Catalysis

Volume 22

A Specialist Periodical Report

Catalysis

Volume 22

A Review of Recent Literature

EditorsJames J. Spivey, Louisiana State University, USAKerry M. Dooley, Louisiana State University, USA

AuthorsNicolas Bion, University of Poitiers, FranceCristina Della Pina, University of Milan, ItalyDaniel Duprez, University of Poitiers, FranceFlorence Epron, University of Poitiers, FranceErmelinda Falletta, University of Milan, ItalyJozsef L. Margitfalvi, Institute of Surface Chemistry and Catalysis, Budapest,HungaryF. C. Meunier, University of Caen, FranceMichele Rossi, University of Milan, ItalyJohannes W. Schwank, University of Michigan, USAK. Seshan, University of Twente, The NetherlandsAndrew R. Tadd, University of Michigan, USAEmılia Talas, Institute of Surface Chemistry and Catalysis, Budapest, Hungary

ISBN 978-1-84755-951-7DOI 10.1039/9781847559630ISSN 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 fornon-commercial purposes, or criticism or review as permitted under the termsof the UK Copyright, Designs and Patents Act, 1988 and the Copyright andRelated Rights Regulations 2003, this publication may not be reproduced,stored or transmitted, in any form or by any means, without the priorpermission in writing of The Royal Society of Chemistry, or in the case ofreprographic reproduction only in accordance with the terms of the licencesissued by the Copyright Licensing Agency in the UK, or in accordance with theterms of the licences issued by the appropriate Reproduction RightsOrganization outside the UK. Enquiries concerning reproduction outside theterms stated here should be sent to The Royal Society of Chemistry at theaddress 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

If you buy this title on standing order, you will be given FREE accessto the chapters online. Please contact sales@rsc.org with proof ofpurchase to arrange access to be set up.

Thank you.

Preface

James J. Spiveya and Kerry M. Dooleya

DOI: 10.1039/9781847559630-FP005

Recent research activity in catalysis has centered on energy related pro-cesses, and the synthesis of higher value compounds using biomass as wellas conventional reactants. The work reported here addresses these twoimportant areas.

First, Nicolas Bion, Florence Epron, and Daniel Duprez (Universitede Poitiers, France) provide a review of bioethanol reforming, particularlyas compared to conventional hydrocarbon reforming. Reforming of bio-derived ethanol is an essential element in an overall process to deliverhydrogen from renewable resources. The authors show important differ-ences between reforming of an oxygenate such as ethanol and reforming ofconventional hydrocarbons—e.g., deactivation.

Johannes Schwank and Andrew Tadd (Univ. Michigan) examine aclosely related reaction—catalytic reforming of liquid hydrocarbons, par-ticularly for application to solid oxide fuel cells, which are being developedcommercially for use as auxiliary power systems. They consider steam re-forming, catalytic partial oxidation, and autothermal reforming, each ofwhich has different challenges in terms of heat transfer, kinetics, andcatalyst deactivation.

Another energy-related subject is the use of spectroscopic methods tostudy reactions of interest in environmental control systems. Specifically,Fred Meunier (CNRS, France) reviews spectrokinetic methods to studyreactions such as NOx reduction using IR spectroscopy. He also shows howspectroscopy can be applied to the water-gas-shift reaction, and importantreaction in the catalysis of fuel reforming for hydrogen production. Heshows how DRIFTS can be combined with isotopic analysis to study thedynamics of the active catalyst surface with a combination of these twomethods.

K. Seshan (Univ. Twente, Netherlands) reports on oxidative conversionof low molecular weight alkanes to the corresponding olefins. These olefinsare important feedstocks for the chemical industry. This particular reviewfocuses solely on oxidative conversion of these alkanes to olefins, which hasadvantages in terms of thermodynamics and kinetics compared to alter-native processes, primarily catalytic or steam cracking.

Asymmetric hydrogenation of activated ketones is a particularly de-manding reaction, producing high-value, enantiopure products for thespecialty chemical industry. Jozef Margitfalvi and Emilia Talas (Institute ofSurface Chemistry and Catalysis, Budapest) review in particular hetero-geneous catalysts for these reactions, although homogeneous catalysts areperhaps more widely used at present. However, homogeneous metal-basedcatalysts are expensive due to the need for chiral ligands. This chapter shows

aGordon A. and Mary Cain Dept. Chemical Engineering, Louisiana State University, BatonRouge, LA 70803

Catalysis, 2010, 22, v–vi | v

�c The Royal Society of Chemistry 2010

recent progress in development of heterogeneous catalysts, such as en-capsulation of a chiral metal complex in micropores.

Finally, Cristina Pina, Ermelinda Falleta, and Michele Rossi (Universitadi Milano, Italy) provides a review of gold catalysis. This focuses onreactions such as selective oxidation of alcohols and carbohydrates, andselective oxidation of hydrocarbons. Within this chapter, a more generalsummary of gold catalysis in the synthesis of advanced materials (e.g., thosebased on PAN) is provided.

We greatly appreciate the efforts of the authors who have contributed tothis volume. We thank the Royal Society of Chemistry for their support ofthis series. Comments are welcome.

vi | Catalysis, 2010, 22, v–vi

CONTENTS

Preface v

James J. Spivey and Kerry M. Dooley

Bioethanol reforming for H2 production. A comparison with

hydrocarbon reforming

1

Nicolas Bion, Florence Epron and Daniel Duprez

1. Introduction 12. The steam reforming of hydrocarbons 23. Steam reforming of ethanol 234. Utilisation of crude bioethanol 375. Conclusions and recommendations 46References 48

Catalytic reforming of liquid hydrocarbons for on-board solid

oxide fuel cell auxiliary power units

56

Johannes W. Schwank and Andrew R. Tadd

1. Introduction 562. Fuel properties and SOFC fuel requirements 583. Catalysts for reforming of liquid hydrocarbons 61

CoverImage provided courtesyof computational sciencecompany Accelrys(www.accelrys.com). Anelectron density isosurfacemapped with the electrostaticpotential for an organometallicmolecule. This shows thecharge distribution across thesurface of the molecule withthe red area showing thepositive charge associatedwith the central metal atom.Research carried out usingAccelrys Materials Studioss.

Catalysis, 2010, 22, vii–ix | vii

�c The Royal Society of Chemistry 2010

4. Fuel reforming methods 625. Deactivation of reforming catalysts 696. On-board reforming of fuels for SOFC APU applications 767. Systems engineering aspects of on-board fuel processing 818. Conclusions 84Acknowledgments 85References 85

Coupling kinetic and spectroscopic methods for the investigation of

environmentally important reactions

94

F. C. Meunier

1. Introduction 942. The bases of spectrokinetic analyses 953. Investigation of the selective reduction of NOx with propene

over Ag/Al2O3

96

4. Spectrokinetic operando investigation of catalytic reactions 1055. Overall conclusions 116References 117

Oxidative conversion of lower alkanes to olefins 119

K. Seshan

1. Introduction 1192. Oxidative conversion of alkanes to olefins over oxide

catalysts with redox properties122

3. Oxidative conversion of alkanes over oxide catalysts withno formal ‘‘redox’’ properties

126

4. Catalytic alkane oxidation at ambient conditions using coldplasma C–C, C–H scission vs C–C bond coupling

131

Acknowledgements 139References 139

Asymmetric hydrogenation of activated ketones 144

Jozsef L. Margitfalvi and Emılia Talas

1. Introduction 1442. Cinchona alkaloids 1513. Alkaloids used in Oritos’s reaction 1604. Methods and approaches used 1645. Specificity of Orito’s reaction 178

viii | Catalysis, 2010, 22, vii–ix

6. Spectroscopic investigations 2207. Theoretical calculations 2358. Reaction mechanisms and related calculations 2439. Conclusions 258Abbreviations used 261References 262

Gold catalysis in organic synthesis and material science 279

Cristina Della Pina, Ermelinda Falletta and Michele Rossi

1. Introduction 2792. Gold catalysis in organic synthesis 2803. Selective oxidation of carbohydrates 2924. Selective oxidation of hydrocarbons 2965. Gold catalysis in material science 3006. Conclusions 313References 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 hydrocarbonfractions (natural gas, naphtha, . . .) on an industrial scale. Replacing fossilfuels by biofuels for H2 production has attracted much attention with anincreased interest for bioethanol steam reforming. Kinetics and mechanismsof hydrocarbon-SR and alcohol-SR present some similarities but also somevery important differences due to alcohol reactivity much more complexthan that of hydrocarbons. The scope of this report is to compare the twoprocesses in terms of reaction mechanisms. Attention will also be paid to thecase of crude bioethanol.

1. Introduction

Whereas hydrogen is the most abundant element of the Universe, it isrelatively 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 inwater, with carbon in fossil hydrocarbons, both with oxygen and carbonin bioresources (carbohydrates, cellulosic and lignocellulosic matter,lignin, . . .) and more rarely with other elements. Water is by far the mainsource of hydrogen on Earth (Table 1).

The stock of hydrogen available in fresh waters (lakes and rivers) is thenof 1.3� 1013 tons while the total ressources in hydrogen in oceans ans seasamount to 1.5� 1017 tons. Comparatively, the ressources in hydrogenavailable 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 naturalgas7 and of 0.8 in coal,8 the stock of hydrogen in fossil fuels would notexceed 111� 109 T (23 in crude oil, 30 in natural gas and 58 GT in coalreserves), i.e. two orders of magnitude less than the amount of hydrogencontained in fresh waters. Unfortunately, water is a stable molecule needinga high energy input to recover hydrogen as H2. This energy may be providedby (i) a chemical source by oxidizing the carbon of an hydrocarbon into COand CO2 (steam reforming); (ii) by electricity (water electrolysis) or (iii) byphotons (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 1 350 000 000 km3 1.5� 1017 T H2

Deep continental waters 36 000 000 km3 4� 1015 T H2

Fresh waters (easily accessible)

Lakes and Rivers 110 000km3 1.3� 1013 T H2

Rivers 1700 km3 1.9� 1011 T H2

aUniversity 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

production9,10 and only this way of hydrogen production will be examinedin this Chapter. On an environmental point of view, steam reforming is not agreen process since all (or almost all) the carbon of the hydrocarbons istransformed into carbon dioxide. To avoid this drawback, fossil fuels maybe replaced by biofuels. Carbon dioxide is still produced but it may berecycled to new biomolecules by photosynthesis. The annual productionof biomass in the World would be comprised between 150 and 420� 109

metric tons.11–13 The mean hydrogen content in biomass being comprisedbetween 5 and 7 wt-%,13 the stock of hydrogen in this renewable matterwould be close to 11� 109 T/year. In other words, ten years of biomassproduction would be sufficient to recover all the hydrogen content of fossilfuels. However, a great part of this biomass is composed of wood, difficultto transform into valuable products. For that reason, only productsderived from cellulosic and hemicellulosic biomass have been consideredfor hydrogen production. Since ten years, intensive researches have been de-voted to the steam reforming of bioethanol which is a fuel well-adapted to theproduction of hydrogen.14 This Chapter deals for a great part with thisprocess, with a special attention paid to the use of crude bioethanol. In a firstpart, however, the steam reforming of hydrocarbons (aromatics and alkanes)will be reviewed as the model of many mechanistic investigations. This willallow one to compare the steam reforming of hydrocarbons with the steamreforming 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 þH2O! COþ 3H2 DH0298 ¼ þ206 kJmol�1 ð1Þ

2. The steam reforming reaction leading to CO2 and H2

CH4 þ 2H2O! CO2 þ 4H2 DH0298 ¼ þ165 kJmol�1 ð2Þ

3. The water gas shift reaction (WGS)

COþH2O! CO2 þH2 DH0298 ¼ �41 kJmol�1 ð3Þ

4. The coking reaction

CH4 ! Cþ 2H2 DH0298 ¼ þ75 kJmol�1 ð4Þ

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

World proven reserves

(109 TOE)

Annual consumption

(109 TOE)

Number of years of

reserve

Oil 181 4.0 45

Gas 172 2.7 65

Coal 610 3.8 160

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

2 | Catalysis, 2010, 22, 1–55

For higher hydrocarbons, a similar set of reactions may be written; forinstance the reactions of n-heptane and of toluene leading to CO and H2

become:

C7H16 þ 7H2O! 7COþ 15H2 DH0298 ¼ þ1107 kJmol�1 ð5Þ

C7H8 þ 7H2O! 7COþ 11H2 DH0298 ¼ þ869 kJmol�1 ð6Þ

Except for the WGS reaction, all the reactions involved in the hydrocarbonsteam reforming are strongly endothermic with an increase of the number ofmolecules. They are thus favored at high temperatures and low pressures.The temperature effect is illustrated in Fig. 1 which shows the change with Tof the gas composition at equilibrium in the methane steam reforming(initial conditions H2O/CH4=1, P=1 bar). Calculations were carried outby minimizing the sum of the Gibbs free energies of formationof all the compounds (reactants and products) while keeping constant thenumber of moles of each element (here C, H and O). Details of the pro-cedure are given in Perry’s Handbook.19 Thermodynamic data (molarGibbs free energy of each compound) are taken from Stull et al.20 MaximalH2 production is observed around 700 1C. Above 700 1C, the H2 mol% doesno longer increase because of the preferential formation of CO at hightemperature. This is coherent with the WGS equilibrium: reaction 3 beingexothermic, CO2 is favored at low temperature while the reverse reaction(RWGS) yielding CO is favored at high temperature. At 900 1C, totalconversion of methane can be achieved yielding quasi-exclusively a syngaswith the composition given in equation 1 (75% H2þ 25% CO).

0%

10%

20%

30%

40%

50%

60%

70%

80%

100 300 500 700 900

Mol

e %

T (°C)

H2O

CH4

H2

CO

CO2

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

The equilibrium gas compositions in methane, n-heptane and toluene steamreforming at 700 1C are compared in Table 3. The initial state is a steam/hydrocarbon mixture with a molar ratio corresponding to the stoichiometryof 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 at700 1C. The vol.% of hydrogen expected in dry gases amounts to 70%. Theformation of methane and CO is favored in the steam reforming of C7compounds at 700 1C, which tends to decrease the hydrogen content in drygases. In every cases, the CO-to-CO2 molar ratio is remarkably constant(around 7.7) whathever the starting hydrocarbon. The conversion of C7hydrocarbons (not reported in Table 3) is total whatever the temperature.Thermodynamics predicts that H2 production from C2þ hydrocarbons iscontrolled by methane formation. If, kinetically, methane formation canbe avoided, the equilibrium gas composition is significantly changed. Anexample is shown in Fig. 2 for toluene steam reforming without CH4 for-mation. The complete conversion of toluene only occurs around 550–600 1Cand the maximal H2 formation is observed at 500 1C, much below thecorresponding value when methane can be formed. Interestingly, benzeneformation can then be observed with a maximum around 400 1C. If methaneformation may be avoided or suppressed, toluene dealkylation can occur inthe 400–500 1C range of temperature: this is the so-called toluene steamdealkylation (TSDA) which has been largely studied in the past as a wayto produce benzene from toluene21–28 (and more generally from alkylben-zenes)29,30 without H2 consumption.

2.2 Kinetics and mechanisms

Methane being a very stable molecule, the steam reforming of natural gasshould be carried out at high temperatures (around 600–700 1C) and it canbe expected that methane activation is a critical step of the reaction (seeSection 2.2.3). Heavier hydrocarbons are more reactive and there are manyindications in the literature that water activation may be the rate deter-mining step in the steam reforming of these compounds, specially at lowertemperatures (400–600 1C).

2.2.1 Aromatics

2.2.1.1 Reaction scheme and catalytic activity. Toluene will be chosen asmodel hydrocarbon illustrating this class of compounds. As mentioned in

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)

H2O H2 CO CO2 CH4 nG

Methane Wet % 5.46 65.62 18.69 2.39 7.84 3.46

Dry % – 69.41 19.77 2.53 8.30

n-Heptane Wet % 4.91 58.04 25.51 3.32 8.23 18.89

Dry % 61.03 26.83 3.49 8.65

Toluene Wet % 4.14 50.52 32.90 4.15 8.29 15.44

Dry % – 52.70 34.32 4.33 8.65

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

4 | Catalysis, 2010, 22, 1–55

Section 2.1, the toluene steam reforming may lead to dealkylation (7) inthe 400–500 1C temperature range where methane formation is kineticallyunfavored (8, 9).

C7H8 þH2O! C6H6 þ COþ 2H2 DH0298 ¼ 164 kJmol�1 ð7Þ

C7H8 þH2 ! C6H6 þ CH4 DH0298 ¼ �42 kJmol�1 ð8Þ

C7H8 þ 10H2 ! 7CH4 DH0298 ¼ �574 kJmol�1 ð9Þ

Equations 8, 9 clearly show that methane formation decreases the hydrogenyield. 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 couldbe established:30

7YH2 ¼ 3YB þ 11YCO þ 18YCO2� 10YCH4

ð10Þ

Intrinsic activity and selectivity to benzene of Group 8-9-10 metals (exGroup VIII) supported on g-Al2O3 (210m

2 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 thatthe flow rate of dry gases (H2þCOþCO2þCH4) is proportional to thetoluene conversion. This property allows one to extrapolate the runningactivity to zero-time. Turnover frequencies given in Table 4 are the intrinsicactivities determined by this technique. The deactivation is due to a carbondeposit on the catalyst (metal and support). It was proved that toluenesteam dealkylation is a relatively ‘‘structure’’ insensitive reaction: the

0%

10%

20%

30%

40%

50%

60%

70%

80%

100 200 300 400 500 600 700 800

Mol

e %

T (°C)

H2H2O

TOL

CO

BENZ

CO2

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

Catalysis, 2010, 22, 1–55 | 5

carbon deposit does not change the turnover frequency per free metalatom which remains very close to the turnover frequency extrapolated atzero-time. Free metal surface area in coked catalysts were measured ac-cording to a procedure detailed in refs.31,32 Initial selectivity to benzeneis the selectivity extrapolated at zero conversion. An interesting feature ofthe toluene steam dealkylation reaction is the increase of SB with con-version. For instance, rhodium selectivity reaches 86% at 10% conversionand 92% around 40% conversion. This is due to a complex behaviour ofthe catalyst including two phenomena: (i) though the dealkylation isstructure insentive, the total steam reforming of toluene (6) is not and isstrongly poisoned by coke deposits; (ii) CO is an inhibitor of both deal-kylation and total reforming but the former reaction is much lessaffected than the latter one.22,33 The relative activities reported in Table 4show that the steam dealkylation reaction is not very sensitive to the natureof metal: there is less than one order of magnitude difference betweenthe most active metal (Rh) and the less active one (Ir). The metal rankingfound by Duprez et al.27 is close to that reported by Grenoble24 in similarconditions (440 1C, alumina support of 175m2 g� 1), except the water-to-toluene ratio (3.25 in the Grenoble’s study). The only difference lies inthe position of Ru found more active by Grenoble. Kim showed that pro-moting alumina by vanadium oxide did not change significantly the metalranking.34

2.2.1.2 Support effects and reaction mechanism on rhodium catalysts.Contrasting with the low sensitivity to nature of metal, the steam deal-kylation reaction appeared as very sensitive to nature of support. This isillustrated in Table 5 with the rhodium as active metal. There are two tothree orders of magnitude difference between the activity of rhodiumcatalysts supported on chromia or aluminochromia catalysts35 and thosesupported on silica or carbon.25,27 The data of Table 5 show that the dif-ferences are not due to metal dispersion effects. Catalyst characterizationcarried out after test also confirmed that there was no sintering duringthe kinetic measurements affecting more certain supports. A bifunctionalmechanism was proposed by Grenoble26 and Duprez et al.27 to explainthis dramatic support sensitivity. In this mechanism schematized on Fig. 3,

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 470 100 81

0.6%Pd 134 29 97

1.1%Pt 88 19 98

5%Ni 77 17 59

5%Co 67 15 51

0.6%Ru 64 14 53

1.15%Ir 60 13 88

6 | Catalysis, 2010, 22, 1–55

the hydrocarbon molecule would be adsorbed and activated on metal siteswhile the water molecule would be activated on support sites.

The molecule of toluene undergoes a dissociative chemisorption on ametal site M leading to a molecule of benzene and an alkyl fragment (11).This reaction should require two metal sites. However, benzene being lessstrongly adsorbed than toluene on metals, it is immediately desorbed onceformed.

C6H5�CH3 þM! CHx�Mþ C6H6 þ2�x2

H2 ð11Þ

Water is activated on support sites (S–O–S) according to equation 12:

H2Oþ S�O�S! 2S�OH ð12Þ

The final step is a transfer of OH groups to metal particles where they reactto form carbon oxides and hydrogen:

CHx�Mþ 2S�OH! COþ x

2H2 þ S�O�SþM ð13Þ

Table 5 Support effect in toluene steam dealkylation at 440–480 1C (base 100 for well dis-

persed Rh on g-Al2O3)

Support

Surface area

(m2 g� 1)

Rh dispersion

(%)

Relative activity of

rhodium (per metal site) Reference

Cr2O3 40 20 300–400 35

g-Al2O3 210 90–100 100 23,25,27

a-Cr2O3 10 90–100 50 36

Al2O3-SiO2 300 50 40–50 30

TiO2 10 30 20 27

Al2O3-SiO2 110 40 15 37

SiO2 330 40 20 27

SiO2 260 90–100 20 37

SiO2 300 70 3 25

C 950 40 1 25

unsupported 5 1.2 0.1 25

Surfacemigration

H2O

CXHY

metalOH OH

CH3

support

Fig. 3 Bifunctional mechanism of toluene steam dealkylation

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 kyCHxx2OH ð14Þ

where yCHx is the surface coverage of CHx fragments on metal siteswhile xOH is the OH group coverage on support sites. As step (13) implies atransfer of OH groups through the metal/support interface, it is assumedthat k=KI0, I0 being the length of the particle perimeter per gram ofcatalyst. Surface coverages are deduced from equations 11, 12 andLangmuir-type hyperbolic expressions for equilibium adsorption of mol-ecules A can be approximated to power-law expressions using the followingequation:

KAPA

1þKAPA¼ aðKAPAÞn ð15Þ

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

TOF ¼ r

M0¼ CðS20l0Þ

1�nPnTP

mð1�nÞW ð16Þ

where M0 and S0 are the numbers of metal sites and of support sites, re-spectively, per gram of catalyst; PT and PW are the partial pressures oftoluene and water. Knowing the weight loading (xm %) and the dispersionD0 (%) of metal, the specific perimeter I0 can be calculated by:

I0 ¼ bD20xm ð17Þ

with b ¼ 1:6� 10�7rA2

mol

Mðcubic particlesÞ ð18Þ

r being the metal density (gm� 3), Amol the molar surface of metal (m2

mol� 1) and M, the atomic weight (gmol� 1). For Rh, r=12.45� 106 gm� 3,Amol=47633m2mol� 1 (equidistribution of low index faces) andM=102.9 gmol� 1, which gives b=4.28� 105mg� 1. It is worth noting that,for a catalyst commonly used in steam reforming (0.6%Rh, mean dispersionof 50%), I0 amounts to 6.4� 108mg� 1, a length greater than the Earth-Moon distance. This justifies the great impact the reactions at metal/supportinterfaces 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 tolueneand to water.25 Most authors found orders with respect to toluene com-prised between 0.03 and 0.25 while those with respect to water are generallyclose 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 metalloading and the dispersion (almost a factor 20 between the most activecatalyst and the less active one). Applying the bifunctional model withn=0 reduces the variation to a factor 4 while a very good fit is obtained

8 | Catalysis, 2010, 22, 1–55

for n=0.3 (factor 1.7 and less than 1.1 for nine of the ten samples). Tosum up, the kinetic model based on the bifunctional mechanism representsadequately the changes of activity of Rh/Al2O3 catalysts in toluene steamdealkylation.

2.2.1.3 Other metals. The support effect evidenced for Rh catalystswas investigated for other metals by comparing their activity on aluminaand silica27 (Table 7). There are two groups of metals: those showing astrong support effect in steam dealkylation (Pt, Rh, Pd and to a lesserextent, Ir) and those having the same activity on alumina and silica (Ni;Co and Ru).

In an investigation of multifuel reforming (including toluene), Wang andGorte confirmed the dramatic effect of the support in these reactions.39 Forinstance, Pd/CeO2 was about 7 to 10 times more active than Pd/Al2O3 in thesteam reforming of toluene between 400 and 500 1C. As a rule, ceria orceria-based oxides were found to be very good supports for metals in severalsteam reactions (steam reforming and water-gas shift).40–43 The mechanismof these reactions on ceria-based catalysts is likely to differ from that pro-posed on alumina catalysts in equations 11–13. Ceria is a reducible support

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

TOF

ðD20xmÞ1�n

Rh loading

(xm % Rh)

Dispersion

(D0 %)

TOF

(h� 1) n=0 n=0.3

0.031 100 60 0.194 1.08

0.063 100 160 0.254 1.76

0.18 95 180 0.111 1.02

0.58 92 430 0.088 1.12

0.60 96 460 0.083 1.10

4.82 45 650 0.067 1.05

4.87 61 1070 0.059 1.12

10.1 13.7 200 0.106 1.02

10.3 32 700 0.066 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 88 15 5.87

0.6% Rh 470 91 5.16

0.6%Pd 134 54 2.48

1.15%Ir 60 45 1.33

5%Ni 77 69 1.11

5%Co 67 64 1.05

0.6%Ru 64 62 1.03

Catalysis, 2010, 22, 1–55 | 9

able to dissociate the water molecule and to produce directly hydrogen in aredox process. As proposed by Wang and Gorte,39 hydrocarbon activationwould be very similar to equation 11 of the OH-migration mechanism whileequation 12, 13 should be replaced by the following steps:

H2Oþ Ce2O3 ! 2CeO2 þH2 ð19Þ

CHx�Mþ 2CeO2 ! CO�Mþ Ce2O3 þ x2H2 ð20Þ

The possible occurrence of step 19 has been demonstrated by direct de-composition of water over reduced ceria.44–46 The writing of equations 19,20 is certainly oversimplified since Ce2O3 very likely cannot be formedunder the conditions of steam reforming. There exist numerous sub-oxidesCeO2� x (with 0oxo0.5) whose existence is more probable in theseconditions.47

Metals of class 2 (Ni, Co, Ru) are not support-sensitive. These metalshave the lowest Mnþ /M1 electrochemical potential and can dissociate thewater molecule in steam reforming conditions. For these metals, a mono-functional mechanism with all steps occuring on the metal was proposed.27

equations 12, 13 are replaced by equations 21, 22 respectively

H2OþM ¼ HO�Mþ 12H2 ðequilibrium constant KWÞ ð21Þ

CHx �MþHO�M! COþ x2H2 þ 2M ðrate constant kÞ ð22Þ

This mechanism led to the following rate equation:

TOF ¼ kTPT

1þK1� kTPT

kKM0

� �ð23Þ

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

K ¼ KWPH2O

P1=2H2

ð24Þ

2.2.1.4 Isotopic exchange studies. 16O/18O and H/D isotopic exchangestudies were performed to measure the rate of diffusion of OH groupsinvolved in the bifunctional mechanism of steam reforming.48–51 Theprinciple of the measurement is depicted in Fig. 4.

Exchange experiments are typically performed with pure 18O2 at t=0. Twoconditions should be fulfilled: (i) adsorption/desorption of O2 on the metalparticles (step 1) should be very fast. This is verified by measuring the rate ofisotopic equilibration between 18O2 and 16O2 on the metal; (ii) the rate ofdirect exchange between 18O2 and the support (step 5) should be negligible.This is verified by measuring the rate of exchange on bare supports. Similarexperiments were carried out with deuterium to measure the rate of hydrogenmigration. A simple kinetic model is sufficient to calculate the coefficient ofsurface diffusion DS for supports having moderate O or H mobility.50 Whensurface mobility is very fast, more complex kinetic models should beused.52,53 Several important conclusions could be drawn from these studies.

10 | Catalysis, 2010, 22, 1–55

1. While hydrogen is very mobile on most supports below 100 1C, oxygensurface diffusion is a relatively slow process and requires temperatures in the300–500 1C range to be measurable.

2. However, activation energy for oxygen surface diffusion (50–100 kJmol� 1) is significantly higher than that for hydrogen (10–20 kJmol� 1). As aconsequence, oxygen and hydrogen migrations occur at similar rates around5001C.

3. Rhodium is the best metal for oxygen activation.54 Adsorption/de-sorption of O2 on this metal is very fast and is not very sensitive to the metalparticle 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 tothe metal particle size and, contrary to the case of Rh, it is much faster onbig Pt particles. Oxygen equilibration is very slow on Pd while Ru and Ircould be good candidate for replacing Rh under certains conditions.55

4. The relative values of coefficients of surface diffusion for oxygen andhydrogen on selected oxides are given in Table 8.

Oxygen and hydrogen diffusing at the same rate in the 400–500 1C rangeof temperature, it is very likely that there is a collective migration of OHgroups at these temperatures on most oxides. A crucial point is that OHs arevirtually immobile on silica while the coefficient of diffusion would be twoorders of magnitude higher on alumina. This result is in agreement with thebifunctional mechanism with the slow step being the OH group migration

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 Hydrogen at 75 1C

CeO2 2300 CeO2 80

MgO 50 MgO 22

ZrO2 28 CeO2/Al2O3 16

CeO2/Al2O3 18 Al2O3 10

Al2O3 10 ZrO2 2.3

SiO2 0.1 SiO2 Very low

Surfacemobility

Metal

18O16O

16O2

18O2

18O16O

18O 18O 16O

18O2

Support

Fig. 418O2 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: surfacemigration; Step 4: place exchange of 18O with 16O Step 5: direct exchange gas/support.

Catalysis, 2010, 22, 1–55 | 11

from support to metal. This explains why silica is a poor support for steamreforming reactions. Unfortunately, carbon supports could not be investi-gated by isotopic exchange. There are some indication in the literature thatO species are quite mobile on carbon: for instance Lim et al. found that NOdecomposition was favored on Pt/C by continuous removal of O speciesfrom the metal by the support itself.56,57 The most probable explanationfor the low activity of carbon-supported metal catalysts might be the highhydrophobicity of carbon surface which cannot easily activate the watermolecule.

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 pro-duction from gasoline, heavy oils, biomass and tars. Toluene was chosen asmodel aromatic hydrocarbon in many studies with the objective to reformtotally the molecule and obviously not to produce benzene. However, steamdealkykation could be observed when the steam reforming of aromatic-reach fractions was used to produce hydrogen.

Fuel reforming. Hydrogen production for on-board applications wasinvestigated by Springmann et al.58 at 600–800 1C, 2–5 bar over rhodiumcatalysts deposited on metallic monoliths. The reaction led essentially to asyngas (30% H2þ 12% CO) with small amounts of benzene, methane andCO2. A detailed analysis of benzene reactivity supports the interpretationthat toluene is first dealkylated to benzene which is then gasified. Due tosignificant formation of coke, the catalyst is not stable in toluene steamreforming. Stable hydrogen production (with less carbon deposit) wasobtained in an autothermal ATR process (H2O:O2:TOL=14:3:1). Steamreforming or ATR of iso-octane, hexene and gasoline were also investigatedon the same catalysts. These hydrocarbons were slightly more reactive thantoluene 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 auto-

thermal reforming at 650–800 1C over a complex Rh catalyst (0.3%Rh/3%MgO/20%CeO2-ZrO2) washcoated on a ceramic monolith.59 Theauthors showed that alkanes led to minor amounts of methane together withthe 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 aswell as the detrimental effect of washcoating the catalyst on a ceramicmonolith of low thermal conductivity. This sulfur effect on coke formation isnot in agreement with previous results of Duprez et al. who showed thatsulfur might hinder coke formation in toluene steam dealkylation.31–32,60

The apparent discrepancy is likely coming from the differences of tempera-tures (440 1C in steam dealkylation and 650–800 1C in ATR of gasoline).

Sulfur poisoning is a critical problem in steam reforming processes. SR isoperating on reduced catalysts under conditions where the sulfur organiccompounds are themselves steam reformed into hydrogen, carbon oxidesand dihydrogen sulfide. Several studies were carried out to attempt solvingthis drawback. Azad et al. have developped several catalyst formulation forthe steam reforming of jet fuels in the presence of large amounts of sulfur

12 | Catalysis, 2010, 22, 1–55

(up to 1000ppm).61–63 The best support tested by Azad et al. was a ternaryoxide composed of 10 mol-% Gd2O3, 25mol-% ZrO2 and 65mol-% CeO2

doped or not by Y2O3 or CuO. Active metals were Rh,61,63 Pd62 or com-bination 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 sig-nificant role in the catalyst thioresistance. Two reactions can occur (25, 26)which help mitigate sulfur-led poisoning and long-term deactivation:

Reduction reaction:

CeO2ðsÞ þ ð2�nÞH2ðgÞ ¼ CeOnðsÞ þ ð2�nÞH2OðgÞ ð25Þ

Sulfidation reaction:

2CeOnðsÞ þH2Sþ ð2n�3ÞH2ðgÞ ¼ Ce2O2SðsÞ þ 2ðn�1ÞH2OðgÞ ð26Þ

The specific role of ceria in sulfur resistance of Rh catalysts for jet fuelreforming applications was confirmed by Strohm et al. who also showedthat Ni added to Rh reinforced the thioresistance of a Rh-CeO2-Al2O3

catalyst.64 Rhodium-based catalysts offered a better resistance to de-activation than Pd-ones and the addition of Cu significantly improved thethioresistance of the GdZrCeOx support. This property may be paralleledwith the exceptional OSC (oxygen storage capacity) of Rh-CuCeOx cata-lysts, showing the great mobility of oxygen (and probably sulfur) in thesematerials.65

To increase the hydrogen yield, combination of steam reforming (SR),autothermal reforming (ATR) and water gas shift is often proposed in in-tegrated processes.66,67 Gasification of a fuel composed of methylcyclo-hexane and toluene was studied by Wang et al. at 530 1C.68 Combining SRand WGS on a multifunctional catalyst (Ni-Re/Al2O3) allows to increasethe CO2/COþCO2 ratio up to 92% (at 16% conversion). This is coherentwith the excellent performances of Ni and Re in WGS as reported byGrenoble et al.69 These authors showed that the WGS reaction was verysensitive 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-% COþ 32 vol-%

H2O). From ref. 69

Support effect

Metal/Al2O3 TOF (h� 1) Catalyst Relative activity

Cu 43 400 Pt/Al2O3 90

Re 1380 Pt/SiO2 9

Co 890 Pt/C 1

Ru 695

Ni 370 Rh/Al2O3 13

Pt, Os 225 Rh/SiO2 1

Au 130

Fe, Pd 50

Rh 31

Ir 12

Catalysis, 2010, 22, 1–55 | 13

As WGS is an exothermic reaction, it is favored at low temperature. Forthis reason, Wang et al. found better perfomances in a classical two-bedprocess with ATR (instead of SR) at relatively high temperature and WGSat low temperature. A Ni/Ce-ZSM5 was developped for this application.The use of Ce as dopant could be anticipated, ceria being a very goodpromoter of the WGS reaction.40,41,70,71 Table 9 shows that Rh is a badWGS catalyst while it is considered as the best SR metal. Pt is a better WGScatalyst than Rh but it is less active in steam reforming. Interestingly, Pdhas often intermediary properties between those of Pt and Rh in severalreactions with the following ranking: RhWPdWPt in SR and PtWPdWRh in WGS. For these reasons, the most advanced catalyst formulationfor multifuel reforming applications include generally both Rh or Pd and aWGS component (most often a ceria-based materials). Wang and Gorteshowed that their Pd/CeO2 catalyst was active in the reforming of manyhydrocarbons: 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, gasolineand 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 correctlyvaporized under the form of very fine droplets (the critical point of theprocess), good hydrogen yields and selectivities may be obtained by ATR ofcomplex mixtures such as Diesel gasoils or gasolines. It should however bementionned that the swedish fuels used in this study contained low amountsof sulfur and that the aromatics content of the diesel was comparativelylower than in other countries. These factors allowed the catalyst to be moreresistant to sulfur poisoning and to coke formation than with more commonfuels.

Steam reactions (HC’s steam reforming or WGS) are key-reactions inThree-Way Catalysis (TWC) for converting pollutants in rich conditionswhen O2 concentration in gas phase falls below stoichiometry.41,42,73,74

However, with aromatic gasolines, there is a risk to form benzene by steamdealkylation. In the temperature window of 550–730 1C, Bruelmann et al.observed the net formation of benzene and toluene in the gases issued froma Pd-Rh-CeZrOx-Al2O3 TWC in real conditions.75 The gasoline used in thisstudy 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 C8þ alkylated com-pounds decreased in the post-catalyst exhaust gases, it was shown that theconcentration of benzene and toluene increased. Bruelmann et al. studied indetail the conversion of 12 alkylbenzenes representing 75% of all possiblebenzene precursors in the post-catalyst exhaust gases. These alkylbenzeneswere individually spiked in the pre-catalyst chamber to follow their con-version by chemical ionization mass spectrometry (CI-MS). Dealkylationreactions of ortho-substituted dialkylbenzenes show a clear preferencetowards benzene formation, whereas meta- and para-isomers mainly formtoluene at 630 1C with some shift of the product distribution towards ben-zene at higher temperature (680 1C). Steam dealkylation is by far the mostimportant reaction leading to benzene formation even though somehydrodealkylation 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 thecombustion chamber.76 The combustion of the gasoline with a gas enrichedin H2 leads to a significant decrease of pollutants and to a better engineyield.

Tar reforming. Gasification process of biomass leads to a variety ofproducts, specially tars which should be steam reformed to increase H2

yields.77 Again, benzene, toluene or oxygenated aromatics such as phenol orcresols are often chosen as model compounds. In the 1970–1980’s, con-siderable 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 thesestudies, the main objective was to hydrogenolyze C–C bonds to producelighter hydrocarbons without paying attention to H2 production. Steam oroxy-steam gasification of tars was studied in the 90’s over different dolo-mites with the objective to favor syngas formation.80–82 These materialswere proved to offer a relatively good gasification activity while minimizingthe catalyst cost. They may be a good support for Ni to increase H2 yieldwhile minimizing the coking rate.83 However, their poor mechanical sta-bility made them useless in fluidized reactors. They were progressivelyreplaced 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 BETarea of 4–5m2 g� 1. It was used alone in biomass gasification or as a supportof metals (mainly Ni) to get more active catalysts.85 A detailed investigationof toluene steam reforming over Ni/olivine was performed by Swierczynskiet al.86,87 The reaction temperature was varied in the 550–850 1C range andthe H2O-to-toluene molar ratio was maintained between 7.5 and 24 with astandard value of 16 for most kinetic studies. Total conversion of toluene isreached at 650 1C over Ni/olivine while the reaction on the bare supportstarts at 750 1C. Reaction selectivities are reported in Table 10 for standardtests at 850 1C. The hydrogen yield is 82% on Ni catalyst and only 28% onolivine (a 100% yield corresponds to 18 moles of H2 produced per moleof toluene injected). The thermodynamic limitation at high temperaturewhich tends to favor CO formation (only 11 moles H2 per mole of tolueneaccording 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 whichproves that methane, benzene and polyaromatic hydrocarbons (still formedon 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

Catalyst

Selectivities %a

H2 yield (%)bCO CO2 CH4 Benzene Polyaromatics

Olivine 69 5 2 6 14 28

Ni/olivine 66 31 0 0 0 82

a defined as the % of C in the product per C in reacted toluene. b based on 100% for 18 molesH2 produced per mole of toluene injected.

Catalysis, 2010, 22, 1–55 | 15

A detailed characterization of the active sites of Ni/olivine was performedby Swierczynski et al. by means of Mossbauer spectroscopy and severalother techniques (XRD, SEM, H2-TPR).88 They showed that the goodactivity of Ni/olivine was linked to the formation of a NiO-MgO solid so-lution at the olivine surface during the oxidation and to Ni alloying by Feduring the reduction. Both phenomena contribute to make the catalystmore resistant to sintering and to coke formation. To increase activity in targasification, Zhang et al. proposed to dope Ni/olivine with ceria.89 The bestresults for the steam gasification of benzene and toluene were obtainedbetween 700 and 830 1C on a 3% NiO-1%CeO2/olivine. Ceria allows toincrease benzene and toluene conversion and to form more H2 and CO2 andless CO and methane (Table 11).

Other supports of Ni or Co were used for the steam reforming of modeltar compounds: ceria-zirconia,90 Al-La coprecipitate in the presence of Niand 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 (100–130m2 g� 1) from 400 to 900 1C.90 Theyfound however a total conversion of toluene at 600 1C and above. Bonaet al. restricted their study to a temperature of 650 1C focusing theirinvestigation on the optimization of La/Ni and Co/Ni ratios.91 Tempera-tures from 700 to 875 1C were chosen by Coll et al.92 to study the steamreforming of five aromatic compounds: benzene, toluene, naphthalene,anthracene and pyrene representative of a biomass gasification tar whosecomposition is given in Table 12.

Benzene and toluene are by far the most reactive of the compounds test-ed, with conversions in the order of 1 gHC gcat

� 1min� 1 at a H2O/C ratioaround 4. On the other hand, pyrene (H2O/C=12.6) and naphthalene(H2O/C=4.2) are the least reactive compounds, with conversions varyingfrom 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 40.5 61.3 32.5 6.3 0.01 45.9 61.2 28.9 10.2 0.16

B 60.5 63.6 23.8 12.6 0.01 64.8 63.6 22.5 14.6 0.11

Table 12 Typical composition of a biomass gasification tar. From ref. 92

Compound wt-% Compound wt-%

Benzene 37.9 Three-ring aromatics 3.6

Toluene 14.3 Four-ring aromatics 0.8

Other alkylbenzenes 13.9 Phenolic compounds 4.6

Naphthalene 9.6 Heterocyclic compounds 6.5

Alkylnaphthalenes 7.8 Other 1.0

16 | Catalysis, 2010, 22, 1–55

the gas composition in these studies90–92 follows the thermodynamic ten-dency: decrease of CO2 and CH4 and increase of CO when the temperatureis increased.

Catalyst deactivation is one of the major problems encountered in targasification. Bain et al. investigated tar reforming with a special attentionto benzene, toluene and light alkane transformation. They modeled thekinetics of reforming and in parallel, the kinetics of deactivation of theirNi-alkali-Al2O3 catalyst at five temperatures from 775 to 875 1C.93 First-order rate equations were found to represent both the kinetics of reformingand that of deactivation. Large differences in activation energies forreforming and deactivation of total tar gasification on one hand and ofindividual hydrocarbon gasification (benzene, ethane, methane, . . .) on theother hand were observed. This shows that tar gasification is a complexprocess and that modeling gasification of some hydrocarbons present inthese tars may not represent the total gasification kinetics. The most criticalparameter for coke formation is the H2O/C ratio. Interestingly, Coll et al.92

gave the limit values of this parameter to prevent massive coke formationwhen different aromatic hydrocarbons are gasified (Table 13).

2.2.2 Non-methanic alkanes. Less attention was paid to the steamreforming of C2

þ alkanes. Kikuchi et al. reported a detailed investigation ofthe reforming of n-heptane at 5501C over Rh catalysts.94 N-heptane maybe converted by two reactions: (i) gasification by steam reforming and (ii)aromatisation into toluene. Toluene itself may be gasified by steam; how-ever, if the reaction is carried out at relatively low space time, it is possible todetermine the initial selectivities to gasification and to aromatisation.Interestingly, it can be seen on Fig. 5 that the intrinsic rates of steam re-forming fit well with the kinetic equation developped for the toluene steamdealkylation (16, 17). In Fig. 5, the ratio TOF/D2

0xm is remarkably constantwhen the metal loading in the catalysts (xm) was varied from 0.1 to 5% withmetal dispersion decreasing from 100% (0.1%Rh) to 45% (5%Rh).Aromatisation is likely to occur on metal sites exclusively: no correlationcan be found between TOF of this reaction and the parameter represen-tative of a bifunctional metal/oxide reaction.

The role of the support was confirmed by Muraki and Fujitani95 whoproposed the following rate equation for n-heptane reforming over Rh/MgAl2O4 catalysts:

r ¼ kKcPc

1þKcPc

� �KWPW

1þKWPW

� �ð27Þ

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 725 2.5

Naphthalene 795 3.7

Anthracene 790 6.6

Pyrene 790 8.4

Catalysis, 2010, 22, 1–55 | 17

where subscript C refers to n-C7 (adsorbed on metal sites) and W towater (adsorbed on support sites). Wang and Gorte investigated the steamreforming of different fuels over a Pd/CeO2 catalyst (see Section 2.2.1.5 fuelreforming).39 The same catalyst was proved to be more active and morestable than Pd/Al2O3 in the steam reforming of n-butane with a remarkableCO2 selectivity even for a relatively low H2O/C ratio.96

The specific role of ceria in steam reforming was linked to its redox prop-erties, 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

H2Oþ Ce2O3 ! 2CeO2 þH2 ð29Þ

2Hads ! H2 þ s ð30Þ

CHx;ads þ 2CeO2 ! COþ ðx=2ÞH2 þ Ce2O3 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 beimproved by doping the support with other oxides such as Gd2O3: verystable performances (for 160 h) were observed in iso-butane steam re-forming on a Pt-Ce0.8Gd0.2O1.9 catalyst.

99

Coke resistance of steam reforming catalysts is a key-point for aromaticprocessing and also for light alkane gasification. A remarkable improve-ment of coke resistance of Ni catalysts in propane steam reforming wasreported by Takenaka et al.100 Three Ni catalysts (Ni/MgO, Ni/Al2O3 andNi/SiO2) were prepared by conventional impregnation of pre-formed sup-ports while a special catalyst was prepared by a microemulsion route in-cluding both the Ni precursor, TEOS and hydrazine. The resulting materialprepared by microemulsion (named coat-Ni) consists in Ni particules ofhomogeneous size coated with an external layer of silica (Fig. 6).

0

5

10

15

20

25

0 1 2 3 4 5 6

TOF

/A

%Rh

Aromatisation

Steam reforming

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

18 | Catalysis, 2010, 22, 1–55

The strong interaction of Ni metal particles with silica, in the silica-coatedNi catalysts, prevents the sintering of Ni metal and carbon depositionduring the steam reforming of propane: the rate of coke formation at 600 1Cwas decreased by a factor 2 to 3 on coat-Ni with propane conversion ap-proaching 100%. Alumina was often used as support of Ni in alkane steamreforming. Ni, Fe and Co are metals able to form carbides, precursors ofcarbon filaments, when they react in a hydrocarbon atmosphere.101,102

Zhang et al. succeeded in preventing sintering and coke formation byusing a one-step sol-gel preparation.103 Nickel nitrate and aluminium tri-secbutoxide dissolved in ethanol were used as precursors. SG catalysts werecompared to conventional catalysts prepared by impregnation. SG samplesare characterized by smaller Ni particles strongly bound to alumina surface.The formation of carbon filaments requiring a detachment of Ni particlesfrom the surface104–106 is significantly slowed down in SG catalysts.

2.2.3 Methane steam reforming (MSR). Natural gas reforming is themain source of hydrogen in the World. Annual production od H2 is esti-mated to 400 billions m3, North America being the most important pro-ducer (230 billions m3).107 About 60% come from narural gas reformingand the major part of rest from petroleum refinery or naphtha reforming.For this reason, methane steam reforming has been the subject of hundredof papers and many reviews or monographies (see for instance thoses ofRostrup-Nielsen,15,16 Ross108 and Inui).109 Only the studies dealing withmechanistic considerations will be reviewed here. The central question inthis section is: does methane reforming obey a specific mechanism, differentof 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 similaritiesbetween the mechanisms of steam and dry reforming of methane (32).110

They suggested that CO2 was shifted to CO and water which could be themain (or unique) reactant of methane gasification.

CH4 þ CO2 ! 2COþ 2H2 ð32Þ

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

Catalysis, 2010, 22, 1–55 | 19

In 1993, Rostrup-Nielsen and Bak Hansen showed that methane steamreforming (MSR, 1) and methane dry reforming (MDR, 32) had relativelyclose rates on most metal catalysts (Ni, Ru, Rh, Pd, Ir, Pt) supported onmagnesia.111

Turnover frequencies were compared at 550 1C in the following con-ditions at the reactor inlet: 18.5 vol-% CH4, 74 vol-% CO2 or H2O and7.5 vol-%H2 (Table 14). The CO2-H2 reverse shift reaction was also in-vestigated 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 ofRostrup-Nielsen and Bak Hansen since the steam reforming reaction wouldbe 3 to 10 times faster than dry reforming on the other metals. Rostrup-Nielsen and Bak Hansen explain this apparent discrepancy by the CO in-hibition of both reactions. As dry reforming produces much more CO, thereaction would be more affected by this inhibition. This explanation wassupported by a clear relationship between the CO inhibition factor and theadsorption heat of CO on metal. Moreover, all metals being very active inRWGS reaction (Table 14, last column), the tendency to form CO in dryreforming would be the same for all the catalysts. The work of Rostrup-Nielsen and Bak Hansen was qualitatively confirmed by Qin et al.112,113 whoshowed that steam and dry reforming had the same types of reactionintermediates.

In 2004, Wei and Iglesia published a series or papers on the crucial role ofthe CH4 activation step both in methane steam reforming and in methanedry reforming (MDR) on Ni,114 Rh,115 Ir,116,117 Ru,118 and Pt.119,120 Themain 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 activationnor CO2 activation intervenes in the rate determining steps of methaneconversion.� 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 allthe metals. Moreover, the rates of CH4/CD4 cross-exchanges are signifi-cantly smaller than those of reforming reactions. All these features suggest

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 1.1 2.2 1.9 1.2 5.3

1.4% Ru 3.0 8.9 2.9 3.1 8.7

1.1% Rh 2.2 8.1 1.9 4.3 5.4

1.2% Pd 1.4 1.6 0.18 8.9 8.0

0.9% Ir 1.3 4.5 0.44 10.2 8.6

0.9%Pt 1.0 2.0 0.36 5.6 7.2

20 | Catalysis, 2010, 22, 1–55

that the steps including C–H breaking in methane are irreversible and arethe slow steps of the mechanism for both MSR and MDR.� Reactions with CH4/CO2/D2 and CH4/CO2/

13CO mixtures show thatH/D distributions in H-containing products (including water) and 12C/13Cdistributions in COx are very close to equilibrium. These results confirmthat both water and CO2 activation are fast processes and as a rule, thewater gas shift reaction is equilibrated on all the metals at 600 1C.Wei and Iglesia also found a strong particle size effect: turnover freqenciesin methane reforming generally increased when decreasing the cluster size ofmetal. This was ascribed to the creation of a high number of low-co-ordination metal sites on small particles favoring the C–H cleavage of themethane molecule. Finally, they showed that there was a moderate supporteffect if any: for instance, TOF are very close when Rh or Pt are supportedon Al2O3, ZrO2 or CeO2-ZrO2.

115,119

The crucial role of CH4 activation in methane steam reforming on Rh/a-Al2O3 was confirmed by Tavazzi et al.121 and Donazzi et al.122,123 Thesekinetic studies were carried out in an annular reactor avoiding any mass andheat transfer artifact. Maestri et al. used the experimental data of Donazzito establish a hierarchy between the different kinetic models of methaneconversions (oxidation, steam reforming, dry reforming) by a microkineticapproach 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� ! CH�3 þH� ð33Þ

CH�3þ� ! CH�2 þH� ð34Þ

CH�2þ� ! CH� þH� ð35Þ

CH�þ� ! C� þH� ð36Þ

Water or CO2 activation steps

H2Oþ 2� ! OH� þH� ð37Þ

CO2 þ 2� ! CO� þO� ð38Þ

CO2 þH�þ� ! CO� þOH� ð39Þ

Carbon oxidation steps

C� þOH� ! CO� þH� ð40Þ

C� þO� ! CO� ð41Þ

CO� þOH� ! CO�2 þH� ð42Þ

This kinetic scheme is completed by CO, CO2 desorption and H re-combination 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 ofmethane activation and very likely the rate-determining step of the wholeprocess. They also showed that all the steps including OH* are moreprobable than those including O*: for instance CO2 would be activatedvia step 39 and not via step 38. The most abundant surface species areCO* and H*. It is shown that CO coverage is significantly higher indry reforming than in steam reforming while the reverse is expectedfor H coverage. This strengthens the hypothesis of a CO inhibitionaffecting more dry reforming than the steam reaction. However, the impactof CO2 in methane conversion remains questionnable. Donazzi et al.concluded that CO2 would not directly intervene in the kinetic of dryreforming which would be a combination of reverse water gas shiftand steam reforming.123 Xu and Saeys confirmed that carbon species wouldbe the most important intermediates in MSR on Ni(111).125 Subsurfacecarbon significantly increases the methane dissociation barrier, thusdecreasing the rate of MSR reaction, while boron promoter has thereverse 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 Rhcatalysts were tentatively extended to different metal/zirconia catalysts byJones et al.127 A complete picture of the reactivity was established at 500 1Cby combining thermodynamic and kinetic models. The reaction is structure-sensitive with turnover frequencies increasing with dispersion, suggestingthat the reaction is dominated by step and corner sites. For a given dis-persion (40%) the following trend was observed: Ru E RhWNi E Ir E PtE Pd while thermodynamic and DFT calculations led to the followingranking: RuWRhWNiW IrWPtEPd, in excellent agreement with ex-perimental results. It was confirmed that the critical steps having the highestenergy barriers on most metals were CH4 dissociation (33–36) and COformation (41). Contrary to the conclusions of the previous studies, COformation may be the rate determining step on most metals at 500 1C.However, there is clear tendency that CH4 dissociation becomes pre-dominent at higher temperatures, in agreement with the conclusions of Weiand Iglesia. A similar picture is proposed by Blaylock et al. who showedthat CH* would be the most important C-containing intermediate overNi(111) while CHO* and CHOH* cannot be excluded even though theformation of these intermediates would be strongly dependent on the re-action conditions.128 Earlier studies in the 1970–1980’s have already pro-posed formaldehyde as an intermediate in methane steam reformingaccording to the reactions:108,129

CH�x þH2Oþ x� ! CH2þxO� þ x� ! CH2O

� þ xH� ð43Þ

CH2O� ! COþH2þ� 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 notonly 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 asa support. Craciun et al. have investigated the steam reforming of methaneover ceria-supported Pd, Rh and Pt catalysts between 350 and 550 1C.130

They found that all metal supported on ceria exhibited virtually the sameactivity, 4 to 5 orders of magnitude higher than the activity of the silica-supported catalyst. A similar result was observed over Pd-CeO2-Al2O3 onwhich MSR reaction rates is two orders of magnitude higher than on Pd-Al2O3.

131 Oxygen species from the ceria support seem to play a major role inthis catalysis.

3. Steam reforming of ethanol

Alcohols are very reactive molecules whose decomposition over catalystsurfaces (or in gas phase) is much faster than with hydrocarbons. In thepresence of steam, the reaction stoichiometry of alkanol steam reforming(ASR) is:

CnH2nþ1OHþ ðn� 1ÞH2O! nCOþ 2nH2 ð45Þ

Coupled with the WGS reaction, ASR may give carbon dioxide andhydrogen:

CnH2nþ1OHþ ð2n� 1ÞH2O! nCO2 þ 3nH2 ð46Þ

The same reaction with alkanes (HSR) is:

CnH2nþ2 þ nH2O! nCOþ ð2nþ 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 reactivealcohol. It decomposes spontaneously at relatively low temperatures with-out water in the reacting gases (n=1 in 45). For this reason, methanol isconsidered as a ‘‘liquid’’ syngas, much easier to transport than the syngasitself.

The Gibbs free energy of reaction (45) per carbon atom (i.e. per mole ofCOþ 2H2 produced) is:

DG0T ¼ 1

n½DG0

CO;T � ðn� 1ÞDG0H2O;T

� DG0A;T� for alkanols ð48Þ

while that of reaction (47) will be:

DG0T ¼ 1

n½DG0

CO;T � nDG0H2O;T

� DG0HC;T� for alkanes ð49Þ

A comparison of the Gibbs free energy of the steam reforming reaction at25 1C (298K) and 427 1C (700K) is reported in Table 15 for n=1 to 4. Thesteam reforming reaction is more facile on alcohols than on correspondingalkanes. Data of Table 15 also confirmed that the thermodynamic tendencyis in the following order: C1WC2WC3WC4 for alcohols and in the re-verse order for alkanes.

However, the global reaction scheme is much more complex than thesimple reactions giving CO and H2. Replacing CO by CO2 does not change

Catalysis, 2010, 22, 1–55 | 23

the thermodynamic tendency but, as seen later, the steam reforming reactionscan also produce methane, which modifies significantly the thermodynamicequilibrium, 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 mostimportant being:

1. The steam reforming leading to CO and H2:

C2H5OHþH2O! 2COþ 4H2 DH0298 ¼ þ255 kJmol�1 ð50Þ

2. The steam reforming leading to CO2 and H2:

C2H5OHþ 3H2O! 2CO2 þ 6H2 DH0298 ¼ þ173 kJmol�1 ð51Þ

3. The hydrogenolysis to methane

C2H5OHþ 2H2 ! 2CH4 þH2O DH0298 ¼ �157 kJmol�1 ð52Þ

4. The ethanol dehydration to ethylene

C2H5OHþ ! C2H4 þH2O DH0298 ¼ þ45 kJmol�1 ð53Þ

5. The dehydrogenation to acetaldehyde

C2H5OHþ ! CH3CHOþH2 DH0298 ¼ þ68 kJmol�1 ð54Þ

6. The cracking to methane, CO and H2

C2H5OH! CH4 þ COþH2 DH0298 ¼ þ49 kJmol�1 ð55Þ

7. The cracking to methane and CO2:

C2H5OH! 1

2CO2 þ

3

2CH4 DH0

298 ¼ �74 kJmol�1 ð56Þ

8. The cracking to carbon, CO and H2:

C2H5OH! Cþ COþ 3H2 DH0298 ¼ þ124 kJmol�1 ð57Þ

9. The cracking to carbon, water and H2:

C2H5OH! 2CþH2Oþ 2H2 DH0298 ¼ �7 kJmol�1 ð58Þ

10. The cracking to carbon, methane and water:

C2H5OH! CH4 þ CþH2O DH0298 ¼ �82 kJmol�1 ð59Þ

Table 15 Gibbs free energy (kJmol� 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 DG0

700 Alkane DG0298 DG0

700

Methanol 25.21 � 69.78 Methane 142.07 37.42

Ethanol 61.00 � 47.76 Ethane 107.72 12.29

Propan-1-ol 69.40 � 26.42 Propane 99.09 3.12

Butan-1-ol 71.80 � 24.52 Butane 95.56 � 0.66

24 | Catalysis, 2010, 22, 1–55

The equilibrium composition of the gases with a water/ethanol inletstoichiometry of 1 (corresponding to reaction 50) is shown in Fig. 7. Exceptcarbon, all compounds present in equations 50–59 are included in thethermodynamic calculations but acetaldehyde and ethylene are neverfavored. Moreover, thermodynamics predicts that ethanol should be totallyconverted in the whole range of temperature. At low temperatures (100–300 1C), the cracking into methane and carbon dioxide (56) is thermo-dynamically favored. Hydrogen and CO contents progressively increasewith temperature; at 900 1C, the steam reforming to H2 and CO (50) is theonly reaction to occur with a H2-to-CO molar ratio of 2. All the reactionsleading to methane are exothermic and are thus favored at low temperature.

As for heavy hydrocarbons, methane formation can severely limit thehydrogen yield. If methane was not formed in ESR, the equilibriumcomposition would change drastically (Fig. 8A). Ethanol is then not totallyconverted below 300 1C and hydrogen is formed as of the lowest tempera-tures (below 100 1C). The reaction favored at low temperature is the steamreforming to CO2 and H2. Carbon dioxide is very briefly observed anddisappears by 350–400 1C. Above this temperature, the steam reforming toCO and H2 would be the unique reaction observed in ESR.

Finally, the formation of carbon may also change the reaction thermo-dynamics. Equilibrium compositions were calculated with possible formationof all the compounds of equations 50–59, including carbon (Fig. 8B). Carbonand methane can be formed at low temperatures with a molar ratio of 1suggesting that the reaction favored at low temperature is the ethanolcracking to methane, carbon and water (59). An interesting fact is thatmethane decreases more rapidly than in absence of carbon (compare Fig. 7

0%

10%

20%

30%

40%

50%

60%

70%

80%

100 300

Mol

e %

T (°C)

CH4

H2

H2O CO

900700500

CO2

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

Catalysis, 2010, 22, 1–55 | 25

and 8B) while the carbon itself can be produced over a large range of tem-perature. One should keep in mind that the initial state (H2O/EtOH=1)strongly favors carbon formation. Most of experimental works were per-formed with a H2O/EtOH ratio of 3 or higher. In this case, equilibrium gascomposition are closer to that of Fig. 7 even if carbon can ever be formed.

Equilibrium compositions were also calculated for different H2O/EtOHmolar ratio R and at higher pressures P. The results are reported in Table 16for two temperatures: 600 and 700 1C. H2 yield (YH) is defined as thenumber of moles of hydrogen produced per mole of ethanol entered:

YH ¼ nG �%H2wet

100ð60Þ

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 H2O H2 CO CO2 CH4 nG YH

600 1 1 Wet % 10.75 41.81 15.89 10.39 21.15 4.22 1.76

Dry % – 46.85 17.81 11.64 23.69

600 1 2 Wet % 18.29 46.15 11.62 11.71 12.22 5.62 2.59

Dry % – 56.48 14.22 14.33 14.96

600 1 5 Wet % 33.89 44.81 6.18 11.89 3.21 9.39 4.21

Dry % – 67.80 9.35 17.98 4.86

600 5 1 Wet % 18.87 25.55 7.71 14.49 33.37 3.60 0.92

Dry % – 31.49 9.51 17.86 41.12

700 1 1 Wet % 4.76 56.40 27.07 3.51 8.26 5.15 2.90

Dry % – 59.21 28.42 3.68 8.68

700 1 2 Wet % 11.59 58.19 20.93 6.40 2.88 6.62 3.85

Dry % – 65.82 23.67 7.24 3.26

700 1 5 Wet % 30.83 49.05 10.08 9.73 0.31 9.94 4.86

Dry % – 70.91 14.57 14.07 0.45

700 5 1 Wet % 12.18 40.80 18.16 8.33 20.51 4.25 1.73

Dry % – 46.47 20.68 9.49 23.36

0%

10%

20%

30%

40%

50%

60%

70%

80%

100

(A) (B)

300 500 700 900

Mol

e %

T (°C)

CO

H2

EtOHH2O

CO2 0%

10%

20%

30%

40%

50%

60%

70%

80%

100 300 500 700 900

Mol

e %

T (°C)

H2

CH4

H2O

COC

CO2

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 methaneand carbon formation

26 | Catalysis, 2010, 22, 1–55

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 beobserved, the H2 yield always increases with R and tends to 6, the maximalvalue of R corresponding to the stoichiometry of equation 51. In themeanwhile, the amount of unreacted water increases with R, which impliesthat the energy balance passes through an optimum when R is increased (thebetter H2 yield is compensated by the energy needed for gasifying a higheramount 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 barinstead of 1 bar. This is essentally due to a strong increase of the methaneyield at the expense of the CO yield. Increasing P has a moderate, positiveeffect on the CO2 yield. Of course, diluting the reactants (H2OþEtOH) inan inert gas has exactly the reverse effect: increase of YH and correlativedecrease 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 catalystperformances are artificially improved and the results should be consideredwith circumspection.

When C1 compounds are possibly formed, thermodynamics does neverfavor dehydration or dehydrogenation of ethanol. And yet, there aremany indications in the literature that these reactions on alcohols can occurat relatively low temperature on acid, basic or metal sites. To know thethermodynamic tendency of each reaction, calculations were made inabsence of C1 compounds. Dehydrogenation (ethanol/acetaldehyde/H2)was first considered, then dehydration (ethanol/ethylene/water) and finallythe two reactions together. The results are shown in Fig. 9. Dehydroge-nation (Fig. 9A) is favored at higher temperatures than is dehydration(Fig. 9B): while total conversion of ethanol can be reached by dehydrationat less than 200 1C, dehydrogenation requires almost 400 1C. It should benoted that acetaldehyde and hydrogen, on one hand and ethylene and wateron the other hand are produced in same amount: the mole percentage ofeach product is 50% at full conversion of ethanol (Fig. 9A and 9B). Whenthe two reactions are possible, dehydration is strongly favored with amaximal production of ethylene around 200 1C while that of acetaldehydeincreases only slowly with temperature (Fig. 9C).

As this will be discussed later, dehydration occurs mainly on acid siteswhile dehydrogenation is catalyzed preferably on basic or metal sites. Tocounterbalance the thermodynamic trend, it should be necessary to elim-inate virtually all the acid sites of the catalyst if one wants avoiding ethyleneformation 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 ofethanol decomposition over acid, basic and metal sites makes that numer-ous surface species can be detected according to the nature of metal andsupports.

Catalysis, 2010, 22, 1–55 | 27

3.2 Noble metals

3.2.1 Rhodium catalysts. Rhodium catalysts were early recognized asvery active in ethanol steam reforming.134,135 Cavallaro, Freni and theGroup of Messina were among the first authors to investigate ethanol steamreforming over Rh/Al2O3 catalysts.136–139 The reaction was carried out attemperatures between 50 and 650 1C with a standard H2O/EtOH ratioof 4.2–8.4136,137 with or without O2 addition for autothermal process.138,139

These Authors concluded that the mechanism starts with ethanol de-hydrogenation and/or dehydration followed by the gasification ofacetaldehyde or ethylene intermediates. Ethylene would be formedon acidic sites of alumina while all other steps (including dehydro-genation and gasification) would be catalyzed by the metal. Catalystcoking seems linked to ethylene formation and can be largely sup-pressed by O2 addition.138,139 Interestingly, it was proved that methaneis a primary product whose selectivity decreases with contact time(Table 17).

A comparison of Rh with other metal catalysts was performed byAupretre et al.,140 Breen et al.141 and Liguras et al.142 The catalysts and theconditions used in these studies are reported in Table 18. Catalysts per-formances are given in the table as H2 yield or ethanol conversion. The totalconversion of ethanol being very high in their conditions, Breen et al.

