320 K the molecular orientation, spatial distribution, and thermal behaviour of the powerful chiral modiﬁer precursor (S)-naphthylethylamine adsorbed on Pt(111)427 No formation of ordered arrays was observed in the presence or the absence of coadsorbed hydrogen. Based on high resolution STM images some speculation was done related to the formation of 1:1 docking complex between MePy reactant and the chiral modiﬁer. NEXAFS revealed that the quinoline ring of 10,11-dihydrocinchonidine is orientated parallel to the surface at 298 K, whereas at 323K the orientation is tilted about 601 to the surface.91 von Arx et al. used STM to reveal that the cinchonidine molecules are randomly distributed on the Pt (111) surface.444 Attard and co-workers studied the inﬂuene of surface structure and surface chirality on the adsorption rate of several modiﬁers.445 However, probably due to the large adsorption energy of these systems, no diﬀerence in the adsorption rate was observed. It was also observed that in a hydrogen-saturated solution, the alkaloid dihydrocinchonine is partially desorbed from a kinked, chiral Pt surface. The adsorption of CD on polycrystalline Pt surfaces in H2SO4 was investigated by cyclic voltammetry.241 The adsorption was found to be irreversible. The results indicated that at maximum coverage, 50% of the Pt atoms were still accessible for hydrogen adsorption. They calculated also the site requirement for CD equal to 13–14 Pt atoms. In another study it was calculated that in the surface modiﬁcation model each enantioselective site requires 25 or so Pt atoms to achieve simultaneous adsorption of modiﬁer, reactant, and hydrogen.446 The adsorption of quinoline and CN on Pt (111), Pt (332) and polycrystalline Pt electrode has been studied by diﬀerential electrochemical mass spectrometry (DEMS). It was shown that benzene is even able to displace some of the alkaloid.447 Electrochemical method was applied to investigate the introduction of cinchona alkaloids with R- and S-kink sites of the Pt(643) surface.447 No interaction was evidenced. Catalysis, 2010, 22, 144–278 | 231

Fig. 43 Demodulated ATR spectra of diﬀerent concentration modulation experiments. The KPL concentration was modulated (modulation period T (184 s) between 0 and 5 10 2 mol/L in CH2Cl2. Spectrum a: clean, uncoated Ge internal reﬂection element; spectrum b: a Pt/Al2O3 ﬁlm in the absence of CD; spectrum c–e were recorded on a Pt/Al2O3 ﬁlm in the presence of CD (5 10 4 mol/L). Before the modulation experiments were started (c–e), the Pt/Al2O3 ﬁlms were diﬀerently treated: (c) 30 min N2 saturated CH2Cl2 only; (d) pretreatment with 5 min H2 saturated KPL solution; (e) directly contacted with modulation solutions. (Reproduced from ref. 314 with permission)

6.3.4 Substrate modiﬁer interactions. Interaction of KPL with CD was investigated by ATRIR concentration modulation spectroscopy using CH2Cl2 as a solvent.314 The results showed that in the presence of CD and KPL a new band appeared at 2580 cm 1 as shown in Figs. 43 and 44. This band was attributed to the formation of protonated quinuclidine by chemisorbed hydrogen. This experimental results is considered as a key prove for the support of authors general view with respect to the reaction mechanism (see more details in Section 8), In this respect the use of CH2Cl2 solvent has to mentioned. Due to its use it is not excluded that the observed protonation is simple an artefact. Recently the formation of a surface complex between adsorbed cis-EtPy and protonated CD has been suggested using ATR-IR method during asymmetric platinum catalysed hydrogenation of EtPy in supercritical ethane solvent.448 These results are shown in Fig. 45. Based on the shifts in the 1200–1300 cm 1 region preferential adsorption of EtPy as cis-conformer was suggested. The appearance of the band at 1660 cm 1 was tentatively ‘‘be attributed to carbonyl stretching vibrations of EtPy’’. 232 | Catalysis, 2010, 22, 144–278

Fig. 44 Comparison between two KPL concentration modulation ATR experiments. The KPL concentration was modulated (modulation period T 184 s) between 0 and 5 10 2 mol/L in CH2Cl2. Use of modiﬁers: (a) N-Methyl CD (5 10 4 mol/L), (b) CD (5 10 4 mol/L). (Reproduced from ref. 314 with permission)

Fig. 45 ATR-IR spectra of adsorption/reaction of EP in ‘‘supercritical’’ ethane on (1) Al2O3 in absence of H2; (2) CD-premodiﬁed Pt-black; (3) unmodiﬁed Pt/Al2O3, but CD dissolved in EP; (4) CD-premodiﬁed Pt/Al2O3. Conditions in all experiments were 40 1C, and 95 bar. Molar ratio EP:H2 : ethane=1:5:200. (Reproduced from ref. 448 with permission)

The above shift to lower wave-numbers occurs ‘‘due to hydrogen bonding between the quinuclidine nitrogen and the keto oxygen atom of EtPy’’. However, in this respect it is necessary to mention the author’s following statement: ‘‘This interpretation is in line with our recently reported study on Catalysis, 2010, 22, 144–278 | 233

KPL adsorption on a CD-modiﬁed Pt/Al2O3 thin ﬁlm’’.431 However, in the above reference there is no words related to the assignment of any bands around 1660 cm 1. The conformational ﬂexibility of the quinuclidine moiety was investigated by ATRIR experiments under nearly in situ conditions, by comparing the adsorption behaviour of CD and O-phenyl-CD on platinum.98,435 It was concluded that the tertiary nitrogen of the quinuclidine moiety can participate in the anchoring of the alkaloid and can be protonated by surface hydrogen. This study is related to the mechanism proposed by Baiker’s group whereby the tertiary nitrogen can promote charge polarization of hydrogen and its transfer to the substrate (see more details in Section 8). There is one more comment on these results. Even if we accept results shown in Figures 43–45 one question still remains: What is the proof that surface species assigned to the [CD-H þ -substrate] complex is really involved in the ED? 6.3.5 HPLC-MS and HPLC-ESI-MS investigations. These studies were related to the investigation of the products formed from diﬀerent alkaloids during enantioselective hydrogenation reactions. Upon investigation of the eﬀect of a-ICN it was demonstrated by HPLC–ESMS measurements that the cyclic ether structure of the alkaloid remained unchanged.449 In another study the product of isomerization of b-ICN, b-isocinchonicine (b-ICNN), was hydrogenated using supported Pt and Pd catalysts. The products were analyzed using HPLC-ESI-ion-trap MS measurements.450 Combined HPLC and ESI-MS method was used to investigate cinchona alkaloid derivatives formed in the hydrogenation of a-ICN and b-ICN.197 The products of reaction are shown in Fig. 46. The hydrogenated compounds were identiﬁed as 1 0 ,2 0 ,3 0 ,4 0 - tetrahydro-a-ICN (A) and 1 0 ,200 ,300 ,400 tetrahydro-b-ICN (B) and decahydro-a-ICN (C). Upon investigating C9-O-substituted cinchona alkaloids in the enantioselective hydrogenation of EtPy ESI-MSD-ion-trap method was applied to follow and identify the hydrogenated derivatives of these cinchona derivatives.206 HPLC-MS method was used to investigate products of H–D exchange measurements of diﬀerent alkaloids.451 As revealed by these measurements, iso-alkaloids are not converted back to CN or QD; (v) in all alkaloids studied, H–D exchange takes place on the quinoline skeleton as well as on carbon atom C9; (vi) H–D exchange on the quinuclidine skeleton appears signiﬁcant only in the case of CN and a-ICN.230 Deuterium exchange in CD was also studied in ref. 48.

Fig. 46 Cinchona alkaloid derivatives formed by hydrogenation of a-ICN and b-ICN.

234 | Catalysis, 2010, 22, 144–278

Potential products of hydrogenation of a-ICN were investigated in another study. The aim of this work was identifying hydrogenated cinchonas formed during enantioselective hydrogenation of EtPy. The target compound was DHCN which might possibly be formed from a-ICN. The results obtained by HPLC/ESI-MS measurements showed that of DHCN was not formed.385 7. 7.1

Theoretical calculations Introduction and ﬁrst attempts

The proposed mechanisms for the asymmetric hydrogenation of activated ketones over cinchona-platinum catalyst system that leads to the observed ED require supports using diﬀerent theoretical-computational studies. These studies are related to the conformational analysis of both substrates and modiﬁers, adsorption of both substrates and modiﬁers into Pt and energetic calculations for the whole complex system Pt-modiﬁer-substrate. It is generally accepted that the CD or its natural or synthetic analogue forms a 1:1 complex, what is hydrogenated on the metal surface. The question is where the chiral discrimination takes place on the Pt surface or in the liquid phase. Both the structure and the conformational complexity of cinchona alkaloids generate several possible interactions with the substrates. As a results a ‘‘chiral pocket’’ is created for the ED either on the Pt surface or in the liquid phase. The way to explore these properties one has to investigate or model theoretically the characteristic features of substrate molecules and the cinchona alkaloids. Based on this knowledge, the modiﬁer-Pt and the substrate-modiﬁer and substrate-modiﬁer-Pt interactions can be investigated. It is useful to extend all these calculations with solvent eﬀect. Various computational methods have been used so far, such as molecular mechanics452 and quantum chemical calculations.453 Molecular geometries can be optimized on MMFF94 molecular mechanic level. Relatively accurate energies (mainly for energy diﬀerences) can be obtained e.g. on HF-SCF/6–31G* or B3LYP/6–31G* level single point energy calculations. The involvement of metal in these calculations requires the use DFT methods. In the next sub-sections we shall review most of the relevant results related to the computation on substrates, modiﬁers. Calculations related to the substrate-modiﬁer interactions and possible interactions of all these components with Pt surface will be discussed in Section 8 related to the reaction mechanism. 7.2

Characteristic features of substrate molecules

Among the substrates, for which the cinchona-Pt catalytic system yields high ee values, the following groups of molecules are in the focus of experiments and computations: (i) pyruvate esters, (ii) ketopantolactone (KPL), (ii) diketones (PPD), (iii) triﬂuoro acetophenone (TFA), and (iv) triﬂuorodiketones. In addition, the ﬂuor substituted derivatives of the ﬁrst Catalysis, 2010, 22, 144–278 | 235

and third groups of compounds were also included into computational studies. It was also shown that s-trans conformer of MePy is more stable in the gas phase by 1–2 kcal/mol, but the relative stability could be strongly inﬂuenced by the metal surface, especially because the s-cis conformer has a considerably larger dipole moment.454 With respect to the formation of substrate CD complex the two conformers of MePy were compared,455 and it was found that the complex yielding (R)-methyl lactate upon hydrogenation was more stable than the corresponding pro-(S) complex by 1.8 kcal/mol (corresponds to an enantiomeric excess of 92%, in good agreement with experiment), however, for the analogous complexes of s-cis-methyl pyruvate the energy diﬀerence is only 0.2 kcal/mol in favour of pro-(R), corresponding to 17% ee value. The relation between the electronic structure of a-substituted ketones and their reactivity456 in the racemic and enantioselective Pt-catalysed hydrogenation was also investigated. A correlation between the keto carbonyl orbital energy and the hydrogenation rate was found, which rationalizes the eﬀect of the substituent on the rate of hydrogenation (the often observed rate acceleration). The ﬁrst model calculations indicated that in the complex responsible for the enantio-diﬀerentiation the a-keto ester existed in trans-conformation.457 Further studies revailed that in the enantio-diﬀerentiation complex the a-keto ester can also exist in its cis conformation.455 The fact that in the hydrogenation of KPL to (R)-PL high enantioselectivites were obtained indicated that the rigid cis conformation has no inﬂuence on the enantiodiﬀerentiation step. Generally speaking, the number of substrate molecules with high ee is in a relatively narrow59,370 range: This strong substrate speciﬁcity has not been answered yet, neither by theory nor with computation. a-keto esters were modelled in diﬀerent studies.49,74,456,458 The trans conformer is more stable than the cis one, the carbon atom in the keto group is partially positively charged, while the oxygen part is negatively. In the hydrogenation of acetophenone and TFA derivatives on CDmodiﬁed Pt/Al2O3, the rates and ee values varied strongly with the nature of the aromatic substituents.332,363 The diﬀerent reactivities were attributed to the electronic (and steric) eﬀect of the substituents and to hydrogen-bonding interactions between the quinuclidine N atom of the alkaloid and the carbonyl group of the substrate.359,456 Theoretical calculations revealed a linear correlation between the logarithm of the reaction rate and the highest occupied molecular orbital and lowest unoccupied molecular orbital stabilization DEorb of the carbonyl compounds, relative to the reference compound (see Fig. 47.).53 The relative orbital stabilization is deﬁned as the sum of two numbers: the diﬀerence between the energy of the anti-bonding orbital of the reference compound acetophenone and that of the substituted acetophenone, and the corresponding energy diﬀerence for the bonding orbitals. The more stabilized the orbitals of the substituted acetophenone are, the larger DEorb and the reactivity of the molecule are. According to these calculations (where the metal surface was not involved), the origin of ‘‘ligand acceleration’’ is the lowering of the p-orbitals in the diastereomeric complex of the substrate and modiﬁer. In the pro(R) and pro(S) complexes, 236 | Catalysis, 2010, 22, 144–278

Fig. 47 Linear correlation between the logarithm of the hydrogenation rate (mmol h 1) of acetophenone and TFA derivatives and the relative orbital stabilization DEorb. (Reproduced from ref. 53 with permission)

the carbonyl -orbitals are diﬀerently stabilized, which results in diﬀerent intrinsic rates in the formation of the two enantiomers. It remains, however, to be proven that the concept can be extended to other substrates and reaction types. The proton aﬃnities of seven diﬀerent ketones, vicinal diketones, and a-keto esters (acetophenone, TFA, 2,3-butanedione, PPD, MePy, EBF and KPL) have been evaluated theoretically using the conventional ab initio HF and several post-HF methods (MP2, MP4, CCSD), density functional methods with the B3LYP hybrid functional, as well as some ab initio model chemistries [CBS-4M, G2(MP2), and G3(MP2)//B3LYP].459 In the most stable protonated species, the proton is bound to one of the carbonyl oxygens in the molecule. The preferred site depends on the molecule. In two a-keto esters (MePy and KPL) the carbonyl oxygen of the ester group is protonated. In the case of EBF and the asymmetric vicinal diketone, PPD it is the carbonyl oxygen next to the phenyl group, which forms a more stable bond with the proton. These phenomena can be understood in terms of resonance stabilization of the resulting cations. It was shown that the protonation of both the modiﬁer and the reactant in acidic solvent hinders the formation of a reactant– modiﬁer complex, which is believed to be crucial for enantio-discrimination, consequently in these cases the ee decreases. This decrease of ee was observed in case of butanedione (14% vs. 47%), KPL (35% vs. 79%) and PPD (6% vs. 65%) comparing results in AcOH and toluene, respectively. It is known that triﬂuoro beta-diketones can also exist in enol form. The adsorption of both the keto and enol forms of 1,1,1-triﬂuoro-2,4-diketone into Pt(111) was modelled and calculated406 DFT calculations including the simulation of the interaction of a protonated amine with the enolate Catalysis, 2010, 22, 144–278 | 237

adsorbed on Pt revealed that only the C–O bond next to the CF3 group of the substrate is in direct contact with Pt and can be hydrogenated. 7.3

Characteristic features of cinchona alkaloids

Characteristic features of cinchona alkaloids have already been discussed in Sections 2 and 3. For this reason, we shall refer to information given in these sections. In the next sub-section we shall focus mainly on the conformational analysis of cinchona alkaloids and their analogues. 7.3.1 Conformational behaviour of cinchonidine. Among the eﬀective chiral modiﬁers used for enantioselective hydrogenation of activated ketones over Pt/Al2O3 catalysts the most eﬀective and widely used one is CD. The characteristic feature of this alkaloid is shown in Figs. 2 and 3. It has already been mentioned earlier cinchona alkaloids were intensively investigated by NMR methods176,186 as described in Section 2. Conformational behaviour of cinchonidine was calculated independently by diﬀerent groups.83,88,460,461 Most of these results are in accordance with earlier results discussed in Section 2.176,183 In our ﬁrst study,84 the conformational analysis was done by using rigid quinoline and quinuclidine moieties. As a result, four stable conformations have been found. In our subsequent study,461 all of these calculations were repeated in such a way that only the phi and psi torsion angles were forced to be constant, while all other freedom of the molecule were left to relax. The conformational analysis indicated that CD might exist at least in nine diﬀerent forms, however only four of them are relatively stable (two open (A1 and A2) and two closed (C1 and C2) conformers). These results are shown in Fig. 48. The solid line in this ﬁgure gives the contour of the possible forms of CD within 8 kcal/mol energy range. For CD the 2–D NOE spectra indicate446 that the major conformation in solution is conformation A2, this is close to that adopted by the molecules in the solid state.179 Thus, the conformational analysis strongly indicates that CD can exist both in open and closed forms and both forms of CD can be involved in the formation of substrate–modiﬁer complex. Detailed conformational analysis of CD in solutions using NMR techniques as well as theoretical calculations was done in ref. 88. Three conformers of CD are shown to be stable at room temperature, cl(1), cl(2), and op(3), with the latter being the most stable in apolar solvents. The stability of the closed conformers relative to that of open(3), however, increased with solvent polarity. In polar solvents the three conformers have similar energy (Fig. 48). Structures and relative energies in kcal/mol of low energy CD conformations were calculated using hybrid density functional (B3LYP/6–31 þ G*/ PCM B3LYP) and AMBER* optimization (AMBER*/GB/SA).416 The relative stability of the conformers is as follows: op(3)Wcl(1)Wcl(2)Wop(4). The eﬀect of protonation on the conformation of CD was investigated in a recent study.417 It was shown that the protonation of cinchonidine appears to hinder the rotation around the C4 0 –C9 and C9–C8 bonds and to favour only a narrow range of the conformational space of the molecule. In terms of the behaviour of CD and CN molecules in solution, 2D NMR 238 | Catalysis, 2010, 22, 144–278

Fig. 48 Conformational analysis of cinchonidine. The calculated energy map has been obtained by changing the torsion angles phi ((C3 0 )–(C4 0 )–(C9)–(C8)) and psi ((C4 0 )–(C9)–(C8) –(C7)). The contours are given in steps 0.5 kcal/mol. (Reproduced from ref. 461 with permission)

experiments indicate a somewhat more restricted rotational conformation space for cinchonine than for cinchonidine.462 Protonation of cinchonidine also signiﬁcantly restricts its rotational conformation space.417 7.3.2 Conformational behaviour of cinchona derivatives. Detailed NMR analysis and ab initio quantum chemical calculations were performed on 1 0 ,2 0 ,3 0 ,4 0 ,10,11-hexahydroderivatives of CD.409 The rotation around the C4 0 –C9 and C9–C8 bonds led to conformers of close energies, providing evidence on the possible presence of other stable conformers in the solution of these cinchonidine derivatives. The conformational analysis of the synthetic chiral modiﬁer O-phenylcinchonidine in vacuum has been performed at semi-empirical level and at DFT level with a medium-size basis set for energetics related to the parent alkaloid cinchonidine.421 The O-phenyl-cinchonidine behaves similarly to cinchonidine and owns the same main stable conformers as mentioned above in vacuum and in CH2Cl2 and CCl4 solvents. Based on combined theoretical–experimental results, the open(3) appears to be the most populated in these solvents, but indication was found that an excess cl(2) conformer has to be also expected in CD2Cl2 in comparison to CD. The authors suggest that the sterical constraints imposed by insertion of O-phenyl at the C9 position shows its eﬀect when the substituted CD adsorbs on the surface via its quinoline part. Isocinchonines belongs to the class of rigid alkaloids (see Section 3.2). In these molecules the rotation of the quinuclidine ring is restricted (see Fig. 6). b-ICN was investigated in a recent study and its conformational analysis was performed. The results conﬁrmed that the numerous conformational changes possible for CD and CN are reduced to a single degree of freedom, namely rotation around C(4 0 )-C(9).463 Catalysis, 2010, 22, 144–278 | 239

Recently detailed NMR, DFT, and X-ray investigation of some cinchona alkaloid O-ethers related to the determination of their structure and conformations was published.391 7.3.3 Other calculations. Conformational analysis of synthetic modiﬁers, such as (R)-2-1-pyrrolidinyl)-1-1-naphthyl)ethanol, (R)-2-1pyrrolidinyl)-1-2-naphthyl)ethanol, and (R)-2-1-pyrrolidinyl)-1-1-8-methylnaphthyl)]ethanol was also performed. Open–(3) conformers appeared to be the most stable. Minimum energy structures of the pro-(R) and pro-(S) interaction complexes between (R)-2-1-pyrrolidinyl)-1-1-naphthyl)ethanol and trans ethyl pyruvate were also performed. Good quantitative agreement between calculated and experimental ee values has been found for the enantioselective hydrogenation of EtPy over Pt catalyst chirally modiﬁed by synthetic pyrrolidinyl– naphthyl–ethanol modiﬁers, assuming that EtPy exists in the trans conformation in the adsorbed enantio-diﬀerentiating complex. The destabilising repulsive interaction between EtPy and the anchoring aromatic moiety within the pro-(S) complex has been identiﬁed to be important for ED.85 7.4

Substrate-Pt interaction

The adsorption of ketones on transition metals has been the topic of various studies.464–466 In general, ketones adsorb on transition metal surfaces via two diﬀerent bonding mode: as Z1(O) in an end-on adsorption conﬁguration in which the oxygen atom is bonded by its lone pair orbital to the metal surface, or as Z2(C, O), with both the carbon and the oxygen atoms of the keto-group p-bonded to the metal and the CQO moiety lying parallel to the surface. The bonding interaction between an adsorbate and a surface is a very complex process.467 To perform ﬁrst principle calculation on the adsorption of substrate molecule97 on a reasonably large (about 20–40 atom) Pt (or Pd) surface or cluster has become feasible only recently, however these results should be handle cautiously. For example a drawback of using metal clusters of this size is that the Pt cluster is strongly paramagnetic (high spin states) in the result of the computation,98 while experimentally it is not magnetic. The interaction of various ketones with Pt surface was investigated in details.97 Fig. 49 shows the adsorption geometries of EtPy for both the cis and in the trans conformations. The cis Z2 adsorption appeared to be the most stable one (see Fig. 49a) In the Z1 adsorption mode only the trans conformation showed an energy minimum (see Fig. 49c), whereas the cis conformer was not stable when Z1 adsorbed. When adsorbed Z2 the main interaction the keto-carbonyl moiety interacts with the metal. Once the preferred keto-carbonyl adsorption had taken place, the ester group interacts only weekly with the Pt surface. The adsorption EBF and its derivatives onto Pt surface was also modelled and calculated.468 The results showed that the introduction of two o-substituents into the aromatic ring completely eliminated the reactivity of the ketone. The dramatic diﬀerence between EBF and ethyl mesithylglyoxylate (5) is their mode of adsorption. The o-substitution suppresses adsorption 240 | Catalysis, 2010, 22, 144–278

Fig. 49 Adsorption modes of EtPy on Pt: (a) Z2-cis, (b) Z2-trans, (c) Z1-trans, and (d) semi-hydrogenated Z7.3 -cis. (Reproduced from ref. 97 with permission)

modes where the keto-carbonyl group is bound to the metal in Z2(C,O) mode involved in the hydrogenation reaction. In a recent study the interaction of PPD with Pt surface was investigated by DRIFT spectroscopic method and DFT calculations.469 Seven diﬀerent adsorption forms were suggested as shown in Fig. 50. The calculated adsorption energies are given in Table 19. DFT calculations demonstrated that Z1-(O2) conﬁguration is the most stable end-on adsorption mode of PPD. Tilted p-bonded adsorption mode of cinchonidine was revealed on the platinum catalyst at higher concentration of CD. 7.5

Modiﬁer-Pt interaction

The interaction of CD with Pt(111) both in ultrahigh vacuum (UHV) and in ethanol solvent has been studied using molecular dynamics (MD) simulation. In UHV at low coverage (0.0125 molecules/Pt atom) and 298.15 K the CD was found to adsorb with the quinoline ring oriented largely parallel (a=61) to the surface.441 Cinchonidine surface attachment was found to be through both p bonding of the aromatic group and adsorption of the CQC double bond of the vinyl group. The interactions between ethanol solutions of CD (0.129 and 1.035 M) and the platinum surface were also simulated. For the less concentrated solution (0.129 M) two diﬀerent equilibrium conformations were found, one in which only part of the quinoline is attached to the surface, and another slightly more stable conformation. In the latter one the quinoline group is adsorbed parallel to the platinum surface. Catalysis, 2010, 22, 144–278 | 241

Fig. 50 PPD adsorption modes. (Reproduced from ref. 469 with permission)

Table 19 Adsorption Energies for Diﬀerent Adsorption Modes of 1-Phenyl-1,2-propanedione (Reproduced from ref. 469 with permission) adsorption conﬁguration

DE (kJ mol 1)

Z1(O1) Z1(O2) pseudo- n1(O1) di- n1(O1, O2) Z2(C1, O1) Z2(C2, O2) Z3(C1, O1, O2)

19a 36 55 3a 88b 38a 59

a

Taken from ref. 470.

b

Phenyl ring only partly adsorbed.

It was also observed that that CD conformation at the surface was aﬀected both by the ethanol solvent and the CD-CD interactions and their competition for surface sites. The conformations of CD adsorbed on a Pt(111) surface were also investigated.442 Eight conformationally diﬀerent adsorption states due to diﬀerent degrees of rotation around the t1 and t2 degrees of freedom were identiﬁed. The possible role of these conformations in the formation of chiral surface sites relevant to enantioselective hydrogenation was also investigated. The comparison of the conformational behaviour of CD in solution and on Pt has revealed the eﬀect of the metal surface on the internal mobility of the alkaloid. In the study the role of the adsorbed op(3) conformer, the observed conformational ﬂexibility on the Pt surface revealed the possibility that other conformers of CD also might be involved in ED. Closed conformations of CD are found to play an important role in the 242 | Catalysis, 2010, 22, 144–278

Fig. 51 The adsorption modes of cinchonidine: (a) parallel p-bonded; (b) tilted p -bonded; (c) a-H-abstracted quinolyl; (d) quinoline-N-lone pair. (Reproduced from ref. 469 with permission)

conformational equilibria on the surface due to their stability and are identiﬁed as precursors of the less stable, but probably more active, open conformers. Although the open and closed conformers are closely related to the correspondent ones found in solution, surface species that are also adsorbed via quinuclidine moiety have been identiﬁed also as possible forms of metal–modiﬁer interaction and can be involved in ED. In a recent study the possible forms of adsorbed CD was given as shown in Fig. 51.469 Density functional theory (DFT) at the B3LYP/T(ON)DZP level was used to model one-to-one reactant-modiﬁer interactions relevant to the enantioselective hydrogenation of PPD and MP over platinum catalysts In an other study205 DFT calculations revealed that protonated cinchonidine and 10,11-dihydrocinchonidine are more stable on Pt when adopting the so-called QA-Open(4) conformation rather than the Open-(3) conformation. Thus, the QA-Open conformations may have some role in the enantioselective hydrogenation over modiﬁed Pt catalysts. The results of these calculations are shown in Fig. 52. 8. 8.1

Reaction mechanisms and related calculations Introduction

In the ﬁrst approaches related to the enantioselective hydrogenation of activated ketones over Pt-cinchona catalytic system mechanistic views developed earlier for Ni-tartaric acid catalyst system.471 It also means that the ﬁrst models were proposed without any solid knowledge about the reaction mechanism, i.e., the proposed reaction mechanism and schemes were based on pure ‘‘presumption’’ related to the knowledge accumulated in studies over the Ni-tartaric acid system. However, careful analysis of these two enantioselective hydrogenation reactions shows deﬁnite diﬀerences as follows: (i) mode of introduction of the modiﬁer, (ii) amount of modiﬁers, (iii) reaction rate, and (iv) reaction temperature In Ni/tartrate system the catalyst requires pre-modiﬁcation under conditions diﬀerent from those used in the hydrogenation reaction. Contrary to that the Pt/cinchona system the introduction of the chiral modiﬁer Catalysis, 2010, 22, 144–278 | 243

Fig. 52 Side and top views of the RI-BP86/SV(P) optimized Open(3) and QA-Open(4) conformations of cinchonidine, 10,11-dihydrocinchonidine, and their protonated counterparts on the Pt38 cluster. The adsorption energies as deﬁned in the text are given in parentheses (in kJ mol 1) (Reproduced from ref. 205 with permission)

into the reaction mixture during racemic hydrogenation instantaneously induces ED. In the Pt/cinchona system the substrate/modiﬁer ratio is very high (in case of KPL it is 276 000), but the optimum substrate/modiﬁer ratio strongly depends on the type of substrate. In Ni/tartrate system this value is several order lower. In the Pt/cinchona system as it has already been discussed the cinchona alkaloid induces not only enantio-diﬀerentiation but a well-pronounced rate acceleration. Contrary to that in the Ni/tartrate system the modiﬁer decreases the reactions rate. It has to be emphasized that the rate acceleration eﬀect has also been observed in homogeneous catalytic reactions in the presence of cinchona alkaloids.353 In Ni/tartrate system the reaction takes place at moderate temperature above 60 1C, while the enantioselective hydrogenation of prochiral ketones requires low temperature around 0–10 1C. High temperatures above 40 1C are not favourable to get high ee values. As it has been mentioned in a recent publication301 ‘‘two types of mechanisms-modiﬁed catalyst,58,265,455,472 and shielding eﬀect74 have been proposed’’. Unfortunately, not all of the authors consider this way. Those who accepted the modiﬁed catalyst model (we shall call it ‘‘surface 244 | Catalysis, 2010, 22, 144–278

modiﬁcation’’ model) did an enormous eﬀort using various spectroscopic methods and computational tools aimed to demonstrate that in fact only one reaction scheme is working, i.e., both in aprotic and protic reaction media the enantioselective hydrogenation reaction proceeds via the protonated form of cinchonidine and the surface entities responsible for the enantio-diﬀerentiation step has a N–H þ –O bond.53 It also means that all key events, i.e., the rate acceleration and enantio-diﬀerentiation are exclusively surface related phenomena. Of course, the surface of Pt has a crucial role in this reaction. First or all Pt provides landing sites for all participants in the given reaction, i.e., both ‘‘actors’’ and ‘‘spectators’’ can land and react on the Pt sites. The question is what of these ‘‘surface induced interactions’’ shall have a direct contribution into the rate acceleration and enantio-diﬀerentiation steps. 8.2

The ‘‘surface modiﬁcation’’ model

The surface modiﬁcation concept was ﬁrst suggested in early nineties, when the number of publications in this area was very scarce. The distinction between ‘‘modiﬁed’’ and ‘‘unmodiﬁed’’ sites over platinum was done by Blaser and coworkers.63 These terms have been widely accepted and used by those who believe in the ‘‘surface modiﬁcation’’ concept. The above distinction was formulated into a kinetic equation describing the ‘‘ligand acceleration’’ phenomenon.58 It has to be mentioned that in the above study no mechanistic views were given just a very simple reaction scheme. According to this scheme enantioselective hydrogenation takes place over ‘‘modiﬁed’’ sites, while racemic hydrogenation over ‘‘unmodiﬁed’’ sites. This model gives a relatively good correlation between rate and ee values, but it does not explain the variation of the ee values with the concentration of the substrate. The ﬁrst mechanistic view or scheme was given by Wells and coworkers in early nineties (‘‘template model’’).65,228 According to this the enantio-differentiating sites are created by an ‘‘ordered layer of the alkaloid’’ with the formation of a ‘‘chiral pocket’’, i.e., a free room between adsorbed chiral entities, where the enantio-diﬀerentiation can take place. However, their concept was not supported by surface spectroscopic methods446 and the original idea was withdrawn45 very soon and a new idea based on the involvement of the half-hydrogenated form of a-keto ester in the enantio-differentiation step was proposed by the same research group.472 It has to be emphasized again that this new idea was suggested without any experimental prove or evidence. The fact that in acetic acid the enantioselective hydrogenation of alkyl pyruvates takes place with higher rates and higher enantioselectivity than in aprotic solvents the original idea given by Wells and coworkers was further extended to the involvement of the protonated form of cinchonidine.49 In this model the quinuclidine nitrogen atom is protonated and the substrate is still in its original state, maintaining the double bond character of the carbonyl group. Later on this model was accepted as a general one even in the absence of acids.53 Baiker and coworkers have published several experimental85,97 and theoretical papers314,430,442 trying to convince the Catalysis, 2010, 22, 144–278 | 245

readers that in the presence of cinchona alkaloids the atomic hydrogen is spontaneously ionized with the formation protonated form of alkaloid. There is a general view that the hydrogen-bonding interactions can trigger for rate acceleration.359,456 8.3

‘‘Shielding eﬀect model’’83

8.3.1 The principle of chemical shielding.183 The basis for this approach is the shielding eﬀect (SE) known in organic chemistry. If a prochiral moiety is preferentially shielded its further reaction can take place only from its unshielded site resulting in an ED. A chiral template molecule can induce SE in a similar way, i.e. it preferentially interacts with one of the prochiral sites of the substrate leaving the unshielded site free for the reaction. Intramolecular steric shielding of an a–keto ester moiety has been observed resulting in enantio-diﬀerentiation in the hydrogenation of the a-keto group.473 ED was observed only in the presence of large aromatic substituent, such as naphthyl, and it was completely lost if it was substituted for a phenyl one. Based on this ﬁnding the ED was attributed to the SE induced by the large aromatic moiety. Similar phenomena was also described for the hydrogenation of an a,b- unsaturated ester moiety.474 The above two examples are shown in Fig. 53. Additional examples for chemical shielding can be found elsewhere.475,476 8.3.2 Application of the principle of chemical shielding to the Orito’s reaction.83,183 Both reacting groups shown in Fig. 53 have a common feature, namely a conjugated double bond system. This feature is also characteristic for most of the substrates what can enantio-selectively be hydrogenated in the presence of cinchona-Pt or cinchona- Pd catalyst477 systems. On the other hand it was also shown that in the hydrogenation of EtPy over Pt/Al2O3 catalyst in the presence of new types of modiﬁers (derivatives of 2-1-pyrrolidinyl)-1-naphthyl)ethanol) the ED was completely lost if the naphthyl ring was replaced by phenyl or pyridyl one.60 It should also be mentioned that in the hydrogenation of a-keto esters over CD-Pt/Al2O3

CO2Me

CO2Me

OAc O

OAc O

O

O

O

O

O

R

O

R

CH2

O

R = Me, Et, Ph

Fig. 53 Intermolecular chemical shielding in the involvement of a-keto ester and a,b-unsaturated ester. (Reproduced from refs. 475 and 476 with permission)

246 | Catalysis, 2010, 22, 144–278

catalysts the ED was partially or fully lost if the quinoline ring of CD was partially or fully hydrogenated.64 If the key p–p interactions given in Fig. 53 are compared with results obtained in the enantioselective heterogeneous hydrogenation experiments using cinchona-Pt systems the following very important elements of similarity can be found: (i) ED can only be observed in the presence of large aromatic shielding groups; (ii) the reactive prochiral group (both the keto carbonyl and the CQC double bond) is activated by an electron withdrawing carbonyl group; (iii) the prochiral keto carbonyl group in most of the activated ketones is in a conjugation with the adjacent carbonyl group (or with the aromatic ring in TFA as it was shown earlier).461 As emerges from these comparisons the presence of a large aromatic substituents in the modiﬁer and a conjugated double bond system in the substrate should play an important role to induce ED in these asymmetric hydrogenation reactions, i.e. these are the key elements responsible for the substrate speciﬁcity. It has to be added that the shielding eﬀect model suggest that substrate modiﬁer interactions responsible for the ED take place in the liquid phase and not on the Pt sites. The term substrate-modiﬁer interaction in liquid phase was also mentioned by other authors. ‘‘The substrate-modiﬁer interaction exists, according to circular dichroism, in solution, probably in the form of aggregates’’.68 Similar views were given in another study.221 It has also been suggested that for some substrates, the solvent is involved in the substrate–modiﬁer interaction.478 It has been suggested that the OH group of the alkaloid should be involved in the substrate–modiﬁer interaction which more likely occurs in the liquid phase.479 8.4

Character of substrate – modiﬁer interaction

In a recent review it has been admitted that ‘‘in the absence of reliable experimental evidence, most mechanistic ideas are based on assumptions and (at best) calculations. In most cases, the models assume two interactions between the amine type modiﬁer and the ketone: an N–H–O52,77,458,480,481 or N-C type attractive interaction67,412,482 and a second attractive or repulsive interaction that directs the adsorption of the ketone on Pt.52,483 In the following section we shall follow the above consideration, i.e., we shall distinguish electrophilic and nucleophilic interactions between the substrate and the modiﬁer. It has to be emphasized that all existing models postulate 1:1 type interactions between CD and the substrate. Barto´k et al.451 and Augustine et al.371 proposed that not only the quinuclidine N, but also the OH function of CD would be involved in the interactions. However this view can be questioned as neither the methylation nor the removal of the OH group in CD hinders the enantio-selection in the hydrogenation of EtPy.57 The ﬁrst attempt aimed to elucidate the character of substrate-modiﬁer interactions was done in an early study458 related to the investigation of interaction between MePy with NH3 and NH4 þ . In this study the ammonia part represented the quinuclidine nitrogen of CD. The results indicated that MePy can interact with both NH3 and NH4 þ and the electrophilic interaction is more favourable than the nucleophilic one. However, the Catalysis, 2010, 22, 144–278 | 247

nucleophilic interaction provided interesting result, namely the reaction pocket of MePy is located between the two carbonyl groups. This ﬁnding indicated that under condition of enantioselective hydrogenation both carbonyl groups are activated. Consequently, if the enantioselective hydrogenation of EtPy is performed in methanol trans-esteriﬁcation reaction can take place. This has been proved experimentally.307 In the next approaches two types of interactions were modelled, namely the Baiker’s group focused on the electrophilic interaction between the halfhydrogenated substrate and CD.53,74,457,484 In our group, based on kinetic results over cinchonidine-Pt/Al2O3, and the proposed ‘‘shielding eﬀect’’ model the nucleophilic interaction between the modiﬁer and substrate was favored.69 8.4.1 Electrophilic interaction. In the ﬁrst theoretical study related to the substrate-modiﬁer interaction the formation of a week complex between protonated CD and MePy methyl pyruvate was investigated.74 In this study molecular mechanics and AM1 semi-empirical methods were used. The calculated surface complex bifurcated electrophilic interaction between the protonated quinuclidine and the keto carbonyl group was considered. The results revealed that adsorption of the complex leading to (R)-methyl lactate is more favorable than that of the corresponding system yielding (S)-methyl lactate. In another study ab initio calculations were used to study the interaction between protonated amines (NH3, (CH3)3N and quinuclidine) and methyl pyruvate (MP), as well as between protonated MP and these amines.481 Based on results it has been suggested that interactions mediated by a proton between the MP and the alkaloid are the main driving force leading to enantiodiﬀerentiation in the hydrogenation of a-ketoesters. MP interacts with protonated amines preferentially in the s-cis conformation, with a proton making two hydrogen bonds to the carbonyl oxygens. This proton may be transferred to MP, forming a new complex in which the amines are bonded to the protonated MP. The last complex is approximately 10 kcal mol 1 less stable than the ﬁrst one. However, this energy diﬀerence decreases to approximately 5 kcal mol 1 when solvent eﬀects are included. Characteristic feature of these models is that both in the absence and presence of acid in the key reaction intermediate the electrophilic interaction prevails with the involvement of N–H–O or N–H þ –O bonds. This interaction can be either monodentate or bidentate (bifurcated) as shown in Figs. 54.

Fig. 54 Schematic representation of mechanistic models suggested by Baiker and co-workers; A: monodentate interaction;49,98 B: monodentate interaction;338 C: bidentate interaction.338 (Reproduced from refs. 49, 98, 338 with permission)

248 | Catalysis, 2010, 22, 144–278

The key issue in this model the adsorption of the chrial modiﬁer with its quinoline ring parallel to the Pt surface as it has already been discussed in Section 6. Several adsorbed conformations of CD were calculated.442 Sequence of events on the Pt surface related to the protonation of quinuclidine nitrogen is shown in Fig. 55, 56, while the calculated substratemodiﬁer complex and its interaction with the Pt surface are given in Fig. 57. In this respect the waging motion of the quinuclidine part has been emphasized. The overall route for proton transfer from protonated CD to adsorbed substrate supported by DFT calculations and in situ ATR-IR spectroscopy is shown in Fig. 55. It has to be added that in the recent publication of this group it was found that the most stable intermediate complex forms without adsorption of the substrate.103 We have serious objection against the exclusiveness of the electrophilic interaction in the Orito’s reaction. Assuming the scheme given in Fig. 55 one would suggest that not only the quinuclidine nitrogen of the cinchona alkaloid can be involved in the transformation of atomic hydrogen formed on the Pt site to a protonated nitrogen base. All tertiary nitrogen bases should have similar ability. Consequently, the hydrogen transfer in the presence of an achiral tertiary amines (ATA) should result in a racemic product. Thus,

Fig. 55 Simpliﬁed scheme for the interaction of the quinuclidine N atom with the Pt-H system and the subsequent transfer of the H to the adsorbed ketone. (Reproduced from ref. 372 with permission)

Fig. 56 Hydrogen uptake of CD from a platinum surface. (Reproduced from ref. 98 with permission)

Catalysis, 2010, 22, 144–278 | 249

Fig. 57 Proposed relative surface structures of adsorbed CD and MePy on a Pt31 cluster (DFT calculations), which allow an H-bonding interaction (not shown). (Reproduced from ref. 389 with permission)

the simultaneous addition of cinchona alkaloids and ATAs should result in a decrease in the ee values. However, our results discussed in Section 5.7.2 showed a completely opposite eﬀect. According to Figs. 55–57 the key issues in the ‘‘surface modiﬁcation’’ model are as follows: (i) adsorption of the modiﬁer with its condensed aromatic ring parallel to the Pt surface, (ii) stabilization of the modiﬁer in its open form, (iii) adsorption of the substrate via its both carbonyl bonds in re-phase, (iv) formation of a hydrogen bond between protonated alkaloid and the substrate and (v) transfer of the proton to the substrate. The condensed aromatic ring is often called as anchoring site. As it has already be mentioned earlier the chiral C8 and C9 carbon atoms of the alkaloid play a vital role in the enantio-diﬀerentiation, i.e. their conformation determines the position of the ‘‘chiral pocket’’ located in the neighbourhood of quinuclidine nitrogen. In addition the mode of adsorption of the chiral modiﬁer (ﬂat or tilted) has a decisive role in the ED step as it has been shown in the series of O-substituted derivatives of CD.435 In a recent study the role of the modiﬁer structure in the reactant-modiﬁer interactions relevant to the heterogeneous enantioselective hydrogenation of PPD and MePy) was studied using DFT calculations.205 Two protonated modiﬁers, CD and MeOCD, in diﬀerent conformations were considered. So-called bifurcated and cyclic hydrogen-bonded reactant-modiﬁer interaction modes were investigated. The results showed that only the bifurcated reactant-modiﬁer(Open3) complexes were found to be relevant in the determination of enantioselectivity. Analysis of the orbital stabilization implies a notable decrease in the enantiomeric excess of the main hydrogenation product of PPD when CD is replaced with MeOCD. On the other hand, according to the theoretical calculations the hydrogenation of MP over modiﬁed Pt is expected to yield an equal ee values in the presence 250 | Catalysis, 2010, 22, 144–278

of both modiﬁers. DFT calculations revealed that protonated CD and DHCD are more stable on Pt when adopting the so-called QA-Open (4) conformation rather than the Open (3) conformation. These conformations are shown in Fig. 58. Thus, the QA-Open conformations may have some role in the enantioselective hydrogenation over modiﬁed Pt catalysts. The QA-Open (4) conformation of a modiﬁer is adsorbed on the surface via both its quinoline and quinuclidine moieties, and a reactant may interact simultaneously with the protonated quinuclidine nitrogen and the functional group at the C(9) position of the modiﬁer. The interaction between KPL (Pro-(R) conformation) and adsorbed o-PyOCD over Pt surface was also modelled.372 In this case both the quinuclidine and pyridine moieties in o-PyOCD were protonated. At the end of the simulation the hydrogen was transferred to the keto-carbonyl group of ketopantolactone, therefore forming a semi-hydrogenated surface species, while protonated o-pyridyl group coordinated to the ester carbonyl group as shown in Fig. 59. Modelling studies revealed also that there is no mode of docking of any low energy conformation of epiquinidine with pyruvate ester that could lead to selective enantioface adsorption of the latter.485

Fig. 58 Calculated stabilized structure of CD over Pt in Open (3) and QA-Open (4). (Reproduced from ref. 204 with permission)

Fig. 59 Interaction of adsorbed o-PyOCD in the most stable position of the o-pyridoxy moiety, with KPL adsorbed in a Pro-(R) conformation. (Reproduced from ref. 372 with permission)

Catalysis, 2010, 22, 144–278 | 251

Recently a new idea was suggested assuming a H-bond between the quinuclidine N of CD and the ester carbonyl or triﬂuoromethyl group of the substrate and a second, monodentate or bidentate H-bond involving two or one aromatic H atoms of the modiﬁer at 5 0 - and 6 0 -positions and the O atom of the ketocarbonyl group.72,316,483 In addition, it was suggested that similar interactions might exist for all activated ketones. General principles of this type of interactions are shown in Fig. 60. This type of interactions has also been proposed in recent studies by the Barto´k’s goup384,486 (see Fig. 61). However, in this case the character of substrate modiﬁer interaction is nucleophilic. However, these models strongly contradicts to experimental ﬁndings as ring-substituted cinchona derivatives, such as QN and QD containing a methoxy group in the 6 0 -position, are highly eﬀective modiﬁers in the hydrogenation of activated ketones (see Section 3.1). Consequently, the interaction of the substrate molecule with the proton of the quinoline ring is not a prerequisite for enantio-diﬀerentiation. In addition, the proposed interaction of the keto group with the aromatic hydrogen (see Fig. 60 cannot give any reasonable explanation for the activation of the keto group resulting in rate acceleration. Based on ESI-MS spectra of EtPy, DHCD and EtPy-DHCD mixtures interesting equilibria (see Fig. 62) were suggested by Barto´k et al.195 In this respect it interesting to note that the semi-ketal formed between the CD and EtPy was also evidenced in another study.109

Fig. 60 Two-point H-bonding model suggested by McBreen et al. A: interaction between protonated CD and MePy, B: interaction between protonated CD and triﬂuoroacetophenone (TFA). (Reproduced from refs. 316 and 483 with permission)

Fig. 61 Hydrogen bounded substrate – modiﬁer complex. (Reproduced from ref. 487 with permission)

252 | Catalysis, 2010, 22, 144–278

+N

COOEt

H

Me

N H

OH

OEt

+N

C

O

N

OH

O

H

OH

C C Me

+N H

N H

O

C

Me COOEt

OH Fig. 62 Possible form of adducts between DHCH and EtPy (Reproduced from ref. 195 with permission)

8.4.2 Nucleophilic interaction. Nucleophilic interaction between the substrate and the modiﬁer has been suggested by diﬀerent authors.83,84,369,412,458,488 In an early mechanistic view67 it was proposed that ‘‘the hydrogenation of pyruvate over either modiﬁed or unmodiﬁed platinum takes place on the more coordinatively unsaturated corner atoms or adatoms on the platinum surface’’. In their further study369 it was suggested that the substrate-modiﬁer complex (association) ‘‘incorporates some interaction between the quinuclidine nitrogen and the ketone carbonyl group of the pyruvate. One such interaction is between the electron pair on the nitrogen and the electron deﬁcient carbon atom of the carbonyl group, which, as discussed above would account for the observed increase in hydrogenation rate in these reactions’’. The involvement of coordinatively unsaturated platinum was also suggested by the Barto´k’s group.486,487 The ‘‘shielding eﬀect’’ model is also based on the nucleophilic interactions. The key issue of this model is the involvement of closed conformer of CD in the substrate modiﬁer interaction. The character of these interactions will be discussed in the next section. Upon investigating enantioselective hydrogenation of KPL in the presence of b-ICN Barto´k an coworkers suggested two possible forms of surface complex representing either electrophilic or nucleophilic interaction as shown in the next scheme (see Fig. 63.) The inversion of the ee was attributed to the change of the reaction mechanism from nucleophilic to electrophilic one. It was also suggested that in the hydrogenation of activated ketones in the presence of cinchona-Pt catalysts proceeds ‘‘through nucleophilic addition of a cinchona alkaloid to the ketone to form a zwitterionic adduct, which is then hydrogenolyzed with inversion of conﬁguration. The enantioselectivity of the reaction is determined by the relative stabilities of the diastereomeric adducts adsorbed on platinum’’.412 However, this mechanism has been ruled out as it was pointed out that this approach does not take into account steric hindrance against the interaction of the amine modiﬁer with cyclic ketones and further critical point is the regioselectivity of the hydrogenolysis of the hypothetical zwitterionic intermediate.413 Catalysis, 2010, 22, 144–278 | 253

Fig. 63 Enantioselective hydrogenation of KPL in the presence of b-ICN. (Reproduced from ref. 419 with permission)

Fig. 64 The proposed structures of adduct complexes of b-ICN (B) and CD (C) with esters of phenylglyoxylic acids. (Reproduced from ref. 486 with permission)

When the inversion of enantioselectivity in the presence of b-ICNs was investigated the formation of diﬀerent nucleophilic adducts was proposed as shown in Fig. 64. 8.4.3 Character of interactions in the ‘‘shielding eﬀect’’ model. Cinchona alkaloids (CA) have two rotational axises, which allow to rotate either the quinoline ring around the C(1) 0 -C(9) axis or the quinuclidine ring around the C(9)-C(8) axis (see Sections 2.1 and 3.1). Molecular mechanics and ab initio calculations performed by diﬀerent groups that in liquid phase CA can exist at least in three diﬀerent stable forms. We suggest that the closed form of the modiﬁer is required both for the RA and the ED. Only the closed form of CA can provide the cooperative required for ED. The possible arrangements of the substrate and the modiﬁer in the shielded complex are shown in Fig. 65. In the hydrogenation of pyruvate esters the complex shown in Fig. 65A would result in the expected (R)-lactate ester, while the complex given in Fig. 65B would give the corresponding (S)-product. The major diﬀerence between the (R) and (S) complexes is the mode of interaction between the lone pair of electrons of the quinuclidine 254 | Catalysis, 2010, 22, 144–278

Fig. 65 Shielded [methyl pyruvate-CDclosed] complexes. A – favourable alignment; B – unfavorable alignment. (Reproduced from ref. 461 with permission)

nitrogen and the keto carbonyl group. In complex (R), the ‘directionality’489 of the nucleophilic attack by quinuclidine nitrogen towards the keto carbonyl group is very favourable to increase the reactivity of the keto carbonyl group because the electron- rich quinuclidine nitrogen and the keto-oxygen of the substrate are on the opposite sides of the keto carbon atom. According to the orbital steering theory,490 a proper ‘reaction window’ or ‘reaction cone’ can result in perturbation of the reacting group. In our case the proper ‘reaction window’ is determined by the relative position of the quinuclidine N1, pyruvate C2 and O3 atoms, i.e., by direct N1–C2 interaction as shown in Fig. 66. In case of proper ‘reaction window’ the overall reactivity of the keto group should increase. We suggest that the above perturbation leads to a pronounced rate increase both in the hydrogenation reaction and the formation of by-products, such as semi-ketal, transesteriﬁcation and deuterium exchange products84,289,307 Thus, in complex (R), the favourable directionality promotes the perturbation of the keto carbonyl group, resulting in the observed RA. Contrary to that in complex (S), due to the misalignment of the interacting groups, i.e., due to the lack of direct N1–C2 interaction, no RA can be expected, consequently, the hydrogenation of (S). complex is not accelerated. Those who favour the modiﬁer-surface or modiﬁer-metal interactions suggest that the quinoline ring is involved in the adsorption of the modiﬁer to the metal.45,82,214 Contrary to that we suggest that the quinoline ring is involved in the p-p interaction with the substrate via ‘‘p-p stacking’’.83,84 We suggest that the RA is a cooperative eﬀect with the involvement of both the quinuclidine nitrogen and the quinoline ring (Fig. 66). Monte-Carlo simulation method was used to investigate the interaction of the [methyl pyruvate– CD]closed complex with Pt (111). surface. The result shown in Fig. 67 indicates that the shielded complex retains its entity even after adsorption. The above ﬁgure gives a good presentation of the SE provided by the large aromatic moiety. With respect to the explanation related to the use of small Pt colloids we should like to refer to results of our calculations (Monte Carlo simulation). These results given in Fig. 68 shows that the closed [substrate – modiﬁer] complex can be accommodated at the Pt(111) face even of a small Pt nanocluster.244 When the ‘‘shielding eﬀect’’ model was proposed it was also mentioned that in the enantioselective hydrogenation of activated ketones cinchona alkaloids behave like an enzyme.93 This view has recently been emphasized without reference to our original idea.103 Catalysis, 2010, 22, 144–278 | 255

Fig. 66 A – Simpliﬁed scheme for the [MePy–CDclosed ] complex; B – the ‘reaction window’ for the substrate–modiﬁer interaction in [MePy–CDclosed ] complex. (Reproduced from ref. 461 with permission)

Fig. 67 Monte-Carlo simulation of the adsorption of the [MePy–CD]closed complex onto Pt (111) surface. (Reproduced from ref. 83 with permission)

8.5

Pros and contras related to existing models

8.5.1 ‘‘Surface modiﬁcation’’ model. The surface modiﬁcation model has been strongly altered since the introduction of the ﬁrst concept, i.e., the division of Pt surface into modiﬁed and unmodiﬁed sites. According to the generalized view the key issues in this model are as follows: (i) adsorption of CD in its open (3) conformation via its quinoline ring parallel to the Pt surface, (ii) conformational changes at the Pt surface to form quinuclidine bonded CD, (iii) formation of protonated quinuclidine moiety, (iv) transformation of the proton to the substrate to form half hydrogenated surface species, and (v) direct addition of the second hydrogen form the Pt surface to get the chiral keto-alcohol (see Fig. 57). However, based on the discussion in this contribution we can emphasize that this model was not able to give appropriate answer to the following 256 | Catalysis, 2010, 22, 144–278

Fig. 68 Monte-Carlo simulation of the adsorption of the [MePy–CD]closed complex onto the Pt(111) surface of small Pt colloid. (Reproduced from ref. 244 with permission)

important observations: (i) rate acceleration (see Section 5.5.1), (ii) the appearance of the initial transient period in the conversion-selectivity dependencies (see Section 5.5.2); (iii) the inﬂuence of ATAs (see Section 5.6.2), (iv) the inﬂuence of compounds with large aromatic ring (see Section 5.6.3), (v) the inﬂuence of the modiﬁers of the Pt and support (see Section 5.6.3), (vi) the stability of nucleophilic [substrate – modiﬁer] complex (see Section 6.1.1) in the presence of AcOH, (vii) high enantioselectivities over very small Pt nano-colloids (see Section 4.2.2). Even if scheme shown in Fig. 55 is valid a simple question can be raised? Why the adsorption and subsequent strong distortion of the alkaloid at Pt surface is needed to pick up a proton from the Pt surface if it could also be done without any preadsorption via quinuclidine N-Pt interaction. Would not be it more logical and energetically more favourable? In this respect we have to address literature data related to the protonation of pyridine and its analogs observed under high vacuum and low temperatures.491,492 These references were cited in several papers as a direct proof related to surface reactions given in Fig. 55. However, it has to be pointed out that none of the authors referred to experimental conditions used in these earlier studies. It has to be emphasized that in refs. 492,493 results obtained under ultra-high vacuum were presented. The key experiments were performed over Pt(110) surface under base pressure between 5 10 11 and 1 10 10 Torr and the temperature was kept between 100–180 K. The intensities of the 3450 cm 1 EELS peak characteristic of cation formation showed strong temperature dependence and above 180 K it was hardly detected. Consequently, there are many speculations related to this model. Finally, it has to be added that this model does not take into account one of the important issues that these alkaloids are used by organic chemist for many years to induce ED or chiral separation. Catalysis, 2010, 22, 144–278 | 257

8.5.2 ‘‘Shielding eﬀect’’ model (SEM). The shielding eﬀect model can explain the following experimental ﬁndings:74,83 (i) substrate speciﬁcity), (ii) inversion of enantioselectivity for enantiopairs CD-CN, QN-QD, (iii) rate acceleration, (iv) the MI character of the ee-conversion dependencies (v) the loss of ee in case of replacing the quinoline ring for pyridyl or phenyl, (vi) formation of transesteriﬁcation and deuterium exchange products; (vii) effectiveness of very small Pt colloids, (viii) the role of achiral tertiary amines, (viii) the need for of large aromatic moieties in cinchona alkaloids to induce ED. The reaction network derived form the shielding eﬀect model provided kinetic equations what can describe the following kinetic events:83 (i) rate acceleration, (ii) increase the reaction rate at the initial period of the reaction,66 (iii) the the MI character of the ee-conversion dependencies. The strongest conﬂict with ‘‘shielding eﬀect’’ model is the ﬁnding that ICNs can also induce enantio-diﬀerentiation. However, in this respect the anomalous behaviour of these rigid alkaloids361 has to be mentioned. The behaviour of these alkaloids needs further elucidation and probably the use of more pure alkaloids. In this respect the lack of rate acceleration in the presence of a-ICN and the inversion of ee in the presence of b-ICN have to be mentioned. Another unclear issue is that shielding eﬀect model requires nucleophilic interaction between the substrate and the modiﬁer, while in the presence of AcOH electrophilic interactions interaction prevails. Although in this relation recent NMR results has to be emphasized, what clearly indicated that in case of KPL even in the presence of AcOH the nucleophilic substrate-modiﬁer adduct can be formed.419 Finally, we have to admit that the ‘‘shielding eﬀect’’ model was not supported by the scientiﬁc community in the ﬁeld of heterogeneous catalysis. This fact can be attributed to the deﬁciency of the model. However, it cannot be excluded that due to the strong inﬂuence of those who favour the ‘‘surface modiﬁcation’’ model, the scientiﬁc community just simple followed the main stream without any criticism. 9.