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

100%

0 200

(A) (B)

(C)

400 600 800

Mol

%

T (°C)

Ethanol

Acetaldehyde/H2

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

100%

0 200 400 600 800

Mol

%

T (°C)

EthanolEthylene/H2O

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0 200 400 600 800

Mol

%

T (°C)

Ethanol

Ethylene

Acetaldehyde

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

28 | Catalysis, 2010, 22, 1–55

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

CCOX% ¼1

2

½COþ CO2�out½Ethanol�in

� 100 ð61Þ

The performances of the alumina-supported metals can be ranked asfollows:

NiWRhcPdWPtWCuWRu, according to Aupretre et al.140

RhcPdWNiEPt, according to Breen et al.141

RhcPtWPdWRu, according to Liguras et al.142

As a rule, Rh appears as the most active metal but interestingperformance can be achieve over highly loaded Ni140 or Ru142 catalysts. Theproduct distribution and the stability of Rh catalysts can also depend on themetal precursor used for the catalyst preparation.143,144

A comparison of MgO-supported metals was also carried out by Frusteriet al. at 650 1C.145 The results reported in Table 19 confirm the very good

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

CCOX (Useful

conversion)

at 700 1C (%)

Catalyst

(% dispersion

in parenthese)

Conversion

at 750 1C (%)

1%Rh 2.3 1%Rh 90 0.5%Rh (62) 52

1%Pt 0.6 5%Ni 15 1%Rh (45) 79

0.75%Pd 1.1 0.5%Pd 30 2%Rh (50) 91

0.67%Ru 0.3 1%Pt 15 1%Ru (14) 18

9.7%Ni 3.1 3%Ru (22) 69

9.1%Cu 0.4 5%Ru (21) 70

1% Rh/CeZrOx 5.1 1%Pd (39) 25

9.7%Ni/CeZrOx 4.4 1%Pt (98) 32

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

Contact time (s)

Ethanol

conversion %

C1-selectivity (%)YH (mol H2/mol

EtOH)CO2 CO CH4

0.120 100.00 69.26 27.41 3.33 5.17

0.048 100.00 55.60 32.00 12.40 4.34

0.033 80.59 49.35 33.69 14.00 3.43

0.020 55.59 47.04 35.37 17.59 2.30

Catalysis, 2010, 22, 1–55 | 29

performance of Rh on this support. On the basis of 100 for Rh, the fol-lowing ranking was obtained:

Rh,100WPd,45WCo,27WNi,17, not so far from the relative activity ofthese metals in toluene steam dealkylation (See Table 7).

Aupretre et al.140 also showed that CeZrOx mixed oxides could beexcellent supports for Rh and Ni, increasing significantly the H2 yield.

Diagne et al. investigated the hydrogen production by ethanol reformingover 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 and500 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 repre-sentative of real conditions. At 450 1C, YH is close to 5.7mol H2/molEtOH 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 chemisorptionand the determination of the adsorption constant KCO2 at 0 1C. It wasshown that ceria was the most basic oxide (KCO2=0.7 kPa� 1) and thatKCO2 decreased linearly with the Zr/Ce ratio down to 0.25 kPa� 1 for purezirconia. Diagne et al. concluded that the hydrogen yield is not favored by ahigh basicity of support. Rh/CeZrOx catalysts were also studied by DeRogatis et al.148 and the group of Wang at Richland.149–151 In spite ofdifferent experimental conditions (R=5 and dilution in Ar for De Rogatiset al.; R=4, no dilution for Wang et al.), the conclusions of these studieswere rather similar:� While alumina or Al-spinel supports favor ethanol dehydration to

ethylene,149,151 CeZrOx supports strongly favor the acetaldehyde route toCOx and H2.

148,149 It is not clear, however, if the CeZrOx support promotesethanol 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 Oxygenstorage capacity (and mobility) on CeZrOx would play a crucial role in thereaction 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 directdecomposition of water into hydrogen (and not only into OH groups).148,151

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 con-

versions are between 7 and 20%

Catalyst

BET area

(m2 g� 1)

Metal

dispersion %

C-product distribution (vol-%)

TOF (s� 1)CO CO2 CH4 CH3CHO

3%Rh 2.1 16.0 19.5 67.1 8.1 5.3 12.1

3%Pd 2.0 12.8 31.8 32.8 21.7 14.5 5.5

21%Ni 6.3 14.0 16.5 58.2 5.7 19.6 2.1

21%Co 7.6 15.6 18.8 44.4 5.6 31.2 3.3

30 | Catalysis, 2010, 22, 1–55

This is in line with direct evidence of water dissociation into hydrogen overreduced Rh/CeO2.

45

� CeZrOx gives more stable catalysts by reinforcing their resistanceto coking.150 This is linked to the inhibition of the dehydration route toethylene which is a strong coke precursor and to the oxygen storage capacitypromoting CHx oxidation and surface cleaning along the steam reformingprocess.Aupretre et al. have studied the ethanol steam reforming at 700 1C under1 or 11bar and R=4 over 0.2%Rh/MgxNi1� xAl2O4 catalysts.152 Metalaccessibility were close to 40% for all the catalysts. Tests under 11 bar wereperformed in view of a coupling of the reformer with a separation mem-brane for H2 purification. The acid-base properties of the supports werecarefully characterized by CO2 chemisorption and dimethyl-3,3-but-1-eneisomerization as well as FTIR of lutidine and DRIFT. The morphology ofthe 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), canexplain the neutralization of most acidic sites. Some of them are createdduring the metal precursor impregnation. The highest H2 yield at 1 bar wasobtained with the Rh/Mg0.75Ni0.25A2O4 catalyst (4.68mol H2/mol EtOHand 5.58mol/mol with WGS) while the best performances under 11 barwere obtained with the nickel-free catalyst (2.74mol H2/mol EtOH and 3.25with WGS). Acid-base properties of alumina can be tailored by addingsome oxide promoters. Recently, Can et al. have investigated the effect ofSc, 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 themost basic catalysts presenting a new type of acid site less able to form cokeprecursor.

Most of the previous studies report that ethanol conversion in gas phaseis far to be negligible and it seems that several products of gas phase con-version are coke precursors. For instance, Aupretre et al. showed that asignificant amount of ethylene was produced when the reactor was loadedwith SiC only (the ‘‘inert’’ material used as diluant of the catalyst).152 At700 1C, 1 bar, ethanol conversion was 71% while it increased with thepressure to reach 98% at 11 bar (residence time effect). The formation ofundesirable by-products in gas phase can be avoided by performing thereaction of steam reforming or autothermal reforming in ultra-short resi-dence time reactors (o10 ms). Deluga et al. reported very interesting resultson this type of reactor with a catayst consisting in a Rh/CeO2 layer de-posited on a ceramic foam.154

Surface species and gas phase products in the steam reforming of ethanolon TiO2 and Rh/TiO2 were studied by Rasko et al. by coupling FTIR andTPD-MS.155 Ethanol dissociation forms ethoxides C2H5O

� at ambienttemperature. Dehydrogenation leads to acetaldehyde whose formation isaccelerated in the presence of rhodium and steam. The same study wasextended to other metals and to other supports (alumina, ceria, . . .).156 Atlow temperature, ethanol is molecularly adsorbed on oxides while it formsethoxides and proton on supported noble metals. Heating these adsorbedspecies 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 theymay deactivate the catalyst. Dehydration to ethylene is observed only onalumina and virtually not on ceria. Surface species issued from ethanoladsorption at room temperature and their decomposition by HT-treatmentwere investigated by Idriss over ceria-supported Rh, Pd, Pt and Aumonometallics as well as over Pt-Rh, Rh-Au, Rh-Pd and Pt-Pd bime-tallics.157 O–H bond dissociation leading to ethoxy species can be observedon Rh, Pt and Pd. In the meanwhile, and only on Rh, a C–H bond dis-sociation of the CH3 group leading to oxometallate species is also observed.As a result, Pd and Pt may produce significant amounts of acetaldehydewhile the C–C bond dissociation, leading to CO and methane, is favoredon Rh.

From all the studies carried out over Rh catalysts, a global picture of thereaction mechanism as illustrated in Fig. 10 can be proposed.

In this mechanism, methane is formed by acetaldehyde (or ethanol)cracking and is then steam reformed in a further step. It would be a primaryproduct of ESR. To support this mechanism, it may be remarked thatthe actual methane concentration in gas phase is sometimes higher thanthe equilibrium concentration. Another route to methane, not evoked inequations 50–58 is the hydrogenation of CO or CO2. This way of methaneformation was investigated by Birot et al.158 over Rh/CeZrOx ESR cata-lysts. They showed that methane can be produced by CO hydrogenation butnot by CO2 hydrogenation whose rate is negligible in ESR conditions.

3.2.2 Other metals. Though group 10 noble metals were compared torhodium in many studies (y 3.2.1), they were sometimes investigated alone,generally with the objective to prove their efficiency when deposited onadequate supports.

Platinum. One of the first study on ethanol steam reforming was reportedby Rampe et al.159 Catalysts were Pt, Ru, Pd and Ni or bimetallics Ni-Ptand 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 thisstudy was to demonstrate the feasibility of an ethanol reformer of 3kW.

CH3CH2OH

Metal

Support

H2O

OH OH

Basic siteAcid site

CH3CHO H2, CO, CO2, CH4CH3CH2OHC2H4

Fig. 10 Mechanism of ethanol steam reforming. Ethanol is either dehydrogenated into acet-aldehyde over basic (or metal site) or dehydrated into ethylene over acid sites. Acetaldehyde isthen decomposed in COþCH4þCO2. Methane can be reformed at elevated temperature whilethe CO/CO2 mixture is shifted to more or less CO2 according to the temperature and the Rratio. Ethylene can be directly reformed into syngas or hydrogenolyzed into methane. This isalso a coke precursor whose formation is preferably to be avoided.

32 | Catalysis, 2010, 22, 1–55

Jacobs et al. have investigated the Pt/CeO2 system in ethanol steamreforming at 200–550 1C in diluted medium (H2O/EtOH/N2þH2/=33/1/29).160 Adding hydrogen to the inlet feedstream allows to maintainceria in a reduced state. Jacobs et al. showed that ceria played a significantrole in the reaction mechanism: ethanol would be adsorbed on defect sites ofceria forming ethoxy species which, in turn, are transformed into acetatespecies. One of the role of the metal would be to hydrogenate surface speciesinto carbon dioxide and methane. In this mechanism, CO2 is the primaryCOx 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 studieswere performed by the Group of Davis at Lexington and the Group ofNoronha at Rio de Janeiro over Pt/CeZrOx catalysts.161,162 A wide range ofby-products were identified. The catalyst is mainly deactivated by carbonwhich creates a barrier at the metal/support interface, thus inhibitingacetate decomposition and hydrogenation. Co-feeding oxygen with waterstrongly reduces coke formation and increases catalyst stability at theexpenses of H2 yield.

162 Platinum was also used to improve Ni stability bydecreasing coke formation in Ni-Pt bimetallic catalysts.163

Ruthenium. Ru/alumina catalysts (in the form of pellets or wash-coatedon cordierite monolith or on ceramic foams) were investigated by Liguraset al.164 On such structured catalysts, the control of temperature alongthe 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 auto-thermal reforming of ethanol. Contrary to the conclusions of the previousstudies on Pt, CO would be a primary product of the reaction and sub-sequently transformed into CO2 and H2 by WGS. As Ru is a very goodcatalyst for this reaction, CO and CO2 concentration would be close toequilibrium.

Palladium. Ethanol steam reforming on Pd catalysts was studied byGoula et al. over Pd/Al2O3

165 and by Casanovas et al. over Pd/ZnO.166 Atlow temperature, Goula et al.165 observe an important formation of CO andmethane corresponding to ethanol (or acetaldehyde) decomposition. In-creasing the temperature favors the formation of CO2 and H2. Above500 1C, the formation of CO increases again in accordance with the WGSequilibrium. Casanovas et al. have tested their 5%Pd/ZnO catalyst between250 and 4501C with a H2O/EtOH ratio of 13.166 They report very goodactivity at medium temperature (300–400 1C) with, however, a significantformation of acetaldehyde. The major problem they encountered was theformation of PdZn alloy at the highest temperature. A two-layer fixed bedreactor was proposed by Galvita et al.167 for ethanol reforming. Ethanol isfirst transformed by dehydrogenation and cracking on a Pd/C catalyst in agas mixture composed of methane, COx and hydrogen (see equations 54–56, section 3.1). Methane is then reformed at high temperature to carbonoxides and hydrogen on a conventional Ni catalyst.

Iridium. Ethanol steam reforming was investigated over Ir/CeO2 byZhang 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 hin ESR or autothermal ESR. It seems that the combination of Ir metaland 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 propertiesof the Ir/CeO2 system and the exceptional stability of Ir in theseconditions. These results are coherent with the OSC and 18O/16O exchangeproperties measured on similar Ir/CeO2 catalysts by Bedrane et al.55 As onmost ceria-supported catalysts, Ir/CeO2 produces a significant amount ofacetaldehyde at low temperature (below 450 1C). In addition, acetone wasalso observed up to 500 1C by secondary reaction of acetaldehyde withwater (62):

2CH3CHOþH2O! CH3COCH3 þ CO2 þ 2H2 ð62Þ

Ethanol conversion is total above 4501C and the product distribution isclose 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 equilibriumvalues: 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 in-dustry for many chemical reactions. Because of its high performances andits low cost, nickel is also one of the most studied metals for ethanol steamreforming. Nickel-containing catalysts are reported to have a high activityfor ethanol steam reforming. Fatsikostas and Verykios171 studied theethanol steam reforming over nickel catalysts supported on g-Al2O3, La2O3,and La2O3/g-Al2O3. They demonstrated that Ni promoted reforming ofethanol and acetaldehyde as well as the water–gas shift and methanationreactions. At temperature below 300 1C, pure nickel causes bond breakingof ethanol in the following order: O–H, –CH2–, –C–C–and �CH3, thekey reaction being ethanol dehydrogenation. At elevated temperature, theoverall reaction network of ethanol steam reforming results in the for-mation of H2, CO, CO2, and traces of CH4. Nickel is also known to presenta high hydrogenation activity and it may help to combine hydrogen atomson the catalyst surface to form molecular hydrogen.135,172 However, nickelpresents a limited WGS activity, limiting the selectivity towards H2 andCO2. Consequently, the support will also play a role in the catalytic per-formances. The ethanol conversion as well as the selectivity to hydrogen ofnickel catalysts on various support reported in the literature for tempera-tures higher than 550 1C are given in Table 20. The results presented inTable 20 show that whatever the experimental conditions and the nickelcontent, the majority of the supports lead to high selectivity to hydrogen.

In fact, one of the major problem to overcome with nickel catalysts is toavoid the catalyst deactivation due to metal particle sintering as well as tocoke deposition. To improve the stability of the nickel catalyst, several wayswere investigated by modifying either the nature of the support or themetallic phase. Thus, various supports were studied, such as Al2O3, SiO2,MgO, MgAl2O4, La2O3, ZnO, CeO2, CeO2–ZrO2, CexTi1� xO2 or per-ovskite-type oxides (LaAlO3, SrTiO3 and BaTiO3).

135,140,173–180,188,189 Thebetter performances were obtained with the most basic supports favoringthe ethanol dehydrogenation and inhibiting ethanol dehydration leading toethylene, which is coke precursor. The addition of dopants to the support

34 | Catalysis, 2010, 22, 1–55

can also improve the activity, the selectivity and the stability of the catalysts.For example, alkali are very often added to the support to neutralizethe acid sites, inhibiting in this way the dehydration reaction leading to theformation of olefins responsible for coke production.181–184 It was alsoshown that the addition of potassium to the support improves the stabilityof Ni/MgO catalysts by limiting the sintering of nickel particles without anyeffect 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 activemetal for the water-gas-shift reaction thus increasing the hydrogen yield,and to relieve the carbon deposition. Moreover, iron has electronic prop-erties and a ion radius similar to that of nickel and then it can easily replacenickel in nickel-containing phase. The iron-promoted nickel-based catalystsshowed high selectivity to hydrogen as well as high resitance to both sin-tering and oxidation in oxidative atmosphere. The improved durability,compared to the iron-free nickel catalyst was attributed to the presence ofNiAl2O4-FeAl2O4 mixed crystals.193

Copper-based catalysts are known to favor the ethanol dehydrogenationroute and copper is also more active in methane steam reforming. Associ-ated to nickel, copper increases the water-gas shift activity as well as thestability of nickel particles.194,195 Biswas and Kunzru evidenced178 thatthe addition of copper to Ni/CeO2-ZrO2 catalyst not only increased thesteam reforming and WGS activities but also enhanced acetaldehydedecomposition and reforming. Klouz et al.190 and Fierro et al.191 also shownthat the addition of Cu to Ni-based catalysts enhanced the selectivity

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

conversion (%)

Selectivity

to H2 (%) Ref.

10 g-Al2O3 4 – 600 1C 100 75 185

15 g-Al2O3 6 – 600 1C 100 87 186

17 g-Al2O3 3 He, 62.5

vol.%

750 1C 100 93 187

17 La2O3 3 – 600 1C 93 87 188

17 La2O3 3 – 700 1C 100 95 188

17 La2O3 3 He, 62.5

vol.%

750 1C 100 90 187

15 La2O3-Al2O3 6 – 600 1C 100 87 186

10 TiO2 4 – 600 1C 100 86 185

10 Ce0.5Ti0.5O2 3 N2, 80

vol.%

600 1C 100 58 189

30 Ce0.74Zr0.26O2 8 N2 600 1C 98 88 178

30 Ce0.74Zr0.26O2 8 N2 650 1C 100 93 178

10 ZnO 4 – 600 1C 95 80 185

17 MgO 3 He, 62.5

vol.%

750 1C 100 79 187

17 YSZ 3 He, 62.5

vol.%

750 1C 100 92 187

Catalysis, 2010, 22, 1–55 | 35

towards hydrogen and improved its stability. This was related to the abilityof copper of modifying the affinity of nickel particles for carbon, thus in-hibiting coke formation. The interaction between the Cu–Ni phase and thesupport plays also an important role in the complex reaction network takingplace during the ethanol steam reforming. Calles et al.,196 prepared Cu-Nibimetallic catalysts on Ce and La modified SBA15 support. They observedthat on the Ce-modified support, the formation of large CeO2 particlesdecreased the metal-support interaction thus increasing the metal particlesize, leading to a low hydrogen selectivity. On the contrary, a high metal-support interaction, as observed with high contents of La2O3 in the modi-fied support, leads to a better metallic dispersion as well as a lower cokeformation.

Another way of improving the stability of Ni-based catalysts consistsof adding small amount of noble metals such as platinum197,198 or pal-ladium198 The addition of promoters caused a decrease in the NiO reductiontemperature. Moreover, the bimetallic catalysts showed a higher ethanolconversion and higher hydrogen yield than the monometallic one, whateverthe nature and concentration of the noble metal. Rhodium was also addedto 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 thesupport. The reduction of the Rh–Ni/CeZr mixed oxide during the catalytictest decreased the nickel-support interaction, thus favoring the carbonfilament production and diminishing the oxidizing properties of thesupport.

3.3.2 Cobalt-based catalysts. Cobalt has also been the object of manypapers on ethanol steam reforming.179,200–209 It was demonstrated that thesupport is of major importance for the catalytic performances in ethanolsteam reforming in the presence of Co-supported catalysts. Among thesupports studied, the alumina support was reported to yield the highestselectivity towards hydrogen200,207 but the basic supports, such as MgO leadto more stable catalyst, resistant to coke formation.207 The metal-oxideinteraction appeared to be also essential for the stability.201 Thus, to favorCo-support interactions, fluorite-type Ce-Zr-Co oxide was prepared andevaluated in real bioethanol steam reforming at 540 1C. At this temperature,the catalyst deactivated, due to the formation of carbonaceous products atthe oxide surface and also of filaments. This result was explained by thereduction of cobalt cations of the mixed oxide, weakening the metal supportinteraction, thus favoring carbon deposition.210 The addition of Rh to thisCeZrCo catalyst increased the catalyst stability by avoiding the accumu-lation of carbon species on the surface.211

As recently developped by Urasaki et al.,179 the Co catalysts supported onperovskites such as SrTiO3 and LaAlO3 showed higher catalytic activitiesand much longer–term stabilities compared to conventional Co/MgO andCo/gAl2O3 catalysts. It was inferred that the lattice oxygen in perovskitesplayed a positive role in inhibiting the coke.

3.3.3 Copper-based catalysts. Copper is a good dehydrogenation cata-lyst212 and is known to present a highWGS 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 ethanolsteam reforming, yielding CO, CO2 and H2 as the main products at lowtemperature, above 360 1C.213 Recently,214 nickel, cobalt and/or manganesewere added to a commercial Cu/ZnO/Al2O3 and the resulting catalysts weretested in ethanol steam reforming. It was demonstrated that, below 480 1C,the presence of modifiers leads to a decrease in the methane formation and anincrease in the hydrogen yield and selectivity. At this temperature, the for-mation 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 introductionof the combustion engine. From 1935 to 1945, ethanol was used pure, ormixed with petrol before to be replaced by gasoline or diesel with the de-velopment of the petrochemical industry. The crisis in the 1970s revived theinterest of ethanol with the running of the Brazilian National AlcoholProduction Progam (PROALCOOL), in 1975. However except for thisdecade the price of the oil remained low making the use of other alternativeenergy resource hardly acceptable. It was necessary to wait for the begin-ning of the 21st century with the realization of two simultaneous concerns,the decreasing of the resources of the fossil fuels and the global warming toreally consider that an alternative energy became essential. One way ofreducing environmental effects and the dependance on fossil fuels is to userenewable bioethanol as direct fuel for transportation. Another way is touse hydrogen as an energy carrier to support sustainable energy develop-ment. Hydrogen can be used in a fuel cell to generate electricity with highefficiency. It is extremely clean as the only by-product is water. In order tosupport sustainable hydrogen economy, it is crucial to produce ethanol andhydrogen cleanly and renewably.

4.1 Bioethanol production and composition

Bioethanol is an ethanol produced renewably by fermentation by yeast ofbiomass materials, such as sugar cane, sugar beet, wheat, potatoes, corns,and other sugar- and starch-rich plants.215 Such crops which have a fooduse are often referred as ‘‘first generation’’ biomass crops. Sugar and starchcrops have been utilized for bioethanol production due to the relative easeat which their constituent reducing sugar unit can be separated in water(hydrolysis) and subsequently fermented. It should be noted that that fer-mentation is not a ‘‘green’’ process since one C6 unit of sugar gives only twomoles of ethanol, two carbons being lost as CO2. Brazil and United Statesproduce large quantities of bioethanol from sugar cane and corn respect-ively. In the European Union, where France and Spain are the largestproducers, the ethanol production is more modest and comes from thefermentation of sugar beet or wheat in a large part. As ethanol is producedfrom 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 dras-tically. Table 21 gives a comparison of the performances obtained in termsof ethanol yields from main ‘‘first generation’’ feedstocks.

Catalysis, 2010, 22, 1–55 | 37

These data corresponding to the cultivation efficiency have a direct im-pact on the climate benefits. Other factors influence the GHG emissions : (i)the fuel and fertilizer used in the ethanol plant; (ii) the efficiency with whichby-product are dealt with; (iii) the type of land for cultivation.218 Accordingto the raw material, to the agricultural yield and to the transformationprocess efficiency, the energy balance (energy used/energy produced) hasbeen 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 ofGHG saving cannot be avoided. Different studies can bring to entirelyopposite conclusions. For Searchinger et al. who assume that a productionof biofuel requires new cultivation of cropland by ‘‘displacement effect’’, theproduction of bioethanol is considered as a threat for the environment.219

From other authors it seems clear that, under most production scenari, thenet greenhouse gas effect of bioethanol is positive.220 This last conclusion isusually accepted even if the data calculted in term of reduction of GHGemissions vary. ADEME reported a detailed comparison of the two firstworks reported in the Table 22 in order to analyse the reasons of thedivergences. One of the conclusions was that as far as normalized results areconcerned the values become comparable.

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 andsugarcane, are used to produce food and sugar as consumer goods. Therecent rise in the price of various crops such as the wheat has been attributedsometimes to the greatly increased demand for biofuels in recent years bymany developed nations. The use of ‘‘second-generation’’ lignocellulosicbiomass, emerged to be the best option available currently. This second-generation bioethanol only requires inexpensive cellulosic biomass as

Table 21 Some examples of ethanol yields from

various crops (from 216 and 217)

Crop Ethanol yield (hl/ha)

Sugar cane 70

Sugar beet 21.7–47.4

Wheat 11.6–20.1

Corn 18.2–32.8

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 37–44 � 6 to þ 43 222

ADEME 75 75 223

PWC 40–60 40–70 224

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 itsmore complex molecular structures, the production of ethanol from lig-nocellulosics is more difficult than from sugar cane or stach-rich materials.Lignocellulose consists of three main components: cellulose, hemicelluloseand lignin, the first two being composed of chains of sugar molecules. Thepercentage of each component differ with the nature of the feedstock(Table 23). The difference between first- and second-generation bioethanolproduction is that, in the latter case, an extra step is required to hydrolyselignocellulose biomass.226,227

The final step, the distillation process, remains the same for both gener-ations of bioethanol. Important factors to consider when evaluating the useof bioethanol are fuel or feedstock is the overall economy of processes in-volved as well as the energy-efficency from biomass to wheel. Fuel gradebioethanol needs to be water-free, thus the production requires distillationbeyond the azeotrope point, and this is one of the major production costs offuel-ethanol.229 On the contrary, ESR process requires, as mentioned pre-viously, a significant amount of water (50, 51), which makes the expensivedistillation superfluous. Only a simple flash distillation to some 50–70% isnecessary thereby considerably lowering the production costs of bioethanol.However the use of the diverse feedstocks as well as the flash distillation canpose different challenges in a subsequent ESR catalyst, due to the variety ofcontaminants present in the different bioethanol fractions. The compositionof two qualities of bioethanol produced in France from sugar beet is pre-sented in Table 24. In the crude bioethanol fraction, the main impurities arealcohols 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 andnitrogen-containing bases. The following paragraph of this review will re-port a study of the effect of impurities present in raw bioethanol, such asesters, aldehydes, higher alcohols or acetic acid, on the stability of theseveral catalysts during bioethanol steam reforming.

As far as ‘‘first generation’’ bioethanol is concerned, the major part of thecontaminants will be always heavy alcohols with a variation in the com-position that depends on the crop (Table 25).

To our knowledge, only one group reported the composition of diversefractions (Table 26) obtained from a ‘‘second generation’’ bioethanol andinvestigated 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

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

Biomass Lignin (%) Cellulose (%) Hemicellulose (%)

Soft wood 27–30 35–42 20–30

Hard wood 20–25 40–50 20–25

Wheat straw 15–20 30–43 20–27

Catalysis, 2010, 22, 1–55 | 39

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

4.2 Crude bioethanol steam reforming

Until now crude bioethanol has been very rarely used as ethanol source forESR reaction, whereas it is a solution to produce hydrogen in a cost-effective manner. The majority of the studies reported in the literature onthe production of hydrogen by ethanol steam reforming were performed

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

From ref. 230

Constituents

Crops

Sugar beet Cereal Potatoes

Methanol 5.1 – –

n-propan-1-ol 31.7 9.1 16.4

Sec-butan-1-ol – – –

Isobutan-1-ol 16.6 19.2 15.9

n-butan-1-ol – 0.2 1.2

2-methylbutan-1-ol 14.9 19.0 13.6

3-methylbutan-1-ol 31.7 52.4 52.9

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

Total acidity (Acetic acid) gm� 3 0.8 0.4

Dry extract gm� 3 20 39

Esters gm� 3 o1 123

Aldehydes gm� 3 o1 108

Methanol gm� 3 42.5 94

Butan-2-ol gm� 3 o0.5 o10

Propan-1-ol gm� 3 o0.5 581

Methyl-2 Propanol-1 gm3 o1 304

Propen-2 ol-1 gm� 3 n.d. o10

Butan-1-ol gm� 3 o0.5 o10

Methyl-2 butanol-1 gm� 3 n.d. 273

Total higher alcohols gm� 3 n.d. 582

Total sulphur gm� 3 o0.5 855

Volatile nitrogenated bases gm� 3 o1 1746

40 | Catalysis, 2010, 22, 1–55

using pure ethanol. The steam reforming of crude ethanol differs from thatof pure ethanol by the fact that the impurities present in the crude ethanolfeed may influence the hydrogen yield and the catalyst stability. Two typesof studies on the effect of the use of crude bioethanol have been performedby either using model ethanol and water mixtures containing one impurityas identified in crude bioethanol or using directly a real crude bioethanolfeed.

Very few studies report the use of crude ethanol for hydrogen productionby steam reforming. Akande et al.233 used directly the mixture provided byPound Maker Agventures (Canada) and obtained by fermentation of highstarch 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 steamreforming was directly performed on the crude ethanol sample at atmos-pheric pressure and 400 1C, at a rather low weight hourly space velocity(WHSV) of 1.68 h� 1. Ni/Al2O3 catalysts containing 10 to 25wt.% ofNi and prepared by various techniques (coprecipitation, precipitation andimpregnation) were used for the experiments. Whatever the nickel contentand the preparation method, the initial ethanol conversion was high, be-tween roughly 55 and 95%, and then decreased with time on stream (TOS)to stabilize at about 200 min. The same catalyst behaviour was obtainedduring the steam reforming of crude bioethanol from sugar beet, in thepresence of a Rh/MgAl2O4 catalyst.

234 In both cases, the deactivation ob-served at the onset of the reaction was attributed to coking.

In another study, Vargas et al.235 observed also a deactivation, but onlyafter 11h of time-on-stream. They used a real bioethanol obtained by sugarcane molasses fermentation followed a simple distillation at 60 1C.

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

From ref. 231

Fraction 1 Vol% Fraction 2 Mol% Fraction 3 Mol%

Major

constituents

Ethanol 5.7 Ethanol 18 Ethanol 44

Xylose 1.2 3-Methyl-1-

butanol

0.7 Ethyl acetate 0.5

Glucose 1.1 2-Methyl-1-

propanol

0.3 1,1-Diethoxyethane 0.2

Lactate 0.5 2-Methyl-1-

butanol

0.3

Acetate 0.5 Propanol 0.1

Glycerol 0.4 Cyclopentanone 0.05

Other

constituents

(traces)

Ethyl acetate Furfural Propanol

Acetic acid 4-Hexen-1-ol 2-Methyl-1-

propanol

1,1-Diethoxyethane 3-Methyl-1-

butanol

2-Methyl-1-

butanol

Cyclopentanone

Catalysis, 2010, 22, 1–55 | 41

The solution was mainly made of water and ethanol (ethanol/water=1/5.95mol/mol) and contained only traces of methanol, n-propanol,n-butanol and 3-methylbutanol (isoamylic alcohol) as impurities. The re-action was also performed at atmospheric pressure, 540 1C and the realbioethanol mixture was diluted in a Ar/N2 gas mixture. The catalyst was aCe-Zr-Co fluorite-type oxide. The results obtained showed that the catalyticbehavior is similar when using a water/pure ethanol mixture or a realbioethanol feed. The only difference concerned the slightly higher hydrogenyields obtained from bioethanol due to the steam reforming of the higheralcohols. Nevertheless, whatever the origin of the ethanol used for thereaction, the ethanol conversion was 100% at the onset of the reaction,remained stable during the first 11h of TOS and then declined continuouslyto reach roughly 45–50% after 25 h TOS. Consequently, the hydrogen yield,which was higher than 0.3 gH2 h

� 1 gcat� 1 at the beginning of the reaction,

falled down 0.1H2 h� 1 gcat

� 1 after 25 h. The apparent stability observedduring the first hours of TOS may be explained by the large amount ofcatalyst used for this reaction, in excess compared to the one needed toreach the complete ethanol conversion. Possibly during the first hours ofTOS a part of the catalyst did not work. In this study, the deactivation wasalso explained by the authors by the formation of carbon during thereaction.

More recently, Rass-Hansen et al.231 used a technical bioethanol pro-duced from wheat straw. Two fractions of technical alcohol were testedin ethanol steam reforming. These fractions were obtained from differentparts of the distillation process. They contained contaminants such asethyl acetate , 1,1-diethoxyethane, higher alcohols and cyclopentanone, thepercentage of each type of impurity depending on the fraction considered.The catalysts used for this study were Ni and Ru catalysts supported onMgAl2O4 spinel. The authors observed that whatever the reaction tem-perature and the catalyst, the rate of carbon formation is generally slightlyhigher 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 contributeto the fast deactivation of catalyst.

In conclusion, whatever the origin of the bioethanol, a deactivation of thecatalyst was observed during the steam reforming reaction that was at-tributed 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 for87% of the impurities, the most important being propan-1-ol (27%) andmethyl-3 butanol-1 (27%), and also esters, aldehydes, acetic acid andnitrogen-containing bases.236 In order to determine which type of impurityis responsible for the deactivation observed during the steam reforming ofcrude bioethanol, the impact of various impurities on this reaction wasstudied. For that purpose, ‘‘model’’ raw ethanol feeds were prepared byaddition 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 dif-ferent functions (butanal, diethylether, butanol, ethylacetate) (ii) moleculeswith acidic and basic properties (acetic acid and diethylamine) and (iii)linear or branched alcohols (methanol, propan-1-ol, butan-1-ol, pentan-1-ol, isopropanol, 2-methylpropan-1-ol, 3-methylbutan-1-ol). The resultsobtained in the presence or not of various impurities after 8h of time onstream on a 1wt.%Rh/MgAl2O4 catalyst with a molar water to ethanolratio R of 4, a weight hourly space velocity of 19.5 h� 1 are summarized inTable 27. These reaction conditions were chosen in order to have less than100% of ethanol conversion (XEtOH %) to discriminate the possible effectsof the compounds present in bioethanol on the catalyst stability.

4.3.1 Acid and basic impurities. Table 27 shows that the presence ofdiethylamine favors the ethanol conversion, and slightly increases thehydrogen yield. This promoting effect of diethylamine has been explainedby a competition of this basic molecule with the alcohol molecules for theacidic sites.236 Diethylamine being adsorbed preferentially on the acidic sitesof the support, the dehydration of ethanol on these sites is thus inhibited.The promoting effect of diethylamine on ethanol conversion was explainedby a modification of the metal electronic properties resulting from anelectron transfer of the free nitrogen doublet toward the metal.236

Contrary to what is observed with diethylamine, the presence of aceticacid leads to a decrease of both ethanol conversion and hydrogen yieldcompared to what is obtained without impurity. Acetic acid may promoteethylene formation by increasing the acidity of the support surface and thenfavors the catalyst deactivation by coke deposition. The yield in ethyleneobserved in the presence of acetic acid is similar to that obtained in theabsence of this impurity, but it is at a lower ethanol conversion, which

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)

Impurity

Distribution of products (mol molEtOH� 1)

XEtOH

CokeaH2 CO CO2 CH4 C2H4 C2H6 CH3CHO H2O (%)

– 2.35 0.57 0.34 0.3 0.07 0.04 0.03 3.5 78 29.5

Diethylamine 2.53 0.67 0.37 0.38 0.06 0.04 0.02 3.45 88 64.0

Acetic acid 2.08 0.49 0.31 0.26 0.07 0.04 0.04 3.6 72 68.0

Butanal 2.6 0.67 0.39 0.35 0.06 0.04 0.02 3.41 86 93.3

Butanol 1.42 0.35 0.19 0.26 0.09 0.04 0.04 3.95 64 77.4

Ethyl acetate 0.97 0.31 0.09 0.29 0.13 0.04 0.05 4.06 57 67.0

MeOH 2.48 0.57 0.37 0.25 0.06 0.04 0.03 3.06 78 74.0

n-C3H7OH 1.52 0.47 0.19 0.38 0.12 0.05 0.04 3.47 74 80.0

n-C4H9OH 1.42 0.35 0.19 0.26 0.09 0.04 0.04 3.63 65 77.4

n-C5H11OH 1.23 0.33 0.14 0.29 0.11 0.04 0.05 3.46 57 62.0

i-C3H7OH 1.22 0.38 0.13 0.33 0.13 0.04 0.04 3.76 70 101.0

i-C4H9OH 1.05 0.31 0.10 0.29 0.11 0.04 0.05 3.61 56 62.5

i-C5H11OH 0.89 0.28 0.09 0.25 0.10 0.04 0.05 3.78 53 75.0

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

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 beexplained by the formation of intermediate acetate species that decomposeinto H2, CO2 and adsorbed carbon on the surface. Then, whateverthe intermediate species invoked (ethylene or acetate) the deactivation of thecatalyst observed in the presence of acetic acid was explained by the for-mation of coke, favored in the presence of this acidic molecule.236

4.3.2 Various types of impurities with four carbon atoms. Various typeof impurities identified in raw bioethanol were studied (aldehyde, higheralcohol 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 thepresence of butanal, the ethanol conversion and the hydrogen yield areincreased compared to the reference test (without impurity), as it was ob-served in the presence of diethylamine. On the contrary, the presence ofbutanol and ethylacetate strongly deactivates the catalyst, but the mostdeactivating impurity is diethylether since the ethanol conversion is of 64and 57% in the presence of the alcohol and the ester, respectively. In thepresence of these impurities, the yields of final products (H2, CO, CO2, CH4)are strongly decreased, whereas the yields of intermediate products, espe-cially ethylene and acetaldehyde are higher than that of the reference test.Then, it can be inferred that intermediate products are rapidly produced byethanol dehydration or dehydrogenation on the catalyst but they are moreslowly transformed. The presence of high amounts of ethylene may explainthe catalyst deactivation by formation of carbonaceous products. Thedeactivation observed in the presence of butanol may also be linked to theproduction of butene, which is also a coke precursor, by butanol de-hydration. The deactivation by the ester may be due to the hydrolysis ofethylacetate on the acidic sites of the support, yielding ethanol and aceticacid, and then to the presence of acetic acid. But, the deactivation observedis much more important than that observed in the presence of acetic acid.The deactivation may be also explained by a competitive adsorption, sinceit has been reported in the literature that ethylacetate is more stronglyadsorbed on alumina than ethanol.237

4.3.3 Effect of the alcohols. In the presence of methanol (Table 27), theethanol conversion is not modified but the hydrogen yield is slightly in-creased, compared to the reference test. Methanol is easily converted bysteam reforming, thus producing also hydrogen, and then its presence doesnot affect the ethanol conversion. In the presence of the higher alcohols(more than three carbon atoms), linear or branched, the ethanol conversionand the hydrogen yield decrease when the amount of carbon atoms in themolecule is increased. This effect is more pronounced in the presenceof branched alcohols compared to the linear ones. It can be seen fromthe value presented in Table 27 that the yield in ethylene is much moreimportant in the presence of these alcohols than with ethanol alone, thusleading to a more important amount of coke. It has been also demonstratedby studying the steam reforming of these higher alcohols234 that they aredehydrated to the corresponding olefin. The olefins may be then polymer-ized 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 branchedalcohols may be explained by the formation of more stable carbocationsthus facilitating the olefin production.210

In conclusion, except methanol, diethylamine and butanal, all the im-purities identified in crude bioethanol from sugar beet lead to the de-activation 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 catalystdeactivation observed during the steam reforming of crude bioethanol isdue to coke formation,233–235 it is of major importance to reconsider thecatalyst formulation, by modifying the support and then metallic phase, inorder to find a stable catalyst able to convert crude bioethanol by steamreforming.

In order to improve the catalyst stability in the presence of crude bioe-thanol, various rare earth elements (Sc, Y, La, Er and Gd) were added to analumina support. Indeed, these species are known to improve the catalyticperformance in steam reforming and the stability of the catalysts, by de-creasing the support acidity, thus disfavoring the olefin formation, and byincreasing its basicity, necessary to activate water. These supports were usedfor depositing 1wt.% of rhodium. The characterization of this catalyst serieshas been described in detail by Can et al.153 Table 28 reports the resultsobtained during the pure ethanol steam reforming in the presence of thesecatalysts, compared to that of a Rh/MgAl2O4 one. All modified samplesclearly show higher ethanol conversions (XEtOH), hydrogen yields andlower intermediate product yields, especially ethylene, compared to theRh/MgAl2O4 catalyst. The highest performances were obtained with theRh/Y2O3-Al2O3 (Table 28).

Nevertheless, whatever the catalyst, the methane yield is much higherthan the value at the thermodynamic equilibrium. This result may be ex-plained if methane is considered as an intermediate product. Then, a wayto 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 secondmetal.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)

Catalyst Impurity

Yield (molmol� 1)

XEtOH (%)H2 CO CO2 CH4 C2H4 C2H6 CH3CHO H2O

Thermodynamic

equilibrium

3.77 0.54 0.71 0.10 0.00 0.00 0.00 3.03 –

Rh/MgAl2O4 2.35 0.57 0.34 0.3 0.07 0.04 0.03 3.50 78.0

Rh/Sc2O3-Al2O3 3.30 0.71 0.63 0.53 0.01 0.02 0.01 2.56 99.4

Rh/Y2O3-Al2O3 3.43 0.67 0.65 0.50 0.01 0.01 0.01 2.50 98.7

Rh/La2O3-Al2O3 3.34 0.62 0.69 0.54 0.01 0.01 0.01 2.50 98.6

Rh/Er2O3-Al2O3 3.09 0.69 0.59 0.57 0.02 0.02 0.01 2.64 98.7

Rh/Gd2O3-Al2O3 3.29 0.67 0.66 0.54 0.01 0.02 0.00 2.54 98.7

Catalysis, 2010, 22, 1–55 | 45

coimpregnated with the Rh precursor onto the Y2O3-Al2O3 support.The results presented in Table 29 show that the presence of nickel or pal-ladium allows one to increase the hydrogen yield, whereas platinum has nobeneficial effect. Nevertheless, the promoting effect of Ni is more pro-nounced than that of palladium. The better performances of the Rh-Nicatalysts may be explained either by a better water-gas shift activity or bya better methane steam reforming activity, both reactions leading to theformation of hydrogen. The yield in CO is slightly increased in the presenceof Rh-Pd and Rh-Ni catalysts compared to that obtained in the presence ofRh (from 0.7 to 0.79 and 0.75 respectively), but the methane yield is muchmore decreased (from 0.55 to 0.49 and 0.46 respectively). It can be inferredfrom these results that the presence of Pd or Ni decreases slightly theactivity of the catalysts for the water gas shift reaction but increases in amajor extent the activity in methane steam reforming. As this reactionyields 3 molecules of hydrogen per molecule of methane converted, thehighest hydrogen yield obtained with the Rh-Ni catalyst may be explainedby its higher activity in methane steam reforming compared to the Rhcatalysts.

The Rh-Ni/Y2O3-Al2O3 was tested in the presence of crude bioethanolfrom 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, thehydrogen yield is very high (3.49mol/molethanol) with the Rh-Ni/ Y2O3-Al2O3 catalyst, and the conversion is only slightly decreased from 100% atthe onset of the reaction to 97% after 24 h of time-on-stream (97% ofconversion compared to 100% at the beginning of the reaction). On thecontrary, in the same reaction conditions, the Rh/MgAl2O4 catalyst isstrongly deactivated by coke deposition, especially during the first 2 h ofreaction.

5. Conclusions and recommendations

Although catalytic steam reforming (SR) is widely used for hydrogenproduction, it is not a green process since all the carbon containedin the feedstock is transformed into CO2. Replacing fossil fuels (mainlynatural gas today) by biofuels may circumvent this drawback. Bio-ethanol is an excellent candidate in the perspective of a hydrogen-basedeconomy.

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)

Metal phase

Yield (molmol� 1)

XEtOH (%)H2 CO CO2 CH4 C2H4 C2H6 CH3CHO H2O

Rh 3.43 0.67 0.65 0.50 0.01 0.01 0.01 2.50 98.7

Rh-Pt 3.53 0.69 0.77 0.57 0.02 0.02 0.01 2.04 93.4

Rh-Pd 3.64 0.79 0.69 0.49 0.01 0.02 0.01 2.24 98.6

Rh-Ni 3.84 0.75 0.76 0.46 0.01 0.01 0.01 2.10 98.2

46 | Catalysis, 2010, 22, 1–55

On a kinetics and catalysis point of view, there are some similaritiesbetween 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 adominant role, specially on Rh. At higher temperatures (550 1C and above),and specially for methane steam reforming, hydrocarbon activation is thedetermining step of reaction.

But there are also some remarkable differences between hydrocarbonsand alcohols.� Alcohols currently lead to complex kinetic schemes including

dehydration, dehydrogenation, cracking,. These reactions may be muchmore rapid than the steam reforming itself, which makes that alcoholconversion 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 ofethane).

Several grades of bioethanol can be processed to produce hydrogen buthigh purification costs should be avoided. Most studies were devoted to thesteam reforming of pure ethanol. The present tendency is to produce bio-fuels of second generation from lignocellulosic biomass or even from lig-nine. It is recommended to pay more attention to the use of raw bioethanol

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

0 4 8 12

Rh/MgAl2O4

Rh-Ni/Y2O3-Al2O3

16 20 24

Time (h)

Hyd

rog

en

yie

ld (

mo

l/m

ol)

Fig. 11 Hydrogen yield as a function of time-on-stream during the steam reforming of crudebioethanol in the presence of the Rh/MgAl2O4 and Rh-Ni/Y2O3-Al2O3 Rh-Ni/Y-Al2O3

catalysts.

Catalysis, 2010, 22, 1–55 | 47

that could be produced by such processes and to develop catalysts moreresistant 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, Born-

trager, 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, Cop-

enhagen, 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. Loffler 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 andM. 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. Hubner, 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. Lindstrom 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. Frances 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. Guillen, J. G. Alcalde, L. Garcıa 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 De-

partment 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-Lopez, F. G. Requejo, A. F. Craievich, J. Wei, M. Avalos-

Borja 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. Knozinger, Appl. Catal. A, 2002, 230, 153.

132 H. Knozinger, 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. Kecskes, M. Toth, M. Domok and K. Baan, 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. Hubner, 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. Orucu, F. Gokaliler, A. E. Aksoylu and Z. I. Onsan, 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 A. N. Fatsikostas and X. E. Verykios, J. Catal., 2004, 225, 439.

172 J. Kugai, S. Velu and C. Song, Catal. Lett., 2005, 101, 255.

173 Y. Yang, J. Ma and F. Wu, Int. J. Hydrogen Energy, 2006, 31, 877.

174 J. Sun, X. P. Qiu, F. Wu and W. T. Zhu, Int. J. Hydrogen Energy, 2005, 30,

437.

175 F. Frusteri, S. Freni, V. Chiodo, L. Spadaro, O. Di Blasi, G. Bonura and

S. Cavallaro, Appl. Catal. A: Gen., 2004, 270, 1.

176 F. Frusteri, S. Freni, V. Chiodo, S. Donato, G. Bonura and S. Cavallaro, Int.

J. Hydrogen Energy, 2006, 31, 2193.

177 Y. Yang, J. X. Ma and F. Wu, Int. J. Hydrogen Energy, 2006, 31, 877.

178 P. Biswas and D. Kunzru, Catal. Lett., 2007, 118, 36.

179 K. Urasaki, K. Tokunaga, Y. Sekine, M. Matsukata and E. Kikuchi, Catal.

Commun., 2008, 9, 600.

180 L. J. I. Coleman, W. Epling, R. R. Hudgins and E. Croiset, Appl. Catal. A,

2009, 363, 52.

181 F. Frusteri, S. Freni, V. Chiodo, L. Spadaro, G. Bonura and S. Cavallaro,

J. Power Sources, 2004, 132, 139.

182 F. Marino, G. Baronetti, M. Jobbagy and M. Laborde, Appl. Catal. A, 2003,

238, 41.

183 F. Marino, M. Boveri, G. Baronetti and M. Laborde, Int. J. Hydrogen Energy,

2001, 26, 665.

184 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.

185 A. Denis, W. Grzegorczyk, W. Gac and A. Machocki, Catal. Today, 2008, 137,

453.

186 J. W. C. Liberatori, R. U. Ribeiro, D. Zanchet, F. B. Noronha and J. M. C.

Bueno, Appl. Catal. A, 2007, 327, 197.

187 A. N. Fatsikostas, D. I. Kondarides and X. E. Verykios, Catal. Today, 2002,

75, 145.

188 A. N. Fatsikostas, D. I. Kondarides and X. E. Verykios, Chem. Comm., 2001,

851.

189 J. L. Ye, Y. Q. Wang, Y. Liu and H. Wang, Int. J. Hydrogen Energy, 2008, 33,

6602.

190 V. Klouz, V. Fierro, P. Denton, H. Katz, J. P. Lisse, S. Bouvot-Mauduit and

C. Mirodatos, J. Power Sources, 2002, 105, 26.

191 V. Fierro, O. Akdim, H. Provendier and C. Mirodatos, J. Power Sources,

2005, 145, 659.

192 M. H. Youn, J. G. Seo, P. Kim, J. J. Kim, H. I. Lee and I. K. Song, J. Power

Sources, 2006, 162, 1270.

193 L. Huang, J. Xie, R. Chen, D. Chu, W. Chu and A. T. Hsu, Int. J. Hydrogen

Energy, 2008, 33, 7448.

194 J. H. Lee, E. G. Lee, O. S. Joo and K. D. Jung, Appl. Catal. A, 2004,

269, 1.

195 T. J. Huang, T. C. Yu and S. Y. Jhao, Ind. Eng. Chem. Res., 2006, 45,

150.

196 J. A. Calles, A. Carrero and A. J. Vizcaıno, Microporous and Mesoporous

Materials, 2009, 119, 200.

197 F. Soyal-Baltacioglu, A. E. Aksoylu and Z. I. Onsan, Catal. Today, 2008, 138,

183.

198 L. P. R. Profeti, J. A. C. Dias, J. M. Assaf and E. M. Assaf, J. Power Sources,

2009, 190, 525.

199 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. Borjesson, 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 energetiques et gaz a effet de serre des filieres 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 hydrocarbonsfor on-board solid oxide fuel cell auxiliarypower units

Johannes W. Schwanka and Andrew R. Tadda

DOI: 10.1039/9781847559630-00056

Catalytic reforming of liquid transportation fuels and their hydrocarboncomponents via steam reforming, catalytic partial oxidation, and auto-thermal reforming is reviewed. The review focuses on fuel reforming togenerate hydrogen-rich syngas for on-board applications, with emphasis onsolid-oxide fuel cell (SOFC) based auxiliary power units (APU). After abrief overview of fuel properties of gasoline, diesel and kerosene-based jet-fuels such as JP-8, reforming methods including steam reforming, catalyticpartial oxidation and autothermal reforming are discussed. Strategies todeal with catalyst deactivation caused by carbon deposition and sulfurpoisoning are delineated. The review also addresses the special engineeringchallenges associated with the development of compact, on-board fuel re-formers 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 electri-fication of drive trains. While the immediate future may see the large-scaledeployment of hybrid and plug-in hybrid electric vehicles, the developmentof hydrogen-powered proton exchange membrane (PEM) fuel cell vehiclesis being pursued as an alternative option for eventually replacing theinternal combustion engine.

With growing interest in fuel cells as primary propulsion systems forautomobiles,1 a concerted research and development effort was mountedtowards on-board reforming of gasoline.2 PEM fuel cells require high-purityH2. Given the challenges of limited on-board hydrogen storage capacity andthe lack of hydrogen refueling infrastructure, on-board processing of gas-oline into pure H2 has been considered as a possible solution.3 Most of theresearch projects focused on partial oxidation or autothermal reformingprocesses, as on-board applications required fast start-up and good transientresponse.4 Large-scale conventional steam reforming technology developedfor stationary applications cannot easily be transferred to automotive ap-plications, due to size and weight limitations, and the limited availability ofsteam on board a vehicle. To generate the amount of hydrogen required for aPEM fuel cell vehicle would have taken a reformer system too large to fit intoa typical passenger car. Therefore, major efforts had to be directed towardsdownsizing of fuel reformer systems and the development of novel catalyticreactor configurations, including microstructured and plate reformers.5–15

aTransportation 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 arequite formidable, especially for applications requiring the fuel processing oflogistic fuels such as JP-8 and diesel.

Catalytic fuel reforming generates a hydrogen-rich gas stream fromhydrocarbons.16,17 Catalysts for onboard fuel reforming must meet a largenumber of requirements. First, they should be able to catalyze reformingreactions of hydrocarbons and their thermal decomposition products, andbe able to function under transient conditions, permitting multiple startupsand shutdowns without requiring elaborate regeneration schemes. Catalystsshould be resistant to coking, and they must be thermally stable withoutsintering at the high temperatures encountered under typical fuel processingconditions. Furthermore, they must be mechanically strong and withstandthe vibrations on moving vehicles. Ideally, they should also be tolerant tosulfur, even if low-sulfur fuels are being used. Finally, catalysts must be ableto cope with the variability in the chemical composition of fuels and fueladditives.

Despite considerable progress towards technical targets set by the U.S.Department of Energy, the complexity and technical difficulty of generatingpure H2 from gasoline on-board of vehicles prompted the U.S. Departmentof Energy in 2004 to make a no-go decision, and funding of research effortsfor on-board fuel reforming of gasoline into PEM-fuel cell grade H2 driedup.18 Instead, the research and development effort was redirected towardscentralized hydrogen generation and on-board storage of H2.

There is, however, another fuel cell application that could greatly benefitfrom on-board reforming of liquid transportation fuels, namely auxiliarypower units (APU) containing solid oxide fuel cells. Heavy-duty trucks,military vehicles, and recreational vehicles have significant needs for electricpower even when the vehicle is not moving.19–23 For example, heavy-dutytrucks tend to idle their diesel engines for many hours at a time, burning fuelat very low efficiency, to provide heating and cooling and generate powerfor sleeper compartment accessories such as televisions, microwave ovens,and refrigerators. According to recent studies, idling diesel engines burnnearly a billion gallons of diesel fuel every year in the United States alone,and shifting to SOFC-APU systems could result in significant fuel savingsand lower emissions.24,25 Depending on the degree of thermal integration,on-board SOFC APUs can be more efficient than conventional combustion-engine based electricity generators and provide the additional benefits oflower emissions and lower noise levels. The major impediment to thecommercial deployment of SOFC-based APU systems is the challenge toconvert liquid transportation fuels such as diesel and JP-8 military fuels on-board into hydrogen-rich syngas suitable for SOFCs.

Several books on the subject of fuel cell technology contain chaptersproviding an introduction into the basic principles of fuel processing.26–29

There have also been several review papers published on the subject of fuelprocessing for transportation and other portable power applications.30–34

Song provided an overview of the fuel processing options for low-temperature and high-temperature fuel cells.35 Qi et al. reviewed the techno-logical progress in the development of integrated fuel processors, with focuson 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 reformingof liquid fuels with special regard to the properties of diesel fuel and lightfuel oil.37,38 Semelsberger et al. analyzed various reforming processes from athermodynamic point of view.39 An extensive review of catalytic reformingof liquid hydrocarbon fuels for fuel cell applications appeared in 2006,40

focusing on reforming of diesel, gasoline, and representative model com-pounds. This earlier review40 provided an excellent treatment of the relevantthermodynamics, 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 duringthe 1980s,41 and there has been considerable interest in smaller scalefuel reforming using monolithic structures for distributed power generationin both stationary and mobile applications.42 Considerable progress hasbeen made towards reforming of C8þ hydrocarbons that are typicallyencountered in gasoline, diesel, or jet fuel.43–52 However, there are stillmany open questions as to how specific types of molecules affect thereforming behavior of commercial fuels. Research publications on thesubject of fuel processing of liquid fuels fall into two general categories:the first category includes fundamental studies of catalytic conversion ofindividual molecular components of a fuel or mixtures of model compoundsserving as surrogates for fuels. The second category contains papers thatdemonstrate a catalyst’s ability to reform gasoline, diesel, or jet fuelat laboratory scales to demonstrate the feasibility of various fuel proces-sing strategies. Here, we are focusing exclusively on reviewing on-boardcatalytic fuel processing of hydrocarbon-based liquid transportationfuels with emphasis on SOFC-based APU applications. This reviewwill not deal with other reforming methods such as plasma-assistedreforming,53 or supercritical reforming that have been covered in earlierreviews.40 This review also does not cover the extensive literature onreforming 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 ppmlevels of CO. Consequently, on-board fuel processing of gasoline or dieselfor a PEM fuel cell requires an elaborate set of catalytic reactors operatingin series, including a desulfurizer, a reformer, high and low temperaturewater gas shift reactors, and a selective oxidation reactor, methanationreactor, 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 fuel-processing scheme. The only requirement is to convert the liquid hydro-carbons into hydrogen-rich syngas and to protect the fuel cell from sulfurcontamination. This, for all practical purposes, requires only a reformer anddesulfurizer. Furthermore, APUs for heavy duty vehicles tend to operateunder more or less steady-state load for long periods of time, and there isless demand for dynamic load following.

58 | Catalysis, 2010, 22, 56–93

To gain a better understanding of the challenges involved in convertinggasoline, diesel, or jet fuels into H2-rich syngas, a brief overview of prop-erties and compositions of these fuels may be helpful. Gasoline, diesel, andcommercial jet fuels consist of hundreds of compounds. The exact fuelcompositions may be adjusted depending on the season and geography.Gasoline is a complex fuel and contains many impurities that can createproblems in catalytic fuel processing. Its catalytic conversion to syngas re-quires high temperatures in excess of 930K. Commercial jet fuels JET-Aand JET-A1 that are primarily used in the US are kerosene-based fuels. Themilitary uses specially formulated kerosene jet fuels, JP-4, JP-5, JP-8, andJP-100. These fuels have been hydrotreated and contain several additives.Jet fuels are considered kerosene-type fuels because fuel vaporization takesplace mainly within a temperature range of 473–623K. This range is muchnarrower compared to diesel, to meet the specifications for turbine engines.The high temperature boiling ranges of diesel and jet fuels make laboratory-scale fuel reforming studies quite difficult.

Diesel and jet fuels contain three major types of hydrocarbons: linear andbranched paraffins, cycloalkanes, and aromatics. The ring structures incycloalkanes and aromatics are generally believed to be more difficult toreform than paraffins. Diesel has typical carbon numbers of 10–16, whileJP-8 can be represented by an average carbon number of 11, excludingsulfur compounds and heteroatom containing additives, and gasoline by acarbon number of 8. Often, studies of fuel reforming of complex fuels arecarried 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 withmixtures 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 arecloser to dodecane, which is therefore often used as a surrogate. Inci-dentally, dodecane is also a suitable surrogate for paraffins in JP-8. Violi etal. developed a surrogate blend of six pure hydrocarbons to simulate thedistillation and compositional characteristics of JP-8.59

Although heteroatoms such as sulfur and copper are present in only smallconcentrations, they can have a significant effect on a fuel’s reformingchemistry and on the performance of reforming catalysts. Jet fuels suchas 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 onreforming catalysts.

Jet fuels contain a number of additives, and the type and concentration ofadditives represents the main difference between Jet-A and JP-8.60 For ex-ample, the military specification MIL-DTL-83133 for JP-8 requires theaddition of compounds that can act as static electricity dissipators, cor-rosion inhibitors, lubricity enhancers, fuel system icing inhibitors, andantioxidants.

Jet fuels tend to be quite susceptible to oxidation, since they mayhave undergone hydrotreating to remove mercaptans. The role of anti-oxidants is to prevent reactions of dissolved oxygen with fuel molecules thatcould 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-ditertiarybutyl-4-methylphenol.

The purpose of adding static dissipators is to increase the electrical con-ductivity of the fuel, thereby preventing the build-up of hazardous staticcharges. The currently used Statdis 450, manufactured by Octel, containsdinonylnaphthalene sulfonic acid along with other organic solvents and twoadditional ingredients consisting of sulfur and nitrogen-containing polymers.61

Corrosion inhibiting and lubricity improving additives protect the fueldistribution system from corrosion and lubricate metal surfaces in the tur-bine engine. These are compounds that strongly interact with metal surfaces,and are added at typically 20mg/L concentrations. There are many differenttrade names for these additives, for example QPL-25017-7 shown below.

Fuel system icing inhibitors consist of di-ethyleneglycol monomethylether(di-EGME).

CH3�O�CH2�CH2�O�CH2�CH2�OH

This additive, present at 0.1–0.15 vol%, prevents water that is dissolved inthe fuel from freezing at low temperatures. The handling of fuels containingdi-EGME in a fuel processor is difficult because di-EGME can precipitateout and form a gelatinous phase when water and fuel are co-fed in liquidform. Therefore, it is of critical importance to completely vaporize both thewater and the fuel prior to feeding them into the fuel processor.

The purpose of the metal deactivating additive N,N0-disalicylidene-1,2-propane diamine, added typically at 10mg/L is to improve the thermalstability 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 require-ment to vaporize the fuel prior to feeding it into the reformer. Heavyhydrocarbons typically present in diesel have boiling points higher than620–670K, and at these temperatures, some fuels begin to pyrolyze. Somereforming methods, as we will see later, require the co-feeding of steam, andthis can cause complications with miscibility and gelation of some of the fuelcomponents.

3. Catalysts for reforming of liquid hydrocarbons

The catalysts used for reforming of liquid hydrocarbons are usually pre-cious metal such as rhodium or nickel supported on oxide supports. Thereseems to be a consensus that the reforming of hydrocarbons in liquidtransportation fuels is possible, but difficult because of problems associatedwith carbon formation and sulfur poisoning. Consequently, the literatureshows 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 reformingof the liquid hydrocarbons, possibly by making the catalyst more resistantto coking by facilitating the oxidation of surface carbon species.62,63 Inaddition to typical supported metal catalysts, doped or substituted metal-oxide catalysts have also been applied to liquid fuel reforming. Of note arepyrochlore catalysts and hexaaluminate catalysts.67

The majority of reforming catalysts reported in the literature have beenprepared by conventional techniques. General descriptions of catalyst syn-thesis may be found in two recent books.68,69 Typically, a catalyst support ischosen and the active metal component is added as a solution. Removal ofthe solvent by drying leads to crystals of the metal precursor distributedthrough the pore structure of the support. These precursor crystals areconverted to metal or metal oxide particles usually by heating underhydrogen or air. The usual goal is to distribute the metal as widely as pos-sible, achieving very small average metal particle sizes so that the greatestpossible portion of metal atoms are in the surface of the particles, availablefor reaction. A number of variables can affect the metal dispersion, includingbut not limited to the precursor, solvent, support, and drying method.

Catalyst supports may be obtained from suppliers or prepared usingvarious techniques. Many mixed oxide supports are prepared by co-pre-cipitation or a sol-gel approach. Both approaches begin with solutions ofthe desired metals. In co-precipitation a base or other precipitating agent isadded or evolved in solution to cause precipitation of the metal hydroxidesor other insoluble forms, such as carbonates. After recovery by filtrationand washing, the solids are calcined to convert to the desired oxides. Sup-ports such as Al2O3 and SiO2 are easily obtained from major suppliers andare less often prepared by investigators.

Most often catalysts are tested in packed beds of small particles, butultimately for mobile applications they must be prepared in some structuredform. A washcoat on planar surfaces, honeycomb-type monoliths, or othersupport is usually required to provide low pressure drop and preventcatalyst attrition.

Catalysis, 2010, 22, 56–93 | 61

There are a large number of catalyst characterization techniques avail-able.70–72 Among the most important physical characteristics of reformingcatalysts are the total surface area and the metal dispersion. High specificsurface areas provide space to widely distribute metal particles, and highmetal dispersions yield efficient use of the active component, particularlyimportant when using noble metals. Often dispersions as high as 25–50%may be obtained when using noble metals.73–75 Nickel-based catalysts aremost often reported to have lower dispersions.74,101,126,127 Some investi-gators, however, have reported obtaining high dispersions when workingwith nickel.76

Catalyst surface areas are usually measured by N2 physisorption atcryogenic temperatures following the BET method. Metal particle sizes areoften estimated by titrating the metal with a chemisorbing species such asH2 or CO, which adsorb selectively on the metal. Metal particle sizes mayalso be estimated using SEM or XRD, provided that the particles are largeenough to yield sufficiently well defined peaks.

Operation under reforming conditions often leads to decreases in thecatalyst surface area.73,77–79,138,141 Surface area losses can be caused by highoperating temperatures or phase transitions in unstable support materials.Blocking of pores by coke can also lead to lower measured surface areas ifthe carbonaceous species are not removed prior to characterization.

The metal dispersion often decreases during reforming.73,141 High tem-peratures tend to cause sintering of the metal particles, leading to lowereddispersions. Components of the feed can also lead to reduction in availablemetal surface area; for example operation with S-containing feeds may leadto lower measured metal dispersions. It is important to note that surfacebound species may block chemisorption sites without changing the actualmetal particle sizes.

4. Fuel reforming methods

In the next section, three methods for fuel processing are reviewed, namelysteam reforming (SR), catalytic partial oxidation (CPOX), and autothermalreforming (ATR).

4.1 Steam reforming

Steam reforming dates back to the late 1800s and early 1900s where Mondand Langer pioneered the use of nickel or cobalt on pumice as catalysts,80

and Mittasch and Schneider demonstrated methane steam reforming overmagnesia supported nickel.81 Today, steam reforming is practiced on a largeindustrial scale for generating hydrogen-rich syngas from natural gas.

Typical steam reforming catalysts contain anywhere from 4–30wt%nickel supported on refractory oxides. More expensive cobalt and noblemetal catalyst have also been used. An extensive review of catalytic steamreforming was written by Rostrup-Nielsen.82 Steam reforming is typicallyconducted at operating temperatures of 950 to 1200K under 101–2500 kPaof pressure and steam to carbon ratios of 2/1 to 3/1. Excess steam isbeneficial, as it drives the reaction to completion, and it also decreases thedeposition of carbonaceous species, ‘‘coke’’, on the catalyst surface. The

62 | Catalysis, 2010, 22, 56–93

endothermic nature of the steam reforming reaction requires high operationtemperatures to achieve good conversion. Since steam reforming reactorsrequire indirect heating, the process does not lend itself well to applicationswhere rapid startup and response to transients is necessary. Steam re-forming of hydrocarbons involves a number of consecutive and parallelreactions.83 The most important reactions can be described by the followingoverall equations:

CnHm þ nH2O! nCOþ ðnþ 1=2mÞH2 ð1Þ

CnHm þ 2nH2O! nCO2 þ ð2nþ 1=2mÞH2 ð2Þ

The latter overall equation includes the formation of CO2, which isformed in the exothermic water gas shift reaction:

COþH2O, CO2 þH2 ð3Þ

In addition to water gas shift, CO and hydrogen can also undergomethanation:

COþ 3H2 , CH4 þH2O ð4Þ

Furthermore, carbon deposition on the catalyst can occur by the fol-lowing reactions:

2CO, Cþ CO2 ð5Þ

CnHm , nCþm=2H2 ð6Þ

Carbon deposition via reactions (5) and (6) and dehydrogenation oflarger hydrocarbons would be detrimental to the maintenance of catalystactivity, but these reactions can in principle be avoided by feeding water inexcess to operate under high steam/carbon ratios. There have been manystudies aimed at minimizing the deposition of carbon during steam re-forming.84–89 The need for high steam/carbon ratios makes on-board steamreforming impractical, because of the limited supply of water on board of avehicle.