Conclusions

In this review an attempt was done to give a retrospective overview about methods, approaches and results obtained in the last three decades in the area of enantioselective hydrogenation of activated ketones. Both practical and theoretical aspects were discussed. Characteristic feature of this review is that the term ‘‘chirally modiﬁed surface’’ was not really used. Although tremendous eﬀort has been done so far to elucidate the peculiarities of this reaction there are still several open questions related to the substrate-modiﬁer and substrate-modiﬁer-platinum interactions involved in the enantio-diﬀerentiation step. It seems to us that starting from the beginning of early nineties there is a permanent desire to demonstrate and prove that in the presence of cinchona-Pt catalyst system all interactions responsible for ED take place on the Pt surface. In addition, last years the mainstream concentrated to prove that the protonated quinuclidine moiety is involved in the ﬁrst step of hydrogen transfer, with the involvement of adsorbed form of the alkaloid by 258 | Catalysis, 2010, 22, 144–278

its quinuclidine moiety and adsorbed hydrogen, i.e., the key mechanism in aprotic and protic solvents is the same. We consider that this mechanistic view is too general and in addition it is artiﬁcial and far-fetched. This view neglects series of experimental evidences obtained by diﬀerent research groups. Let us remind the reader only three of these neglecting facts: Use of Pt colloid. As it was mentioned in Section 4.2.2 high rates and high enantioselectivities were obtained upon using very small Pt colloids. In none of the reviews published so far by catalytic scientists this anomalous ﬁndings were not really discussed. However, in a recent review written by organic chemists it was emphasized that upon using Pt colloids in enantioselective hydrogenation of MePy in the presence of CD ‘‘the smallest Pt clusters gave the best results despite having no ﬂat surface large enough for the adsorption of cinchonidine’’.493 Addition of quinoline. In this respect one of the earlier results has also to be mentioned.345 In this study it was shown that the addition of quinoline to the reaction mixture at very low concentration (0.1 g/L) increased both the rate and the ee values. The authors attributed this observation to some sort of base eﬀect. Unfortunately, due to the dominance of the general view, i.e. the ‘‘formation of chirally modiﬁed surfaces’’ this result has completely been forgotten and in the last eighteen years it was only very seldom cited. In this respect we should like to refer to our results discussed in Section 5.6.2). These results clearly indicated that the addition of quinoline has no negative eﬀect either on the reaction rate and the ee values. Based on these ﬁndings the scheme shown in Fig. 69 has been suggested. Fig. 69 shows four diﬀerent situations described in ref. 402. A represents the racemic hydrogenation in the absence of any additive, where due to the poisoning eﬀects of by-products the rate is controlled by free Pt sites left. B corresponds to the situation when quinoline is added prior to the addition of hydrogen. In this case the poisoning eﬀect of by-products decreased resulting in a rate increase in racemic hydrogenation. C represents the enantioselective hydrogenation upon injecting CD, where the initial surface coverages are identical to those established in case A.

Fig. 69 Surface coverages in the presence of quinoline added to the reaction mixture. (Reproduced from ref. 402 with permission)

Catalysis, 2010, 22, 144–278 | 259

After injection of CD instantaneous RA and MI type the ee – conversion dependencies are evidenced. D corresponds to the situation when cinchonidine is injected to surface B containing preadsorbed quinoline. In case D, due to the established competition between CD and quinoline the Pt sites are covered by both CD and quinoline. However, the net results are unexpected, i.e., increased rate and increased ee. Consequently, these results might indicate that the general view that ‘‘strongly bonded to the platinum CD via its quinoline ring’’ needs some corrections. Spectroscopic results in liquid phase. Both NMR83,84,418,419 and Circular Dichorism68,93 spectroscopic results clearly indicated that there is a complex formation between the substrates and cinchona alkaloids in the liquid phase. These facts were completely neglected by the scientiﬁc community. In this respect let us refer to a very recent results indicating that the nucleophilic complex between KPL and b-ICN can exist even in the presence of AcOH. The authors showed a nice dependence between the solution concentration of the 1:1 substrate-modiﬁer complex and the enantioselectivity as shown in Fig. 70. In addition it was shown that there is an excellent correlation between the ee values, the concentration of the 1:1 substrate/modiﬁer complex and the amount of AcOH added as shown in Fig. 71. It is known that in the enantioselective hydrogenation of KPL all polar solvents have a negative eﬀect.76 Results presented in ref. 419 (see Figs. 2 and 3) deﬁnitely show the importance of the complex formation in the liquid phase. However, even in the light of these unambiguous evidences the authors of this study were not brave enough as they made the following remark: ‘‘we did not doubt the role of the protonated cinchona despite the fact that the spectroscopy data published previously, obtained under the conditions of the Orito reaction in toluene, are not totally convincing in terms of protonation of the N atom of quinuclidine’’. We believe that in the near future further high quality and unambiguous experimental data will be obtained related to the character of substratemodiﬁer interactions as well to the formation of substrate-modiﬁer complex at the Pt surface. We also hope that those who have diﬀerent views on the mechanism of Orito’s reaction in the future will get more wide open platforms to publish their results and ideas.

Fig. 70 Dependence of the ee of the concentration of substrate-modiﬁer complex in the liquid phase. (Reproduced from ref. 419 with permission)

260 | Catalysis, 2010, 22, 144–278

Fig. 71 Comparison of the total concentration of the 1:1 b-ICN–KPL complexes measured by NMR (circles) with the enantioselectivities (diamonds) obtained at identical AcOH (CD3COOD for the NMR) concentrations (logarithmic scale). (Reproduced from ref. 419 with permission)

Abbreviations used AcOH AS ATA ATR-IR CD CI CN De DHCD DHCN DRIFT EBF ECD ED ee EOG EOPB Et EtLa Etpy HHCD HRTEM ICN IR ITP KPL M M/S MBF Me MeLa

acetic acid anchoring sites achiral tertiary amine attenuated total reﬂection infrared (spectroscopy) cinchonidine chiral induction cinchonine diastereomeric excess 10,11-dihydrocinchonidine 10,11-dihydrocinchonine diﬀuse reﬂectance FT infrared ethyl benzoylformate enantioselectivity–conversion dependencies enantio-diﬀerentiation enantiomeric excess (%) diethyl 2-oxoglutarate ethyl-2-oxo-4-phenyl butirate ethyl ethyl lactate ethyl pyruvate hexahydrocinchonidine high-resolution transmission electron microscopy isocinchonine infrared spectroscopy initial transient period ketopantolactone metal modiﬁer–substrate molar ratio methyl benzoylformate methyl methyl lactate Catalysis, 2010, 22, 144–278 | 261

MeOCD MeODHCD MePy MI NED NLP PADA Ph PhOCD PNEA PPD PSC Q QD QN QND RA RAIRS RAIRS RE SBA SERS STM TEA TFA TFAP TMS TOF Y

O-methyl-cinchonidine O-methyl-10,11-dihydrocinchonidine methyl pyruvate monotonic increase 1-naphthyl-1,2-ethanediol non-linear phenomena pyruvaldehyde dimethyl acetal phenyl O-phenyl-cinchonidine pantoylnaphthylethylamine 1-phenyl-1,2-propanedione primary surface complex quinoline quinidine quinine quinuclidine rate acceleration reﬂection absorption infrared spectroscopy reﬂection adsorption infrared spectra rate enhancement stabilized bimetallic alloy surface-enhanced Raman scattering scanning tunnelling microscopy triethylamine triﬂuoroacetic acid triﬂuroacetophenone trimethylsilane turnover frequency (h 1) yield (%)

References 1 R. A. Sheldon, Chirotechnology, Marcel Dekker, New York, 1993. 2 Chiral Catalyst Immobilization and Recycling, D. E. De Vos, I. F. J. Vankelecom and P. A. Jacobs (Eds), Wiley-VCH, 2000. 3 Comprehensive Asymmetric Catalysis, E. N. Jacobsen, A. Pfaltz, and H. Hisashi (Eds.) 2004, Springer, Berlin, Heidelberg, New York, LXXVIII, 1856. 4 Applied Homogeneous Catalysis by Organometallic Complexes, (second ed.), B. Cornils, W. A. Herrmann (Eds.) Wiley-VCH, Weinheim, 2002 5 C. E. Song and S. Lee, Chem. Rev., 2002, 102, 3495. 6 M. T. Reetz, Adv. Catal., 2006, 49, 1. 7 M. T. Martin, Drug Discovery Today, 1996, 1, 239. 8 R. Noyori, Tetrahedron, 1994, 50, 4259. 9 W. S. Knowles, Angew. Chem. Int. Ed., 2002, 41, 1998. 10 H. C. Kolb, M. S. VanNieuwenhze and K. B. Sharpless, Chem. Rev., 1994, 94, 2483. 11 K. B. Sharpless, Angew. Chem. Int. Ed., 2002, 41, 2024. 12 T. T. L. Au-Yeung and A. S.C. Chan, Coord. Chem. Rev., 2004, 248, 2151. 13 J. M. Brown, Platinum Metals Rev., 1987, 31, 137. 262 | Catalysis, 2010, 22, 144–278

14 H. Sasai, E. Emori, T. Arai and M. Shibasaki, Tetrahedron Letters, 1996, 37, 5561. 15 H. U. Blaser, B. Pugin and F. Spindler, J. Mol. Catal. A: Chem., 2005, 231, 1. 16 I. F. J. Vankelecom and P. A. Jacobs, in Chiral Catalyst Immobilization and Recycling, D.E. de Vos, I.F.J. Vankelecom, P.A., Jacobs, (Eds.), Wiley-VCH, Weinheim, 2000, p. 1. 17 H. U. Blaser, B. Pugin and M. Studer, in Chiral Catalyst Immobilization and Recycling, D. E. de Vos, I. F. J. Vankelecom and P. A., Jacobs (Eds.), WileyVCH, Weinheim, 2000, p.19. 18 S. J. Bae, S. W. Kim, T. Hyeon and B. M. Kim, Chem. Commun., 2000, 31. 19 M. D. Jones, R. Raja, J. M. Thomas and B. F. G. Johnson, Top. Catal., 2003, 25, 71. 20 J. A. M. Brandts, C. V. Cavenaghi, A. Gerlach and M. J. Burk, Chimica Oggi, 2000, 18, 47. 21 S. K. Tanielyan, R. L. Augustine, US Patent 6,005,148 (1999) and 6,025,295 (2000) to Seton Hall University 22 R. L. Augustine, S. K. Tanielyan, N. Mahata, Y. Gao, A. Zsigmond and H. Yang, Appl. Catal. A: Gen., 2003, 256, 69. 23 R. L. Augustine, P. Goel, N. Mahata, C. Reyes and S. K. Tanielyan, J. Mol. Catal. A: Chem., 2004, 216, 189. 24 M. Reggelin, M. Schultz and M. Holbach, Angew. Chem. Int. Ed., 2002, 41, 1614. 25 C. R. Landis and J. Halpern, J. Am. Chem. Soc., 1987, 109, 1746. 26 E. Burello, D. Farruseng and G. Rothenberg, Adv. Synth. Catal., 2004, 346, 1844. 27 H. U. Blaser, Chem. Rev., 1992, 92, 935. 28 A. M. Thayer, Chem. Eng. News, 2005, 83, 40. 29 G. M. Schwab and L. Rudolph, Naturwiss., 1932, 20, 362. 30 E. I. Klabunovskii, in Absolute asymmetric synthesis and asymmetric catalysis, Aspects of the Origin of Life, M. Florkin (Ed.), Pergamon Press, N.Y., Oxford, 1960, p. 105. 31 S. Akabori, S. Sakurai, Y. Izumi and Y. Fuji, Nature, 1956, 178, 323. 32 A. Tai and T. Sigimura, in Chiral Catalyst Immobilization and Recycling, D. E. de Vos, I.F.J. Vankelecom and P.A. Jacobs (Eds.), Wiley-VCH, Weinheim, 2000, p. 173. 33 D. Y. Murzin, P. Maki-Arvela, E. Toukoniitty and T. Salmi, Catal. Rev.-Sci,. Eng., 2005, 7, 175. 34 T. Osawa, T. Harada and O. Takayashu, Top. Catal., 2000, 13, 155. 35 E. Klabunovskii, G. V. Smith and A´. Zsigmond, in Heterogeneous Enantioselective Hydrogenation, Theory and Practice, Catalysis by Metal Complexes, Vol 31, Springer, Dordrecht (The Netherlands), 2003. 36 S. Nakagawa, T. Sugimura and A. Tai, Chem. Lett., 1997, 859. 37 T. Osawa, A. Ozawa, T. Harada and O. Takayasu, J. Mol. Catal. A: Chem., 2000, 154, 271. 38 A. Hoek and W. M. H. Sachtler, J. Catal., 1979, 58, 276. 39 Y. Orito, S. Imai and S. Niwa, Collected papers of the 43rd Catalysis Forum, Japan, 1978, 30. 40 Y. Orito, S. Imai and S. Niwa, J. Chem. Soc. Jpn., 1979, 1118. 41 Y. Otito, S. Imai, S. Niwa and G.-H. Nguen, J. Synth. Org. Chem. Jpn., 1979, 37, 173. 42 Y. Orito, S. Imai and S. Niwa, J.Chem. Soc. Jpn., 1980, 670. 43 Y. Orito, S. Imai and S. Niwa, J.Chem. Soc. Jpn., 1982, 137.

Catalysis, 2010, 22, 144–278 | 263

44 R. Hess, S. Diezi, T. Mallat and A. Baiker, Tetrahedron: Asymmetry, 2004, 15, 251. 45 G. E. Webb and P. B. Wells, Catal. Today, 1992, 12, 319. 46 A. Baiker, J. Mol. Catal. A: Chem., 1997, 115, 473. 47 H. U. Blaser, H. P. Jalett, M. Muller and M. Studer, Catal. Today, 1997, 37, 441. 48 P. B. Wells and A. Wilkinson, Top. Catal., 1998, 5, 39. 49 A. Baiker, J. Mol. Catal. A: Chem., 2000, 163, 205. 50 M. Von Arx, T. Mallat and A. Baiker, Top. Catal., 2002, 19, 75. 51 M. Studer, H. U. Blaser and C. Exner, Adv. Synth. Catal., 2003, 345, 45. 52 T. Bu¨rgi and A. Baiker, Acc. Chem. Res., 2004, 37, 909. 53 T. Mallat, E. Orglmeister and A. Baiker, Chem. Rev., 2007, 107, 4863. 54 H. U. Blaser, B. E. Pugin, F. Spindler and M. Thommen, Acc. Chem. Res., 2007, 40, 1240. + and M. Barto´k, Ultrasonics 55 B. To¨ro¨k, B. Bala´zsik, M. To¨ro¨k, Gy. Szo¨llosi Sonochem., 2000, 7, 151. 56 M. Sutyinszki, K. Szo¨ri, K. Felfo¨ldi and M. Barto´k, Catal. Comm., 2002, 3, 125. 57 H. U. Blaser, H. P. Jalett, D. M. Monti, A. Baiker and J. T. Wehrli, Stud. Surf. Sci. Catal., 1991, 67, 147. 58 M. Garland and H. U. Blaser, J. Am. Chem. Soc., 1990, 112, 7048. 59 A. Baiker, Catal. Today, 2005, 100, 59. 60 M. Schu¨rch, T. Heinz, R. Aeschimann, T. Mallat, A. Pfaltz and A. Baiker, J. Catal., 1998, 173, 187. 61 H. U. Blaser, H. P. Jalett, M. Garland, M. Studer, H. Thies and A. WirthTijani, J. Catal., 1998, 173, 282. 62 H. Wynberg, in: Topics in Stereochemistry, E. L. Eliel, S. H. Wilen and N. L. Allinger (Eds.), Wiley, New York, 1986, Vol. 16, p. 87. 63 H. U. Blaser, H. P. Jalett, D. M. Monti, J. F. Reber and J. T. Wehrli, Stud. Surf. Sci. Catal., 1988, 41, 153. 64 J. L. Margitfalvi, P. Marti, A. Baiker, L. Botz and O. Sticher, Catal. Lett., 1990, 6, 281. 65 I. M. Sutherland, A. Ibbotson, R. B. Moyes and P. B. Wells, J. Catal., 1990, 125, 77. 66 J. Wang, Y. K. Sun, C. Leblond, R. N. Landau and D. G. Blackmond, J. Catal., 1996, 161, 752. 67 R. L. Augustine, S. K. Tanielyan and L. K. Doyle, Tetrahedron: Asymmetry, 1993, 4, 1803. 68 A. Tungler, T. Ma´the´, T. Tarnai, K. Fodor, G. To´th, J. Kajta´r, I. Kolossva´ry, B. Here´nyi and R. A. Sheldon, Tetrahedron: Asymmetry, 1995, 6, 2395. + J. Catal., 1995, 156, 175. 69 J. L. Margitfalvi and M. Hegedus, 70 B. To¨ro¨k, K. Felfo¨ldi, G. Szakonyi and M. Barto´k, Ultrasonics Sonochem., 1997, 4, 301. 71 (a) P. Ma¨ki-Arvela, M. Kuzma, R. Sjo¨holm, R. Leino and T. Salmi, in: Proceedings of the 9th International Symposium on Relations between Homogeneous and Heterogeneous Catalysis, July 20–24, 1998, Southampton, Book of Abstracts, p. 23; and (b) E. Toukoniitty, P. Maki-Arvela, A. N. Villea, A. K. Neyestanaki, T. Salmi, R. Leino, R. Sjoholm, E. Laine, J. Vayrynen, T. Ollonquist and P. J. Kooyman, Catal. Today, 2000, 60, 175. 72 S. Lavoie, M.-A. Laliberte, I. Temprano and P. H. McBreen, J. Am. Chem. Soc., 2006, 128, 7588. 73 H. U. Blaser, M. Garland and H. P. Jallet, J. Catal., 1993, 144, 569.

264 | Catalysis, 2010, 22, 144–278

74 O. Schwalm, J. Weber, B. Minder and A. Baiker, J. Mol. Struct. (Theochem), 1995, 330, 353. + and E. Tﬁrst, Tetrahedron: Asymmetry, 1996, 7, 75 J. L. Margitfalvi, M. Hegedus 571. 76 T. Mallat, M. Bodmer and A. Baiker, Catal. Lett., 1997, 44, 95. 77 M. Schu¨rch, O. Schwalm, T. Mallat, J. Weber and A. Baiker, J. Catal., 1997, 169, 275. 78 K. E. Simon, G. Wang, T. Heinz, T. Giger, T. Mallat, A. Pfaltz and A. Baiker, Tetrahedron: Asymmetry, 1995, 6, 505. 79 H. Bo¨nnemann and G. A. Braun, Angew. Chem. Int. Ed. Engl., 1996, 35, 1992. 80 S. Basu, H. Paul, C. S. Gopinath, S. Bhaduri and G. K. Lahiri, J. Catal., 2005, 229, 298. 81 D. Ferri, T. Bu¨rgi, K. Borsze´ki, T. Mallat and A. Baiker, J. Catal., 2000, 193, 139. 82 T. Mallat, Z. Bodnar, B. Minder, K. Borszeky and A. Baiker, J. Catal., 1997, 168, 183. + and E. Tﬁrst, Stud. Surf. Sci. Catal., 1996, 83 J. L. Margitfalvi, M. Hegedus 101, 241. + J. Mol. Catal. A: Chem., 1996, 107, 281. 84 J. L. Margitfalvi and M. Hegedus, 85 A. Vargas, T. Bu¨rgi and A. Baiker, J. Catal., 2001, 197, 378. 86 Y. Sun, J. Wang, C. Leblond, R. N. Landau, J. Laquidara, J. R. Sowa and D. G. Blackmond, J. Mol. Catal. A: Chem., 1997, 115, 495. 87 D. Ferri, T. Bu¨rgi and A. Baiker, J. Phys. Chem. B, 2001, 105, 3187. 88 T. Bu¨rgi and A. Baiker, J. Am. Chem. Soc., 1998, 120, 12920. 89 D. Ferri, T. Bu¨rgi and A. Baiker, J. Chem. Soc., Perkin Trans., 2002, 2, 437. 90 Z. Ma, J. Kubota and F. Zaera, J. Catal., 2003, 219, 404. 91 D. Ferri, T. Bu¨rgi and A. Baiker, J. Phys. Chem. B, 2004, 108, 14384. 92 R. J. LeBlanc, W. Chu and C. T. Williams, J. Mol. Catal. A: Chem., 2004, 212, 277. 93 J. L. Margitfalvi, E. Ta´las, E. Tﬁrst, C. V. Kumar and A. Gergely, Appl. Catal. A: Gen., 2000, 191, 177. 94 G. A. Attard, J. Phys. Chem. B, 2001, 105, 3158. 95 T. Evans, A. P. Woodhead, A. Gutierrez-Soza, G. Thornton, T. J. Hall, A. A. Davis, N. A. Young, P. B. Wells, R. J. Oldman, O. Plashkevych, O. Vahtras, H. Agren and V. Carravetta, Surf. Sci., 1999, 436, L691. 96 J. M. Bonello, E.C.H. J. Sykes, R. Lindsay, F. J. Williams, A. K. Santra and R. M. Lambert, Surf. Sci., 2001, 482, 207. 97 A. Vargas, T. Bu¨rgi and A. Baiker, J. Catal., 2004, 222, 439. 98 A. Vargas, D. Ferri and A. Baiker, J. Catal., 2005, 236, 1. 99 Y. Sun, J. Wang, C. LeBlond, R. N. Landau and D. G. Blackmond, J. Catal., 1996, 161, 759. 100 J. U. Ko¨hler and J. S. Bradley, Langmuir, 1998, 14, 2730. 101 L. Michaelis and M. Menten, Biochem. Z., 1913, 49, 333. 102 E. Ta´las and J. L. Margitfalvi, Catal. Comm., 2008, 9, 984. 103 A. Vargas, G. Santarossa, M. Iannuzzi and A. Baiker, J. Phys. Chem. C, 2008, 112, 10200. 104 J. Margitfalvi, L. Guczi and A. H. Weiss, React. Kinet. Catal. Lett., 1980, 15, 475. 105 L. L. Kesmodel, L. H. Dubois and G. A. Somorjai, J. Chem. Phys., 1979, 70, 2180.

Catalysis, 2010, 22, 144–278 | 265

106 G. A. Somorjai, M. A. Van Hove and B. E. Bent, J. Phys. Chem., 1988, 92, 973. 107 E. Toukoniitty, P. Ma¨ki-Arvela, A. K. Neyestanaki, T. Salmi and D.Yu. Murzin, Appl. Catal. A: Gen., 2002, 235, 125. 108 C. LeBlond, J. Wang, J. Liu, A. T. Andrews and Y.-K. Sun, J. Am. Chem. Soc., 1999, 121, 4920. 109 E. Ta´las, L. Botz, J. Margitfalvi, O. Sticher and A. Baiker, J. Planar Chromatogr., 1992, 5, 28. 110 H. M. R. Hoﬀmann and J. Frackenpohl, Eur. J. Org. Chem., 2004, 4293. 111 K. Kacprzak and J. Gawronski, Synthesis, 2001, 961. 112 S. K. Tian, Y. Chen, J. Hang, L. Tang, P. McDaid and L. Deng, Acc. Chem. Res., 2004, 37, 621. 113 L. Pasteur, C.R. Acad. Sci., 1853, 37, 110. 114 Optical Resolution Procedures for Chemical Comounds, Optical Resolution Information Center, Manhattan College, Riverdale, New York, 1981 115 S. Rogozhin and V. A. Davankov, Russ. Chem. Rev. (Uspekhi Khimii), 1968, 37, 1327. 116 C. Czerwenka, N. M. Maier and W. Lindner, J. Chromatography A, 2004, 1038, 85. 117 W. R. Oberleitner, N. M. Maier and W. Lindner, J. Chromatography A, 2002, 960, 97. 118 T. Sakai and A. Hirose, Talanta, 2003, 59, 167. 119 A. A. Smaardijk and H. Wynberg, J. Org. Chem., 1987, 52, 135. 120 V. Prelog and M. Wilhelm, Helv. Chim. Acta, 1964, 37, 1634. 121 H. Pracejus, Fortschr. Chem. Forsch., 1967, 8, 493. 122 A. Ojida, T. Yamano, N. Taya and A. Tasaka, Org. Lett., 2002, 4, 3051. 123 B. Vakulya, Sz. Varga, A. Csa´mpai and T. Soo´s, Org. Letts., 2005, 7, 1967. 124 H. C. Kolb, P. G. Andersson and K. B. Sharpless, J. Am. Chem. Soc., 1994, 116, 1278. 125 K. B. Sharpless, W. Amberg, Y. L. Bennani, G. A. Crispino, J. Hartung, K. S. Jeong, H. L. Kwong, K. Morikawa, Z. M. Wang, D. Xu and X. L. Zhang, J. Org. Chem., 1992, 57, 2768. 126 G. A. Crispino, K. S. Jeong, H. C. Kolb, Z. M. Wang, D. Xu and K. B. Sharpless, J. Org. Chem., 1993, 58, 3785. 127 L. Wang and K. B. Sharpless, J. Am. Chem. Soc., 1992, 114, 7568. 128 U. H. Dolling, P. Davis and E. J. J. Grabowski, J. Am. Chem. Soc., 1984, 106, 446. 129 A. Bhattacharya, U. H. Dolling, E. Grabowski, S. Karady, K. Ryan and L. Weinstock, Angew. Chem. Int. Ed. Engl., 1986, 25, 476. 130 R. S. E. Conn, A. V. Lovell, S. Karady and L. M. Weinstock, J. Org. Chem., 1986, 51, 4710. 131 D. L. Hughes, U. H. Dolling, K. M. Ryan, E. F. Schoenewaldt and E. J. J. Grabowski, J. Org. Chem., 1987, 52, 4745. 132 M. Masui, A. Ando and T. Shioiri, Tetrahedron Letters, 1988, 29, 2835. 133 M. J. O’Donnell, W. D. Benett and S. Wu, J. Am. Chem. Soc., 1989, 11, 2353. 134 W. Nerinckx and M. Vandewalle, Tetrahedron: Asymmetry, 1990, 1, 265. 135 K. B. Lipkowitz, M. W. Cavanaugh, B. Baker and M. J. O’Donnell, J. Org. Chem., 1991, 56, 5181. 136 T. B. K. Lee and G. S. K. Wong, J. Org. Chem., 1991, 56, 872. 137 A. Ando, T. Miura, T. Tatematsu and T. Shioiri, Tetrahedron Letters, 1993, 34, 1507. 266 | Catalysis, 2010, 22, 144–278

138 M. J. O’Donnell, S. Wu and J. C. Huﬀman, Tetrahedron, 1994, 50, 4507. 139 B. Lygo and P. G. Wainwright, Tetrahedron Letters, 1997, 38, 8595. 140 M. J. O’Donnell, Z. Fang, X. Ma and J. C. Huﬀman, Heterocycles, 1997, 46, 617. 141 E. J. Corey, F. Xu and M. C. Noe, J. Amer. Chem. Soc., 1997, 119, 12414. 142 B. Lygo and P. G. Wainwright, Tetrahedron Letters, 1998, 39, 1599. 143 B. Lygo, J. Crosby and A. J. Peterson, Tetrahedron Letters, 1999, 40, 8671. 144 E. V. Dehmlow, S. Wagner and A. Mu¨ller, Tetrahedron, 1999, 55, 6335. 145 B. Lygo and D. C. M. To, Tetrahedron Letters, 2001, 42, 1343. 146 H. Park, B. S. Jeong, M. S. Yoo, J. H. Lee, M. K. Park, Y. J. Lee, M. J. Kim and S. Jew, Angew. Chem., Int. Ed., 2002, 41, 3036. 147 B. Lygo and D. L. Humphreys, Tetrahedron Letters, 2002, 43, 6677. 148 R. Chinchilla, P. Mazon, C. Najera and F. J. Ortega, Tetrahedron: Asymmetry, 2004, 15, 2603. 149 S. Kumar and U. Ramachandran, Tetrahedron, 2005, 61, 7022. 150 S. Elango, M. Venugopal, P. S. Suresh and P. S. Eni, Tetrahedron, 2005, 61, 1443. 151 M. S. Yoo, B. S. Jeong, J. H. Lee, H. Park and S. Jew, Organic Letters, 2005, 7, 1129. 152 M. B. Andrus, Z. Ye and J. Zhang, Tetrahedron Letters, 2005, 46, 3839. 153 A. Siva and E. Murugan, J. Mol. Catal. A: Chem., 2005, 241, 111. 154 K. Hermann and H. Wynberg, Helv. Chim. Acta, 1977, 60, 2208. 155 N. Kobayashi and K. Iwai, J. Am. Chem. Soc., 1978, 100, 7071. 156 P. Hodge, E. Khoshdel and J. Waterhouse, J. Chem. Soc., Perkin Trans., 1983, 1, 2205. 157 M. Inagaki, J. Hiratake, Y. Yamamoto and J. Oda, Bull. Chem. Soc. Jpn., 1987, 60, 4121. 158 A. Sera, K. Takagi, H. Katayama, H. Yamada and K. Matsumoto, J. Org. Chem., 1988, 53, 1157. 159 B. M. Kim and K. B. Sharpless, Tetrahedron Letters, 1990, 31, 3003. 160 D. Pini, A. Petri, A. Nardi, C. Rosini and P. Salvadori, Tetrahedron Letters, 1991, 32, 5175. 161 B. B. Lohray, A. Thomas, P. Chittari, J. R. Ahuja and P. K. Dhal, Tetrahedron Letters, 1992, 33, 5453. 162 B. B. Lohray, Tetrahedron: Asymmetry, 1992, 3, 1317. 163 H. Han and K. D. Janda, J. Am. Chem. Soc., 1996, 118, 7632. 164 H. Han and K. D. Janda, Tetrahedron Letters, 1997, 38, 1527. 165 E. Nandanan, A. Sudalai and T. Ravindranathan, Tetrahedron Letters, 1997, 38, 2577. 166 L. Canali, E. S. Song and D. C. Sherrington, Tetrahedron: Asymmetry, 1998, 9, 1029. 167 C. E. Song, C. R. Oh, S. W. Lee, S. Lee, L. Canali and D. C. Sherrington, Chem. Commun., 1998, 2435. 168 R. Alvarez, M. A. Hourdin, C. Cave, J. d’Angelo and P. Chaminade, Tetrahedron Letters, 1999, 40, 7091. 169 R. Chinchilla, P. Mazon and C. Najera, Tetrahedron: Asymmetry, 2000, 11, 3277. 170 D. Bernstein, S. France, J. Wolfer and T. Lectka, Tetrahedron: Asymmetry, 2005, 16, 3481. 171 P. Hodge, E. Khoshdel, J. Waterhouse and J. M. J. Frechet, J. Chem. Soc., Perkin Trans., 1985, 1, 2327. 172 R. A. Aitken and J. Gopal, Tetrahedron: Asymmetry, 1990, 1, 517. Catalysis, 2010, 22, 144–278 | 267

173 S. Kawahara, A. Nakano, T. Esumi, Y. Iwabuchi and S. Hatakeyama, Organic Letters, 2003, 5, 3103. 174 M. Aune, A. Gogoll and O. Matsson, J. Org. Chem., 1995, 60, 1356. 175 G. D. H. Dijkstra, R. M. Kellogg and H. Wynberg, Recl. Trav. Chim. PaysBas., 1989, 108, 195. 176 G. D. H. Dijkstra, R. M. Kellogg and H. Wynberg, J. Org. Chem., 1990, 55, 6121. 177 L. Dupond, A. Konsur, K. Lewinski and B. Oleksyn, Acta Cryst., 1985, C41, 616. 178 S. Kashino and M. Haisa, Acta Cryst., 1983, C39, 310. 179 B. J. Oleksyn, Acta Cryst., 1982, B38, 1832. 180 B. J. Oleksyn, L. Lebioda and M. Ciechanowicz-Rutkowska, Acta Cryst., 1979, B35, 440. 181 L. Meurling, Chem. Scr., 1975, 7, 90. 182 H. Hiemstra and H. Wynberg, J. Am. Chem. Soc., 1981, 103, 417. 183 G. D. H. Dijkstra, R.M. Kellogg, H. Wynberg, J. S. Svendsen, I. Marko and K. B. Sharpless, J. Am. Chem. Soc., 1989, 111, 8069. 184 U. Berg, M. Aune and O. Matsson, Tetrahedron Letters, 1995, 36, 2137. 185 W. J. Colucci, S. K. Jungk and R. D. Gandour, Magn. Reson. Chem., 1985, 23, 335. 186 T. Sai, N. Takao and M. Sugiura, Magn. Reson. Chem., 1992, 30, 1041. 187 T. H. A. Silva, A. B. Oliveira and W. B. De Almeida, Bioorg. Med. Chem., 1997, 5, 353. 188 J. Thiel and P. Fiedorow, J. Mol. Struct., 1997, 405, 219. 189 T. Williams, R. G. Pitcher, P. Bommer, J. Gutzwiller and M. R. Uskokovic, J. Am. Chem. Soc., 1969, 91, 1871. 190 G. Ucello-Barretta, L. Di Bari and P. Salvadori, Magn. Reson. Chem., 1992, 30, 1054. 191 J. Gawronsky, M. Brzostowska and J. Koput, Croatica Chemica Acta, 1989, 62, 97. 192 H. U. Blaser, H. P. Jalett, W. Lottenbach and M. Studer, J. Am. Chem. Soc., 2000, 122, 12675. 193 E. Ta´las and J. L. Margitfalvi, Chirality, 2010, 22, 3. 194 H. U. Blaser, H. P. Jalett and F. Spindler, J. Mol. Catal. A: Chem., 1996, 107, 85. 195 M. Barto´k, Curr. Org. Chem., 2006, 10, 1533. + and T. Barto´k, React. Kinet. Catal. Lett., 196 M. Barto´k, K. Felfo¨ldi, Gy. Szo¨llosi 1999, 68, 371. + F. Bartha and 197 M. Barto´k, M. Sutyinszki, I. Bucsi, K. Felfo¨ldi, Gy. Szo¨llosi, T. Barto´k, J. Catal., 2005, 231, 33. 198 H. U. Blaser, in ‘‘Advances in Catalyst Design. Proceedings of the UNIDO Workshop on Catalyst Design’’, Trieste, Italy (11-13 December 1990), M. Grazini, C. N. R. Rao (Eds.), Singapore, World Scientiﬁc, 1991, p. 317 199 A. Pfaltz and T. Heinz, Top. Catal., 1997, 4, 229. 200 E. Orglmeister, T. Bu¨rgi, T. Mallat and Baiker, J. Catal., 2005, 233, 333. 201 C. Exner, A. Pfaltz, M. Studer and H. U. Blaser, Adv. Synth. Catal., 2003, 345, 1253. 202 S. Diezi, T. Mallat, A. Szabo´ and A. Baiker, J. Catal., 2004, 228, 162. 203 K. Felfo¨ldi, K. Bala´zsik and M. Barto´k, J. Mol. Catal. A: Chem., 2003, 202, 163. 204 K. Bala´zsik, K. Szo¨ri, K. Felfo¨ldi, B. To¨ro¨k and M. Barto´k, Chem. Commun., 2000, 555.

268 | Catalysis, 2010, 22, 144–278

205 A. Taskinen, V. Nieminen, M. Hotokka and D.Yu. Murzin, J. Phys. Chem. C, 2007, 111, 5128. + I. Bucsi and M. Barto´k, 206 Sz. Csere´nyi, K. Felfo¨ldi, K. Bala´zsik, Gy. Szo¨llosi, J. Mol. Catal. A: Chem., 2006, 247, 108. 207 G. Wang, T. Heinz, A. Pfaltz, B. Minder, T. Mallat and A. Baiker, J. Chem. Soc. Chem. Commun., 1994, 2047. 208 B. Minder, T. Mallat, A. Baiker, G. Wang, T. Heinz and A. Pfaltz, J. Catal., 1995, 154, 371. 209 E. Orglmeister, T. Bu¨rgi, T. Mallat and A. Baiker, J. Catal., 2005, 232, 137. 210 S. Diezi, M. Hess, E. Orglmeister, T. Mallat and A. Baiker, J. Mol. Catal. A: Chem., 2005, 239, 49. 211 T. Mallat and A. Baiker, Appl. Catal. A: Gen., 2000, 200, 3. 212 B. Minder, M. Schu¨rch, T. Mallat, A. Baiker, T. Heinz and A. Pfaltz, J. Catal., 1996, 160, 261. 213 H. U. Blaser and M. Studer, Acc. Chem. Res., 2007, 40, 1348. 214 H. U. Blaser and M. Mu¨ller, Stud. Surf. Sci. Catal., 1991, 59, 73. 215 S. P. Griﬃths, P. Johnston, W. A. H. Vermeer and P. B. Wells, J. Chem. Soc. Chem. Commun., 1994, 2431. 216 K. Felfo¨ldi, T. Varga, P. Forgo´ and M. Barto´k, Catal. Lett., 2004, 97, 65. 217 K. Fodor, A. Tungler, T. Ma´the´, S. Szabo´ and R. A. Sheldon, in: Catalysis of Organic Reactions, Chem. Ind., E.F. Herkes, (ed.), Vol. 75, New York, Marcel Dekker, 1998, p. 115. 218 A. Tungler, T. Ma´the´, K. Fodor, R. A. Sheldon and P. Gallezot, J. Mol. Catal. A: Chem., 1996, 108, 145. 219 G. Farkas, K. Fodor, A. Tungler, T. Ma´the´, G. To´th and R. A. Sheldon, J. Mol., Catal. A: Chem., 1999, 138, 123. 220 G. Farkas,_ . Sı´ pos, A. Tungler, A. Sa´rka´ny and J. L. Figueiredo, J. Mol. Catal. A: Chem., 2001, 170, 101. 221 A. Tungler and G. Fogassy, J. Mol. Catal. A: Chem., 2001, 173, 231. 222 E´´. Sı´ pos, A. Tungler and I. Bitter, in: Catalysis of Organic Reactions. Chemical Industries, D.G. Morrell, (ed.), Vol 89, New York-Basel, Marcell Dekker, 2003, p. 653. 223 A. Tungler, M. Kajta´r, T. Ma´the´, T. To´th, E. Fogassy and J. Petro´, Catal. Today, 1989, 5, 159. 224 E´´ . Sı´ pos, A. Tungler and I. Bitter, J. Mol. Catal. A: Chem., 2003, 198, 167. 225 B. Minder, M. Schu¨rch, T. Mallat and A. Baiker, Catal. Lett., 1995, 31, 143. 226 A. Marinas, T. Mallat and A. Baiker, J. Catal., 2004, 221, 666. 227 R. Hartung, J. G. E. Krauter, S. Seebald, P. Kukula, U. Nettekoven and M. Studer, Asymmetric Hydrogenation of Activated Ketones with catASium F214, http://www.degussacatalysts.com/asp/documents/79 Document.pdf. 228 P. A. Meheux, A. Ibbotson and P. B. Wells, J. Catal., 1991, 128, 387. 229 R. L. Augustine and S. K. Tanielyan, J. Mol. Catal. A: Chem., 1997, 118, 79. + and K. Felfo¨ldi, Catal. Lett., 1999, 230 M. Barto´k, T. Barto´k, G. Szo¨llosi 61, 57. 231 B. Toukoniitty, E. Toukoniitty, J. Kuusisto, J.-P. Mikkola, T. Salmi and D.Yu. Murzin, Chem. Eng. J., 2006, 120, 91. 232 D. Y. Murzin, E. Toukoniitty and J. Hajek, React. Kinet. Catal. Lett., 2004, 83, 205. 233 D. J. Jenkins, A. M. S. Alabdulrahman, G. A. Attard, K. G. Griﬃn, P. Johnston and P. B. Wells, J. Catal., 2005, 234, 230.

Catalysis, 2010, 22, 144–278 | 269

234 E. Toukonitty and D.Yu. Murzin, J. Catal., 2006, 241, 96. 235 A. Vargas, F. Hoxha, N. Bonalumi, T. Mallat and A. Baiker, J. Catal., 2006, 240, 203. + Gy. Szo¨llosi + and M. Barto´k, J. Catal., 2006, 244, 255. 236 K. Szori, 237 E. Toukoniitty, S. Franceschini, A. Vaccari and D.Yu. Murzin, Appl. Catal. A: Gen., 2006, 300, 147. 238 S. Diezi, A. Szabo´, T. Mallat and A. Baiker, Tetrahedron: Asymmetry, 2003, 14, 2573. 239 D. G. Blackmond, J. Catal., 1998, 176, 267. 240 D. Ferri, T. Bu¨rgi and A. Baiker, Chem. Commun., 2001, 1172. 241 I. Bakos, S. Szabo´, M. Barto´k and E. Ka´lma´n, J. Electroanal. Chem., 2002, 532, 113. 242 X. Zuo, H. Liu and M. Liu, Tetrahedron Lett., 1998, 39, 1941. 243 H. Bo¨nnemann and G. A. Braun, Chem. Eur. J., 1997, 3, 1200. 244 J. L. Margitfalvi, E. Ta´las, L. Yakhyaeva, E. Tﬁrst and I. Berto´ti, in: Catalysis of Organic Reactions, D.G. Morrell, (Ed.), Chem. Ind., Vol. 89, Dekker, New York, 2003, p. 393. 245 J. U. Ko¨hler and J. S. Bradley, Catal. Lett., 1997, 45, 203. 246 A. Kraynov and R. Richards, Appl. Catal. A: Gen., 2006, 314, 1. 247 K. E. Simons, A. Ibbotson, P. Johnston, H. Plum and P. B. Wells, J. Catal., 1994, 150, 321. 248 T. Marzialetti, M. Oportus, D. Ruiz, J. L. G. Fierro and P. Reyes, Catal. Today, 2008, 133, 711. 249 C. F. Yang, H. Y. Jiang, J. Feng, H. Yan Fu, R. X. Li, H. Chen and X. J. Li, J. Mol. Catal. A: Chem., 2009, 300, 98. 250 S. Bhaduri, V. S. Darshane, K. Sharma and D. Mukesh, J. Chem. Soc. Chem. Commun., 1992, 1738. 251 M. Maris, D. Ferri, L. Konigsmann, T. Mallat and A. Baiker, J. Catal., 2006, 237, 230. 252 Y. Huang, J. Chen, H. Chen, R. Li, Y. Li, L.-E. Min and X. Li, J. Mol. Catal. A: Chem., 2001, 170, 143. 253 H. X. Ma, H. Chen, Q. Zhang and J. X. Li, J. Mol. Catal. A: Chem., 2003, 196, 131. 254 M. Maris, T. Mallat and A. Baiker, J. Mol. Catal. A: Chem., 2005, 242, 151. 255 W. Xiong, H. X. Ma, Y. Y. Hong, H. Chen and X. J. Li, Tetrahedron: Asymmetry, 2005, 16, 1449. 256 R. Hess, F. Krumeich, T. Mallat and A. Baiker, J. Mol. Catal. A: Chem., 2004, 212, 205. 257 F. Hoxha, N. van Vegten, A. Urakawa, F. Krumeich, T. Mallat and A. Baiker, J. Catal., 2009, 261, 224. 258 T. Marzialetti, J. L. G. Fierro and P. Reyes, Catal. Today, 2005, 107–108, 235. 259 T. J. Hall, J. E. Halder, G. H. Hutchings, R. L. Jenkins, P. McMorn, P. B. Wells and R. P. K. Wells, Top. Catal., 2000, 11–12, 351. 260 W. Reschetilowski, U. Bo¨hmer and J. Wiehl, Stud. Surf. Sci. Catal., 1994, 84, 2021. 261 U. Bo¨hmer, F. Franke, K. Morgenschweis, T. Bieber and W. Reschetilowski, Catal. Today, 2000, 60, 167. 262 K. Bala´zsik, B. To¨ro¨k, G. Szakonyi and M. Barto´k, Appl.Catal. A: Gen., 1999, 182, 53. 263 M. A. Fraga, M. J. Mendes and E. Jordao, J. Mol. Catal. A: Chem., 2002, 179, 243. 264 L. Xing, F. Du, J. J. Liang, Y. S. Chen and Q. L. Zhou, J. Mol. Catal. A: Chem., 2007, 276, 191. 270 | Catalysis, 2010, 22, 144–278

265 J. T. Wehrli, A. Baiker, D. M. Monti and H. U. Blaser, J. Mol. Catal., 1990, 61, 207. 266 A. Kraynov, A. Suchopar and R. Richards, Catal. Lett., 2006, 110, 91. 267 M. Barto´k, K. Bala´zsik, T. Barto´k and Z. Kele, Catal. Lett., 2003, 87, 235. 268 K. Bala´zsik and M. Barto´k, J. Mol. Catal. A: Chem., 2004, 219, 383. 269 W.-R. Huck, T. Bu¨rgi, T. Mallat and A. Baiker, J. Catal., 2003, 216, 276. 270 E. Schmidt, D. Ferri and A. Baiker, Langmuir, 2007, 23, 8087. + P. Forgo´ and M. Barto´k, Chirality, 2003, 15, S82. 271 Gy. Szo¨llosi, 272 M. Barto´k, P. T. Szabo´, T. Barto´k, G. Szo¨llo¨si and K. Bala´zsik, Rapid Commun. Mass Spectrom., 2001, 15, 65. + and T. Barto´k, Ultrasonics Sonochem., 273 M. Barto´k, K. Bala´zsik, Gy. Szo¨llosi 1999, 5, 149. 274 J. L. Margitfalvi and E. Ta´las, Appl. Catal. A: Gen., 1999, 182, 65. + K. Bala´zsik and T. Barto´k, J. Mol. Catal., A: Chem., 275 M. Barto´k, Gy. Szo¨llosi, 2002, 177, 299. 276 H. U. Blaser, H. P. Jalett, D. M. Monti and J. T. Wehrli, Appl. Catal., 1989, 52, 19. 277 T. Mallat, S. Frauchiger, P. J. Kooyman, M. Schu¨rch and A. Baiker, Catal. Lett., 1999, 63, 121. 278 S. P. Griﬃths, P. Johnson and P. B. Wells, Appl. Catal. A: Gen., 2000, 191, 193. 279 B. Minder, T. Mallat, P. Skrabal and A. Baiker, Catal. Lett., 1994, 29, 115. + unpublished results. 280 J. L. Margitfalvi and M. Hegedus, 281 K. Bala´zsik and M. Barto´k, J. Catal., 2004, 224, 463. 282 B. To¨ro¨k, K. Felfo¨ldi, G. Szakonyi, K. Bala´zsik and M. Barto´k, Catal. Lett., 1998, 52, 81. 283 B. To¨ro¨k, K. Bala´zsik, Gy. Szo¨llo˜si, K. Felfo¨ldi, I. Kun and M. Barto´k, Ultrasonics Sonochem., 1999, 6, 97. 284 K. Bala´zsik, B. To¨ro¨k, K. Felfo¨ldi and M. Barto´k, Ultrasonics Sonochem., 1999, 5, 149. 285 B. Toukoniitty, O. Roche, J.-P. Mikkola, E. Toukoniitty, F. Klingstedt, K. Era¨nen, T. Salmi and D. Yu. Murzin, Chem. Eng. J., 2007, 126, 103. 286 (a) A. Tai, T. Hiraki and S. Murakami, Bull. Chem. Soc. Japan, 1983, 56, 1414; and (b) K. Ito, T. Harad and A. Tai, Bull. Chem. Soc. Japan, 1980, 53, 3367. 287 K. E. Simons, P. A. Meheux, A. Ibbotson and P. B. Wells, Stud. Surf. Sci. Catal. (ICC Budapest), 1993, 75, 2317. 288 (a) M. Arx, N. Dummer, D. J. Willock, S. H. Taylor, R. P. K. Wells, P. B. Wells and G. J. Hutchings, Chem. Commun., 2003, 1926; and (b) N. F. Dummer, R. P. K. Wells, S. H. Taylor, P. B. Wells and G. J. Hutchings, in Catalysis in Application (International Symposium on Applied Catalysis, Glasgow, Jul. 16–18, 2003) 2003, p. 278. 289 J. L. Margitfalvi, B. Minder, E. Ta´las, L. Botz and A. Baiker, Stud. Surf. Sci. Catal. (ICC Budapest), 1993, 75, 2471. 290 N. Ku¨nzle, R. Hess, T. Mallat and A. Baiker, J. Catal., 1999, 186, 239. 291 N. Ku¨nzle, T. Mallat and A. Baiker, Eur. Pat. Appl. 99112078.3. 292 N. Ku¨nzle, J. W. Soler, T. Mallat and A. Baiker, J. Catal., 2002, 210, 66. 293 N. Ku¨nzle, J. W. Soler and A. Baiker, Catal. Today, 2003, 79–80, 503. 294 E. Toukoniitty, P. Ma¨ki-Arvela, A. Kalantar Neyestanaki, T. Salmi, R. Sjo¨holm, R. Leino, E. Laine, P. J. Kooyman, T. Ollonqvist and J. Va¨yrynen, Appl. Catal. A: Gen., 2001, 216, 73. 295 D. M. Meier, D. Ferri, T. Mallat and A. Baiker, J. Catal., 2007, 248, 68. Catalysis, 2010, 22, 144–278 | 271

296 X. Li and C. Li, Catal. Lett., 2001, 77, 251. 297 N. Ku¨nzle, T. Mallat and A. Baiker, Appl. Catal. A: Gen., 2003, 238, 251. 298 R. Wandeler, N. Ku¨nzle, M. Schneider, T. Mallat and A. Baiker, J. Catal., 2001, 200, 377. 299 F. Gao, L. Chen and M. Garland, J. Catal., 2006, 238, 402. 300 D. Ruizand and P. Reyes, J. Chilean Chem. Soc., 2008, 53, 1740. 301 D. M. Meier, T. Mallat, D. Ferri and A. Baiker, J. Catal., 2006, 244, 260. 302 J. T. Wehrli, A. Baiker, D. M. Monti, H. U. Blaser and H. P. Jalett, J. Mol. Catal., 1989, 57, 245. 303 W. Reschetilowski, U. Bo¨hmer and K. Morgenschweis, Chem. Ing.Tech., 1995, 67, 205. 304 B. Panella, A. Vargas and A. Baiker, J. Catal., 2009, 261, 88. 305 V. Morawsky, U. Pru¨sse, L. Witte and K. D. Vorlop, Catal. Comm., 2000, 1, 15. 306 R. Hess, F. Krumeich, T. Mallat and A. Baiker, Catal. Lett., 2004, 92, 141. 307 J. L. Margitfalvi, in: Catalysis of Organic Reactions, M.G. Scaros and M.L. Prunier (Eds.), Chem. Ind., Vol 62, Dekker, New York, 1995, p. 189.. 308 J. M. Bonello, R. M. Lambert, N. Ku¨nzle and A. Baiker, J. Am. Chem. Soc., 2000, 122, 9864. 309 D. Ferri, S. Diezi, M. Maciejewski and A. Baiker, Appl. Catal. A: Gen., 2006, 297, 165. 310 S. Lavoie, M. A. Laliberte´, G. Mahieu, V. Derners-Carpentire and P. McBreen, J. Am. Chem. Soc., 2007, 129, 11668. 311 Z. Liu, X. Li, P. Ying, Z. Feng and C. Li, J. Phys. Chem. C, 2007, 111, 823. 312 J. M. Bonello, F. J. Williams, A. K. Santra and R. M. Lambert, J. Phys. Chem. B, 2000, 104, 9696. 313 M. Barto´k, K. Bala´zsik, G. Szo¨llo¨si and T. Barto´k, J. Catal., 2002, 205, 168. 314 N. Bonalumi, T. Bu¨rgi and A. Baiker, J. Am. Chem. Soc., 2003, 125, 13342. 315 S. Lavoie and P. H. McBreen, J. Phys. Chem. B., 2005, 109, 11986. 316 S. Lavoie, M. A. Liberte´ and P. H. McBreen, Catal. Lett., 2004, 97, 111. 317 P. Johnston and P. B. Wells, J. Catal., 1995, 156, 180. 318 O. J. Sonderegger, T. Bu¨rgi and A. Baiker, J. Catal., 2003, 215, 116. 319 T. Salmi, D.Yu. Murzin, J.-P. Mikkola, J. Wa¨rna˚, P. Ma¨ki-Arvela, E. Toukoniitty and S. Toppinen, Ind. Eng. Chem. Res., 2004, 43, 4540. 320 B. Toukoniitty, J.-P. Mikkola, D.Yu. Murzin and T. Salmi, Appl. Catal. A: Gen., 2005, 279, 1. 321 E. Toukoniitty and D.Yu. Murzin, Catal. Lett., 2004, 93, 171. 322 H. U. Blaser, B. Pugin and F. Spindler, J. Mol. Catal. A: Chem., 2005, 231, 1. 323 M. Schu¨rch, N. Ku¨nzle, T. Mallat and A. Baiker, J. Catal., 1998, 176, 569. 324 W. A. H. Vermeer, A. Fulford, P. Johnston and P. B. Wells, Chem. Commun., 1993, 1053. 325 E. Toukoniitty, P. Ma¨ki-Arvela, J. Kuusisto, V. Nieminen, J. Paivarinta, M. Hotokka, T. Salmi and D. Y. Murzin, J. Mol. Catal. A: Chem., 2003, 192, 116. 326 J. A. Slipszenko, S. P. Griﬃths, P. Johnston, K. E. Simons, W. A. H. Vermeer and P. B. Wells, J. Catal., 1998, 179, 267. 327 E. Toukoniitty, P. Ma¨ki-Arvela, M. Kuzma, A. Villela, A. K. Neyestanaki, T. Salmi, R. Sjo¨holm, R. Leino, E. Laine and D. Y. Murzin, J. Catal., 2001, 204, 281. 328 M. Studer, S. Burkhardt and H. U. Blaser, Chem. Commun., 1999, 1727.