While many contributions to the steam reforming literature focus onsteam reforming of methane or natural gas, there is some literature dealingwith the steam reforming of heavier hydrocarbons. For example, Gorte andcoworkers have investigated steam reforming of benzene, toluene, cyclo-hexane and n-octane, molecules that are relevant as surrogates for gasolinereforming.90 This study, conducted on ceria-supported Pd catalysts, dem-onstrated that paraffins are more amenable to steam reforming than aro-matic compounds. Benzene in particular had a much higher activationbarrier than paraffinic molecules. An extensive review of steam reforming ofaromatic compounds over Al2O3 and SiO2 supported group VIII metals,published by Duprez, delineated trends that can provide guidance for fuelprocessing of kerosene-based liquid fuels.91 As a normalized measure ofcatalytic 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, de-hydrogenation, and total gasification reactions. For fuel processing aimedat generating hydrogen-rich gas for fuel cells, total gasification is the mostimportant reaction. In terms of total gasification activity, Rh proved to bethe most active catalyst, closely followed by Ni. Pt, however, was less activeby an order of magnitude. The activity of Rh appeared to be strongly af-fected by the nature of the support, while Pt and Ni seemed to be lesssensitive. Rh/Al2O3 had much higher activity than Rh/SiO2.

There has been some work on Ru-based steam reforming catalysts. Forexample, Suzuki et al. demonstrated that hydrodesulfurized kerosene couldbe reformed over a Ru/CeO2–Al2O3 catalyst at 1073K and a steam tocarbon ratio of 3.5, and 100% conversion of hydrodesulfurized kerosenewith high H2 yield could be maintained for 8000 h on stream.92 However,under their reaction conditions, significant amounts of methane, approxi-mately 5%, were formed. Hu et al. used a proprietary Pd/ZnO catalyst forsteam 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-8fuel, but also desulfurized JP-8 deactivated the catalyst within a few hourson stream.

In summary, it can be stated that steam reforming of hydrocarbonstypical for kerosene-type fuels is possible under relatively high steam/car-bon ratios and at high temperatures as long as catalyst poisons, especiallysulfur, are removed. In general, aromatic hydrocarbons are more difficult toreform than paraffins. It needs to be noted that steam reforming, due to itsslow dynamic response and large water supply requirements, is not a veryattractive option for on-board fuel processing systems.

4.2 Catalytic partial oxidation (CPOX)

Partial oxidation of hydrocarbons in presence of substoichiometric amountsof oxygen is an efficient method for generating CO and H2.

93 Partial oxi-dation of hydrocarbons is a highly exothermic process and raises the un-converted reactants and products to very high temperatures. Thenoncatalytic, homogeneous partial oxidation occurs at temperatures ofabout 1600–1700K, but in presence of suitable catalysts, the temperaturecan be significantly lowered to about 1150K. The incomplete combustion ofhydrocarbons leading to formation of CO and H2 is attractive for on-boardreforming because it does not require water to reform the fuel, and thereaction has rapid light-off, providing fast start-up and good transient re-sponse. The overall reaction is depicted below:

CnHm þ 1=2nO2 þ 1:88nN2 ! nCOþ 1=2mH2 þ 1:88nN2 ð7Þ

CPOX reactions are highly exothermic with fast kinetics, and maybecome mass transfer limited. Nevertheless, the high reaction rates and theconcomitant rapid release of heat facilitate rapid start-up of the fuel pro-cessor, an important issue for on-board applications. A major concern is thesusceptibility 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 absenceof steam in the feed leads to lower H2 yield compared to steam reforming. Itis very important to carefully control the oxygen to carbon ratio (O/C), aseven small variations in the O/C ratio can lead to significant changes incatalyst bed temperature, product yields, and coking. The sensitivity to O/Cratios and the highly exothermic nature of the process pose a major chal-lenge for controlling CPOX reactors, and there has been major emphasis ondeveloping coke resistant CPOX catalysts.94

There is an extensive body of literature dealing with partial oxidation ofmethane, but there have also been many studies regarding CPOX of liquidhydrocarbons and transportation fuels.95 Early developments of POX re-actors focused on the use of methanol, which thanks to its lower heat ofreaction permits a relatively simple reactor design.96 An example of this isthe HotSpotTM reactor for methanol fuel processing developed by JohnsonMatthey.97 POX technology for gasoline or diesel reforming is very chal-lenging, mainly in terms of temperature control.

On the fundamental research side, the Schmidt group at the University ofMinnesota has investigated the partial oxidation of C1–C16 hydrocarbonsand low sulfur diesel into syngas with Rh catalysts on monoliths at highspace velocities under very short contact times.50,98,99 They also exploredthe contributions of homogeneous and heterogeneous reactions in thecatalytic partial oxidation of n-octane, iso-octane, and their mixtures onRh-coated a-alumina foams.100 They found that the syngas product selec-tivities were independent of the structure of the reacting fuel, but not theolefin product distribution. Iso-octane produced mostly propylene and iso-butylene, 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 amountsof 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 theNi surface, thereby limiting the number of Ni atom surface ensemblesavailable for C–C bond formation and coke build-up, while leaving theactive sites for partial oxidation more or less intact. Since the reaction en-vironment in a CPOX reactor is rather complex, involving highly exo-thermic combustion and partial combustion reactions, followed byendothermic steam reforming using H2O generated in the combustion zoneand slightly exothermic water gas shift that becomes significant in the coolerregions of the reformer, it would be important to experimentally measureproducts, intermediates, and temperatures as a function of time and pos-ition in the reformer. The boundaries between the different reaction zones inthe reformer are most likely dynamic, and significant temperature andconcentration gradients will be present due to variations in the steam/C andO/C ratio. To probe these spatial variations in reforming reactors and togain insight into the complex interplay of reactions and intermediates,analytical techniques such as Spatially Resolved Capillary Inlet MassSpectrometry (SpaciMS) have been used by Galen Fisher at DelphiResearch Labs in collaboration with a group at ORNL led by WilliamPartridge and Jae-Soon Choi.102 This approach permits mapping of

Catalysis, 2010, 22, 56–93 | 65

composition and temperature profiles within the working reactor, in bothaxial as well as radial directions. The SpaciMS method utilizes capillariesinserted into monolith channels, and the gas collected at various positions inthe monolith is fed to a mass spectrometer for analysis. Temperatures aremeasured by thermocouples. A schematic of the experimental setup isshown in Fig. 1.

This method was originally pioneered for time-resolved measurements ofemission transients103 and in situ measurement of reactions occurring inlean-NOx trap catalysts.104–107 While diesel lean NOx traps operate at lowertemperatures in the range of 473–773K, reformers reach much highertemperatures creating a challenge for the thermal and mechanical stabilityof the capillary probes. At ORNL, temporal resolutions of 7–1Hz havebeen obtained for capillaries with lengths of 0.4 and 2.5m, respectively. Thecapillaries can be moved via a translation device within the monolithchannels, thereby probing gas compositions as function of position. To maptemperature profiles, either thermocouples or phosphor-tipped non-con-ductive optical fibers can be used, with different types of phosphors offeringthe ability to monitor different temperature regimes.108

The SpaciMS analysis has been used to investigate the CPOX reaction oflight hydrocarbons, methane, and propane. Fig. 2 shows typical resultsobtained for propane reforming under O/C of about 1 and at space velocityof 30 000 h� 1.109 In the oxidation zone, which is about 0.6mm wide, syngasis generated and both combustion and then reforming appear to occur, andCO2, H2O, and temperature peaks are observed.

This zone is followed by a 0.6mm wide C3H8 depletion zone, where steamand dry reforming occurs, and CH4 generation is observed. Steam and dryreforming dominate the downstream section of the monolith, and water gasshift equilibrium is established. A similar approach was taken by theSchmidt group110 to obtain species and temperature profiles for cata-lytic partial oxidation of methane on Rh-coated a-Al2O3 foams.111,112

Their experimental set-up included a moveable thin quartz capillary forsampling of gases via mass spectrometry and a thermocouple. Applyingsuch spatially resolved analysis methods to liquid fuel reforming would bevery useful.

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

66 | Catalysis, 2010, 22, 56–93

4.3 Autothermal reforming (ATR)

Autothermal reforming, a combination of steam reforming and partialoxidation, may be more suitable for on-board reforming than steam re-forming because of its better response to transient operation. In auto-thermal reforming, a portion of the fuel is oxidized by air, and the heatreleased in the process is used to drive the endothermic steam reformingreactions in the downstream sections of the reactor. The overall heat duty ofthe reactor can be adjusted by using varying O/C and steam/C ratios. Theintroduction of superheated steam leads to the onset of steam reforming andwater gas shift reactions, causing the downstream section of the catalyst bedto cool. The overall stoichiometry of the autothermal reforming process canbe described by equation (8):

CnHmOy þ xðO2 þ 3:76N2Þ þ ð2n� 2x� yÞH2O

¼ nCO2þð2n� 2x� yþm=2ÞH2 þ 3:76xN2

ð8Þ

Since the autothermal reforming process uses both water and oxygen inthe feed stream and produces CO2 and H2, it might intuitively appear to bea poor choice for a fuel processor because of CO2 dilution of the SOFCanode feed stream. However, in reality the reaction does not proceed tocompletion and there is a significant concentration of CO present in thereactor effluent, according to the water gas shift equilibrium equation (3).

There is a large body of literature on autothermal reforming of methaneand light hydrocarbons.113 However, for heavier liquid hydrocarbons, it hasproven 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 thecatalyst bed entrance, followed by gradual decrease in temperature in the

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

Catalysis, 2010, 22, 56–93 | 67

downstream sections of the catalyst bed. In addition, there is the possibilityof homogeneous and heterogeneous thermal cracking reactions of hydro-carbons. A schematic of these reaction zones in shown in Fig. 3. Thephotograph of a nickel-containing monolith shows clearly the demarcationbetween the partial oxidation and steam reforming zones.

The generation of hydrogen through autothermal reforming of iso-octanewith lowering of CO concentrations through water-gas shift and preferen-tial oxidation was studied by several research groups, for example Moonet al.114 and Thompson and coworkers.115 A comparison of the iso-octanereforming performance of Ni/CeZrO2 catalysts in packed beds versusmicroreactors showed the importance of properly managing heat transfereffects, an issue that is very critical in microchannel reactors.116

A group at Argonne National Laboratories has worked on reformingliquid hydrocarbons with a Pt catalyst supported on cerium oxide andgadolinium oxide.117 A temperature of 1123K was required to achievesatisfactory hydrogen yields and bring the methane concentration to lowlevels. To describe the reaction kinetics relevant for autothermal reformingof iso-octane, Pacheco et al.118 developed a mathematical model using theAspen Plus process simulator, and using the experimental results obtainedby the Argonne team. Palm et al. reformed a mixture of C13–C19 hydro-carbons in presence of precious metal based catalyst, and they achieved highhydrogen and low methane yields.119 To capture the characteristics ofnaphthenes, aromatics, and sulfur in diesel fuel, which are thought to beresponsible for coking and catalyst poisoning, the effect of model com-pounds 1,2,3,4-tetrahydronaphthalene, decahydronaphthalene and 1-ben-zothiophene on ATR conversion was also investigated. It was found thatdecahydronaphthalene did not have a detrimental effect on conversion, but1,2,3,4-tetrahydronaphthalene gave a decrease in conversion. It was hy-pothesized that the aromatic character of the 1,2,3,4-tetrahydronaphthalenemade it more difficult to reform, thus lowering the conversion. The additionof 1-benzothiophene, at 30 ppm sulfur, sharply decreased the conversionand also altered the temperature profile of the reactor. The changes intemperature profile suggest that the sulfur poisons primarily the endo-thermic steam reforming sites of the catalyst.

Fig. 4 shows a schematic autothermal reforming reaction network fordodecane over ceria-zirconia supported Ni catalysts, taking into accounthomogeneous and heterogeneous reaction pathways.120 These proposedroutes do not explicitly show the contributions of water gas shift, but thecontribution of water gas shift is implied by the product distribution. Theprimary route to syngas is CPOX of dodecane, with steam reforming acting

Fig. 3 Reaction zones in autothermal reforming reactor.

68 | Catalysis, 2010, 22, 56–93

primarily on conversion of C1–C4 hydrocarbons produced in varioushomogeneous and heterogeneous cracking reactions.

A recent study reported the use of perovskite oxides containing rare earthelements for autothermal reforming of isooctane in a fixed bed micro-reactor.121 It turned out that the binary oxides LaNiO3 and LaCoO3 wereactive and gave high yields of H2, but suffered from structural stabilityproblems, as the oxides decomposed under the reducing conditions of thereformer. Other binary oxides such as LaCrO3, LaFeO3, and LaMnO3 werestable but less active.

Similar to steam reforming, aromatic compounds are more difficult toprocess by ATR than ring and chain compounds. The effect of fuel additiveson catalyst activity and durability remain largely unknown. The review ofthe literature on the processing of JP-8 into syngas suggests that sulfurpoisoning is the dominant challenge with coking being the second mostimportant concern.

5. Deactivation of reforming catalysts

5.1 Carbon deposition

A major technical obstacle to reforming of liquid hydrocarbon fuels iscarbon deposition and coking on the catalyst. During reforming, hydro-carbons are broken down on catalytic surfaces, their constituent fragmentsreact with steam or oxygen, and CO, CO2, and H2 are released from thecatalyst surface. Accumulation of carbon can lead to catalyst deactivation,reactor fouling and pressure drop problems, and in some cases evenstructural failure of catalysts. However, the detailed mechanisms for carbon

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

Catalysis, 2010, 22, 56–93 | 69

deposition and coke formation remain open to debate, and a clear under-standing does not currently exist how the chemical structure of fuels in-fluences coking behavior. One school of thought is that coke forms by thepolymerization of large hydrocarbons into arrays of polycyclics. It is be-lieved that aromatics are detrimental to reforming because the compoundsact as precursors for the formation of the polycyclic networks. On nickelcatalysts, 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 accu-mulation than are those based on precious metals. In his general review,Sehested has summarized the development of different types of carbonstructures formed from higher hydrocarbons during steam reforming onnickel catalysts.122 Of significant importance is the mechanism of carbonfilament formation, which can lead to macroscopic catalyst failure beyondsimple deactivation. It has been proposed that the formation of filamentouscarbon involves the diffusion of carbon atoms formed via decomposition ofadsorbed hydrocarbons through Ni particles, forming a nickel carbideintermediate phase.123

One approach to increase the coking resistance of Ni catalysts is the useof supports with mobile oxygen, for example mixed oxides of ceria andzirconia, but even with this support, some carbon deposition mainly in formof filaments was observed.124,125

Cordierite monoliths loaded with Ni used in the reforming of dodecaneshowed significant carbon deposition following reaction.126 At high nickelloadings (W7 wt%) the carbon deposition was so severe that the monolithsdisintegrated during the reaction. TPO of the recovered monoliths showedthat carbon deposited during POX, SR, and ATR had different oxidationtemperatures, indicating differences in composition or morphology. WhenNi/Ce0.75Zr0.25O2 was loaded on the monoliths, carbon deposition wasgreatly reduced.

Carbon deposition was also observed during ATR of isooctane overNi/Ce0.75Zr0.25O2 catalysts of varying nickel loadings.127 Here the amountof carbon was found to increase with increasing nickel loading, whichwas associated with increasing average nickel particle size. SEM analysis ofthe spent catalysts showed carbon filaments of various sizes, as well as whatappeared to be amorphous carbon coating the particles.

Chen et al. carried out a detailed spatially-resolved analysis of carbondeposition in monoliths coated with Ni/ceria-zirconia or just Ni alone underATR, CPOX, and SR conditions.128 They mapped the amount and types ofcarbon deposited in different sections of the monoliths with a combinationof scanning electron microscopy, energy-dispersive X-ray spectroscopy(EDS), and temperature-programmed oxidation. They found significantdifferences in the amounts and location of carbon, depending on the type ofreforming reaction, reflecting axial variations in temperature and oxygenconcentration. Fig. 5 shows a typical SEM image of filamentous carbondeposited in the downstream sections of a monolith catalyst.

An alternative approach to impart better resistance to carbon depositionand coking is to modify the nickel surface. Sn has been successfully used todecrease carbon deposition in several catalytic processes, for example

70 | Catalysis, 2010, 22, 56–93

aromatization and dehydrogenation of paraffins. Motivated by the bene-ficial effect of Sn in these reactions, the effect of Sn was explored for Ni-catalyzed steam reforming, partial oxidation,129,130 and dry reforming.131

The beneficial effect of adding small amounts of tin to nickel was also ob-served on reducible oxides, such as ceria-zirconia.124,125

Nikolla et al. showed by combined DFT and experimental studies thatthe carbon surface chemistry of nickel catalysts can be very effectivelycontrolled by alloying the nickel surface with tin.132 Long-term activitymaintenance of a nickel catalyst is governed by the prevention of carbon-carbon bond formation leading to coke while selectively facilitating theformation of C–O bonds. On Ni surfaces, there is no differentiation in theactivation barriers for C–C versus C–O bond formation, while introducingSn into the Ni surface leads to a situation where the overall rate of carbonoxidation is much greater than the rate of C–C bond formation. It is for-tunate that in the limit of small Sn concentrations, the formation of Sn/Nisurface 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 alloycatalysts supported on yttria stabilized zirconia, a ceramic used in solidoxide fuel cell anodes were measured using a variety of experimental probes,and the results supported theoretical models indicating a relationship be-tween catalytic activity and the position of the center of the electronic dband.134 Contrary to the common belief that the change in the electronicstructure of metals in an alloy is caused by charge transfer among the alloycomponents, this work showed that the interaction of Sn and Ni leads toshared electronic states. Consequently, the filling of the Ni d band remainsunchanged (Fig. 6). This has important consequences for the catalyticperformance of Sn/Ni alloy catalysts. The Sn-induced decrease of theaverage energy of d electrons decreases the binding strength of carbonspecies. This lowers the surface concentration of carbon-containing reaction

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

Catalysis, 2010, 22, 56–93 | 71

intermediates during reforming. The result is a significant decrease in thedriving force for coke deposition.

While more resistant to carbon deposition than nickel, precious metalcatalysts do exhibit varying levels of carbon accumulation during re-forming. 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/gcatalyst after reaction, respectively.48 Carbon deposition levels appearedto be correlated with the presence of unsaturated intermediates. Addition of5% 1-methylnaphthalene to the feed increased carbon deposition forall the catalysts. Although significant carbon was found in the spent cata-lysts, no significant impact on performance was reported. As with severalnickel-based investigations, TPO of the spent catalysts revealed very dif-ferent oxidation temperatures for the carbon species deposited on thecatalyst.

Dreyer et al. observed carbon deposition during POX and ATR of decaneand hexadecane.135 Although the carbon was not strictly quantified, carbonburn-off temperature profiles indicated there was more carbon depositedunder pure POX conditions than ATR. No effect on the catalytic per-formance due to carbon was noted.

The presence of heavy polyaromatic compounds in Diesel creates a ten-dency for deposition of large amounts of carbon. This can become a seriousissue especially under autothermal conditions where the downstream sec-tions of the catalyst bed may experience a decrease in temperature due toendothermic steam reforming reactions. To prevent carbon formation, thereactor temperature must be kept at temperatures above 1070K. Onestrategy to decrease carbon formation has been to introduce additives suchas tin, to use rare earth oxide supports,136 and to avoid conditions thatwould lead to the formation of carbides.88

After reviewing carbon deposition data across many investigations it maybe concluded that carbon deposition remains a concern for both nickel- andprecious-metal based catalysts. It is important to note, however, that themere presence of carbon on a spent catalyst is not necessarily indicative ofcarbon poisoning. Surface carbon is a necessary intermediate of reformingand may be expected to be present on all reforming catalysts; it is thequantity and form of carbon that lead to performance issues.

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

72 | Catalysis, 2010, 22, 56–93

5.2 Deactivation by sulfur

There are various strategies that can be pursued to prevent deactivation ofcatalysts during reforming of sulfur-containing fuels. The first strategy aimsat development of catalysts that can tolerate significant levels of sulfurwithout experiencing long-term deterioration of performance.35,137 This ischallenging, as sulfur strongly binds to the active sites on the catalyst andprevents access by other reactants. In addition to simple site blocking, sulfuratoms can have detrimental electronic effects on active metals, which caneffectively poison surrounding sites. Mitigation of sulfur poisoning requiresoperation of the catalysts at high temperatures, and may still require theremoval of sulfur compounds from the fuel reformer effluent to protect theSOFC anode catalyst from sulfur poisoning. Significant research efforts,however, have been directed towards developing hydrocarbon-reformingcatalysts capable of tolerating moderate to high concentrations of sulfur.

Several factors in catalyst formulation and operation can be used to en-hance sulfur tolerance. The most straightforward is the use of a noble metalsuch as Pt, Pd, or Rh as the active component. Noble metals exhibit lowersusceptibility to poisoning by sulfur than other transition metals. Within thenoble metals, some appear better than others. Azad’s group at the Uni-versity of Toledo has reported that 1wt% Rh or Rh-Pd supported onGd0.1Ce0.9O1.9 or Zr0.25Ce0.75O2 was stable for steam reforming toluene inpresence of 50 ppm thiophene.138 Catalysts of the same formulation wherePd was substituted for Rh showed greater deactivation.139 Schmidt’s grouphas reported partial oxidation of methane with 28 ppm CH3SH (0.5wt%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 inhydrogen yield observed.

Shekhawat et al. directly compared Rh/ZrO2–CeO2, Pt/ZrO2–CeO2, andPt/Al2O3.

48 The Rh/ ZrO2–CeO2 catalyst was the only one to show stableoperation under 1000 ppmw S content in the fuel. They also demonstratedthat the Rh catalyst can substantially recover activity when sulfur is re-moved from the feed. While Rh appears to be superior, Pt has also beenshown to act as a stable reforming component under some conditions. Luet al. have reported that 1.5 wt% Pt/ Gd0.1Ce0.9O1.9 was stable for steamreforming of isooctane with a sulfur level of 300 ppmw.75 Some deactivationwas observed at sulfur loadings of 500 ppmw, but hydrocarbon conversionwas maintained above 90% for 100 hours.

The use of reducible metal oxide supports has been shown to be superiorto conventional supports such as Al2O3. Cheekatamarla and Lane havecompared various bimetallic catalysts supported on both Al2O3 and CeO2

for reforming of JP-8.141 They found that the catalyst activity was higher forall metal combinations when using CeO2 as a support. Mixed oxides in-cluding TiO2 and CeO2 are a central claim in BASF’s patent application fora sulfur tolerant natural gas steam reforming catalyst.142

A group at NETL, in conjunction with Spivey, has studied pyrochlores ascatalysts for liquid fuel processing. Catalytically active metals may be sub-stituted into the pyrochlore structure, which has excellent thermal propertiesof 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 ofn-tetradecane and n-tetradecane/1-methylnaphthalene/dibenzothiophenemixtures (50 ppmw).64 Both catalysts showed reduction features assignable toRu during temperature programmed reduction, although the reduction in thepyrochlore took place at higher temperature. The Ru dispersion was 2.3% forthe pyrochlore and 27% for the Ru/g–Al2O3 catalyst. The pryochlore catalystwas significantly more resistant to deactivation during S-feeding, maintaingH2 and CO yields in excess of 70% with only small increases in CO2 andCH4 production. The Ru/g–Al2O3 catalyst showed catastrophic loss of H2

and CO yields with larger increases in CO2 and CH4 production. The sup-ported Ru catalyst also gave a much higher olefin yield during S-feeding thanthe pyrochlore. Upon removal of S from the feed, neither catalyst completelyrecovered its prior activity, but the pyrchlore recovered to a much higherH2 and CO yield than the alumina-supported Ru catalyst. The alumina-supported Ru catalyst also showed much higher C deposition.

Rh-substituted La-Zr pyrochlores also showed better S tolerance thancomparable supported Rh catalyst.65 During catalytic partial oxidation ofn-tetradecane with addition of 1000 ppmw dibenzothiophene, the pyro-chlore (La1.50Sr0.50Rh0.10Zr1.90O6.70) showed a drop in H2 and CO yieldfrom 90% to 70–75% on S addition. The original H2 and CO yields werealmost totally recovered 200 minutes after removing S from the feed.In contrast, the supported Rh catalyst showed a drop in H2 and CO from75–80% to less than 40%, with continuing deactivation. Neither H2 norCO yield recovered to above 50% upon removing S from the feed stream.Olefin production during S-feeding was lower on the pyrochlore thanon the supported Rh catalyst. The level of activity recovery was nega-tively correlated to the amount of C found deposited on the catalystfollowing reaction. The amounts of C deposited on the catalysts were higherfollowing reaction with S than during partial oxidation with n-tetradecaneonly.66

Hexaaluminates are oxide materials with a spinel structure and refractoryproperties. 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 ofthe formula ANi0.4Al11.6O19� d (A=La, Sr, Ba) have been studied for thecatalytic partial oxidation of n-tetradecane.67 The Ba and Sr substitutedcatalysts showed a b-alumina structure, while the La substituted materialexhibited a magnetoplumbite structure. Catalyst surface areas ranged from14 to 22m2/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 50L g� 1 h� 1. The La-substitued catalyst showed a peak inH2 and CO production, followed by slight decreases then steady perform-ance. This was attributed to carbon deposition deactivating the most activenickel sites. Both the Ba and Sr-substituted catalysts showed no loss in H2

and CO production. All catalysts produced some CO2 and some lighthydrocarbons. The tolerance of the catalysts to S was tested by introductionof 50 ppmw dibenzothiophene to the fuel stream during reaction. TheLa-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 afterdibenzothiophene was removed whereas the Sr-substituted catalyst did not.

Additives such as promoters or sacrificial components have also beenused to improve sulfur tolerance. Azad et al. examined the effect of addingY2O3 and CuO to Rh-based steam reforming catalysts. Y2O3 had a bene-ficial effect on performance, apparently by increasing or stabilizing the Rhdispersion.139 CuO was added as a sacrificial component to react withsulfur, thereby removing sulfur from the active metal sites.138 While thisimproved the hydrogen yield obtained during reforming with sulfur, in theabsence of sulfur more coking was observed as compared to the catalystwithout CuO. Similarly, BASF’s methane steam reforming catalyst alsoincludes a transition metal component to capture sulfur during reforming,which is released during a subsequent regeneration step.143 Dinka has re-ported that addition of 2wt%K to a La0.6Ce0.4Fe0.8Ni0.2O3 perovskitecatalyst increased sulfur tolerance, allowing an increase from 50 to 225 ppmS in autothermal reforming of JP-8 with no change in performance.144 Theeffect of 2wt%K addition was similar to the effect of 1 wt% Ru. Mawdsleyand Krause reported that introducing Cr as stabilizing element into LaNiO3

improved the sulfur tolerance.121 Modification of Ni/Sr/ZrO2 catalysts withRe or La was also found to improve the sulfur tolerance of the catalystduring autothermal reforming of hydrocarbon fuels.145 Molybdenum car-bide catalysts have also shown some tolerance to sulfur poisoning in a studyof steam reforming of tri-methylpentane.146

Many sulfur tolerant catalysts have been studied at modest space vel-ocities, allowing for greater contact time with the catalyst region and op-portunity for reaction. Reactions can be carried to high conversion in veryshort contact times when no sulfur is present in the feed. Schmidt’s grouphas published extensively on successful CPOX using precious metal cata-lysts of hydrocarbons from methane to tetradecane at contact times of a fewmilliseconds.98,147 Schwank’s group at the University of Michigan hasdemonstrated ATR of isooctane, dodecane, and isooctane at gas hourlyspace velocities of 200 000 hr� 1.148,149 Of the sulfur tolerant catalysts sur-veyed here, the space velocities range from 8000 hr� 1 to 62 000 hr� 1. Theselower space velocities allow catalysts with lower activities, a result of partialsulfur poisoning, to achieve the required hydrocarbon conversion.

The organic sulfur compounds present in liquid fuels can react withcatalytic surface sites to form stable metal sulfides, thus causing severedeactivation. One possible solution is to remove sulfur compounds fromfuels via selective adsorption on adsorbents such as zeolites that can beregenerated.150,151 Yang and co-workers developed a p-complexationmethod for selective sorption of sulfur compounds by copper and palladiumhalide sorbents and CuY zeolites.152–156 They also investigated the use ofcarbon-based sorbents.157 For on-board applications, the need for re-generation of the sorbents introduces an additional degree of complexity,and in some cases there may also be some issues with effective removal ofsterically hindered sulfur compounds in presence of aromatic hydrocarbons.

Given the difficulties in fuel processing of heavy hydrocarbon fuels, athird strategy may be pursued, namely catalytic hydrodesulfurization of thefuel 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 suitablecatalyst, and the H2S is then absorbed in a sulfur absorption bed, for ex-ample zinc oxide. This method is widely used on an industrial scale,158 butsince hydrodesulfurization reactors require high pressure H2, it may not bepractical for on-board deployment. A group from PNNL has recently de-scribed a very interesting alternative.159 Their process uses an integratedsteam reformer to generate hydrogen for hydrodesulfurization and amicrochannel distillation unit upstream of the hydrodesulfurizer. This ap-proach makes it possible to process a lighter feed fraction instead of theunmodified JP-8 fuel. The light fraction from the microchannel distillationunit contains smaller concentrations of refractory sulfur compounds,thereby facilitating the hydrodesulfurization. The U.S. Navy has also pur-sued a fuel processing system capable of handling JP-8 type fuels thatutilizes 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 forSOFC APUs. The first strategy is external reforming, where the catalyticconversion of liquid fuels into syngas takes place in a separate catalyticreactor. The H2-rich syngas product is then fed to the anode compartmentsof the fuel cell stack. From a reaction engineering perspective, external re-forming is the simplest method for reforming of complex liquid transpor-tation fuels such as diesel and gasoline, but suffers from low overallefficiency and high system cost.

The second strategy is indirect internal reforming. This method is verysimilar to external reforming, but in this case the catalytic reformer is de-signed in such a way that it is in direct thermal contact with the anodecompartment of the solid oxide fuel cell. This method was initially appliedto molten carbonate fuel cells.161 Having the reforming reactor in thermalcontact with the high-temperature fuel cell facilitates better thermal inte-gration of the heat released during the electrochemical reactions on theanode with the heat requirements for vaporizing steam and fuel, and can beutilized to provide additional heat for endothermic steam reforming re-actions in the fuel reformer. The practical implementation of this concepthas to deal with the problem of thermal mismatch between the relatively fastendothermic steam reforming reactions and the much slower exothermicelectrochemical reactions in the fuel cell. This thermal mismatch can lead tocold spots in sections of the reformer, and in a worst case scenario couldcause fractures of the system.162

The third possible strategy is direct internal reforming, which is perhapsthe ideal heat integration strategy. Due to high SOFC operating tempera-tures, internal reforming of methane or light hydrocarbons such as propaneis possible. Internal reforming is attractive because of the SOFCs ability toutilize CO along with H2. Carrying the reforming reaction out directly in theanode compartment of the fuel cell provides the most efficient way totransfer heat from the electrochemical reactions to the catalytic sites wherethe reforming reactions take place. The direct reforming of liquid fuels isvery 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 weredesigned with guidance from DFT calculations.163 Barnett and Zhan car-ried out internal partial oxidation of iso-octane on SOFC anodes, but hadto place an additional Ru-CeO2 layer between the fuel stream and the anodeto obtain stable operation without anode coking.164

6.1 Gasoline

In view of the extensive infrastructure for gasoline distribution, there havebeen substantial efforts towards the development of gasoline reformers inindustry, involving companies such as Hydrogen Burner Technology andArthur D. Little,165 which later partnered with the Italian company DeNora to start Nuvera. A. D. Little developed a system that included a POXreactor as primary fuel processor and water gas shift and preferential partialoxidation (PROX) reactors to generate PEM-fuel cell grade pure H2.Gasoline fuel processor development also was carried out at ExxonMobil incollaboration with General Motors. Shell worked on the development ofgasoline CPOX reactors for Daimler Chrysler in partnership with Ballardand UTC.

Delphi Corporation has been one of the leading developers of on-boardSOFC 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 ‘‘proofof concept’’ gasoline powered SOFC APU was demonstrated on a BMW7-series sedan.167 The fuel processor for this system involved a catalyticpartial oxidation (CPOX) reformer containing alumina or zirconia basedcatalyst formulations.168

This fuel processor shown in Fig. 7 does not include provisions for re-moval of sulfur, requiring that the catalysts must have adequate sulfurtolerance. 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 anenergy recovery unit.

On a more fundamental basis, researchers at Argonne National La-boratory investigated the effect of the major constituents of gasoline, fueladditives, and impurities on fuel processor performance.169 They found thatat high space velocities and/or low reforming temperatures antioxidantadditives in gasoline decreased the hydrogen yield in the reformate.

6.2 Diesel

Diesel fuel has a relatively high hydrogen content, making it an attractivefuel for on-board reforming.170 H2-rich reformate gas can be generatedfrom diesel fuel not only through reforming, but also though directhydrocarbon decomposition. A group at Argonne National Lab has utilizedsimplified mixtures of fuel components to understand how factors such asH2O:C and O2:C ratios, temperature, and fuel composition affect the re-actions in diesel reforming.171 They indentified intermediates in the oxi-dation and coke-forming reactions.

While steam reforming provides in principle the highest H2 yield, theendothermic nature of the process and the need to supply large amounts ofsteam makes this process unfavorable for on-board, transient operation.The exothermic catalytic partial oxidation (CPOX) lends itself much betterto transient operation. Diesel CPOX reactors can be operated withoutcatalyst, and after start-up in air, the reactors operate at very high tem-peratures in excess of 1500K. By adding catalysts, the reactor temperaturecan be significantly lowered, making such systems attractive for on-boardvehicle applications. The major components of diesel fuel, n-decane and n-hexadecane, can be converted to syngas with high selectivity.172 CPOX,thanks to its exothermic nature, requires only very short contact times in theorder of milliseconds.173

However, catalytic diesel reforming is quite challenging due to the pro-pensity of diesel to pyrolyze, and coke formation and sulfur deactivation ofthe catalysts are major issues that need to be carefully managed. Due to thelarge number of hydrocarbon components in the fuel, a multitude of re-actions are involved, making it very difficult to unravel mechanistic de-tails.88,174 For diesel reformers, typical catalysts contain either preciousmetals (Pt, Rh, Ru) or less expensive base metals (Ni) that are supported oncarefully engineered oxide supports. For example, Pt/ceria and bimetallicPt-Pd ceria catalysts175,176 and Pt catalysts supported on Al2O3–CeO2 orAl2O3–La2O3

177 have been used for autothermal reforming of syntheticdiesel. As Spivey has pointed out in a recent ACS meeting, the tendency ofreforming catalysts to deactivate by coke deposition and sulfur poisoningcalls for innovative approaches to novel catalytic materials and reactordesigns, guided by computational catalysis methods.178 In a study ofautothermal diesel fuel reforming over Ru-doped lanthanum chromite andaluminite catalysts, adequate fuel mixing prior to feeding diesel, steam andair into the autothermal reformer was identified as a critical issue.179

There has been recent patent activity regarding thermoneutral reformingprocesses 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 patentsclaim the use of multicomponent catalysts containing Ni, Ce2O3 La2O3, Pt,ZrO2, Pt, Rh, and Re for reforming of feedstocks containing o 200 pmsulfur without coke formation.

Alternate strategies to circumvent carbon formation have been employed.Low and intermediate temperatures can favor the formation of C, as canlow local O/C ratios. Mundschau and coworkers have reported the use of amembrane reactor comprising YSZ walls.181 Fuel is fed inside the mem-brane and air passes through the membrane walls to mix with the fuel abovethe catalyst bed. The air passing through the walls increases the local O/Cratio in the cool regions of the reactor, suppressing the driving force for Cformation. The catalyst employed inside the reactor is a La0.5Sr0.5CoO3–d

perovskite, operated at 1223K to suppress carbon formation in the catalystregion. The perovskites catalyzed both total and partial oxidation of com-mercial diesel, depending on the O/C ratio employed.

6.3 Jet fuel

As already discussed above, the reforming of kerosene-based jet fuels suchas JP-8 is difficult due to the presence of heavy hydrocarbons in these fuels.Among heavy hydrocarbons, paraffins and cycloalkanes are relatively easyto convert while the aromatics are known to be the most difficult ones toprocess. The challenges in reforming heavy aromatics stem both from theirlower reactivities as well as from their higher propensities to form cokeunder typical reforming conditions.

Sung and Ibaretta developed a model for reforming of a kerosene sur-rogate over a wide range of conditions with varying feed temperatures,operating pressures, steam/carbon ratio and O/C ratio.182 Based on calcu-lations using finite gas-phase chemistry, the concluded that short residencetimes and partial oxidation with minimal water addition gave the most ef-ficient reforming performance.

Gould et al. examined the performance of a nickel-ceria-zirconia catalystfor autothermal reforming of n-dodecane, tetralin, and their mixture, asrepresentative compounds for alkanes and bicyclic compounds in jet fuel.149

It was found that the mixture of tetralin and n-dodecane did not react asexpected for a linear combination of the two types of molecules. Instead, thereforming behavior was dominated by reforming characteristics of puretetralin. This observation appeared to be counterintuitive, as one wouldexpect that the aromatic character of tetralin would make it more difficult toreform. It is well documented that compared to alkanes, aromatic moleculeshave higher activation energies for steam reforming.183 The simple ex-planation for the tetralin-dominated reforming behavior of the mixture wasthat the tetralin-containing mixture led to higher temperature profiles in thereactor compared to n-dodecane, due to the higher adiabatic equilibriumtemperature obtained with tetralin.

As potential low-cost coking-resistant catalysts for reforming of JP-8 fuelsurrogate, 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 catalystsurface.

An alternative strategy for avoiding the coking problem is to remove theheavy hydrocarbons, especially the heavy aromatics from the feedstocksbefore reforming. In refineries, aromatics can easily be separated frompetroleum fractions through solvent extraction or adsorption. Clearly, suchstrategies are not applicable for compact, on-board fuel processors. Theunit operations required for the regeneration of solvent or adsorbent wouldintroduce far too much complexity. Unfortunately, the boiling points of thevarious hydrocarbon species in these types of fuels are too close to eachother for effective separation via distillation. Therefore, alternative ap-proaches have been considered for selective removal of heavy polynucleararomatics prior to reforming. One such approach involves the catalyticcracking of JP-8, followed by separation of light cracked gases from heaviesbefore 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 depositionand catalyst deactivation, pre-reforming strategies relying on crackingmight require frequent catalyst regeneration. As an alternative, reactiveseparation of heavy aromatics appears possible, taking advantage of thedifferences in the relative reactivities of various hydrocarbons. Prereforminghas been practiced widely in industry as a preliminary step to convertdistillate fuels for the production of synthesis gas.185 Naphtha feed is pro-cessed first in a prereformer to produce an equilibrium mixture of methane,hydrogen, CO and CO2 that is then subsequently sent to the main reformerwhere the conversion to synthesis gas takes place. In this approach, reactivehydrocarbons would be converted into lighter components such as methane,hydrogen and oxides of carbon, while keeping the less reactive heavyaromatics intact. The product mixture will then be cooled down to atemperature where the unconverted heavy aromatics condense as liquid.The methane-rich gaseous stream would be fed into the reformer toproduce synthesis gas, while the condensed heavy aromatics would beused as fuel to generate heat. To deal with non-volatile residues thatcan constitute up to 1.5 vol% in jet fuels, and to decrease the danger ofcoking due to heavy hydrocarbons, catalytic cracking of the fuel withzeolite catalysts and manganese/alumina catalysts and subsequentseparation of the light cracked gas from the non-volatile aromatic specieshas been investigated.186

The Air Force Research Laboratory (AFRL) has developed a JP-8 fuelprocessor capable of generating 3kWe of SOFC grade feed reformate.187

Their fuel processor uses a combination of partial fuel vaporization andcatalytic cracking to pre-reform the fuel. Their strategy for JP-8 processingis twofold. First, the fuel is partially vaporized so that the bulk of the sulfurcontaining species remain in the liquid phase, which can be fed to a com-bustor that heats a steam reformer. Second, the remaining organosulfurcompounds are converted into H2S by catalytic cracking, which can sub-sequently be adsorbed on ZnO. Additionally, the cracking unit converts thelarger hydrocarbons, which have a propensity to coke, into smaller non-coking species.

80 | Catalysis, 2010, 22, 56–93

McDermott Technology Inc has developed a fuel processor for Navy shipservice capable of converting military logistic fuel (NATO F-76) into a H2

rich gas.188 In conjunction with Siemens Westinghouse and Phillips Pet-roleum they developed a 250 kWe low-sulfur diesel powered SOFC.189 Theirdesign incorporates an ATR with a sulfur removal unit operation down-stream of the ATR to protect the SOFC. They have operated a pilot scaleprocessor 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 microreformersystem used Pt supported on Gd-doped CeO2, but this system was designedfor stationary, residential applications, rather than for on-board deploy-ment.190 The reformer was coupled with a downstream ZnO bed to removesulfur, to protect the SOFC.

7. Systems engineering aspects of on-board fuel processing

No discussion of on-board fuel reforming can be complete without ad-dressing system and engineering considerations. Issues of reactant delivery,heat integration, and others must be dealt with to achieve on-board re-forming of liquid fuels. This balance of plant introduces parasitic losses onthe system, reducing the amount of power that can be effectively extractedfrom the fuel. High operating pressures or long processing trains, whichintroduce significant pressure drops across the system, will lead to increasedenergy costs for reactant delivery. Separate heat exchange steps will increasethe system weight and volume, adversely affecting power density of the fullsystem.

The primary driver for system-related considerations is the choice of re-forming scheme. While SR offers high hydrogen yields without nitrogendilution, it requires a supply of water, a high-duty vaporization step, andhigh rates of heat transfer into the reforming reactor from a separate heatsource. ATR also requires water, but the feed ratio to the fuel will be muchlower than SR. POX operation requires only air as a co-reactant for thefuel, but may be prone to carbon deposition and higher operating tem-peratures than SR and ATR. For mobile applications, POX or ATR arelikely the most practical reactions schemes.

The literature regarding design of reformers and systems integration isrelatively sparse. Studies of reforming trains for PEM fuel cells, which in-clude water gas shift and preferential oxidation reactors, are available.191

There has also been an appreciable research effort directed towards mem-brane reactors, particularly for hydrogen separation using Pd-based mem-branes, which can accommodate the PEM’s requirement for high-purityhydrogen.192–194 As SOFCs are able to utilize CO as well as hydrogen, themembrane approach is unnecessary.

In contrast to large-scale stationary reformer operation,195 on-boardsystems have to deal with several additional major technical challenges,including frequent transient operation during start-up and shutdown, op-eration under high space velocities, and ability to function under harshthermal conditions and mechanical vibrations. Conventional packed-bedreactor 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 theyare more responsive to fast transients.196–198

Delphi Corporation teamed up with PACCAR Incorporated and VolvoTrucks North America to define what system level requirements must bemet for diesel fuelled SOFC based auxiliary power units on commercialtrucks.199 Battery power was used to bring the SOFC up to operatingtemperature. Once the reformer was operating in partial oxidation modeand supplied hydrogen-rich syngas to the SOFC, it became possible to re-cycle anode tail gas, providing additional steam thereby moving fromCPOX mode towards autothermal reforming. A potential hazard en-countered during warm-up was that at temperatures below 773K, H2

leaking from the reformer or fuel cell stack could collect in the system andignite as the temperature increases. To deal with this challenge, sensors hadto be added to detect leakage of H2 and CO, and the sensor signal would asnecessary trigger appropriate control action to shut the system down. Asadditional safety precaution, special efforts were made to tightly seal thecomponents to minimize leakage and maximize containment.

In principle, water required for ATR can be recovered from the SOFCanode exhaust or tail gas burner, as schematically shown in Fig. 8. Ifcondensation is used as a recovery step, re-vaporization will be required. Ifwater 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 asCO2 and N2 will be recycled along with the steam (Fig. 9).200

A group from Los Alamos National Laboratory found that changing therecycle rate from 20% to 30% resulted in a much slower temperature rise inthe autothermal catalyst bed and required operation under higher O/Cratios 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

profiles.201 An ASPEN simulation showed how the condensed exhaust gasescould 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 thepast decade significant efforts have been made to advance fuel cellscloser to commercialization. A great portion of the effort, however, has beendirected at PEM fuel cells and reforming systems designed to supportthem. General Motors has disclosed work on heat management in aPEM-supporting reformer system wherein water vaporization is used tocontrol the PrOX reactor temperature and fuel cell exhaust is combustedto increase heat recovery, integration with a water-cooled high-temperaturePEM stack, and heat-exchange networks to enable rapid startup.203–205

They have worked on design considerations for decreasing the volumeand mass of the reformer train, and managing steam within the fuelprocessor/fuel cell.206,207 Volvo has also published some work on fuel cellsystems, but as with GM it is aimed at motive power and low temperaturefuel cells.208

Ford Motor Company made a significant number of patent applicationsregarding fuel cells, but apparently only one that is inclusive of an on-boardreformer.209 The company’s focus appeared to be on the fuel cell stack,hydrogen separation membranes, and the use of metal-hydrides as a loadleveling and startup reservoir of hydrogen.

Delphi has perhaps the strongest patent literature record of fuel cellsystem integration development. Their work has been geared towardsSOFCs as the electrical generation device. Work has covered the range ofprocess considerations including but not limited to startup strategies,210,211

SOFC tail gas recycling,212,213 heat transfer,214 waste heat recovery,215 andtemperature control.216

In work which is more applicable to SOFC-APUs, General Motors hasinvestigated solutions to reformer startup using electrical preheat to initiatelight-off.217 This pre-heated reformer is nominally intended for use within

Fig. 9 Steam/carbon ratio as function of anode tail gas recycle ratio.200

Catalysis, 2010, 22, 56–93 | 83

the PEM-supporting fuel processor train disclosed in other General Motorspatents.

Besides supplying hydrogen rich gas for fuel cell applications, on-boardreformers can also be used for hydrogen-assisted engine operation, en-hanced rapid 3-way catalyst light-off in emission control systems, and forproviding reductant for enhanced diesel aftertreatment and selective cata-lytic 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 6H2O ð9Þ

2NOþ 2NH3 þ 1=2O2 ! 2N2 þ 3H2O ð10Þ

10NOþ C3H8 ! 5N2 þ 3CO2 þ 4H2O ð11Þ

2NOþ CH4 þO2 ! N2 þ 2H2Oþ CO2 ð12Þ

While ammonia and urea, which can be thermally decomposed intoammonia, are very effective for selective catalytic reduction of NOx, they arecumbersome to use, as they require separate tanks to be installed on thevehicle. Furthermore, there is the danger that unconverted ammonia slipsthrough the catalytic converter. An alternative strategy is to rely onhydrocarbons from the fuel as reducing agents.218 Compared to ammonia,hydrocarbons are less effective NOx reducing agents, but adding smallamounts of hydrogen can drastically improve the rates of reduction.219 Anon-board fuel reformer could provide not only hydrogen, but also CO andlight hydrocarbons to facilitate NOx reduction to N2. Not surprisingly,supported Pt catalysts are among the most active catalysts for NOx re-duction with H2, but their N2 selectivity is not very good.220 Other groupVIII noble metals, such as Pd, are also active and selective at moderatetemperatures.221 While CO by itself can be used as a reductant under oxi-dizing conditions, it appears to be less effective compared to hydrocarbonsor hydrogen, but on a Pd/Al2O3 catalyst good NOx conversion and se-lectivity was achieved with a syngas mixture of CO and H2.

222 Anotherpromising catalyst system is Ag/Al2O3.

223–228 On Ag catalysts, hydrogenaddition to ammonia229 or hydrocarbons230,231 has proven to be verybeneficial for NOx conversion and N2 selectivity, and it significantly lowersthe light-off temperature of the catalyst.219,232,233

8. Conclusions

While there has been considerable effort devoted to the development ofcompact fuel processors for liquid fuels, there are still many challenges re-maining that need to be addressed. From a fundamental science standpoint,more detailed reaction kinetics and rate laws need to be determined, notonly for pure hydrocarbons, but also for mixtures of hydrocarbons wheresynergistic effects might be at work, and ultimately for actual fuels. Theextension of spatially resolved analysis of axial and radial composition andtemperature profiles in monolith reactors from light gases to liquid

84 | Catalysis, 2010, 22, 56–93

hydrocarbons and actual fuels appears promising. Guided by density-functional theory, the design and synthesis of advanced carbon- and sulfurtolerant catalyst formulations comes within reach. From a reaction-engin-eering standpoint, major issues that remain to be addressed are adequateperformance under transient operation, ability to mitigate coking and sulfurpoisoning of catalysts, thermal integration and efficiency, and overall sys-tem integration within the weight and space constraints of vehicles. There isalso a need to demonstrate long term maintenance of catalytic activityunder typical on-board conditions with repeated start-up and shutdowncycles. Start-up sequences for auxiliary power units appear to be closelyguarded secrets in the industrial community.

There is a need for more extensive system level auxiliary power unit re-search to gain a better understanding of how fuel processors can be bestintegrated with the fuel cell stacks and the balance of plant. Focusing onoptimization of individual components is unlikely to lead to optimizedoverall system performance.

Acknowledgments

The authors would like to acknowledge the financial support provided bythe U.S. Army Tank-Automotive Research, Development & EngineeringCenter under Cooperative Agreement Number W56HZV-05-2-0001, anddetails 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. Lowe 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 Inter-

national 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. Kurschner, 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 Com-

mittee 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 Y. Lu, J. Chen, Y. Liu, Q. Xue and M. He, J. Catal., 2008, 254, 39.

76 H. H. Ibrahim, P. Kumar and R. O. Idem, Energy and Fuels, 2007, 21, 570.

77 A. S. Mukasyan and P. Dinka, J. Power Sources, 2007, 167, 472.

78 A. Qi, S. Wang, G. Fu and D. Wu, D. Appl. Catal. A: General, 2005,

293, 71.

79 A. C. McCoy, M. J. Duran, A.-M. Azad, S. Chattopadhyay and M. A.

Abraham, Energy Fuels, 2007, 21, 3513.

80 L. Mond and C. Langer, British Patent 12608 (1888).

81 A. Mittasch and C. Schneider, U.S. Patent 1,128,804 (1915).

82 J. R. Rostrup-Neilsen, Catalysis: Science and Technology, vol. 5, Chap. 1,

Catalytic Steam Reforming, Springer-Verlag, 1984.

83 J. Rostrup-Nielsen, J. Catal., 1973, 31, 173.

84 J. Rostrup-Nielsen and D. L. Trimm, J. Catal., 1977, 48, 155.

85 J. R. Rostrup-Nielsen, NATO ASI Ser. Ser. E: Appl. Sci., 1982, 54, 127.

86 D. L. Trimm, Catal. Rev.-Sci. Eng., 1977, 16, 155.

87 T. Borowiecki, A. Machocki and J. Ryczkowski, Stud. Surf. Sci. Catal., 1994,

88, 537.

88 D. L. Trimm, Catal. Today, 1999, 49, 3.

89 E. Nikolla, J. Schwank, and S. Linic, J. Catal. 2009, doi:10.1016/j.jcat.2009.

02.006 (in press).

90 X. Wang and R. J. Gorte, Appl. Catal., A: General, 2002, 224, 209.

91 D. Duprez, Appl. Catal. A: General, 1992, 82, 111.

92 T. Suzuki, H. Iwanami and T. Yoshinari, Int. J. Hydrogen Energy, 2000,

25, 119.

93 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.

94 K. A. Avci, Z. I. Onsan and D. L. Trimm, Appl. Catal. A, 2001, 216, 243.

95 W. Yanhui and W. Diyong, Int. J. Hydrogen Energy, 2001, 26, 795.

96 R. Kumar, S. Ahmed and M. Krumpelt, Proceedings of the US Fuel Cell

Seminar, Kissimee, Florida, November 1996, 17.

97 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.

98 L. D. Schmidt, E. J. Klein, C. A. Leclerc, J. J. Krummenacher and K. N. West,

Chem. Engin. Science, 2003, 58, 1037.

99 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.

100 G. J. Panuccio, K. A. Williams and L. D. Schmidt, Chem. Engin. Sci., 2006, 61,

4207.

101 S. Pengpanich, V. Meeyoo, T. Rirksomboon and J. Schwank, Catal. Today,

2008, 136, 214.

102 Galen B. Fisher, personal communication.

103 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 Tech-

nical Paper, 2000-01-2952.

104 J.-S. Choi, W. P. Partridge and C. S. Daw, Appl. Catal. A, 2005, 293, 24.

105 J.-S. Choi, W. P. Partridge, W. S. Epling, N. W. Currier and T. M.

Yonushonis, Catal. Today, 2006, 114, 102.

106 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.

107 J. Parks, S. Huff, J. Pihl, J.-S. Choi and B. West, SAE Technical Paper 2005-

01-3876.

108 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. Monzon, A. Guerrero-Ruız 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: En-

vironmental, 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: Environ-

mental, 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 andM. 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 de-

velopment: 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 Fran-

cisco, 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, Indian-

apolis, 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 Inte-

gration 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 andW. H. Pettit, W. H.Methods for ReducingMass 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 sys-

tems. 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. Konig, 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. Sobalık, 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. Eranen, 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 methodsfor the investigation of environmentallyimportant reactions

F. C. Meuniera

DOI: 10.1039/9781847559630-00094

1. Introduction

The improvement of the activity and selectivity of catalysts is a perpetualobjective for researchers in catalysis and can rely on a number of approaches.While high-throughput combinatorial methods are raising a lot interest andfinding some success in heterogeneous catalysis,1 the understanding of re-action mechanism through detailed kinetic and spectroscopic studies is an-other proven approach to support catalyst development. The purpose of thisreport is to present some examples as to how insights into catalyst struc-ture 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 ofthe author) related to the production of hydrogen via the water-gas shiftreaction (WGSR) and the selective reduction of NOx, which are both ofcurrent interest with respect to environmental issues. The most recent workput the emphasis on using so-called ‘‘operando’’ conditions, in which thespectroscopic data were recorded while the reaction rate was simultaneouslymeasured over the catalyst using a single bed reactor. One of the aims of thisreport was to highlight some of the benefits of combining kinetic andspectroscopic 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 werecompared to the rate of CO2 production during WGSR obtained over thesame sample in a different experiment. The second example describes how acombination of kinetic and spectroscopic data (here recorded on two sep-arate apparatuses) helped understanding, in parts, a very complex reaction,which is the selective catalytic reduction of NO with propene. The subtlerole of NO oxidation on the reaction pathway of NO reduction over Ag-Al2O3-based catalysts is discussed. The different mechanisms taking placedepending on the Ag loading are highlighted. The formation of oxidisedNOx(ads) species, important reaction intermediates, was evidenced byin situ DRIFTS and thermogravimetric measurements. The isocyanatesobserved by DRIFTS were also proposed as a crucial surface intermediate.

The third and final example documents the use of operando DRIFTSwith steady-state isotopic kinetic analysis (SSITKA), using a mass spectro-meter (MS) to follow mass transients, to investigate catalytic reactions. Thewater-gas shift (WGS) reaction over noble metals supported on cerium-containing oxides is presented in details. The DRIFTS-MS-SSITKA tech-nique proved invaluable in determining the true role of formates ‘‘seen by

aLaboratoire Catalyse et Spectrochimie, ENSICAEN, Universite de Caen, CNRS, 6 BdMarechal 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. essentiallyspectators). Many advantages of coupling kinetic and spectroscopic meas-urements are discussed, e.g. cell validation as a kinetic reactor, role of theobserved surface species, oxidation state of the working catalyst, spectatorversus potential reaction intermediate.

2. The bases of spectrokinetic analyses

Tamaru and co-workers reported investigative work in heterogeneous cata-lysis combining spectroscopic and kinetic data dating back to the 1960’s.2–5

The corresponding studies represented some of the first attempts to relate theconcentration and reactivity of surface species to the rate of the reactionmeasured over the same catalyst. The water-gas shift reaction, COþH2O-CO2þH2, over base metal oxides was one of the reaction investigated. Thetype of experiments was highly challenging at the time, bearing in mind thetechnological limitation of the equipment (e.g. dispersive IR, since FTIRonly become widespread much later)6 and supply of high purity gases (e.g.CO was sometimes obtained from the decomposition of sodium formate bysulphuric acid and using a liquid nitrogen trap).4 Custom-made cells weredesigned 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 thetransmission IR analysis of a single wafer (e.g. with a mass of 300mg) andthe second bed contained a much larger mass of catalyst (e.g. 11 g) to ensurea measurable conversion.4 The utilisation of a dual bed implies a non-negligible risk that each bed experienced different experimental such astemperature and concentration gradients.

The decomposition rate of surface formates were compared to the rate ofCO2 production during the water-gas shift reaction (WGSR) obtained over aMgO sample at 280 1C. The WGSR was measured under a feed of CO andwater in a recirculation mode. In a different experiment, the rate of formatedecomposition was obtained by following the decay of the formate bands forvarious initial surface coverages. The quantitative relation between IR bandintensity and formate concentration was obtained via calibration curverealised using reference samples derived from adsorption of know amounts offormic acid on the catalyst. The values of WGSR rate and formate de-composition were sufficiently similar (Table 1) so that the formates seen byIR could be conclusively proposed as the main reaction intermediate.

Table 1 Comparison of the rate of formate decomposition to CO2þH2 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

CO2þH2 (mm3 g� 1 h� 1)

Rate of the water-gas

shift reaction

(mm3 g� 1 h� 1)

0.06 17 11

0.07 25 23

0.08 37 31

Catalysis, 2010, 22, 94–118 | 95

This type of study, based on a transient involving a change in thechemical potential of one or more of the chemical elements present,assumes that the reactivity of the surface species is the same under reactioncondition and under concentration changes. This is clearly not always thecase7 and will be discussed in more details in section four of the presentreport.

A large number of spectrokinetic studies were also carried out in the1970’s in the former USSR by Mathyshak and co-workers.8 A recent reviewgathering many examples of this work has been published in CatalysisToday.9 In essence, these authors varied many experimental parameters (inparticular reactants concentrations) for the reaction of interest and meas-ured the consumption rate of the reactant(s), formation rates of the prod-uct(s) and the surface coverage of adsorbates observable by IR. Amicrokinetic model was then developed and the experimental and simulatedvariations of rates and surface coverages were compared to ascertain themodel.9 The procedure, leading to a possible reaction mechanism, appearsto be experiment and time-intensive. Unfortunately, the methods used tocarry out spectral decomposition are often unclear, while this point is oftenthe bottle neck when complex spectra are considered. Other difficultiesassociated with the techniques regard the determination of molarabsorption coefficients and, sometimes, the use of chemical transients.In summary, the full microkinetic analysis combined with spectroscopy is anelegant method but clearly requires a significant amount of work and stillsome assumptions.

Many teams around the World have since used spectrokinetic methodsderived from these seminal investigations, in particular by using isotopictransients (e.g. for studying syn-gas conversion,10–14 CO2-reforming ofmethane15,16 and nitrogen oxides decomposition,17,18) but it is not the scopeof 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 inour investigation of the selective catalytic reduction of NOx with hydro-carbons, a highly complex reaction. The data corresponding to differentexperimental techniques were recorded on separate apparatuses, each hav-ing its own reactor. Therefore, the experimental parameters (e.g. tempera-ture 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 thevarious techniques were consistent and worthwhile qualitative conclusionscould be drawn.

A commercial cell from Spectra-Techs was used to carry out the diffusereflectance FTIR (DRIFTS) measurements. The bottom part of thecatalyst bed, which was not probed by the IR beam, was possibly by-passedby the reaction mixture, due to the high pressure drop of the originalreactor 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 thatcould be placed in the DRIFTS cell (typically 20mg) was not sufficientto induce a significant conversion. The thermogravimetric data wererecorded on an IGA microbalance. More experimental details can be foundelsewhere.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 pro-moted by, e.g., Co, Ni, Cu, Fe, Sn, Ga, In, Ag) are active catalysts for theselective reduction of NOx (NO and NO2) with hydrocarbons (HC-SCR).22

Note that under typical lean-burn conditions, the promoting metals arealmost exclusively in an oxidised state. Catalysts based on Ag supported ong-alumina received a particular attention, as these materials are among themost 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 highertemperatures (W400 1C), under these experimental conditions. It is inter-esting 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 be-fore complete propene conversion. The 1.2% Ag/g-Al2O3 yielded similarconversions to N2 as those obtained over the alumina but at lower tem-peratures. Low concentrations of N2O, NO2 and NH3 were also obtained inthis case.

The activity of the high loading silver catalyst was significantly differentfrom that of the g-Al2O3 and the 1.2% Ag/g-Al2O3. Complete combustionof the reductant was achieved at 350 1C over the 10% Ag/g-Al2O3, incontrast to the temperatures of 100% combustion of the reductant over1.2% Ag/g-Al2O3 and g-Al2O3 of 500 1C and 565 1C, respectively. In add-ition, the N2 yield remained significantly lower than that of N2O (obtainedat the lower temperatures) and NO2 (obtained at the higher temperatures).Over the 10% Ag/g-Al2O3 and at the higher temperatures, the conversion toNO2 was limited by the thermodynamics of the reaction (see dotted line inFig. 1, NO2 yield):

NOþ 12O2 , NO2 ð1Þ

One of the striking features of the catalytic data reported in Fig. 1 was thesharp increase in the NO2 yield obtained over the alumina as soon ascomplete conversion of propene was achieved, at ca. 565 1C. The values ofNO2 yield obtained at these temperatures were significantly higher than thatallowed by the thermodynamics of the reaction represented by equation (1).It has to be stressed that identical plots were obtained independently ofusing increasing or decreasing temperature profiles and no change in theyield of NO2 was observed after several hours on stream. A NO2 yield inslight 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

error, as indicated by the fact that the NO2 yield value measured in the caseof the 10% Ag/g-Al2O3 was exactly equal to that expected by the thermo-dynamics (Fig. 1).

Such high NO2(g)/NO(g) ratios largely exceeding the value expected fromthermodynamics were also observed during the course of the SCR reactionover other excellent SCR catalysts based on cobalt/alumina.23 These obser-vations clearly indicates that the main route to NO2(g) over these catalysts isnot 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. hadproposed that NO2 was an intermediate, based on the facts (i) that the re-action starting from this molecule was much faster than that starting fromNO and (ii) that the activity measured in the presence of NO2 did not requirethe presence of the promoter.

0

20

40

60

80

100

150 350 550

C3H

6 co

nver

sion

/%

0

20

40

60

80

100

150 350 550

NO

con

vers

ion

/%0

10

20

30

150 350 550

NO

2 Y

ield

/%

0

20

40

60

150 350 550N

2 Y

ield

/%

0

2

4

6

150 350 550

Temperature /°C Temperature /°C

NH

3 Y

ield

/%

0

10

20

30

150 350 550

N2O

Yie

ld /%

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 NOþ 500 ppm C3H6þ 2.5% O2 /He,W/F¼ 0.06 g s cm� 3 (GHSVB50 000h� 1). The dotted line in the plot giving the NO2 yieldrepresents the thermodynamic limit associated with the reaction NOþ 1/2O23NO2.(Reprinted from reference 20.)

98 | Catalysis, 2010, 22, 94–118

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 measuredand the best SCR catalysts (i.e. those leading to N2, alumina and 1.2%Ag-promoted alumina) displayed a very poor activity at the optimumtemperatures for the SCR reaction (Fig. 2). The activity of the 1.2% Ag-alumina was essentially identical to that of the support, while the highloading sample (10 wt.% Ag) was very active for NO2 formation.

IR and thermogravimetric data were also collected in order to unravel therole of the Ag promoter. The in situ DRIFTS spectra reported in Fig. 3qualitatively describe the growth of various surface nitrate species (bandsat ca. 1560, 1305 and 1255 cm� 1) when the samples were exposed to aNOþO2 stream. Both silver-promoted materials appeared to exhibit afaster uptake rate of NOx(ads) than that observed over the plain alumina.This trend was confirmed quantitatively using the mass uptake measuredwith the microbalance (Fig. 3). It is clear that at least one of the main role ofthe 1.2% silver promoter was to favour the oxidation of NO(g) to stronglybound 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 10wt.% 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 occuron these catalysts. A NO decomposition-type mechanism was suggested tooccur on the high loading material, which probably consisted of large me-tallic silver particles deposited on the alumina (Fig. 4). To this respect, theactivity pattern of this sample resembles that of platinum group metals. Inthe case of the low loading Ag/Al2O3, dispersed Agþ cations and/or small

0

5

10

15

20

150 250 350 450

Temperature /°C

550 650

NO

2 Y

ield

/%

Thermal Equilibrium

10% Ag/Al2O3

Al2O3

1.2% Ag/Al2O3

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% NOþ 5% O2 in Ar,W/F¼ 0.06 g s cm� 3. (Reprinted from reference 20.)

Catalysis, 2010, 22, 94–118 | 99

12001300 1400150016001700

120013001400150016001700

12001300Wavenumber /cm-1 Time under the NO/O2/He stream /min

1400150016001700

12551305

1550

1550

13001560

Absorbance

0.3

a

b

c

1235

0

2

4

6

0 1 2 3 4 5

Rel

ativ

e w

eigh

t upt

ake

/ 10−3

Al2O3

1.2% Ag/Al2O3

10% Ag/Al2O3

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 streamwas 15min (lower spectrum), 60min (middle spectrum) and 180min (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%NOþ 2.5%O2/He. (Reprinted from reference 20.)

R-NO2

R-ONONOx-

N2O + N2

alumina

Ag °

Ag+

ON OO

C3H6NO / O2

H2OCOx

CxHy

NO + O2

C3H6

N2NO2

R-NH2R-NCO NH3

O2

ON

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

100 | Catalysis, 2010, 22, 94–118

electropositive silver clusters in strong interaction with the alumina pre-vailed. The exact nature of the reaction mechanism(s) occurring over alu-mina and low loading Ag/Al2O3 is not yet fully understood, but N2 isprobably formed via a series of parallel and consecutive reactions involvingnumerous intermediates over both the silver and alumina phases.

Oxidised species of nitrogen (e.g., inorganic nitrates) are thought to reactwith reduced forms of this element (e.g., isocyanate, ammonia, formed viaorgano-nitrogen compounds) to produce N2 (Fig. 4). The typical volcano-shape of the N2 yields plots is ascribed to the competitive reactions betweenNO and O2 for the reductant. The unselective combustion of the reducingagent with O2 becomes much faster than the SCR at higher temperaturesand diminishes the number of reductant molecules available for theSCR. 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 bythe evidence described thereafter. The formation of organonitrogen com-pounds was observed over some SCR catalysts.25 The formation and de-composition of such compounds would rationalise the high concentrations ofNO2 observed. The example given below equations (2–4) is arbitrarily basedon nitromethane, simply because thermodynamic data are readily available forthis molecule. The organonitrogen compound would be formed along with CO(or CO2) from the reaction of propene, O2 and NO equation (2). The sub-sequent oxidation of this organo-NOx species would yield NO2 equation (3):

C3H6 þ 32O2 þ 2NO! 2CH3NO2 þ CO ð2Þ

2CH3NO2 þ 72O2 ! 2NO2 þ 2CO2 þ 3H2O ð3Þ

Overall, the equations above combine to give:

C3H6 þ 5O2 þ 2NO! 2NO2 þ 2CO2 þ COþ 3H2O ð4Þ

While it is likely that the actual mechanism occurring on the low loadingAg materials involves more than one organonitrogen species according to amore complex reaction scheme, these equations are important in thesense that the reactions are strongly exergonic over the temperature rangeinvestigated here. The standard Gibbs free energy of reaction at 813Kassociated with eqn. 4.3.2 and 4.3.3, are DrG1¼ � 298 kJmol� 1 and� 734 kJmol� 1, respectively. The global reaction equation (4) is thereforealso strongly favoured and provides a rational explanation for the highNO2/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 reactionof hydrocarbon, dioxygen and nitric oxide in the liquid or gas phase.26 Thedecomposition products of organo-nitrogen species yield similar productsto those observed during the SCR reactions (e.g., cyanide, isocyanates),supporting their role as intermediates. NH3 can be obtained from nitro-methane through the tautomerisation to the corresponding oxime followedby dehydration to a nitrile N-oxide equation (5) which isomerise to an

Catalysis, 2010, 22, 94–118 | 101

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 organo-nitrile N-oxide were formed from organo-nitroso compounds, via enol andcyanide formation (eqn. 7, only the N-containing fragment is shown).

�CH2�NO2 ! �CHQNOðOHÞ ! �CQNQOþH2O ð5Þ

�CQNQO! �NQCQO! �NH2 ! NH3 ð6Þ

�CH2�NO! �CHQNðOHÞ ! �CRN! �CQNQO ð7Þ

We have shown that the nature of the organonitrogen compounds greatlyinfluenced the nature of the products formed, as two main reactivity

1258

1306

1554

1376

1393

1447

1598

2094

2228

2902

30033390

12001600200024002800320036004000

ν / cm−1

(a)

(b)

0.1

Abs

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

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 mainsurface species (Fig. 5). In contrast, nitrito-type compounds led to NO2 andsurface nitrates and nitrites. The latter type of compounds is therefore anobvious 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 canprobably be extended to other oxide-based SCR catalysts (Fig. 6).22,23 Therole of dioxygen is intricate and paradoxal, as this oxidiser strongly favoursthe reduction of NO over most catalytic formulations. Most authors ac-knowledge that the two main functions of O2 are the oxidation of NO andof the reductant to form the various reaction intermediates.

As far as the hydrocarbon is concerned, one of the important initial stepsproposed is its oxidation to strongly bound oxidised species such as acet-ates. The acetate species or other adsorbed oxidised hydrocarbon species(e.g. acrylates)30 are then believed to react with the surface nitrates (orpossibly with gas-phase NOx) to yield organo-nitrogen species, the exactnature of these species remaining unclear. The formation of the organoni-trogen species is likely to be the rate-determining step of the reaction, as

NO (g) + O2 (g) + CxHy (g)

Inorganic NOx (ads)(several species, in

particular, monodentatenitrate)

CxHyOz (ads)(several species, inparticular acetate)

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

R-ONO (g or ads)

R-CNR-NCOR-NH2

NH3

NO2 (g)

N2 (g)

adNOx

Minor route

Fig. 6 Schematic representation of the two main reaction pathways taking place on oxide-based catalysts for the selective reduction of NO with hydrocarbons. (Reprinted fromreference 23.)

Catalysis, 2010, 22, 94–118 | 103

these can only be observed during transient experiments such as tempera-ture-programmed surface reaction monitored by in situ IR.

Like some other researchers, we have therefore proposed that the coup-ling of nitrogen atoms to form N2 could occur via the reaction between theoxidised (e.g., NO(g), nitrate) and reduced (e.g, –NCO, NH3) forms of ni-trogen. This observation stresses that the reaction mechanism is verycomplex since NO will react through a series of parallel pathways to formnumerous intermediates.

The relevance and rate of each step of the scheme represented in Fig. 6depends on the nature of the reductant, the catalyst and experimentalconditions. The overall rate-determining step and the surface concen-trations of each species will vary accordingly. For instance, the rate at whichacetates formed was shown to be dependent on the chain length of the al-kane.31 As a result, the relative surface coverage of acetates and nitratespecies can vary as a function of type of feed, catalyst and experimentalconditions used. In the case of the C3H6–SCR of NO over Al2O3, the rate ofnitrate formation is slower than their rate of consumption and as a resultacetates predominate (Fig. 7, bottom spectrum). When Ag is added to theAl2O3, the oxidation of NO to ad-NOx species is promoted and surfacenitrate species now predominate (Fig. 7, upper spectrum).

Other experimental parameters such as temperature, water vapour pres-sure will affect the weight of the reactions as reported in Fig. 6. The possibleparticipation of homogeneous reactions must also be considered, especiallyat the higher temperatures and bearing in mind that NO and NO2 areradicals.32 Yet, it appears the majority of data reported on oxides/basemetals are consistent with this scheme.

1380

1400

1600

2910

3005

1555

1400180022002600300034003800

Wavenumbers /cm−1

CO2

Absorbance

a.u.

C

H

OO1460

1460

γ-Al2O3

1% Ag /γ-Al2O3

NOx (ads)

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 NOþ 500 ppm C3H6þ 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 de-termining some of the main aspects of the hydrocarbon selective catalyticreduction of NO over oxide-based catalysts. One of the role of the Agpromoter at low loadings was to favour the formation of surface nitrate/nitrite species, but not directly NO2(g). The formation of NO2(g) occurredvia a complex reaction scheme involving the oxidation of organonitrogenspecies, probably of the nitrito-type. A second major outcome of our workwas to evidence the occurrence of another distinct reactions mechanism athigher silver loadings, similar to that taking place of platinum, based on adecomposition-type mechanism. While no unambiguous conclusion couldbe given on the true nature as reaction intermediates of the surface speciesobserved by in situ IR, the observed species nonetheless helped building arealistic model of this complex reaction pathway.

4. Spectrokinetic operando investigation of catalytic reactions

As described in the previous section, the collection and comparison ofkinetic and spectroscopic data can be useful in gaining some understandingof the mechanism of a catalytic reaction. However, data pertaining to dif-ferent techniques are usually collected on separate apparatuses, each havingits own reactor. The simultaneous collection of various spectroscopic datain a single reactor is currently receiving much attention as a means toovercome the possibility of differences in the actual experimental conditionsprevailing 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’’ refersto spectroscopic measurements of catalysts under working conditions withsimultaneous on-line product analysis. This term was used in the literaturestarting from 200235,36 with the aim to distinguish work in which on-lineactivity 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 spectroscopicdata simultaneously measured, which is particularly useful when carryingout isotopic transients. Secondly, it ensures that the activity data obtainedin the operando reactor are consistent with those observed in a conventional‘‘ideal’’ reactor. These conditions required the (often challenging) devel-opment of reaction cells operating in a kinetically relevant mode, able towithstand extreme conditions, while still allowing the electromagnetic ra-diation to escape from the reactor. In spite of the technical difficulties, thesuccess of operando techniques is such that the number of teams switchingtheir studies towards this methodology is increasing.37

The investigations reported in this section will describe how a DRIFTScell can be checked for kinetic relevance and the modification that can bemade 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 de-termine the actual role of surface species seen by DRIFTS. The water-gasshift reaction will be investigated in details and some data related to COhydrogenation will also be presented.

4.1 Development of a kinetically relevant DRIFTS cell reactor

Diffuse reflectance FT-IR spectroscopy (DRIFTS38,39) is increasingly beingused as a means to investigate the reactivity of surface species under re-action conditions, but it is usually considered only as a qualitative techni-que. However, it was demonstrated that DRIFTS spectroscopy can be anaccurate quantitative tool for operando studies, providing that an appro-priate analytical transformation of the diffused intensity is used (i.e. in mostcases the pseudo-absorbance rather than the Kubelka-Munk function40)and that a calibration curve relating band intensity to adsorbate concen-tration is available.41 We also showed that an appropriately modifiedDRIFTS cell reactor19 led to reaction rates identical to those measured in alinear quartz tube plug flow reactor.41 DRIFTS reactors are particularlysuited to operando investigations since the catalyst powder can be used assuch, whereas FTIR-transmission techniques require pressing wafers, whichcan lead to mass-transport limitations and catalyst modifications.

The environmental DRIFTS chambers proposed by Spectra-Tech havebeen widely used to carry out in situ and operando analyses. It has beenknown for many years42 that these cells present some bypass of the catalystbed. As a result, a large proportion of the incoming gas directly goes to theoutlet, without passing through the catalyst bed located on top of the cer-amic crucible. This problem is made worse by the fact that the porous fritused to support the catalyst presents a very high pressure drop. More, in thecase of the high-pressure cell, the ZnSe dome can be very close to the rim ofthe crucible limiting its accessibility. As a result, the volume delimited by thecatalyst bed and the void above it can become a dead-zone, with very slowdiffusion pathway to the circulating gas passing through the cell. The extentof bed bypass can be assessed (i) by comparing the catalytic activitymeasured in the DRIFTS reactor to that measured in a traditional tubularreactor and (ii) measuring the time needed to purge the bed area, moni-toring an IR-sensitive tracer.19

The DRIFTS cell that was used throughout this section was the high-temperature model, in which flat ZnSe windows are several millimetersapart from the crucible, therefore not limiting the access to the catalyst bed.The bed bypass when using a flowrate of 100mlmin� 1, which was theflowrate value typically used during our operando experiments, was esti-mated by measuring the extent of CO conversion during oxidation with anexcess 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 takingplace. The oxidation data showed that almost 80 % of the feed bypassed thecatalyst bed in the case of the original cell, as the CO conversion leveled offat ca. 20% after the light-off.19

The cell was modified by replacing the original crucible with a custom-made 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 in-stead of the ceramic porous frit of the original cell. The catalyst powder wassieved using the same mesh as that located in the crucible and only thefraction 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 theintensity of the IR signal obtained.

The gap between the crucible stem and the metallic base was sealed bywrapping the lower part of the crucible stem with Teflon-tape before in-serting into the base. The Teflon tape was in close contact with the ther-mostated cell metal base, the temperature of which always remainedmoderate (typically less that 120 1C even when the crucible itself was heatedup to 400 1C). Therefore, the PTFE tape did not show any significant levelof degradation under standard conditions of use. A level of CO combustionhigher than 98.5% was achieved above the temperature light-off on themodified system, indicating that the bed bypass was negligible. A furthervalidation of the kinetic relevance of our modified DRIFTS cell come fromthe fact that the catalytic activity measured during the WGS reaction usingthe modified cell was shown to be equal to that measured in a conventionaltubular plug-flow reactor.41 Note that no conversion of reactant was ob-served when the crucible was heated up to the reaction temperature in theabsence of the catalyst.

The DRIFTS and mass spectrometry (MS) data collected during ourDRIFTS-MS experiments actually relate to different regions of the reactor.The DRIFTS data are collected at the front (and top) part of the bed, whilethe MS data are collected at the exit of the cell, i.e. after the bed (Fig. 8a). Ifthe evolution of the MS signals of reactions products during an isotopicswitch is to be compared to the evolution of the DRIFTS intensities ofsurface intermediates, then the gas phase all along the catalyst bed should be

H2+

CO2adsorbates

CO + H2O

Pt /CeO2

DRIFTSsignal

ca. 0.2 mm

2 mm

CO2 + H2 MS signal

0

0.2

0.4

0.6

0.8

1

0 100 200 300 400 500Time after isotopic switch /s

Rel

ativ

e M

S or

DR

IFTS

inte

nsity

Kr

13CO2

(IR)

x (MS)

(a) (b)

Fig. 8 (a) Schematic representation of the catalyst bed and the zone probed by the DRIFTSbeam. The MS data are collected at the exit of the DRIFTS cell, i.e. after the bed. (b) Relativeintensity 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 understeady-state WGS conditions over a 2% Pt/CeO2. T¼ 473 K, feed: 1% 13COþ 10% H2O in2% Kr/Ar. The feed was 1% 12COþ 10% H2O in Ar before the switch. (Reprinted fromreference 19.)

Catalysis, 2010, 22, 94–118 | 107

homogeneous (to avoid chromatographic effects). This is to ensure that thevariation of the gas-phase composition at the front of the bed, which dir-ectly relates to the surface species observed by the DRIFTS, is equivalent tothat measured by MS after the bed.

It is therefore important to use differential conditions throughout the bedby keeping a conversion lower than ca. 15%. In this way, the concentrationof surface species should be essentially constant throughout the catalystbed. The fact that only a fraction of the bed volume is analysed by DRIFTSstill allow doing quantitative analysis knowing the full bed mass, providingcalibration curves are available based on the very same sample impregnatedby known concentrations of adsorbates (see Section 4.5). The DRIFTSsignal of the surface species of interest measured under reaction conditionand that measured over the calibrated samples are related to a same analysisvolume (or mass) and knowing the concentration (in mol/g) in the sampleallows deriving the full amount of sorbate present (in mol) using the actualsample mass. It is clear that using a different sample for calibration pur-poses (e.g. with a different particle size, surface area or composition) wouldlead to a different calibration curve, as shown in reference 45, and does notallow quantitative insights. Bed packing must also be carried out in a re-producible manner, which is usually the case when using the same operator.

In an example treated here based on water-gas-shift data, we are fortu-nate that the reaction product CO2 can be observed both by using theDRIFTS signal and by the MS. Note that the bands measured by DRIFTSare related to the surface species present in the upper volume of the catalystbed (typically at less than 200 microns depth for a total bed length of ca.2mm), while the signal of CO2(g) is both coming from the same volume andthe free gas volume above the bed.

An isotopic exchange of CO2 was followed during a SSITKA-DRIFTS-MS experiment over the Pt/CeO2 catalyst (see subsequent sections for thedescription of the method). The CO conversion was 10% in these con-ditions, ensuring differential conditions. The evolution of the 13CO2 signalsfollowing the switch to the 13CO-containing feed is given in Fig. 8b. The MSsignal of the Kr tracer is also reported for the sake of completeness. The Krprofile (Fig. 8b) (and that of the reactant CO, not shown here) was essen-tially a step-function in comparison with the CO2 signals, indicating that thevariation 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 limitedby a very slow gradual introduction of the labeled gas (as the switchingvalve is located before the CO mass flow controller). The disadvantage ofthe technique used by Jacobs and Davis is that all processes with a timeconstant lower than that of the supply of the labeled compounds (severalminutes) will not be resolved, while the time constant of our method withthis modified cell is ca. 7 s.

It is clear that the relative variations of the DRIFTS and MS signalsassociated with 13CO2 (Fig. 8b) were identical, displaying an almost ex-ponential increase with a 50%-exchange time of about 55 s. These dataunambiguously 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 atthe bed entrance, which was measured using the DRIFTS signal. Therefore,the gas-phase profile in the thin (ca. 2mm) catalyst bed was homogeneous,in the reaction conditions used here. This fact justifies the comparison of thecurves obtained by DRIFTS for the surface species at the top of the bed andthe curves obtained for the products of reaction by MS (or any otheranalytical techniques) after the reactor, at least as long as differential con-ditions apply.

4.2 Quantitative DRIFT analysis of adsorbates concentrations

There is a widespread misconception that the IR signal measured in thediffuse reflectance mode of species adsorbed at surfaces must always bereported as Kubelka-Munk units, which are given by the equation below(eqn. 8), in which RN¼ I/Io is the reflectance of the sample, that is the ratioof the intensity diffused (noted I) to that incident (noted Io):

FðR1Þ ¼K

S¼ ð1� R1Þ2

2R1ð8Þ

In fact, a different expression (eqn. 9) has been proposed to account forthe concentration of the adsorbates,44 in which R is the reflectance measuredin the presence of the adsorbate:

Mathyshak� KrylovðRÞ ¼ ½c� ¼ aðR1 � RÞ

R1

1

R�R1

� �ð9Þ

Based on these equations, we have shown40 that the relative absorbance,i.e. ¼ � log R0 with R0 ¼ relative reflectance¼R/RN, is a more linearfunction of the surface concentration than the Kubelka-Munk function inthe range of relative reflectance R0 comprised between 100 and 60%. Thisrange of relative reflectance is pertinent to most DRIFT studies of surfacespecies, for which the signal loss due to the absorption of the adsorbedspecies is weak. The use of the absorbance function also overcomes theproblem associated with baseline drifts during measurements.40 It is only inthe case of low values of reflectance that using the Kubelka-Munk trans-form may be more appropriate.

Further evidence that units of absorbance are appropriate when investi-gating surface species by DRIFTS is given by the data reported in Fig. 9,which relate to the decomposition of 12C-containing surface formates spe-cies during the water-gas shift reaction over a ceria-based catalyst (videinfra). The intensity of the formate bands was expressed in absorbanceunits. An almost perfect exponential decay was observed (Fig. 9a), as ex-pected in the case of first order processes, since the corresponding semi-logarithmic plot yielded a straight line (Fig. 9b). As a consequence, theabsorbance units were always used during our investigations whenever anyquantitative work was carried out. Another a common misconception isthat DRIFTS work cannot be quantitative, contrary to the case of trans-mission IR data. On the contrary, we were able to draw calibration curvesto accurately quantify the concentration of formates and use these curves toquantify the concentration (in mol/g) during our operando work.41,45

Catalysis, 2010, 22, 94–118 | 109

4.3 The operando SSITKA-DRIFTS-MS method

Spectroscopic studies are more powerful when combined with isotopictransient methods (SSITKA46,47), which allow operating at the chemicalsteady-state. The operando DRIFTS-SSITKA method described here relieson using a single catalytic bed, which allows the characterisation by DRIFTspectroscopy of the surface of the very same catalyst particles that are re-sponsible for the catalytic activity measured at the exit of the cell by gas-chromatography or mass-spectrometry.48 The group at Queen’s UniversityBelfast was the first to report data coupling in situ DRIFTS with activitymeasurement using a mass spectrometer (MS) during a steady-state isotopicexchange kinetic analysis (SSITKA) using a single bed reactor. Thismethodology is similar to that developed earlier for transmission FTIR byChuang et al.14,49 Note that mass transport limitations may occur withwafers and that the temperature control can also be difficult (as is forDRIFTS cells at high temperatures). These techniques derived from the so-called ‘‘isotopic jump’’ technique of Tamaru et al.,5 which relied on a two-bed IR cell (see Section 2 of this report).

The principle of the DRIFTS-MS-SSITKA method that was developedfor the operando investigation of catalytic reactions is schematically rep-resented in Fig. 10. The use of SSITKA method allows us assessing thechemical reactivity of surface species with respect to the formation of areaction product under chemical steady-state conditions. This techniqueconsists in replacing one of the reactants (e.g. 12CO) by an isotopomer (e.g.13CO) during reaction and following simultaneously the exchange of thelabelled reaction product (here 13CO2) by mass-spectrometry and the sur-face species (e.g. 13 or 12C-containing carbonyl and formates) by DRIFTS.The DRIFTS bands of the surface species typically shift in wavenumbersduring the analysis, and various integration methods can be used toquantify accurately the band heights or areas.19

A surface species observable by IR can only be a main reaction inter-mediate if it exchanges at least as fast as the reaction product in the

0

0.2

0.4

0.6

0.8

1

0 500 1000 1500

Time after isotopic switch /s

Prop

ortio

n of

12C

-con

tain

ing

surf

ace

form

ates

-2

-1

00 500 1000 1500

Time after isotopic switch /s

Log

10 o

f th

e pr

opor

tion

of 12

C-

cont

aini

ng s

urfa

ce f

orm

ates

(a) (b)

Fig. 9 (a) Relative DRIFTS signals measured in absorbance units associated with the 12C-containing formate species following a 12CO–13CO isotopic switch under steady-state WGSconditions over a 2% Pt/CeO2. T¼ 473K, feed: 1% 13COþ 10% H2O in 2% Kr/Ar. Thesample was initially at steady-state under 1% 12COþ 10% H2O in Ar. (b) Logarithmic plot ofthe data reported in (a). The dotted line is a straight line used as an eye guide. (Reprinted fromreference 19.)

110 | Catalysis, 2010, 22, 94–118

gas-phase (e.g., species I in Fig. 10). A significantly slower exchange or noexchange at all indicates a minor reaction intermediate and a spectatorspecies, respectively (e.g., species S in Fig. 10). Using a single bed and dif-ferential conditions ensures that any observed variations of surface speciesconcentrations can be related to that of products in the gas-phase. We havereported the first use of this setup in an investigation of the reverse water-gas shift (RWGS) reaction:48

RWGS : CO2 þH2 ) COþH2O ð10Þ

The importance of using chemical steady-state conditions for the study ofthe reverse water-gas shift reaction was highlighted during our investigationcomparing the rate of removal of surface species during a purge in an inertgas 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 thedifference of reactivity observed was related to the difference in the oxi-dising/reducing nature of the feed, which altered ceria surface oxidationstate, which in turn modified the strength of the adsorbate bonding to theceria surface.52

Another example of the importance of using chemical steady-state wasrecently reported by Mims and co-workers in the case of methanol synthesisover Cu/SiO2 catalysts.

53 The authors measured the rate of decompositionof the formates formed at the surface of Cu supported on silica duringmethanol synthesis from CO/H2 mixtures. The formate decomposition rate

Fig. 10 Schematical representation of the DRIFTS-MS-SSITKA technique for the operandoinvestigation 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 phaseproducts 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 isa ‘‘spectator’’ (better called a minor intermediate). I* and S* are the corresponding labeledsurface species.

Catalysis, 2010, 22, 94–118 | 111

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

4.4 Operando SSITKA-DRIFTS-MS study of the RWGS

An example of the application of our DRIFTS-MS-SSITKA technique isgiven below for the RWGS reaction.48 Fig. 12 shows the typical DRIFTspectra of formate, carbonyl and carbonate species during the isotope ex-change 12CO2 to 13CO2 of the reactant. The replacement of 12C with aheavier isotope led to a red-shift of most of the wavenumber of the vibrationof the surface species of interest. Note that we were able to integrate in anunequivocal manner the concentrations of the formate, carbonyl and car-bonate species.

The corresponding exchange data of the surface species are reported inFig. 13, which showed that carbonates and carbonyl species were typicallyhalf-exchanged in about 60 s. On the contrary the exchange time of formatespecies was much longer, i.e. about 11min. The corresponding MS datashowing 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 car-bonyl and carbonates, and therefore these species are potentially main re-action intermediates, contrary to the formates. Formates should be namedas minor intermediates, rather than spectator species, since those still ex-changes and probably led to some CO, albeit at a much lower rate thanthose formed via the other surface intermediates.

0

0.2

0.4

0.6

0.8

1

0 5 10 15 20 25 30 35 40Time /min

Rel

ativ

e D

RIF

TS

inte

nsity

Fig. 11 Relative intensity of the IR bands of the formate (7,3), carbonyl (’,&) and car-bonate (�,�) species as a function of time on stream under Ar (solid symbols) and underRWGS stream containing 13CO2 (open symbols). The sample was at steady-state state in 1%12CO2þ 4% H2 and T¼ 498K before switching to either Ar or the 13CO2-containing feed.(Reprinted from reference 50.)

112 | Catalysis, 2010, 22, 94–118

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 : COþH2O) CO2 þH2 ð11Þ

The comparison of the exchange curves of the reaction product (hereCO2) and formate was more intriguing that in the case discussed above. The

2700280029003000

Wavenumber / cm-1

0.05 log 1/R

2841 cm-1

2947 cm-1

2825 cm-12916 cm-1

a

f

dcb

e

1800190020002100

Wavenumber / cm-1

0.1 log 1/R

2057 cm-1

1977 cm-1

2010 cm-11904 cm-1

a

f

dcb

e

800820840860880900

Wavenumber / cm-1

0.1 log 1/R 851 cm-1866 cm-1

831 cm-1

862 cm-1

a

f

dcb

e

Fig. 12 Typical DRIFTS data of a DRIFTS-MS-SSITKA experiment during the RWGS overPt-CeO2: exchange of (top) formate (middle) carbonyl and (bottom) carbonate species atvarious times after an isotope exchange 12CO2 to

13CO2. Feed: 1% CO2þ 4% H2. (Reprintedfrom reference 41.)

Catalysis, 2010, 22, 94–118 | 113

formate exchange was significantly slower than that of CO2 at 160 1C,suggesting that formates were unimportant reaction intermediates at thesetemperatures (Fig. 14). However, the exchange of these two species wasessentially identical at 220 1C, suggesting that formates could potentially bea 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 ismore precise than that obtained by MS) and the specific rate of formatedecomposition. The latter was calculated as the product of the formate

0

0.2

0.4

0.6

0.8

1

0 5 10 15 20 25 30

Time (min)

Rel

ativ

e IR

/ M

S in

tens

ity (

a.u.

)

Formate

Carbonyl

Carbonate

13CO

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

0

0.2

0.4

0.6

0.8

1

0 5 10 15 20Time / min

Rel

ativ

e IR

and

MS

sig

nals

Formate, 220 °C

Formate, 160 °C

CO2, 220 °C CO2, 160 °C

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

114 | Catalysis, 2010, 22, 94–118

concentration by the pseudo-first-order rate constant (noted k) of formatedecomposition:45

Rate of formate decomposition ¼ k� ½formate� � 100%

The value of k was determined from the formate exchange curve fol-lowing the isotopic switch, which showed a first order decay (i.e. ex-ponential curve) (see Fig. 14). The formate concentration was determinedunder reaction conditions via the calibration curves described else-where.41,45 The term 100% is introduced to indicate that we arbitrarilyassumed that the formate totally decomposed to CO2þH2 and not at all toCOþH2O. Therefore, the decomposition rate of the formate species yieldedthe upper limit of the rate of formate decomposition to CO2, since it is likelythat formates were not only decomposing to give CO2, but instead gave amixture of CO and CO2. Even assuming that formates decomposed solely toCO2, the rate of formate decomposition for highly active Pt/CeO2 WGScatalysts accounted for less than 10% of the total rate of CO2 formation inthe range of temperatures and experimental conditions used here, respect-ively (Fig. 15). Therefore, the formates seen by DRIFTS are minor reactionintermediates over this very active WGS catalyst. These observationsstresses that the similarity of the time constant of formates and CO2 is not asufficient condition to guarantee that formates seen by DRIFTS are mainreaction intermediates.

In some cases, formates seen by IR have been also considered essentiallyas ‘‘buffer’’ surface species;55 whether those participate or not to the mainreaction pathway is unclear. In any case, while our work cannot indicatewhat the main WGS reaction pathway is, it certainly stresses that observingformates displaying some sort of reactivity is not a sufficient criterion toelect those as important reaction intermediates as proposed in manyinstances.56,57 Of course, our conclusions are based on the samples thatwe investigated under our experimental conditions; therefore any extrapo-lation outside this system is hazardous. DFT-based work carried out both onceria-supported noble metals and copper seems to favour the role of

Rat

e /1

0-6 m

ol s

-1g-1

CO2 formation

Formate

decomposition

0

2

4

6

8

10

12

160 °C

Reaction temperature

220 °C180 °C

Fig. 15 Rate of CO2 production and rate of formate decomposition over the Pt/CeO2 at threedifferent temperatures under 2% COþ 7% H2O in Ar. (Reprinted from reference 45).

Catalysis, 2010, 22, 94–118 | 115

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

4.6 Conclusions on the spectrokinetic investigation of catalytic reactions

A reliable study of a catalytic system using spectroscopic tools must to takeinto account the following points:

i. It is crucial to check the kinetic relevance of the spectroscopic cell to beused for operando work. Appropriately modified DRIFTS reaction cellscan behave as true kinetic reactor.

ii. DRIFTS work can be performed in a fully quantitative manner. Inmost cases, the absorbance units are more appropriate than Kubelka-Munkunits.

iii. Chemical steady-state must be used to determine the true operandoreactivity of surface species, particularly for catalysts whose compositionare affected by the reaction conditions.

iv. The DRIFTS-SSITKA-MS is a truly operando technique and proveduseful in unravelling the (minor) role of formate species seen by DRIFTSfor the water gas shift reaction or CO hydrogenation over our samples. Thenature of the true reaction intermediates remains unknown over thesematerials.

5. Overall conclusions

The examples reported here show that combining spectroscopic and kineticstudies can greatly benefit to the understanding of heterogeneous catalyticreaction. The example dealing with the selective reduction of NOx showedthat reaction steps and/or elusive reaction intermediates could be postulatedby the observation of side-products and the reactivity of model compounds.The use of fully quantitative methods also revealed that ‘‘slowly’’ reactingsurface species (i.e. spectators) can sometimes be easily mistaken for truereaction intermediates. The conclusions derived from in situ/operandospectroscopic work should therefore always be carefully thought throughand the corresponding limits of validity clearly defined.

We must be aware that far too often species observed by in situ/operandospectroscopic analysis are spectators or minor reaction intermediates (be-longing to a slow parallel pathway), and it is careless to jump to the con-clusion that these species are important, simply because those can be seenand 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 reactionpathway’’ and a proper quantitative spectrokinetic analysis is then carriedout to relate the specific rate of decomposition of these surface species to thespecific 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. Mir-

odatos, 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 andH. 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. Martınez-Arias, M. Fernandez-Garcıa, 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. Banares, 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 alkanesto olefins

K. Seshana

DOI: 10.1039/9781847559630-00119

1. Introduction

In the current scenario of decreasing oil reserves, light alkanes offer choiceas feedstock for chemicals, fuels and energy. They are available as NaturalGas (NG, methane, ethane) and Liquefied Petroleum Gas (LPG, propane,butanes) in the earth’s crust. Gas reserves are expected to last for a numberof 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 synthesisof: (i) bulk chemicals, e.g., ethylene oxide, acrolein (ii) polymers, e.g., poly-ethylene, -propylene or -butenes or (iii) fuels such as diesel, gasoline, e.g., bybutene/butane alkylation.

Direct synthesis of these bulk products from alkanes is still not exten-sively commercialized maleic anhydride from butane being one of the fewexample.4 Alkane conversion to olefins is therefore a key step in themanufacture of bulk chemicals and fuels. As the title suggests, this review isrestricted in scope to oxidative alkane conversions to lower olefin, which areintermediates to fuels e.g., gasoline, and chemicals e.g. oxygenates such asacrolien, acetone, maleic anhydride etc, via various chemical routes aspracticed in a refinery/petrochemical complex.

Direct/single step conversion of alkanes to such chemicals is equally inte-resting from the scientific and economic point of view. There has been tre-mendous amount of attention in the academic and industrial laboratories forthe development of efficient catalysts for such routes. This topic is beyond thescope 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 ofmethane to methanol or formaldehyde,5–7 ethane to acetic acid/acetal-dehyde8–10 (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 –C=

4 ) is expected to beinsufficient, as the demand for these important components in the modernrefinery/petrochemical industry grows.18 These light olefins are currentlyproduced using fossil oil, e.g., from catalytic or steam cracking of naphtha andassociated gas, or from fluid catalytic cracking of vacuum gas oil. While thesetwo routes are well developed, increasing the capacity of these processes is onlypossible to some extent, as changing regulations limit the use of byproducts(notably aromatic molecules) in fuel. In this context, methane as a feedstockhas drawn wide attention. Tremendous efforts both at academic and industriallaboratories in the 80’s and 90’s to convert methane to ethylene via catalyticoxidative coupling did not lead to commercial success.19,20 The research failed

aCatalytic 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 maximumyields 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 CHd

3 or CHd2 radicals,

and (ii) coupling of these radicals to produce ethane or ethylene, respect-ively. The former is an endothermic reaction and considering the strength ofthe C–H bond (232 kJ/Mol) in methane, requires very high temperatures(W750 1C).22 Unfortunately, the C–C coupling is an exothermic reactionfavored at lower temperatures. There is thus a fundamental barrier to at-tempts to convert methane via oxidative coupling, and the failure to achieveappreciable ethane or ethylene yields is not surprising. Methane activationat lower temperatures where the exothermic coupling is more favorablecould 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 theproduction of C2 to C4 olefins. Steam cracking follows a radical chemistryroute, the carbon radicals (primary or secondary) formed initially via C–Hbond cleavage result in smaller primary radicals after subsequent b-cleavage(see Fig. 1). As radicals do not undergo isomerisation, every furtherb-cleavage of the primary radicals formed results in C2 product. Steamcracking therefore maximizes ethylene yields, and this route becomes lessattractive when propylene demand grows faster than that of ethylene.23

Catalytic cracking over zeolitic and other solid acid catalysts, on the otherhand, follows an ionic mechanism (Fig. 2) involving carbo-cations (carbe-nium ions mostly, CnH

þ2nþ 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, whichon b-cleavage lead to olefins with three or more carbon atoms depending onthe carbon chain length in the reactant hydrocarbon. Catalytic cracking isthus an option for higher olefins, however, most of the capacity in FluidCatalytic Cracking units is used for fuels, e.g. gasoline production.

Dehydrogenation of alkanes to olefins is an excellent option for threereasons: (i) alkanes are cheap feedstocks (ii) it has the advantage that itgenerates olefins, e.g.,

C3H8 ! C3H6 þH2 DH ¼ 117 kJmol�1

with the same carbon number as the alkanes and (iii) byproduct hydrogen isin 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 thermodynamicequilibrium, and there is a strong tendency to coking and consequentlycatalyst deactivation can be severe, leading to short lifetimes.24–28 Catalystlifetimes are indeed small in practical commercial applications, e.g., eighthours to a few seconds (continuous catalyst regeneration) are reported. Theexisting processes for the dehydrogenation of light paraffins such as OLE-FLEX (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 consistof catalyst regeneration (i.e., carbon burn-off) in combination with heatintegration. These processes appeared in the early 80’s but have made onlylimited breakthrough commercially.

Continuously increasing global demand for light alkenes has, therefore,spurred substantial interest in the development of alternative routes in-volving light alkanes as feedstocks. The incentive for alternative processes ispredicted to grow in the future as the increase in the availability of lightalkenes from refineries is expected to be quite limited.29 Therefore oxidativeconversion of alkanes to olefins presents an available option. This isachieved via selective combustion of the hydrogen formed in the conversionof alkanes to olefins,

C3H8 þ 1=2O2 ! C3H8 þH2O DH ¼ �126 kJmol�1

and the reaction is termed oxidative dehydrogenation.30 The major ad-vantages of oxidative dehydrogenation over conventional dehydrogenationis that it: (i) overcomes the thermodynamic equilibrium limitations on olefinyield faced in the direct catalytic dehydrogenation (ii) minimizes the cokeformation and the related catalyst deactivation during conversion due to thepresence of oxygen and (iii) avoids the need for heat input, as the reaction isexothermic and can be run adiabatically and at lower temperatures.30,31

Oxidative conversion (coupling) of methane was extensively studied, asmentioned earlier, however olefin yields have not been impressive enoughfor commercial applications. Oxidative dehydrogenation of ethane has alsobeen studied by more or less the same researchers and over similar catalystsalso without success.32 Currently, there is tremendous interest in the oxi-dative conversion of propane and butane.

Oxidative dehydrogenation of alkanes to olefins is therefore still at a de-velopmental stage despite the enormous amount of research and develop-ment activities carried out by academic/industrial laboratories. Nocommercial process is operative at the moment. Development of catalyststhat minimize combustion and maximize olefin yields is still the bottleneck.33

This review addresses the efforts over the last few years, by us specificallyand by others, toward characterizing the catalyst materials and discerningtheir relation to the kinetic and mechanistic sequences involved in theconversion of alkanes to olefins in the presence of oxygen. This review isspread into three general areas. Many catalysts reported are essentiallyoxides,34 and they operate via a Mars van Krevelen ‘redox’ type mechanismand generate olefins with the same carbon number. These are discussedfirst. Recent studies on oxidic catalysts with no formal ‘redox’ propertieshave 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–Cbond cleavage is also observed and this results in a mixture of olefins withdifferent carbon numbers. This process is termed oxidative cracking anddiscussed next. Finally, recent efforts in alkane activation using cold plasmaat ambient conditions in micro reactors shows that oxidative C–C bondcoupling is possible at these conditions. This leads to molecular weightbuildup, i.e., the formation of olefins with a higher C-number than thestarting alkanes, and promises exciting new chemistry and applications.This is discussed last. A few thoughts on futuristic scenarios complete themanuscript.

2. Oxidative conversion of alkanes to olefins over oxide catalysts with

redox properties

Among the alkanes, methane conversion to olefins has been the moststudied process. Most of these developments took place in the 80’s and90’s. All sorts of oxidic catalysts were attempted, with alkali and alkalineearth oxides the most studied.1,2 In the case of methane, attemptswere focused in the catalyst/oxygen assisted cleavage of C–H bonds toyield CH2, CH3 radicals. Coupling of these radicals was expected to yieldethylene or ethane, respectively, and the process was termed oxidativecoupling. In spite of the enormous amounts of research and development,C2 yields were limited (o30 mol%), breakthroughs did not appear andefforts stopped in the 90’s. The major barrier to conversion was that se-lective C–H bond activation in methane required very high tempera-tures (700–800C), even in the presence of a catalyst and oxygen. At thesehigh temperatures, radical coupling, which is essentially exothermic, wasnot a facile process and hence C2 yields never were appreciable. We do notattempt to go into these details and readers are directed to excellentreviews from M. Baerns, J.R.H. Ross1 and Lunsford.2 Later in thismanuscript, we will show that if the activation of C–H bonds in methanecan be achieved at lower temperatures, C–C coupling reactions can befacilitated.18

Ethane is a relatively cheap feedstock and is mostly associated with me-thane in the natural gas stream. As in the case of methane, ethane con-version focused on (oxy)dehydrogenation to ethylene. Most of the researchwas carried in the same laboratories that were studying oxidative couplingof methane. The catalysts studied were therefore similar to those attemptedfor methane.32 Again, no significant breakthroughs were made. Resultsfrom Union Carbide on V-Nb-Mo based catalysts showed initial promise(very high ethylene selectivities were reported), but did not lead to com-mercial success.35,36

Conversion to syngas (COþH2) by oxidation with water (steam re-forming), carbon dioxide (dry reforming) or oxygen (partial oxidation) re-mains still the main route for the valorization of these alkanes, especiallymethane.

The ever increasing demand for higher olefins such as propene andbutene, their limited availability from steam/catalytic cracking operationsand 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 stocksis still an attractive option.

Most of this work was focused on vanadia- or molybdena-based oxidesystems which undergo redox changes during the conversion. Acidic (Al2O3,SiO2) and basic (MgO, CaO) oxides have been used as supports. Basicsupports such as MgO have been the most beneficial, because they minimizere-adsorption of product olefin and thereby minimize combustion. Rela-tively high yields were reported for Mo-Ni-O catalysts,37 (30% conversionof 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 strongolefin adsorption (on cationic Lewis acid sites) and the subsequent oxi-dation to carbon oxides. This is clearly visible in the results shown inFig. 3.31 This behavior is typical for almost all the catalysts studied. Athigher propane conversions, sequential oxidation of the propylene productto carbon oxides becomes significant. Among the vanadia based catalystsreported, use of niobia as a support has been shown to improve propyleneselectivity31 up to 80mol%. Niobia is a non acidic support it is reported tolose all its Bronsted and Lewis acidity when calcined above 350 1C.38 Theabsence of acidity clearly reduces product olefin re-adsorption, and min-imizes its sequential combustion. Such catalysts give higher olefin selectiv-ities, but the presence of residual acidity, which cannot be completelyremoved, is still detrimental. However, even on these catalysts, higherpropane conversions lead to loss of olefin selectivity, as can be seen fromFig. 4. Higher temperatures give better yields, but the secondary com-bustion of olefin cannot be avoided.

Lattice oxygen in these oxides participates in the oxidation (oxidativeabstraction of hydrogen) of the alkane. It is suggested by Kung et al.34 thatthe C–H bond is split homolytically over the vanadia catalysts, creatingpropyl radicals. Kung et al. further elucidated the role of homogeneousradical reactions in oxidative dehydrogenation of propane over V-Mg-Ocatalysts, revealing the contribution of each component (homogeneous and

0 2.5 5 7.5 100

2

4

6

Contact time [s]

Yie

ld [

mol

%]

Propene

CO2

CO

425 °C

Fig. 3 Oxidative dehydrogenation of propane over V2O5/MgO catalyst, influence of contacttime (conversion) on yields.

Catalysis, 2010, 22, 119–143 | 123

heterogeneous) in the overall process.39 The hydroxyls thus formed arereleased to the gas phase as water, reducing the catalyst. Alternatively, thelattice oxygen also takes part in the combustion of the alkane. Re-oxidationof the catalyst by gas phase oxygen completed the catalytic cycle and re-generates the catalyst.

The nature and concentration of surface active sites determine propaneconversion and product selectivity (olefin vs. COx). Low Energy Ion Scat-tering is a useful technique to determine surface concentrations of oxidespecies. Compared to XPS, which probes a few atom layers, LEIS allowsmeasurement at the outermost layer. Hence it is an optimal method todetermine activity correlations. Fig. 5 shows a correlation between intrinsicactivity (TOF, s� 1) for propane conversion as a function of surface vanadiaconcentration for a series of catalysts determined by LEIS.40 An increase in

0 5 1 15 200

4

8

Propane conversion [mol %]

Prop

ene

yiel

d [m

ol %

]

V-Nb-O bulk

V-Nb-O monolayer

Vanadia

Niobia

425°C

Fig. 4 Conversion–yield plots for a series of vanadia based catalysts at 425 1C.

0

1

2

3

4

0 10 20 30 40 50 60

Intr

insi

c ac

tivit

y pe

r si

te(A

rbitr

ary

Uni

ts)

Surface V concentration (mol %)

Fig. 5 The intrinsic activity per V site as a function of the surface V site concentration de-termined by LEIS. The intrinsic activity per site is determined by dividing the propane con-version by the surface V concentration and is normalized to 1 for the lowest activity.

124 | Catalysis, 2010, 22, 119–143

intrinsic activity (activity per site) as a function of vanadium concentrationsuggests that different vanadia sites of varying catalytic activity must bepresent at the surface.

51V solid-state NMR allows one to discriminate various vanadia species41

based on chemical shift anisotropies (see Fig. 6). Our measurements showedthat three types of vanadia species were present i.e., vanadium in an isolatedtetrahedral oxygen environment, in corner sharing tertrahedra, and in dis-torted octahedra. The number of octahedral sites increased with increasingvanadium concentration. Thus when the vanadium concentration is chan-ged the types of species on the surface and their concentrations change.Larger vanadia clusters (Q1, Oct) have linking oxygen bonds between themwhich are susceptible to redox changes under the oxidative dehydrogenationconditions. Vanadium in different oxygen environments and their differentredox natures would be the reason for the changes in intrinsic activitieswhen catalysts with different vanadia loadings are considered.

Oxides typically contain a variety of oxygen species, e.g., vanadia exhibitsboth V–O–V bonds holding the vanadia octahedra together or danglingVQO bonds. These bonds can be easily characterized by IR spectroscopy.There is debate over the role of these oxygens in the selective oxidation andcombustion of propane. It has been claimed that VQO bonds lead tocombustion while V–O–V bonds are involved in selective oxidation toolefins.42 This is based on the observation that V2O5, which exhibits VQObonds, mostly gives combustion. Supporting V2O5 on MgO results in theformation of magnesium orthovanadates, which do not have VQO bonds.These catalysts are more selective to propylene formation and combustion isminimized. Alternatively, the oxygen present in the V–O–V bond has beencharacterized as responsible for olefin formation as well as combustion.However, oxide based catalysts which show facile redox properties tend tocatalyze sequential combustion of the olefin formed. A detailed survey byCavani and Trifiro showed,37 the limits achieved olefin yields were almostalways below 30mol%. This situation has changed very little in the last few

Fig. 6 Width of the static 51V chemical shielding anisotropy as a function of the vanadiumcoordination.

Catalysis, 2010, 22, 119–143 | 125

years, and the search for an efficient catalyst to convert C3–C4 alkanes toolefins is still on.

In order to efficiently convert alkanes to olefins, the critical issue forcatalyst 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 litera-ture,19,20,33,34,37–43 a commercial breakthrough for oxidative conversion ofalkanes to olefins is still lacking. As discussed in the previous section, one ofthe major difficulties in developing efficient catalysts has been the fact thatthe product olefin is more prone to combustion than the starting alkane.45

Thus higher olefin selectivities can be obtained only at low conversions, andolefin yields decrease at higher conversions due to sequential oxidation ofthe olefin.31 Olefin re-adsorption and subsequent combustion on the re-ducible oxide catalyst is indeed claimed as the reason for the lack of excitingprogress in the field of oxidative conversion/dehydrogenation of al-kanes.23,31–33,45

One way to overcome this problem may be to minimize the olefin sorp-tion on the oxide catalyst, even modifying it to favor desorption. Olefins arenucleophilic, electron rich, and thus an oxidic surface with basic (non-acidic) properties would be a proper choice.23

Basic alkali and alkaline earth oxides have been attempted as catalysts forthe oxidative coupling of methane to ethylene9,43 and the oxidative de-hydrogenation of ethane to ethylene.32,44 One of the most studied catalystsis Li/MgO. Work in the last years shows that it is also a promising catalystsfor the oxidative conversion of propane and butane.23 As expected (Fig. 7),the Li/MgO catalyst shows olefin selectivity for the oxidation of propanethat is almost independent of conversion. This is different from the case ofredox-type catalysts reported in the previous section. It can be seen from

00 30 60

2

4

6

8

10

conversion (%)

mol

%

olefin selectivity

yield

585°C

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

126 | Catalysis, 2010, 22, 119–143

Fig. 8 that the olefin yields vs. conversion are almost linear and no sec-ondary conversion of olefins to COx is observed. This also implies that higholefin yields (50mol%) can be achieved, as seen from Fig. 8.

As reported often in the case of methane coupling, the catalytic activity ofLi-promoted MgO is determined by surface [O� ] species, and their exist-ence in MgO was mainly shown using the electron paramagnetic resonance(EPR) technique.43,46,47 Lunsford suggested that the [O� ] species was cre-ated by the substitution of Liþ for Mg2þ ions to allow charge balance, andwas stabilized in the MgO lattice as [LiþO� ] centers.48 Remarkably, [O� ]are reported to be very stable at high temperatures and can exist in thecrystal lattice of metal oxides even in the absence of oxygen in the gasphase.49

The similar ionic radii of Liþ (rLiþ=0.76 A) and Mg2þ (rMg2þ=0.72 A)allows easy accommodation of Liþ in the lattice of MgO.50 Replacement ofMg2þ by Liþ creates lattice defects, i.e., oxygen vacancies (positive holes)(Scheme below). The proposed active site [LiþO� ] is produced by a holeadjacent to a Liþ site trapping an oxygen atom.51,52

2Li0MgOxO þ VO

dþ 1=2O2 ! 2Li0MgOOdþOx

O

Scheme: Proposed mode of formation of the [LiþO� ] (shown in Kroger-Vink notation above) active site in Li/MgO catalysts. A hole trapped atthe O2� is adjacent to Liþ sites.51,52

Unlike conventional Li/MgO catalysts prepared by impregnation, sol gelmethods used recently give high surface area Li/MgO, where such defectsites are enhanced and lead to improved catalyst activity.53

Evidence for the presence of [LiþO� ] defect sites in these catalysts alsocomes from IR spectroscopy measurements.54 In the case of MgO-basedmaterials, surface sites of low coordination, i.e., Mg2þLCO

2�LC pairs can be-

have as strong acid-base pairs. The weak adsorption of CO (hence the low

0

25

50

0 25 50 75 100

Conversion (mol%)

Yie

ld (m

ol%

)

600°C

Total olefins

COx

Propene

Fig. 8 Oxidative dehydrogenation of propane over Li-MgO based catalysts.

Catalysis, 2010, 22, 119–143 | 127

temperatures required to observe them, at � 193 1C) on Lewis acid sitessuch as Mg2þLC produces IR bands at frequencies higher than the stretchingfrequency of the free CO molecule in the gas phase,55 and thus provides atool for probing these surface sites.

IR spectra of CO adsorbed on Li/MgO catalysts prepared by sol gelmethods, at � 193 1C, (Fig. 9) show evidence for the presence of such sites.In the case of 1 and 3wt% Li/MgO, the spectra show three bands at ap-proximately 2164–2168, 2152–2155 and 2146–2148 cm� 1. These bands werealso present in the spectra of CO adsorbed on MgO-sg and are assigned toCO adsorbed on Mg2þ4C single sites, Mg2þ5C single sites, and anchored toMg2þ4C and Mg2þ5C at the step site, respectively. For the sample with thehighest lithium loading (5wt%), additional weak bands at 2200 and2184 cm� 1 were also present. The band at 2200 cm� 1 is attributed to a COmolecule interacting with a Mg2þ3C site.56 Based on DFT calculations, theadsorption at 2184 cm� 1 can be attributed to the addition of a second COmolecule to the CO-Mg2þ3C adduct, resulting in the formation of dicarbonylspecies.56,57 In addition, changes in the relative distribution of all thecomponents present in the spectra were observed. The appearance of sites oflower coordination of lithium atoms due to a decrease of the particle sizecan be certainly excluded, since the addition of lithium causes a substantialdecrease in surface area and increase in particle size. Thus, sites of lowcoordination such as Mg2þ3C , which require oxygen vacancies near them,indicate incorporation of lithium in the lattice structure of MgO, creatingthe oxygen vacancies.

210521252145216521852205

Wavenumber cm-1

Abs

orba

nce

(a.u

.)

a

C=O

C=O

C=O

C=O

b

c

Mg2+5C

Mg2+5C

Mg2+4C

Mg2+3C

Mg2+4C

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

128 | Catalysis, 2010, 22, 119–143

Extensive work from the groups of Lunsford et al., and Ross et al., Seshanet al., have shown, in the case of Li/MgO catalysts, that the first step in theoxidative conversion of methane involves the homolytic scission of C–Hbonds forming surface –OH groups and alkyl radicals (1):19,23,43,46,47

½O�ðSÞ þ CH4 ! ½OH�ðSÞ þ CH3d ð1Þ

The resulting radicals are released from the catalyst surface and sub-sequently initiate gas-phase chain propagation reactions to yield products.23

In the case of vanadia-type redox catalysts, the only products obtained inthe case of propane or butane were the corresponding olefins (i.e., propeneor butene), and carbon oxides. The product distribution obtained during theoxidative dehydrogenation of propane over a Li/MgO catalyst is given inTable 1. The presence of C1–C2 hydrocarbons in the products indicates that

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 ofoxidative coupling of methane over Li/MgO catalysts have been proposed(Fig. 10). These sequences involve initiation at the catalyst surface followed bygas phase homogeneous reactions.43 Leveles et al.23 discussed the perform-ance of the catalyst tested in relation to the reaction mechanism for propaneODH and reported that propane activation occurs via a C–H bond splitting

Table 1 Oxidative dehydrogenation of propane over Li/MgO catalysts at 550 C,

propane conversion 15%

Component COþCO2 Methane C2þC=2 C=

3

Selectivity (mol%) 15 10 30 45

HO2·

O2

H2O22 HO·

·CH3C3H8

C3H8

CH4

H2

C2H4

C3H6H·

½nC3H7·

½iC3H7·

O2 COxCH2O

H2O

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[LiþO� ] active site by homolytic hydrogen abstraction, forming [LiþOH� ]and propyl radicals equation (1). There is general evidence that the two dif-ferent primary radicals formed after heterogeneous activation, n- and iso-propyl radicals, are released from the catalyst surface to the gas phase, andradical chain reactions lead to the final products. For the propyl radicalsformed, two different decomposition routes have been proposed for the gasphase: (i) the isopropyl radicals can undergo scission of C–H bond at the a-position and decompose into propylene and H an radical and (ii) the n-propylradical undergoes C–C cleavage in the b-position, forming a methyl radicaland ethylene (Fig. 10). Thus, the cleavage of a C–H bond (dehydrogenation)versus cleavage of a C–C bond (cracking) is the processes controlling theselectivity to, respectively, propylene and ethylene. At a relatively low T(o600 1C), and at a low conversion level of propane, the contribution of thegas phase reactions is less, and a propylene to ethylene ratio higher than 1 wasobserved.23 At higher temperatures (Z600 1C), a decrease in the rate of de-hydrogenation and an increase in cracking was recorded.

Surprisingly, for identical conversions, increasing the number of defect[LiþO� ] sites improved selectivity to olefins and higher propylene toethylene ratio in the products was observed. Indeed, as Kondratengo et al.reported, in the case of propane ODH over vanadium oxides systems in-creasing the density of active sites affected the olefin distribution.58 Inparticular, they suggested that a high density of oxidizing sites is essentialfor further ODH of the propyl radicals to propylene, and therefore sup-pression of the concurrent cracking reaction pathway to ethylene.

Similarly, in the case of Li/MgO-sg catalysts with a high density of activesites (per volume of catalytic bed), heterogeneous H-atom abstraction fromC3H

d7-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.

½LiþO�� þ C3H!7 ½LiþOH�� þ C3H6 ð2Þ

Propane activation on [LiþO� ] active centers, leading to the propylradical, is a single site interaction. The chance that the propyl radicals mayfurther react with a second [LiþO� ] active site leading to propylene(multiple site interactions) strongly depends on site density. The finalstep43,59 is the regeneration of the active site, which involves electrontransfer to the anion vacancy and the dissociative chemisorption of oxygen,respectively eqn. (3) and (4):

½LiþO2�� þ ½LiþVa� ! ½LiþO�� þ ½LiþV�a � ð3Þ

½LiþV�a � þO2 ! ½LiþO�� þO ð4Þ

However, certain features in the Ito-Lunsford mechanism appear to beunlikely, especially at lower temperatures. In particular, removal of oxygenfrom the lattice is not facile and this would be the rate-limiting step of thecatalytic cycle rather than hydrogen abstraction.59 Moreover, migration ofa proton would require: (i) substantial energy to overcome the electrostaticbarrier60 and (ii) the proximity of [LiþO� ] centers.61

130 | Catalysis, 2010, 22, 119–143

Alternatively, Sinev62,63 proposed a new mechanism for regeneration ofthe active sites that does not require the removal of lattice oxygen and thusthe formation of oxygen vacancies. In fact, the re-oxidation of the catalystcan proceed by the mechanism of oxidative dehydrogenation of surface OHgroups, 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�� þHO2d ð5Þ

HO2dþ½OH�� ! ½O�� þH2O2 ð6Þ

H2O2 ! 2OH ð7Þ

OHþ ½OH�� ! ½O�� þH2O ð8Þ

It is appropriate to stress that the overall reaction for regeneration is thesame as in the mechanism proposed by Ito and Lunsford.59 In fact, it in-volves the participation of two surface [OH� ] groups and the formation ofwater. However, it does not require removal of lattice oxygen. It is knownthat 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 nottake 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 catalystspretreated 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 ofLi/MgO catalysts proceeds as some sort of oxidative dehydrogenation ofsurface hydroxyl groups (Fig. 15). Our observations suggest that only athigher temperatures (700 1C) i.e., at methane coupling conditions, thescheme proposed by Lunsford is operative. But, in the case of oxidativedehydrogenation of higher alkanes, which occur at lower temperatures(500 1C) the catalyst regeneration goes via traditional re-oxidation with ade-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

O O

HH

½ O2

H2O

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

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 oxy-gen of the catalyst, occur on the oxide surface in a typical ‘‘Mars-vanKrevelen’’ type redox mechanism. Oxygen from the gas phase regeneratesthe oxide, which is reduced during alkane oxidation.40 Accordingly, Kon-dratengo et al. have shown recently that in the case of vanadium oxidesystems, a high density of oxidizing sites can favour H-atom abstractionfrom 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 split-ting forming alkyl radicals. Because the oxygen in these type oxides is lessnucleophilic, the radicals are then released to the homogeneous gas phasewhere radical chain reactions determine the composition of olefinsformed.53,58 Thus alkane to olefin conversion on such catalysts involves aroute of heterogeneously-initiated homogeneous reactions. Results alsoindicate that total oxidation (combustion) reactions occur both in the gasphase and on the surface of the oxide.68

The contribution of such homogeneous gas phase routes in heterogeneouscatalysis has been especially discussed in the last 20 years in the case ofoxidative coupling of methane to higher hydrocarbons43 and oxidative de-hydrogenation of light alkanes (ODH) to olefins.23 By varying the postcatalytic volume of the reactor and observing that this caused an increase inconversion,66 it was concluded that gas phase radical reactions make amajor contribution to such processes. In addition, direct evidence for thepresence of surface-generated gas-phase radicals has been provided byspectroscopic methods, in particular by techniques such as matrix isolationelectron spin resonance (MI-ESR) and infrared spectroscopy (MI-IR), intandem with a catalytic reactor.46

In the case of oxidative dehydrogenation of alkanes, using Li/MgOcatalysts, an increase in olefin selectivity could be achieved by increasing thenumber of active sites per volume of catalytic bed.68 We suggested recently,that the propyl radicals formed

½LiþO�� þ C3H7d! ½LiþOH�� þ C3H6 ð9Þ

could undergo a second hydrogen abstraction at the active sites leading topropene,68 indicating that olefin formation on the catalyst surface ispossible.

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 [LiþO� ]. Lowering the temperature of the reactioncan facilitate further radical conversions on the catalyst surface as inequation (1). Therefore the olefin selectivity can be manipulated by de-veloping more active catalysts.

The use of cold plasma is one way to achieve C–H bond activation atambient temperatures. Stable and cold gaseous plasma can be generated atroom temperature inside a micro reactor by dielectric barrier discharge.70,71

The advantage of using a microreactor is that it allows generation of non-thermal 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 bondsas a result of inelastic collisions.71 Additionally, performing the reaction in amicro scale system (channel dimensions 10–1000 mm) with intrinsically highsurface to volume ratio provides extreme quenching conditions on thecatalyst surface.72,73

Thus a micro plasma reactor containing catalyst can be used to convertlight alkanes into more valuable molecules at room temperature and at-mospheric pressure.74 Incorporation of a stable catalyst layer on the reactorwall is crucial in such a situation. In the case of catalytic reactions manyattempts have been made to deposit catalysts on the walls of micro reactors,instead of introducing powder-type materials in the micro channels inpacked-bed fashion, which leads to a high pressure drop.75

Figs. 12,13 show, respectively, the top and schematic cross sectional viewof the microplasma reactor used by Trionfetti et al.68 It consists of a Pyrexrectangular chip of 50mm length� 15mm width. Microchannels of 30mmlength� 5mm width and a channel depth of 500 mm were realized in thechip by means of chemical etching using aqueous HF. Sandwich thermalbonding of three Pyrex plates (one with, two without a channel) allowedfabrication of the micro reactor. Details of the processing scheme are givenelsewhere.76 Gas inlet and outlet holes were created by powder blastingusing Al2O3 particles. The two copper ribbon electrodes connect to an ex-ternal power supply in order to generate plasma by dielectric barrier dis-charge (DBD) at atmospheric pressure.90,91 A power output between 2 and25W is easy to stabilize. The power absorbed by the plasma can be calcu-lated from the corresponding V–Q Lissajous Figures.77 The plasma gener-ated in the DBD configuration consists of high energy electrons and ischaracterized by a large number of micro-discharge filaments (ionization ofthe medium by the electrons), each lasting few nanoseconds.78 These highenergy electrons (in the range 3–4 eV) are able to activate hydrocarbons andoxygen at room temperature and atmospheric pressure.79 The short life timeof 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 letare indicated (A, B). The copper plate is also well visible (C). This is connected to a powersupply using adhesive copper foils (D).

Catalysis, 2010, 22, 119–143 | 133

small volume and the large surface to volume ratio of the micro reactorallow fast removal of the heat produced during oxidation of alkanes.

The channels of the micro reactor were coated with Li/MgO catalyst layerby wetting the walls with sol via micro-pipetting, allowing the sol to gel andsubsequently calcining at 500 1C to get the catalyst coating.68,69,80–83 Detailsof the characteristics of the reactor and catalyst system are extensivelydiscussed elsewhere.68,80–83

Results obtained by us for propane oxidation highlight some very excitingchemistry.68 Fig. 14 shows the optical emission spectrum of C3H8–Heplasma with 3W power input. Electronic excitation of ‘CH’ correspondingto the A2D-X2P transition at 431.5 nm was used to determine the kineticgas temperature in this emission region.77 The resolution of the spec-trometer, calibrated using a UV lamp, was determined to be 0.7 nm (fullwidth at half maximum). This is not enough to resolve the individual ro-tational lines of the Q, R and P branches of the CH band. However, therotational temperature, which reflects the gas temperature inside the fila-mentary discharge, was calculated by comparing the CH band in Fig. 14with those in simulated spectra at various temperatures using the LIFBASEsoftware.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 thehighest power of 24W supplied from the source. In the case of the 3W usedin our experiments an average gas temperature of 25� 50 1C was obtained.Thus, the spectra in Fig. 14 indicate that propane activation occurs at closeto room temperature.

The existence of CH and H bands in the spectra shown in Fig. 14 indi-cates decomposition of propane via both C–C and C–H bond scission. Inagreement with this supposition, the product stream showed componentstypical of the oxidative cracking of propane as discussed earlier, i.e., COx,ethane, propene and acetylene. Most remarkably, in the case of the Plasma

Surface

Discharges

Dielectric-Pyrex

Catalyst layeradditional dielectric

∼

CopperElectrode

Fig. 13 Schematic drawing of the cross-sectional view of the employed microplasma reactor.The 3 Pyrex plates forming the microreactor are schematically represented.

134 | Catalysis, 2010, 22, 119–143

He

He

He

CH A2Δ→X2Π

C2(d3Π-a3Π) Hα

400 500 600 700

Wavelength (nm)

Inte

nsity

(a.

u.)

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

C3H6 COx C2H6 C2H2 CH4 C4 >C4

Sel

ectiv

ity (

mol

%)

C2H4

Products

0

10

20

30

Fig. 15 Selectivity to the main products for a plasma microreactor. Conditions: flow rate15ml/min, feed composition 10% propane, 1% oxygen and balance helium; activation at roomtemperature. Formation of hydrogen was detected but not quantified.

Catalysis, 2010, 22, 119–143 | 135

microreactor a substantial amount (37mol%) of products with highermolecular weights than propane, i.e., C4, C

þ4 , were also observed (Fig. 15).

The presence of products containing four or more C atoms requires C–Cbond formation under the conditions present in the microreactor.

For Li/MgO catalysts, propane activation requires the presence of oxygenand higher temperatures (TW600 1C)).35,53,85 In these catalysts [LiþO� ]type defect centers are the active sites and their formation requires thehigher temperatures. The [LiþO� ] site is claimed to have a high H-atomaffinity and, at relative high temperature, is able to abstract Hd from pro-pane molecules forming n- and iso-propyl radicals as primary intermedi-ates.85–87 These propyl radicals are formed depending on whether primaryor secondary hydrogen is abstracted from propane. At the higher tem-peratures required for the reaction, endothermic decompositions arefavored and the two types of radicals (n & iso) undergo different unim-olecular reaction routes in the homogeneous phase: iso-propyl radicalsundergo C–H bond scission at the a-position, yielding propene and ahydrogen atom, while n-propyl radicals preferentially undergo b-scission ofthe C–C bond, forming a methyl radical and ethylene.88 The variousproducts formed can be accounted for by a series of radical chain reactionsin the gas phase, details of which were discussed earlier.89

In a plasma microreactor, C3H8 molecules are directly activated/con-verted via collisions with energized electrons. Activation produces radicalssuch as C3H7

d due to cleavage of C–H bonds (2). These can initiate radicalchain reactions. Thus, reaction (10) is strongly influenced by the number ofcharges transferred or accumulated on the dielectric surface,92 and thereforeby the relative permittivity.

C3H8 þ e� ! C3H7dþHdþ e� ð10Þ

The reactivity of micro-plasma towards propane is improved when a layerof Li/MgO was present. This is due to the larger permittivity of oxide layer(9.7 C2/N*m

2 for MgO) compared to Pyrex (4.8C2/N*m2).93–95 This allows

more impacts (and more inelastic collisions) giving rise to excitation of ahigher 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 thepresence of oxygen and plasma the propane conversions are higher. Amongthe atomic processes taking place in a non-thermal plasma, the electronimpact dissociation of O2 to form charged and

e� þO2 ! 2Oþ e� ! O� þO ð11Þ

e� þO2 ! O�2 þ e� ! OþOþ e� ð12Þ

neutral oxygen has been reported in literature and is described in reactionequations 3 and 4.96 The O� species, present in the homogeneous phase, hasbeen reported to cause C–H bond scission in alkanes e.g., methane,97 eth-ane.98 In the case of propane this will result in the formation of propyl andhydroxyl radicals as shown below:

½O�� þ C3H8 ! ½OH�� þ C3H7d ð13Þ

136 | Catalysis, 2010, 22, 119–143

Further, C3H7 radicals react fast with O2 forming hydroperoxyl (HO2d)

radicals, which can react with propane molecules to form H2O2. De-composition of H2O2 results in hydroxyl radicals (OHd), which become themain chain propagators.47

[LiþO� ] sites are defect sites and their existence at lower temperatureshas been investigated and confirmed, using EPR spectroscopy, by Lunsfordand others.48,59,100 Improved propylene yields suggest that defect [LiþO� ]sites may already be present at low temperature in the presence of plasmaand oxygen, and that these sites abstract hydrogen from propyl radicalsforming propylene.99 Alternatively, the presence of plasma can also help tocreate other defect sites which are able to selectively terminate radicals.Detailed studies are present in the literature describing the emission of UVlight during dielectric barrier discharge (DBD).101 Nelson et al.102 and laterE. Knozinger et al.103 reported, using EPR studies, that interaction betweenUV light and MgO particles can give rise to surface paramagnetic centers(trapped electrons, typically F-centers, [Vd

O]). Goodman et al. suggest thatthese oxygen vacancies, containing one or two electrons (F-type defects), areable to activate C–H bonds during methane oxidative coupling.104 Bondscission occurs and the Hd is trapped at the defect site. Thus, the presence of[Vd

O]-type defect sites caused by the plasma may allow Hd abstraction frompropane and even from propyl radicals, and in the latter case enhancedselectivity to propene can be expected.

Optical spectra recorded in presence of plasma (Fig. 18) show that CH-type radical species are formed. Dimerization of such species can result inethyne, which is often detected in experiments with plasma and hydro-carbons.108 Hydrogen redistribution during this reaction, associated withthe formation of dehydrogenated products such as ethyne, may account forthe observation of methane.