272 | Catalysis, 2010, 22, 144–278

329 B. To¨ro¨k, K. Felfo¨ldi, K. Bala´zsik and M. Barto´k, Chem. Commun., 1999, 1725. 330 M. Bodmer, T. Mallat and A. Baiker, in: Catalysis in Organic Reactions, F. E. Herkes, (Ed.), Chemical Industries, Vol. 75, Marcel Dekker, New York, 1998, p. 75. 331 M. von Arx, T. Mallat and A. Baiker, J. Catal., 2000, 193, 161. 332 M. von Arx, T. Mallat and A. Baiker, Tetrahedron: Asymmetry, 2001, 12, 3089. 333 M. von Arx, T. Mallat and A. Baiker, Catal. Lett., 2002, 78, 267. 334 G. Z. Wang, T. Mallat and A. Baiker, Tetrahedron: Asymmetry, 1997, 8, 2133. 335 N. Ku¨nzle, A. Szabo´, M. Schu¨rch, G. Z. Wang, T. Mallat and A. Baiker, Chem. Commun., 1998, 1377. 336 A. Szabo´, N. Ku¨nzle, T. Mallat and A. Baiker, Tetrahedron: Asymmetry, 1999, 10, 61. 337 S. Diezi, M. Hess, E. Orglmeister, T. Mallat and A. Baiker, Catal. Lett., 2005, 102, 121. 338 S. Diezi, S. Reimann, N. Bonalumi, T. Mallat and A. Baiker, J. Catal., 2006, 239, 255. 339 K. Szo¨ri, Gy. Szo¨llo¨si and M. Barto´k, Adv. Synth. Catal., 2006, 348, 515. 340 R. Hess, A. Vargas, T. Mallat, T. Bu¨rgi and A. Baiker, J. Catal., 2004, 222, 128. 341 K. Szo¨ri K, B. To¨ro¨k and K. Felfo¨ldi, M., in: ‘‘Catalysis in Organic Reactions’’ M. E. Ford (Ed.) Chemical Industries Vol. 82, Marcel Dekker, New York, 2001. p. 489 342 E. Ta´las, and J. L. M. Margitfalvi, unpublished results. 343 H. U. Blaser, H. P. Jalett and J. Wiehl, J. Mol. Catal., 1991, 68, 215. 344 B. To¨ro¨k, K. Bala´zsik, G. Szo¨llo¨si, K. Felfo¨ldi and M. Barto´k, Chirality, 1999, 11, 470. 345 H. U. Blaser and H. P. Jalett, Stud. Surf. Sci. Catal., 1993, 78, 139. 346 E. Tukonitty, P. Ma¨ki-Arvela, J. Wa¨rna and T. Salmi, Catal. Today, 2001, 66, 411. 347 M. Studer, S. Burkhardt, A. F. Indolese and H. U. Blaser, Chem. Commun., 2000, 1327. 348 M. von Arx, T. Mallat and A. Baiker, J. Catal., 2001, 202, 169–176. 349 S. D. Jackson, J. Willis, G. D. McLellan, G. Webb, M. B. T. Keegan, R. B. T. Moyes, S. Simpson, P. B. Wells and R. Whyman, J. Catal., 1993, 139, 191. 350 H. U. Blaser, D. Imhof and M. Studer, Stud. Surf. Sci., 1997, 108, 175. 351 F. Hoxha, T. Mallat and A. Baiker, J. Catal., 2007, 248, 11. 352 T. Mallat and A. Baiker, J. Catal., 2007, 251, 246. 353 E. N. Jacobsen, I. Marko´, W. S. Mungal, G. Schroder and K. B. Sharples, J. Am. Chem. Soc., 1988, 110, 1968. 354 M. Sutyinszki, K. Szo¨ri, K. Felfo¨ldi and M. Barto´k, Catal. Lett., 2002, 81, 281. 355 X. B. Zuo, H. F. Liu and C. Yue, J. Mol Catal. A: Chem., 1999, 147, 63. 356 H. U. Blaser, H. P. Jalett and F. Spindler, J. Mol. Catal. A: Chem., 1996, 107, 85. + and M. Barto´k, J. Mol. Catal. 357 K. Bala´zsik, I. Bucsi, Sz. Csere´nyi, Gy. Szo¨llosi A: Chem., 2008, 285, 84. 358 J. L. Margitfalvi, H. P. Jalett, E. Ta´las, A. Baiker and H. U. Blaser, Catal. Lett., 1991, 10, 325. 359 A. Vargas, T. Bu¨rgi and A. Baiker, New J. Chem., 2002, 26, 807.

Catalysis, 2010, 22, 144–278 | 273

360 T. Varga, K. Felfo¨ldi, P. Forgo´ and M. Barto´k, J. Mol. Catal. A: Chem., 2004, 216, 181. 361 J. L. Margitfalvi and E. Ta´las, Appl. Catal. A: Gen., 2006, 301, 187. 362 R. Hess, T. Mallat and A. Baiker, J. Catal., 2003, 218, 453. 363 K. Felfo¨ldi, T. Varga, P. Forgo´ and M. Barto´k, Catal. Lett., 2004, 97, 65. 364 E. Toukoniitty and D. Y. Murzin, Catal. Lett., 2004, 93, 171. 365 I. Busygin, E. Toukoniity, R. Sillanpaa, D.Yu. Murzin and R. Leino, Eur. Org. Chem., 2005, 2811. 366 E. Toukoniity and D. Yu. Murzin, J.Catal., 2007, 251, 244. 367 E. Ta´las, J. L. Margitfalvi and O. Egyed, J.Catal., 2009, 266, 191. + 368 Gy. Szo¨llosi, Sz. Csere´nyi, F. Fu¨lo¨p and M. Barto´k, J. Catal., 2008, 260, 245. 369 R. L. Augustine and S. K. Tanielyan, J. Mol. Catal. A: Chem., 1996, 112, 93. 370 H. U. Blaser, H. P. Jalett, M. Mu¨ller and M. Studer, Catal. Today, 1997, 37, 441. 371 X. Li, R. P. K. Wells, P. B. Wells and G. J. Hutchings, J. Catal., 2004, 221, 653. 372 F. Hoxha, L. Ko¨nigsmann, A. D. Ferri, T. Mallat and A. Baiker, J. Am. Chem. Soc., 2007, 129, 10582. 373 C. Puchot, O. Samuel, E. Dunach, S. H. Zhao, C. Agami and H. B. Kagan, J. Am. Chem. Soc., 1986, 108, 2353. 374 D. Guillaneux, S. H. Zhao, O. Samuel, D. Rainford and H. B. Kagan, J. Am. Chem. Soc., 1994, 116, 9430. 375 H. B. Kogan, C. Girard, D. Guillaneux, D. Rainford, O. Samuel, S. Y. Zhang and S. H. Zhao, Acta Chemica Scandinavica, 1996, 50, 345. 376 Z. Ma and F. Zaera, J. Am. Chem. Soc., 2006, 128, 16414. 377 Z. Ma, I. Lee and F. Zaera, J. Am. Chem. Soc., 2007, 129, 16083. 378 L. Balazs, T. Mallat and A. Baiker, J. Catal., 2005, 233, 327. 379 K. Bala´zsik, Sz. Csere´nyi, Gy. Szo¨llosi, F. Fu¨lo¨p and M. Barto´k, Catal. Lett., 2008, 125, 401. 380 M. Barto´k, M. Sutyinszki, K. Bala´zsik and G. Szo¨llosi, Catal. Lett., 2005, 100, 161. + and M. Barto´k, Catal. Lett., 2008, 124, 46. 381 K. Bala´zsik, Gy. Szo¨llosi 382 N. F. Dummer, R. Jenkins, X. Li, S. M. Bawaked, P. McMorn, A. Burrows, C. J. Kiely, R. F. K. Wells, D. J. Willock and G. J. Hutchings, J. Catal., 2006, 243, 165. + J. Catal., 2006, 239, 74. 383 M. Bartu´k, K. Bala´zsil, I. Bucsi and Gy. Szo¨llosi, 384 M. von Arx, T. Mallat and A. Baiker, Angew. Chem. Int. Ed., 2001, 40, 2303. + G. Fogassy, M. Bartu´k and 385 K. Bala´zsik, T. A. Martinek, I. Bucsi, Gy. Szo¨llosi, G. A. Olah, J. Mol. Catal. A: Chem., 2007, 272, 265. 386 M. Barto´k, M. Sutyinszki and K. Felfo¨ldi, J. Catal., 2003, 220, 207. 387 N. Bonalumi, A. Vargas, D. Ferri and A. Baiker, J. Phys. Chem. C, 2007, 111, 9349. 388 N. Bonalumi, A. Vargas, D. Ferri and A. Baiker, Chem. Eur. J., 2007, 13, 923. 389 A. Vargas, T. Bu¨rgi and A. Baiker, J. Catal., 2004, 226, 69. 390 I. Busygin, J. Wa¨rna˚, E. Toukoniitty, D. Yu. Murzin and R. Leino, J. Catal., 2008, 254, 339. 391 I. Busygin, V. Nieminen, A. Taskinen, J. Sinkkonen, E. Toukoniitty, Rr. Sillanpa, D. Yu. Murzin and R. Leino, J. Org. Chem., 2008, 73, 6559. + ¨ lo¨s, Catal. Today, 1989, 6, 73. 392 J. L. Margitfalvi, E. Ta´las and S. Gobo 393 J. L. Margitfalvi, Gy. Vanko´, I. Borba´th, A. Tompos and A. Ve´rtes, J. Catal., 2000, 190, 474. 274 | Catalysis, 2010, 22, 144–278

+ and A. Tompos, Appl. Catal. A: 394 J. L. Margitfalvi, I. Borba´th, M. Hegedus Gen., 2002, 229, 35. 395 L.-Ch. de Me´norval, A. Chaqroune, B. Coq and F. Figueras, J. Chem. Soc., Faraday Trans., 1997, 93, 3715. 396 S. J. Pratt, S. J. Jenkins and D. King, Surf. Sci., 2005, 585, L159. 397 M. Barto´k, P. T. Szabo´, T. Barto´k and G. Szo¨llosi, Rapid Commun. Mass Spectrom., 2000, 14, 509. 398 D. Ferri, T. Bu¨rgi and A. Baiker, A. J. Chem. Soc. Perkin Trans., 1999, 2, 1305. 399 J. L. Margitfalvi, E. Ta´las and M. Hegedus, Chem. Comm, 1999, 656–646. 400 J. L. Margitfalvi, E. Ta´las F. Zsila and S. Kristya´n, Tetrahedron: Asymmetry, 2007, 18, 750. 401 J. L. Margitfalvi, E. Ta´las, N. Marı´ n-Astorga N and T. Marzialetti, in: Catalysis of Organic Reactions, Sowa, (ed.), Chem. Ind., Vol. 104, New York, Taylor&Francis, 2005, p. 535. 402 E. Ta´las E and J. L. Margitfalvi, Catal. Commun., 2008, 9, 984. 403 G. A. Attard, D. J. Jenkins, O. A. Hazzazi, P. B. Wells, J. E. Gillies, K. G. Griﬃn and P. Johnston, P., In: Catalysis in Application, S. D. Jackson, J. S. J. Hardgreaves and D. Lennon, D., (Eds.), The Royal Society of Chemistry, Cambridge, 2003, p. 71 (483). 404 G. A. Attard, J. E. Gillies, C. A. Harris, D. J. Jenkins, P. Johnston, M. A. Price, D. J. Watson and P. B. Wells, Appl.Catal. A: General, 2001, 222, 393. 405 (a) S. Szabo´ and F. Nagy, J. Electroanal. Chem., 1978, 87, 261; and (b) S. Szabo´ and F. Nagy, J. Electroanal. Chem., 1978, 88, 259. 406 S. Diezi, D. Ferri, A. Vargas, T. Mallat and A. Baiker, J. Am. Chem. Soc., 2006, 128, 4048. + Chem Comm., 2002, 407 M. Barto´k, M. Sutyinszki, K. Felfo¨ldi and G. Szo¨llosi, 1130. 408 R. Hess, A. Vargas, T. Mallat, T. Bu¨rgi and A. Baiker, J. Catal., 2004, 222, 117. + A. Chatterjee, P. Forgo´, M. Barto´k and F. Mizukami, J. Phys. 409 G. Szo¨llosi, Chem. A, 2005, 109, 860. 410 T. Mallat and A. Baiker, J. Catal., 1998, 176, 271. 411 O. J. Sonderegger, T. Bu¨rgi, L. K. Limbach and A. Baiker, J. Mol. Catal., A: Chemical, 2004, 217, 93. 412 G. Vayner, K. N. Houk and Y. K. Sun, J. Am. Chem. Soc., 2004, 126, 199. 413 E. Orglmeister, T. Mallat and A. Baiker, J. Catal., 2005, 234, 242. 414 A. Solladie 0 -Cavallo, F. Hoernel, M. Schmitt and F. Garin, Tetrahedron Letters, 2002, 43, 2671. 415 M. J. Karplus, Chem. Phys., 1959, 30, 11. 416 T. Bu¨rgi, Zh. Zhou, N. Ku¨nzle, T. Mallat and Alfons Baiker, J. Catal., 1999, 183, 405. 417 R. A. Olsen, D. Borchardt, L. Mink, A. Agarwal, L. J. Mueller and F. Zaera, J. Am. Chem. Soc., 2006, 128, 15594. 418 T. A. Martinek, T. Varga, F. Fu¨lo¨p and M. Barto´k, J. Catal., 2007, 246, 266. 419 T. A. Martinek, T. Varga, K. Bala´zsik, G. Szo¨llosi, F. Fu¨lo¨p and M. Barto´k, J. Catal., 2008, 255, 296. 420 O. J. Sonderegger, G. M.-W. Ho, T. Bu¨rgi and A. Baiker, J. Catal., 2005, 230, 499. 421 A. Vargas, N. Bonalumi, D. Ferri and A. Baiker, J. Phys. Chem. A, 2006, 110, 1118. 422 Y. Zhao, F. Gao, L. Chen and M. Garland, J. Catal., 2004, 221, 274. 423 L. Chen, Y. Zhao, F. Gao and M. Garland, Appl. Spectrosc., 2003, 57, 797. Catalysis, 2010, 22, 144–278 | 275

424 T. Bu¨rgi, F. Atamny, R. Schlo¨gl and A. Baiker, J. Phys. Chem. B, 2000, 104, 5953. 425 A. F. Carley, M. K. Rajumon, M. W. Roberts and P. B. Wells, J. Chem. Soc., Faraday Trans., 1995, 91, 2167. 426 T. E. Jones and C. J. Baddeley, Surf. Sci., 2002, 519, 237. 427 J. M. Bonello, F. J. Williams and R. M. Lambert, J. Am. Chem. Soc., 2003, 125, 2723. 428 J. Kubota and F. Zaera, J. Am. Chem. Soc., 2001, 123, 11115. 429 F. Zaera, J. Phys. Chem., 2008, 112, 16196. 430 W. Chu, R. J. LeBlanc and C. T. Williams, Catal. Comm., 2002, 3, 547. 431 D. Ferri and T. Bu¨rgi, J. Am. Chem. Soc., 2001, 123, 12074. 432 D. Ferri, T. Bu¨rgi and A. Baiker, J. Catal., 2002, 210, 160. 433 T. Bu¨rgi and A. Baiker, J. Phys. Chem. B, 2002, 106, 10649. 434 W. Chu, R. J. LeBlanc, C. T. Williams, J. Kubota and F. Zaera, J. Phys. Chem. B., 2003, 107, 14365. 435 N. Bonalumi, A. Vargas, D. Ferri, T. Bu¨rgi, T. Mallat and A. Baiker, J. Am. Chem. Soc., 2005, 127, 8467. 436 J.-D. Gru¨nwaldt and A. Baiker, Phys . Chem. Chem. Phys., 2005, 7, 3526. 437 M. Castonguay, J.-R. Roy, A. Rochefort and P. H. McBreen, J. Am. Chem. Soc., 2000, 122, 518. 438 T. Bu¨rgi, F. Atamny, A. Knop-Gericke, M. Ha¨vecker, T. Schedel- Niedrig, R. Schlo¨gl and A. Baiker, Catal. Lett., 2000, 66, 109. 439 J. Lai, Z. Ma, L. J. Mueller and F. Zaera, J. Phys. Chem. B, 2009, 113, 11696. 440 I. Lee, Z. Ma, Sh. Kaneko and F. Zaera, J. Am. Chem. Soc., 2008, 130, 14597. 441 S. R. Calvo, R. J. LeBlanc, C. T. Williams and P. B. Balbuena, Surf. Sci., 2004, 563, 57. 442 A. Vargas and Baiker, J. Catal., 2006, 239, 220. 443 M. Wahl, M. von Arx, T. A. Jung and A. Baiker, J. Phys. Chem. B, 2006, 110, 21777. 444 M. von Arx, M. Wahl, T. A. Jung and A. Baiker, Phys. Chem. Chem. Phys., 2005, 7, 273. 445 O. A. Hazzazi, G. A. Attard and P. B. Wells, J. Mol. Catal. A: Chem., 2004, 216, 247. 446 K. E. Simons, P. A. Meheux, S. P. Griﬃths, I. M. Sutherland, P. Johnston, P. B. Wells, A. F. Carley, M. K. Rajumon, M. W. Roberts and A. Ibbotson, Recl. Trav. Chim. Pays-Bas., 1994, 113, 465. 447 N. Fietkau, R. Bussar and H. Baltruschat, Electrochimica Acta, 2006, 51, 5626. 448 M. S. Schneider, A. Urakawa, J. D. Grunwaldt, T. Bu¨rgi and A. Baiker, Chem. Comm., 2004, 744. 449 M. Barto´k, K. Felfo¨ldi, B. To¨ro¨k and T. Barto´k, Chem. Comm., 1998, 2605. 450 I. Bucsi, M. Sutyinszki, K. Felfo¨ldi and M. Barto´k, Catal. Comm., 2006, 7, 104. 451 M. Barto´k, K. Felfo¨ldi, Gy. Szo¨llosi and T. Barto´k, Catal. Lett., 1999, 61, 1. 452 W. J. Hehre, W. W. Huang, P. E. Klunzinger, B. J. Deppmeier and A. J. Driessen: Spartan Manual, Wavefunction, Inc., 18401 Von Karman Ave., Suite 370, Irvine, CA 92612 453 Gaussian 98, Revision A.6, J. A. Pople et al., Gaussian, Inc., Pittsburgh PA, 1998 454 D. Ferri, T. Bu¨rgi and A. Baiker, J. Chem. Soc., Perkin Trans., 2000, 2, 221. 455 T. Bu¨rgi and A. Baiker, J. Catal., 2000, 194, 445.

276 | Catalysis, 2010, 22, 144–278

456 A. Vargas, T. Bu¨rgi, M. von Arx, R. Hess and A. Baiker, J. Catal., 2002, 209, 489. 457 O. Schwalm, B. Minder, J. Weber and A. Baiker, Catal. Lett., 1994, 23, 271. 458 O. Schwalm, J. Weber, J. L. Margitfalvi and A. Baiker, J.Mol.Struct., 1993, 297, 285. 459 A. Taskinen, V. Nieminen, E. Toukoniitty, D.Yu. Murzin and M. Hotokka, Tetrahedron, 2005, 61, 8109. 460 O. Schwalm, Ph.D. Thesis, University of Geneva, No 2754, 1995 461 J. L. Margitfalvi and E. Tﬁrst, J. Mol. Catal. A: Chemical, 1989, 139, 81. 462 L. Mink, Z. Ma, R. A. Olsen, J. N. James, D. S. Sholl, J. L. Mueller and F. Zaera, Top. Catal., 2008, 48, 120. 463 E. Schmidt, D. Ferri, A. Vargas and A. Baiker, J. Phys. Chem. C, 2008, 112, 3866. 464 C. Houtman and M. A. Barteau, J. Phys. Chem., 1991, 95, 3755. 465 Z. M. Liu and M. A. Vannice, Surf. Sci., 1994, 316, 337. 466 M. Mavrikakis and M. A. Barteau, J. Mol. Catal. A: Chem., 1998, 131, 135. 467 F. Ruette, F. M. Poveda, A. Sierraalta and J. Rivero, Surf. Sci., 1996, 349, 241. 468 A. Vargas, S. Reimann, S. Diezi, T. Mallat and A. Baiker, J. Mol. Catal. A: Chemical, 2008, 282, 1. 469 I. Busygin, O. P. Tkachenko, V. Nieminen, V. Yu. Borovkov, R. Sillanpa, E. Toukoniitty, L. M. Kustov, D. Yu. Murzin and R. Leino, J. Phys. Chem. C, 2007, 111, 9374. 470 V. Nieminen, A. Taskinen, M. Hotokka and D. Yu. Murzin, J. Catal., 2007, 245, 228. 471 C. J. Baddeley, Top. Catal., 2003, 25, 17. 472 G. Bond, P. A. Meheux, A. Ibbotson and P. B. Wells, Catal. Today, 1991, 10, 371. 473 U. Maitra and P. Mathivanan, Tetrahedron: Asymmetry, 1994, 5, 1171. 474 B. Xiang, K. Snow and M. Belly, J.Org. Chem., 1993, 58, 993. 475 (a) J. F. Maddaluno, N. Gresh and C. Giessner-Prettre’d, J. Org. Chem., 1994, 59, 793; and (b) W. Oppolzer, M. Kurth, D. Reichlin, C. Chapuie, M. Mohnhaupt and F. Moﬀatt, Helv. Chim. Acta, 1981, 64, 2802. 476 G. B. Jones and B. J. Chapman, Synthesis-Stuttgart, 1995, 475. 477 T. J Hall, P. Johnston, W. A. H. Vermeer, S. R. Watson and P. B. Wells, Stud. Surf. Sci. Catal., 1996, 101, 221. 478 V. Nieminen, A. Taskinen, E. Toukoniitty, M. Hotokka and D.Yu. Murzin, J. Catal., 2006, 237, 131. 479 I. Kun, B. To¨ro¨k, K. Felfo¨ldi and M. Barto´k, Appl. Catal. A: General, 2000, 203, 71. 480 X. Zuo, H. Liu, D. Guo and X. Yang, Tetrahedron, 1999, 55, 7787. 481 J. W. D. Carneiro, C. D. de Oliveira, F. B. Passos, D.A. G. Aranda, P. Rogerio, P. R. N. de Souza and O. A. Antunes, Catal. Today, 2005, 107, 31. 482 M. Barto´k, K. Bala´zsik and F. Notheisz, React. Kinet. Catal. Lett., 2002, 77, 363. 483 S. Lavoie, G. Mahieu and P. H. McBreen, Angew. Chem., Int. Ed., 2006, 45, 7404. 484 O. Schwalm, J. Weber, B. Minder and A. Baiker, J. Quantum. Chem., 1994, 52, 1191. 485 K. E. Simons, Ph.D. thesis (University of Hull), 1994.

Catalysis, 2010, 22, 144–278 | 277

486 K. Szo¨ri, K. Bala´zsik, K. Felfo¨ldi and M. Barto´k, J. Catal., 2006, 241, 149. 487 K. Szori, K. Bala´zsik, K. Felfo¨ldi, I. Bucsi, S. Csere´nyi, G. Szo¨llosi, E. Vass, M. Hollo´si and M. Barto´k, J. Mol. Catal. A: Chemical, 2008, 294, 14. 488 J. W. D. Carneiro, C. D. B. de Oliveira, F. B. Passos, D. A. G. Aranda, P. R. N. de Souza and O. A. C. Antunes, J. Mol. Catal. A Chem., 2005, 226, 221. 489 F. Manger, Tetrahedron, 1983, 39, 1013. 490 D. R. Storm and D. E. Koshland Jr, J. Am. Chem. Soc., 1972, 94, 5805. 491 I. C. Lee and R. I. Masel, J. Phys. Chem. B, 2002, 106, 368–373. 492 I. C. Lee and R. I. Masel, J. Phys. Chem. B, 2002, 106, 3902–3908. 493 F. Fache, E. Schulz, M. L. Tommasino and M. Lemaire, Chem. Rev., 2000, 100, 2159.

278 | Catalysis, 2010, 22, 144–278

Gold catalysis in organic synthesis and material science Cristina Della Pina,a Ermelinda Fallettaa and Michele Rossia DOI: 10.1039/9781847559630-00279

1.

Introduction

One of the most exciting and unforeseen developments of the chemical research has been the recent application of gold in catalysis. In fact, this metal has become an important tool in organic synthesis several years after the ﬁrst reports on ethyne hydrochlorination and CO oxidation and now it is widely employed in many fundamental catalytic processes as oxidation, hydrogenation and coupling reactions.1–4 New applications of gold have been also proposed for commercial syntheses by academic and industrial researchers.5–8 An ultimate project concerns an inorganic reaction, that is the direct synthesis of hydrogen peroxide which has been developed by the impressive work of Hutchings’ group.1–3 Strategic application of gold is the selective transformation of renewable biological resources, a key importance task for balancing the CO2 cycle. In particular, valuable oxygenated compounds can be produced as new building blocks for further transformation. The catalytic conversion of carbohydrates and alcohols to the corresponding carbonylic or carboxylic compounds still maintains an attracting power, being the products employed as chemical intermediates and high value components for perfumery, food and pharmaceutical industry.1–8 Selective oxidations using the eco-friendly air or pure dioxygen, as the oxidant, and supported metals as catalysts, are earning a general praise: they represent the promising response to the environmental restrictions for the progressive shutdown of the traditional methods, causing undesired and toxic by-products. Gold catalysis enjoyed an important progress owing to the rapid advancement of nanotechnology and nanoscience, thus resulting in new applications for commercial syntheses by both academic and industrial research worlds.9–12 The strong scientiﬁc appeal towards the ‘‘precious metal’’ can be easily realized considering its peculiar property to discriminate inside chemical groups and geometrical positions,13–15 and its chemical stability, strictly related to the unique features of gold itself. The kinetic studies clearly show how much the activity is highly dependent on the size of metallic gold particles. In particular, many investigations on the liquid-phase oxidation of polyols, alcohols, carbohydrates demonstrate that only small gold particles are catalytically active,16,17 which is in line with the behaviour of gold particles employed in the gas-phase oxidation of carbon monoxide.1

a

Department of Chimica Inorganica, Metallorganica e Analitica ‘‘L. Malatesta’’ e ISTM-CNR, Universita` degli Studi di Milano, Via Venezian 21 20133, Milano, Italy

Catalysis, 2010, 22, 279–317 | 279 c

The Royal Society of Chemistry 2010

2. 2.1

Gold catalysis in organic synthesis Selective oxidation of alcohols

The reason why the oxidation of alcohols represents an attractive topic mainly lies in the wide diﬀusion of hydroxy-compounds, their easy availability from renewable sources and the proﬁtable employment of their derivatives as chemicals for organic synthesis. A great number of recent reviews deals with the catalytic oxidation of the C–OH group, outlining the evolution of the catalytic system from conventional Pt and Pd to more sophisticated Pt-Pd-Bi polymetallic systems, in order to increase selectivity and drop the deactivation process.18,19 A novel generation of catalysts for alcohols and polyols oxidation is represented by supported gold: its application leads to a dramatic improvement in selectivity and stability, thus fomenting an exciting competition among ruthenium, platinum, palladium catalysts.1–3,20 Corma and Hutchings’ research groups are quite active in this ﬁeld, resulting in fundamental achievements which have added a key progress in this topic.2,3 In particular, Hutchings et al.21 have shown the advantage in using continuous ﬂow reactors for the oxidation of glycerol under mild conditions: both monolith and meso-scale structured downﬂow slurry bubble column designs lead to an increment in the reaction rate and selectivity towards glyceric acid over autoclave. An ultimate method has now been proposed,22 based on a gold-immobilized microchannel ﬂow reactor for the oxidation of alcohols with molecular oxygen: Kobayashi and co-workers have shown how the oxidation of various alcohols proceeded easily to give the corresponding aldehydes and ketones in good to excellent yields. No leaching of gold was observed and the gold-immobilized capillary column could be continuously used for at least four days without loss of activity. Probably, it is the ﬁrst example of a microreactor that allowed full conversion of alcohols by aerobic oxidation of alcohols in microchannels. However, the ﬁrst systematic study on gold catalysis for selective liquid phase oxidation has been performed at Milan University, with the ambitious aim to ﬁnd a substitute for palladium, platinum and, particularly, copper in the aerobic oxidation of the alcoholic group. The most challenging drawbacks to be overcome were metal leaching and scarce selectivity of the traditional catalysts. The early experiments for testing the activity of metal gold were disheartening: whereas bulk copper quickly reacted with O2 and ethane-1,2-diol in basic solution to produce oxoethanoate- and formate-derivatives,23 gold powder resulted totally inert towards any transformation of the glycol. The high chemical stability of bulk gold was overcome by discovering the new peculiarities of gold nanoparticles and the logic of Scheme 1 was soon adopted also by other diﬀerent research groups for liquid and gas phase applications. The favourite methods for preparing catalysts were co-precipitation, deposition-precipitation and colloidal particles immobilisation. In our case, ﬁnely dispersed gold supported on carbon by metal sol immobilisation allowed to achieve an active and selective catalyst for liquid phase oxidation. 280 | Catalysis, 2010, 22, 279–317

Gold Long life but low activity metal

Nanotechnologies

High dispersion

Kinetics, Mechanism, Models

Metal-support interaction

Particle size Nature and role of the support Scheme 1 Tailoring eﬃcient gold catalysts.