This argument regarding dimerisation and formation of ethyne105–107 canalso be logically connected to the large amounts of C4 and Cþ4 products thatwe observe. The role of plasma is in the activation of propane and the for-mation of radicals at ambient temperatures. The formation of C4 and Cþ4products from propane essentially requires C–C bond formation, whichis an exothermic process and therefore favored at lower temperatures.Therefore it is not surprising to see C–C bond formation reactions under ourconditions. In conventional fixed bed reactors, propane activation occurs athigher temperatures (TW600 1C) in the presence of catalyst. These con-ditions favor the rupture of C–C bonds, and thus we see only products ofcracking or combustion, i.e., with molecular weights lower than that ofpropane.

Considering that similar amounts of C4þCþ4 products are formed in themicroreactor both with and without catalyst, it is suggested that thecoupling 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 thesemolecules 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 (C4þC3) were observed.

Catalysis, 2010, 22, 119–143 | 137

C2 products were the most abundant species when methane was thefeedstock (Table 3). Appreciable amounts of C3 (11%) and C4 (2%)products, demanding multiple C–C couplings, were also found. Thusactivation by cleavage of the C–C and C–H bonds and C–C coupling atambient temperatures are consistently observed for all three hydrocarbonsstudied, viz, propane, ethane and methane.109

Alkane activation occurs in the presence of plasma via collisions withenergized electrons, e.g., C3H8þ e�-C3H7

dþHdþ e� .110 In the presenceof helium this mechanism might become more efficient because the meanfree path of the electrons increases111 due to the small cross section of highlyexcited helium species created during electric discharge. These excited spe-cies of helium, can also possess higher energy, up to nearly 20 eV, enough toactivate a hydrocarbon by collisions.112

Alkane activation can also occur via an indirect route, i.e., activation ofgas phase oxygen by plasma. Among the atomic processes taking place in anon-thermal plasma, the electron impact dissociation of O2 to form chargedand neutral oxygen has been reported in the literature and is described inequations 11 and 12.113,114 The O� species, present in the homogeneousphase, 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 theformation of propyl and hydroxyl radicals as in equation 13. The reactionbetween C3H7

d radicals with O2 results in hydroperoxyl (HO2d) radicals

which react further with a propane molecule to form H2O2. Decompositionof the latter gives two hydroxyl radicals (OHd), which become the mainchain propagators.23 All these reactions can initiate radical chain reactions,but the main difference is that these propagation and termination reactionsnow 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 thatmethane activation required high temperatures and at the high temperaturesC–C bond formation would be expected to be less favorable. Typically,formation of C–C bonds is an exothermic process and therefore favored atlower temperatures. Therefore, it is not surprising to see C–C bond

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 15ml/min, RT; 15% conversion level of ethane (mol%)

COx C3H8þC3H6 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 15ml/min, RT; 10% conversion level

of methane (mol%)

COx C3H8þC3H6 C2H4þC2H6 C2H2 C4

30 11 49 8 2

138 | Catalysis, 2010, 22, 119–143

formation reactions under the mild conditions present during plasmaexperiments. C–C bond coupling is also observed during olefin metathesistype reactions, also at room temperature, but requiring a transition metalcatalyst. Current scenarios for upgrading light hydrocarbons (increasingmolecular weight) for, e.g., alkylation,118 metathesis119,120 or oligomerisa-tion121,122 reactions (e.g., SHOP process123,124) involve at least one olefin.However, in our studies direct homologation of alkanes is observed. Thisrepresents an interesting development for upgrading cheap low molecularweight alkanes to commercially useful fuels and/or feedstock materials forthe chemical industry.

Acknowledgements

The author wishes to state his pleasure and thanks to students and col-leagues who helped to carry out the work. In the hope of not missinganyone, 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 Pro-

ceedings, 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 andM. 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 activatedketones

Jozsef L. Margitfalvia and Emılia Talasa

DOI: 10.1039/9781847559630-00144

In this contribution key features of the asymmetric hydrogenation ofactivated ketones over cinchona–-Pt catalyst system are reviewed. Bothhistorical backgrounds and recent results are evaluated and discussed.The focus is laid on the peculiarities of these reactions, such as (i) methodsand 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-Ptinteractions.

1. Introduction

1.1 General information

In order to reduce the environmental and health hazards the demand foroptically active compounds in high enantiopurity is increasing in the field ofpharmaceutical, agrochemical and cosmetic products. Although the mostcommon applications are bio-related, there is also a great interest in the areaof materials science for chiral compounds, such as chiral polymers or chiralliquid crystals. For this reason the interest to produce chiral compounds inhighly pure form is expanded over the past decades. There are differentapproaches to obtain compounds in high enantiopurity. One of the mostenvironmentally friendly methods is the use of chiral catalysts.1–3 There aredifferent approaches in chiral catalysis, such as homogeneous,4 hetero-geneous,5 enzymatic,6 and ‘‘artificial catalytic antibodies’’.7 The commonfeature of all approaches is that relatively small amount of chiral auxilaritiesis required to produce chiral compounds in high enantiopurity. In this re-spect the term ‘‘multiplication of chirality’’8 is often quoted. When het-erogeneous catalysts are applied, the chiral auxilarities used are often calledas chiral modifiers.

In the past decades, significant progress has been achieved in homo-geneous enantioselective catalysis, what is reflected by the Nobel Prizeawarded in 2001 to Sharpless, Noyori, and Knowles. Variety of transitionmetal complexes containing unique chiral ligands have been developed toinduce enantioselectivity by homogeneous catalysts. These chiral metalcomplexes were used in various catalytic reactions, such as hydrogenation,9

dihydroxilation,10 epoxidation,11 Diels-Alder reaction,12 C–C bond for-mation12,13 Michaels reaction,14 etc. These reactions are considered havegreat importance for the production of fine chemicals and pharmaceuticalproducts.15

aInstitute of Surface Chemistry and Catalysis, Chemical Research Centre, Hungarian Academyof 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 rela-tively small amount of catalyst, however the chiral ligands and the metalcomplexes used are relatively expensive. Homogeneous catalysts are sensi-tive both to oxygen and moisture, for this reason the handling of thesecompounds is quite troublesome. Additional serious problem is the catalystrecovery. As the remaining trace impurities of metals have a definite en-vironmental and health hazard the development of complex methods for theremoval of traces of transition metals increases significantly the productioncosts in homogeneous catalytic reactions. The immobilization of metalcomplexes into inorganic or organic supports is one of the approaches toovercome the above problems as immobilized catalysts can be separated byfiltration or can be used in continuous-flow reactors.

Homogeneous catalysts can be immobilized both to inorganic16 andpolymeric supports.17 Most of the cases the solid contains a ligand, what isconsidered as an anchoring site.

In case of inorganic supports the reactivity of surface OH groups is usedto immobilize a given type of ligand upon using the reactive trimethoxy (ortrichloro) silane derivatives.18 However, in this type of immobilizedhomogeneous catalysts leaching of the metal-complex has often beenobserved.

Among the new strategies to heterogenize transition metal complexes theencapsulation of a chiral metal complex in micropores19 and the use oftethered type metal complexes have to be mentioned.20 The tethered metalcomplex catalysts were recently developed by Augustine and coworkers.21–23

These catalysts showed high activity and high chemo- and/or enantioselec-tivity in various hydrogenation reactions.

When metal-complexes (both homogeneous and immobilized one) areused the enantio-differentiation is controlled by the structure of the metalcomplex. In this respect the molecular character of the catalytic step has tobe emphasized, although the exact form or structure of the [metal complex –substrate] adduct is often unknown. In these enantioselective catalytic re-actions the chirality and the helicity of ligands has a great importance tocontrol the enantio-differentiation step.24 This control can be either ther-modynamic or kinetic in character. The distinction between these twomodes of control is often very difficult. Here we should like to refer to theclassical contribution made by Halpern and coworkers.25

The most important class of ligands used in asymmetric reactions has achiral organic backbone with tertiary phosphino, amino and alcoholicfunctional groups. Highly effective chiral ligands are often bidentate; i.e.they have two coordinating sites for the coordination of the substratemolecule. In chiral homogeneous catalysis, due to the high substrate spe-cificity high enantioselectivities can only be achieved for a defined class ofsubstrate molecules, i.e., the right combination of metal and ligands has tobe found for each individual catalytic reaction.26

Even though enantioselective homogeneous catalysis is still a relativelyyoung discipline, several enantioselective homogeneous catalytic processeshave already been used on an industrial scale.27,28

As concerns the performance (enantioselectivity), the mechanistic insightand general understanding heterogeneous enantioselective catalysis is far

Catalysis, 2010, 22, 144–278 | 145

behind to its homogeneous analogue. Only a relatively narrow range ofprochiral compounds with CQO and CQC bonds can be hydrogenatedwith high enantioselectivity. However, due to the above-mentioned draw-backs in homogeneous catalysis, such as separation, reuse, and stability, theinterest for heterogeneous asymmetric catalysis increases permanently.

The first publication related to asymmetric heterogeneous catalytichydrogenation was published by Schwab in the early thirties.29 In the firstattempts intrinsically chiral solids, such as quartz has been applied.30

In other approaches chiral biopolymers or natural fibers, such as silk fibroinwere used,31 but due to the low optical yield and severe reproducibilityproblems this approach has been almost completely discarded.

The discovery of the Ni/tartrate system for the enantioselective hydro-genation of beta-diketones or their analogs was the first real breakthroughin this field. This area has been recently reviewed.32–35 Optical yields over90% were obtained for various substrates using the Ni-NaBr/tartrate sys-tem.36 In general this catalytic system requires a pre-modification of theparent nickel catalyst with tartaric acid prior to the reaction in order toform chirally modified sites involved in enantio-differentiation. Recently,the use of in situ modification procedures for Ni/tartrate system has beendescribed.37

In this catalytic system reactive chemisorption of tartaric acid to thenickel surface resulting in some leaching of the nickel to the liquid phase hasbeen evidenced.38 Due to the interaction with tartaric acid the formation ofimprinted sites on the nickel surface has been suggested. The enantio-dif-ferentiation takes place over this type of chirally modified sites.

Since the discovery of the cinchona-platinum catalyst system by Orito’sgroup39–43 for the enantioselective hydrogenation of methyl pyruvate(MePy) or ethyl pyruvate (EtPy) the platinum-cinchona system has beensuccessfully applied in the enantioselective hydrogenation of a variety ofa-functionalized activated ketones, such as various a-ketoesters, keto-pantolactone (KPL), a-ketoamides, a-diketones, a-keto acetals, a-ketoe-thers, trifluoromethyl ketones, and pyrrolidine-2,3,5-triones. Recently themethod has been extended to other types of ketones.44 This type of asym-metric hydrogenation reaction is considered as the most intensively studiedheterogeneous enantioselective hydrogenation reaction.

Since the early eighties great amount of knowledge accumulated on thiscatalytic system. The topic has been reviewed by different authors.15,33,35,45–54

The most important characteristic features of the platinum-cinchona systemare 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 ifthe basic nitrogen at the quinuclidine moiety of the cinchona alkaloid isblocked by alkylation;57

� Upon using various substrates the addition of the cinchona alkaloidresults in significant rate acceleration;58

� The amount of modifier required to induce high enantioselectivity is inthe range of 1� 10� 5M, 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 was276.000);59

� The replacement of the quinoline ring of the modifier by phenyl orpyridyl 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 pro-

cess of understanding the peculiarities of the platinum-cinchona system.This progress covers the following main issues: (i) elucidation of both therate acceleration and the origin of enantio-differentiation, (ii) clarificationof the nature of species formed both in liquid phase and on Pt surface byusing various spectroscopic methods, (iii) establishment of general andspecific kinetic patterns, and (iv) theoretical calculations and relatedmodeling.

As far as several reviews have already been published on the enantiose-lective hydrogenation of activated ketones in the presence of cinchona-Ptcatalyst in this review an attempt was done focusing on (i) methods andapproaches used, (ii) recent achievements, and (iii) recalling historicalevents.

Contrary to earlier reviews in this contribution the mechanistic con-siderations will be treated without any preconception. It means that it willbe a priori not accepted that all phenomena involved in the rate accelerationand induction of enantio-differentiation can be related to events takingplace exclusively over Pt surface.61 Consequently, in this review we shallalso refer to the general aspects of chiral induction generated by cinchonaalkaloids. Possible interactions in the liquid phase will also be discussed.

Enantio-differentiation is a phenomenon characteristic mostly for organicreactions. There are various synthetic methods in organic chemistry to in-duce chirality. It has to be emphasized that cinchona alkaloids are well-known chiral compounds used by organic chemists to bring about chiralinduction.62 This issue will be briefly discussed in Section 2.1

We shall try to demonstrate that the enantioselective hydrogenation ofactivated ketones is a very complex reaction. Depending on the conditionsof catalyst pretreatment, the accomplishment of the hydrogenation reactionand the type of substrate and modifier used the key interactions responsiblefor the transformation of the chiral information can takes place both at thePt surface and in the liquid phase. Methods and approaches used in thisarea will also be discussed as these issues were not treated in previousreviews.

1.2 Orito’s followers

After Orito’s publications39,40–43 intensive research programs were startedby different research groups. First two groups in Switzerland, one at CibaGeigi in Basel under supervision by Dr. H.U. Blaser,63 the other in Zurich atETH with professor A. Baiker.64 Parallel to that professor P.B. Wells65 inHulls (Great Britain) started a program related to the use cinchona alkal-oids in heterogeneous catalytic asymmetric hydrogenation.

Later on other groups in the USA (Dr. D. Blackmond at Merck,66 Pro-fessor 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 HungarianAcademy of Sciences and Prof. M. Bartok70 at University of Szeged ), inFinnland ( professors T. Salmi and D. Murzin33,71at Abo Academy Uni-versity) joined to this research area. Today there are around 15–20 in-dependent research groups all around the word that are involved in theinvestigation of one of the aspects of Orito’s reaction.

It is interesting to emphasize that the method developed by Orito wasquickly modified as the ‘‘pre-modification’’ approach was replaced by insitu modification technique. In this respect the pioneering works were doneby the two Swiss groups. Only one group has continued for a while to applythe ‘‘pre-modification’’ approach: (the group in Hulls), however today thisapproach is almost forgotten. Further details about modification pro-cedures will be given in Section 4.

In the first approaches mostly Pt/Al2O3 catalysts and EtPy were used inorder to elucidate the general kinetic patterns.58,73 Later on the focus waslaid on (i) the elucidation of the reaction mechanisms,61,65,67,74,75 (ii) the useof new substrates,76,77 (iii) the application of new modifiers,60,78 (iv) newtype of catalysts,79,80 (v) the formation of by-products,81 (vi) the role ofimpurities,82 (vii) modeling substrate modifier interaction.83–85 Today so-phisticated experimental and computational techniques, such as reactioncalorimetry,86 the AFM,87 NMR,88 FTIR,89 in situ reflection-adsorptioninfrared spectroscopy (RAIRS),90 attenuated total reflection infraredspectroscopy (ATRIR),91 surface-enhanced Raman spectroscopy (SERS),92

circular dichroism,93 electrochemical methods,94 Near-edge X-ray ab-sorption fine structures (NEXAFS)95,96 studies, DFT calculations97,98 areused to get as much as possible information about these unique asymmetrichydrogenation reactions.

Orito’s approach later on was extended to other type of activated ke-tones. The substrates were classified according to the observed rates andenantioselectivities. This classification is given in Fig. 1. Further discussionof substrate specificity will be given in Section 5.3.

From the above discussion it can be concluded that the enantioselectivehydrogenation of activated ketones is the most detailed studied asymmetrichydrogenation reaction. However, despite of the extensive studies there areplenty of unanswered questions related to the origin of enantio-differentiation.

High rate – high ee

(a) (b) (c)

Medium rate – medium ee Low rate – low ee

8 9 10

1 2 3 4

5 6 7

11 12 13

14 15

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 additionalknowledge in the field of materials science, surface chemistry, surface sci-ence, surface analysis, computational chemistry, chemical engineering, etc.Heterogeneous enantioselective hydrogenation requires an additionalknowledge, i.e. an education in the field of organic reactions. As far asenantioselective reactions are a specific area of organic catalysis theknowledge in this area should also be very specific.

It has to be emphasized that those who joined to this research area havedifferent scientific background and different research interest. In most of thecases we are witnessing the prevalence of approaches and views reflectingthe mode of thinking of a chemical engineer or a surface scientist.

Although excellent reviews were written in the last ten years,46–52 thosewho have a solid background in organic chemistry can realize that both thestructure and the conclusions of these reviews reflect the view of those whohave a ‘‘surface science’’ or ‘‘chemical engineering’’ background.

The investigation of heterogeneous catalytic reactions requires a complexknowledge and the use of various experimental techniques. Among thesemethods reaction kinetics and surface characterization has the greatestimportance. Detailed reaction kinetic studies can provide sufficient infor-mation to suggest or create a proposed reaction mechanism. However, evenif reaction kinetics can completely be described by the given system ofdifferential equations the elucidation of the exact reaction mechanismcannot be guaranteed. This would require the comparison of several pos-sible or potential kinetic equations derived from other possible reactionmechanisms. Unfortunately, such kind of comparison is only seldom isperformed in studies related to the elucidation of reaction mechanisms ofasymmetric hydrogenation of activated ketones.

In the enantioselective hydrogenation of activated ketones the accom-plishment of detailed reaction kinetic measurements is strongly hindered bythe peculiarities of the reaction, such as (i) the ability of modifiers to induceenantio-differentiation at very low modifier/substrate ratio, (ii) the highreactivity of the substrates resulting in the formation of various by-prod-ucts, (iii) the loss of alkaloid during the reaction, and (iv) the catalyst poi-soning. All these issues will be discussed latter in separate sub-paragraphs.

Due to the formation of byproducts, the loss of alkaloids and catalystpoisoning the analysis of the full kinetic curve is almost impossible. Con-sequently, the use of initial (or maximum) reaction rates cannot provide aproper background for exact kinetic analysis or for the establishment ofcorrect reaction mechanism. The use of in situ calorimetry is one of the mostprecise methods to obtain direct reaction rates,99 although this method isnot common. The preliminary analysis of results obtained by in situ cal-orimetry indicates that the rate follows the Michaelis-Menten mech-anism,100 what is characteristic to enzyme catalytic reactions.101

The fact, that only trace amount of cinchona alkaloid can result in eevalues above 90%, strongly indicate that this catalytic system is extremelyunique. 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/modifierratio is 276 000:1.59 In other studies related to the enantioselective hydro-genation of EtPy the above ratio is in the rage of 50 000 to 166 000.102 Sucha high value of ‘‘chiral amplification’’ is characteristic only to enzymes.Cinchona alkaloids, due to their high extent of ‘‘chiral amplification’’, theirdistinct substrate specificity and their flexible structure, can be considered asa ‘‘mini-enzyme’’.93 This aspect is often forgotten by those working in thisarea; although in a recent study the similarity to enzymes has already beenmentioned.103

The other key approach for the elucidation of reaction mechanism is theinvestigation of the formation of different surface intermediates by differentsurface analytical tools. These approaches will be described in details inSection 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 ordoes it represent a dead-end in the given reaction scheme?’’ In the formercase we talk about ‘‘actors’’, while in the letter case about ‘‘spectators’’.Unfortunately, this kind of questions is only seldom raised. It has to beemphasized that the differentiation between ‘‘actors’’ and ‘‘spectators’’ is along dispute in the area of heterogeneous catalysis. Even in gas phase re-actions taking place at atmospheric pressure neither the ‘‘in situ’’ nor the‘‘operando’’ spectroscopy can always provide an unambiguous exact answerabout the involvement of a given surface species in the reaction network.When we are dealing with heterogeneous hydrogenation reactions takingplace in a three-phase system at high hydrogen pressure the accomplishmentof in situ or operando spectroscopy is very difficult.

Catalysis scientists like very much to refer to sophisticated spectroscopicdata and use these data in favor of their ideas with respect to the reactionmechanism. Even one of the authors of this contribution walked into thiscatch. In early eighties the direct route for the hydrogenation of acetylene toethane was confirmed by a sophisticated ‘‘double isotope labeling’’ techni-que.104 Based on surface science results we proposed that the formedethylidyne species (RC–CH3) are responsible for the direct route of ethaneformation. Of course, we did not detect these species; we just referred to oneof recently published LEED results.105 Later on it turned out that theethylidyne species are so stable that they cannot be removed from the Pdsurface by hydrogen at room temperature.106 Of course, several similarmisinterpretations can be found in the literature.

It will be discussed latter on that in the hydrogenation of activated pro-chiral ketones the presence of hydrogen at the Pt surface is very crucial.It strongly determines the performance of the catalyst. The hydrogenationof ketones requires relatively large abundance of hydrogen at the Pt surface.Probably, it is the reason that these enantioselective hydrogenation re-actions cannot be performed under conditions of transfer hydrogenation.Any disturbance in the amount of available hydrogen can result in signifi-cant alteration is the performance of Pt-cinchona catalytic system. It isespecially notable when the hydrogenation is performed in a continuous-flow reactor under trickle bed condition.107

However, too much hydrogen at the Pt sites will result in the hydro-genation 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 thathydrogen ‘‘starving’’ conditions increases the chance for undesired side re-actions, such as oligomerization and polymerization. Consequently, if in situmeasurements cannot be performed under optimum hydrogen coverage thechance to detect ‘‘spectators’’ is very high. For this reason all spectroscopicdata presented so far should be treated with definite precaution.

2. Cinchona alkaloids

2.1 Characteristic features of cinchona alkaloids

Cinchona alkaloids are used in many fields of our everyday life. They arewidely used in the pharmaceutical and chemical industry. Quinine, derivedfrom the bark of Cinchona ledgeriana Moens ex Trimen is the oldest knownnatural antimalarial drug. Cinchona alkaloids as easy available chiralagents have great importance both in the academic research and in largescale industrial use. In this respect the classical separation of racemicnaproxene can be mentioned.110 An explanation why cinchona alkaloids areuniversal molecules for so many purposes was given by Wynberg.62 This isshown in Fig. 2.

Various parts of the molecule fill the following functions: (a) hydrogenbond formation (interaction with metals); (b) basic amine; (c) bulk aliphatichydrocarbon moiety; (d) ‘‘handle’’ to modify; (e,f) chiral pocket (epimersavailable; conformer formation); (g) bulk aromatic hydrocarbon, polariz-able, p-p interaction; (h) ‘‘handle’’ to modify; steric and polar influence. Inthis section the use of cinchona alkaloids in chiral separation and chiralcatalysis will be reviewed very briefly. More detailed reviews can be foundelsewhere.62,111,112

2.2 Use of cinchona alkaloids as chiral auxiliaries

2.2.1 Chiral separation. The first example of resolution through for-mation of diastereomeric salts was made by Pasteur113 who used quinicineand 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 withpermission, (Figure 19))

Catalysis, 2010, 22, 144–278 | 151

separation of various racemic mixtures.114 In the sixties quinine and someother cinchona alkaloids were used to prepare chiral sorbents.115 Alkaloidsof this type were covalently bonded to a silica support via their olefinicgroup. In this way several functional groups in a bulky chiral system pro-vided multiple contact points with the racemate to be resolved.

Up to now big variety of separation techniques using cinchona alkaloidshas been reported. A two-dimensional liquid chromatography–mass spec-trometry (LC–MS) system was developed for the separation of both dia-stereomers and enantiomers of peptides.116 Generally the presence ofelectron-deficient aromatic N-acyl constituents and bulky, highly lipophilicside chains enhances enantioselective adsorption, reflecting the importanceof intermolecular p-donor-acceptor and hydrophobic interaction with thechiral selector.117 Mixed ternary ion associate formation between xanthenedye, cinchona-alkaloid and quaternary ammonium ion has been applied todeterminate the trace amount of quaternary ammonium salts in pharma-ceuticals by UV spectrophotometry.118

2.2.2 Chiral catalyst. In the field of chiral catalysis huge amount ofwork has been done and cinchona alkaloids have been involved the in thefollowing areas:� chiral basic and nucleophilic catalysts in organo-catalytic reactions,� chiral ligands coordinating metals like osmium in homogeneous cata-

lytic 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 haslead to optically active alcohols having enantiomeric excess up to 92%.119

Cinchona alkaloids have been used both in solute form in liquid phaseand in bounded form immobilized into polymer or oxide type supports. Inorgano-catalysis based on cinchonas large variety of substrates and types ofthe reactions has been reported. In 1954 Prelog and Wilhelm described thebehaviour of different cinchona alkaloids and some of their derivatives inthe asymmetric cyanhydrin synthesis.120 A review of cinchona alkaloid-catalyzed reactions covering the period prior 1968 was given by Pracejus.121

Cyanohydrin reaction, the Michael reaction, the 1,4-thiol and thiolace-tate additions, selenophenol addition reactions, epoxidation of electron-poor olefins, formation of the phosphorus-carbon bond using chiral aminecatalysis, 1,2-additions in the presence of cinchonas has been detailed byWynberg in 1986.62

Highly enantioselective Reformatsky reaction of ketones has been ac-complished using cinchona alkaloids as chiral templates.122 Cinchona al-kaloid-derived chiral bifunctional thiourea organocatalysts weresynthesized and applied in the Michael addition between nitromethane andchalcones with high ee and chemical yield.123

The osmiumtetraoxide catalyzed asymmetric dihydroxilation (AD) is avery important field of cinchona utilization (see Fig. 3).10,124 Cinchona al-kaloid backbone is ideally suited for providing high ligand acceleration aswell as enantioselectivity in AD. It has been found that the

152 | Catalysis, 2010, 22, 144–278

enantioselectivity is influenced mainly by the nature of O9 substituent of thecinchona alkaloid backbone. Three different classes of ligands are veryeffective for the dihydroxylation of almost any olefin (PHAL-,125 PYR-126

and IND-class127). Large scale of substrates (monosubstituted, 1,1-di-substituted, 1,2-disubstituted, trisubstituted even tetrasubstituted olefins,enol ethers, cyclic olefins, amides, enones, sulfur-containing olefines, pro-tected divinyl ketones, conjugated dienes, trienes etc) can be successfullydihydroxylated.10

Phase transfer catalysis (PTC) based on cinchona alkaloids128–153 is acontinuously developing practical method for asymmetric synthesis becausethese catalysts are very selective. Enantioselective epoxidation of a,b-un-saturated ketones utilizing cinchona alkaloid-derived quaternary ammo-nium phase-transfer catalysts bearing an N-anthracenylmethyl functiongave also appropriate results.

Common to all successful applications of cinchona alkaloid derived phasetransfer catalysts is that the reaction conditions have to be optimized,consequently structures of substrate, reagent, and catalyst must ‘‘fit toge-ther’’: An attempt has been made to understand the role of differentstructure units of cinchona derivatives in PTC. The N-anthracenylmethylgroup introduced by Lygo139,143 and Corey141 has been good for an increasein selectivity, although 1-naphthylmethyl was almost as effective.144 Phasetransfer catalysts having diaryl substitution at the 3-and 4-positions of theN-benzyl group in cinchonidinium salts were prepared to check how sub-stituted aryl groups affect the asymmetric induction in the benzylation re-action as compared to those having flat linear aryl systems likenaphthylmethyl and anthracenylmethyl groups.149

Tremendous amount of work has aimed the preparation and investigationof supported cinchona alkaloids as catalyst.154–170 Polymer bound cinchonaalkaloids have been employed for a number of heterogeneous catalytic re-actions e.g. asymmetric Michael additions,154–156,171 asymmetric synthesis ofa-amino acids,169 enantioselective a-chlorination of acid chlorides,170 asym-metric aminohydroxilation167 asymmetric dihydroxilation of alkenes.160,163

To recycle the alkaloid-OsO4 complex in asymmetric hydroxilation reaction

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

Catalysis, 2010, 22, 144–278 | 153

Kim and Sharpless159 have synthesized four different polymers, Piniet al. have examined copolymers of acrylonitrile and substituted quinidineand quinine and reported very low optical purity of diols (up to 45% ee).160

Lohray et al. prepared several copolymers of styrene and 4-phenylstyrenewith 10% DHQD-4-vinylbenzoate affording the most effective catalyst forheterogeneous AD.161 A nice example of immobilized quinine used as acatalyst for enantioselective a-chlorination of acid chlorides was given inref. 170. Large scale of different type of polymers has been reported forimmobilization of cinchonas.155,157,158,163,165,166 Silica gel supportedcinchona alkaloids have been used also as catalysts in asymmetric dihydroxy-lation and aminohydroxylation.167 Norcinchol supported on silica viaethoxy-silane compound has been applied for enantioselective hydrogenationof a-keto esters with moderate success.47 For different purposes differentcinchona alkaloids are suitable. In methanolysis of different tricyclicanhydrides quinidine and quinine has given higher ee than cinchonidine orcinchonine.172

For asymmetric dihydroxylation also quinine (QN) and quinidine(QD) have been found as the most effective ligand.10 In a few cases etheraltype cinchona alkaloids173 are also successfully used as chiral template,e.g. b-isocupreidine has been used in the asymmetric Baylis-Hillmansreactions of aromatic imines with 1,1,1,3,3,3-hexafluoroisopropyl acrylategiving (S)-enriched, N-protected-a-methylene-b-amino acid esters. Incontrast to the corresponding aldehydes, imines have shown the oppositeenantioselectivity.173

Based on the above short review it can be concluded that cinchona al-kaloids and their derivatives due to their multifunctional structure and easyavailability have been widely used for chiral induction in asymmetric syn-theses as well as chiral separations for a century. When new problems inasymmetric techniques have to be solved the application of cinchonas oftenprovides the proper solution again.

All of these results clearly indicate that cinchona alkaloids have been usedby organic chemists in various areas in order to induce chiral induction orchiral recognition. Consequently, the use of these alkaloids in Orito’s re-actions is only one of the opportunities provided by the unique chemicalproperties of these natural compounds. Any attempt to describe the action ofcinchona 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 cinchonaalkaloids based on computational or spectroscopic methods have beenmade with the aim of understanding of chiral discrimination process.88,174

The most frequently investigated cinchona alkaloids and cinchona deriva-tives are summarized in Fig. 4. Cinchona alkaloids consist of two rigid cyclicsystems, a heteroaromatic quinoline ring and a saturated cyclic quinuclidinering connected by two carbon-carbon single bonds. They have four asym-metric centers: C3, C4, C8, and C9. However, their configurations differfrom each other only at C9 and C8.

154 | Catalysis, 2010, 22, 144–278

N

H

N

R2

R4

R3

R1

4′

4′9

8

3N R

4

N

R2

R3

H

R1

8 94′

3

R1

R2

R3

R4

R2

R3

R4

quin

ine

C2H

3O

CH

3O

HH

cinc

honi

dine

C2H

3H

OH

Hdi

hydr

oqui

nine

C2H

5O

CH

3O

HH

dihy

droc

inch

onid

ine

C2H

5H

OH

Hep

iqui

nine

C2H

3O

CH

3H

OH

benz

oylq

uini

neC

2H5

OC

H3

OB

zH

(p-C

l)-be

nzoy

l-di

hydr

oqui

nidi

neC

2H5

OC

H3

pClB

zOH

chlo

roqu

inin

eC

2H3

OC

H3

Cl

Hde

oxyc

inch

onid

ine

C2H

3H

HH

met

hoxy

dihi

dro-

cinc

honi

dine

C2H

3H

OC

H3

H

R1

quin

idin

eC

2H3

OC

H3

OH

Hci

ncho

nine

C2H

3H

OH

Hdi

hydr

oqui

nidi

neC

2H5

OC

H3

OH

Hdi

hidr

ocin

chon

ine

C2H

5H

OH

H

epiq

uini

dine

C2H

3O

CH

3H

OH

epid

ihyd

roqu

inid

ine

C2H

5O

CH

3H

OH

(p-C

l)-be

nzoy

l-di

hydr

oqui

nidi

neC

2H5

OC

H3

PClB

zOH

acet

ildih

ydro

quin

idin

eC

2H5

OC

H3

OA

cH

(dim

ethy

lcar

bam

oyl)-

dihi

droq

uini

dine

C2H

5O

CH

3O

CO

NM

eH

met

hoxy

dihy

dro-

quin

idin

eC

2H5

OC

H3

OC

H3

H

Fig.4

Structure

andconfigurationofthecinchonaalkaloidsmost

frequentlyinvestigated(R

eproducedfrom

ref.176withpermission)

Catalysis, 2010, 22, 144–278 | 155

2.3.2 Conformational analysis. Crystallographic structure of QN,177

QD,178 CD179 and cinchonine (CN) 180 is well described. In solution how-ever, existence of several other conformers can be supposed. Conformationof quinine and quinidine was a key issue in different studies62,120,121,179,180

and the C8-C9 and C40-C9 bonds were considered most important in de-termining the overall conformation of these compounds. Hiemstra andWynberg181 have proposed that the most stable conformation of quininehave the largest substituent-the quinuclidine ring-on one side of the quinolinering, 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 conformationalproperties of QN and QD.175,176,183 By use of the molecular modelingprogram CHEMX, the conformational freedom with respect to the C8-C9and C9-C40 bond was investigated.175 Molecular mechanics studies showedthat cinchona alkaloids can in principle adopt four different conformations:two ‘‘open’’ one in which the quinuclidine nitrogen points away from thequinoline ring and two ‘‘closed ’’ one in which the quinuclidine nitrogenpoints toward the quinoline ring176(see Fig. 5). One of the calculated con-formations of QD (open conformation (3) in Fig. 5) has almost the samegeometry as the crystal structure.175 Different dihydroquinidine (DHQD)

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

156 | Catalysis, 2010, 22, 144–278

derivatives as model substances behave in different way. Acetyldihy-droquinidine exhibits the closed conformation in CDCl3.

183

Hydroxy cinchona alkaloids exist in open conformation (3) (see Fig. 5) atleast in 90%, wherein some conformational freedom of the quinuclidinering exists. Methoxy derivatives predominantly adopt the open conform-ation (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 quinuclidinering exists. Methoxy derivatives predominantly adopt the open conform-ation (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 andcircular dichroism spectroscopy as well as molecular mechanics compu-tations have shown that in ether solution dihydroquinidine existed in twoconformations, the open conformation (3) (anti, open) (80–90%) andclosed conformation (1) (syn, closed) (10–20%) separated by a barrier of8.3 kcal/mol.184

The results of computations favoured to the closed conformation (1) inthe gas phase and this discrepancy was explained by preferential solvatationof hydroxy group, which is sterically more available in anti conformation.Bulky substituents on the hydroxy group, such as in the p-chlorobenzoateester have the same effect.184

The conformation of cinchonidine in solution has been investigated byNMR techniques as well as with theoretical tools.88 Three conformers ofcinchonidine (CD) are shown to be substantially populated at room tem-perature, closed conformation (1), closed conformation (2), and openconformation (3). The latter is the most stable in apolar solvents. The sta-bility of the closed conformers relative to that of open conformer (3),however, increases with solvent polarity. In polar solvents the three con-formers have similar energies. The relationship between relative energiesand the dielectric constant of the solvent is not linear but resembles the formof an Onsager function.88

In o-dichlorobenzene or dimethyl sulfoxide solution the dihydroquinidine(DHQD) and (p-chlorobenzoyl)dihydroquinidine (p-ClBzDHQD) werefound to exist as an equilibrating mixture of two main conformers, seeTable 1.174 The relative amounts of these two conformers depend on con-centration as well as on solvent and temperature. Changes in the ratio of thetwo conformers of DHQD can explain the observed changes of the enan-tioselectivity in the indene rearrangement when the solvent was changedfrom o-dichlorobenzene to dimethyl sulfoxide.

Solute-alkaloid interactions are also able to influence the conformationalbehavior.183 In case of ester derivatives the energy difference between closedand open conformation is less and is probably of the same order of mag-nitude as the amount of stabilization caused by interactions with solutes,such as methanol or weak acids, or with strong electrophiles, such as os-mium tetraoxide.

In case of the methoxy derivatives the energy difference between closedconformation (2) and open conformation (3) has vanished. In non-

Catalysis, 2010, 22, 144–278 | 157

coordinating solvents like CD2C12, the methoxy derivatives are still pre-dominantly found in the closed conformation (2), but in the presence of anyelectrophile the equilibrium shifts towards the open conformation (3).Quinine and quinidine (and other hydroxy derivatives) by themselves al-ready possess a distinct preference for the open conformation (3) and thusdo not depend on extra stabilization caused by interactions with solute.183

Upon investigation of the above mentioned cinchona alkaloid catalyzedMichael addition of thiols to enones, it was found from the NOESY spectraof QD and QN in the presence of 4-methylbenzenethiol that the alkaloidconformations do not change on formation of an ion pair.175

When cinchona alkaloids are used as chiral bases, the main interactionwith the substrate comes from protonation of the tertiary nitrogen in thequinuclidine ring and a subsequent formation of an ion pair between theprotonated alkaloid and the deprotonated substrate molecule.183

When the alkaloids are used as chiral ligands, the main interaction is theformation of a dative bond between the tertiary nitrogen of the quinuclidinering and the metal atom of the substrate molecule (osmium tetraoxide).183

Investigation of the temperature effect has led the authors to come to animportant finding. Low temperature experiments at �20 1C and �60 1C inCDCl3, did not alter the 1H NMR spectra, no line broadening has beenobserved, and averaged spectra were still recorded at 40 1C. Thus, even atthese low temperatures, it was not possible to freeze out different con-formers. This was an indicative of a fast exchange between the differentconformations 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-openbarrier only half of this.174

The 1H NMR relaxation method was applied to QD. The proposedconformation had the following dihedral angles: o(C11–C10–C3–C4)=1501 and o(C40–C9–C8–C7)=701. The conformation of side-chaino(C11–C10–C3–C4) was found to be different from the one found forcrystalline form by X-ray analysis.186 Potential energy surface (PES) for QDhas been comprehensively investigated using the molecular mechanics

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 CDCl3, 0.2M 2.6 8.29 4.2 0.72 0.28

DHQD CDCl3, 0.02M 2.6 8.29 5.0 0.58 0.42

DHQD THF-d8, 0.02M 2.6 8.29 4.2 0.72 0.28

DHQD o-DCB-d4, 0.005M 2.6 8.29 4.8 0.61 0.39

DHQD o-DCB-d4, 0.02M 2.6 8.29 5.0 0.58 0.42

DHQD dioxane-d8, 0.02M 2.6 8.29 5.0 0.58 0.42

DHQD acetone-d6, 0.02M 2.6 8.29 6.3 0.35 0.65

DHQD DMSO-d8, 0.02M 2.6 8.29 7.2 0.19 0.81

p-ClBzDHQD CDCl3, 0.2M 2.5 8.73 7.4 0.21 0.79

p-ClBzDHQD o-DCB-d4, 0.02M 2.5 8.73 7.8 0.15 0.85

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

158 | Catalysis, 2010, 22, 144–278

(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 alsodetermined by NMR and supported with molecular mechanics.188 NOEinteractions in quinuclidine moiety of the cinchona ethereal isomers areshown in Fig. 6. Further structural information on cinchona derivatives willbe given in Section 6.1.

2.3.3 Solute-solute interaction. Intermolecular interaction of the alkal-oid molecules in solution can also be observed. Significant difference be-tween the NMR spectra of optically pure and racemic dihydroquinidineswas found under the same conditions (0.35M in CDCl3). The spectraldifferences were greatly reduced when CD3OD was used as solvent. Theacetates of optically active and racemic dihydroquinine showed significantlysmaller differences than those observed with dihydroquinine. The authorshave explained the observations by solute-solute interactions of theenantiomers.189

Osmometry was used to measure average molecular weight for quinine.Results of these experiments have indicated the presence of particles largerthan monomeric quinine at 37 1C for a 16mM solution in toluene. Forconcentrationso4mM the quinine was almost completely monomeric.182

The coexistence of monomer and dimers of quinidine was established inquinine-chloroform solutions by investigating the temperature and con-centration dependence of the NMR spectral parameters by combination of2D NOESY and proton-selective relaxation rate measurements. Similarconformation of the alkaloid was found both in the dimer and monomerforms. It was shown that the quinuclidine ring is on one side of the quin-oline ring and the CHOH moiety on the other, with the quinoline planealmost bisecting the angle between C-H8 and C-OH190 see Fig. 7.

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

Catalysis, 2010, 22, 144–278 | 159

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

3. Alkaloids used in Oritos’s reaction

Studies aimed at systematic variation of cinchona modifiers and theiranalogs have played a definite role in building up hypotheses for themechanism of Orito’s reaction. The structural units of cinchona alkaloidshave been discussed in previous section. There are different reviews46,192,193

related to the analysis of modifiers used in asymmetric hydrogenation ofactivated ketones. For this reason in this section only the key issues will bebriefly mentioned.

We shall apply the following classification for chiral modifiers applied (i)flexible cinchona alkaloids, (ii) flexible cinchona derivatives, (iii) rigid cin-chona derivatives, (iv) flexible cinchona analogues, and (v) other type ofchiral templates.

3.1 Flexible cinchona alkaloids and their derivatives

Chiral templates most often used in the heterogeneous catalytic asymmetrichydrogenation of activated ketones are natural cinchona alkaloids such as,

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

160 | Catalysis, 2010, 22, 144–278

CD, CN, QN and QD (see Fig. 8).33,40,51 CD is the most frequently in-vestigated 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 Ingeneral CD is a better modifier than CN. This difference is more pro-nounced in ethanol than in toluene, but in AcOH the difference is negligible.With the exception of epicinchona alkaloids196 and isocinchonines197 it is ageneral observation, that the configuration of C8 or C8 and C9 atoms of thecinchona alkaloid determines the product distribution.57,192,198 Already inone of the first studies it was evidenced that tertiary N in the quinuclidinemoiety of cinchonas plays crucial role194,198,199 although in recent studies itwas shown that in case of ketopantolactone200 even N-alkylated derivativesof CD can induce very slight enantioselection.

Surprisingly the N-oxide derivative of CD has also resulted in enantio-selection. A possible reason is that N-oxide can be reduced very fast underthe 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, butits quinoline ring. This is an undesired side reaction leading to the sub-stantial loss of enantioselectivity.57,195,199 It has been suggested that thedecrease in the ee values upon using CD derivatives with partially hydro-genated quinoline be attributed to a weaker adsorption of the alkaloid tothe Pt surface.192 The phenomenon can also be explained by the loss of theshielding effect of the aromatic p-system required for chemical shielding viap–p interaction83 (see Section 8.3).

In a detailed study18 different cinchona analogs and 8 different substrateswere investigated. The results indicated that no ‘‘best’’ chiral template exists

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

Catalysis, 2010, 22, 144–278 | 161

for all substrates.201 This finding indicates that interactions between thesubstrate and the chiral modifier template depend on various factors.201

C9 substituted compounds represent an important group of flexible cin-chona alkaloids. 9–O-methyl-10,11-dihydrocinchonidine (MeODHCD) themost frequently used C9 derivative generally behaves slightly better thanCD in the hydrogenation of a-ketoesters.194,199,202,203 Similar positive re-sults were obtained upon using other substrates.44,199,204 However dike-tones, such as 1-phenyl-1,2-propanedione produces lower ee in the presenceof MeODHCD than in the presence of CD.205 Detailed studies on the use ofthese alkaloid derivatives can be found elsewhere.202,206 With respect to theuse 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 thecase of bulky O-substituted derivatives of CD relative to DHCD has beenattributed to a tilted mode of adsorption of these chiral templates to the Ptsurface206 (see Sections 6 and 8).

3.2 Rigid cinchona derivatives

C8(S) C9(R) type of cinchona alkaloids, such as CN, QD and cupreidinecan form inner ether derivatives. These derivatives are called ‘‘rigid’’ as inthese alkaloids the rotation around the bond C8–C9 as an axis is notpossible.

These alkaloids were used to demonstrate that the formation of closedconformation of alkaloids is not a prerequisite for the formation of sub-strate-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 cinchonaalkaloids, such as aromatic ring, chiral moiety, and basic nitrogen. Im-portant feature of these new analogues is the presence of an aromatic groupin the close neighbourhood of the stereogenic center. Different types ofenantiomerically pure primary and secondary aminoalcohols78 and amineshave been tested as chiral templates in the hydrogenation of pyruvateesters,207,208 ketopantolactone,209 trifluoromethyl ketones,210 1,1,1-tri-fluoro-2,4-diketones, etc.44

3.3.1 Aminoalcohols. Series of enantiomerically pure 2-hydroxy-2-ary-lethylamines (see Fig. 9) has been prepared from the corresponding ole-fins.78 Upon using compound A in the hydrogenation of EtPy ee valueshigher than 75% was achieved.207,208 The replacement of the naphtyl ringby 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 significantdecrease in both ee (o5%)60 and reaction rate.

3.3.2 Amines. Upon using (R)-1-(1-naphthyl) ethylamine as chiraltemplate in the asymmetric hydrogenation of EtPy 82% ee has beenachieved in AcOH. It has been shown that (R)-1-(1-naphthyl) ethylamine isonly a precursor of the actual chiral template, which is a secondary amine(aminoester) formed in situ from (R)-1-(1-naphthyl) ethylamine and EtPyby condensation to the corresponding imine and subsequent reduction of

162 | Catalysis, 2010, 22, 144–278

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

3.4 Other type of chiral templates

Other natural alkaloids and their derivatives were also applied as chiraltemplates in the asymmetric hydrogenation of activated ketones althoughthe ee values obtained were much lower than over CD and its derivatives.Blaser and coworkers tested about 100 different chiral auxiliaries, but theynever found any meaningful enantioselectivity.213 Ephedrine has given lowor moderate optical yields in the hydrogenation of a-ketoesters.214 Codeine,strychnine, and brucine have provided only 2–12% ee.215 Using tri-fluoromethylcyclohexyl ketone substrate brucine has not resulted in opticalyield.216 (� )-Dihydro-apovincaminic acid ethyl ester has also been appliedas chiral template for EtPy substrate (27–30% ee).68,217–219 Other vincaderivatives have also been tested but (� )-dihydro-apovincaminic acid ethylester has been found to be the most effective one.68,220,221 Upon using othercompounds as chiral templates in the hydrogenation of EtPy (S)-a,a-diphenyl-2-pyrrolidinemethanol222 resulted in moderate ee (max 25%) de-pending on the type of the solvent. (S)-proline chiral auxilary has also beentested.223 During the hydrogenation of EtPy in the presence of (S)-prolineresulted in the formation of N-alkylated proline while in case of isophoronesubstrate a diastereoselective oxazolidine type intermediate was formed in acondensation reaction. Hydrogenation reaction itself proved to be

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

Catalysis, 2010, 22, 144–278 | 163

diastereoselective. Isophorone has produced ee up to 60% but (R)-ethyllactate has been formed in very low optical purity (1–5% ee).223 (S)-prolinederivatives, such as Z-(S)-proline 2-naphthyl ester, Z-(S)-proline 2-(2-naphthyl)-ethyl ester, Z-(S)-proline 3-ethyl-indole ester and (Z)-(S)-proline-3-ethyl-indolamide, (S)-proline-2-naphthylamide hydrochloride has alsobeen tested as chiral templates of new type in case of EtPy.224 a,a-Diphenyl-L-prolinol chiral template resulted in 14% (S) in the hydrogenation of tri-fluoromethylcyclohexyl ketone.216 Dextrocarbinol base has induced noenantioselectivity in the same reaction.216 ‘‘Troger’s base’’ ((5R,11S)-(þ )-2,8-dimethyl-6H,12H-5,11methanodibenzo[d,f][1,5]diazocin) as a chiraltemplates has given 65% ee using acetic acide solvent in the asymmetrichydrogenation of EtPy.225 (R)-(� )-2-phenylglycinol has induced poorenantioselectivity 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-(a-methylbenzyl) phtalic acid monoamide, (R)-(� )-1-1-naphthyl) ethyliso-cyanate has given moderate ee (23%, 59% respectively).225

It has been shown that 1-naphthyl-1,2-ethanediol226 is an effective chiralmodifier in the hydrogenation of KPL and ethyl-4,4,4-trifluoroacetoacetate.It is the first effective nonamine-type chiral template used in Orito’sreactions.

4. Methods and approaches used

4.1 General information

In this section methods and approaches applied in the enantioselectivehydrogenation of activated ketones will be described. One of the charac-teristic features of this catalytic system is the need for catalyst pretreatmentin hydrogen at 350–400 1C prior to the reaction. The omitting of pre-re-duction step resulted in low rates and low enantioselectivities. The need forcatalyst pretreatment has already been described by Orito’s group.40 In alater study it has been shown that the modification of the Pt surface by thealkaloid requires pure Pt sites.63 Recently a new type of Pt/Al2O3 catalysthas been developed by Degussa (catASium F214) which can be used withoutpre-reduction.227 This catalyst gives high rates and high ee values when it isused as received.

Some authors claimed that the aerobic treatment of the catalyst, i.e. theformation of chemisorbed oxygen on the Pt sites, is needed to improve boththe reaction rate and the ee values.65,228,229 This issue will be discussed inSection 4.3. The use of ultrasound resulted in also an improved performanceof supported Pt catalyst.70,230 The other important issue is the mode ofintroduction of the modifier. In Orito’s approach pre-modification has beenused.40 The discovery of in situ modification was the next important find-ing.63 Upon using in situ modification the ‘‘ligand acceleration’’ phenom-enon has been discovered.58 However, it has to be mentioned that rateacceleration (RA) was not observed for all substrates and all modifiersinvestigated.

Based on this fact recently same groups questioned the validity of the rateacceleration phenomenon.232–234 This issue will be discussed in Section5.5.1.

164 | Catalysis, 2010, 22, 144–278

Most of the authors are calculating either the initial rate or the maximumrate observed after a short induction period. Unfortunately, due to sidereactions and the transformation of the alkaloid during the hydrogenationreaction (see Section 5.1) the analysis of whole kinetic curve is very trou-blesome, although there were attempts to do that.100

We should like to emphasize that in order to get a full picture about thepeculiarities of these unique reactions reliable data with respect to both thereaction rates and the optical yields should be provided. One of the mostserious problems is that in large number of recent publications the rates arenot given at all.210,235–237 Only conversion or yields values measured at theend of the reaction are compared.

There were couples of papers devoted to the problems of reproducibilityand 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 theorigin of the substrate as well as on the batch number.69,93 Systematicstudy of this effect was done by the Ciba group. These results are shown inTable 2.195

Of course, the purity of the reactant and solvent has a great impact on thevalidity of kinetic results, consequently results obtained upon using un-purified substrates, especially ketoesters, has to be treated with greatconcern.

Researchers with sufficient background in organic chemistry realized veryearly that the purification of the substrates before the use is a must. The useof purified modifier is also very important.238 There were strong disputes inthe literature concerning to the use of unpurified or contaminated sub-strates.66,69,82,228 The lack of background in organic chemistry often re-sulted in strange and unreliable results. For instance, the reactionmechanism was investigated by different groups in alcoholic solvent, despitethe fact that the most investigated substrates, i.e., EtPy or MePy, react withlinear 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. (Re-

produced from ref. 194 with permission)

Origin

Undistilled Distilled

J, T, 20 J, T, 20 E, T, 20 J, Ac, 20 E, Et, 20 J,T,100 Average

rate ee rate ee rate ee rate ee rate ee rate ee Rate ee

Fluka91 3 63 12 69 10 69 12 83 25 74 56 82 20 73

Fluka92 4 78 44 80 50 80 64 88 90 74 96 83 58 81

Lancaster 7 71 14 78 14 77 26 87 36 77 48 85 24 79

ICN, Ohio 9 77 15 79 05 79 38 89 40 78 64 86 30 81

Sigma 9 76 24 80 20 80 46 87 40 78 114 87 42 81

Jansen 18 80 24 83 18 83 46 90 50 80 102 89 43 84

R.de Haen 5 73 30 81 21 79 46 87 50 78 148 87 50 81

Aldrich 50 84 70 85 36 85 76 91 70 84 132 90 72 87

TCI, 50 82 48 83 62 83 78 90 68 70 164 80 78 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 disputesand argues even these days it is possible to find papers, where no words issaid how the substrate was purified or what even is worst, there are publi-cations where unpurified substrate has been used.233,239 These facts oftenresulted in misinterpretation of experimental results. These issues will bedescribed in Sections 5.5.1 and 5.5.2.

In all pioneering studies, i.e., in early nineties, the determination of op-tical yield was not an easy task, especially at low conversion. For this rea-son, the changes in the ee values with conversion were not investigated,consequently the anomalous monotonic increase type ee-conversion de-pendencies 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, ATR�IR, AFS, Raman spectroscopy, etc. has been used inorder to elucidate the reaction mechanism or the origin of RA and enantio-differentiation (ED) (see Section 6.6). The common problem related to thesestudies is that conditions of these measurements are far away from thoseused in real hydrogenation reaction, although some measurement methods,such as (ATR-IR) were performed under condition close to hydrogen-ation.240 A typical misused situation is when the chemisorption of CD wasinvestigated by electrochemical methods in concentrated H2SO4. Based onthese results it was concluded that the adsorption of CD on Pt (111) is ir-reversible.241 The problem with these results is that those who need someadditional proof with respect to ‘‘surface induced’’ RA and ED like theseresults 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 cata-lytic step from those, what are formed on the surface of platinum, but arenot involved in the catalytic act? The latter species are often called as‘‘spectators’’ in a given catalytic reaction. In many cases the surface con-centration of ‘‘spectators’’ can be much higher than that of the ‘‘actors’’. Inthis respect let us remind the reader for the classical problem in homo-geneous catalysis discussed by Halpern.225 In his classical study it wasdemonstrated that in homogeneous catalytic enantioselective hydrogen-ation not the most stable [substrate-catalyst] complex is involved in the EDstep, but the less stable one, what reacts with hydrogen much faster than theformer.

In connection to the above discussion the use of sophisticated surfacetechniques for the elucidation of the origin of ED has to be mentioned.None of these methods can fully guarantee that the observed surface speciesis really involved in the given step of enantioselective hydrogenation.Consequently, it is almost impossible to distinguish, whether an identifiedsurface entity is an ‘‘actor’’ or just a ‘‘spectator’’.

4.2 Catalysts applied

4.2.1 Supported metal catalysts. In the enantioselective hydrogenationof activated ketones supported Pt is the most commonly used catalyst. Pt/Ccatalysts have been used by Orito in his original approach. Alumina sup-ported Pt catalysts containing around 5wt% metal are the most commonly

166 | Catalysis, 2010, 22, 144–278

used catalyst. Two industrial catalysts, E 4759 from Engelhard and JMC 94from 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 hasrather small pores and a low pore volume, while JMC 94 is a wide-porecatalyst with a large pore volume. There are reports related to the use of Ptcolloids 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 andRh252–256 is considered as a curiosity, although recent results using rhodiumis very promising.257 Supported iridium catalysts were used in the enan-tioselective hydrogenation of diketones in order to suppress the hydrogen-ation of the second carbonyl group.258 Palladium is not a suitable metal forthe hydrogenation of keto carbonyl group.

Pt supported on HNaAY259 and ZSM-5 zeolites,260 MCM-41261 meso-porous materials, clays262 and ion exchanges resins250 were also tested in theenantioselective hydrogenation of EtPy, however, the performance of thesecatalysts 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 lowee values (below 35%) and very moderate reaction rates.263 There are re-ports on the use of carbon nanotubes as support.264 We consider that allhigh surface area materials are inefficient supports for this reaction, due totheir high adsorption power resulting in high modifier concentration at thesupport and lowering the modifier concentration in the liquid phase.

In earlier studies it has been suggested that Pt dispersion has a decisiveinfluence on both the activity and ee and it was suggested that in order toobtain high optical yields the dispersion should be r0.2.63 It has beensuggested that an appropriate flat Pt surface be required to accommodatethe modifier or the modifier-substrate complex in order to get pronouncedED.265

Contrary to the above results and suggestions results upon using aPt/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 Furtherresults on Pt nanocolloids prepared,79,243–246 indicated also that there is noreal 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 arestabilized by nitrogen and oxygen containing ligands. Under properlychosen experimental condition these Pt colloids show high activity andrelatively high enantioselectivity.246 Pt colloids were also used in kineticinvestigations.100 It was demonstrated that the RA could also be observedwhen Pt colloids were used.

In this respect Pt colloids stabilized by cinchona alkaloids have thegreatest interest. The concept of using chiral stabilizing agent for thepreparation of Pt colloids has been applied by Bonnemann.79 These colloidswere used to hydrogenate EtPy. Upon using DHCD or CD as stabilizingagent the mean size of Pt colloids was in the range of 1.5–2.8 nm. It isinteresting to note that upon using these colloids in the hydrogenation ofEtPy ee values in the range of 75–80% were obtained. In a recent study Ptnanocolloids stabilized by cinchona alkaloids were used in enantioselective

Catalysis, 2010, 22, 144–278 | 167

hydrogenation of EtPy in as received form or immobilized on varioussupports.266

It is interesting to mention that colloids prepared by Bonnemann’smethod required addition of free alkaloid to induce high rate and high eevalues,. The addition of alkaloid to the solution increased both the reactionrate and ee values in the range of 80–85% were obtained in theAcOHþMeOH mixture.244 The results indicated that two forms of thealkaloids 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 alkaloidsadded into the liquid phase. Under condition of enantioselective hydro-genation these two forms are in dynamic equilibrium. The Pt colloid pre-pared upon using cinchonine (PtCN) was used to investigate the possibleexchange between the two forms of the alkaloid. These results are presentedin Fig. 10.

In the first experiment the PtCN colloid was used and the concentration of(CN)ex was 6.8� 10�3M. In this experiment the ee was independent of theconversion 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� 3M. Theinitial ee values (ee=� 0.6) show that at low conversions the initial (CN)stform is involved in the events controlling the asymmetric induction. As thereaction proceeded the (CN)st form was exchanged by (CD)ex resulting in adecrease in the ee values. The final ee value (ee=0) indicates that the aboveexchange is almost quantitative. This result indicates that there is an ex-change 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

-0.6

-0.2

0.2

0.6

1.00.0 0.2 0.4 0.6 0.8 1.0

Conversion

ee

Fig. 10 The ee – conversion dependencies obtained in the presence of PtCN varying thecharacter of excess alkaloid ((CN)ex or (CD)ex).7 – experiment in the presence of (CN)ex; D –experiment in the presence of (CD)ex without ultrasound treatment,E,B – experiment in thepresence of (CD)ex after treatment with ultrasound. The concentration of excess alkaloids is6.8� 10� 3M; nanocolloid=0.020 g [EtPy]0=0.6M, pH2=5 bar, solvent: CH3COOH/MeOH(5/1), Tr=12 1C. (Reproduced from 244 with permission)

168 | Catalysis, 2010, 22, 144–278

the ee value was constant but was opposite in sign, i.e., during the ultrasonictreatment full exchange between the two forms of the alkaloid ((CN)st and(CD)ex) took place (see Fig. 10). The phenomenon appeared to be fullyreproducible.

According to computer modeling the above Pt nanocolloids have particlesize in the range of 1.6–2.8 nm, i.e., the size of accessible Pt (111) surface isvery small (3� 4 or 5� 5 Pt atoms). This small Pt colloid can accommodatethe ‘‘shielded’’ modifier-substrate complex, while the accommodation of theopen modifier-substrate complex would require much larger surface sites.

4.2.3 Characteristic features of supported Pt catalysts used. The resultsdiscussed so far indicated there is no need to have a preferred particle size ashigh ee values were obtained over catalyst having broad Pt dispersion range.However, in active and enantioselective catalysts the Pt sites should berelatively clean. Both the pre-treatment in hydrogen at 300–400 1C and thetreatment in ultrasound can provide clean Pt surface.

Another important issue is that the catalyst used has to be relatively inertrelated to in the hydrogenation of the quinoline ring of the alkaloid. Con-ditions for ring hydrogenation of modifiers were investigated in variousstudies.267–271

Although it has been shown that upon using Al2O3 support in AcOH Aloxonium ions and their adducts with the alkaloid has been detected272 theinvolvement of these species in the catalytic reaction is quite doubtful. Thesuggested ‘‘electrostatic acceleration’’273 needs further experimental proof.

As a rule the support should be relatively inert. Highly acidic supportscan induce acid catalyzed undesired side reactions. Both the high acidityand the high surface area of supports decrease the amount of alkaloidavailable for ED. It was shown that Cl containing alumina precursor andchlorine-containing platinum salts exhibit significantly higher optical yieldthan similar catalysts prepared from chlorine free starting-materials.265 Ithas also been demonstrated that the modification of alumina support byalkoxy-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 thebeneficial effect of preheating the catalyst in hydrogen at 300–400 1C priorto the modification was emphasized.40,41 The selection of a proper pre-treatment procedure for supported Pt catalysts is one of the basic issues.Several other pretreatment methods were applied and different explanationswere given for the favourable effect of reductive, aerobic and ultrasonictreatments. Fig. 11 shows the general scheme for catalysts pretreatment.76

The common feature is the reduction of the catalyst used at relatively hightemperature (300–400 1C). It is called reductive treatment. The catalyst canbe cooled either in hydrogen or in an inert atmosphere. In oxidativetreatment after the reductive treatment the catalyst is treated in air andcooled down in an inert atmosphere. Most of the authors agree thatupon using supported Pt catalyst a reductive treatment is a must and specialcare has to be done to prevent contamination of reduced catalyst with

Catalysis, 2010, 22, 144–278 | 169

oxygen. However, there are groups using pre-reduced catalysts kept orstored in air.275

4.3.1 Prereduction in hydrogen. In one of the first publications usingin situ modification of Pt/Al2O3 catalyst63,276 it was mentioned that thethermal treatment at 400 1C hydrogen has pronounced effect both on theactivity and the enantioselectivity. After thermal treatments in hydrogen at400 1C, 15–20% higher ee values were obtained than over untreatedcatalysts.

In a recent review195 the role of pretreatment was formulated as follows:(i) pretreatment cleans up the surface of the catalyst by removing oxygen aswell as impurities; (ii) residual Pt salts are converted to metallic Pt; (iii) theaverage particle size of Pt increases, (iv) the morphology of Pt particles, i.e.the distribution of exposed face, edge and corner atoms is also alteredfavourably; (v) it promotes adsorbate-induced surface restructuring.

Restructuring during pretreatment of Pt/alumina catalyst used in enan-tioselective hydrogenation of KPL was also studied.277 The influence ofreductive and oxidative heat treatment on the enantioselectivity of chirallymodified Pt/alumina has been reinvestigated. Enhancement in ee by39–49% has been observed after treatment in hydrogen at 250–600 1C, ascompared to untreated or pre-oxidized catalysts. The changes in ee afterreductive and oxidative treatments are reversible, and always the finaltreatment is decisive. A HRTEM study indicates that adsorbate-inducedrestructuring of Pt crystallites during hydrogen treatment at elevated tem-perature can play a role in the selectivity improvement, but the changes aresuperimposed by the strong structure-directing effect of the aluminasupport.

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

170 | Catalysis, 2010, 22, 144–278

4.3.2 Influence of oxygen. The effect of oxygen on the performance ofcinchona – Pt catalyst system has been studied by different groups undervarious conditions.65,229 In these studies different solvents, different type ofsupported Pt catalysts and different experimental conditions were used. Forthis reason it is very difficult to make any right conclusion or interpretationrelated to the given observation; is a particular finding a general phenom-enon or an experimental artifact?

In a recent review33 a generalized comment was given, namely the effect ofmodification atmosphere of Pt-CD catalysts affects on the ee values. Forinstance under air higher ee values has been achieved in EtPy hydrogen-ation, whereas under anaerobic condition ee decreases drastically.229,278 Inref. 229 it was demonstrated that the addition of oxygen during the enan-tioselective hydrogenation of EtPy has a positive effect both on the rate andthe ee values. The observed effect was attributed to restructuring of thesurface of Pt in the presence of oxygen. There is only one remark with re-spect to these findings, i.e., the final ee value (below 40%) is extremely lowfor the experimental conditions applied.

When anaerobic and aerobic treatments of Pt/SiO2 catalyst was com-pared after anaerobic treatment decreased enantioselectivity and greatlyreduced activity was observed using DHCD as modifier in the hydrogen-ation of MePy.65 It has to be mentioned that this pretreatment was per-formed in ethanol. In further studies it was shown that during this aerobictreatment ethanol was oxidized over platinum catalyst into acetic acid279

and probably the formed AcOH was responsible for the increased per-formance. Similar prove has been obtained in our laboratory.281

Table 3 shows the results obtained in the enantioselective hydrogenationof trifluoroacetophenone (TFAP).76 These results clearly show that eithertreatment in an oxygen atmosphere or stirring in air resulted in a decrease inthe enantioselectivity.

Bartok and coworkers have applied a reductive treatment prior to the useof catalyst, but the catalyst is stored in air before its final use. It was shownthat the increase of the storage time up to one week has no pronouncedeffect on the performance of the catalyst.277 With respect to the role ofoxygen it was also suggested that during this treatment PtO could beformed. During the hydrogenation reaction PtO is reduced to metallic Ptand water. It is not excluded that the presence of small amount of water canresult in some improvement in the performance.76 In the enantioselectivehydrogenation of MePy or butane-2,3-dione over Pt in the presence of CDthe coadsorption of oxygen with the alkaloid resulted in a positive effect

Table 3 Influence of catalyst pretreatment in the hydrogenation of TFAP (90mg 5w%

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 – toluene 2 63 89 16

2 reductive A toluene 2 50 96 33

3 reductive B toluene 2 86 95 33

4 reductive A 1,2-dichlorobenzene 4 105 95 45

5 oxidative 1,2-dichlorobenzene 4 105 91 29

Catalysis, 2010, 22, 144–278 | 171

both on the rate and ee.278 It was suggested that the presence of a strong co-adsorbate, such as oxygen, the surface was not poisoned by CD. In a recentstudy the promoting effect of helium treatment was also mentioned.281

However, it was admitted that the effect is due to the small oxygen impurityin the helium used.

4.3.3 Use of ultrasound and microwave heating. The effect of ultrasoundradiation was investigated in details by Bartok’s group.55,70,282–284 Themethod appeared to be highly efficient as the ultrasound radiation resultedin and increase both in the reaction rate and the enantioselectivity. Table 4shows representative results using three different substrates and threemodifiers.55 The decrease in metal particle size was given as an explanationof improved performance.282 In another study TFAP was applied as asubstrate and the use of sonication resulted in positive effect.273 The resultsindicated also that both the frequency of ultrasound and the duration ofsonication have a strong influence of the enantioselectivity.55

However, it was also pointed out that the presence of the modifier isabsolutely crucial during sonication. It was proposed that the ultrasonicirradiation created a more effective surface modification, resulting in theformation of surface sites required for optimum enantio-differentiation.Besides it an additional positive effect of oxygen was also observed.273 Itwas suggested that ultrasonic irradiation helps the removal of the impuritiesfrom 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)

Entry Substrate Modifier Catalyst Major product

Optical yield (ee %)

‘‘silent’’ ‘‘sonication’’

1 1 4 E40665 R 85 97

2 1 5 E40665 S 78 83

3 1 6 E40665 R 93 98

4 2 4 E4759 R 88 92

5 2 5 E4759 S 34 57

6 2 6 E4759 R 60 68

7 3 4 E4759 R 79 92

8 3 5 E4759 S 50 92

6 3 6 E4759 R 83 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 over5wt% Pt/SF (silica fiber) catalyst a notable enhancement of reaction rate, eeand rs was observed under ultrasound compared to silent conditions.231 Inmesitylene solvent four-fold increase in the reaction rate was observedunder ultrasound compared to identical silent conditions, while in methylacetate and in toluene the rate enhancement was only minor. Upon usingPt/SF catalyst it was suggested that surface smoothening and cleaning takeplace under ultrasound irradiation. However, no significant differences in Ptparticle size distribution between sonic and silent treated catalysts wereobserved by TEM.