2.2

Catalyst preparation

Eﬃcient, carbon-supported gold catalysts for liquid phase oxidation can be prepared starting from colloidal dispersions containing metallic gold (sol). Diﬀerently sized gold particles, in the range 2–10 nm, could be obtained by reducing chloroauric acid with NaBH4 in the presence of stabilizing agents as polyvinylalcohol (PVA), polyvinylpyrrolidone (PVP) and tetrahydroxymethylphosphonium chloride (THMP), glucose. Au (III) concentration is a key factor for tuning particle size. Either high resolution electron transmission microscopy (HR-TEM) or X-ray diﬀraction (XRD) techniques are applied for particle size determination after immobilisation of the sol on a useful supporting material, as copper grid and carbon powder. TEM shows the direct image of the metal particles, whereas Scherrer equation allows the calculation of the mean diameter from the half height width of the XRD pattern. Colloidal gold nanoparticles were generally collected on two types of activated carbons: for catalytic tests, Au was immobilised on a coconut derived carbon powder (AS=1300 m2 g 1 from Camel) at a level of 0.2– 0.8% (w/w), which was chosen for the low sulphur content, while for XRD determination 1–2% Au (w/w) was contacted with a pyrolytic carbon powder (AS=254 m2 g 1 from Cabot), which was selected for its fast adsorption property. The modulation of gold clusters size was achieved using initial solutions ranging from 25 mg L 1 (small particles) to 500 mg L 1 (large particles) of gold. 2.3

Oxidation of diols

Aliphatic 1,2 diols can be oxidised to the corresponding monocarboxylates with O2 under low pressure (1–3 bar) in the presence of the stoichiometric Catalysis, 2010, 22, 279–317 | 281

Table 1 Catalytic activity and selectivity of carbon dispersed metals in the oxidation of vicinal diols

HO

1

TOF (h ) Selectivity (%)

OH

Au/C

Pd/C

Pt/C

3500 98

500 77

475 71

Au/C

Pd/C

Pt/C

2000 99

720 90

650 89

OH OH

1

TOF (h ) Selectivity (%)

amount of NaOH. Supported gold particles were shown to be the best catalytic system, by comparing with Pd and Pt metals (Table 1).13,24 Developing the sol immobilisation technique and improving our ability in preparing small colloidal gold particles, gold catalyst activity could be grown from a few hundred TOF units up to 3500 h 1 in the case of glycolate and 2000 h 1 for lactate. Selectivity at 100% conversion was raised to surprisingly high values, diﬀerently from the lower performance of Pd and Pt catalysts. Regarding phenylethane-1,2-diol oxidation, however, a worse selectivity scenario appeared, probably due to a strong induction eﬀect of the phenyl group: two abundant by-products, namely benzoate and phenylglyoxylate, were detected together with the expected mandelate, in basic conditions (Scheme 2).

OH

OH OH +

O2

COO

Au/C

+

NaOH

By-products

COO+

By-products

O

-

COO

Scheme 2 Reaction products detected during the oxidation of phenylethane-1,2-diol (P) with Au/C catalyst. [P]=0.4 M; P/Au=500; P/NaOH=1; T=343 K.

An accurate comprehension of the experimental results can not set aside the alkali catalysed keto-enolic equilibrium c and internal Cannizzaro-type reaction f, reported in Scheme 3. The original gold selectivity could be evaluated by dropping the reaction pH to the value 7 when reactions c and f, together with the overoxidation of mandelate, were inhibited. 282 | Catalysis, 2010, 22, 279–317

O

O OH

b OH

O

e

h

OH c

O O

f OH

OH O

a

O

d

O-

g

OO O

Scheme 3

-

+ HCO3

Reaction pathway of phenylethane-1,2-diol oxidation in basic solution.

Scheme 4 shows that the formation of mandelate supports path a, due to the oxidation at the terminal carbon atom, while phenylglyoxylate and benzoate underpin path b favouring the oxidation at the internal carbon. O

O OH

b

O

OH OH

O O

OH

OH

O

a

O-

O O-

O O

-

Scheme 4 Reaction pathway of phenylethane-1,2-diol oxidation at pH 7.

The oxidation observed under these conditions concerned mainly the secondary alcoholic function, 62.5%, followed by the primary one, 45%. The selectivity towards mandelate can be increased by promoting reactions c and f. Table 2 visualizes how we were able to improve the selectivity from 45 to 83% by increasing alkali concentration and temperature.25 Table 2 Optimization of phenylethane-1,2-diol (PED) oxidation for producing mandelate with Au/C catalyst NaOH/PED

T (K)

Conversion %

Selectivity %

1 2 2 4

343 343 363 363

52 100 100 100

45 60 70 83

Non-vicinal glycols can also undergo selective oxidation: however, Table 3 shows the worse reactivity of 1,3 propanediol and diethyleneglycol oxidation Catalysis, 2010, 22, 279–317 | 283

Table 3 Oxidation of isolated diols with gold catalysts. Substrate=0.4M; substrate/Au=100; T=343 K; pO2=3 bar; pH=9.5

HO

1

TOF (h ) Selectivity (%)

OH Au/C

Au/TiO2

430 100

490 95

HO

OH O

1

TOF (h ) Selectivity (%)

Au/C

Au/TiO2

240 99

240 98

with respect to vicinal diols reported in Table 1, while the selectivity to monocarboxylates maintains always high values. The great interest in the synthesis of dicarboxylic acids has suggested a thorough study on the oxidation of diethyleneglycol, in order to force the reaction towards the double oxidation, by changing the amount of alkali, nature of catalyst and temperature. The optimization of the experimental conditions at O2 pressure 3 bar and substrate: Au ratio 1000 led to the production of monocarboxylate by gold on carbon, whereas gold on titania resulted in 45% of the diacarboxylate in the presence of 2 mol of NaOH at 363 K (Table 4). Other by-products were absent.26 In the case of Au/TiO2 catalyst, a positive metal-support interaction was therefore outlined.

Table 4 Inﬂuence of catalyst and experimental conditions on the oxidation of diethyleneglycol to mono- and dicarboxylates Catalyst

NaOH/Substrate

T (K)

T (h)

Conv %

Monoacid%

Diacid%

1%Au/C 1%Au/C 1%Au/C 1%Au/TiO2 1%Au/TiO2 1%Au/TiO2 1%Au/TiO2

1 2 2 1 1 2 2

343 343 363 343 363 343 363

4 4 1 4 2 3 6

96 80 83 95 95 94 100

99 97 98 98 96 70 55

1 3 2 2 4 30 45

2.4

Oxidation of other polyols

2.4.1 Glycerol. The huge and easy availability of glycerol as a byproduct of biodiesel has recently prompted research to transform this cheap compound into valuable chemicals.27,28 The application of gold catalysis in glycerol oxidation under mild conditions has been experimented mainly by two Groups. In spite of the variety of potential reaction products, originated by the general oxidative pathway reported in Scheme 5, Hutchings’group has underlined the high selectivity of gold: using graphite as a 284 | Catalysis, 2010, 22, 279–317

O O

O

OH OH Mesoxalic (or β-ketomalonic) acid

O HO

OH

HO

OH

OH OH Tartronic acid

Glycerol OH

O

OH

HO

O

HO

O

H Glyceraldehyde

Glyceric acid

O

O

OH Dihydroxyacetone

OH

O

HO

OH

OH

Hydroxypiruvic acid

Scheme 5 Reaction pathway of the aerobic oxidation of glycerol.

support, in water solution at 333 K and in the presence of NaOH, 100% selectivity to sodium glycerate could be readily achieved at 50–60% conversion.29 Another thorough investigation on glycerol oxidation has been carried out by Prati et al. in Milano. In a ﬁrst study, the relationship between catalyst morphology and selectivity was explored at total conversion: it has been found that larger gold particles (20 nm), supported on suitable carbons, show low TOFs but favour glycerate formation under mild conditions (303 K, 3 bar), thus allowing yields up to 92%.30 Exciting results were achieved by the Authors using bimetallic nanoparticles as supported catalyst.31 Two important points were highlighted: a) the activity can be improved by using the bimetallic Au-Pt and Au-Pd systems, thus demonstrating a synergistic eﬀect between the metals; b) the selectivity to the desired product can be aﬀected by the nature of the catalyst (particle size, alloyed phases and support) and experimental conditions. A smart use of the metals allowed a precise modulation of selectivity, whereas pure gold promotes glyceric acid formation, Pd addition favours further transformation to tartronic acid and Pt addition leads to carbon-carbon bond ﬁssion resulting in glycolic acid. Also in this case, the best performance was recorded with smaller particle size catalysts. Another important parameter is the atomic ratio of the metals in (AuxPdy)/C catalyst, which deeply inﬂuences activity and selectivity, as well as the supporting materials (Carbon, Graphite, TiO2, Ti/SiO2, SiO2). 2.4.2 Sorbitol. Supported gold catalysts have been employed in sorbitol oxidation (Scheme 6) and a comparison with Pd and Pt catalysts was carried Catalysis, 2010, 22, 279–317 | 285

OH

OH OH

HO OH

OH

Sorbitol

OH

OH

OH

OH

OH

O

HO

O

HO

OH

OH

O-

OH

O-

Gulonate

Gluconate -O

OH

OH O

O OH

OH

O-

Glucarate Scheme 6 Aerobic oxidation of sorbitol

180

Conversion % Selectivity % (Gluconate+Gulonate)

150

Selectivity % (Glucarate)

120

TOF (h-1)

90 60 30 0 1%Au/C

1%Pd/C

1%Pt/C

Fig. 1 Activity and selectivity of carbon dispersed metals in sorbitol oxidation.

out (Fig. 1).32 The presence of alkali and the use of carbon as the supporting material allowed the successful performance of the reaction. Monometallic gold catalysts led to humble TOFs, favouring the oxidation of the primary alcoholic function to monocarboxylates and leading to gluconate and gulonate with very low amounts of dicarboxylate (glucarate). An improvement of activity was achieved by employing bimetallic Au þ Pd and Au þ Pt catalysts, resulting in full conversion. The bimetallic system was superior to monometallic gold also for the selectivity to monocarboxylates at a given conversion, resulting in superior values almost independently from the nature of the second metal. 2.5

Oxidation of allyl alcohol to 3-hydroxypropionic acid

The importance of 3-hydroxypropionic acid as a new building block has been highlighted in a oﬃcial classiﬁcation.33 Presently, 3-hydroxypropionic acid is a rare and expensive chemical which is commercialized as an aqueous solution 286 | Catalysis, 2010, 22, 279–317

by a few suppliers. Beside the traditional stoichiometric reactions, also catalytic methods are reported for its synthesis and many patents have recently raised.34–40 In spite of all the eﬀorts, however, none of the proposed new processes is eﬀectively operating to our knowledge. Recently, a new, unexplored route to 3-hydroxypropionic acid based on the aerobic oxidation of allyl alcohol has been reported.41 Its economical advantage lies on the fact that allyl alcohol (2-propene-1-ol) is a large scale chemical (over 50 000 tonne per year production) derived from the petrochemical industry via propene and propene oxide as intermediates.42 As in many fortuitous discoveries, gold catalysis, very selective in the aerobic oxidation of many hydroxylated molecules under mild conditions, was tested in order to transform allyl alcohol to acrylic acid. Surprisingly, reacting allyl alcohol in aqueous alkali solutions, a slow oxidation takes place which produces 3-hydroxypropionate beside minor amounts of acrylate and glycerate. Indeed, the oxidation can be controlled by tuning the alkali amount and temperature, as reported in Table 5. Table 5 Allyl alcohol oxidation in the presence of 0.3%Au/C catalyst. ALA=allyl alcohol, ACA=acrylic acid, GLA=glyceric acid, GLY=glycerol. Reaction conditions: [allyl alcohol]=1M, pO2=3 bar, allyl alcohol/metal=4000 (molar ratio), t=24 h. Yields by HPLC analysis on the crude reaction product. Yield% Test

NaOH/ALA (molar ratio)

T (K)

Conv %

3-HPA

ACA

GLA

GLY

1 2 3 4 5

1 3 1 3 3

298 298 323 323 353

98 100 98 100 100

19 16 42 79 74

15 30.5 13 9.5 18

38.5 5 33 11 6

traces 0 traces 0 0

According to the reported data, a high selectivity at full conversion to the high value 3-hydroxypropionic acid can be achieved with an excess of NaOH at mild temperature, 323 K. Further experiments were carried out for comparing the home-made 0.3% Au/C catalyst with the reference catalyst 1.5% Au/TiO2 provided by World Gold Council,43 under the same conditions (Table 6). Table 6 Allyl alcohol oxidation in the presence of 1.5%Au/TiO2 catalyst. Experimental conditions as in Table 5. Yield % Test

NaOH/ALA (molar ratio)

T (1C)

Conv %

3-HPA

ACA

GLA

GLY

6 7 8 9 10

1 3 1 3 3

298 298 323 323 353

37 25 94 97 97

8 7 50 53 11

23.5 18 37 32 21.5

0 0 8 12 6

0 0 0 0 0

Catalysis, 2010, 22, 279–317 | 287

It can be outlined that Au/C catalyst presents the best results, in terms of yields of 3-hydroxypropionic acid (79% for Au/C against 53% for Au/TiO2). The reaction mechanism of allyl alcohol aerobic oxidation leading to the unexpected product is of hard interpretation. The progressive sampling during the 24 h, in fact, does not help to ﬁnd the ﬁrst transformation of the reagent, because no transient species could be detected beside acrylic acid and glyceric acid, while glycerol was detected only in trace amounts (o1%). Further tests showed that glycerol could be transformed into glyceric acid under the above reported conditions but neither acrylic acid, nor glyceric acid and 3-hydroxypropeneoxide (glycidol) could produce 3-hydroxypropionic acid in detectable amounts. Diﬀerent pathways for the various products during allyl alcohol oxidation should be guessed. Moreover, using acrolein as a one pot reagent no 3-hydroxypropionic acid was recorded beside the yellow solid material, probably a condensation product. The slow addition of acrolein (1.1 ml), in progressive small portions (ten microliters every 0, 5 h), to the reacting mixture resulted in the formation of 3-hydroxypropionate (10% yield after 24 h reaction). This test suggests acrolein as a probable non detectable intermediate, undergoing a base catalyzed Michael-addition of water to give 3-hydroxypropanal, whose rapid oxidation would give the observed 3-hydroxypropionate (Scheme 7). A speciﬁc catalyst, 0.3% Au/C, was prepared as in Section 2.2, using glucose as a protecting agent, decreasing the pH of the gold colloid to 4 by means of HCl and calcining at 673 K under H2 for 2 hours.41 2.6

Other alcohols

The conversion of aliphatic and aromatic alcohols to aldehydes under neutral conditions can be successfully performed by gold catalysis. The facile achievement of the corresponding carboxylates is reached in the presence of alkali. Hutchings et al. have reported exciting results using Au-Pd bimetallic catalysts, owing to a synergistic eﬀect between the metals,44 while Corma et al. have demonstrated the synergistic eﬀect between Au nanoparticles and the supporting nanometric CeO2 material under solvent-free conditions.45 For many practical applications in organic synthesis, gold catalysis applied to alcohols oxidation is relatively slow (TOF values around dozens or hundreds h 1). To overcome the intrinsic low activity of gold, Hutchings and co-workers studied the contribute of a second metal, in particular palladium. Thus, Au-Pd on TiO2 was reported to show an extraordinary enhancement compared to monometallic gold. Benzylic alcohol, under solventless conditions, could be oxidized to aldehyde ﬁve times faster in the presence of Au-Pd/TiO2 than Au/TiO2 at 373 K and 2 atm of O2 with selectivity to aldehyde over 90% at 75% conversion. Under similar conditions, an exceptional value of TOF=26 9000 h 1 was reported for 1-phenylethanol oxidation, using Au-Pd/TiO2. Other supports, as Al2O3 or Fe2O3, were not producing catalysts as active and selective as Au-Pd/TiO2. A variety of alcohols were selectively oxidized in the absence of solvent; however, when toluene or water were used as solvents, a general dropping in activity of the catalyst could be observed. 288 | Catalysis, 2010, 22, 279–317

OH HO

OH Glycerol

(traces) OH O

OH

HO

OH

Allyl alcohol

OH

O Glyceric acid

Acrylic Acid

O

Acrolein

H

OH

HO

O 3-hydroxypropionic acid Scheme 7 Pathways during the oxidation of allyl alcohol.

Corma et al. developed a peculiar catalyst by supporting Au on nanosized CeO2. This catalyst appeared to be not only highly selective toward the oxidation of alcohols to carbonylic derivatives, but also very active operating, without any solvent, at 353 K and atmospheric pressure. Benzylic and cinnamyl alcohols were smoothly oxidized to aldehydes, whereas secondary alcohols were transformed to ketones. Primary alcohols, such as 3-phenyl-1propanol, however produced 3-phenylpropyl-3-phenylpropanoate in 83% selectivity at 73% conversion. The detailed description of this wide and important research area has been recently reported.3 2.7

Aminoalcohols

The doping eﬀect of the amino-group on traditional metal catalysts may be the cause of the lack of literature regarding the aerobic oxidation of aminoalcohols. Gold has been shown to be a pleasant exception for oxidizing this important class of aminoacids.26,46–49 In fact, we ﬁrstly discovered that nanometric gold represents a much better catalytic system if compared to palladium and platinum under similar conditions (Table 7). Alkali increases Catalysis, 2010, 22, 279–317 | 289

Table 7 Catalytic oxidation of aminoalcohols with carbon dispersed metals in the: a) absence of alkali [Substrate]=0.4M, substrate/metal=1000, pO2=3 bar, T=343 K, t=2h. b) presence of alkali, [Substrate]=0.4M, substrate/metal=1000, substrate/NaOH =1, pO2=3 bar, T=343 K, t=2h.

HO NH2

Conversion (%) a) Conversion (%) b)

1% Au/C

5% Pd/C

5% Pt/C

3 20

0 0

0 0

NH2 OH H3C

Conversion (%) a) Conversion (%) b)

1% Au/C

5% Pd/C

5% Pt/C

22 65

0 0

0 0

HO NH2

Conversion (%) a) Conversion (%) b)

1% Au/C

5% Pd/C

5% Pt/C

3 20

0 0

0 0

NH2 OH H3C

Conversion (%) a) Conversion (%) b)

1% Au/C

5% Pd/C

5% Pt/C

22 65

0 0

0 0

the oxidation rate, despite the amino group already ensures the presence of a basic solution.26 The material employed as the gold support is relevant for aﬀecting the catalytic performance, as it has been outlined also in other cases: alumina is a better supporting material for gold nanoparticles than carbon (Table 8). An interesting problem of chemo-selectivity is present in the catalytic oxidation of aminoalcohols of formula R1R2N–(CH2)n–CH2OH. As highlighted in former experiments on primary aminoalcohols26 and in more recent results on tertiary amines,49 this oxidation can result into the corresponding aminoacid as well as the N-oxide. The resulting product is determined by the nature of the nitrogen substituents, experimental conditions and catalytic system. 290 | Catalysis, 2010, 22, 279–317

Table 8 Catalytic oxidation of aminoalcohols with 1% Au/Al2O3 in the presence of alkali. [Substrate]=0.4M; substrate/metal=1000; substrate/NaOH=1; pO2=3 bar; T=343 K; t=2 h. Substrate

H 2N

Conversion %

OH

23

100

NH2 OH H2N

OH

H2N

OH

27 32

OH

In the case of N-substituted aminoalcohols, the oxidation takes place exclusively at the nitrogen atom. Thus, the reaction of 3-dimethylamino-1propanol with O2 in the presence of gold-containing catalysts produces the corresponding N-oxide with 100% regioselectivity.49 Table 9 and Fig. 2 show that the oxidation of the amino group is possible both in the absence and in the presence of alkali. In the absence of alkali, 100% selectivity has been also observed with diﬀerent metal catalysts, but only gold containing catalysts allowed 100% conversion, while Pt/C resulted inert and Rh/C led to only 20% conversion towards unidentiﬁed compounds.

Table 9 Aerobic oxidation of 3-dimethylamino-1-propanol. Reaction conditions: substrate 0.4 M, substrate/M=1000, pO2=2 atm, T=363 K, t=24 h. Selectivity of N-oxide as a sum of the free N-oxide and its hydrated form. Catalyst

Conversion %

Selectivity to N-oxide %

Selectivity to Aminoacid %

1%Au/C 1%Au/Al2O3 1%Au/TiO2 1%Rh/C 1%Pt/C 0.5%Au-0.5%Rh/C 0.5%Au-0.5%Pt/C

100 100 95 20 0 33 40

100 100 100 0 0 100 100

0 0 0 0 0 0 0

These data underpin the observation that aliphatic amines cause the catalytic deactivation of the traditional noble metals. The same test carried out in the presence of NaOH at pH 10.8 resulted in worse performances with all the catalysts, except for Rh/C which arose its activity up to 33% (Fig. 2). A general evaluation on the role of alkali in promoting or depressing the catalyst activity is risky to be done, because of the limited number of tested substrates.26,49 The nature of substrates, however, has been outlined to be essential for aﬀecting gold catalytic performance. Catalysis, 2010, 22, 279–317 | 291

100

80

%

60

40

20

0 Au/C

Au/Al2O3 Au/TiO2

Conversion

Rh/C

Selectivity (N-oxide)

Pt/C

Au-Rh/C Au-Pt/C

Selectivity (Aminoacid)

Fig. 2 Oxidation of 3-dimethylamino-1-propanol. Reaction conditions: substrate 0.4M, substrate/M=1000, pO2=1, T=343 K, t=2h, pH=10.8.

3.

Selective oxidation of carbohydrates

The worldwide availability of carbohydrates is attracting the interest in their oxidative transformation. As a general trend, in the oxidation of aliphatic oxygenated compounds with supported gold particles the following order of reactivity has been observed: aldehydesWprimary alcoholsWsecondary alcohols; tertiary alcohols and carboxylic acids are almost inert under moderate conditions (up to 363 K and 3 bar). In particular, the aerobic oxidation of aldehydes can be performed using water, organic solvents and solvent free conditions, also in the absence of alkali. We discovered that Au is able to easily oxidize aldehydes in water solution and, diﬀerently from Pt, no deactivation was observed on recycling. According to the expected trend, catalytic aldose oxidation occurs at the aldehydic group leading to carboxylic acid or carboxylates.

3.1

Glucose to sodium gluconate

Gluconic acid and gluconates are industrial intermediates widely employed in food chemistry, surfactants and cleansing agents. That is why the oxidation of glucose, a cheap renewable starting material, appears as a charming topic. Moreover, the present industrial production is only via fermentation by enzyme (Aspergillus niger mould), but the low productivity of this process has prompted the interest in ﬁnding eco-friendly technologies based on the use of oxygen in aqueous solution, under mild conditions by heterogeneous catalysis. Many eﬀorts have been so far carried out: Pt-based catalysts, for example, allowed high conversion and good selectivity, but quickly deactivated because of leaching, self-poisoning and over-oxidation. These limits could be partly overcome by using bi- and tri-metallic catalysts and the promoting eﬀect of Bi.50,51 292 | Catalysis, 2010, 22, 279–317

The ﬁrst application of gold catalysis for transforming glucose into gluconates was soon excellent in terms of activity, selectivity and durability. A comparison among diﬀerent Pd, Pt and Au catalysts, performed at 323 K and atmospheric pressure, demonstrate the great peculiarity of gold: while palladium and platinum catalysts led to selectivityo95%, Au resulted in a selectivity close to 100% at total conversion.52 Although superior to Pt and Pd catalysts, in the ﬁrst experiments gold was shown to be less eﬃcient with respect to enzymatic catalysis. Encouraged by the promising achievements, eﬀorts were addressed to the deep improvement of the catalytic system, in order to ﬁnd an alternative process to the biochemical route. A fundamental contribute was also played by the mechanism studies (Section 3.1.1) which were carried out side by side. Starting from TOF of a few hundred h 1 units, we reached the exceptional value close to sixty thousand units,17 which is similar to the behaviour of enzymatic catalysis. Following a detailed study on the kinetics of enzymatic oxidation,53 we compared Hyderase (from Amano Enzyme Co., U.K.), a biological preparation containing glucose oxidase and catalase as active components and ﬂavine-adenine dinucleotide (FAD) as the rate controlling factor (1.3 10 6 mol g 1), with the most eﬃcient gold catalyst, 0.5% Au/C prepared as in Section 2.2, containing metal particles of 3.6 nm mainly at the surface. Kinetic data of glucose oxidation to gluconate were recorded using a glass reactor interfaced to an automatic titration device equipped with NaOH as a reagent. Magnetic stirring and high speed turbine were alternatively used during the tests. A careful control of various parameters such as pH, temperature, glucose concentration and stirring speed led to the comparative results reported in Scheme 8.53 Hyderase

Au/C

1M

3M

Catalyst/Glucose

6

5

pH

5-7

9.5

T

303 K

323 K

Stirring

900 rpm

39000 rpm

[Glucose]

Spec. Activity

Productivity

145 h-1

122 kg m-3 h-1

218 h-1

514 kg m-3 h-1

Scheme 8 Comparison between biological and inorganic catalysis in glucose oxidation.

Catalysis, 2010, 22, 279–317 | 293

Considering the molecular eﬃciency of the active FAD sites, a turnover frequency of 600 000 h 1 was obtained, being this value undoubtedly better than the eﬃciency of active external gold atoms in the inorganic catalyst, calculated as 90 000 h 1. However, taking into account the lower FAD concentration in the enzymatic extract and the potentiality of gold, which allows a threefold higher glucose concentration, the ﬁnal productivity is superior with gold by using a similar amount of total catalyst.

3.1.1 Kinetics and models. Kinetic investigations were performed on the selective liquid phase oxidation of glucose, using carbon supported gold particles54 or unsupported colloidal gold particles55 as the catalysts. The ﬁrst study was carried out by Claus and co-workers: they proposed a Langmuir-Hinshelwood model, considering the surface oxidation reaction as the limiting factor of the whole reaction rate, while adsorption of substrate and desorption of the product were regarded as fast steps. The Authors ﬁnally suggested a dehydrogenation mechanism converting glucose to sodium gluconate, supposing water as the reduction product of dioxygen and detecting a scarce eﬀect of glucose concentration on the reaction rate. In the second paper, the application of ‘‘naked’’ gold particles as a catalyst led to detect hydrogen peroxide-instead of water-as the reduction product of O2. Kinetic tests at low glucose concentrations (o0.1 M) recorded a ﬁrst reaction order, leading to an asymptote at higher concentrations (0.5 M). A ﬁrst order dependence was also found for the O2 concentration. An Eley– Rideal mechanism, characterised by the adsorption of glucose in its hydrated form on gold, was then suggested because of the optimum ﬁtting with the experimental evidence. As a consequence, a rate equation which ﬁts both the ﬁrst-order with respect to oxygen and the decreasing order with respect to glucose was found.55 In order to compare the enzymatic catalysis with the gold catalytic system, a kinetic investigation of glucose catalysed by the above cited Hyderase was performed under similar conditions applied for the gold catalysis. According to measurements of initial rate as a function of initial glucose concentration were recorded. A Michaelis-Menten mechanism resulted to be coherent with the experimental data.56 It is worth noting that both gold catalysis and enzymatic catalysis are able to promote the oxidation of glucose with the same stoichiometry where the 2 electrons reduction of molecular oxygen produces hydrogen peroxide as the transient by-product. Nevertheless, enzymatic and inorganic catalysis apply diﬀerent strategic mechanisms. Regarding the enzymatic system, the rate determining step is the oxidation of the substrate by the enzyme, which is transformed into the reduced form according to a faster step and showing a zero order with respect to dioxygen. The oxidation of glucose by dioxygen dissolved in water represents the rate determining step of the gold catalytic process, with a ﬁrst order dependence of the reaction rate on pO2. The corresponding rate determining steps interestingly recorded similar activation energies (47.0 kJ/mol for gold and 49.6 kJ/mol for enzymatic catalysis).55,56 294 | Catalysis, 2010, 22, 279–317

OH

O HO−

+

R

O−

R H

H

OH

OH O−

R

Au

+

Au− Au O

R H

H

O2

slow

degradation H

Au

O H2O2

+

+

R O−

Scheme 9

Au Au

O

R

O− O

OH

Molecular mechanism of aerobic glucose oxidation with gold nanoparticles.

Scheme 9 visualizes the proposed molecular mechanism of glucose oxidation on a gold nanoparticle.57 This mechanism suggests that the presence of alkali is fundamental for activating gold particles.52 Kinetic experiments conﬁrmed that there is no diﬀerence in the activity of unsupported and supported gold nanoparticles, except for a better stability of the supported gold catalyst in the time, thus excluding any role of the support in promoting the oxidation.57 The key role of the particle size in aﬀecting gold catalytic activity is a leit motiv also characterizing glucose oxidation. The sudden loss of activity of particles larger than 10 nm was recorded using unsupported colloidal particles (10 4 M Au, 0.38 M glucose, 303 K under oxygen at atmospheric pressure), due to a quasi-linear correlation between the initial speciﬁc molar activity (moles of reacted glucose/ mole total gold h) and the inverse of the mean diameter in the range 2–7 nm.17 A correlation between surface structure and catalytic behaviour in solid materials is of strategic importance for producing quick and clean industrial reactions. This has prompted to derive a geometrical model for describing the morphological properties of two catalysts made of carbon supported gold particles, prepared as in Section 2.2, having a known size distribution centred at 3.30 nm and 7.89 nm respectively.58 For this purpose, the progressive poisoning eﬀect of diﬀerent molecules on these catalysts, performed during the aerobic oxidation of glucose, has been used as a diagnostic tool. The observed deactivation trend follows the order thiocyanateWcyanide EcysteineWthiourea and each of them obeys an exponential law. The kinetics of catalyst deactivation has been interpreted by considering the Catalysis, 2010, 22, 279–317 | 295

contribute of electronic factors which overlap the space shielding of active sites, due to long range poison-catalyst interaction inﬂuencing the entire metal particle. The consequent insight in the aerobic oxidation of glucose suggested a molecular model for electronic interactions in gold nanoparticles: correlating the nature of the molecules, which caused a consistent poisoning effect, and considering the promoting eﬀect of OH , we found that the dioxygen reduction step is diﬀerently inﬂuenced by soft and hard-nucleophiles. In conclusion, it has been underlined how the competition of the poisoning molecule with the reagents can be discussed considering two extreme cases: for a hard nucleophile, no back-donation from metal to the Lewis base is expected, leaving in the reacting solution the original or a higher catalytic eﬀect, as in the case of OH . Regarding a nucleophile, N, showing p back-bonding ability, the removal of the electron density from the metal inhibits dioxygen reduction thus decreasing the catalytic property of the entire gold particle (Schemes 9, 10).