Summing up results related to the use of ultrasound its positive effect hasbeen ascribed to the following facts:55 (i) ‘‘through the decrease in metalparticle size, the platinum dispersion becomes close to optimal using acatalyst of large metal particle size,’’ (ii) ‘‘the surface density of the modifierincreases as a result of insonation, providing more chiral sites for the hy-drogenation and, in parallel, suppresses the background reaction, i.e. ra-cemic hydrogenation’’.

Enantioselective and racemic hydrogenation of EtPy over Pt/Al2O3

catalyst was investigated under microwave dielectric and conventionalheating.285 A homemade laboratory microwave loop reactor was appliedallowing differentiating between dielectric and conventional heating. Theeffects of polar and non-polar solvents on enantioselective hydrogenation ofEtPy were studied in toluene and ethyl alcohol. In case of toluene, which ismicrowave transparent, no significant differences in the reaction rate andenantioselectivity were observed between dielectric and conventional heat-ing. In case of EtOH, the reaction rate remained unaffected. However, the eedramatically decreased from 60 to 40% under microwave heating. No sig-nificant improvement of the reaction rate with an increasing microwavepower input was observed. The authors suggested that this observationcaused by the local superheating of the polar EtOH in the cavity, which isnot possible in the non-polar toluene. Our explanation is different; namelythe decreased ee is due to the formation of semi-ketal from the substrate andthe 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 apremodification procedure was used to introduce the chiral modifier intothe Pt/C catalyst pretreated in hydrogen at 400 1C.40 It has to be emphasizedthat the premodification was performed at higher temperature than that ofthe hydrogenation reaction. During premodification the catalyst and thecinchona alkaloid has been stirred in a given solvent for a relatively longperiod (24 hours). This premodification procedure strongly resembled themodification process used for Ni/tartaric acid catalysts developed earlier.286

The premodification was followed by filtration and mild washing and thepremodified catalyst was introduced into the reactor containing the solventand 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. Furtherresults indicated if the filtration and the washing is not carefully performedsolvated 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 alkaloidintroduced was not really known. Probably it was the reason that in ref. 287the authors admitted the ‘‘erratic variation of the initial rates’’. After suc-cessful introduction of the in situ methods57 and demonstration of the ad-vantage of premixing technique64 the premodification method was almostentirely forgotten. This approach is still used when the enantioselectivehydrogenation is investigated in the gas phase.288

4.5 In situ modification

4.5.1 General Information. In situ modification of the catalysts prior tothe hydrogenation of activated ketones can experimentally be performed indifferent way. These various approaches provides different surface coverageat t=0, i.e., prior to the start of the hydrogenation reaction.

Let us consider that the Pt catalyst is pre-reduced in hydrogen around400 1C and kept in an inert atmosphere prior to its use. The catalyst in thisform 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 completelydifferent surface coverage at t=0, what can have different influence both onthe 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 pre-mixed prior to the hydrogenation reaction. This method provides highcoverage 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 proportionalto the amount of alkaloid used. This phenomenon will be discussed inSection 5.5.1.

During premixing the substrate can decompose resulting in carbonmonoxide, which is considered as a strong catalyst poison. The substrateinteracts also with the modifier and induces the formation of high-molecularweight byproducts (see Section 5.1). These by-products have a negativeinfluence on the initial rate by their poisoning effects. If the amount ofhydrogen in the overall hydrogen pool is high (i.e., when the catalyst iscooled in the hydrogen atmosphere from the temperature of re-reduction) inthis case partial hydrogenation of both the cinchona alkaloid and thesubstrate can also take place prior to the introduction of hydrogen.

Consequently, upon using the premixing technique a new problem ap-peared, i.e. the reproducibility of the reaction rate. It was observed that therate of reaction depended on the duration of premixing, i.e. the time re-quired to close the high-pressure autoclave, purge the reactor with nitrogenand hydrogen, and pressurize the reactor and start stirring.64 Reliable andreproducible 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 alcoholshave been used as solvents.

4.5.3 Injection technique

Injection of the modifier. This method excludes all undesired interactionsbetween the substrate and modifier. It provides high coverage of the sub-strate at the Pt site. However, if the catalyst is not cooled in an inert at-mosphere racemic hydrogenation of the substrate can also take place beforestarting the reaction (i.e. before t=0). Neither the spontaneous oligomer-ization/condensation of the substrate or its decomposition into CO can beexcluded in this case.

Injection of the substrate. This method excludes also all undesired inter-actions between the substrate and modifier. It provides a definite surfaceconcentration of the modifier. No oligomers or condensation products existat t=0. However, partial hydrogenation of the modifier cannot be excludedif the catalyst has been cooled in a hydrogen atmosphere.

The injection method gives an opportunity to get reliable kinetic dataprovided a proper ratio between the batch and the injected volumes ischosen. In this way the influence of some of the undesired side reaction canbe eliminated, providing more chance to get intrinsic kinetic data. In thisapproach either the modifier or the substrate is injected by high-pressurehydrogen.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 heterogeneouscatalysts become very efficient when the fixed-bed reactor is used; con-sequently it is very promising to introduce fixed bed reactors with the aim toindustrialize the asymmetric catalysis. However, up to the late nineties therewere only scarce data related to the use of continuous-flow reactors inasymmetric hydrogenation reactions.

As far as only trace amount of modifier is required to induce high ee anattempt was done to use a continuous fixed-bed reactor for the enantiose-lective hydrogenation a KPL.290 This approach resulted in significant pro-cess intensification; consequently upon using a small tubular reactor (size ofa 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 continuousfeeding of minute amounts of chiral modifier to the reactant stream. The eevalues for KPL and EtPy without optimization was 83.4 and 89.9%, re-spectively. Transient measurements by stopping of the flow of CD indicatethat 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-phenyl-1,2-propanedione giving relatively high enantiomeric excesses.294 Theknitted silica fiber catalyst gave encouraging results in the continuous fixedbed operation with enantiomeric excesses comparable to those obtained inthe batch reactor.

EtPy was hydrogenated in a continuous-flow fixed-bed reactor and highee 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 ethyl-2-oxo-4-phenylbutyrate (EOPB) on Pt/g-Al2O3 catalyst in the presence ofCD was investigated in a fixed-bed reactor with the aim to synthesizeenantiomerically pure (R)-(þ )-EHPB, a building block for the synthesisof several commercially important A.C.E. inhibitors. The highest ee valuearound 69% was obtained in toluene at 6MPa hydrogen pressure. Al-though stable conversion values were obtained in time on stream experi-ments, the ee values decreased in time.

The transformation of isopropyl-4,4,4-trifluoroacetoacetate to the cor-responding hydroxyester was studied in a fixed bed reactor over Pt/Al2O3 inthe presence of MeO-CD and trifluoroacetic acid.297 Around 0 1C and uponvarying the pressure, the total liquid flow rate and the feed composition eevalues up to around 90% were achieved. However, the time on streampattern showed quite notable activity decrease.

Enantioselective hydrogenation of EtPy was performed in a continuous-flow fixed-bed reactor using supercritical carbon dioxide and supercriticalethane (scC2H6).

299 In the latter solvent much higher catalytic activity wasobserved.

Ethyl benzoylformate (EBF) was hydrogenated in a continuous-flowfixed-bed reactor over Pt/Al2O3 catalyst in the presence of CD and CN.299

Variety of chemical and physico-chemical methods was applied to pretreator clean the chiral fixed bed between multiple hydrogenation reactions. Itwas observed that after an enantioselective hydrogenation with CD asmodifier at 0 1C, the continuous-flow reactor could be effectively cleaned at0 1C, and that a racemic unmodified hydrogenation could be performedthereafter. This implies the effective desorption of chiral species from thesurface during cleaning. Contrary to that cleaning of the reactor at 50 1Cresulted in a reproducible unmodified enantioselective hydrogenation, with amarked inversion of enantiomeric excess. The inversion of ee will be dis-cussed in Section 5.5.4.

In a recent study enantioselective hydrogenation of EtPy was also per-formed in continuous-flow reactor in the presence of CD over Pt/Al2O3.

300

All these results indicate that the use of continuous-flow reactors shouldbe applicable to different substrates based on the use of chiral modifiers andsupported metal hydrogenation catalysts. This approach provides moreefficient screening method for potential chiral modifiers establishing thebasis for future technical applications. Based on the use of continuous-flowreactors the so called ‘‘chiral switch’’ methodology has been developed forthe investigation of the relative adsorption strength or the competition ofchiral 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 isstrongly connected to the catalyst deactivation phenomena. Deactivation willalso be discussed in Section 5.2. Due to catalyst deactivation for the reuse ofsupported Pt catalysts fresh modifier has to be added before each hydrogen-ation cycle,302,303 or the modifier is fed permanently in continuous man-ner108,290 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 for30min, the liquid was removed, and solvent, EtPy and occasionally modifierwere added to the reactor. Hydrogenation was repeated as describedabove’’.268

An intelligent approach has been developed for the reuse of hydrogen-ation catalyst.304 Magnetic Pt/SiO2/Fe3O4 catalyst was prepared and suc-cessfully applied for the enantioselective hydrogenation of various activatedketones. 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 externalmagnetic field and recycled several times with almost complete retention ofactivity and enantioselectivity.

Figs. 12A and B show two types of reuse experiments. In Fig. 12A freshmodifier has been added in each run, consequently the ee values are con-stant; however there is a significant catalyst deactivation. However, asshown in Fig. 12B both the activity and ee decrease on reuse of the catalystif no fresh modifier is added to the reaction mixture at the beginning of eachnew run. It is known from other studies246,248,261 that almost constant eevalues can be achieved in reuse experiments, where ‘‘fresh’’ modifier isadded in every reuse.

Interesting observation was described in ref. 305. Stopping the enantio-selective hydrogenation of EtPy at a conversion of approximately 70%,long-term stability of the catalyst can be achieved. During 10 cycles ofhydrogenation, the activity and enantioselectivity of the repeatedly usedcatalyst remain constant at high values even without adding fresh modifierat the beginning of each new run. These observations indicate that the lossof modifier takes place only at high conversion, i.e., the presence of excess ofsubstrate 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: 5mlof toluene, [DHCD]=0.1mmol/l, fresh DHCD was added each time to the reaction solution.B: 5ml of toluene, [DHCD]=0.01mmol/l, no DHCD was added for reuse of catalyst.(Reproduced from ref. 268 with permission)

Catalysis, 2010, 22, 144–278 | 177

Different reuse methods and influence of the solvent were investigated byBartok et al.268 In the course of enantioselective hydrogenation of EtPy intoluene, ee increased with catalyst reuse. The increase was in the range of10–20%, however this increase was not observed in acetic acid. The authorsconsidered that the phenomenon of ‘‘increase in ee on reuse’’ is an intrinsicfeature of the catalyst system used, i.e. new chiral centers making higher eepossible are formed. It was suggested that during reuse due to the intensiveinteraction of a solid surface and the chiral modifier reconstruction ofthe Pt surface takes place. This is in a good agreement with recent find-ings306 that during reuse due to the intensive interaction of a solid surfaceand the chiral modifier reconstruction of the Pt surface takes place. Theauthors believe that ‘‘the surface atoms of the catalyst are continuouslyreorganized during the reaction as a result of adsorption/chemisorptionsteps. In this respect it was supposed that the presence of trace amounts ofoxygen might also play an important role in the reaction studied. The lackof increase in ee on reuse in acetic acid may indicate that in this solvent thereaction mechanism is different. The authors most important conclusionswith respect to the reuse of catalyst are as follows: ‘‘(i) Pt/Al2O3 catalystswith a Pt-dispersion of 0.2–0.3 and a mean Pt particle size of 3–5 nm are thebest; (ii) prior to use, the catalyst should be prereduced at 673K for 1–1.5 hin hydrogen flow; (iii) it is necessary to add fresh modifier for each reuse ofthe catalyst’’.268

5. Specificity of Orito’s reaction

As it has already been discussed that enantioselective hydrogenation ofactivated ketones has several specificities. The main specificities are as fol-lows: (i) side reactions, (ii) catalyst deactivation, (iii) solvent effect, (iv)substrate specificity, (v) rate acceleration (enhancement), (vi) enantioselec-tivity–conversion (time) dependencies, (vii) non-linear phenomenon, and(viii) inversion of enantioselectivity. In order to understand all peculiaritiesof these unique reactions the specificities have to be discussed separately.

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

178 | Catalysis, 2010, 22, 144–278

5.1 Side reactions

The intrinsic reactivity of activated ketones is high. In this respect differenttype 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 enantioselec-tivity. The occurrence of side reactions depends on the mode of introductionof the reaction components. The by-product formation can take place bothin racemic and enantioselective hydrogenation of activated ketones. In mostof the side reactions the substrates are involved. With respect to the modifierthe hydrogenation of the quinoline ring has to be mentioned.271

The first thorough information related to the side reactions and by-products formation was given in ref. 307 The formation of semi-ketals cantake place between the substrate and alcoholic solvent,279 the substrate andthe reaction product,233 the substrate and CD.109 It has been demonstratedin ref. 279 that CD catalyses the formation of semi-ketals. The extent ofsemi-ketal formation increases in the following order: t-BuOHoi-PrOHoEtOHoMeOH. Semi-ketals can be hydrogenolized to the correspondingalcohols in a racemic reaction, i.e. this side reaction strongly decreases tooverall ee of the enantioselective hydrogenation.

Oligomerization and polymerization reactions have been discussed bydifferent authors.109,308,309 In these reactions both Pt and alumina sites canbe involved. A dimer of the substrate is formed in the condensation reactionof the enol and keto forms of EtPy.233 The keto-enol transformation ofMePy has been recently investigated by using thermal programmed de-sorption (TPD), STM and reflectance adsorbance infrared spectroscopy(RAIRS).310 It was shown that MePy undergoes CH bond scission at roomtemperature on clean Pt(111) leading to surface mediated enol formationand assembly into H-bonded superstructure. The latter was severely in-hibited by addition of hydrogen (10�6 Torr). STM data show no evidencefor 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-hydroxylketone, which is subsequently dehydrated to generate CQC containingspecies (see Scheme 1). The formed condensation product can be involved incyclization reactions as shown in Scheme 2. These cyclic products wereconsidered as one of the key compounds poisoning the catalyst during thehydrogenation of EtPy.81 In ref. 311 side reactions with the involvement ofPt sites were also investigated. Scheme 3 shows these reactions and the roleof hydrogen in their suppression.

The fact that the aldol condensation of EtPy on Al2O3 can be suppressedby adsorbed acetic acid may be interpreted as that the acetic acid adsorbsand blocks some basic sites on alumina.309 An additional set of side re-actions was discussed in ref. 312. It was suggested that these reactions mighttake place over the Pt sites resulting in strong poisoning effect.312

Several adducts originating from the base-catalysed EtPy conden-sation 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

ATR-FTIR method. It has been shown that both hydrogen and CD sup-presses the decomposition of EtPy taking place on Pt surface.91 The sta-bilization of MePy against decomposition has been suppressed bybenzene315 and 1-(1-naphthyl)ethylamine.316

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

O

O

OH

O

O

OO

O

OH

OH

O

O

OO

OO

O

O

O

OHO

O

-EtOH

2a 2b

1a 1b

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

180 | Catalysis, 2010, 22, 144–278

In a recent study it was shown that although adsorbed CD suppresses thedecomposition of EtPy it cannot suppress condensation and hydrolysis ofEtPy on g-alumina. Coexistence of CD and hydrogen is needed to suppressall side reaction of EtPy over Pt/g-Al2O3.

311

Recently it has been suggested that the polymerization of EtPy takesplace preferentially at steps sites of Pt and CD or other tertiary aminesinhibits propagation of polymerization over Pt sites.233

It is known that most of the products of the above side reactions areconsidered as catalyst poison, consequently their presence significantly alterthe intrinsic kinetic patterns of these reactions. All these facts strongly in-dicate that experimental conditions have to be strictly standardized in orderto minimize the effect of by-products formed. The role of by-products to therate acceleration phenomena will be discussed in Section 5.5.1.

5.2 Catalyst deactivation

In the enantioselective hydrogenation of activated ketones catalyst de-activation takes place both in batch and continuous-flow reactors. Thedeactivation can be attributed to both of chemical and physical processes.

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

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 isinevitable, but it can be retarded or prevented and some of its consequencescan be avoided with clever process design.231 One of the main reasons forcatalyst deactivation can be ascribed to the formation of by-products asdescribed 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-flowreactors provides useful information with respect to catalyst deactivation.

It has been shown that that irreversible deactivation of Pt catalysts bycondensation and oligomerization polymerization can be related tohydrogen starvation.312 In a recent study311 it was found that only thecoexistence of CD and H2 could thoroughly inhibit the side reactions ofEtPy on Pt/g-Al2O3.

The decrease of the catalytic activity due to the use of unpurified sub-strates has been discussed in different studies.66,69,82,317 Consequently,purification of the reactant seems to be a necessary prerequisite to avoidcatalyst deactivation.318 With respect to catalyst deactivation the disputerelated to the origin of rate acceleration has to be mentioned.234,295 Furtherdiscussion of this dispute will be given in Section 5.5.1.

Catalyst deactivation has been observed in kinetic experiments performedin batch reactor.233,289 Deactivation was also evidenced in continuous-flowregime using various experimental designs.291 It was also observed onheating the platinum catalyst in either hydrogen or helium at 350 1C for twohours and then using it without exposure to oxygen.229

General principles of catalyst deactivation both in batch and continuousflow reactors were given in ref. 319. In this work enantioselective hydro-genation of 1-phenyl-1,2-propanedione was studied. An elegant way topromote catalyst durability, activity and selectivity is to apply on-lineacoustic irradiation during the course of reaction.320 Ultrasound can retardcatalyst deactivation by (catalyst) surface cleaning and exposing fresh,highly active surface as well as by the reduction of diffusion length in thecatalyst pores by alteration of the surface of catalyst. Furthermore, stronglyabsorbed organic impurities that block active sites can also be removed bysonification. In ref. 231 the deactivation was studied both under con-ventional and microwave heating using EtPy as a substrate. No catalystdeactivation was observed in three consecutive experiments with re-usedcatalyst. Previously, it has been reported107,321 that during continuoushydrogenation of EtPy and 1-phenyl-1,2-propanedione; a notable catalystdeactivation takes place.

Based on these results it can be concluded that in the enantioselectivehydrogenation of activated ketones deactivation is an inevitable phenom-enon. 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 aminemodifier and the alumina support for aldol reaction;295 (ii) decreasing themodifier/substrate ratio to reduce the rate of side reactions in solution;(iii) working at high surface hydrogen concentrations, that is, at highhydrogen 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 ofmodifier at low surface hydrogen concentrations, and (v) minimizing thecontact 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 hy-drogenated over Pt-cinchona catalyst system. This issue has been discussedin various earlier reviews.33,46,52,195,322 Characteristic feature of these sub-strates is the presence of an activating group with strong electron-withdrawing properties (ester, carbonyl, acetal, amido and a,a,a-trifluorogroup) in the a-position to the prochiral keto group.76 As most of thesubstrates applied so far were extensively discussed in earlier reviews in thissection we shall give only a brief summary.

The first substrates successfully hydrogenated asymmetrically were linearand 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-ketoa-mides.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-trifluoro-ketones236 and substituted aromatic a,a,a-trifluoroketones,53,333,337 trifluorosubstituted a, b-ketones,44,338 b-ketoesters.339 Pt-cinchona system has alsobeen 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 eevalues in the presence of optimum modifier are shown in Table 5.50 Most ofthe substrates, with the exception of the amido derivatives resulted in eevalues above 80%.

Additional results related to the influence of substituents in a-keto esterswere published in refs. 338,341. In these studies both R1 and R2 substituents(see Scheme 4) were systematically altered.

The Bartok’s group found that in AcOH when R1 was methyl, iso-propyl,terc.-butyl, phenyl and phenylethyl or the R2 group was methyl, ethyl andisopropyl the sense of the enantio-differentiation was not altered. The in-crease of the size of R1 and R2 resulted in slight decrease in the reaction rateand ee values.

Additional results presented by Baiker’s group are summarized inTable 6.338 In this series a of experiments the variation of the bulkiness bothat the keto and ester sides in nine different a-ketoesters was investigatedupon using CD and QN, as chiral modifiers. In toluene in the presence ofCD good to high ee values (eemax=94%) were achieved. Consequently, inthe presence of CD the enantio- differentiation is controlled by the estergroup, notwithstanding of the steric bulkiness or electronic structure of the

O

O

O

O

O

O

O

NH

O

O

R CF3

O

O

O

O

Fig. 14 Structure of first substrates successfully hydrogenated asymmetrically.

Catalysis, 2010, 22, 144–278 | 183

Table

5Hydrogenationofvariousactivatedketones

over

Pt/Al 2O

3-cinchonacatalysts.(R

eproducedfrom

ref.50withpermission)

No

Substrates

Modifier

Solvent

Substrate-M

odi-

fier

ratioa

Substrate-Ptratiob

ee(%

)Ref.

1O

O

O

QD

AcO

H1540

1640

98c

344

2Ethylbenzoylform

ate

CD

AcO

HþToluene

868

300

98

56

3

OH

O

O

MeO

CD

EtO

H/H

2O

350

440

82

345

4

O

O

OCD

Toluene

296000

1040

91d

323

5O

O

NHCD

AcO

H300

160

58

334

6butanedione

CD

Toluene

40

2100

63

326

7O

O

CD

EtO

Ac

143

66

94e

346

184 | Catalysis, 2010, 22, 144–278

8

N

O

OCD

Toluene

320

210

91

335

9

OE

t

O

O

ODHCD

Toluene

720

710

86

347

10

O

OR

OR

MeO

CD

AcO

H1050

1320

97

330

CD

AcO

H130

180

97

328

11

CF 3

O

O

OMeO

CD

THF-TFA

290

180

96f

50

12

O

CF 3

CD

Toluene-TFA

290

180

92

332

asubstrate/m

odifier

molarratio.bsubstrate/Ptmolarratio.creaction

mixture

(withoutsubstrate)ultrasonic

treatm

ent.

d�81C.ekinetic

resolution.f�201C,CD:

cinchonidine,

DHCD:10,11-dihydrocinchonidine,

MeO

HCD:methoxy-H

CD,MeO

CD:methoxy-C

D,THF:tetrahydrofurane,

TFA:trifluoroaceticacid.

Catalysis, 2010, 22, 144–278 | 185

alkyl and functionalized aryl group on the other side of the keto group.Other studies reveal also that ester, carboxyl, amido, carbonyl, acetal, andtrifluoromethyl functions have similar directing effects. Results presented inTable 6 show that structurally more demanding substrates show significantdifferences between toluene and acetic acid. However, none of the mech-anistic models developed for the enantioselective hydrogenation of acti-vated ketones over Pt-cinchona catalyst system can explain the followinganomalies found in Table 6: (i) higher ee values in toluene in the presence ofCD for substrates (2), (4), (5) and (9); (ii) the strong increase of ee in aceticacid in the presence of QN for substrate (1); (iii) almost complete loss of eein AcOH in the presence of QN for substrates (3) and (9).

In the presence of CD the greatest drop both in the conversion and eevalues was measured for the hydrogenation of 9 (see Table 6) when chan-ging 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 avery plausible explanation. We consider this observation as an artifact, as inour 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 theenantioselectivities in toluene and acetic acid are similar.53 This statement isnot really correct as in various other studies73 it has been shown that inacetic acid both the rate and ee values are higher than in toluene. It hasalready been shown that the addition of a small amount of acetic acid eitherto toluene or ethanol has a very pronounced effect.84

Derivatives of trifluoroacetophenone with different substituents at thearomatic ring (CF3, N(Me)2 and Me) were hydrogenated over Pt/Al2O3 inthe presence of CD, CD�HCl and 9–O-methyl-CD.332 It was shown thatelectron-withdrawing substituents increased and electron-releasing one de-creased the rate and enantioselectivity in these reactions, although stericeffects (with m- or p-substituents) were also substantial. Cinchona alkaloidswere also used in the asymmetric hydrogenation of non-activated ketones.In this case the enantioselectivity is rather moderate as it was emphasized ina recent review.195

Upon using various nonactivated trifluoromethyl ketones, such as me-thyl-, adamantyl, and terc-butyl235 in the presence of CD low ee values wereobtained (eemax=44% for adamantyl derivatives). Positive effect of TFAwas also demonstrated, while the use of AcOH as a solvent resulted in lowyields and low ee values. In propanol inversion of the ee was observed.These results confirmed again that in the hydrogenation of non-activatedketones high ee values couldn’t be expected. Unfortunately, the initial rateswere not determined in this study. The authors claimed that their result‘‘indicates that enantioselectivity is guided by the trifluoromethyl

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

186 | Catalysis, 2010, 22, 144–278

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

5.4 Solvent effect

In enantioselective hydrogenations of activated ketones both the rates andthe enantioselectivities are greatly influenced by the type of solvents used.

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

R1O

R2

O

O

CD

ee(%)

[conv.]

QN

ee(%)

[conv.]

CD

ee(%)

[conv.]

QN

ee(%)

[conv.]

1 –CH3 –CH2CH3 80 (R) 21 (R) 88 (R) 92 (R)

[100] [100] [100] [100]

2 H3C

H3C CH3

–CH2CH3 56 (R) 12 (R) 27 (R) 2 (R)

[91] [61] [99] [99]

3 –CH2CH3 86 (R) 79 (R) 86 (R) 4 (R)

[100] [100] [100] [100]

4 H3C

H3C CH3

95 (R) 89 (R) 70 (R) 24 (R)

[94] [43] [73] [64]

5 –CH2CH3 92 (R) 75 (R) 80 (R) 31 (R)

[100] [197] [100] [100]

6

F

F –CH2CH3 87 (R) 76 (R) 86 (R) 76 (R)

[100] [99] [100] [100]

7 F3C

CF3

–CH2CH3 66 (R) 47 (R) 72 (R) 72 (R)

[94] [90] [98] [98]

8 O

O

–CH2CH3 94 (R) 84 (R) 94 (R) 84 (R)

[95] [100] [94] [91]

9 –CH2CH3 86 (R) 60 (R) 46 (R) 0 (R)

[100] [32] [31] [7]

Catalysis, 2010, 22, 144–278 | 187

Both protonic and aprotonic solvents have been applied. The solvent caninfluence the enantioselective hydrogenation in different ways; it changesthe solubility of the hydrogen99 and the substrate, the mass transportproperties of the reaction mixture, the adsorption behavior of substratesand modifier on the Pt active sites.

Solvents have also a great influence on the conformation of alkaloidused.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 optimumsolvent 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; howeveras it was shown latter the alcohols react with a-ketoester with the formationof semi-ketals (see Section 5.1). This effect was also discussed in a recentstudy related to the enantioselective hydrogenation of ethyl-4,4,4-tri-fluoroacetoacetate348 in ethanol and propanol.

All these results indicate that the use of alcohols in kinetic or mechanisticinvestigations65,247,348,349 should be avoided as it was emphasized in ref. 53.Upon using O-alkylated CD (R-O-CD) derivatives the use of AcOH orTFA is not recommended as in their presence hydrolysis of R-OCD cantake place. AcOH was not a proper solvent for KPL.77

In one of the first studies the reaction rates and enantioselectivities werecompared in ethanol, toluene and acetic acid.73 These results un-ambiguously show the advantage of using AcOH. In another studies it wasshown that upon hydrogenating EtPy AcOH is the best solvent as thehighest ee values (ee=98%)55 and highest reaction rates were obtained inthis solvent.

The influence of the two most commonly solvents, i.e., toluene and aceticacid, on the rates and ee in enantioselective hydrogenation of EtPy areshown in Table 7. These data clearly show the superior influence of AcOHon the ee values. Not only higher ee values were obtained in AcOH, but alsothe amount of modifier required to get high ee values is one order less inAcOH than in toluene. However, it is interesting to note that contrary toearlier observations at atmospheric hydrogen pressure the rate of hydro-genation 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)

Modifier,

Mmol/l

Hydrogenation

in AcOH at

100 bar350

Hydrogenation

in AcOH at

1 bar267

Hydrogenation

in toluene at

100 bar350

Hydrogenation

in toluene at

1 bar267

rate,

mmol/sec ee, %

rate,

mmol/sec ee, %

rate,

mmol/sec ee, %

rate,

mmol/sec ee, %

0.001 35 83 16 69 8 35 35 16

0.01 80 92 55 91 20 75 57 64

0.1 105 94 85 92 48 82 96 78

1 135 94 63 92 25 83 71 77

188 | Catalysis, 2010, 22, 144–278

mentioning that in racemic hydrogenation of EtPy higher rates weremeasure in toluene than in AcOH.342

Other organic acids were also used to improve the enantioselectivity in Pt-cinchona system. Fig. 15 shows the influence of added trifluoroacetic acid(TFA) on the enantioselectivity in the hydrogenation of KPL in the pres-ence of synthetic modifier (R,R)-PNEA. The results showed that excessTFA was needed to get maximum ee values. It was suggested that part ofthe excess TFA be required to neutralize basic sites of the Al2O3 support.

Excellent correlation was found between the dielectric constant of thesolvents used and both the enantiomeric excess and the population ofconformer Open(3) as calculated by density functional theory in combin-ation with a reaction field model (POpen(3)).

88 This dependence is shown inFig. 16. Earlier results indicated that both reaction rates and ee values de-creased with the polarity of the solvent. Figs. 17A–C show the dependenceof ee on the empirical solvent parameter EN

T in three different systems.In the hydrogenation of EtPy (see Fig. 17A) good ee values are obtained inmoderately apolar solvents, in which the reactant and modifier dissolve.46

Interestingly, primary alcohols are also good solvent, although they reactrapidly with the substrate with the formation of corresponding semi-ketals.The highest ee that time was 95% obtained in acetic acid, while the lowestone in water. The former result was attributed to the protonation of thequinuclidine nitrogen of CD by AcOH and suggesting the alteration of thereaction mechanism in AcOH,283 while the low activity in water can beattributed to the side reaction between EtPy and water, i.e. to the formationof corresponding vicinal diol, and the racemic hydrogenolysis of the diolformed. The latter reaction is responsible for the loss of ee. Fig. 17A showsthat the solvent influence is notable, although the slope in this figure isrelatively moderate. Contrary to that in Rh/Al2O3-b-ICN system used in thehydrogenation of KPL the above slope is higher indicating that the solventhas more pronounced influence on the enantioselectivity (see Fig. 17B).340

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

Catalysis, 2010, 22, 144–278 | 189

The third system behaves in a completely different way. The correlationbetween ee and EN

T is not really good, however it is more interesting that thecharacter of dependence is opposite compared to the systems given inFigs. 17A and B. Nevertheless, the striking effect of solvent properties onthe ee values is obvious. The highest ee to (S)-3,5-di(trifluoromethyl)phenylethanol was obtained in the weakly polar solvent toluene and ethylacetate. The ee decreased in polar solvents. In dimethylformamide, iso-propanol, and ethanol the ee inverted from the (S) to the (R) enantiomer.This behavior indicates that in case of substituted acetophenones thereaction mechanism is strongly altered.

There is another unusual behaviour of this type of substrates. In theenantioselective hydrogenation of 3,5-bis(trifluoromethyl)acetophenone theaddition of trifluoroacetic acid (TFA) resulted in strong decrease in thereaction rate at TFA/CD=around 5 and full inversion of ee at TFA/CDW50.340

Among the solvents AcOH and its triflourinated derivative (TFA) hastheir own peculiarities. The difference in the rates in AcOH and othersolvents is well documented in one of the earlier results, where the additionof small amount of acetic acid strongly increased the overall performance ofthe 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 inalcoholic solvents on the reaction rate and ee. The rate decreases in thefollowing order: n-butanolWethanolWmethanol, i.e. it follows the reactivitytrend of alcohols to form semi-ketals: n-butanoloethanolomethanol.

Very pronounced solvent effect was observed in the hydrogenation of1,1,1-trifluoro-5,5-dimethyl-2,4-hexanedione using Pt/Al2O3 in the presence

Fig. 16 Combined plot of the ee values obtained in the hydrogenation of KPL over Pt/Al2O3-CD system (left axis) and the population of conformer Open(3) as calculated by DFT incombination with a reaction field model (P Open(3), right axis) v.s. the dielectric constant of thesolvent (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)

190 | Catalysis, 2010, 22, 144–278

of synthetic modifier, pantoyl-naphthylethylamine.210 The results show adramatic influence of solvents on the ee values. No correlation can beobtained between ee and relative permittivity (er,) or empirical solventparameter (EN

T). Surprisingly high ee values were obtained only in halogen

Fig. 17 Effect of solvent on enantiodifferentiation of in different enantioselective hydrogen-ation reactions. Dependence of the ee value on empirical parameter EN

T (reproduced fromrefs. 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/Al2O3-b-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 Methanol 0.022 0.019 62.2

2 ethanol 0.057 0.034 72.0

3 n-butanol 0.099 k2Wk1 75.0

4 toluene (EtPy)a 0.057 k2Wk1 86.3

5 MCH (EtPy)a 0.106 0.106 78.1

6 toluene (AcOEt)a 0.063 k2Wk1 84.0

7 MCH (AcOEt)a 0.069 0.060 75.3

8 ethanolþAcOHb 0.074 0.048 91.4

9 tolueneþAcOHb 0.120 0.120 93.1

Reaction conditions: T=23 1C, P=50 bar, [EtPy]0=1.0M, [CD]=8.4� 10� 4M, CD in-jection. a Ethyl pyruvate or ethyl acetate (1.5 cm3) is added to dissolve cinchonidine in thesesolvents (8.5 cm3). b The solvent is mixed with acetic acid; [AcOH]0=5.0M. c Measured at90–100% conversion.

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/aluminawas found in this solvent and steady-state conditions could not be reachedin the continuous-flow reactor at 10 bar.295

In one of the recent publications352 it was mentioned the use ofdichloromethane should be avoided in hydrogenation reactions. As it wasemphasized ‘‘dehalogenation of this solvent on Pt, particularly at highhydrogen pressure, affords HCl, which induces a new set of acid-catalyzedside reactions. It is a very correct remark what was addressed against the useof CH2Cl2 at high pressure by an English group.233

However, we consider also that the use of CH2Cl2 as a solvent by Baiker’sgroup in ATR-FTIR spectroscopic studies314 can also be criticized. TheseATR-FTIR studies are considered as one of the crucial proves for theformation of protonated CD in the absence of AcOH. However, as it wasmentioned in ref. 352 HCl can be formed from CH2Cl2. Consequently, theuse of CH2Cl2 would result in the protonation of CD in the absence ofAcOH and hydrogen.

5.5 Kinetic aspects

5.5.1 Rate acceleration. The rate enhancement of enantioselective hy-drogenation of activated ketones has been observed by various researchgroups. The first results were published by the Ciba Group.57,58,63 Thephenomenon was also observed by others.65,67,289 In all of these studiesusing EtPy as a substrate the common observation was that the modifiedreaction 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 wasproposed that ‘‘a reaction is considered ligand accelerated if there is a slowerunmodified (unselective) cycle and a faster modified (selective) cycle’’.58

This term has been used for many years, although its chemical meaning isquite doubtful or even misleading.

Most of the authors use the term rate acceleration (RA) or rate en-hancement (RE). Not only cinchona alkaloids, but also other tertiaryamines, such as quinuclidine, triethylamine, etc. can induce RA63. Thisbehaviour was evidenced in various solvents.73,354 The addition of smallamount of acetic acid into ethanol or toluene resulted in even more pro-nounced RA84. This behaviour is very characteristic for a-keto esters(Etpy, MePy, KPL77) and was observed not only over Pt, but other metalssuch as Rh252,254 and Ir.355 In ref. 341 it was found that in a-ketoester theincrease of the size of R1 and R2 groups resulted in slight alteration in theextent of RA.

With respect to kinetics studies the reproducibility problems related todifferent impurities has to be mentioned.73,84 Different batches of ethylpyruvate can give completely different kinetic results.84,356 The impuritiesalter both the initial rates and the enantioselectivity. When unpurified EtPyis used in this case the rate of racemic hydrogenation is extremely low.233

For this reason, kinetic results using the given substrate without any purifi-cation65,66,227,232 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 differenttertiary amines increases the rate significantly.

The other problem is that initial rates or related kinetic informationare only seldom published. Most authors prefer to provide activity data inthe form of conversion values measured after a given reaction time. It isespecially notable in recent publications.235,351,357 Consequently, less andless information is available with respect to the RA phenomena. It hasto be mentioned that RA has been observed only in liquid phasehydrogenations; in gas phase hydrogenation of MePy no RA wasmeasured.288

Direct measurement of the reaction rates by using reaction calorimetry ina transient experiment provided an unambiguous evidence for the RA100.The results show that upon injection of CD the rate increase is instant-aneous and the rate increase is in the range of 5–12 depending on the type ofcatalyst used. Similar pronounced RA was observed in other studies over5% Pt/Al2O3 catalyst using EtPy upon injecting cinchona type modifiers ortertiary amines at higher pressures.289,307

With respect to the RA one important question can be raised: is anydirect coupling between the reaction rate and the enantioselectivity. Thiscoupling was first described by Blaser et al. in their ‘‘ligand accelerationmodel’’.58 Although the above coupling was clearly presented recently moreand more evidences have been accumulated that this coupling is not aprerequisite to obtain high enantioselectivities.

The first evidences against the direct coupling were obtained in ourstudies.289,358 In ref. 358 it was shown that the modification of Pt/Al2O3 bySn(C2H5)x moieties strongly alter the reaction rates, but their effect on the eeis negligible. Based on the analysis of the form of conversion-selectivitydependencies in ref. 289 it was stated that, ‘‘at low concentrations of sub-strate and modifier, contrary to instantaneous rate acceleration, the maximumee values are obtained only after a certain time delay. The increase of the rateof enantioselective hydrogenation with respect to the racemic one was welldocumented 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 thatdo not show any RA or even the rate of enantioselective reaction is slowerthan the rate of racemic one. In most of these substrates the prochiral ketogroup is not activated. These substrates are as follows: acetophenone,359

3,5-bis(trifluoromethyl)acetophone,340 alkylsubstituted trifluoromethylketones.359,360

Classification of substrates according to their extent of RA and ED wasgiven recently (see Fig. 1).72 In addition, the following experimental con-ditions 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 trifluoromethylderivatives;363 (v) hydrogenation of EtPy in the presence of CD at very lowsubstrate concentrations.234

The appearance and the disappearance of RA acceleration are welldocumented in a series of experiments using trifluoro acetophenone andcyclohexyl analog. In the case of the former substrate pronounced rate

Catalysis, 2010, 22, 144–278 | 193

acceleration was observed as shown in Fig. 19A. Contrary to that uponusing trifluoro cyclohexyl ketone in the presence of modifier the rate ofreaction decreased as presented in Fig. 19B.

Further insight on the origin of RA was obtained upon comparing thehydrogenation of acetophenone and trifluoromethyl derivatives of

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

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

194 | Catalysis, 2010, 22, 144–278

acetophenone over Pt/Al2O3-CD system.332,362 In these studies both the rateand the ee values strongly depended on the electronic and steric effect of thesubstituents and on the hydrogen-bonding interactions between the qui-nuclidine N atom of the alkaloid and the carbonyl group of the substrate.

In the hydrogenation of acetophenone and trifluoromethylacetophenonederivatives on CD-modified Pt/Al2O3, the conversion rates and enantio-selectivities varied strongly with the nature of the aromatic substitu-ents.332,362 Different reactivities were attributed to the electronic (and steric)effect of the substituents and to hydrogen-bonding interactions between thequinuclidine N atom of the alkaloid and the carbonyl group of the sub-strate. A linear correlation has been found between the logarithm of thereaction rate and the highest occupied molecular orbital and lowest un-occupied molecular orbital stabilization of the carbonyl compounds(DEorb), relative to the reference compound.359

The by-products are high-molecular weight compounds and are con-sidered as strong catalyst poisons reducing the number of available Pt sitesresulting in substantial decrease in the reaction rates. Consequently, inthis case the RA is masked by a catalyst poisoning effect. Recently theRA phenomenon has been questioned by two research groups.233,234,364

In ref. 232 it was concluded that ‘‘rate enhancement is now attributed toreaction occurring at a normal rate at an enhanced number of sites, not (aspreviously proposed) to a reaction occurring at an enhanced rate at aconstant number of sites’’.

The final conclusion was that the ‘‘rate enhancement in the presence of analkaloid modifier is attributed to the inhibition of the pyruvate ester poly-merization at the Pt surface’’. In another recent study365 it was emphasizedthat ‘‘the reaction rate was lower in all chirally modified reactions ascompared to the racemic reaction in the absence of modifier’’.

We believe that in references cited above experimental conditions werenot properly chosen as their findings strongly contradict to results observedearlier by several groups. In a recent paper352 the use of ‘‘reactive’’ solventin ref. 232, such as dichloromethane, was strongly criticized.

Recently, with respect to the RA phenomena an open dispute has beenemerged in Journal of Catalysis.295,352,366 In a recent study continuous–flowexperiments were performed providing clear evidence that the rate accel-eration exists and it was concluded that ‘‘it is not the suppression of catalystdeactivation by addition of chiral modifier, because under appropriateconditions catalyst deactivation is negligible in pyruvate hydrogenation’’.295

This statement was strongly opposed in ref. 366. Those who favor the roleof deactivation defended their view referring to their earlier results shown inFig. 20.234 This figure shows that the rate acceleration appears only at highconcentration of substrate, while at low concentration the rate of enantio-selective hydrogenation is lower than that of the racemic one. In this respectit has to be mentioned that the determination of reaction rate at low sub-strate concentration is very plausible. We consider that the minor differ-ences shown in Fig. 20 cannot be considered as a real prove for the lackof RA.

We have to emphasize again that the decrease of the reaction rate inenantioselective hydrogenation of EtPy at low substrate concentration366

Catalysis, 2010, 22, 144–278 | 195

can be related to the decreased concentration of substrate-modifier com-plexes formed under condition of catalytic hydrogenation. In this respect itis irrespective where the above complex has been formed in the liquid phaseor at the Pt surface. The decrease of the concentration of intermediatecomplex can result in pronounced rate decrease; similar to the kinetic pat-terns observed in enzymatic kinetics.

In our recent study transient experiments with injection of CD duringracemic hydrogenation of EtPy were investigated using purified substratesand a ‘‘distillation residue’’. The ‘‘distillation residue’’ contained 20% ofcompound 1a (see Scheme 2 in Section 5). Fig. 21A shows that the increasein the time delay between the start of racemic hydrogenation and the in-jection of CD has no influence on the rate of enantioselective

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

0.0

0.2

0.4

0.6

0.8

1.0A

0 50 100 150

time, min

conv

ersi

on

0.0

0.2

0.4

0.6

0.8

1.0

0 50 100 150

time, min

conv

ersi

on

Fig. 21 Influence of the time delay in CD injection during raceme hydrogenation. Kineticcurves of EtPy hydrogenation upon using purified substrate; [CD]=5� 10� 5M, T=20 1C,PH2=50 bar, catalyst=5% Pt/Al2O3 (E 4759), 0.125 g; 7 – CD injection at 0min; & – CDinjection at 15min; � – CD injection at 30min; } – CD injection at 90min; (�) – no CD(racemic hydrogenation); A: [EtPy]0=1.0M (purified by distillation prior to the use); B:

[EtPy]0=1.0M (‘‘distillation residue’’ containing dimer 1a in the amount of 20%). (Repro-duced from ref. 367 with permission)

196 | Catalysis, 2010, 22, 144–278

hydrogenation. It means that racemic products formed up to 10% con-version have no measurable influence on the reaction rate, consequently thesize of free Pt surface available for enantioselective hydrogenation is notaltered during racemic hydrogenation. When similar series of experimentswere performed in ethanol in the above conversion range slight decrease inthe initial rates was observed.289

Results given in Fig. 21B clearly show that upon using the ‘‘distillationresidue’’ the rate of racemic hydrogenation decreases, the decrease in rate isaround eleven-fold compare to purified EtPy. The rate decrease in the ra-cemic hydrogenation is due to the strong poisoning effect induced bycompound 1a. The poisoning effect can also be observed in enantioselectivehydrogenation, its extent is around three-fold. The introduction of CDduring racemic hydrogenation of ‘‘distillation residue’’ resulted in also in-stantaneous rate acceleration in all cases (see Fig. 21B).

Contrary to results obtained in the previous series of experiments uponincreasing the time delay from zero to 90 minutes the rate of enantioselec-tive hydrogenation decreases. All these results unambiguously show that thestatement given in ref. 232 ‘‘rate enhancement is now attributed to reactionoccurring at a normal rate at an enhanced number of sites, not (as previ-ously proposed) to a reaction occurring at an enhanced rate at a constantnumber of sites’’ cannot be valid. It is hard to suggest that the addition of5� 10� 5M modifier will compete with 0.2M high molecular weigh productand can remove their adsorbed forms instantaneously from the Pt surface.

In an analogous series of experiments shown in Fig. 22, upon usingmethyl-benzoyl formate (MBF) substrate367 similar trend in the concen-tration dependences was obtained as in the case of EtPy (see Fig. 20).However, the rate of the enantioselective hydrogenation was higher thanthat of the racemic one in the whole concentration range.

Results obtained in series of experiments using MBF shows that the RAeffect is maintained in a relatively broad concentration range. It is a goodexample for the appearance of RA for the class of substrate with decreasedability to form by-products. Finaly let us conclude that we completely

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

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

0.000

0.005

0.010

0.015

0.020

0.025

0.030

0 0.2 0.4 0.6

Concentration MBF, M

d[R

+S]/

dt, M

/min

1

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

Catalysis, 2010, 22, 144–278 | 197

disagree with the new views advertised in refs. 233, 234. Both our earl-ier69,289,307 and recent results269 show that upon introduction of CD duringracemic hydrogenation of EtPy the RA are instantaneous. This conclusionhas been supported by recent results obtained in continuous flowreactor.295,366

5.5.2 Enantioselectivity–conversion dependencies. In enantioselectivehydrogenation of EtPy one of the most disputed kinetic pattern is the form ofthe enantioselectivity–conversion (time) dependencies (ECD). In the hydro-genation of EtPy monotonic increase (MI) type dependencies were obtainedat low conversion in various studies83,84,268,289,369 as shown in Fig. 23. TheMI type behaviour is often called as initial transient period (ITP).195

In first publications integrated ee values were calculated from actualconcentration of (R) and (S) products according to the general formula

ee ¼ ð½R� � ½S�Þ=ð½R� þ ½S�Þ:

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 measuredoptical 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 reactionmixture or when the loss of the modifier during the enantioselective hydro-genation is very pronounced. In addition kinetic ee values can also be cal-culated from the corresponding reactions rates:

eekin ¼ ð½rR� � ½rS�Þ=ð½rR� þ ½rS�Þ:

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

198 | Catalysis, 2010, 22, 144–278

This kind of behaviour was also observed upon using other activatedketones, such as PADA,281 MBF281 and KPL.53 In earlier studies this be-haviour was not discovered as no attempt was done to determine the eevalues at low conversion.

Other characteristic feature of ee-conversion (time) dependencies is thedecrease of the ee values at high conversion.108 This decrease was attributedto the loss of alkaloid during the enantioselective hydrogenation. Theaddition of further amount of alkaloid during the hydrogenation experi-ment 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 useof injection method for the introduction of reaction compo-nents. Fig. 23A shows the ee-time dependencies in EtPy hydrogenationin ethylacetate injecting the substrate into the mixture containing thecatalyst and the modifier. When CD was injected into the reactorusing toluene or ethanol as a solvent MI type enantioselectivity-conversiondependencies were also observed provided the concentration of CD wasless than 10� 4M (see Fig. 23B). In both solvents the appearance of MIcharacter was independent whether CD or the substrate was injected. In allcases the increase part was very pronounced in the first 15–25% ofconversion.

Later on similar behaviour was also described by two other groups in thehydrogenation of EtPy in ethanol using the premixing technique.66,82 Therewas a very tough dispute between these two groups as they had completelydifferent view on this new kinetic phenomenon.82,239

Further results clearly indicated281,350,371 that the appearance of MI typeof enantioselectivity – conversion (time) dependencies strongly depends onthe following experimental conditions: (i) concentration of CD, (ii) themode 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 obtainedin toluene upon using premixing and injection techniques, respectively.93

Fig. 24 Enantioselectivity (eecalc)-conversion dependencies during the hydrogenation of EtPyin toluene; & – [CD]0=1� 10� 4M; E – [CD]0=1.2� 10� 5M; T=23 1C, PH2=50 bar,[Etpy]0=1.0M, catalyst: 5wt% 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 premixingthe MI type behaviour disappears completely and there is only a very slightincrease or decrease of enantioselectivity with conversion. Contrary to thatwhen the injection technique was used the MI type behaviour was observedand its character depended strongly on the initial concentration of CD. Thelower the concentration of CD the more pronounced the MI character (seeFig. 24B). The MI character was completely maintained when eekin valueswere used instead of eecalc. It has to be emphasized that when similar ex-periments were performed in ethanol the MI character appeared upon usingboth premixing and injection methods.84 This behaviour was attributed tothe formation of semi-ketal from the substrate and the solvent during theperiod of premixing.

Similar experiments were performed in the hydrogenation of 3,5-bis(tri-fluoromethyl)-acetophenone over Pt/Al2O3 in the presence of CD.306 MItype dependence was observed under general experimental condition, whilealmost constant ee values were obtained after premixing the reaction mix-ture in nitrogen as shown in Fig. 25A.53 Consequently, identical observationwas obtained as in the hydrogenation of EtPy.93 The results clearly indicatethat both surface cleanness and interactions in the liquid phase have theircontribution for the appearance of MI type ee–conversion (time)dependencies.

Different aspects of the appearance of initial transient period were dis-cussed by Bartok upon using various substrates, such as EtPy, PADA,MBF.281 One of the most interesting observations was the dependence ofinitial transient period on the pre-treatment conditions and the purity of thesubstrate as shown in Fig. 25B. After pretreatment in helium ee values are15–20% higher than those without this pre-treatment. This behaviour wasattributed to the of 5 ppm oxygen in helium. It was suggested that theoxygen can alter the Pt surface, what is more favourable for the interactionwith DHCD or CD.

20

40

60

80

100B

0 20 40 60 80 100

Conversion (%)

ee (

%)

98% purity / He98% purity / H299.9% purity / He99.9% purity / H2

H2

He A

Fig. 25 Effect of catalyst pretreatment on ee-time (conversion) dependencies. A: sub-strate=3,5-bis(trifluoromethyl)-acetophenone, catalyst=Pt/Al2O3, chiral modifier=CD, sol-vent=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 stirredunder N2 for 1 h before H2 was introduced (premixing); B: Pretreatment of re-reduced catalystin toluene with H2 or He in the enantioselective hydrogenation of EtPy. (Reproduced fromref. 281 with permission.)

200 | Catalysis, 2010, 22, 144–278

The influence of the reuse of the catalyst on the initial transient periodwas investigated by the Bartok’s group. Typical MI type dependencies areshown in Fig. 26.268 The results showed that the expression of MI characterdepended on the amount of modifier, but it was less pronounced after reuseof the catalyst. However, in repeated use, i.e. after removal of the reactionmixture and addition of fresh toluene, EtPy, and modifier resulted in10–20% increase in the ee values.

On the basis of these results it was concluded that the ‘‘the phenomenonof 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 thisrespect restructuring of the Pt surface was suggested based on analogousanalysis of results given in refs. 55, 90, 277. It was also supposed that oxygenplays a definite role in this process.

A sequential introduction of different substrates was performed in whichthe hydrogenation of MePy was carried out following the initial hydro-genation of EtPy using Pt/Al2O3-cinchona catalyst. In these experiments,the MI type character was obtained in both hydrogenation experiments (seeFig. 27A). The character of ee-conversion dependencies was maintainedafter reversed order of introduction of substrates (see Fig. 27B). The ob-servation that the initial transient effect is still observed with the sequentialhydrogenation of EtPy and MePy indicates that the phenomenon cannot beattributed to impurity effects. Consequently, it is more probable that thereaction-driven equilibrium of the chiral environment play a role in the MIcharacter of ee-conversion dependencies.

In one of the recent studies three different modifiers, such as CD, 9–O-phenyl-CD (PhOCD), 9–O-pyridil-CD (PyrOCD) were investigated in thehydrogenation of EtPy.372 Well-expressed MI type behaviour was obtainedfor all three modifiers. However, despite all the convincing results presented

20

40

60

80

100

0 20 40 60 80 100

Conversion (%)

ee (

%) 12

2r1r

3

3r

[DHCD](mmol/L)

Temp.(°C)

1

1r

2

2r33r

0.01 use

0.01 reuse

0.1 use

0.1 reuse0.01 use0.01 reuse

-10

-10

-10

-10-20-20

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

Catalysis, 2010, 22, 144–278 | 201

in this section the following statement was done: ‘‘We assume that the slowremoval of surface impurities is the major reason for this behavior’’ (i.e. theMI type ee-conversion dependencies). Consequently, there are groups whodo not learn from results obtained by other groups and keeping their oldviews as a dogma.

Finally we can summarize the general views with respect to the origin ofthe initial transient behaviour of ECD: (i) it is related to impurities or otherexperimental 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 homo-geneous asymmetric catalysis have been investigated for many years sincethe pioneering work of H.B. Kagan.373 First the phenomenon was attrib-uted to the diastereomeric association inside or outside the catalyticcycle.374 Later on the approach was extended to the use of mixtures ofdiastereomeric ligands.375

Recently this approach was been extended to the Orito’s reaction. It wassuggested that the nonlinear behavior be due to the deviation from theexpected ideal behaviour assuming that the molar ratios of the modifiers insolution and on the metal surface are identical. Consequently, the nonlinearbehavior of mixtures of two diastereoisomers or two completely differentchiral modifiers has been attributed mainly to their different adsorptionstrength,287 however the contribution of the adsorption geometries on themetallic sites was also emphasized and new term ‘‘non-linear phenomenon’’(NLP) has been introduced.269 Besides it was also suggested269 that modi-fier–modifier interactions may also be involved in the NLP, but no ex-perimental evidence has been found yet. It was concluded that theinvestigation of NLP behavior of mixtures of two modifiers is a powerfultool in heterogeneous catalysis for characterizing the relative adsorptionstrength of modifiers under truly in situ conditions.

However, in this respect the controversy between catalytic and spectro-scopic investigations related to the evaluation of the relative adsorption

Fig. 27 Enantioselectivity-conversion dependencies in sequential hydrogenation experiments.a: EtPy hydrogenation to 100% conversion prior to addition of MePy; b: MePy hydrogenationto 100% conversion prior to addition of EtPy; ’ – EtPy conversion, K – e.e. in (R)-EtLa,

7 – MePy conversion,E – e.e. in (R)-MeLa. T=20 1C, reaction pressure 30 bar H2 for EtPyhydrogenation, 50 bar H2 for MePy introduction. (Reproduced from ref. 371 with permission.)

202 | Catalysis, 2010, 22, 144–278

strength of various chiral modifiers has to be mentioned.376–378 We have toadmit that this controversy can be attributed to the absence of substrate andhydrogen in spectroscopic investigations.

In a recent review based on the above results the following statement wasdone: ‘‘An essential conclusion from this study is that, although strongadsorption is a crucial requirement for an efficient modifier, there is nopositive correlation between the adsorption strength (AS) and the enan-tioselectivity achieved with the modifier alone’’.53

In a recent study it was emphasized that the NLP ‘‘can presumably beinterpreted on the basis of differences in the structure of the substrate-modifier complexes formed and in the adsorption-desorption processes ofthe complexes, thus the NLP is not solely dependent on the adsorption ofcinchona alkaloids, as suggested by earlier experimental data’’.379

The above view is very close to our one. In one of our study361 we turnedback to the original idea given by Kagan, namely to the formation of dia-stereomeric associations between the substrate and different competingchiral entities as modifiers in the liquid phase. Consequently, in our inter-pretation the nonlinear behaviour is due to different enantio-differentiationability of two modifiers acting simultaneously in the liquid phase resultingin different substrate-modifier complexes (associations). Of course theenantio-differentiation ability of two modifiers is further influenced bydifferent factors, such as the adsorption-desorption behaviour, the abun-dance and the reactivity of the formed associates.

It has to be added that kinetically the difference between the two inter-pretations (different adsorption strengths v.s. differences in the structureand stability of substrate-modifier complexes) for the NLP cannot be done.Only careful analysis of the chemical and surface properties can providesome hints inside the origin of these observations.

Different experimental techniques were used to investigate the NLP oftwo alkaloids: (i) variation of the initial ratio of two modifiers measuring theee values at the end of the reaction,287 (ii) applying a fixed initial ratio of twomodifiers and following the ee-conversion dependencies,361 (iii) usingtransient method in a batch reactor, where one of the modifiers is intro-duced at t=0, while the other one after a given time lap,269 (iv) usingcontinuous flow reactors and creating transient conditions by switchingfrom one modifier to another one.301

The deviation from the expected linear correlation was first observed inthe hydrogenation of EtPy in the presence of CD–CN and QN–QD mix-tures.287 At mole fraction of 0.5 of these alkaloids the ee value was higherthan zero indicating that the ED ability of CD and QN is higher than that ofthe CN and QD, respectively. Similar result was also obtained in otherpublications.60,379 In all of these studies the findings were attributed to thedifferences in the adsorption strength of the alkaloids.

The most striking NLP behaviour was observed in the hydrogenationof KPL over Pt/Al2O3 in the presence of CD-PhOCD mixtures (seeFig. 28A).238 It is known that PhOCD gives (S)-pantolactone, whereas CDaffords (R)-pantolactone as major enantiomer. The addition of smallamount of CD (XCDo0.05) to a reaction mixture containing PhOCD re-sulted in drastic change from (S)-pantolactone to (R)-pantolactone as the

Catalysis, 2010, 22, 144–278 | 203

major product. This non-linear behavior is attributed to the much strongeradsorption of CD compared to PhOCD. The weaker strengths of ad-sorption of PhOCD were related to its tilted form of adsorption as shown inFig. 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 andQD is shown in Fig. 28B.378 Based on these results the following order wasestablished 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 adifferent order for the strength of adsorption: QN,QDWCDWCN.376 In amore recent RAIRS experiments the order in the adsorption equilibriumconstants (Kads) the following sequence was established: CNWQDW

CDWQN.377 Probably based on these results in a recent review the fol-lowing statement was done: ‘‘An essential conclusion from this study isthat, although strong adsorption is a crucial requirement for an efficientmodifier, there is no positive correlation between the adsorptionstrength (AS) and the enantioselectivity achieved with the modifieralone’’.53

Upon investigating the behaviour of O-alkylated derivatives of CD it wasnicely demonstrated that the adsorption strength of this type of modifiers onPt decreases in the following order: CDWMeOCDWEtOCDWPhOCDETMSOCD.202

Fig. 28 Non-linear effect in enantioselective hydrogenation reactions over Pt/Al2O3 catalyst.A: substrate=KPL, chiral modifier=CD-PhOCD mixtures; schematic illustration of theadsorption 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 addedat 10–20% conversion of EtPy.380 (Reproduced from refs. 238 and 380 with permission.)

204 | Catalysis, 2010, 22, 144–278

A continuous-flow fixed-bed reactor was applied in the enantioselectivehydrogenation of EtPy on Pt/Al2O3 using the principle of ‘‘chiral switch’’.

301

These time on stream experiment start with the introduction of one of themodifiers and a given moment this modifier is switched to another one. TheQD-CD and CD-QD switch is shown in the Fig. 29A and B.295 The resultsclearly show that the enantio-differentiation ability of CD is stronger thanthat of the QD. All of the above results strongly support the general viewthat NLP and the adsorption strength of the modifiers are coupled. How-ever, there are experimental findings indicating that the origin of NLP hasmore complex basis.

In this respect let us refer to a series of transient experiments performed ina batch reactor. Fig. 28B shows the influence of solvents when CD wasadded to the reaction mixture containing QN. In the opposite situationwhen QN was added to the reaction mixture containing CD only minorchanges in the ee values were observed. The most interesting finding is thatthe ‘‘rate of replacement’’ of QN by CD shows strong solvent dependency.No real explanation was given for this finding, although the possibility forthe involvement of solvent polarity and the formation of an alkaloid–acidion pair has been mentioned. Our view is that these experimental findingsindicate that not only the difference in the adsorption strength controlsNLP. Even more striking results related to NLP were observed when theamount of modifiers was varied in the above experiments. These results areshown in Figs. 30 A–C.269 When CD was added to the reaction mixturecontaining QD the direction of the enantioselectivity was immediatelyaltered. The time period required to reach the maximum Dee showedstrong concentration dependence, what was attributed to the fast hydro-genation of the quinoline ring of QN in the first 30 minutes at low con-centration of modifiers. In the opposite situation, i.e. when QD was addedto CD (see Fig. 30C) the decrease part was also explained by the fasthydrogenation of CD, however no acceptable explanation was given for theincrease part.

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

Catalysis, 2010, 22, 144–278 | 205

The mixing of two different alkaloids was also applied to study theanomalous behaviour of both ICN. The results shown in Fig. 31A and Bindicate that the addition of a-ICN to CD, CN and b-ICN has no influenceon the enantio-differentiation ability of these alkaloids.361 Based on theseresults it was suggested that the origin of enantio-differentiation ability ofa-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 theaddition of a-ICN resulted in less loss of ee at high conversions. It is due tothe suppression of the hydrogenation of the quinoline ring of CN in thepresence of a-ICN. Consequently, a-ICN should be strongly adsorbed onthe Pt surface.

The first attempt to compare the behaviour of two substrates (EtPy andKPL) in NLP under identical conditions using CN and QN was done in arecent study. The investigations were performed in two solvents (tolueneand AcOH).379 Three different methods were applied. Here we show resultsobtained in a batch reactor using conventional and transient experiments.According to results given in Fig. 32A, in the hydrogenation of EtPy thedirection of enantio-selection changes almost linearly with the concen-trations of the two modifiers. On the contrary, in the hydrogenation of KPLthe direction of enantio-selection is affected to a much higher extent by CN

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

206 | Catalysis, 2010, 22, 144–278

Fig. 31 Enantioselectivity-conversion dependencies in the presence of mixtures of a-ICN andflexible alkaloids. T=20 1C, pH2=50 bar, injection method, 500 rpm, reaction time=90min;A: ’ – 1.2� 10� 5MCD; & – 1.2� 10� 5MCDþ 1.2� 10� 5Ma-ICN; K – 1.2� 10� 5Mb-ICD; � – 1.2� 10� 5Ma-ICNþ 1.2� 10� 5Mb-ICN; B: � – 1.2� 10� 5Ma-ICN;7 – 1.2�10� 5MCN; D – 1.2� 10� 5MCNþ 1.2� 10� 5Ma-ICN. (Reproduced from ref. 361 withpermission)

Fig. 32 Comparing the behavior of EtPy (EP) and KPL in NLP. A: QN-CNmodifier mixture,total modifiers concentration: 0.1mM, solvent: toluene/AcOH=9/1; B: QN-CN modifiers;C: CN-QNmodifiers. In B and C concentration of each modifier=0.05mM, first abbreviation-modifier used first, second abbreviation-modifier added afterwards; solvent=toluene (T) andAcOH; 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 adsorptionstrengths are identical, while in the presence of KPL the adsorption strengthof 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 bedrawn: (i) CN desorbs QN more readily in the hydrogenation of KPL thanin the hydrogenation EtPy, (ii) in the case of EtPy, CN cannot fully desorbQN from the surface; (iii) in the hydrogenation of KPL CN can nearly fullydesorb QN. Based on these findings the order of adsorption strength inthese two substrates is different, namely in EtPy CNBQN, while in KPLCNWQN. When CN was used as the first chiral modifier (see Fig. 32C), inthe hydrogenation of EtPy CN cannot be desorbed by QN, while in thehydrogenation of KPL under identical conditions CN acted as if QN wasnot present at all. Consequently, the order of the adsorption strength of thetwo cinchonas is different in these to substrates, in case of EtPy CNBQN,while in case of KPL CNWWQN. These findings are similar as those ob-served in ref. 361, where CD, CN and b-ICN acted as if a-ICN was notpresent at all. Such observation was also made in another earlier study intransient experiments using CN and QN.197

Summing up all investigations related to the elucidation of the origin ofNLP we accept the conclusions given in a recent study that ‘‘the NLP de-pends not only on the chiral modifier but also on the substrate to be hy-drogenated. This observation can presumably be interpreted on the basis ofdifferences in the structure of the substrate-modifier complexes formed andin the adsorption-desorption processes of the complexes, thus the NLP isnot solely dependent on the adsorption of cinchona alkaloids, as suggestedby earlier experimental data.197 The statement is in full agreement with ourview related to the importance of the formation of substrate-modifiercomplexes.

5.5.4 Inversion of enantioselectivity. To find relationship betweenthe configuration of chiral centers of the modifier and the chirality of theproduct was one of the early tasks. It has been generally accepted that theconfiguration of C8 or C8 and C9 atoms of the cinchona alkaloid moleculedetermines the product distribution.57,192 Changes in the sense of enantio-selection were first observed by Augustine et al. in 1993. Upon varying theDHCD/ catalyst ratio in the hydrogenation of EtPy over Pt/Al2O3 catalyst(S)-ethyl lactate formed at low modifier concentrations and (R)-enantiomerat higher modifier levels.67 However, the extent of inversion is within theexperimental error. The other intriguing fact is that the ee values are ex-tremely low. It is unprecedented that in this reaction the ee values are lessthan 20%.