N

σ

π

H

O

Au R

O

O

OH

Scheme 10

3.2

Free gluconic acid

Many eﬀorts have been devoted to ﬁnd a one-pot synthesis of free gluconic acid, because the present method-from calcium gluconate and sulfuric acidleads to large amounts of CaSO4 as a by-product. Despite all the attempts, following either enzymatic or inorganic routes, no interesting results were so far recorded due to the inhibition of the catalytic systems at low pH values. Focusing on this challenging topic, multi-component catalysts resulted more active than monometallic gold particles. In particular, an important synergistic eﬀect between gold and platinum was observed (Fig. 3). The most promising Au þ Pt combination was further optimized, leading to a quite active catalyst for alkali-free oxidation of glucose containing gold and platinum in the ratio 2:1 (w/w). The synergistic eﬀect was detected in a series of experiments which compare colloidal catalysts and supported catalysts.59 A direct synthesis of gluconic acid by aerobic oxidation of glucose seems to be possible with gold-based catalysts, starting from the encouraging TOF values around 1000 h 1. 4.

Selective oxidation of hydrocarbons

In order to highlight the versatility of heterogeneous gold catalysis, some applications of unsubstituted hydrocarbon oxidation are considered. 296 | Catalysis, 2010, 22, 279–317

300 Conversion % TOF (h-1)

250

200 150 100 50 0 Au

Pt

Pd

Rh Au-Pt Au-Pd Au-Rh

Fig. 3 Aerobic oxidation of glucose with monometallic and bimetallic catalysts. Glucose/ Au=3000; T=343 K; pO2=3 bar; t=6.5h.

In particular, oxidation of propene to propene oxide (PO), ethene to vinylacetate monomer (VAM) and cyclohexane to cyclohexanol-cyclohexanone mixture (KA oil) were investigated in a more systematic manner. Here we present an overview of the partial oxidation of hydrocarbons by gold catalysis.

4.1

Propene epoxidation

All the eﬀorts to produce propene epoxide (PO) commercially by direct oxidation of propene, similarly to the silver promoted synthesis of ethene oxide, have been so far vain and the largest amount of PO is still manufactured by chlorohydrine process (49%) and the indirect hydroperoxide processes.42 The progressive request for eco-friendly processes, which can exclude chlorine dependence and huge amounts of undesired by-products, has prompted basic research to ﬁnd alternative catalytic routes. In this context, gold has been shown to be a very promising catalyst. In fact, the pioneering Haruta’s work has reported that supported gold catalyst is able to promote the gas phase epoxidation of propene by O2 in the presence of H2.60 The behaviour of gold is unique, as shown by comparing diﬀerent metals dispersed on titania (M=Au, Pt and Pd) under moderate conditions (298–353 K) when equimolecular amounts of H2 and O2 are reacted with C3H6. Only Au produces propene oxide (PO), while Pd and Pt promote mainly the hydrogenation of C3H6 to C3H8 and the formation of small amounts of acetone and carbon dioxide. All the experiments leading to the total selectivity to PO, underlined the strategic role of TiO2 as the support. However, the low conversion needed to reach high selectivity represents a weak point. As a consequence, most of the subsequent investigations were devoted to improve PO yields meanwhile maintaining a high selectivity. Catalysis, 2010, 22, 279–317 | 297

Insights on gold catalysts supported on non-porous and mesoporous titania-silica allowed further progress in PO productivity, also indicating the eﬀect of preparation conditions and pre-treatments on their activity and stability. A series of Au/TS-1 catalysts with diﬀerent gold and titanium contents was examined at 413–473 K at a space velocity of 7000 mL (h gcat) 1. A catalyst prepared with a Si/Ti=36 (atomic ratio) and a gold loading of 0.05 wt% produced 116 gPO(h kgcat) 1 at 473 K, which was the highest rate at that time reported for a TS-1-based catalyst with no deactivation during 40 h. Catalysts prepared with lower titanium and gold contents resulted in very active catalysts, up to 350 gPO(h gAu) 1 at 473 K for 0.01 wt% Au/TS-1 (Si/Ti=500), indicative of a more eﬃcient use of gold and titanium for the epoxidation reaction. The low gold loading coupled with non-detectable gold particles in TEM micrographs suggested that, in these materials, signiﬁcant activity is due to gold entities smaller than 2 nm.61 An eﬃcient Au capture on TS-1 support by a NH4NO3 treatment led to a fourfold increase in Au/TS-1 catalysts. The higher gold amount produced catalysts allowing quite high conversions of propene (5–10%) with acceptable selectivity (75–85%), at 473 K and a space velocity of 7000 ml (h g cat) 1. The related productivity resulted in 134 gPO (h kg cat) 1.62 4.2

Oxidation of ethene to vinyl acetate (VA)

Beside ethene oxide synthesis, successfully performed by silver catalysts,42 another important target in the selective oxidation of ethene is represented by the acetoxylation to vinylacetate (VAM), the latter being the monomeric unit for the production of polyvinylacetate (PVAc). Pd-Au bimetallic silicasupported catalyst, promoted by potassium acetate, is the well-known system commercially applied for the production of the monomer (VAM).3 The relevant added value of vinyl acetate monomer has attracted the interest of both academic and industrial research groups, thus leading to a number of patents. Researchers at Celanese International Corporation were particularly active in this area describing in details successful preparation methods63: the general procedure, similar to the deposition-precipitation technique, provides the active metals on the surface ﬁrstly as water-insoluble compounds which are reduced by a second step to the metallic form. A shell-impregnated catalyst, Pd-Au on a silica support, was also described for the synthesis of vinyl acetate which allowed a selectivity over 90%.64 4.3

Oxidation of higher alkenes

Supported gold catalysts have been employed in the aerobic oxidation of other alkenes without using a second reagent, H2, as a sacriﬁcial reductant or CH3COOH as the acetilating reagent. Hutchings et al.65,66 and other research groups67–69 found that alkene oxidation fairly proceeds by adding a catalytic amount of peroxides (either hydrogen peroxide or tert-butylhydroperoxide) as an oxygen chain initiator. These works fairly emphasize how much selectivity and conversion are dependent on substrate, catalyst and experimental conditions. The oxidation of cyclohexene, styrene, stilbenes and cyclooctene was performed, the catalysts being home made gold 298 | Catalysis, 2010, 22, 279–317

supported on carbon, alumina, titania which were compared with World Gold Council reference catalysts.43 Cyclohexene oxidation presented the highest selectivity to epoxide (50%) and ketone (26%) at 30% conversion using Au/C as a catalyst in 1,2,3,5tetramethylbenzene solvent (TMB). Moreover, a promoting eﬀect of bismuth on Au/C catalyst led to 98% selectivity of a valuable mixture of products.65 Styrene could be converted by aerobic oxidation into epoxide with a low selectivity (29%), by using either a mixture of 1,2,4,5-tetramethylbenzene (TMB)/1,4-dimethylbenzene (DMB) or hexaﬂuorobenzene as a solvent and 1% Au/C as a catalyst. However, the major oxidation product was benzaldehyde with a selectivity around 46% for both solvents, while acetophenone was achieved with a selectivity of 15% with 1,2,4,5-TMB/ 1,4-DMB and 11% with hexaﬂuorobenzene.65 Tert-butylhydroperoxide (TBHP) as the oxidant was employed by Ying et al.68 in styrene oxidation in the presence of gold on mesoporous alumina, obtaining 70% selectivity to epoxide along with a consistent production of benzaldehyde. Similarly, nanosized-gold deposited on TiO2 by deposition-precipitation method was shown to be an active and selective catalyst (around 50% selectivity) for the epoxidation of styrene by TBHP. Furthermore, more exotic oxides as supporting materials for the same reaction were investigated, such as gallium, indium and thallium oxides, thus revealing a pretty good performance of Au/Tl2O3 (around 60% selectivity).67 Cis-stilbene oxidation with dioxygen using a 1% Au/Graphite catalyst led to the corresponding epoxide, with cis : trans ratio depending on the solvent but always in favour of the trans conformation: 74% selectivity to transstilbene epoxide at 48% conversion was found by means of i-propylbenzene as a solvent.65 The oxidation of cis-cyclooctene was investigated using 1% Au/C catalysts both in the presence and absence of solvents by Hutchings et al.65 The best result was obtained with 1,2,3,5-TMB as a solvent, leading to 94% selectivity to the epoxide at 28% conversion while in the absence of solvent 81% epoxide at 8% conversion was obtained. 4.4

Oxidation of alkanes

The activation of C–H bond in the selective oxidation with dioxygen by gold catalysis has appeared to be a promising reality. Research has particularly focused on the synthesis of cyclohexanone and cyclohexanol, because it is a matter of growing interest for chemical industry. These oxidation products, in fact, are fundamental intermediates for making e-caprolactam and adipic acid, thus leading to nylon-6 and nylon-6,6 manufacture beside less important applications as stabilizers, homogenizers for soaps and synthetic detergent emulsions, and as solvents for lacquers and varnishes. Zhao and co-workers70 ﬁrst applied gold catalysis in the activation of cyclohexane: Au/ZSM-5 and Au/MCM-41 favoured a selectivity around 90% and conversions of 10-15% at 423 K, even though a loss in both activity and selectivity after their recycle is a drawback for industrial application. Catalysis, 2010, 22, 279–317 | 299

A number of eﬀorts has been carried out in order to reach a satisfactory one-pot oxidation of cyclohexane65,70–72: Au/graphite without any solvent, but using a halogenated benzene as an additive, led to 92% selectivity (cyclohexanone þ cyclohexanol) at low conversion (1%).65 Higher conversions (20–30%) and selectivity (95%) were achieved by Zhu et al.71 with gold on mesoporous silica catalysts, a clearly better result compared to the current industrial process leading to 70–85% selectivity at 4% conversion and based on the use of cobalt salt or metal-basic acid as catalysts. Although of great interest for petrochemical and natural gas conversion, the selective oxidation of other alkanes has been scarcely investigated.3 5. 5.1

Gold catalysis in material science Conducting polymers: polyaniline and polypyrrole

Since the ﬁrst preparation of the highly conducting polyacetylene (PA) in 1977, much eﬀort has been focused on the synthesis of other organic conducting polymers such as polyaniline (PANI) (Fig. 4), polypyrrole (PPy) (Fig. 5) and polythiophene (PTh) and their applications in devices combining optical, electrochemical and conducting properties, owing to their great versatility.73,74 PANI, in particular, is unique because of its tuneable conductivity being connected to the degree of acid-doping (pH) and oxidation state of the material. Equal numbers of oxidized and reduced units (emeraldine form), with one proton doping every two units, guarantee optimum conductivity of the polymer. Recently, great attention has been put on the morphology of PANI nanostructure, as nanoﬁbers and nanotubes. In general, PANI nanostructures are achieved by a ‘‘template synthesis’’ route using zeolite channels, track-etched polycarbonate and nanosized alumina membrane as templates, which address the growth of the nanostructures.75 Even though this method allows the perfect control of the length and diameter of the products, by template selection, in most applications the template must be removed, requiring additional workup and causing disorder or modiﬁcation of the micro/nanostructures. It has been demonstrated that the morphology and chemical properties of PANI are closely associated with the preparation method and many synthetic procedures have been experimented.76 Generally, aniline polymerization is performed through oxidative coupling of aniline or its dimer, N-(4-aminophenyl)aniline, using oxidants such as ammonium persulfate (APS), K2Cr2O7, KIO3. The most common oxidant used for the preparation of conducting polymers is APS, but its inorganic byproduct (ammonium sulfate) represents a limit for further applications.77 The use of other metals in a high oxidation state does not overcome this drawback.75–79 A recent application of chloroauric acid (HAuCl4), as an oxidant for the polymerization of aniline, resulted to be eﬀective for achieving nanoﬁbers, nanotubes,80 and nanoballs.73 The introduction of metal species can deeply inﬂuence the electronic and chemical properties of the polymer. Chattopadhyay and co-workers reported a new route for synthesizing Au–PANI composites based on the use of H2O2 as the bifunctional reagent for the reduction of HAuCl4 and the oxidation of aniline, leading to the formation of interesting PANI–gold composites.78 300 | Catalysis, 2010, 22, 279–317

Catalysis, 2010, 22, 279–317 | 301

NH

N

NH

NH

N

NH

NH

NH

N

NH

NH

NH

Pernigraniline

N

-

+

Cl

N

H

N

NH

N

Fig. 4 Oxidation states of PANI

N

Conducting Emeraldine

NH

Emeraldine

NH

Leucoemeraldine

NH

N

Cl

N

H

N

NH

-

+

N

H

Cl

NH

N

NH

-

+

NH

NH2

N

H

H

H N

N N

N H

n

H

Fig. 5 Polypyrrole

Metal particles can be introduced in the polymeric framework also in two separate steps.81 Mallick et al.82 have described further preparation methods and applications of gold–polyaniline composites. Nanoparticles of preformed metallic gold have not yet been used as a catalyst in the oxidative polymerization of aniline, but Bicak and Karagoz performed the synthesis of emeraldine base from aniline and gaseous oxygen with Cu(II) as the catalyst.83 The use for organic solvents and a soluble catalyst, however, makes this route far to be a real eco-friendly method, even though the achieved yields were interesting. Also the application of H2O2 for aniline polymerization appears to be attractive, mainly for large-scale applications, because its reduction product, H2O, improves recycling of the reagent. A limit, however, is the slow reaction rate which is prompting research to ﬁnd suitable catalytic systems for improving the kinetics:77 Sivakumar and Gedanken have successful applied ultrasonic irradiation for achieving conductive polyaniline84 and Au–polyaniline composites have been produced in the presence of chloroauric acid.78,84 The exciting performance of gold nanoparticles as the catalyst in the oxidative polymerization of pyrrole85 (Section 5.4), has suggested the investigation on this catalytic method for oxidising aniline. While gold was shown to be almost inert towards the aerobic oxidation of aniline, polyaniline could be synthesized using hydrogen peroxide as the oxidant under mild conditions.86 As pointed out before, the request of a simple and clean preparation method is required for particular technological applications, such as changeable conducting materials,87 electronic displays,88 electrode materials,89 molecular electronic circuit elements,90 restoration of data,91 detectors92 and biochemical analysis.93 In fact, the electrical conductivity of PPy, due to the electrons hopping along and across the polymer chains with conjugating bonds,94,95 is particularly sensitive to residues of reagents and organic solvents disturbing the co-planarity between interchains.96 Similar considerations must be taken in consideration for supporting new catalytic procedures for the synthesis of conducting polypyrrole, which presently can be prepared by chemical,97 electrochemical,98 plasma,99,100 vapour phase101,102 and enzymatic routes.103 5.2

Oxidative polymerization procedure of aniline

Polyaniline can be easily prepared by aniline oxidation in aqueous medium. Various supramolecular structures of the ﬁnal product are obtained, depending on the conditions of the reaction, but the mechanism of their formation has not yet been elucidated. When aniline is oxidized in an acidic aqueous medium with ammonium peroxydisulfate, a PANI precipitate is produced. The blue pernigraniline form, present during the polymerization, converts into the green protonated emeraldine at the end of the 302 | Catalysis, 2010, 22, 279–317

polymerization. The reaction is exothermic and leads, besides PANI formation, to sulfuric acid as a by-product. The progress of polymerization can be followed in situ by recording either the temperature or the pH. It has been recently reported104 that the mechanism of aniline oxidation with ammonium peroxydisulfate in aqueous solution of strong (sulfuric) or weak (acetic) acids is substantially diﬀerent. In sulfuric acid solution a granular PANI was produced; in acetic acid solution, on the contrary, PANI nanotubes were obtained. It has been demonstrated that aniline polymerization proceeds well even in water, without any added acid, when ammonium peroxydisulfate was used as an oxidant. The sulfuric acid produced by the decomposition of peroxydisulfate, in fact, gradually provides the necessary acidity and the ﬁnal PANI is protonated with this acid. Nevertheless, the conductivity of PANI prepared with this simple method is rather low. The use of other oxidants, such as oxygen and hydrogen peroxide, in the presence of gold-based catalysts has led to new exciting results. 5.3

Polymerization of aniline with gold catalysts

Aniline, dissolved in aqueous HCl, is inert towards oxidation with H2O2; on adding a gold catalyst (Aniline: Au=100–1000) the insoluble green polymer ‘‘emeraldine’’ is formed. Puriﬁcation of the product can be achieved by extraction with 1-methyl-2-pirrolidone and evaporation of the solvent under vacuum. Among many catalytic systems, as colloidal gold, gold supported on carbon and gold supported on TiO2, this latter produced the highest yield86 The aniline/emeraldine emichloridrate redox potential is quite high (ca. 1.46 V),105 thus suggesting a thermodynamic barrier to the aerobic oxidation of aniline (E1 O2/H2O=1.23 V). As H2O2 presents a higher value of E1=1.78 V, one could expect this oxidant to be eﬀective in aﬀording the conductive polymer. Actually, while no aerobic oxidative polymerisation of aniline was observed using diﬀerent catalytic systems with O2 at 3 bar and room temperature, a modest catalytic eﬀect (typically 4–5% yield) was detected when a small amount of colloidal gold (Au:Aniline=0.001) was used as a catalyst in combination with the H2O2 reagent. No product was isolated without any catalyst, and no beneﬁt was recorded from using H2O2 in excess (Table 10). However, the PANI yield could be increased by increasing the gold amount in the range Au:Aniline=0.001–0.004 (molar ratio) reaching the asymptotic value of 27% in 24h-tests (Fig. 6). The humble catalytic life of gold ‘‘naked’’ particles in oxidation reactions could be the cause for the limited conversion of aniline to PANI. The

Table 10 Oxidative polymerization of aniline in the presence colloidal gold and diﬀerent H2O2 amounts Test

Au:Aniline

H2O2:Aniline

Yield%

1 2 3 4

0 0.001 0.001 0.001

1 1 2 4

0 4.8 4.8 4.0

Catalysis, 2010, 22, 279–317 | 303

30

Yield %

25 20 15 10 5 0 0

0.005

0.01

0.015

0.02

Au/Aniline Fig. 6

Dependence of aniline polymerization by the amount of colloidal gold.

oxidative polymerisation of aniline was consequently experimented also in the presence of supported gold catalyst (0.5% Au/C, 1% Au/TiO2), which was proven to be more stable in former catalytic applications.17 The achievements, reported in Table 11, show improved performances. Table 11 Polymerization of aniline by supported gold catalysts. (Aniline:Au=1000) Test

Catalyst

Aniline/H2O2

Yield%

1 2

0.5% Au/C 1% Au/TiO2

1 1

11.4 70.1

In particular, the high activity of Au/TiO2 can be ascribed to a strong contribution of the TiO2 support. Diﬀerently from unloaded carbon, which was inert in aniline polymerisation by H2O2, P25 titania catalyzed the partial oxidation of aniline to soluble dark oligomers. However, no solid material was formed. The products obtained in all of the preparations were identiﬁed as ES (emeraldine salt) according to the FT-IR.81 UV–vis,74 and XRD spectra75 and84 reported in Figs. 7–9. The morphology of the products, investigated by transmission electron microscopy (TEM) and scanning electron microscopy (SEM), revealed emeraldine in form of nanospheres of 44–160 nm (Fig. 10A) alternating with micrometric rods (Fig. 10B and Fig. 11). Concerning the morphological properties, similar nanospheres were obtained for the Au/C and TiO2 catalysts. In the case of Au/TiO2, however, the nanospheres were assembled in a cluster-like organization, which was absent in the product derived using the carbon-supported catalyst. The conductivity of the polymer obtained in the high-yield conversion of aniline with Au/TiO2 catalyst (Table 11, test 2) reached the value of 1.5 10 1 S/cm and was determined with a standard conductivity cell (CON–H Material Mates). In conclusion, it has been shown that conductive PANI, mainly in form of nanospheres, can be easily fabricated from aniline by H2O2 oxidation in the presence of gold nanoparticles as a catalyst. The conductivity values, correlated with the bulk resistance of this polymer, are similar to those obtained through other polymerisation methods.84,106 304 | Catalysis, 2010, 22, 279–317

45 40 35

%T

30 25 20 15 10 4000

3500

3000

2500

2000

1500

1000

500

Wavenumber (cm-1) Fig. 7 FT-IR spectrum of emeraldine salt in KBr pellet.

A

0.5

0.0 400

600

800

Wavelength (nm) Fig. 8 UV-vis spectrum of emeraldine salt in NMP as the solvent.

5.4

PANI-based composites

Conducting polymers have been proven to be suitable host matrices for dispersing metallic particles. Conducting polymer composites with metal nanoparticles allow a facile ﬂow of electronic charges through the polymer matrix during electrochemical processes. Through a suitable combination of conducting polymer and metal nanoparticles, novel promising electrodes Catalysis, 2010, 22, 279–317 | 305

700

600

Lin (Counts)

500

400

300 b

200

a

100

20

40

60

80

2-Theta-Scale Fig. 9 X-ray diﬀraction pattern of emeraldine salt.

Fig. 10 TEM images of PANI synthesized with (A) (Bar=200 nm), H2O2 and (B) (Bar=1 mm), ‘‘naked’’ gold nanoparticles.

could be generated with higher surface areas and enhanced electrocatalytic activities, particularly interesting in the fuel cell technology. 5.4.1 Synthesis of PANI/MCM-41. PANI is insoluble in common solvents and impossible to be molten. These drawbacks can be overcome by modifying the polymer with the incorporation of inorganic materials. As PANI/inorganic nanocomposites combine the advantages of PANI and inorganic nanoparticles, extensive research has been carried out in this ﬁeld. Silica nanoparticles, in particular, have received great attention because of their unique properties and wide applications. Primarily, the main objective is to keep the conducting polymer in a stable colloidal form. 306 | Catalysis, 2010, 22, 279–317

Fig. 11 SEM images of PANI synthesized with H2O2 and ‘‘naked’’ gold nanoparticles (Bar=500 nm)

Armes et al.107–110 succeeded in incorporating silica nanoparticles into the core of PANI. Aniline was polymerized by ammonium peroxydisulfate in the presence of silica colloids at low concentration of monomer and oxidant. This technique allows to slow down the rate and degree of polymerization and promote the polymerization on the colloidal surface rather than into the bulk. Eﬀectively, the outer PANI layer of PANI/SiO2 colloids becomes soluble to some extent. Porous silica possesses high surface area and the size of pores can be tuned from 2 to 10 nm in a narrow diameter distribution by changing the experimental conditions. PANI synthesis inside porous silica channels has attracted more attention. Wu and Bein111 fabricated a conductive ﬁlament of PANI in MCM-41, by using MCM-41 as the host which was contacted with aniline gas (not beneﬁcial to the environment) at 313 K for 24 h, thus obtaining signiﬁcant conductivity. PANI/MCM-41 composite was synthesized by chemical oxidative polymerization of aniline on the surface of MCM-41 in the presence of HCl. It was shown that PANI was present not only on the surface of MCM-41 but also inside the pores. Moreover, the higher HCl concentration, the higher conductivity of the composite, because of a wider delocalization of the resulting emeraldine salts. 5.4.2 Preparation of PANI nanoparticles by Fe3O4. Colloidal particles with magnetic dipole moments are able to self-assemble into ﬂexible chains. In particular, they can form chains with threefold junctions resulting from a delicate balance between the dipolar interactions. These chains can be thought as a ‘‘living polymer’’ whose length and organization are determined not by the reaction conditions, but by a thermodynamic balance of forces. Water ‘‘soluble’’ Fe3O4 nanoparticles, with a coating of polyethylene glycol nonylphenyl ether (NP5) and cyclodextrin (CD), were synthesized and used as templates for the preparation of PANI Catalysis, 2010, 22, 279–317 | 307

nanostructures.112 NP5 was used to stabilize the magnetic nanoparticles by formation of a suitable surface coating and CD for improving their miscibility with water. Y-junction PANI nanorods and nanotubes have been synthesized by the use of in situ self-assembled magnetic nanoparticles as templates and pH control of the reaction system. It has been found that an initial pH of the reaction system around 8–10 favours the formation of nanorods, whereas a starting pH range of 5–6 leads to the formation of nanotubes. The morphology of Y-junction PANI nanostructures depends on the reaction conditions such as on aniline concentration and the presence of an organic solvent. 5.4.3 Preparation of PANI nanoparticles by Pt. The incorporation of Pt alone or Pt with secondary metals including Ru, Os, Sn and Mo into PANI has been primarily focused on exploiting the catalytic activity of the metals.113 For example, polymer supported Pt has been used in the catalytic oxidation of methanol, formic acid and hydrazine, in addition to the reduction of dioxygen. The reaction of methanol at Pt surfaces has been the focus of numerous studies. Laborde et al.114 have studied the oxidation of methanol with a platinum modiﬁed polyaniline electrodes in acidic medium. The results indicate that methanol oxidation involves the direct formation of CO2 through reactive intermediates. The synthesis of polyaniline/platinum composites has been achieved using controlled electrochemical reduction of PtCl26 and PtCl24 . The electrochemical and chemical syntheses of PANI/Pt by reduction of PtCl26 and PtCl24 produce morphologically diﬀerent composite materials. 5.4.4 Preparation of PANI nanoparticles by CuSO4. A simple, one pot and in situ chemical synthesis route for the preparation of polyaniline uses cupric sulfate as an oxidizing agent. The reaction has been conducted in methanol and the cupric sulfate has been added drop by drop. The size of the copper nanoparticles varied in the range 2–5 nm. The formation mechanism of micro or nanostructured polymer is not yet clear. In the ﬁrst step, we can assume that an oligomeric form of aniline can be formed due to the presence of a suitable oxidizing agent. The oligomerized aniline thus produced may act as a nucleation centre, which catalyses the oxidation of the remaining monomers present in the solution. The oxidizing agent gets reduced and forms copper nanoparticles that lead to the formation of a metal-polymer composite material, which could represent an important advantage for further applications. This method involving copper has been extended to gold and palladium systems.74 5.4.5 Preparation of PANI-gold composites. A simple route for polyaniline-gold composites synthesis uses tetrachloroauric acid (HAuCl4) as the oxidizing agent, forming metallic gold nanoparticles (10–50 nm) at the end of oxidation. The reaction is conducted in toluene with a phase-transfer catalyst. Other works report the same reaction conducted in 1M HCl aqueous73 or in a solution of D-camphor-10-sulfonic acid.115 A new method of synthesis of an Au nanoparticle-conducting polyaniline composite has been proposed using H2O2 both for the reduction of HAuCl4 and polymerization of aniline in the same aqueous medium.78 308 | Catalysis, 2010, 22, 279–317

5.5

Oxidative polymerization of pyrrole by gold

The polymerization of pyrrole can be fairly performed by aerobic oxidation by using gold catalysis.85 Stirring pyrrole (Py) in acidic (HCl) aqueous solution in air at room temperature (292–295 K) and in the presence of colloidal gold a slow oxidation takes place producing a dark product. The observed yields, up to 75%, are depending on the total gold amount in the range Py:Au=1000–10000, as reported in Table 12.