Analogous observation was found in gas phase hydrogenation of EtPyover 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 modifierconcentration. The inversion of ee was found to be dependent on the natureof the substituent at C9.382 The appearance of inversion of enantioselectivitywas 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 concen-tration206 and even changes of the substrate.210 Inversion has been reported

208 | Catalysis, 2010, 22, 144–278

both on Pt56,202 and Rh254,256,351 catalysts. Summary of recent results wasgiven in ref. 383. Certain C9 substituted derivatives of cinchonas such 9–O-alky,202 -aryl192,202,238 and -silyl202,206,238 derivatives of cinchonidine andSharpless-ligands,192,206 furthermore b-isocinchonine56,197,384 (the rigid de-rivative of CN) have resulted in product with the opposite sense of ee thanthe underivatized alkaloid. Fig. 33 shows both the diminished enantio-selectivity and its inversion with increasing bulkiness of the ether function ofthe modifiers.202

Upon using b-isocinchonine-Pt/Al2O3 catalyst system in the hydrogen-ation of EtPy ee decreases continuously and turn to opposite value withdecreasing of pH (see Fig. 34). Investigation of unexpected inversion hasgiven a new possibility for mechanistic studies.

In the hydrogenation of ethyl-4,4,4-trifluoroacetoacetate over O-methyl-cinchonidine-Pt/Al2O3 catalyst system a significant variation of ee value wasobserved with the conversion in the presence of even trace amounts of wateror catalytic amounts of a strong acid.349,385 This issue has been discussed inSection 5.1. The explanation for the inversion of enantioselectivity is notcompletely clear at the molecular level. It is obvious that the inversion of eecan be related to increasing bulkiness of the substituent at C9 and the in-creased rigidity of the alkaloid molecule. However, these factors alone givenot sufficient answer why O-pyridoxy derivative372 of CD does not lead toinversion in spite of the fact that O-phenyl and O-pyridyl moieties havealmost identical van der Waals volumes. Further arguments are necessary toexplain why a-ICN197 and a-isoquinine385 owning similar rigid structure asb-ICN shows no inversion.

To understand the origin of inversion different physical chemical methodshave been applied. HPLC-MS and GC-MS measurements have shown that

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-tri-methylsylil-cinchonidine, PhOCD: 9-O-phenyl-cinchonidine, XylOCD: 9-O-(3,5-dimethylphe-nyl)-cinchonidine, HFXylOCD: 9-O-[3,5-bis(trifluoromethyl) phenyl]-cinchonidine,NaphOCD: 9-O-naphthyl-cinchonidine. (Reproduced from ref. 202 with permission)

Catalysis, 2010, 22, 144–278 | 209

b-isocinchonine modifier keeps its inner ether structure during the enan-tioselective hydrogenation.386 NMR measurements have proved that thehydrogenation of phenyl group in 9–O-phenyl-CD (PhOCD) is not re-sponsible for inversion.202 From ATR-FTIR spectroscopic measurementsand DFT calculations it has been concluded that the shape of the chiralspace formed by the adsorption of PhOCD onto the metal is altered com-pared to that formed by CD238,387 (see Fig. 28). The phenyl group has acomplex interaction with platinum; it can adsorb via its p system influencingthe strength of adsorption of the modifier. However, at the same time it cangenerate steric repulsion in the proximity of the chiral site.388 The presenceof the phenyl group in (PhOCD) can also hurt the efficiency of the shieldingeffect (see Section 8.3).

Chiral pocket389 has been defined as a physical space that is able toaccommodate, via bonding and repulsive interactions, a pro-chiral ad-sorbate, and that is able to discriminate between its enantiomers. It wassuggested that although CD and PhOCD display similar adsorption modes,the different adsorption strengths and the change of the chiral pocket aresufficient to induce the inversion of enantioselectivity.387 The different roleof the bulky ether groups i.e. repulsion by the phenoxy and attraction by the2-pyridoxy group explains the different behaviour of these derivatives.372

Based on results obtained in the hydrogenation of EtPy197 and keto-pantolactone in the presence of b-ICN381 in toluene interactions responsiblefor the inversion were proposed.

The conformational rigidity of both the chiral modifier and the reactantmay inhibit the geometrical fit of the three components (modifier, reactant,and Pt), consequently the formation of the adsorbed intermediate respon-sible for enantio-selection is hindered. Beside the interaction between thenucleophilic N atom of the quinuclidine skeleton and the electrophilicC atom of the keto group of KPL or EtPy, H-bonded interaction verified byMcBreen are also suggested.72 It is proposed that the sense of enantio-selection is controlled by the conformation of the adsorbed reactant–chiral

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

210 | Catalysis, 2010, 22, 144–278

modifier (1:1) complex which can be influenced by the solvent.384 It has beenconcluded that the adsorption mode and conformation of the modifierduring interaction with the substrate play the crucial importance in thechange of the sense of enantio-selection.53

In a recent study the inversion of enantioselectivity was investigated in thehydrogenation of of 1-phenyl-1,2-propanedione (PPD) using differentO-ether derivatives of CD.391 The data confirmed that the origin of inversionis related to depletion of the H-bonding interaction between the modifierOH-group and the carbonyl group of the reactant rather than to a decreasedpopulation of the Open(3) conformation in the solutions of O-ether de-rivatives when compared with the solution behavior of the parent alkaloid.

In another recent study different O-ethers of CD and CN were used inboth the enantioselective hydrogenation of PPD and the kinetic resolutionof the 1-hydroxyketones formed over Pt catalysts390 Characteristic resultsfor PPD are shown in Fig. 35. As emerges from these results all O-ethersshowed inversion of enantioselectivity. Similar trend was also observed inthe kinetic resolution of 1-hydroxyketones. These results are different fromthat obtained in the enantioselective hydrogenation of EtPy, where in-version was observed only in case of large substituents. Another importantfinding is that in the presence of AcOH the above modifiers showed onlyvery 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 variousadditives on the behaviour of Pt/cinchona catalyst. These additives can beconsidered as modifiers either of the platinum or the support.

Fig. 35 Enantioselectivities of PPD hydrogenation in toluene over Pt/Al2O3 catalysts modifiedwith: (K) – CD; (�) CN; (’) – MeOCD; (&) – MeOCN; (7) – PhOCD; (D) – PhOCN; (E) –TMSOCD; (}) – TMSOCN. (Reproduced from ref. 390 with permission.)

Catalysis, 2010, 22, 144–278 | 211

5.6.1 Modification of Pt by tin tetraethyl. In one of our earlier studiesthe Pt sites in a Pt/Al2O3 catalyst were modified by Sn(C2H5)4.

358 Thismodification is a two step anchoring type surface reaction.392

Pt�Ha þ SnðC2H5Þ4��! Pt�SnðC2H5Þð4�xÞ þxC2H6 ðPSCÞ ð1Þ

Pt�SnðC2H5Þð4�xÞ þH2��! Pt�Snþðn� xÞC2H6 ðSBAÞ ð2Þ

In reaction (1) primary surface complexes (PSC) are formed. After tinanchoring the surface of Pt is covered by surface organometallic complexeswith general formula of Pt-SnR3 or Pt-SnR2. These PSCs are stable at roomtemperature. Upon heating in a hydrogen atmosphere (see reaction (2)) theydecompose 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 catalystsare summarized in Table 9. As emerges from this Table the modificationsresulted in both PSC and SBA forms of Pt-Sn/Al2O3 catalysts with differentSnanch/Pts ratios. The activity of modified catalyst was strongly altered bythe amount and the type of surface species. However, over catalysts con-taining PSC (see Exps. 5, 13, 14, 15, 16 in Table 9) the ee values were almostconstant, i.e. they were not affected by modification of Pt (ee=86–89%).

Striking observation was that upon using PSC the hydrogenation re-action was completely blocked at relatively small Snanch/Pts ratios. This wasattributed 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)

Exp. No

Catalyst

Code No

Temperature of

H2 treatment, 1C Sn/Pts, g

Rate of

hydrogenation,

mol (kgcat sec)� 1

Optical

yield, %

1 Pt no 0.000 0.83 64

2 Pt 150 0.000 1.70 82

3 Pt 200 0.000 1.66 87

4 Pt 400 0.000 2.00 88

5 PtSn-1 no 0.025 3.00 89

6 PtSn-1 100 0.025 2.60 84

7 PtSn-1 200 0.025 2.12 80

8 PtSn-1 200 0.025 2.17 86

9 PtSn-1 400 0.025 1.15 85

10 PtSn-2 200 0.036 1.06 72

11 PtSn-2 400 0.036 1.08 81

10 PtSn-3 200 0.056 0.02 72

11 PtSn-3 400 0.056 0.74 81

13 PtSn-4 no 0.030 1.77 86

14 PtSn-4 no 0.030 1.55 86

15 PtSn-5 no 0.008 2.00 89

16 PtSn-5 no 0.008 1.77 88

212 | Catalysis, 2010, 22, 144–278

hydrogen activation by Sn(R)(4� x) moieties. The highest rates in theenantioselective hydrogenation were obtained upon using modified catalystscontaining PSC. This finding was attributed to the suppression of the poi-soning effect induced by byproducts formed.

There is one interesting remark: in a related study it was confirmed that inthis kind of surface modification of Pt at low tin coverage tin prefers kinkand corner sites.395 Based in this old results it can be concluded that theinvolvement of kink and corner sites of Pt in the ED step, as it has beensuggested by different authors,233,369,396 is highly questionable. In additionthe above results can be considered as a first real hint that the reaction ratesand the ee values are not well correlated, i.e. relatively high ee values can beobtained even when the reaction rate is strongly suppressed. Unfortunately,this result was almost forgotten, as its conclusion did not fit into the conceptof ‘‘ligand acceleration model ’’. This model a priori suggests a definite re-lationship between reaction rate and enantioselectivity.

5.6.2 Addition of achiral amines. Based on the use of different experi-mental methods182,189–191,397,398 it has been suggested that cinchona alkal-oids can form dimers. As far as any dimer would decrease the virtualconcentration of CD in the liquid phase attempts were done to use differentachiral tertiary amines (ATAs) with the aim to shift the equilibrium betweenthe dimer and the monomer form of CD as shown in the following scheme:

½CD�2 ! 2½CD� ð4Þ

½CD�2 þATA! ½CD�ATA� þ CD ð5Þ

This concept has been tested in the enantioselective hydrogenation ofEtPy and hexanedione.93,399–401 It is suggested that the modifier in the formof a dimer is a spectator in the asymmetric hydrogenation reaction.

Results in the presence of various ATAs at different experimental con-ditions are summarized in Tables 10 and 11. These results clearly show the

Table 10 Influence of different added achiral tertiary amines on the enantioselective hydro-

genation 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,

k1 (min� 1)

Rate constant,

k2 (min� 1)

Enantioselectivity

(eemax)

No – 0.0352 0.0465 0.750 (0.714)b

TEA 1.2� 10� 5 0.0407 0.0676 0.841 (0.793)b

DABCO 1.2� 10� 5 0.0886 0.1588 0.915b

QND 1.2� 10� 5 0.1289 0.1645 0.898b

QND 1.2� 10� 5 0.1297 0.1757 0.909b

QNDc 1.2� 10� 5 0.0832 0.1346 0.926b

QNDc 6.0� 10� 5 0.1267 n.m.d 0.936b

QNDc 1.2� 10� 4 0.1219 n.m.d 0.946b

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

Catalysis, 2010, 22, 144–278 | 213

strong effect of ATAs in toluene, however no effect has been observed inEtOH and AcOH. In the presence of quinuclidine unprecedented high eevalues were obtained at low CD concentration (1.2� 10� 5). The ee valueequal to 0.946 is close to those obtained in pure AcOH.

Further evidence with respect to the involvement of alkaloids dimers inthe 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 OHgroup 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 dichroismspectroscopy. In table 12 the intensities of the Cotton shift around 237 nmare shown in the presence of various ATAs. This Cotton shift has beenascribed to the dimer form.192 In case of CH3OCD no similar Cotton shift isobserved. The calculated De values well correlated with the ability of ATAsto increase the reaction rate and ee values. These results suggest that theATA added into the solution of CD be involved in new type of solute–soluteinteraction.

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 inEtOH and AcOH; (iii) the ATA effect depends on its concentration; (iv)ATAs containing bulky substituents show more pronounced effect; (v) theOH 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 variousnitrogen containing aromatic and condensed aromatic compounds wasstudied. The aim of these studies was testing of the validity of the

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 ATAs k1, min� 1 k2, min� 1 eemax eeend

1 Noa 0.0045 0.0068 – –

2 No 0.0236 0.0747 0.838 0.819

3 quinuclidinea 0.0077 0.0109 – –

4 quinuclidine 0.0482 0.0997 0.901 0.882

5 Dabco 0.0486 0.1267 0.909 0.905

6 MPD 0.0322 0.0989 0.895 0.872

7 TEA 0.0300 0.0905 0.849 0.843

8 Edcha 0.0220 0.0735 0.850 0.785

9 Edipa 0.0243 0.1280 0.824 0.796

10 3-quinuclidinol 0.0468 0.0867 0.888 0.870

11 Nob 0.0409 0.1268 0.945 0.945

12 quinuclidineb 0.0699 0.1518 0.946 0.946

[EtPy]0=1.0 M, [CD]=1.2� 10� 5M, ATA=6� 10� 5M, 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 10minutes and between 25–60 minutes, respectively. a in the absence of cinchonidine.b solvent=1M acetic acid in toluene.

214 | Catalysis, 2010, 22, 144–278

‘‘surface model.’’ The ‘‘surface model’’ assumes that CD adsorbs by itsaromatic 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 com-petitive adsorption between CD and condensed aromatic compounds can beestablished, the number of ‘‘chirally modified sites’’ should decrease re-sulting in definite loss of enantioselectivity.

Table 13 shows the influence of added quinoline (Q) on the reactionkinetic and ee values. These results unambiguously show that the additionof quinoline increases both the rate and the ee values. The effect of quinolineis very pronounced at low concentration of CD, while upon increasing the

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� 3M)

ATA-CD

molar

ratio solvent De (M� 1cm� 1)

1 no – – ethanol –

2 no – – CH2Cl2 0.94

3 quinuclidine 0.4 1 CH2Cl2 0.91

4 quinuclidine 0.8 2 CH2Cl2 0.74

5 quinuclidine 2.0 5 CH2Cl2 0.70

6 quinuclidine 4.0 10 CH2Cl2 0.61

7 quinuclidine 8.0 20 CH2Cl2 0.49

8 Dabco 2 5 CH2Cl2 0.92

9 Dabco 4 10 CH2Cl2 0.69

10 Dabco 8 20 CH2Cl2 0.39

11 TEA 2 5 CH2Cl2 0.94

12 TEA 4 10 CH2Cl2 0.90

13 TEA 8 20 CH2Cl2 0.79

[CD]=4� 10� 1M, 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� 5M [QN], 10� 5M k1, min� 1 k2, min� 1 eemax eeend

1 0.6 no 0.027 0.015 0.573 0.235

2 0.6 6.0 0.040 0.069 0.832 0.699

3 0.9 no 0.032 0.058 0.719 0.575

4 0.9 6.0 0.045 0.090 0.867 0.813

5a 1.2 no 0.034 0.073 0.830 0.798

6 1.2 1.2 0.054 0.114 0.880 0.847

7 1.2 6.0 0.056 0.118 0.874 0.860

8 1.2 12.0 0.059 0.123 0.880 0.869

9 6.0 no 0.065 0.187 0.894 0.894

10 6.0 6.0 0.060 0.140 0.898 0.898

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

Catalysis, 2010, 22, 144–278 | 215

concentration of CD the effect disappears. Due to the presence of quinolineunusually high ee values were obtained in toluene at [CD]0=1.2� 10� 5M.

Figure 36A–C shows the ee-conversion dependencies at different CDconcentration. Characteristic feature of these dependencies is that at lowCD concentration ee decreases at high conversion (see Fig. 36A). The loss ofee at high conversion is attributed to the loss of CD due to the hydrogen-ation of its quinoline ring. However, this decrease is strongly suppressed bythe addition of quinoline; consequently the results indicate that quinolinereplaces CD from the Pt surface. This replacement reduces the chance of CDto be hydrogenated by its quinoline ring.

Table 14 shows the influence of various nitrogen containing and con-densed aromatic compound on the rate and enantioselectivity. These resultswere obtained on highly dispersed Pt/SiO2 catalysts. Over this catalyst thering hydrogenation of the quinoline ring was relatively fast. It is the reasonthat over this catalyst the ee decreases with conversion. It is reflected by thelow value of eeend/eemax.

The results show that none of the additives used (see Table 14) resulted inmeasurable rate decrease. However, substantial rate increase was observed

A B C

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

0.0 0.2 0.4 0.6 0.8 1.0

Conversion

0.0 0.2 0.4 0.6 0.8 1.0

Conversion

0.0

0.2

0.4

0.6

0.8

1.0

0.0

0.2

0.4

0.6

0.8

1.0

ee

Fig. 36 Influence of CD concentration on the hydrogenation of EtPy in the presence of Q.[EtPy]0=1.0M, PH2

=50 bar, solvent: toluene, 500 rpm, catalyst: 0.125 g, 5wt.% Pt/Al2O3

(Engelhard 4759), mode of introduction: Pr-I for Q followed by Inj-I of CD; A:[CD]0=0.6� 10� 5M; B: [CD]0=1.2� 10� 5M; C: [CD]0=6.0� 10� 5M, E – no Q; & –6� 10� 5M 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 – 0.031 0.053 98.8 0.666 0.343 0.515

2 Acridine 0.045 0.047 98.4 0.690 0.434 0.629

3 Quinoline 0.048 0.078 99.0 0.686 0.468 0.682

4 Pyridine 0.078 0.069 99.4 0.585 0.446 0.762

5 4-Picoline 0.042 0.059 99.5 0.575 0.447 0.780

6 Naphthalene 0.037 0.044 97.7 0.625 0.323 0.517

7 Antracene 0.033 0.049 98.3 0.620 0.323 0.521

8 Pyrene 0.039 0.048 98.4 0.619 0.337 0.544

[EtPy]0=1 M, [CD]=1.2� 10� 5M, [Additive]=1� 10� 4M, treact=90min, catalyst: 0.07 g2.7wt% Pt/SiO2, solvent: toluene, mode of introduction: Pr-I for additives followed by Inj-I ofCD.

216 | Catalysis, 2010, 22, 144–278

in the presence of acridine, quinoline and pyridine. Two classes of con-densed aromatic compounds can be differentiated: (i) nitrogen containingone resulting in increased eeend/eemax values and (ii) condensed aromaticcompounds having no influence on the eeend/eemax values.

Summing up these results it was concluded that in the enantioselectivehydrogenation 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 CDand quinoline over Pt sites can be considered as a competition. Theconsequence of this competition is the decrease of the number of sites in-volved in ED. Therefore, one would expect a decrease in the ee values.However, it is not the case, the ee increases when the CD/quinolineratio is properly chosen and the CD concentration is low. The observedeffect appeared both on alumina and silica supported Pt catalysts and wasfound to be strongly concentration dependent. However, according to the‘‘Surface model’’ the co-adsorption of these compounds should result in adecrease in the enantioselectivity, what is not observed in our study. Con-sequently, our results might indicate that the ‘‘surface model’’ needs somecorrections.

5.6.4 Modification of the support. There are only scarce data on themodification of the support. In one of our studies the influence of themodification of alumina support by alkyl silanes was investigated. Afterdehydroxilation of the support at 400 1C it was modified by different alkylsilanes resulting in anchored –Si(CH3)3, or –Si(CH3)2C8H17 moieties274. Therate of this anchoring type surface reaction can be controlled by the con-centration of the modifier, the temperature of anchoring reactions and thelength of the R group in the alkyl silanes.

Catalysts modified in this way were used in the enantioselective hydro-genation of EtPy in the presence of CD. The above modification was notbeneficial for the above reaction as the modified catalysts showed pro-nounced decrease in reaction rates and slight loss in enantioselectivitiescompared to the unmodified Pt/Al2O3. These results are shown in Figs. 37Aand 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

significantly altered upon using various pre-activation procedures as shownin Table 15. The removal of the anchored –O–Si(CH3)2R moieties the ori-ginal activity and enantioselectivity was restored completely.

The catalytic performance of these modified catalysts was significantlyaltered upon using various pre-activation procedures as shown in Table 15.The removal of the anchored –O–Si(CH3)2R moieties the original activityand enantioselectivity was restored completely. The behaviour of thistype of modified catalysts was explained by the following phenomena:(i) partitioning or retention of CD or the [substrate-CD] complex by an-chored –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 furtherindirect evidences that interactions in the liquid phase play a very importantrole. Results obtained in this study strongly indicate that in this enantio-selective hydrogenation reaction the enantio-differentiation cannot be at-tributed exclusively to the interaction between the half-hydrogenatedsubstrate and CD on the Pt surface.

5.6.5 Modification of Pt by other components. These studies were per-formed by English groups. These groups use the classical aerobic Pre-modification procedure (see Section 4.4), In the enantioselectivehydrogenation of MePy or butane-2,3-dione in the presence of cinchoni-dine-modified platinum catalysts it was shown that at the catalyst prepar-ation stage, the co-adsorption of the alkaloid with a strong co-adsorbate hasa strong positive effect.278 One of these coadsorbates was oxygen or airdissolved in reactant and solvent. In addition acetylene, methyl acetyleneand butadiene appeared to be effective co-adsorbates. It was suggested thatin the absence of a strong co-adsorbate the surface is poisoned bycinchonidine.

It has been shown that the modification under methylacetylene providesreaction rates and ee values excess under standard conditions (10 bar,293K) that are comparable to, or higher than those obtained with normalaerobic modification.

The importance of surface morphology of small supported Pt particleswas confirmed in refs. 233, 403. In these studies Pt/C and Pt/SiO2 catalysts

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 (%)

1 150 o0.001 58.3

2 250 0.012 77.2

3 400 0.020 84.7

4 400, blanka 0.044 85.0

5 400b 0.054 84.3

6 400, parent 0.057 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 atmo-sphere. c Catalyst without modification.

218 | Catalysis, 2010, 22, 144–278

modified by bismuth and sulphur and were characterized by electrochemicalmethods.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 de-

creased in sulfur modified one.� The formation of polymeric residues were strongly reduced over Bi

modified catalysts.� Both the enantioselective hydrogenation and the formation of poly-

meric 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 oc-curring at an enhanced rate at a constant number of sites.

It has to be stressed out that in this study EtPy was used without anypurification. It is the reason for extremely low rates in the racemic hydro-genation (see Table 16).

This fact strongly questions both the results and the conclusions. There isan additional serious drawback, i.e., the lack of information on racemichydrogenation over catalysts modified by Bi. In this respect it has tomention that the adsorption of Bi on the Pt can give different species, in-cluding metallic and ionic one.405 We suggest the rate increase and thedecrease of the ee over Bi modified catalysts is due to the acceleration effectin the racemic hydrogenation by bismuth cations created over the Ptsurface.

In this respect we should like to revert to our results on tin modifiedcatalysts (see Section 5.6.1). This method clearly indicated that the in-volvement of kink sites in the ee is excluded as tin is located on the kink siteand ee was independent on the amount of tin anchored to the platinum,while the rate showed a strong dependence.

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 None –b 0 2c

Pt CD 850 41 100

Pt QND 440 0 40

Pt 1:1 CD:QND 1205 37 –

Pt-Bi ((YBi)ch=0.35) CD 1350 35 49

Pt-Bi ((YBi)ch=0.35) QND 1710 0 36

Pt-Bi ((YBi)ch=0.35) 1:1 CD:QND 4600 15 –

Pt-S ((YS)ch=0.19) CD 440 52 109

Pt-S ((YS)ch=0.19) QND 310 0 –

Pt-S ((YS)ch=0.19) 1:1 CD:QND 645 51 –

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

� 1. c ConversionW20%.

Catalysis, 2010, 22, 144–278 | 219

6. Spectroscopic investigations

6.1 NMR

NMR techniques related to substrates, reaction products and chiral modi-fiers have widely been used in the investigation of enantioselective hydro-genation of activated ketones. It has also been applied for the determinationof by-products, alternative reaction routes and intermediate complexes andattempts were also made to use NMR for elucidation of different hypo-thetical 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 byNMR. It was also applied as a tool to check the purity of substrate.82 Inmany cases NMR gives an opportunity for identification of reactionproducts.203,210,235,331,336,338,406

It is known long before that cinchona alkaloids are extremely active chiralmodifiers in various organic reactions. Mapping the role of differentstructural elements of cinchonas in the enantio-selection several derivatives,especially C9 substituted cinchonas,202 inner ethers,407 N-alkyl deriva-tives314 have been prepared and checked by NMR,44,238,256,408 Uponidentification of the structure of ether derivatives of cinchonidine such as9–O-phenyl-202,206 9–O-pyridyl-,372 9–O-sylil-cinchonidine202 etc. NMR wasan indispensable tool. Beside of the aforementioned chiral modifiers, thestructure of several cinchona analogues, i.e. amines and amino alcohols44,378

and aryl alcohols226 prepared for chiral template in the Orito’s reaction hasalso been confirmed by NMR.

Chiral modifiers itself very often suffer changes during the enantioselec-tive hydrogenation. To follow the conversion of 9–O-pyrydil-cinchoni-dine372 and the saturation of naphthalene ring of 1-naphthyl ethylaminederivatives208,212 NMR was applied as well as to check the resistance ofphenyl group in 9–O-phenyl-cinchonidine203 and the stability of methoxy-cinchonidine372 and isocinchonines.351 NMR analysis of the reactionmixture showed that 1-naphthyl ethylamine derivatives is quantitativelyconsumed during the hydrogenation reaction and converted to the sec-ondary amine.212 It has been shown by NMR that quaternary ammoniumderivatives of CD as new chiral modifiers remained stable during hydro-genation.200 Formation and structure of hexahydro-cinchonidines andhexahydro-cinchonines has been investigated by NMR.271,409 According toNMR analysis, at 36% saturation of the quinoline rings of CD in acetic acidthe ratio of homoaromatic and heteroaromatic hydrogenation products was2.5 to 1.269

Upon hydrogenation of b-trifluoro ketones the sense of enantioselectivitychanges when the polarity of the solvent changes. The phenomenon wasexplained by the shift of keto-enol equilibrium confirmed by NMR.44,385 Inalcoholic solvent ethyl-4,4,4-trifluoroacetoacetate has shown a reactionroute via semi-ketal.348 Semi-ketal was also detected by use of NMR inother cases.331,410 IR and NMR experiments have revealed that the enan-tioselective hydrogenation of EtPy in nonacidic solvents is complicated bythe simultaneously occurring self-condensation (aldol reaction) of thereactant.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 pathwayswere followed by NMR in the enantioselective reduction of isatin deriva-tives over cinchonidine modified Pt/alumina.411

In the study of hydrogenation of 1,1,1-trifluoro-2,4-diketones strongacid-base interaction has been revealed by NMR. The high chemo- andenantioselectivities in the above reaction were attributed to the formation ofan ion pair involving the protonated quinuclidine part of the chiral modifierand the enolate form of the substrate.406 In the hydrogenation of KPL theformation of dimer was confirmed by NMR.209,323

Recently a mechanistic model involving nucleophilic catalysis and zwitter-ionic adduct formation between a cinchona alkaloid and activated ketone hasbeen suggested.412 Upon investigation of hydrogenation of flouoroketones theformation of the ionic species in the solution was studied by 13C-NMRspectroscopy. Trifluoromethylcyclohexyl ketone was used as a strong elec-trophile agent and the complex was created by the addition of excess tertiaryamine quinuclidine. The adduct formation was studied in two different solventsystems such as deuterated chloroform and acetone but the formation of thezwitterionic species was observed only in acetone.216 13C-NMR confirmed theexistence of other adducts of the zwitterionic type.360 However an NMR studyhas indicated that the zwitterions model is probably based on erroneousinterpretation of the experimental data; the NMR spectra that had beenreported for zwitterion formation may arise from an aldol addition the a,a,a-trifluoromethyl ketone and the solvent acetone, and the reaction is catalyzedby 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 isnot the reacting species, but under condition of enantioselective hydro-genation deuterium exchange takes place not only at the quinoline ring, butat C9 carbon atom, too.414

6.1.1 Conformation analysis of the modifier. The study of the con-formation of cinchona alkaloids investigated by NMR has been brieflymentioned in Section 2.3. The question of which conformer of cinchonidineis involved in the enantio-differentiation step is regarded as a key issue.Baiker el al. have used ab initio calculations and NMR measurements toinvestigate the conformers of (dihydro)cinchonidine in different solvent(such as benzene, toluene, ethyl ether, acetone, etc.).88 The existence of agiven conformer has been rendered by nuclear Overhauser enhancementspectroscopy. NOESY experiments have suggested that Open (3) andClosed (1) and Closed (2) conformers appears. This observation is inqualitative agreement with their calculations. The dihedral angles for dif-ferent conformers have been calculated by ab initio methods. The measuredcoupling constants (13JH8H9(exp)) and the dihedral angles by applying theKarplus equation415 have given possibility for calculation of couplingconstants of different conformers (13JH8H9(i)). The above method is limitedto the determination of only two conformers from one coupling constantsmeasured, however the dihedral angle has been found very similar for thetwo closed conformers of CD, so the population of Open (3) and the sum ofpopulations of closed conformers has been possible to calculate. The resultsare represented in Table 17.

Catalysis, 2010, 22, 144–278 | 221

According to the above approach the conformer Open (3) has been foundthe most stable one.88 Based on parallel solvent dependence of ee and popu-lation of Open (3) it has been suggested that Open (3) conformer plays crucialrole in the asymmetric induction. In case of 9-deoxy-CD derivatives similarmethod has been used.416 The coupling constants of 2-phenyl-9-deoxy-10,11-dihydrocinchonidine (13JH9H8a (5.0Hz); 13JH9H8b (9.3Hz)) is similar to thatof 9-deoxy-10,11-dihydrocinchonidine (5.5 and 8.5Hz) in CDCl3 indicatingthat the relative stability of conformers is also similar. NMR experiments andab initio calculations revealed that conformer Closed (1) is stabilized relativeto 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 CDwas investigated.417 It was shown that protonation strongly hinders therotation around the C40–C9 and C9–C8 bond. Structures and conforma-tional behaviour of several cinchona alkaloid O-ethers in solution (NMRand DFT) were also investigated.391 It was demonstrated that the con-formation found to be abundant in the liquid phase has no direct correl-ation with the enantioselectivity of the PPD hydrogenation reaction. Theauthors concluded that the driving force for production of one of the en-antiomers in excess is due to the specific adsorption of the modifier on thecatalyst surface, a phenomenon that does not correlate with the populationof the conformers in the liquid phase.

6.1.2 Substrate-modifier interaction. In an early work NMR measure-ments have already shown an interaction between CD and EtPy in the liquidphase.84 It was shown that in CD3OD in the presence of 0.15M MePy thecharacteristic doublet of CD at 5.65 ppm was shifted to 5.85 ppm and a newsmall singlet was observed at 6.0 ppm. Upon increasing the concentrationof MePy to 0.6 or 1.0M the doublet vanished and only the new singletat 6.0 ppm was found. More noticeable shift of the C(9) proton, up to6.3–6.4 ppm with a formation of a singlet was observed in neat CD3COODor if small amount of CD3COOD was added into the solution of CD inC6D6. These NMR results suggested that the torsional angle between thehydrogen atoms at C(8) and C(9) carbon atom of CD has been changedresulting in a new conformer of CD.

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 hT3JH8H9 Popen (3) PClosed

benzene 2,28 5.0 0.58 0.42

toluene 2,34 4,1 0,7 0,3

ethylether 4,3 4 0,71 0,29

tetrahydrofurane 7,6 4,7 0,62 0,38

acetone 20,7 6,4 0,4 0,6

dimethylfomamide 36,7 7 0,33 0,67

dimethylsulfoxide 40 7,5 0,27 0,73

water 78,5 7,2 0,3 0,7

ethanol 24,3 3,5 0,77 0,23

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

222 | Catalysis, 2010, 22, 144–278

In Pt-catalysed hydrogenation of 1,1,1-trifluoro-2,4-diketones the com-bined catalytic, NMR and FTIR spectroscopic, and theoretical study re-vealed that high chemo- and enantioselectivities are attributed to theformation of an ion pair involving the protonated amine function of thechiral modifier and the enolate form of the substrate.406

On the basis of NOE studies and theoretical calculations related to thehydrogenation of ketopantolactone in the presence of the (R,R) and (R,S)diastereomers of a new chiral modifier, pantoyl-naphthylethylamine, differ-ent properties of the above diastereomers were investigated, in particular theeffect of acid on the modifier structure.209 The results indicated that in case ofthe (R,R)-diastereomer in apolar solvent, a loose, extended structure changesto a compact one via an additional intra-molecular hydrogen bond, resultingin a more suitable ‘‘chiral pocket’’ available for the reactant on the Pt surface.

Standard 2D NMR spectroscopic methods and diffusion-ordered NMRspectroscopy combined with theoretical calculations has been used to verifythe formation of supramolecular complexes between the pairs O-methyl-cinchonine–ketopantolactone (KPL) and b-isocinchonine–KPL.418 WhenO-methyl-cinchonine or b-isocinchonine and KPL were mixed in dry deu-terobenzene solution, time-dependent chemical shift changes for the cin-chonas and new signals for KPL bound to the modifier have been detected.The spatial pattern of the chemical shift differences and the conformationsof the modifiers determined by NOESY demonstrated that the substratebinding occurs at the quinuclidine N atom, H9, and the quinoline H50 re-gion for O-methyl-cinchonine (H8 and H50 for b-isocinchonine. Based ondiffusion measurements hydrodynamic radii has been estimated which hasproved the co-diffusion of the cinchonas and KPL in a complex. The resultshave shown that not only 1:1, but also 2:1 cinchona-KPL complexes mustbe taken into account. NMR evidences has also been found for the cor-relation between the solution-state concentration of the nucleophilic 1:1modifier-substrate complex and the ee on enantioselective hydrogenation ofKPL using Pt–b-isocinchonine chiral catalyst.418,419

6.2 Circular dichroism

Vibrational circular dichroism (VCD) is a useful tool to determine the ab-solute configuration of the enantiomer produced in excess in an enantio-selective reaction when reference data on the enantiomer are not available.The absolute configurations of the enantiomers can be obtained by com-paring the theoretically calculated VCD spectrum of one enantiomer withthe experimental VCD spectrum of the product of the asymmetric reaction.It is important to know that VCD signal is about three orders of magnitudeless intensive than the corresponding signal in the ordinary transmissionspectrum.235 This method has been successfully applied for the determin-ation of product alcohols in the studies related to the directing effect oftrifluoromethyl group235 or ester group338 of the substrates. It was also usedupon investigating CD modified Rh/alumina catalyst420 in the hydrogen-ation of various aromatic ketones possessing an a-hydroxy or a-methoxygroup and, in case of the enantioselective reduction of isatin derivatives overCD modified Pt/alumina.411

Catalysis, 2010, 22, 144–278 | 223

Theoretical (DFT) and VCD spectroscopy study has been applied for theconformational analysis of the synthetic chiral modifier 9-O-phenyl-CD.421

According to these results 9-O-phenyl-cinchonidine behaves similarly toCD and shows four main conformers, denoted as Closed (1), Closed (2),Open (3), and Open (4). A combined theoretical-experimental VCD spec-troscopy approach has given possibility to increase the spectroscopic sen-sitivity toward changes in the distribution of conformers upon changing thesolvent polarity. The VCD spectra confirm that Open (3) is the most stableconformation in CCl4. Changing from CCl4 to CDCl3 the equilibrium be-tween the conformers does not change significantly. Upon increasing solv-ent polarity besides similar non-coordinating properties the fraction ofClosed (2) species increases considerably.

Relating the conformational results to the enantio-differentiation shownby this modifier (9-O-phenyl-CD) in the platinum-catalysed asymmetrichydrogenation of KPL the inversion of the sense of enantio-differentiationobserved cannot be traced to the conformational behaviour.421

HPLC and UV-vis/circular dichroism299 has been used to assess con-version and selectivity in chiral fixed bed reactor for stereoselective het-erogeneous catalysis.422 The UV-CD method and the HPLC-CD methodhave been used to simultaneously determine ee values and concentration ofeach enantiomer.423

Tungler and coworkers have described that the circular dichroism spec-trum of dihydrovinpocetine changes upon addition of both isophorone aswell as EtPy indicating interactions between these two substrates and thechiral modifier.68 The similar method has been applied in case of (S)-prolinebased chiral modifiers.224 The circular dichroism spectra of CD in toluenehas been found to change by addition of EtPy as shown in Fig. 38.93 Add-ition of EtPy to cinchonidine in chloroform has also resulted in changes inthe circular dichroism spectra of cinchonidine, although these changes wereless pronounced than those in toluene. The above results strongly indicatethat there is an interaction between CD and EtPy in the liquid phase.

The results of circular dichroism spectroscopy401 have provided furtherproof for dimer formation of CD in liquid phase. These results are related tothe addition of ATAs discussed in Section 5.6.2.

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

224 | Catalysis, 2010, 22, 144–278

The Cotton shift of CD was measured at different [CD]0/[QN]0 ratios.The corresponding circular dichroism spectra are shown in Fig. 39. TheCotton shift around 235 nm appeared to be very sensitive to the amount ofquinuclidine added. This Cotton shift is related to the dimer form ofCD.191,400 Analogous results using MeO-CD have shown that this alkaloidcannot 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 farto investigate the interaction of substrates and modifiers or the substrate-modifier complex with the Pt surface. Most of these methods are usingconditions (often high-vacuum) far from those applied in catalytic hydro-genations. For this reason, although some important details of the ad-sorption behaviour of CD and substrates have been revealed, the resultshave to be treated with certain precaution.

Surface characterization methods applied so far are as follows: (i) near-edge absorption fine structure spectroscopy (NEXAFS),91 (ii) X-ray photo-electron spectroscopy424,425(XPS), (iii) low-energy electron diffraction(LEED,425 (iv) scanning tunnelling microscopy (STM),425–427 (v) reflection-absorption infrared spectroscopy (RAIRS),428,429 (vi) surface-enhancedRaman 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 aflow-through cell433 can be considered as appropriate methods approachingalmost real in situ conditions. It has to be emphasized that above twotechniques have the advantage to obtain information about adsorptionprocesses at the solid-liquid interface. In this respect it is important tomention that exact vibrational assignments for adsorbed CD on Pt surfaceusing combination of experimental vibrational spectroscopic measurementsand ab initio computational methods were also reported.92,434,435 Recently a

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

Catalysis, 2010, 22, 144–278 | 225

new method for in situ spectroscopic investigation of heterogeneous cata-lysts 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 Theresults indicated that EtPy adsorbed strongly to the Pt. The ketone carbonylis more strongly involved in the chemisorption bond than the carboxyl one.Further analysis showed that EtPy is predominantly adsorbed in a tiltedrather than a completely flat mode.

The behaviour of EtPy during adsorption on alumina-supported plat-inum films and on a commercial 5wt% Pt/ Al2O3 catalyst has been studiedin the absence and presence of coadsorbed CD.91 The in situ ATR-IR studyat room temperature using hydrogen-saturated CH2Cl2 as solvent demon-strated that upon adsorption on the Pt EtPy decomposes with the formationof strongly adsorbed CO and other organic residues. The presence of CD(10� 4M) strongly decrease the rate of decomposition of EtPy.

Upon using STM method self-condensation of MePy over Pt surface wasobserved.308 This reaction took place in the absence of cinchona modifier atlow hydrogen coverages. Based on this finding new set of side reactions withthe involvement of MePy was proposed and conditions to avoid the by-product 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 (aldolcondensation) under conditions of hydrogenation. These side reactions takeplace both on the Pt site and the Al2O3 support.

Based on analysis of the RAIRS spectra of MePy it has been shown thaton Pt(111) at room temperature MePy undergoes ‘‘surface mediated enolformation’’ leading to an assembly of H-bonded superstructures.310 Thedecrease of the temperature and the use of low background hydrogenpressure suppress these surface reactions.

In a recent study adsorption and reaction of EtPy on Pt/g-Al2O3 wasstudied 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-highvacuum conditions on Pt single-crystal surfaces using X-ray and UVphotoelectron spectroscopies (XPS and UPS),424 NEXAFS,96 and(RAIRS).437 The results indicated that alkyl pyruvates adsorbs via lonepair-metal interaction of both carbonyl groups, i.e., in cis conformationwith their plane oriented normal or tilted with respect to the surface. Athigh coverage, a minority species was assigned to an Z1-transconfiguration.437

The coadsorption of hydrogen resulted in significant influence on theadsorption of alkyl pyruvates by lowering the tilting angle of the adsorbedspecies438 and suppressing surface polymerization of the adsorbed enedio-late species observed earleir.308

6.3.3 Investigation of the modifiers. Adsorbed forms of cinchona al-kaloids display different IR spectra from each other and from the solutionform 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 sur-faces. Although this method seems to be very powerful it could not answerthe key question, namely, which of these species interacts with the substratein the enantio-differentiating step.

The first information that the mode of adsorption of CD on Pt dependson its coverage was obtained by using in situ ATR-IR spectroscopy.240 Atlow coverage the flat mode, while at high one the tilted mode prevails.Further study revealed that abstraction of hydrogen form the quinoline ringcan also take place resulting in a so called a-H abstracted form.431 Study onPt/Al2O3 in the presence of an organic solvent and hydrogen revealed threedifferent adsorption modes of CD as shown in Fig. 40.

Infrared spectroscopy (IR), Raman spectroscopy, surface-enhancedRaman 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 usingATR-IR spectroscopy. The adsorption appears to be more complex thanthat observed on Pt and Pd. Strongly adsorbed flat form was observed onRh when adsorption was performed in the absence of dissolved hydrogen.This form is responsible for the fast hydrogenation of the quinoline ring anddoes not allow the detection of the flat form in the presence of dissolvedhydrogen.270 Contrary to Pt it has been discovered that on Rh hydrogen-ation of the heteroaromatic part of the quinoline ring takes place. AdsorbedCD in the flat geometry is the intermediate of the hydrogenation reaction,

Fig. 40 Suggested adsorption mechanism of cinchonidine on Pt/Al2O3 at 283K based on ATRexperiments; y represents the surface coverage. Species 1: p-bonded, 2: a-H abstracted and 3: Nlone 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 situRAIRS technique.428 Good correlation has been found between the surfaceconcentration of flat lying CD and the enantioselectivity in EtPyhydrogenation.

In another study it was found that the oxygen present in most CD so-lutions from dissolved air blocks the surface toward any CD uptake, pre-sumably via its dissociation to atomic surface oxygen atoms (and maybe bypartial oxidation of the platinum surface), while Ar, N2, or CO2 has noinfluence on the adsorption of CD. Hydrogen plays a unique role, initiallyfacilitating the uptake of CD.90

Non-linear effects (see Section 5.5.3) has also been characterized byin situ ATR-IR spectroscopy comparing the behaviour of CD andPhOCD.435 It was shown that both alkaloids are adsorbed via the quinolinering and that the spatial arrangement of the quinuclidine ring is crucial forthe chiral recognition. The result helped to elucidate the role of the phenylgroup played in the creation of the chiral space responsible for the inversionof ED.

Surface-enhanced Raman spectroscopy has been applied to investigatethe adsorption of CD on polycrystalline Pt.92 The effects of liquid-phaseconcentration in ethanol and that of co-adsorbed hydrogen were studied.It was found that CD is strongly and irreversibly adsorbed through itsquinoline ring via p-bonding. Stronger adsorption of DHCD comparedwith CD was also suggested.

The room-temperature adsorption of four cinchona alkaloids and threereference quinoline-based compounds from CCl4 solutions onto a poly-crystalline Pt surface was characterized by in situ RAIRS.377 The results areshown in Fig. 41. Data show Langmuir type adsorption kinetics. The

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

228 | Catalysis, 2010, 22, 144–278

calculated adsorption equilibrium constants (Kads) are given in Table 18and were found to follow the sequence CNWQDWCDWQNW6–methoxy-quinolineWlepidineWquinoline. Some of this ordering can be explained bydifferences in solubility, but QD displays a much larger Kads than expectedon the basis of its large relative solubility.

Results indicated also that each alkaloid binds differently on Pt at sat-uration coverages. At low concentrations all alkaloids adsorb with theirquinoline ring flat on the surface and then tilt abruptly upon increasingcoverages, but the switch-over takes place at significantly different solutionconcentrations 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 ringdistortion. The most surprising result is the fact that CN shows a higherKads than CD, QN, or QD. In this respect results obtained in an earlierstudy has to be mentioned. As shown in Fig. 42 when in sequential intro-duction of CCl4 solutions of CN, CD, and back to CN was applied CN wasreplaced by CD, while CN cannot replace CD.376 Similar conclusions can beobtained from other studies using the ‘‘chiral switch’’ technique.238,379 Thedifference in the adsorption mode of CD and CN was investigated in arecent study.439

The main message from these studies is that the solvent has to be takeninto consideration in the formation of the above discussed adsorbed formsof alkaloids.

The adsorption of 1-(1-naphthyl)ethylamine (NEA) on platinum surfaceshas also been characterized by RAIRS and temperature-programmed de-sorption (TPD) both under ultra-high vacuum and in situ from liquidsolutions.440

ATR-IR spectroscopy was also used to prove the flexibility of the qui-nuclidine moiety resulting in surface quinuclidine bound CD. It was done bycomparison of the ATR-IR spectra of CD and PhOCD adsorbed onPt.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 thepossible interaction of the quinuclidine moiety of CD with the Pt surface.Based on the comparison of the ATR-IR spectra of CD and CD hydro-chloride adsorbed on Pt under similar condition the authors came to theconclusion that at the Pt the quinuclidine moiety of CD has identicalstructure as in the protonated quinuclidine of CD hydrochloride.98 Thisfinding was considered as an additional evidence that CD can be protonated

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. (Repro-

duced from ref. 377 with permission)

Compounds Kads� 1 mM Solubility in CCl4, mM Kads

� 1/solubility

Quinoline 30 infinite

lepidine 11 infinite

6-methoxyquinoline 6.5 infinite

QN 0.65 8.63 0.075

CD 0.5 1.56 0.32

QD 0.25 16.3 0.015

CN 0.1 0.30 0.33

Catalysis, 2010, 22, 144–278 | 229

by chemisorbed hydrogen in aprotic solvent. However, the authors admitthat ‘‘this conclusion is tempting, further studies are needed to confirm itsvalidity. Doubts may arise from the presence of HCl originating fromCH2Cl2 solvent decomposition on Pt’’.327 We also have serious doubt withrespect to these interpretations.

ATR-IR method was also used to give some new information about theformation of quinuclidine bonded form of alkaloids. However, these resultsare quite dubious. In this respect in a recent study the following informationwas given:270 ‘‘Recent results indicate that the quinuclidine moiety is alsoinvolved in adsorption on Pt.98,441 At low coverage, the energetically fa-voured geometry exhibits the aromatic ring parallel to the surface and thequinuclidine moiety oriented toward the metal surface in a geometry thathas been named surface quinuclidine bound (SQB).442

However, the careful analysis of references given above clearly indicatesthat there is no experimental evidence for the above statement. In ref. 441upon using molecular dynamics simulation ‘‘CD was found to adsorb withthe quinoline ring oriented largely parallel (ao61) to the surface. CD sur-face attachment was found to be through both p- bonding of the aromaticgroup and adsorption of the CQC double bond of the vinyl group’’. It wasalso mentioned that ‘‘we found that CD conformation at the surface wasnot 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

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)

230 | Catalysis, 2010, 22, 144–278

quinuclidine moiety in the adsorption. In ref. 98 only computational resultswere given, while ref. 442 contains scientific speculations and not real evi-dences. Consequently, there is no exact experimental evidence for the for-mation of surface quinuclidine bound CD.

The adsorption of CD and CN on Pt(111) and Pd(111) single crystals hasbeen investigated by means of time-lapse STM in an ultra-high vacuumsystem.443 CD and CN showed similar adsorption modes and diffusionbehaviour on Pt(111). The only exception is that the flatly adsorbed CNmolecules were significantly more mobile than CD.

NEXAFS and corresponding STEX calculations have been appliedto investigate the orientation of DHCD on Pt(111) at 298 and 323K.91

The results indicate that at 298K the quinoline ring is almost parallelto the Pt surface but is tilted up from the surface by 60� 101. However, theresults show that at higher temperatures the alkaloid dissociates toquinoline.

Various techniques, such as NEXAFS, XPS, STM, and temperatureprogrammed reaction was applied to investigate at 320K the molecularorientation, spatial distribution, and thermal behaviour of the powerfulchiral modifier precursor (S)-naphthylethylamine adsorbed on Pt(111)427

No formation of ordered arrays was observed in the presence or the absenceof coadsorbed hydrogen. Based on high resolution STM images somespeculation was done related to the formation of 1:1 docking complex be-tween MePy reactant and the chiral modifier.

NEXAFS revealed that the quinoline ring of 10,11-dihydrocinchonidineis orientated parallel to the surface at 298K, whereas at 323K the orien-tation is tilted about 601 to the surface.91 von Arx et al. used STM to revealthat the cinchonidine molecules are randomly distributed on the Pt (111)surface.444

Attard and co-workers studied the influene of surface structure andsurface chirality on the adsorption rate of several modifiers.445 However,probably due to the large adsorption energy of these systems, no differencein the adsorption rate was observed. It was also observed that in a hydro-gen-saturated solution, the alkaloid dihydrocinchonine is partially desorbedfrom a kinked, chiral Pt surface.

The adsorption of CD on polycrystalline Pt surfaces in H2SO4 was in-vestigated by cyclic voltammetry.241 The adsorption was found to be ir-reversible. The results indicated that at maximum coverage, 50% of the Ptatoms were still accessible for hydrogen adsorption. They calculated alsothe site requirement for CD equal to 13–14 Pt atoms. In another study it wascalculated that in the surface modification model each enantioselective siterequires 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 poly-crystalline Pt electrode has been studied by differential electrochemical massspectrometry (DEMS). It was shown that benzene is even able to displacesome of the alkaloid.447

Electrochemical method was applied to investigate the introduction ofcinchona alkaloids with R- and S-kink sites of the Pt(643) surface.447 Nointeraction was evidenced.

Catalysis, 2010, 22, 144–278 | 231

6.3.4 Substrate modifier interactions. Interaction of KPL with CD wasinvestigated by ATRIR concentration modulation spectroscopy usingCH2Cl2 as a solvent.314 The results showed that in the presence of CD andKPL 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 bychemisorbed hydrogen. This experimental results is considered as a keyprove for the support of authors general view with respect to the reactionmechanism (see more details in Section 8), In this respect the use of CH2Cl2solvent has to mentioned. Due to its use it is not excluded that the observedprotonation is simple an artefact.

Recently the formation of a surface complex between adsorbedcis-EtPy and protonated CD has been suggested using ATR-IR methodduring asymmetric platinum catalysed hydrogenation of EtPy in super-critical ethane solvent.448 These results are shown in Fig. 45. Basedon the shifts in the 1200–1300 cm� 1 region preferential adsorption ofEtPy as cis-conformer was suggested. The appearance of the band at1660 cm� 1 was tentatively ‘‘be attributed to carbonyl stretching vibrationsof EtPy’’.

Fig. 43 Demodulated ATR spectra of different concentration modulation experiments. TheKPL concentration was modulated (modulation period T (184 s) between 0 and 5� 10� 2mol/Lin 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� 4mol/L). Before the modulation experiments were started (c–e), the Pt/Al2O3 filmswere differently treated: (c) 30min N2 saturated CH2Cl2 only; (d) pretreatment with 5 min H2

saturated KPL solution; (e) directly contacted with modulation solutions. (Reproduced fromref. 314 with permission)

232 | Catalysis, 2010, 22, 144–278

The above shift to lower wave-numbers occurs ‘‘due to hydrogen bondingbetween the quinuclidine nitrogen and the keto oxygen atom of EtPy’’.However, in this respect it is necessary to mention the author’s followingstatement: ‘‘This interpretation is in line with our recently reported study on

Fig. 44 Comparison between two KPL concentration modulation ATR experiments. TheKPL concentration was modulated (modulation period T 184 s) between 0 and 5� 10� 2mol/Lin CH2Cl2. Use of modifiers: (a) N-Methyl CD (5� 10� 4mol/L), (b) CD (5� 10� 4mol/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 inEP; (4) CD-premodified Pt/Al2O3. Conditions in all experiments were 40 1C, and 95 bar. Molarratio EP:H2 : ethane=1:5:200. (Reproduced from ref. 448 with permission)

Catalysis, 2010, 22, 144–278 | 233

KPL adsorption on a CD-modified Pt/Al2O3 thin film’’.431 However, in theabove reference there is no words related to the assignment of any bandsaround 1660 cm� 1.

The conformational flexibility of the quinuclidine moiety was investigatedby ATRIR experiments under nearly in situ conditions, by comparing theadsorption behaviour of CD and O-phenyl-CD on platinum.98,435 It wasconcluded that the tertiary nitrogen of the quinuclidine moiety can par-ticipate in the anchoring of the alkaloid and can be protonated by surfacehydrogen. This study is related to the mechanism proposed by Baiker’sgroup whereby the tertiary nitrogen can promote charge polarization ofhydrogen and its transfer to the substrate (see more details in Section 8).

There is one more comment on these results. Even if we accept resultsshown in Figures 43–45 one question still remains: What is the proof thatsurface species assigned to the [CD-Hþ -substrate] complex is really in-volved in the ED?

6.3.5 HPLC-MS and HPLC-ESI-MS investigations. These studies wererelated to the investigation of the products formed from different alkaloidsduring enantioselective hydrogenation reactions. Upon investigation of theeffect of a-ICN it was demonstrated by HPLC–ESMS measurements thatthe cyclic ether structure of the alkaloid remained unchanged.449 In anotherstudy the product of isomerization of b-ICN, b-isocinchonicine (b-ICNN),was hydrogenated using supported Pt and Pd catalysts. The products wereanalyzed using HPLC-ESI-ion-trap MS measurements.450

Combined HPLC and ESI-MS method was used to investigate cinchonaalkaloid derivatives formed in the hydrogenation of a-ICN and b-ICN.197

The products of reaction are shown in Fig. 46. The hydrogenated com-pounds were identified as 10,20,30,40- tetrahydro-a-ICN (A) and 10,200,300,400-tetrahydro-b-ICN (B) and decahydro-a-ICN (C).

Upon investigating C9-O-substituted cinchona alkaloids in the enantio-selective hydrogenation of EtPy ESI-MSD-ion-trap method was applied tofollow and identify the hydrogenated derivatives of these cinchonaderivatives.206

HPLC-MS method was used to investigate products of H–D exchangemeasurements of different alkaloids.451 As revealed by these measurements,iso-alkaloids are not converted back to CN or QD; (v) in all alkaloidsstudied, H–D exchange takes place on the quinoline skeleton as well as oncarbon atom C9; (vi) H–D exchange on the quinuclidine skeleton appearssignificant only in the case of CN and a-ICN.230 Deuterium exchange in CDwas 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 an-other study. The aim of this work was identifying hydrogenated cinchonasformed during enantioselective hydrogenation of EtPy. The target com-pound was DHCN which might possibly be formed from a-ICN. The resultsobtained by HPLC/ESI-MS measurements showed that of DHCN was notformed.385

7. Theoretical calculations

7.1 Introduction and first attempts

The proposed mechanisms for the asymmetric hydrogenation of activatedketones over cinchona-platinum catalyst system that leads to the observedED require supports using different theoretical-computational studies.These studies are related to the conformational analysis of both substratesand modifiers, adsorption of both substrates and modifiers into Pt andenergetic calculations for the whole complex system Pt-modifier-substrate.

It is generally accepted that the CD or its natural or synthetic analogueforms a 1:1 complex, what is hydrogenated on the metal surface. Thequestion is where the chiral discrimination takes place on the Pt surface orin the liquid phase.

Both the structure and the conformational complexity of cinchona al-kaloids generate several possible interactions with the substrates. As a re-sults a ‘‘chiral pocket’’ is created for the ED either on the Pt surface or in theliquid phase.

The way to explore these properties one has to investigate or modeltheoretically the characteristic features of substrate molecules and the cin-chona alkaloids. Based on this knowledge, the modifier-Pt and the sub-strate-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 molecularmechanics452 and quantum chemical calculations.453 Molecular geometriescan be optimized on MMFF94 molecular mechanic level. Relativelyaccurate energies (mainly for energy differences) can be obtained e.g. onHF-SCF/6–31G* or B3LYP/6–31G* level single point energy calculations.The involvement of metal in these calculations requires the use DFTmethods.

In the next sub-sections we shall review most of the relevant results re-lated to the computation on substrates, modifiers. Calculations related tothe substrate-modifier interactions and possible interactions of all thesecomponents with Pt surface will be discussed in Section 8 related to thereaction mechanism.

7.2 Characteristic features of substrate molecules

Among the substrates, for which the cinchona-Pt catalytic system yieldshigh ee values, the following groups of molecules are in the focus of ex-periments 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 computationalstudies.

It was also shown that s-trans conformer of MePy is more stable in the gasphase by 1–2kcal/mol, but the relative stability could be strongly influencedby the metal surface, especially because the s-cis conformer has a consider-ably larger dipole moment.454 With respect to the formation of substrate CDcomplex the two conformers of MePy were compared,455 and it was foundthat the complex yielding (R)-methyl lactate upon hydrogenation wasmore stable than the corresponding pro-(S) complex by 1.8 kcal/mol(corresponds to an enantiomeric excess of 92%, in good agreement withexperiment), however, for the analogous complexes of s-cis-methyl pyruvatethe energy difference is only 0.2 kcal/mol in favour of pro-(R), correspondingto 17% ee value. The relation between the electronic structure ofa-substituted ketones and their reactivity456 in the racemic and enantiose-lective Pt-catalysed hydrogenation was also investigated. A correlation be-tween the keto carbonyl orbital energy and the hydrogenation rate wasfound, which rationalizes the effect of the substituent on the rate of hydro-genation (the often observed rate acceleration).

The first model calculations indicated that in the complex responsible forthe enantio-differentiation the a-keto ester existed in trans-conformation.457

Further studies revailed that in the enantio-differentiation complex thea-keto ester can also exist in its cis conformation.455 The fact that in thehydrogenation of KPL to (R)-PL high enantioselectivites were obtainedindicated that the rigid cis conformation has no influence on the enantio-differentiation step.

Generally speaking, the number of substrate molecules with high ee is in arelatively narrow59,370 range: This strong substrate specificity has not beenanswered yet, neither by theory nor with computation. a-keto esters weremodelled in different studies.49,74,456,458 The trans conformer is more stablethan the cis one, the carbon atom in the keto group is partially positivelycharged, while the oxygen part is negatively.

In the hydrogenation of acetophenone and TFA derivatives on CD-modified Pt/Al2O3, the rates and ee values varied strongly with the nature ofthe aromatic substituents.332,363 The different reactivities were attributed tothe electronic (and steric) effect of the substituents and to hydrogen-bondinginteractions between the quinuclidine N atom of the alkaloid and the car-bonyl group of the substrate.359,456 Theoretical calculations revealed a linearcorrelation between the logarithm of the reaction rate and the highestoccupied molecular orbital and lowest unoccupied molecular orbital stabili-zation DEorb of the carbonyl compounds, relative to the reference com-pound (see Fig. 47.).53 The relative orbital stabilization is defined as the sumof two numbers: the difference between the energy of the anti-bonding or-bital of the reference compound acetophenone and that of the substitutedacetophenone, and the corresponding energy difference for the bondingorbitals. The more stabilized the orbitals of the substituted acetophenoneare, the larger DEorb and the reactivity of the molecule are. According tothese calculations (where the metal surface was not involved), the origin of‘‘ligand acceleration’’ is the lowering of the p-orbitals in the diastereomericcomplex of the substrate and modifier. In the pro(R) and pro(S) complexes,

236 | Catalysis, 2010, 22, 144–278

the carbonyl -orbitals are differently stabilized, which results in differentintrinsic rates in the formation of the two enantiomers. It remains, however,to be proven that the concept can be extended to other substrates andreaction types.

The proton affinities of seven different ketones, vicinal diketones, anda-keto esters (acetophenone, TFA, 2,3-butanedione, PPD, MePy, EBF andKPL) have been evaluated theoretically using the conventional ab initio HFand several post-HF methods (MP2, MP4, CCSD), density functionalmethods with the B3LYP hybrid functional, as well as some ab initio modelchemistries [CBS-4M, G2(MP2), and G3(MP2)//B3LYP].459

In the most stable protonated species, the proton is bound to oneof the carbonyl oxygens in the molecule. The preferred site depends on themolecule. In two a-keto esters (MePy and KPL) the carbonyl oxygenof the ester group is protonated. In the case of EBF and the asym-metric vicinal diketone, PPD it is the carbonyl oxygen next to the phenylgroup, which forms a more stable bond with the proton. These pheno-mena can be understood in terms of resonance stabilization of theresulting cations. It was shown that the protonation of both the modifierand 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 eewas 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. Theadsorption of both the keto and enol forms of 1,1,1-trifluoro-2,4-diketoneinto Pt(111) was modelled and calculated406 DFT calculations including thesimulation of the interaction of a protonated amine with the enolate

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

Catalysis, 2010, 22, 144–278 | 237

adsorbed on Pt revealed that only the C–O bond next to the CF3 group ofthe 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 inSections 2 and 3. For this reason, we shall refer to information given in thesesections. In the next sub-section we shall focus mainly on the conforma-tional analysis of cinchona alkaloids and their analogues.

7.3.1 Conformational behaviour of cinchonidine. Among the effectivechiral modifiers used for enantioselective hydrogenation of activated ke-tones 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 intensivelyinvestigated by NMR methods176,186 as described in Section 2. Conforma-tional behaviour of cinchonidine was calculated independently by differentgroups.83,88,460,461 Most of these results are in accordance with earlier resultsdiscussed in Section 2.176,183 In our first study,84 the conformational analysiswas done by using rigid quinoline and quinuclidine moieties. As a result,four stable conformations have been found. In our subsequent study,461 allof these calculations were repeated in such a way that only the phi and psitorsion angles were forced to be constant, while all other freedom of themolecule were left to relax. The conformational analysis indicated thatCD might exist at least in nine different forms, however only four of themare 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 thisfigure gives the contour of the possible forms of CD within 8 kcal/molenergy range. For CD the 2–D NOE spectra indicate446 that the majorconformation in solution is conformation A2, this is close to that adoptedby the molecules in the solid state.179 Thus, the conformational analysisstrongly indicates that CD can exist both in open and closed forms and bothforms of CD can be involved in the formation of substrate–modifiercomplex.

Detailed conformational analysis of CD in solutions using NMR tech-niques as well as theoretical calculations was done in ref. 88. Three con-formers of CD are shown to be stable at room temperature, cl(1), cl(2), andop(3), with the latter being the most stable in apolar solvents. The stabilityof the closed conformers relative to that of open(3), however, increased withsolvent polarity. In polar solvents the three conformers have similar energy(Fig. 48).

Structures and relative energies in kcal/mol of low energy CD conform-ations were calculated using hybrid density functional (B3LYP/6–31þG*/PCM B3LYP) and AMBER* optimization (AMBER*/GB/SA).416 Therelative 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 ina recent study.417 It was shown that the protonation of cinchonidine ap-pears to hinder the rotation around the C40–C9 and C9–C8 bonds and tofavour only a narrow range of the conformational space of the molecule. Interms of the behaviour of CD and CN molecules in solution, 2D NMR

238 | Catalysis, 2010, 22, 144–278

experiments indicate a somewhat more restricted rotational conformationspace for cinchonine than for cinchonidine.462 Protonation of cinchonidinealso significantly restricts its rotational conformation space.417

7.3.2 Conformational behaviour of cinchona derivatives. Detailed NMRanalysis and ab initio quantum chemical calculations were performed on10,20,30,40,10,11-hexahydroderivatives of CD.409 The rotation around theC40–C9 and C9–C8 bonds led to conformers of close energies, providingevidence on the possible presence of other stable conformers in the solutionof these cinchonidine derivatives.

The conformational analysis of the synthetic chiral modifier O-phenyl-cinchonidine in vacuum has been performed at semi-empirical level and atDFT level with a medium-size basis set for energetics related to the parentalkaloid cinchonidine.421 The O-phenyl-cinchonidine behaves similarly tocinchonidine and owns the same main stable conformers as mentionedabove in vacuum and in CH2Cl2 and CCl4 solvents. Based on combinedtheoretical–experimental results, the open(3) appears to be the mostpopulated in these solvents, but indication was found that an excesscl(2) conformer has to be also expected in CD2Cl2 in comparison to CD.The authors suggest that the sterical constraints imposed by insertionof O-phenyl at the C9 position shows its effect when the substituted CDadsorbs 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 (seeFig. 6). b-ICN was investigated in a recent study and its conformationalanalysis was performed. The results confirmed that the numerous con-formational changes possible for CD and CN are reduced to a single degreeof freedom, namely rotation around C(40)-C(9).463

Fig. 48 Conformational analysis of cinchonidine. The calculated energy map has been ob-tained by changing the torsion angles phi ((C30)–(C40)–(C9)–(C8)) and psi ((C40)–(C9)–(C8)–(C7)). The contours are given in steps 0.5 kcal/mol. (Reproduced from ref. 461 withpermission)

Catalysis, 2010, 22, 144–278 | 239

Recently detailed NMR, DFT, and X-ray investigation of some cinchonaalkaloid O-ethers related to the determination of their structure and con-formations was published.391

7.3.3 Other calculations. Conformational analysis of syntheticmodifiers, such as (R)-2-1-pyrrolidinyl)-1-1-naphthyl)ethanol, (R)-2-1-pyrrolidinyl)-1-2-naphthyl)ethanol, and (R)-2-1-pyrrolidinyl)-1-1-8-methyl-naphthyl)]ethanol was also performed. Open–(3) conformers appeared to bethe most stable. Minimum energy structures of the pro-(R) and pro-(S)interaction complexes between (R)-2-1-pyrrolidinyl)-1-1-naphthyl)ethanoland trans ethyl pyruvate were also performed. Good quantitative agreementbetween calculated and experimental ee values has been found for theenantioselective hydrogenation of EtPy over Pt catalyst chirally modified bysynthetic pyrrolidinyl– naphthyl–ethanol modifiers, assuming that EtPyexists in the trans conformation in the adsorbed enantio-differentiatingcomplex. The destabilising repulsive interaction between EtPy and the an-choring aromatic moiety within the pro-(S) complex has been identified tobe important for ED.85

7.4 Substrate-Pt interaction

The adsorption of ketones on transition metals has been the topic of variousstudies.464–466 In general, ketones adsorb on transition metal surfaces via twodifferent bonding mode: as Z1(O) in an end-on adsorption configuration inwhich 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-groupp-bonded to the metal and the CQO moiety lying parallel to the surface.

The bonding interaction between an adsorbate and a surface is a verycomplex process.467 To perform first principle calculation on the adsorptionof substrate molecule97 on a reasonably large (about 20–40 atom) Pt (or Pd)surface or cluster has become feasible only recently, however these resultsshould be handle cautiously. For example a drawback of using metalclusters of this size is that the Pt cluster is strongly paramagnetic (high spinstates) in the result of the computation,98 while experimentally it is notmagnetic.