Table 12 Aerobic polymerization of pyrrole with air. Reaction time 3 days Test

Py/Au (molar ratio)

Yield%

1 2 3 4

No gold 10000 5000 1000

0 42 50 75

Table 13 Catalytic polymerization of pyrrole with O2 at 0.3 MPa and diﬀerent reaction times Test

t (h)

Py/Au (molar ratio)

Yield%

5 6 7 8 9 10 11 12 13 14 15 16 17 18

8 8 8 16 16 16 24 24 24 24 72 72 72 72

No gold 10000 1000 No gold 10000 1000 No gold 10000 5000 1000 No gold 10000 5000 1000

0 0 1 5 31 34 10 50 60 82 12 61 72 99

An acceleration of the reaction can be obtained by carrying out the reaction under pure oxygen as reported in Table 13 thus allowing a higher yield (82%) of polypyrrole in less time (24 h). A great improvement of pyrrole polymerisation has been observed by using H2O2 instead of O2 as the oxidant. In this case we must outline that pyrrole plus HCl, dissolved in water under nitrogen atmosphere, undergoes a slow oxidative polymerisation also in the absence of catalyst, producing polypyrrole in 57% yield in 24 h (Table 14). The addition of gold improves the polymerisation which can be implemented to 99% of the product. Fig. 12 visualizes the catalytic eﬀect of gold in polypyrrole formation with various oxidative agents. Fig. 13 reports the kinetics of polymeric material Catalysis, 2010, 22, 279–317 | 309

Table 14 Catalytic polymerization of pyrrole using H2O2 as the oxidant. Reaction time 24 h Test

Py/Au (molar ratio)

Yield%

19 20 21 22

No gold 10000 5000 1000

57 66 90 99

O2, 24h

120

H2O2, 24h 100

O2, 3 days Air, 3days

Yield %

80 60 40 20 0 0

0.0002

0.0004

0.0006

0.0008

0.001

0.0012

Au/PY Fig. 12 Gold eﬀect in the polymerization of pyrrole by air, O2 and H2O2.

formation, during pyrrole polymerisation by dioxygen in the presence of diﬀerent quantities of gold. Using pure oxygen gas, a slow auto-oxidation produced a modest yield of insoluble polymer (12%) after 3 days. On adding colloidal gold, a strong catalytic eﬀect was detected, as the polymeric material was formed with almost total yield (99%) after 3 days (Fig. 13). Using hydrogen peroxide as the oxidant, a consistent polymerization of pyrrole was observed also in the absence of catalyst. In fact, the reaction occurred with 57% yield with respect to pyrrole in 24 h. However, also in this case, a catalytic contribute of gold was demonstrated because the yield increased up to 99% by adding a small quantity of colloidal metal (Fig. 12). A concert of diﬀerent analytical techniques allowed the characterization of the obtained polypyrrole structure.116–119 Independently from the synthetic method, all the prepared polymers show a similar IR spectrum which is represented in Fig. 14. Fig. 15 shows typical X-ray diﬀraction patterns of PPy synthesized in the presence and in the absence of gold nanoparticles.120 The TEM images of the various polymeric materials are reported in Figs. 16 A–E. The comparison among the diﬀerent products highlights how much the morphology is aﬀected by the nature of the oxidising reagent. By using H2O2, the quick polymerization led to a partially reticulated structure 310 | Catalysis, 2010, 22, 279–317

100

no gold Py/Au=10000 Py/Au=1000

Yield %

80

60

40

20

0 0

8

16

24

32

40

48

56

64

72

Reaction time (h) Fig. 13 Kinetic data of pyrrole polymerization under O2.

40

%T

30

20

10 3200

2400

1600

800

Wavenumber (cm-1) Fig. 14 FT-IR spectrum of a PPy sample in KBr.

without gold (Fig. 16A) and to amorphous material using gold catalysis (Fig. 16B). More ordered structures were achieved by the non-catalyzed polymerization with gaseous oxygen (Figs. 16C, D). Abundant peculiar thin squares were observed in the high yield polymerization catalysed by gold (Fig. 16E). This latter structure is unusual also in the context of conventional polymerizations of pyrrole and could be interesting for tailor-made composite applications. The conductivity of the synthesized polypyrrole materials is similar to that obtained in conventional chemical polymerization using stoichiometric Catalysis, 2010, 22, 279–317 | 311

700

600

Lin (Counts)

500

400

300 b

200

a

100

20

40

60

80

2-Theta-Scale Fig. 15 X-ray diﬀraction patterns of PPy synthesized in the absence (a) and in the presence (b) of gold nanoparticles

200 nm

A) By H2O2, no Au

200 nm

C) By air,no Au.

100 nm

B) By H2O2, Py/Au=1000.

100 nm

D) By O2,no Au

200 nm

E) By O2,Py/Au=5000

Fig. 16 TEM images of diﬀerent polypyrrole samples.

reagents and it was determined with the standard conductivity cell (CON–H Material Mates) also used for polyaniline (Section 5.2).121,122 The s values, determined for various preparations and measured in samples of diﬀerent thickness under diﬀerent voltages (Table 15), ranged from 2.7 10 4 to 312 | Catalysis, 2010, 22, 279–317

Table 15 Conductivity data of polypyrrole samples Sample (thickness [mm])

0V (S c 1)

0.5V (S cm 1)

1V (S cm 1)

2 (0.69) 3 (1.10) 4 (0.63) 18 (0.64) 19 (0.60) 20 (0.67) 21 (0.64) 22 (0.68)

2.46 10 3 5.09 10 3 8.94 10 4 8.47 10 4 1.33 10 3 4.81 10 3 9.00 10 4 2.74 10 4

5.88 10 4 1.38 10 3 1.39 10 3 9.97 10 4 6.17 10 4 2.70 10 3 6.04 10 4 5.53 10 4

4.80 10 4 3.84 10 3 4.09 10 3 7.23 10 4 3.34 10 4 4.85 10 3 4.56 10 4 4.97 10 4

5 10 3 S cm 1. These conductivities are typical for polymers obtained by using FeCl3, AgNO3, Cu(NO3)2–AlCl3,123 VO(acac)2–AlCl3–O2,124 AlCl3–CuCl–O2125 but much lower than the values generally found in electrochemical polypyrrole ﬁlm deposition.117 A synthetic comparison of s values is reported in Table 16, showing how the wide range in the conductivities (from 10 10 to 102 S cm 1) of the diﬀerently synthesized polymers allows a wide choice of application of these materials, according to the required conductivity. Table 16 Comparison among diﬀerent conductivity data

6.

Research groups

Conductivity (S cm 1)

Rossi et al.85 Chao and March116 Toshima and Tayanagis125 Izumi and Toshima124 Diaz and Fanazawa117

2.7 10 4 5.1 10 3 5.0 10 3 2.8 10 2 10 10 2.1 10 2 10–100

Conclusions

The last twenty years represent a milestone in the catalytic selective oxidation of organic compounds under eco-friendly conditions and this state of ferment goes on. In this context, a key role has been played by gold, whose catalytic properties are rapidly ﬂowing into applications rich in promising results. Starting from liquid phase processes carried out under mild conditions, gold catalysis is fast extending to gas phase processes at higher temperature with comparable perspectives. The focus of the extraordinary peculiarities of gold in catalysis is often represented by its highly dispersed nanometric particles. A novel scenario is now emerging: the application of gold in material science, particularly promising in the facile polymerization of aniline and pyrrole via environmentally friendly routes. Even though methods and techniques have still to be reﬁned, the progress in oxidation reactions by gold and its successful versatility will contribute to the deﬁnitive decline of the so-called ‘‘stoichiometric oxidants’’, so far used in organic synthesis but producing undesired wastes. Catalysis appears to be the powerful tool for making environmentally friendly processes a wide reality. Catalysis, 2010, 22, 279–317 | 313

References 1 G. C. Bond, C. Louis and D. T. Thompson, Catalysis by Gold, Imperial College Press, Singapore, 2006. 2 A. S. K. Hashmi and G. J. Hutchings, Angew. Chem. Int. Ed., 2006, 45, 7896. 3 (a) A. S. K. Hashmi, Chem. Rev., 2007, 107, 3180; (b) C. Della Pina, E. Falletta and L. Prati, M. Rossi Chem. Soc. Rev., 2008, 37, 2077. 4 M. C. Daniel and D. Astruc, Chem. Rev., 2004, 104, 293. 5 M. Haruta and M. Date, Appl. Catal. A, 2001, 222, 427. 6 R. A. Sheldon and J. K. Kochi, Metal-Catalyzed Oxidations of Organic Compounds, Academic Press, New York, 1981. 7 M. Hudlicky, Oxidations in Organic Chemistry, ACS, Washington DC, 1990. 8 U. R. Pillai and E. Sahle-Demessie, Appl. Catal. A, 2003, 245, 103. 9 IP 99A002611 to Lonza, 1999. 10 WO patent WO2004/099114 to Su¨dzucker, 2004. 11 EPA patent 1 393 800 to Nippon Shokubai Co., 2004. 12 WO 2005/003072 to University of Milano, 2004. 13 L. Prati and M. Rossi, J. Catal., 1998, 176, 552. 14 S. Biella and M. Rossi, Chem. Commun., 2003, 378. 15 C. Milone, R. Ingoglia, M. Tropeano, G. Neri and S. Galvagno, Chem. Commun., 2003, 1359. 16 G. C. Bond and D. T. Thompson, Catal. Rev. Sci. and Eng., 1999, 41, 39. 17 M. Comotti, C. Della Pina, R. Matarrese and M. Rossi, Angew. Chem. Int. Ed., 2004, 43, 5812. 18 R. A. Sheldon Catalytic Oxidation (Sheldon, R. A. and Van Santen R. Eds., W.S.P., Singapore, 1995). 19 T. Mallat and A. Baiker, Chem. Rev., 2004, 104, 3037. 20 D. J. Gorin and F. D. Toste, Nature, 2007, 446, 395. 21 S. D. Pollington, D. I Enache, P. Landon, S. Meenakshisundaram. N. Dimitratos, A. Wagland, G. J. Hutchings and E. H. Stitt, Catal. Today, 2009, 145, 69. 22 N. Wang, T. Matsumoto, M. Ueno and H. Miyamura, S. Kobayashi Angew. Chem., 2009, 121, 4838. 23 M. Lanfranchi, L. Prati, M. Rossi and A. Tiripicchio, Chem. Commun., 1993, 1698. 24 L. Prati and M. Rossi, Stud. Surf. Sci. Catal., 1997, 110, 509. 25 S. Biella, L. Prati and M. Rossi, Inorg. Chim. Acta, 2003, 349, 253. 26 S. Biella, G. Castiglioni, C. Fumagalli, L. Prati and M. Rossi, Catal. Today, 2002, 72, 42. 27 M. Pagliaro and M. Rossi, The Future of Glycerol: New Usages for a Versatile Raw Material, RSC Publishing, Cambridge, UK, 2008. 28 M. Pagliaro, R. Ciriminna, H. Kimura, M. Rossi and C. Della Pina, Angew. Chem, Int. Ed., 2007, 46, 4434. 29 S. Carrettin, P. McMorn, P. Johnston, K. Griﬃn and G. Hutchings, Chem. Commun., 2002, 696. 30 F. Porta and L. Prati, J. Catal., 2004, 224, 397. 31 C. Bianchi, P. Canton, N. Dimitratos, F. Porta and L. Prati, Catal. Today, 2005, 102, 203. 32 N. Dimitratos, F. Porta, L. Prati and A. Villa, Catal. Lett., 2005, 99, 181. 33 T. Werpy, J. Holladay, J. White. 2004. Top Value Added Chemicals From Biomass. PNNL-14808, Pac. Northw. Nat. Lab., Richland, WA. http:// environment.pnl.gov/staﬀ/staﬀ_info 314 | Catalysis, 2010, 22, 279–317

34 A. Behr, A. Botulinski, F.-J. Carduck, M. Schneider, US5321156, 1994. 35 D. Arnzt, T. Haas, A. Muller, N. Wiegand, US Pat. 5276201, 1994. 36 M. Emptage, S. L. Haynie, L. A. Laﬀeend, J. P. Pucci, G. M. Whited, US Pat. 7067300, 2006. 37 Shell Chemicals Magazine, n1 10, 2nd Quarter, 1998. 38 D. Arntz et al., EP412337, 1990 to Degussa AG. 39 H. Ishida, E. Ueno, JP2000159724, 2000. 40 F. Suthers, D. C. Cameron, WO0116346, 2001. 41 C. Della Pina, E. Falletta and M. Rossi, Chem Sus Chem, 2009, 2, 57. 42 K. Weissermel and H. J. Arpe, Industrial Organic Chemistry, VCH Weinheim,1997. 43 http://www.gold.org/ 44 D. Enache, J. Edwards, P. Landon, B. Solsona-Espriu, A. Carley, A. Herzing, M. Watanabe, C. Kiely, D. Knight and G. Hutchings, Science, 2006, 311, 362. 45 A. Abad, P. Conception, A. Corma and H. Garcia, Angew. Chem. Int. Ed., 2005, 44, 4066. 46 M. Lazar and R. J. Angelici, J. Am. Chem. Soc., 2006, 128, 10613. 47 B. Zhu and R. J. Angelici, J. Am. Chem. Soc., 2006, 128, 14460. 48 B. Zhu and R. J. Angelici, Chem. Commun., 2007, 20, 2157. 49 C. Della Pina, E. Falletta and M. Rossi, Top. Catal., 2007, 44, 325. 50 M. Besson, F. Lahmer, P. Gallezot, P. Fuertes and G. Fleche, J. Catal., 1995, 152, 116. 51 M. Besson and P. Gallezot, Catal. Today, 2000, 57, 127. 52 S. Biella, L. Prati and M. Rossi, J. Catal., 2002, 206, 242. 53 M. Comotti, C. Della Pina, E. Falletta and M. Rossi, J. Catal., 2006, 244, 122. + 54 Y. Onal, S. Schimpf and P. Claus, J. Catal, 2004, 223, 122. 55 P. Beltrame, M. Comotti, C. Della Pina and M. Rossi, Appl. Catal. A, 2006, 297, 1. 56 P. Beltrame, M. Comotti, C. Della Pina and M. Rossi, J. Catal., 2004, 228, 282. 57 M. Comotti, C. Della Pina, E. Falletta and M. Rossi, Adv. Synth. Catal., 2006, 348, 313. 58 C. Della Pina, E. Falletta and M. Rossi, A. Sacco J. Catal., 2009, 263, 92. 59 M. Comotti, C. Della Pina and M. Rossi, J Molec. Cat.A, 2006, 251, 92. 60 H. Haruta, Catal. Today, 1997, 36, 153. 61 B. Taylor, J. Lauterbach and W. N. Delgass, Appl. Cat. A, 2005, 291, 188. 62 L. Cumaranatunge and W. N. Delgass, J. Catal., 2005, 232, 38. 63 (a) B. Herzog, M. Mark, R. Stein, T. Wang, (Celanese Chemicals Europe Gmbh, Germany; Celanese International Corporation) Ger. Oﬀen. DE10030039 A1 2002; (b) T. Wang, L. E. Wade, V. Wong, J. Han, A. Hagemeyer, D. Lowe, V. Sokolovskii (Celanese International Corporation, USA) PCT Int. Appl. WO2005065819 A1 2005. 64 A. K. Khanmamedova (Saudi Basic Industries Corporation, Saudi Arabia) PCT Int. Appl. WO2002004392 A1 2002. 65 (a) M. D. Hughes, Y.-J. Xu, P. Jenkins, P. McMorn, P. Landon, D. I. Enache, A. F. Carley, G. A. Attard, G. J. Hutchings, F. King, E. H. Stitt, P. Johnston, K. Griﬃn and C. J. Kiely, Nature, 2005, 437, 1132; (b) Y.-J. Xu, P. Landon, D. I. Enache, A. F. Carley, M. W. Roberts and G. J. Hutchings, Catal. Lett., 2005, 101, 175. 66 J. Edwards, P. Landon, A. F. Carley, A. A. Herzing, M. Watanabe, C. J. Kiely and G. J. Hutchings, J. Mat. Research, 2007, 22, 831. Catalysis, 2010, 22, 279–317 | 315

67 (a) N. S. Patil, R. Jha, B. S. Uphade, S. K. Bhargava and V. R. Choudhary, Appl. Catal. A, 2004, 275, 87; (b) N. S. Patil, B. S. Uphade, D. G. McCulloh, S. K. Bhargava and V. R. Choudhary, Catal. Commun., 2004, 5, 681; (c) N. S. Patil, R. Jha, B. S. Uphade, S. K. Bhargava and V. R. Choudhary, Appl. Catal. A, 2004, 275, 87. 68 D.-H. Ying, L.-S. Qin, J.-F. Liu, C.-Y. Li and Y. Lin, J. Mol. Catal. A, 2005, 240, 40. 69 P. Lignier, F. Morﬁn, S. Mangematin, L. Massin, J. Rousset and V. Caps, Chem. Commun., 2007, 20, 186. 70 (a) R. Zhao, D. Ji, G. Lu, G. Qian, L. Yan, X. Wang and J. Suo, Chem. Commun., 2004, 904; (b) G. Lue, R. Zhao, G. Qian, Y. Qi, X. Wang and J. Suo, Catal. Lett., 2004, 97, 115. 71 K. K. Zhu, J. C. Hu and R. Richards, Catal. Lett., 2005, 100, 195. 72 L.-X. Xu, C.-H. He, M.-Q. Zhu, K.-J. Wu and Y.-L. Lai, Catal. Lett., 2007, 118, 248. 73 K. Mallick, M. J. Witcomb, A. Dinsmore and M. S. Scurrell, Macromol. Rapid Commun., 2005, 26, 232. 74 K. Mallick, M. J. Witcomb, A. Dinsmore and M. S. Scurrell, J. Polym. Res., 2006, 13, 397. 75 K. Mallick, M. J. Witcomb, A. Dinsmore and M. S. Scurrell, Langmuir, 2005, 21, 7964. 76 Y. Wang, Z. Liu, B. Han, Z. Sun, Y. Huang and G. Yang, Langmuir, 2005, 21, 833. 77 Z. Sun, Y. Geng, J. Li, X. Wang, X. Jing and F. Wang, J. Appl. Polym. Sci., 1999, 72, 1077. 78 T. K. Sarma, D. Chowdhury, A. Paul and A. Chattopadhyay, Chem. Commun., 2002, 1048. 79 D. K. Moon, K. Osokada, T. Maruyama and T. Yamamoto, Macromol. Chem., 1992, 193, 1723. 80 X. Feng, G. Yang, Q. Xu, W. Hou and J.-J. Zhu, Macromol. Rapid Commun., 2006, 27, 31. 81 X. Feng, C. Mao, G. Yang, W. Hou and J.-J. Zhu, Langmuir, 2006, 22, 4384. 82 K. Mallick, M. J. Witcomb and M. S. Scurrell, Gold Bull., 2006, 39/4, 166. 83 N. Bicak and B. Karagoz, J. Polym. Sci. A, 2006, 44, 6025. 84 M. Sivakumar and A. Gedanken, Synth. Met., 2005, 148, 301. 85 C. Della Pina, E. Falletta, M. Lo Faro and M. Pasta, M. Rossi Gold Bull., 2009, 42/1, 27. 86 Z. Chen, C. Della Pina, E. Falletta, M. Lo Faro, M. Pasta and M. Rossi, N. Santo J. Catal., 2008, 259, 1. 87 X. Lu et al., Materials letters, 2006, 60, 2851. 88 H. Yoneyama, K. Wakamoto and H. J. Tamure´, J. Electrochem. Soc., 1985, 132, 2414. 89 K. Kaneto, M. Maxﬁeld, D. P. Nairns, A. G. MacDiarmid and A. J. Heeger, J. Chem. Soc. Faraday Trans, 1982, 78, 3417. 90 J. W. Thackeray, H. S. White and M. S. Wrighton, J. Phys. Chem., 1985, 89, 5133. 91 S. Kuwabata, S. Ito and H. Yoneyama, J. Electrochem. Sci. Technol., 1998, 135, 1691. 92 P. N. Bartlett and S. K. Ling Chung, Sens Actuators, 1989, 20, 287. 93 D. T. Hao, T. N. Suresh Kumar, R. S. Srinivasa, R. Lal and N. S. Punekar, Contractor AQ. Anal. Chem., 1992, 64, 2645. 94 J. He, D. E. Tallmann and G. D. Bierwager, J. Electrochem. Soc., 2004, 151, B644. 316 | Catalysis, 2010, 22, 279–317

95 A. J. F. Romero, J. J. L. Cascales and T. F. Otero, J. Phys. Chem. B, 2005, 109, 907. 96 Y.-C. Liu and K.-H. Yang, Electrochimica Acta, 2006, 51, 5376. 97 N. Toshima, Makromol. Chem., Macromol. Symp., 1992, 59, 123. 98 N. C. Billingham and P. D. Calvert, Adv. Polym. Sci., 1989, 90, 1. 99 G. J. Cruz, J. Morales and R. Olayo, Thin Solid ﬁlms, 1999, 324, 119. 100 K. Hosono, I. Matsubara, N. Murayama, W. Shin and N. Izu, Mater. Lett., 2004, 58, 1371. 101 S. N. Tan and H. Ge, Polymer, 1996, 37, 965. 102 J. Kim, D. Sohn, Y. Sung and E. R. Kim, Synth. Met., 2003, 132, 309. 103 H. K. Song and G. T. R. Palmore, J. Phys. Chem. B, 2005, 109, 19278. 104 E. N. Konyushenko, J. Stejskal, I. Sedenkova, M. Trchova, I. Sapurina and M. Cieslar, J. Prokes Polym. Int., 2006, 55, 31. 105 S. K. Manohar and A. G. MacDiarmid, Synth. Met., 1991, 41, 711. 106 W. Li and H.-L. Wang, J. Am. Chem. Soc., 2004, 126, 2278. 107 M. Gill, S. P. Armes, D. Fairhurst, S. Emmett, T. Pigott and G. Idzorek, Langmuir, 1992, 8, 2178. 108 M. Gill, F. L. Baines and S. P. Armes, Synth. Met., 1993, 55–57, 1029. 109 S. Maede and S. P. Armes, J. Colloid. Interf. Sci., 1993, 159, 257. 110 S. Maede and S. P. Armes, J. Mater. Chem., 1994, 14, 935. 111 C. G. Wu and T. Bein, Science, 1994, 264, 1757. 112 H. Xia, D. Cheng, C. Xiao and H. S. O. Chan, J. Mat. Chem., 2005, 15, 4161. 113 D. W. Hatchett, R. Wijeratne and J. M. Kinyanjur, J. Electroanal. Chem., 2006, 593, 203. 114 H. Laborde, J.-M. Leger and C. Lamy, J. Appl. Electrochem., 1994, 24, 219. 115 H. Huang, Y. Zhang, Y. Long, J. Yuan, D. Han, Z. Wang, L. Niu and Z. Chen, Chem. Eur. J., 2006, 12, 5314. 116 T. H. Chao and J. March, J. Polym. Sci., Part A: Polym. Chem., 1988, 26, 743. 117 A. F. Diaz, K. K. Kanazawa and G. P. Gardini, J.C.S. Chem. Comm., 1979, 635. 118 S. T. Selvan, J. P. Spatz, H. A. Klok and M. Mo¨ller, Adv. Mater., 1998, 10, 132. 119 G. R. Mitchell, Polym. Commun., 1986, 27, 346. 120 T. Hanawa, S. Kuwabata and H. Yoneyama, J. Chem. Soc. Faraday Trans 1, 1988, 84, 1587. 121 F. Berthier, J.-P. Diard and R. Michel, Journal of Electroanalytical Chemistry, 2001, 510, 1. 122 G. Meszaros, I. Szenes and B. Lengyel, Electrochemistry Communications, 2004, 6, 1185. 123 T. H. Chao, J. March J. Polym. Sci., 1988, 26, 743. 124 S. Izumi, N. Toshima Bull. Chem. Soc. Jpn., 1994, 67, 539. 125 N. Toshima and J. Tayanagi, Chem. Lett., 1990, 1369.

Catalysis, 2010, 22, 279–317 | 317

Catalysis, Volume 22 (RSC Specialist Periodical Reports)

Catalysis: v. 18 (Specialist Periodical Reports)

Catalysis: Vol. 19 (Specialist Periodical Reports)

Catalysis: Vol 17 (Specialist Periodical Reports)

Photochemistry, Volume 37 (Specialist Periodical Reports)

Organometallic Chemistry, Volume 36 (Specialist Periodical Reports)

Organometallic Chemistry, Volume 31 (Specialist Periodical Reports)

Alkaloids Volume 4: Specialist Periodical Rep (Specialist Periodical Reports) (v. 4)

Organosphophorus Chemistry (Specialist Periodical Reports) (v. 4)

Heterocyclic Chemistry: v.4 (Specialist Periodical Reports)

Photochemistry (Specialist Periodical Reports) (Vol 13)

Nuclear Magnetic Resonance: Volume 37 (Specialist Periodical Reports)

Terpenoids and Steroids: Volume 12 (Specialist Periodical Reports)

Terpenoids and Steroids: Volume 3 (Specialist Periodical Reports)

Dielectric and Related Molecular Processes: Volume 1 (Specialist Periodical Reports)

Organophosphorus Chemistry (Volume 18) Specialist Periodical Report

Chemical Modelling: Vol. 2: Applications and Theory (Specialist Periodical Reports)

Heterocyclic Chemistry: v.3 (Specialist Periodical Reports) (Vol 3)

Theoretical Chemistry,vol 2 (Specialist Periodical Reports) (v. 2)

Catalysis: A Review of Chemical Literature: Vol 15 (Specialist Periodical Reports)

Catalysis: A Review of Chemical Literature: Vol 16 (Specialist Periodical Reports)

Inorganic Chemistry of the Transition Elements, Volume 2 (Specialist Periodical Reports) (v. 2)

Chiral Sulfur Ligands: Asymmetric Catalysis (RSC Catalysis Series)

Organophosphorus Chemistry - Volume 22 (SPR Organophosphorus Chemistry (RSC))

Ribozymes and RNA Catalysis (RSC Biomolecular Sciences)

Gas Kinetics and Energy Transfer: v.4: A Review of Chemical Literature (Specialist Periodical Reports)

The Alkaloids: A Review of Chemical Literature: v. 8 (Specialist Periodical Reports)

Electrochemistry: v.8: A Review of Chemical Literature (Specialist Periodical Reports) (Vol 8)

Inorganic Chemistry of the Transition Elements: v. 4: A Review of Chemical Literature (Specialist Periodical Reports)

Electrochemistry Vol 6: The Periodic Table of the Elements Wall Chart (Specialist Periodical Reports) (v. 6)

Electrochemistry: v.10: A Review of Chemical Literature (Specialist Periodical Reports)

Catalysis, Volume 22 (RSC Specialist Periodical Reports)

Catalysis: v. 18 (Specialist Periodical Reports)

Catalysis: Vol. 19 (Specialist Periodical Reports)

Catalysis: Vol 17 (Specialist Periodical Reports)

Photochemistry, Volume 37 (Specialist Periodical Reports)

Organometallic Chemistry, Volume 36 (Specialist Periodical Reports)

Organometallic Chemistry, Volume 31 (Specialist Periodical Reports)

Alkaloids Volume 4: Specialist Periodical Rep (Specialist Periodical Reports) (v. 4)

Organosphophorus Chemistry (Specialist Periodical Reports) (v. 4)

Heterocyclic Chemistry: v.4 (Specialist Periodical Reports)

Photochemistry (Specialist Periodical Reports) (Vol 13)

Nuclear Magnetic Resonance: Volume 37 (Specialist Periodical Reports)

Terpenoids and Steroids: Volume 12 (Specialist Periodical Reports)

Terpenoids and Steroids: Volume 3 (Specialist Periodical Reports)

Dielectric and Related Molecular Processes: Volume 1 (Specialist Periodical Reports)

Organophosphorus Chemistry (Volume 18) Specialist Periodical Report

Chemical Modelling: Vol. 2: Applications and Theory (Specialist Periodical Reports)

Heterocyclic Chemistry: v.3 (Specialist Periodical Reports) (Vol 3)

Theoretical Chemistry,vol 2 (Specialist Periodical Reports) (v. 2)

Catalysis: A Review of Chemical Literature: Vol 15 (Specialist Periodical Reports)

Catalysis: A Review of Chemical Literature: Vol 16 (Specialist Periodical Reports)

Inorganic Chemistry of the Transition Elements, Volume 2 (Specialist Periodical Reports) (v. 2)

Chiral Sulfur Ligands: Asymmetric Catalysis (RSC Catalysis Series)

Organophosphorus Chemistry - Volume 22 (SPR Organophosphorus Chemistry (RSC))

Ribozymes and RNA Catalysis (RSC Biomolecular Sciences)

Gas Kinetics and Energy Transfer: v.4: A Review of Chemical Literature (Specialist Periodical Reports)

The Alkaloids: A Review of Chemical Literature: v. 8 (Specialist Periodical Reports)

Electrochemistry: v.8: A Review of Chemical Literature (Specialist Periodical Reports) (Vol 8)

Inorganic Chemistry of the Transition Elements: v. 4: A Review of Chemical Literature (Specialist Periodical Reports)

Electrochemistry Vol 6: The Periodic Table of the Elements Wall Chart (Specialist Periodical Reports) (v. 6)

Electrochemistry: v.10: A Review of Chemical Literature (Specialist Periodical Reports)

Recommend Documents