The interaction of various ketones with Pt surface was investigated indetails.97 Fig. 49 shows the adsorption geometries of EtPy for both the cisand in the trans conformations. The cis Z2 adsorption appeared to be themost stable one (see Fig. 49a) In the Z1 adsorption mode only the transconformation showed an energy minimum (see Fig. 49c), whereas the cisconformer was not stable when Z1 adsorbed. When adsorbed Z2 the maininteraction the keto-carbonyl moiety interacts with the metal. Once thepreferred keto-carbonyl adsorption had taken place, the ester group inter-acts only weekly with the Pt surface.

The adsorption EBF and its derivatives onto Pt surface was also modelledand calculated.468 The results showed that the introduction of two o-sub-stituents into the aromatic ring completely eliminated the reactivity of theketone. 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

modes where the keto-carbonyl group is bound to the metal in Z2(C,O) modeinvolved in the hydrogenation reaction.

In a recent study the interaction of PPD with Pt surface was investigatedby DRIFT spectroscopic method and DFT calculations.469 Seven differentadsorption forms were suggested as shown in Fig. 50. The calculated ad-sorption energies are given in Table 19. DFT calculations demonstrated thatZ1-(O2) configuration is the most stable end-on adsorption mode of PPD.Tilted p-bonded adsorption mode of cinchonidine was revealed on theplatinum catalyst at higher concentration of CD.

7.5 Modifier-Pt interaction

The interaction of CD with Pt(111) both in ultrahigh vacuum (UHV) and inethanol solvent has been studied using molecular dynamics (MD) simu-lation. In UHV at low coverage (0.0125 molecules/Pt atom) and 298.15Kthe CD was found to adsorb with the quinoline ring oriented largely parallel(a=61) to the surface.441 Cinchonidine surface attachment was found to bethrough both p bonding of the aromatic group and adsorption of the CQCdouble bond of the vinyl group. The interactions between ethanol solutionsof CD (0.129 and 1.035M) and the platinum surface were also simulated.For the less concentrated solution (0.129M) two different equilibriumconformations were found, one in which only part of the quinoline is at-tached to the surface, and another slightly more stable conformation. In thelatter one the quinoline group is adsorbed parallel to the platinum surface.

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)

Catalysis, 2010, 22, 144–278 | 241

It was also observed that that CD conformation at the surface was affectedboth by the ethanol solvent and the CD-CD interactions and their com-petition for surface sites.

The conformations of CD adsorbed on a Pt(111) surface were also in-vestigated.442 Eight conformationally different adsorption states due todifferent degrees of rotation around the t1 and t2 degrees of freedom wereidentified. The possible role of these conformations in the formation ofchiral surface sites relevant to enantioselective hydrogenation was also in-vestigated. The comparison of the conformational behaviour of CD in so-lution and on Pt has revealed the effect of the metal surface on the internalmobility of the alkaloid. In the study the role of the adsorbed op(3) con-former, the observed conformational flexibility on the Pt surface revealedthe 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

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 (kJmol� 1)

Z1(O1) � 19a

Z1(O2) � 36

pseudo- n1(O1) � 55

di- n1(O1, O2) � 3a

Z2(C1, O1) � 88b

Z2(C2, O2) � 38a

Z3(C1, O1, O2) � 59

a Taken from ref. 470. b Phenyl ring only partly adsorbed.

242 | Catalysis, 2010, 22, 144–278

conformational equilibria on the surface due to their stability and areidentified as precursors of the less stable, but probably more active, openconformers. Although the open and closed conformers are closely related tothe correspondent ones found in solution, surface species that are also ad-sorbed via quinuclidine moiety have been identified also as possible forms ofmetal–modifier interaction and can be involved in ED.

In a recent study the possible forms of adsorbed CD was given as shownin Fig. 51.469 Density functional theory (DFT) at the B3LYP/T(ON)DZPlevel was used to model one-to-one reactant-modifier interactions relevantto the enantioselective hydrogenation of PPD and MP over platinumcatalysts In an other study205 DFT calculations revealed that protonatedcinchonidine and 10,11-dihydrocinchonidine are more stable on Pt whenadopting the so-called QA-Open(4) conformation rather than the Open-(3)conformation. Thus, the QA-Open conformations may have some role inthe enantioselective hydrogenation over modified Pt catalysts. The results ofthese calculations are shown in Fig. 52.

8. Reaction mechanisms and related calculations

8.1 Introduction

In the first approaches related to the enantioselective hydrogenation ofactivated ketones over Pt-cinchona catalytic system mechanistic views de-veloped earlier for Ni-tartaric acid catalyst system.471 It also means that thefirst models were proposed without any solid knowledge about the reactionmechanism, i.e., the proposed reaction mechanism and schemes were basedon pure ‘‘presumption’’ related to the knowledge accumulated in studiesover the Ni-tartaric acid system.

However, careful analysis of these two enantioselective hydrogenationreactions shows definite differences as follows: (i) mode of introduction ofthe modifier, (ii) amount of modifiers, (iii) reaction rate, and (iv) reactiontemperature

In Ni/tartrate system the catalyst requires pre-modification under con-ditions different from those used in the hydrogenation reaction. Contraryto that the Pt/cinchona system the introduction of the chiral modifier

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 withpermission)

Catalysis, 2010, 22, 144–278 | 243

into the reaction mixture during racemic hydrogenation instantaneouslyinduces ED.

In the Pt/cinchona system the substrate/modifier ratio is very high (in caseof KPL it is 276 000), but the optimum substrate/modifier ratio stronglydepends on the type of substrate. In Ni/tartrate system this value is severalorder lower.

In the Pt/cinchona system as it has already been discussed the cinchonaalkaloid induces not only enantio-differentiation but a well-pronounced rateacceleration. Contrary to that in the Ni/tartrate system the modifier de-creases the reactions rate. It has to be emphasized that the rate accelerationeffect has also been observed in homogeneous catalytic reactions in thepresence of cinchona alkaloids.353

In Ni/tartrate system the reaction takes place at moderate temperatureabove 60 1C, while the enantioselective hydrogenation of prochiral ketonesrequires low temperature around 0–10 1C. High temperatures above 40 1Care not favourable to get high ee values.

As it has been mentioned in a recent publication301 ‘‘two types ofmechanisms-modified catalyst,58,265,455,472 and shielding effect74 have beenproposed’’. Unfortunately, not all of the authors consider this way. Thosewho accepted the modified catalyst model (we shall call it ‘‘surface

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

244 | Catalysis, 2010, 22, 144–278

modification’’ model) did an enormous effort using various spectroscopicmethods and computational tools aimed to demonstrate that in fact onlyone reaction scheme is working, i.e., both in aprotic and protic reactionmedia the enantioselective hydrogenation reaction proceeds via the proto-nated form of cinchonidine and the surface entities responsible for theenantio-differentiation step has a N–Hþ–O bond.53 It also means that allkey events, i.e., the rate acceleration and enantio-differentiation are ex-clusively surface related phenomena.

Of course, the surface of Pt has a crucial role in this reaction. First or allPt 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 iswhat of these ‘‘surface induced interactions’’ shall have a direct contributioninto the rate acceleration and enantio-differentiation steps.

8.2 The ‘‘surface modification’’ model

The surface modification concept was first suggested in early nineties, whenthe number of publications in this area was very scarce. The distinctionbetween ‘‘modified’’ and ‘‘unmodified’’ sites over platinum was done byBlaser and coworkers.63 These terms have been widely accepted and used bythose who believe in the ‘‘surface modification’’ concept. The above dis-tinction was formulated into a kinetic equation describing the ‘‘ligand ac-celeration’’ phenomenon.58 It has to be mentioned that in the above studyno mechanistic views were given just a very simple reaction scheme. Ac-cording 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 concentrationof the substrate.

The first mechanistic view or scheme was given by Wells and coworkers inearly nineties (‘‘template model’’).65,228 According to this the enantio-dif-ferentiating sites are created by an ‘‘ordered layer of the alkaloid’’ with theformation of a ‘‘chiral pocket’’, i.e., a free room between adsorbed chiralentities, where the enantio-differentiation can take place. However, theirconcept was not supported by surface spectroscopic methods446 and theoriginal idea was withdrawn45 very soon and a new idea based on the in-volvement of the half-hydrogenated form of a-keto ester in the enantio-dif-ferentiation step was proposed by the same research group.472 It has to beemphasized again that this new idea was suggested without any experi-mental prove or evidence.

The fact that in acetic acid the enantioselective hydrogenation of alkylpyruvates takes place with higher rates and higher enantioselectivity than inaprotic solvents the original idea given by Wells and coworkers was furtherextended to the involvement of the protonated form of cinchonidine.49 Inthis model the quinuclidine nitrogen atom is protonated and the substrate isstill in its original state, maintaining the double bond character of thecarbonyl group. Later on this model was accepted as a general one even inthe absence of acids.53 Baiker and coworkers have published severalexperimental85,97 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 isspontaneously ionized with the formation protonated form of alkaloid.There is a general view that the hydrogen-bonding interactions can triggerfor rate acceleration.359,456

8.3 ‘‘Shielding effect model’’83

8.3.1 The principle of chemical shielding.183 The basis for this approachis the shielding effect (SE) known in organic chemistry. If a prochiral moietyis preferentially shielded its further reaction can take place only from itsunshielded site resulting in an ED. A chiral template molecule can induceSE in a similar way, i.e. it preferentially interacts with one of the prochiralsites of the substrate leaving the unshielded site free for the reaction.

Intramolecular steric shielding of an a–keto ester moiety has been ob-served resulting in enantio-differentiation in the hydrogenation of thea-keto group.473 ED was observed only in the presence of large aromaticsubstituent, such as naphthyl, and it was completely lost if it was substitutedfor a phenyl one. Based on this finding the ED was attributed to the SEinduced by the large aromatic moiety. Similar phenomena was also de-scribed for the hydrogenation of an a,b- unsaturated ester moiety.474 Theabove two examples are shown in Fig. 53. Additional examples for chemicalshielding 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 fea-ture, namely a conjugated double bond system. This feature is also char-acteristic for most of the substrates what can enantio-selectively behydrogenated 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 (derivativesof 2-1-pyrrolidinyl)-1-naphthyl)ethanol) the ED was completely lost if thenaphthyl ring was replaced by phenyl or pyridyl one.60 It should also bementioned that in the hydrogenation of a-keto esters over CD-Pt/Al2O3

R = Me, Et, Ph

OAc

O O

OR

O

CO2Me

O

CH2

OO

OAc

O O

R

CO2Me

Fig. 53 Intermolecular chemical shielding in the involvement of a-keto ester and a,b-un-saturated 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 waspartially or fully hydrogenated.64

If the key p–p interactions given in Fig. 53 are compared with resultsobtained in the enantioselective heterogeneous hydrogenation experimentsusing cinchona-Pt systems the following very important elements of simi-larity can be found: (i) ED can only be observed in the presence of largearomatic shielding groups; (ii) the reactive prochiral group (both the ketocarbonyl and the CQC double bond) is activated by an electron with-drawing carbonyl group; (iii) the prochiral keto carbonyl group in most ofthe 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 aromaticsubstituents in the modifier and a conjugated double bond system in thesubstrate should play an important role to induce ED in these asymmetrichydrogenation reactions, i.e. these are the key elements responsible for thesubstrate specificity.

It has to be added that the shielding effect model suggest that substratemodifier interactions responsible for the ED take place in the liquid phaseand not on the Pt sites. The term substrate-modifier interaction in liquidphase was also mentioned by other authors. ‘‘The substrate-modifierinteraction exists, according to circular dichroism, in solution, probably inthe form of aggregates’’.68 Similar views were given in another study.221 Ithas also been suggested that for some substrates, the solvent is involved inthe substrate–modifier interaction.478 It has been suggested that the OHgroup of the alkaloid should be involved in the substrate–modifier inter-action 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 ex-perimental evidence, most mechanistic ideas are based on assumptions and(at best) calculations. In most cases, the models assume two interactionsbetween 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 re-pulsive interaction that directs the adsorption of the ketone on Pt.52,483

In the following section we shall follow the above consideration, i.e., weshall distinguish electrophilic and nucleophilic interactions between thesubstrate and the modifier. It has to be emphasized that all existing modelspostulate 1:1 type interactions between CD and the substrate. Bartok et al.451

and Augustine et al.371 proposed that not only the quinuclidine N, but alsothe OH function of CD would be involved in the interactions. However thisview can be questioned as neither the methylation nor the removal of the OHgroup in CD hinders the enantio-selection in the hydrogenation of EtPy.57

The first attempt aimed to elucidate the character of substrate-modifierinteractions was done in an early study458 related to the investigation ofinteraction between MePy with NH3 and NH4

þ . In this study the ammoniapart represented the quinuclidine nitrogen of CD. The results indicated thatMePy can interact with both NH3 and NH4

þ and the electrophilic inter-action is more favourable than the nucleophilic one. However, the

Catalysis, 2010, 22, 144–278 | 247

nucleophilic interaction provided interesting result, namely the reactionpocket of MePy is located between the two carbonyl groups. This findingindicated that under condition of enantioselective hydrogenation bothcarbonyl groups are activated. Consequently, if the enantioselective hy-drogenation of EtPy is performed in methanol trans-esterification reactioncan take place. This has been proved experimentally.307

In the next approaches two types of interactions were modelled, namely theBaiker’s group focused on the electrophilic interaction between the half-hydrogenated substrate and CD.53,74,457,484 In our group, based on kineticresults over cinchonidine-Pt/Al2O3, and the proposed ‘‘shielding effect’’ modelthe nucleophilic interaction between the modifier and substrate was favored.69

8.4.1 Electrophilic interaction. In the first theoretical study related to thesubstrate-modifier interaction the formation of a week complex betweenprotonated CD and MePy methyl pyruvate was investigated.74 In this studymolecular mechanics and AM1 semi-empirical methods were used. The cal-culated surface complex bifurcated electrophilic interaction between the pro-tonated quinuclidine and the keto carbonyl group was considered. The resultsrevealed that adsorption of the complex leading to (R)-methyl lactate is morefavorable than that of the corresponding system yielding (S)-methyl lactate.

In another study ab initio calculations were used to study the interactionbetween protonated amines (NH3, (CH3)3N and quinuclidine) and methylpyruvate (MP), as well as between protonated MP and these amines.481

Based on results it has been suggested that interactions mediated by aproton between the MP and the alkaloid are the main driving force leadingto enantiodifferentiation in the hydrogenation of a-ketoesters. MP interactswith protonated amines preferentially in the s-cis conformation, with aproton making two hydrogen bonds to the carbonyl oxygens. This protonmay be transferred to MP, forming a new complex in which the amines arebonded to the protonated MP. The last complex is approximately10 kcalmol� 1 less stable than the first one. However, this energy differencedecreases to approximately 5 kcalmol� 1 when solvent effects are included.

Characteristic feature of these models is that both in the absence andpresence of acid in the key reaction intermediate the electrophilic interactionprevails with the involvement of N–H–O or N–Hþ–O bonds. This interactioncan 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 itsquinoline ring parallel to the Pt surface as it has already been discussed inSection 6. Several adsorbed conformations of CD were calculated.442

Sequence of events on the Pt surface related to the protonation of qui-nuclidine nitrogen is shown in Fig. 55, 56, while the calculated substrate-modifier 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 em-phasized. The overall route for proton transfer from protonated CD toadsorbed substrate supported by DFT calculations and in situ ATR-IRspectroscopy is shown in Fig. 55. It has to be added that in the recentpublication of this group it was found that the most stable intermediatecomplex forms without adsorption of the substrate.103

We have serious objection against the exclusiveness of the electrophilicinteraction in the Orito’s reaction. Assuming the scheme given in Fig. 55 onewould suggest that not only the quinuclidine nitrogen of the cinchona al-kaloid can be involved in the transformation of atomic hydrogen formed onthe Pt site to a protonated nitrogen base. All tertiary nitrogen bases shouldhave similar ability. Consequently, the hydrogen transfer in the presence ofan 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 systemand the subsequent transfer of the H to the adsorbed ketone. (Reproduced from ref. 372 withpermission)

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

Catalysis, 2010, 22, 144–278 | 249

the simultaneous addition of cinchona alkaloids and ATAs should result ina decrease in the ee values. However, our results discussed in Section 5.7.2showed 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 condensedaromatic ring parallel to the Pt surface, (ii) stabilization of the modifier in itsopen form, (iii) adsorption of the substrate via its both carbonyl bonds inre-phase, (iv) formation of a hydrogen bond between protonated alkaloidand the substrate and (v) transfer of the proton to the substrate. The con-densed aromatic ring is often called as anchoring site.

As it has already be mentioned earlier the chiral C8 and C9 carbon atomsof the alkaloid play a vital role in the enantio-differentiation, i.e. theirconformation determines the position of the ‘‘chiral pocket’’ located inthe neighbourhood of quinuclidine nitrogen. In addition the mode ofadsorption of the chiral modifier (flat or tilted) has a decisive role inthe ED step as it has been shown in the series of O-substituted derivatives ofCD.435

In a recent study the role of the modifier structure in the reactant-modifierinteractions relevant to the heterogeneous enantioselective hydrogenationof PPD and MePy) was studied using DFT calculations.205 Two protonatedmodifiers, CD and MeOCD, in different conformations were considered.So-called bifurcated and cyclic hydrogen-bonded reactant-modifier inter-action modes were investigated. The results showed that only the bifurcatedreactant-modifier(Open3) complexes were found to be relevant in the de-termination of enantioselectivity. Analysis of the orbital stabilization im-plies a notable decrease in the enantiomeric excess of the mainhydrogenation product of PPD when CD is replaced with MeOCD. On theother hand, according to the theoretical calculations the hydrogenation ofMP over modified Pt is expected to yield an equal ee values in the presence

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)

250 | Catalysis, 2010, 22, 144–278

of both modifiers. DFT calculations revealed that protonated CD andDHCD are more stable on Pt when adopting the so-called QA-Open (4)conformation rather than the Open (3) conformation. These conformationsare shown in Fig. 58. Thus, the QA-Open conformations may have somerole in the enantioselective hydrogenation over modified Pt catalysts. TheQA-Open (4) conformation of a modifier is adsorbed on the surface via bothits quinoline and quinuclidine moieties, and a reactant may interact sim-ultaneously with the protonated quinuclidine nitrogen and the functionalgroup at the C(9) position of the modifier.

The interaction between KPL (Pro-(R) conformation) and adsorbedo-PyOCD over Pt surface was also modelled.372 In this case both the qui-nuclidine and pyridine moieties in o-PyOCD were protonated. At the end ofthe simulation the hydrogen was transferred to the keto-carbonyl group ofketopantolactone, therefore forming a semi-hydrogenated surface species,while protonated o-pyridyl group coordinated to the ester carbonyl groupas shown in Fig. 59.

Modelling studies revealed also that there is no mode of docking of anylow energy conformation of epiquinidine with pyruvate ester that could leadto selective enantioface adsorption of the latter.485

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

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

Catalysis, 2010, 22, 144–278 | 251

Recently a new idea was suggested assuming a H-bond between thequinuclidine N of CD and the ester carbonyl or trifluoromethyl group of thesubstrate and a second, monodentate or bidentate H-bond involving two orone aromatic H atoms of the modifier at 50- and 60-positions and the O atomof the ketocarbonyl group.72,316,483

In addition, it was suggested that similar interactions might exist for allactivated ketones. General principles of this type of interactions are shownin Fig. 60.

This type of interactions has also been proposed in recent studies by theBartok’s goup384,486 (see Fig. 61). However, in this case the character ofsubstrate modifier interaction is nucleophilic.

However, these models strongly contradicts to experimental findings asring-substituted cinchona derivatives, such as QN and QD containing amethoxy group in the 60-position, are highly effective modifiers in the hy-drogenation of activated ketones (see Section 3.1). Consequently, theinteraction of the substrate molecule with the proton of the quinoline ring isnot a prerequisite for enantio-differentiation.

In addition, the proposed interaction of the keto group with the aromatichydrogen (see Fig. 60 cannot give any reasonable explanation for the ac-tivation of the keto group resulting in rate acceleration.

Based on ESI-MS spectra of EtPy, DHCD and EtPy-DHCD mixturesinteresting equilibria (see Fig. 62) were suggested by Bartok et al.195 In thisrespect it interesting to note that the semi-ketal formed between the CD andEtPy was also evidenced in another study.109

Fig. 60 Two-point H-bonding model suggested by McBreen et al. A: interaction betweenprotonated 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 withpermission)

252 | Catalysis, 2010, 22, 144–278

8.4.2 Nucleophilic interaction. Nucleophilic interaction betweenthe substrate and the modifier has been suggested by differentauthors.83,84,369,412,458,488 In an early mechanistic view67 it was proposedthat ‘‘the hydrogenation of pyruvate over either modified or unmodifiedplatinum takes place on the more coordinatively unsaturated corneratoms or adatoms on the platinum surface’’. In their further study369

it was suggested that the substrate-modifier complex (association) ‘‘in-corporates some interaction between the quinuclidine nitrogen and theketone carbonyl group of the pyruvate. One such interaction is betweenthe electron pair on the nitrogen and the electron deficient carbon atomof the carbonyl group, which, as discussed above would account for theobserved increase in hydrogenation rate in these reactions’’. The involve-ment of coordinatively unsaturated platinum was also suggested by theBartok’s group.486,487

The ‘‘shielding effect’’ model is also based on the nucleophilic inter-actions. The key issue of this model is the involvement of closed conformerof CD in the substrate modifier interaction. The character of these inter-actions will be discussed in the next section. Upon investigating enantio-selective hydrogenation of KPL in the presence of b-ICN Bartok ancoworkers suggested two possible forms of surface complex representingeither electrophilic or nucleophilic interaction as shown in the next scheme(see Fig. 63.) The inversion of the ee was attributed to the change of thereaction mechanism from nucleophilic to electrophilic one.

It was also suggested that in the hydrogenation of activated ketones in thepresence of cinchona-Pt catalysts proceeds ‘‘through nucleophilic additionof a cinchona alkaloid to the ketone to form a zwitterionic adduct, which isthen hydrogenolyzed with inversion of configuration. The enantioselectivityof the reaction is determined by the relative stabilities of the diastereomericadducts adsorbed on platinum’’.412 However, this mechanism has beenruled out as it was pointed out that this approach does not take into accountsteric hindrance against the interaction of the amine modifier with cyclicketones and further critical point is the regioselectivity of the hydro-genolysis of the hypothetical zwitterionic intermediate.413

+

NH OH

NC

COOEt

OHMe

+

N

HOH

NH

C

COEt

Me

O

O

NO

HC COOEt

Me

OH

N

H

+

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

Catalysis, 2010, 22, 144–278 | 253

When the inversion of enantioselectivity in the presence of b-ICNs wasinvestigated the formation of different nucleophilic adducts was proposed asshown in Fig. 64.

8.4.3 Character of interactions in the ‘‘shielding effect’’ model. Cinchonaalkaloids (CA) have two rotational axises, which allow to rotate either thequinoline ring around the C(1) 0-C(9) axis or the quinuclidine ring aroundthe C(9)-C(8) axis (see Sections 2.1 and 3.1). Molecular mechanics andab initio calculations performed by different groups that in liquid phase CAcan exist at least in three different stable forms. We suggest that the closedform of the modifier is required both for the RA and the ED. Only theclosed form of CA can provide the cooperative required for ED.

The possible arrangements of the substrate and the modifier in the shiel-ded complex are shown in Fig. 65. In the hydrogenation of pyruvate estersthe complex shown in Fig. 65A would result in the expected (R)-lactateester, while the complex given in Fig. 65B would give the corresponding(S)-product. The major difference between the (R) and (S) complexes is themode of interaction between the lone pair of electrons of the quinuclidine

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

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

254 | Catalysis, 2010, 22, 144–278

nitrogen and the keto carbonyl group. In complex (R), the ‘directionality’489

of the nucleophilic attack by quinuclidine nitrogen towards the keto car-bonyl group is very favourable to increase the reactivity of the keto carbonylgroup because the electron- rich quinuclidine nitrogen and the keto-oxygenof the substrate are on the opposite sides of the keto carbon atom. Accordingto 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 relativeposition of the quinuclidine N1, pyruvate C2 and O3 atoms, i.e., by directN1–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 theabove perturbation leads to a pronounced rate increase both in the hy-drogenation reaction and the formation of by-products, such as semi-ketal,transesterification and deuterium exchange products84,289,307 Thus, incomplex (R), the favourable directionality promotes the perturbation of theketo carbonyl group, resulting in the observed RA. Contrary to that incomplex (S), due to the misalignment of the interacting groups, i.e., due tothe 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 interactionssuggest that the quinoline ring is involved in the adsorption of the modifierto the metal.45,82,214 Contrary to that we suggest that the quinoline ring isinvolved in the p-p interaction with the substrate via ‘‘p-p stacking’’.83,84 Wesuggest that the RA is a cooperative effect with the involvement of both thequinuclidine nitrogen and the quinoline ring (Fig. 66).

Monte-Carlo simulation method was used to investigate the interactionof the [methyl pyruvate– CD]closed complex with Pt (111). surface. The resultshown in Fig. 67 indicates that the shielded complex retains its entity evenafter adsorption.

The above figure gives a good presentation of the SE provided by thelarge aromatic moiety. With respect to the explanation related to the use ofsmall Pt colloids we should like to refer to results of our calculations (MonteCarlo simulation). These results given in Fig. 68 shows that the closed[substrate – modifier] complex can be accommodated at the Pt(111) faceeven of a small Pt nanocluster.244

When the ‘‘shielding effect’’ model was proposed it was also mentionedthat in the enantioselective hydrogenation of activated ketones cinchonaalkaloids behave like an enzyme.93 This view has recently been emphasizedwithout reference to our original idea.103

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

Catalysis, 2010, 22, 144–278 | 255

8.5 Pros and contras related to existing models

8.5.1 ‘‘Surface modification’’ model. The surface modification modelhas been strongly altered since the introduction of the first concept, i.e., thedivision of Pt surface into modified and unmodified sites. According to thegeneralized view the key issues in this model are as follows: (i) adsorption ofCD in its open (3) conformation via its quinoline ring parallel to the Ptsurface, (ii) conformational changes at the Pt surface to form quinuclidinebonded CD, (iii) formation of protonated quinuclidine moiety, (iv) trans-formation of the proton to the substrate to form half hydrogenated surfacespecies, and (v) direct addition of the second hydrogen form the Pt surfaceto get the chiral keto-alcohol (see Fig. 57).

However, based on the discussion in this contribution we can emphasizethat this model was not able to give appropriate answer to the following

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

256 | Catalysis, 2010, 22, 144–278

important observations: (i) rate acceleration (see Section 5.5.1), (ii) theappearance of the initial transient period in the conversion-selectivity de-pendencies (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 Section6.1.1) in the presence of AcOH, (vii) high enantioselectivities over very smallPt nano-colloids (see Section 4.2.2).

Even if scheme shown in Fig. 55 is valid a simple question can beraised? Why the adsorption and subsequent strong distortion of thealkaloid at Pt surface is needed to pick up a proton from the Pt surface if itcould also be done without any preadsorption via quinuclidine N-Ptinteraction. Would not be it more logical and energetically morefavourable?

In this respect we have to address literature data related to the proto-nation of pyridine and its analogs observed under high vacuum and lowtemperatures.491,492 These references were cited in several papers as a directproof related to surface reactions given in Fig. 55. However, it has to bepointed out that none of the authors referred to experimental conditionsused in these earlier studies. It has to be emphasized that in refs. 492,493results obtained under ultra-high vacuum were presented. The key experi-ments were performed over Pt(110) surface under base pressure between5� 10� 11 and 1� 10� 10 Torr and the temperature was kept between100–180K. The intensities of the 3450 cm� 1 EELS peak characteristic ofcation formation showed strong temperature dependence and above 180Kit 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 theimportant issues that these alkaloids are used by organic chemist for manyyears to induce ED or chiral separation.

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

Catalysis, 2010, 22, 144–278 | 257

8.5.2 ‘‘Shielding effect’’ model (SEM). The shielding effect model canexplain the following experimental findings:74,83 (i) substrate specificity), (ii)inversion of enantioselectivity for enantiopairs CD-CN, QN-QD, (iii) rateacceleration, (iv) the MI character of the ee-conversion dependencies (v) theloss of ee in case of replacing the quinoline ring for pyridyl or phenyl, (vi)formation of transesterification and deuterium exchange products; (vii) ef-fectiveness of very small Pt colloids, (viii) the role of achiral tertiaryamines, (viii) the need for of large aromatic moieties in cinchona alkaloidsto induce ED.

The reaction network derived form the shielding effect model providedkinetic equations what can describe the following kinetic events:83 (i) rateacceleration, (ii) increase the reaction rate at the initial period of the re-action,66 (iii) the the MI character of the ee-conversion dependencies.

The strongest conflict with ‘‘shielding effect’’ model is the finding thatICNs can also induce enantio-differentiation. However, in this respect theanomalous behaviour of these rigid alkaloids361 has to be mentioned. Thebehaviour of these alkaloids needs further elucidation and probably the useof more pure alkaloids. In this respect the lack of rate acceleration in thepresence of a-ICN and the inversion of ee in the presence of b-ICN have tobe mentioned. Another unclear issue is that shielding effect model requiresnucleophilic interaction between the substrate and the modifier, while in thepresence of AcOH electrophilic interactions interaction prevails. Althoughin this relation recent NMR results has to be emphasized, what clearly in-dicated that in case of KPL even in the presence of AcOH the nucleophilicsubstrate-modifier adduct can be formed.419

Finally, we have to admit that the ‘‘shielding effect’’ model was notsupported by the scientific community in the field of heterogeneous cata-lysis. This fact can be attributed to the deficiency of the model. However, itcannot be excluded that due to the strong influence of those who favour the‘‘surface modification’’ model, the scientific community just simple followedthe main stream without any criticism.

9. Conclusions

In this review an attempt was done to give a retrospective overview aboutmethods, approaches and results obtained in the last three decades in thearea of enantioselective hydrogenation of activated ketones. Both practicaland theoretical aspects were discussed. Characteristic feature of this reviewis that the term ‘‘chirally modified surface’’ was not really used.

Although tremendous effort has been done so far to elucidate the pecu-liarities of this reaction there are still several open questions related to thesubstrate-modifier and substrate-modifier-platinum interactions involved inthe enantio-differentiation step.

It seems to us that starting from the beginning of early nineties there is apermanent desire to demonstrate and prove that in the presence of cin-chona-Pt catalyst system all interactions responsible for ED take place onthe Pt surface. In addition, last years the mainstream concentrated to provethat the protonated quinuclidine moiety is involved in the first step ofhydrogen 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 inaprotic and protic solvents is the same. We consider that this mechanisticview is too general and in addition it is artificial and far-fetched. This viewneglects series of experimental evidences obtained by different researchgroups. 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 highenantioselectivities were obtained upon using very small Pt colloids. In noneof the reviews published so far by catalytic scientists this anomalous findingswere not really discussed. However, in a recent review written by organicchemists it was emphasized that upon using Pt colloids in enantioselectivehydrogenation of MePy in the presence of CD ‘‘the smallest Pt clusters gavethe best results despite having no flat surface large enough for the adsorptionof cinchonidine’’.493

Addition of quinoline. In this respect one of the earlier results has also tobe mentioned.345 In this study it was shown that the addition of quinoline tothe reaction mixture at very low concentration (0.1 g/L) increased both therate and the ee values. The authors attributed this observation to some sortof base effect. Unfortunately, due to the dominance of the general view, i.e.the ‘‘formation of chirally modified surfaces’’ this result has completely beenforgotten and in the last eighteen years it was only very seldom cited. In thisrespect 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 negativeeffect either on the reaction rate and the ee values. Based on these findingsthe scheme shown in Fig. 69 has been suggested.

Fig. 69 shows four different situations described in ref. 402. A representsthe racemic hydrogenation in the absence of any additive, where due to thepoisoning effects of by-products the rate is controlled by free Pt sites left. Bcorresponds to the situation when quinoline is added prior to the additionof hydrogen. In this case the poisoning effect of by-products decreased re-sulting in a rate increase in racemic hydrogenation. C represents theenantioselective hydrogenation upon injecting CD, where the initial surfacecoverages 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 – conversiondependencies are evidenced. D corresponds to the situation when cincho-nidine is injected to surface B containing preadsorbed quinoline. In case D,due to the established competition between CD and quinoline the Pt sitesare covered by both CD and quinoline. However, the net results are un-expected, i.e., increased rate and increased ee. Consequently, these resultsmight indicate that the general view that ‘‘strongly bonded to the platinumCD via its quinoline ring’’ needs some corrections.

Spectroscopic results in liquid phase. Both NMR83,84,418,419 and CircularDichorism68,93 spectroscopic results clearly indicated that there is a complexformation between the substrates and cinchona alkaloids in the liquidphase. These facts were completely neglected by the scientific community. Inthis respect let us refer to a very recent results indicating that the nucleophiliccomplex between KPL and b-ICN can exist even in the presence of AcOH.The authors showed a nice dependence between the solution concentrationof the 1:1 substrate-modifier complex and the enantioselectivity as shown inFig. 70. In addition it was shown that there is an excellent correlation be-tween the ee values, the concentration of the 1:1 substrate/modifier complexand the amount of AcOH added as shown in Fig. 71.

It is known that in the enantioselective hydrogenation of KPL all polarsolvents have a negative effect.76 Results presented in ref. 419 (see Figs. 2and 3) definitely show the importance of the complex formation in the liquidphase. However, even in the light of these unambiguous evidences theauthors of this study were not brave enough as they made the followingremark: ‘‘we did not doubt the role of the protonated cinchona despite thefact that the spectroscopy data published previously, obtained under theconditions of the Orito reaction in toluene, are not totally convincing interms of protonation of the N atom of quinuclidine’’.

We believe that in the near future further high quality and unambiguousexperimental data will be obtained related to the character of substrate-modifier interactions as well to the formation of substrate-modifier complexat the Pt surface. We also hope that those who have different views on themechanism of Orito’s reaction in the future will get more wide open plat-forms to publish their results and ideas.

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

260 | Catalysis, 2010, 22, 144–278

Abbreviations used

AcOH acetic acidAS anchoring sitesATA achiral tertiary amineATR-IR attenuated total reflection infrared (spectroscopy)CD cinchonidineCI chiral inductionCN cinchonineDe diastereomeric excessDHCD 10,11-dihydrocinchonidineDHCN 10,11-dihydrocinchonineDRIFT diffuse reflectance FT infraredEBF ethyl benzoylformateECD enantioselectivity–conversion dependenciesED enantio-differentiationee enantiomeric excess (%)EOG diethyl 2-oxoglutarateEOPB ethyl-2-oxo-4-phenyl butirateEt ethylEtLa ethyl lactateEtpy ethyl pyruvateHHCD hexahydrocinchonidineHRTEM high-resolution transmission electron microscopyICN isocinchonineIR infrared spectroscopyITP initial transient periodKPL ketopantolactoneM metalM/S modifier–substrate molar ratioMBF methyl benzoylformateMe methylMeLa methyl lactate

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

Catalysis, 2010, 22, 144–278 | 261

MeOCD O-methyl-cinchonidineMeODHCD O-methyl-10,11-dihydrocinchonidineMePy methyl pyruvateMI monotonic increaseNED 1-naphthyl-1,2-ethanediolNLP non-linear phenomenaPADA pyruvaldehyde dimethyl acetalPh phenylPhOCD O-phenyl-cinchonidinePNEA pantoylnaphthylethylaminePPD 1-phenyl-1,2-propanedionePSC primary surface complexQ quinolineQD quinidineQN quinineQND quinuclidineRA rate accelerationRAIRS reflection absorption infrared spectroscopyRAIRS reflection adsorption infrared spectraRE rate enhancementSBA stabilized bimetallic alloySERS surface-enhanced Raman scatteringSTM scanning tunnelling microscopyTEA triethylamineTFA trifluoroacetic acidTFAP trifluroacetophenoneTMS trimethylsilaneTOF turnover frequency (h� 1)Y 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. Vank-

elecom 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.), Wiley-

VCH, 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 Enantio-

selective 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. Burgi 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.

55 B. Torok, B. Balazsik, M. Torok, Gy. Szoll +osi and M. Bartok, Ultrasonics

Sonochem., 2000, 7, 151.

56 M. Sutyinszki, K. Szori, K. Felfoldi and M. Bartok, 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. Schurch, 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. Wirth-

Tijani, 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. Mathe, T. Tarnai, K. Fodor, G. Toth, J. Kajtar, I. Kolossvary,

B. Herenyi and R. A. Sheldon, Tetrahedron: Asymmetry, 1995, 6, 2395.

69 J. L. Margitfalvi and M. Heged +us, J. Catal., 1995, 156, 175.

70 B. Torok, K. Felfoldi, G. Szakonyi and M. Bartok, Ultrasonics Sonochem.,

1997, 4, 301.

71 (a) P. Maki-Arvela, M. Kuzma, R. Sjoholm, R. Leino and T. Salmi, in:

Proceedings of the 9th International Symposium on Relations between Homo-

geneous 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.

75 J. L. Margitfalvi, M. Heged +us and E. Tfirst, Tetrahedron: Asymmetry, 1996, 7,

571.

76 T. Mallat, M. Bodmer and A. Baiker, Catal. Lett., 1997, 44, 95.

77 M. Schurch, 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. Bonnemann 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. Burgi, K. Borszeki, 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.

83 J. L. Margitfalvi, M. Heged +us and E. Tfirst, Stud. Surf. Sci. Catal., 1996,

101, 241.

84 J. L. Margitfalvi and M. Heged +us, J. Mol. Catal. A: Chem., 1996, 107, 281.

85 A. Vargas, T. Burgi 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. Burgi and A. Baiker, J. Phys. Chem. B, 2001, 105, 3187.

88 T. Burgi and A. Baiker, J. Am. Chem. Soc., 1998, 120, 12920.

89 D. Ferri, T. Burgi 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. Burgi 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. Talas, 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. Burgi 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. Kohler and J. S. Bradley, Langmuir, 1998, 14, 2730.

101 L. Michaelis and M. Menten, Biochem. Z., 1913, 49, 333.

102 E. Talas 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. Maki-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. Talas, L. Botz, J. Margitfalvi, O. Sticher and A. Baiker, J. Planar Chro-

matogr., 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. Csampai and T. Soos, 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. Muller, 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, Tetra-

hedron 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. Pays-

Bas., 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. Talas 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. Bartok, Curr. Org. Chem., 2006, 10, 1533.

196 M. Bartok, K. Felfoldi, Gy. Szoll +osi and T. Bartok, React. Kinet. Catal. Lett.,

1999, 68, 371.

197 M. Bartok, M. Sutyinszki, I. Bucsi, K. Felfoldi, Gy. Szoll +osi, F. Bartha and

T. Bartok, 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. Burgi, 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. Felfoldi, K. Balazsik and M. Bartok, J. Mol. Catal. A: Chem., 2003, 202,

163.

204 K. Balazsik, K. Szori, K. Felfoldi, B. Torok and M. Bartok, 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.

206 Sz. Cserenyi, K. Felfoldi, K. Balazsik, Gy. Szoll +osi, I. Bucsi and M. Bartok,

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. Burgi, 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. Schurch, 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. Muller, 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. Felfoldi, T. Varga, P. Forgo and M. Bartok, Catal. Lett., 2004, 97, 65.

217 K. Fodor, A. Tungler, T. Mathe, 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. Mathe, K. Fodor, R. A. Sheldon and P. Gallezot, J. Mol. Catal.

A: Chem., 1996, 108, 145.

219 G. Farkas, K. Fodor, A. Tungler, T. Mathe, G. Toth and R. A. Sheldon,

J. Mol., Catal. A: Chem., 1999, 138, 123.

220 G. Farkas,_ . Sıpos, A. Tungler, A. Sarkany 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. Sıpos, 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. Kajtar, T. Mathe, T. Toth, E. Fogassy and J. Petro, Catal.

Today, 1989, 5, 159.

224 E. Sıpos, A. Tungler and I. Bitter, J. Mol. Catal. A: Chem., 2003, 198, 167.

225 B. Minder, M. Schurch, 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.

230 M. Bartok, T. Bartok, G. Szoll +osi and K. Felfoldi, Catal. Lett., 1999,

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.

236 K. Sz +ori, Gy. Szoll +osi and M. Bartok, J. Catal., 2006, 244, 255.

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. Burgi and A. Baiker, Chem. Commun., 2001, 1172.

241 I. Bakos, S. Szabo, M. Bartok and E. Kalman, J. Electroanal. Chem., 2002,

532, 113.

242 X. Zuo, H. Liu and M. Liu, Tetrahedron Lett., 1998, 39, 1941.

243 H. Bonnemann and G. A. Braun, Chem. Eur. J., 1997, 3, 1200.

244 J. L. Margitfalvi, E. Talas, L. Yakhyaeva, E. Tfirst and I. Bertoti, in: Catalysis

of Organic Reactions, D.G. Morrell, (Ed.), Chem. Ind., Vol. 89, Dekker, New

York, 2003, p. 393.

245 J. U. Kohler 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. Bohmer and J. Wiehl, Stud. Surf. Sci. Catal., 1994, 84,

2021.

261 U. Bohmer, F. Franke, K. Morgenschweis, T. Bieber and W. Reschetilowski,

Catal. Today, 2000, 60, 167.

262 K. Balazsik, B. Torok, G. Szakonyi and M. Bartok, 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. Bartok, K. Balazsik, T. Bartok and Z. Kele, Catal. Lett., 2003,

87, 235.

268 K. Balazsik and M. Bartok, J. Mol. Catal. A: Chem., 2004, 219, 383.

269 W.-R. Huck, T. Burgi, T. Mallat and A. Baiker, J. Catal., 2003, 216, 276.

270 E. Schmidt, D. Ferri and A. Baiker, Langmuir, 2007, 23, 8087.

271 Gy. Szoll +osi, P. Forgo and M. Bartok, Chirality, 2003, 15, S82.

272 M. Bartok, P. T. Szabo, T. Bartok, G. Szollosi and K. Balazsik, Rapid

Commun. Mass Spectrom., 2001, 15, 65.

273 M. Bartok, K. Balazsik, Gy. Szoll +osi and T. Bartok, Ultrasonics Sonochem.,

1999, 5, 149.

274 J. L. Margitfalvi and E. Talas, Appl. Catal. A: Gen., 1999, 182, 65.

275 M. Bartok, Gy. Szoll +osi, K. Balazsik and T. Bartok, J. Mol. Catal., A: Chem.,

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. Schurch 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.

280 J. L. Margitfalvi and M. Heged +us, unpublished results.

281 K. Balazsik and M. Bartok, J. Catal., 2004, 224, 463.

282 B. Torok, K. Felfoldi, G. Szakonyi, K. Balazsik and M. Bartok, Catal. Lett.,

1998, 52, 81.

283 B. Torok, K. Balazsik, Gy. Szollosi, K. Felfoldi, I. Kun and M. Bartok,

Ultrasonics Sonochem., 1999, 6, 97.

284 K. Balazsik, B. Torok, K. Felfoldi and M. Bartok, Ultrasonics Sonochem.,

1999, 5, 149.

285 B. Toukoniitty, O. Roche, J.-P. Mikkola, E. Toukoniitty, F. Klingstedt, K.

Eranen, 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. Talas, L. Botz and A. Baiker, Stud. Surf. Sci.

Catal. (ICC Budapest), 1993, 75, 2471.

290 N. Kunzle, R. Hess, T. Mallat and A. Baiker, J. Catal., 1999, 186, 239.

291 N. Kunzle, T. Mallat and A. Baiker, Eur. Pat. Appl. 99112078.3.

292 N. Kunzle, J. W. Soler, T. Mallat and A. Baiker, J. Catal., 2002, 210, 66.

293 N. Kunzle, J. W. Soler and A. Baiker, Catal. Today, 2003, 79–80, 503.

294 E. Toukoniitty, P. Maki-Arvela, A. Kalantar Neyestanaki, T. Salmi, R. Sjo-

holm, R. Leino, E. Laine, P. J. Kooyman, T. Ollonqvist and J. Vayrynen,

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. Kunzle, T. Mallat and A. Baiker, Appl. Catal. A: Gen., 2003, 238, 251.

298 R. Wandeler, N. Kunzle, 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. Bohmer 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. Prusse, 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. Kunzle 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. Bartok, K. Balazsik, G. Szollosi and T. Bartok, J. Catal., 2002,

205, 168.

314 N. Bonalumi, T. Burgi 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. Burgi and A. Baiker, J. Catal., 2003, 215, 116.

319 T. Salmi, D.Yu. Murzin, J.-P. Mikkola, J. Warna, P. Maki-Arvela, E. Tou-

koniitty 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. Schurch, N. Kunzle, 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. Maki-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. Maki-Arvela, M. Kuzma, A. Villela, A. K. Neyestanaki,

T. Salmi, R. Sjoholm, 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. Torok, K. Felfoldi, K. Balazsik and M. Bartok, 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. Kunzle, A. Szabo, M. Schurch, G. Z. Wang, T. Mallat and A. Baiker,

Chem. Commun., 1998, 1377.

336 A. Szabo, N. Kunzle, 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. Szori, Gy. Szollosi and M. Bartok, Adv. Synth. Catal., 2006, 348, 515.

340 R. Hess, A. Vargas, T. Mallat, T. Burgi and A. Baiker, J. Catal., 2004, 222,

128.

341 K. Szori K, B. Torok and K. Felfoldi, M., in: ‘‘Catalysis in Organic Reactions’’

M. E. Ford (Ed.) Chemical Industries Vol. 82, Marcel Dekker, New York,

2001. p. 489

342 E. Talas, 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. Torok, K. Balazsik, G. Szollosi, K. Felfoldi and M. Bartok, Chirality, 1999,

11, 470.

345 H. U. Blaser and H. P. Jalett, Stud. Surf. Sci. Catal., 1993, 78, 139.

346 E. Tukonitty, P. Maki-Arvela, J. Warna 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. Szori, K. Felfoldi and M. Bartok, 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.

357 K. Balazsik, I. Bucsi, Sz. Cserenyi, Gy. Szoll +osi and M. Bartok, J. Mol. Catal.

A: Chem., 2008, 285, 84.

358 J. L. Margitfalvi, H. P. Jalett, E. Talas, A. Baiker and H. U. Blaser, Catal.

Lett., 1991, 10, 325.

359 A. Vargas, T. Burgi and A. Baiker, New J. Chem., 2002, 26, 807.

Catalysis, 2010, 22, 144–278 | 273

360 T. Varga, K. Felfoldi, P. Forgo and M. Bartok, J. Mol. Catal. A: Chem., 2004,

216, 181.

361 J. L. Margitfalvi and E. Talas, Appl. Catal. A: Gen., 2006, 301, 187.

362 R. Hess, T. Mallat and A. Baiker, J. Catal., 2003, 218, 453.

363 K. Felfoldi, T. Varga, P. Forgo and M. Bartok, 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. Talas, J. L. Margitfalvi and O. Egyed, J.Catal., 2009, 266, 191.

368 Gy. Szoll +osi, Sz. Cserenyi, F. Fulop and M. Bartok, 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. Muller 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. Konigsmann, 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. Balazsik, Sz. Cserenyi, Gy. Szollosi, F. Fulop and M. Bartok, Catal. Lett.,

2008, 125, 401.

380 M. Bartok, M. Sutyinszki, K. Balazsik and G. Szollosi, Catal. Lett., 2005, 100,

161.

381 K. Balazsik, Gy. Szoll +osi and M. Bartok, Catal. Lett., 2008, 124, 46.

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.

383 M. Bartuk, K. Balazsil, I. Bucsi and Gy. Szoll +osi, J. Catal., 2006, 239, 74.

384 M. von Arx, T. Mallat and A. Baiker, Angew. Chem. Int. Ed., 2001, 40, 2303.

385 K. Balazsik, T. A. Martinek, I. Bucsi, Gy. Szoll +osi, G. Fogassy, M. Bartuk and

G. A. Olah, J. Mol. Catal. A: Chem., 2007, 272, 265.

386 M. Bartok, M. Sutyinszki and K. Felfoldi, 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. Burgi and A. Baiker, J. Catal., 2004, 226, 69.

390 I. Busygin, J. Warna, 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.

392 J. L. Margitfalvi, E. Talas and S. G +obolos, Catal. Today, 1989, 6, 73.

393 J. L. Margitfalvi, Gy. Vanko, I. Borbath, A. Tompos and A. Vertes, J. Catal.,

2000, 190, 474.

274 | Catalysis, 2010, 22, 144–278

394 J. L. Margitfalvi, I. Borbath, M. Heged +us and A. Tompos, Appl. Catal. A:

Gen., 2002, 229, 35.

395 L.-Ch. de Menorval, 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. Bartok, P. T. Szabo, T. Bartok and G. Szollosi, Rapid Commun. Mass

Spectrom., 2000, 14, 509.

398 D. Ferri, T. Burgi and A. Baiker, A. J. Chem. Soc. Perkin Trans., 1999,

2, 1305.

399 J. L. Margitfalvi, E. Talas and M. Hegedus, Chem. Comm, 1999, 656–646.

400 J. L. Margitfalvi, E. Talas F. Zsila and S. Kristyan, Tetrahedron: Asymmetry,

2007, 18, 750.

401 J. L. Margitfalvi, E. Talas, N. Marın-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. Talas 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.

407 M. Bartok, M. Sutyinszki, K. Felfoldi and G. Szoll +osi, Chem Comm., 2002,

1130.

408 R. Hess, A. Vargas, T. Mallat, T. Burgi and A. Baiker, J. Catal., 2004, 222,

117.

409 G. Szoll +osi, A. Chatterjee, P. Forgo, M. Bartok and F. Mizukami, J. Phys.

Chem. A, 2005, 109, 860.

410 T. Mallat and A. Baiker, J. Catal., 1998, 176, 271.

411 O. J. Sonderegger, T. Burgi, 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. Burgi, Zh. Zhou, N. Kunzle, 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. Fulop and M. Bartok, J. Catal., 2007, 246, 266.

419 T. A. Martinek, T. Varga, K. Balazsik, G. Szollosi, F. Fulop and M. Bartok,

J. Catal., 2008, 255, 296.

420 O. J. Sonderegger, G. M.-W. Ho, T. Burgi 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. Burgi, F. Atamny, R. Schlogl 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. Burgi, J. Am. Chem. Soc., 2001, 123, 12074.

432 D. Ferri, T. Burgi and A. Baiker, J. Catal., 2002, 210, 160.

433 T. Burgi 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. Burgi, T. Mallat and A. Baiker, J. Am.

Chem. Soc., 2005, 127, 8467.

436 J.-D. Grunwaldt 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. Burgi, F. Atamny, A. Knop-Gericke, M. Havecker, T. Schedel- Niedrig, R.

Schlogl 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. Burgi and A. Baiker, Chem.

Comm., 2004, 744.

449 M. Bartok, K. Felfoldi, B. Torok and T. Bartok, Chem. Comm., 1998, 2605.

450 I. Bucsi, M. Sutyinszki, K. Felfoldi and M. Bartok, Catal. Comm., 2006, 7,

104.

451 M. Bartok, K. Felfoldi, Gy. Szollosi and T. Bartok, 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. Burgi and A. Baiker, J. Chem. Soc., Perkin Trans., 2000, 2, 221.

455 T. Burgi and A. Baiker, J. Catal., 2000, 194, 445.

276 | Catalysis, 2010, 22, 144–278

456 A. Vargas, T. Burgi, 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. Torok, K. Felfoldi and M. Bartok, 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. Bartok, K. Balazsik 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. Szori, K. Balazsik, K. Felfoldi and M. Bartok, J. Catal., 2006, 241, 149.

487 K. Szori, K. Balazsik, K. Felfoldi, I. Bucsi, S. Cserenyi, G. Szollosi, E. Vass,

M. Hollosi and M. Bartok, 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 andmaterial 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 chemicalresearch has been the recent application of gold in catalysis. In fact, thismetal has become an important tool in organic synthesis several years afterthe first reports on ethyne hydrochlorination and CO oxidation and now itis widely employed in many fundamental catalytic processes as oxidation,hydrogenation and coupling reactions.1–4 New applications of gold havebeen also proposed for commercial syntheses by academic and industrialresearchers.5–8 An ultimate project concerns an inorganic reaction, that isthe direct synthesis of hydrogen peroxide which has been developed by theimpressive work of Hutchings’ group.1–3

Strategic application of gold is the selective transformation of renewablebiological resources, a key importance task for balancing the CO2 cycle. Inparticular, valuable oxygenated compounds can be produced as newbuilding blocks for further transformation.

The catalytic conversion of carbohydrates and alcohols to the corres-ponding carbonylic or carboxylic compounds still maintains an attractingpower, being the products employed as chemical intermediates and highvalue components for perfumery, food and pharmaceutical industry.1–8

Selective oxidations using the eco-friendly air or pure dioxygen, as theoxidant, and supported metals as catalysts, are earning a general praise:they represent the promising response to the environmental restrictions forthe progressive shutdown of the traditional methods, causing undesired andtoxic by-products. Gold catalysis enjoyed an important progress owing tothe rapid advancement of nanotechnology and nanoscience, thus resultingin new applications for commercial syntheses by both academic andindustrial research worlds.9–12 The strong scientific appeal towards the‘‘precious metal’’ can be easily realized considering its peculiar property todiscriminate inside chemical groups and geometrical positions,13–15 and itschemical stability, strictly related to the unique features of gold itself. Thekinetic studies clearly show how much the activity is highly dependent onthe size of metallic gold particles. In particular, many investigations on theliquid-phase oxidation of polyols, alcohols, carbohydrates demonstrate thatonly small gold particles are catalytically active,16,17 which is in line with thebehaviour of gold particles employed in the gas-phase oxidation of carbonmonoxide.1

aDepartment 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. Gold catalysis in organic synthesis

2.1 Selective oxidation of alcohols

The reason why the oxidation of alcohols represents an attractive topicmainly lies in the wide diffusion of hydroxy-compounds, their easy avail-ability from renewable sources and the profitable employment of their de-rivatives as chemicals for organic synthesis. A great number of recentreviews deals with the catalytic oxidation of the C–OH group, outlining theevolution of the catalytic system from conventional Pt and Pd to moresophisticated Pt-Pd-Bi polymetallic systems, in order to increase selectivityand drop the deactivation process.18,19

A novel generation of catalysts for alcohols and polyols oxidation isrepresented by supported gold: its application leads to a dramatic im-provement in selectivity and stability, thus fomenting an exciting com-petition among ruthenium, platinum, palladium catalysts.1–3,20 Corma andHutchings’ research groups are quite active in this field, resulting in fun-damental achievements which have added a key progress in this topic.2,3 Inparticular, Hutchings et al.21 have shown the advantage in using continuousflow reactors for the oxidation of glycerol under mild conditions: bothmonolith and meso-scale structured downflow slurry bubble column designslead to an increment in the reaction rate and selectivity towards glyceric acidover autoclave.

An ultimate method has now been proposed,22 based on a gold-immo-bilized microchannel flow reactor for the oxidation of alcohols with mo-lecular oxygen: Kobayashi and co-workers have shown how the oxidationof various alcohols proceeded easily to give the corresponding aldehydesand ketones in good to excellent yields. No leaching of gold was observedand the gold-immobilized capillary column could be continuously used forat least four days without loss of activity. Probably, it is the first example ofa microreactor that allowed full conversion of alcohols by aerobic oxidationof alcohols in microchannels.

However, the first systematic study on gold catalysis for selectiveliquid phase oxidation has been performed at Milan University, with theambitious aim to find a substitute for palladium, platinum and, particularly,copper in the aerobic oxidation of the alcoholic group. The most chal-lenging drawbacks to be overcome were metal leaching and scarceselectivity of the traditional catalysts. The early experiments for testingthe activity of metal gold were disheartening: whereas bulk copperquickly reacted with O2 and ethane-1,2-diol in basic solution to produceoxoethanoate- and formate-derivatives,23 gold powder resulted totally inerttowards any transformation of the glycol. The high chemical stability ofbulk gold was overcome by discovering the new peculiarities of goldnanoparticles and the logic of Scheme 1 was soon adopted also byother different research groups for liquid and gas phase applications.The favourite methods for preparing catalysts were co-precipitation, de-position-precipitation and colloidal particles immobilisation. In our case,finely dispersed gold supported on carbon by metal sol immobilisationallowed to achieve an active and selective catalyst for liquid phaseoxidation.

280 | Catalysis, 2010, 22, 279–317

2.2 Catalyst preparation

Efficient, carbon-supported gold catalysts for liquid phase oxidation can beprepared starting from colloidal dispersions containing metallic gold (sol).Differently sized gold particles, in the range 2–10nm, could be obtained byreducing chloroauric acid with NaBH4 in the presence of stabilizing agents aspolyvinylalcohol (PVA), polyvinylpyrrolidone (PVP) and tetrahydroxymethyl-phosphonium chloride (THMP), glucose. Au (III) concentration is a keyfactor for tuning particle size. Either high resolution electron transmissionmicroscopy (HR-TEM) or X-ray diffraction (XRD) techniques are applied forparticle size determination after immobilisation of the sol on a useful sup-porting material, as copper grid and carbon powder. TEM shows the directimage of the metal particles, whereas Scherrer equation allows the calculationof the mean diameter from the half height width of the XRD pattern.

Colloidal gold nanoparticles were generally collected on two types ofactivated carbons: for catalytic tests, Au was immobilised on a coconutderived carbon powder (AS=1300m2 g� 1 from Camel) at a level of 0.2–0.8% (w/w), which was chosen for the low sulphur content, while for XRDdetermination 1–2% Au (w/w) was contacted with a pyrolytic carbonpowder (AS=254m2 g� 1 from Cabot), which was selected for its fast ad-sorption property. The modulation of gold clusters size was achieved usinginitial solutions ranging from 25mgL� 1 (small particles) to 500mgL� 1

(large particles) of gold.

2.3 Oxidation of diols

Aliphatic 1,2 diols can be oxidised to the corresponding monocarboxylateswith O2 under low pressure (1–3 bar) in the presence of the stoichiometric

Kinetics, Mechanism, ModelsNanotechnologies

High dispersion

Particle size

Metal-support interaction

Nature and role of the support

GoldLong life but low activity metal

Scheme 1 Tailoring efficient gold catalysts.

Catalysis, 2010, 22, 279–317 | 281

amount of NaOH. Supported gold particles were shown to be the bestcatalytic system, by comparing with Pd and Pt metals (Table 1).13,24

Developing the sol immobilisation technique and improving our ability inpreparing small colloidal gold particles, gold catalyst activity could begrown from a few hundred TOF units up to 3500 h� 1 in the case of gly-colate and 2000 h� 1 for lactate. Selectivity at 100% conversion was raisedto surprisingly high values, differently from the lower performance of Pdand Pt catalysts.

Regarding phenylethane-1,2-diol oxidation, however, a worse selectivityscenario appeared, probably due to a strong induction effect of the phenylgroup: two abundant by-products, namely benzoate and phenylglyoxylate,were detected together with the expected mandelate, in basic conditions(Scheme 2).

An accurate comprehension of the experimental results can not set asidethe alkali catalysed keto-enolic equilibrium c and internal Cannizzaro-typereaction f, reported in Scheme 3.

The original gold selectivity could be evaluated by dropping the reactionpH to the value 7 when reactions c and f, together with the overoxidation ofmandelate, were inhibited.

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

diols

HOOH

Au/C Pd/C Pt/C

TOF (h� 1) 3500 500 475

Selectivity (%) 98 77 71

OH

OH

Au/C Pd/C Pt/C

TOF (h� 1) 2000 720 650

Selectivity (%) 99 90 89

+ O2Au/C

NaOH+ By-products

COO

OHOH

OH -

By-productsCOO

O

COO+

- -

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

282 | Catalysis, 2010, 22, 279–317

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

The oxidation observed under these conditions concerned mainly thesecondary alcoholic function, 62.5%, followed by the primary one, 45%.The selectivity towards mandelate can be increased by promoting reactionsc and f. Table 2 visualizes how we were able to improve the selectivity from45 to 83% by increasing alkali concentration and temperature.25

Non-vicinal glycols can also undergo selective oxidation: however, Table 3shows the worse reactivity of 1,3 propanediol and diethyleneglycol oxidation

OHOH

OOH

OHO

OO

OHO

O

OO

O

-

-

O

O-

a

b

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

OHOH

OOH

OHO

OO

OHO

O

OO

O

-

-

O

O

-

a

b

+ HCO3-

c

e

d

f

g

h

Scheme 3 Reaction pathway of phenylethane-1,2-diol oxidation in basic solution.

Table 2 Optimization of phenylethane-1,2-diol (PED) oxidation for

producing mandelate with Au/C catalyst

NaOH/PED T (K) Conversion % Selectivity %

1 343 52 45

2 343 100 60

2 363 100 70

4 363 100 83

Catalysis, 2010, 22, 279–317 | 283

with respect to vicinal diols reported in Table 1, while the selectivity tomonocarboxylates maintains always high values.

The great interest in the synthesis of dicarboxylic acids has suggested athorough study on the oxidation of diethyleneglycol, in order to force thereaction towards the double oxidation, by changing the amount of alkali,nature of catalyst and temperature. The optimization of the experimentalconditions at O2 pressure 3 bar and substrate: Au ratio 1000 led to theproduction of monocarboxylate by gold on carbon, whereas gold on titaniaresulted in 45% of the diacarboxylate in the presence of 2mol of NaOH at363K (Table 4). Other by-products were absent.26 In the case of Au/TiO2

catalyst, a positive metal-support interaction was therefore outlined.

2.4 Oxidation of other polyols

2.4.1 Glycerol. The huge and easy availability of glycerol as a by-product of biodiesel has recently prompted research to transform this cheapcompound into valuable chemicals.27,28 The application of gold catalysis inglycerol oxidation under mild conditions has been experimented mainly bytwo Groups. In spite of the variety of potential reaction products, origin-ated by the general oxidative pathway reported in Scheme 5, Hutch-ings’group has underlined the high selectivity of gold: using graphite as a

Table 3 Oxidation of isolated diols with gold catalysts. Substrate=0.4M; substrate/Au=100;

T=343K; pO2=3 bar; pH=9.5

OH OH

Au/C Au/TiO2

TOF (h� 1) 430 490

Selectivity (%) 100 95

HO

OOH

Au/C Au/TiO2

TOF (h� 1) 240 240

Selectivity (%) 99 98

Table 4 Influence of catalyst and experimental conditions on the oxidation of di-

ethyleneglycol to mono- and dicarboxylates

Catalyst NaOH/Substrate T (K) T (h) Conv % Monoacid% Diacid%

1%Au/C 1 343 4 96 99 1

1%Au/C 2 343 4 80 97 3

1%Au/C 2 363 1 83 98 2

1%Au/TiO2 1 343 4 95 98 2

1%Au/TiO2 1 363 2 95 96 4

1%Au/TiO2 2 343 3 94 70 30

1%Au/TiO2 2 363 6 100 55 45

284 | Catalysis, 2010, 22, 279–317

support, in water solution at 333K 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 carriedout by Prati et al. in Milano. In a first study, the relationship betweencatalyst morphology and selectivity was explored at total conversion: it hasbeen found that larger gold particles (20 nm), supported on suitable car-bons, show low TOFs but favour glycerate formation under mild conditions(303K, 3 bar), thus allowing yields up to 92%.30 Exciting results wereachieved by the Authors using bimetallic nanoparticles as supported cata-lyst.31 Two important points were highlighted: a) the activity can be im-proved by using the bimetallic Au-Pt and Au-Pd systems, thusdemonstrating a synergistic effect between the metals; b) the selectivity tothe desired product can be affected by the nature of the catalyst (particlesize, alloyed phases and support) and experimental conditions. A smart useof the metals allowed a precise modulation of selectivity, whereas pure goldpromotes glyceric acid formation, Pd addition favours further transfor-mation to tartronic acid and Pt addition leads to carbon-carbon bond fis-sion resulting in glycolic acid. Also in this case, the best performance wasrecorded with smaller particle size catalysts. Another important parameteris the atomic ratio of the metals in (AuxPdy)/C catalyst, which deeply in-fluences 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 sorbitoloxidation (Scheme 6) and a comparison with Pd and Pt catalysts was carried

Dihydroxyacetone

O

OHHOOH

OHHO

Glycerol

Mesoxalic (or β-ketomalonic) acid

O

OO

OH OH

Hydroxypiruvic acid

O

OHO

OH

Glyceraldehyde

OH

OHO

H

Glyceric acid

OH

OHO

OH

Tartronic acid

OH

OO

OH OH

Scheme 5 Reaction pathway of the aerobic oxidation of glycerol.

Catalysis, 2010, 22, 279–317 | 285

out (Fig. 1).32 The presence of alkali and the use of carbon as the supportingmaterial allowed the successful performance of the reaction. Monometallicgold catalysts led to humble TOFs, favouring the oxidation of the primaryalcoholic function to monocarboxylates and leading to gluconate andgulonate with very low amounts of dicarboxylate (glucarate).

An improvement of activity was achieved by employing bimetallicAuþPd and AuþPt catalysts, resulting in full conversion. The bimetallicsystem was superior to monometallic gold also for the selectivity tomonocarboxylates at a given conversion, resulting in superior values almostindependently 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 beenhighlighted in a official classification.33 Presently, 3-hydroxypropionic acid is arare and expensive chemical which is commercialized as an aqueous solution

HOOH

OH

OH

OH

OH

HOO

OH

OH

OH

OH

O-

HOO

OH

OH

OH

OH

O-

Sorbitol

GluconateGulonate

OO

OH

OH

OH

OH

O

O

-

-

Glucarate

Scheme 6 Aerobic oxidation of sorbitol

0

30

60

90

120

150

180

1%Au/C 1%Pd/C 1%Pt/C

Conversion %

Selectivity %(Gluconate+Gulonate)

Selectivity % (Glucarate)

TOF (h-1)

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

286 | Catalysis, 2010, 22, 279–317

by a few suppliers. Beside the traditional stoichiometric reactions, also catalyticmethods are reported for its synthesis and many patents have recentlyraised.34–40

In spite of all the efforts, however, none of the proposed new processes iseffectively operating to our knowledge.

Recently, a new, unexplored route to 3-hydroxypropionic acid based onthe aerobic oxidation of allyl alcohol has been reported.41 Its economicaladvantage lies on the fact that allyl alcohol (2-propene-1-ol) is a large scalechemical (over 50 000 tonne per year production) derived from the petro-chemical industry via propene and propene oxide as intermediates.42

As in many fortuitous discoveries, gold catalysis, very selective in theaerobic 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 slowoxidation takes place which produces 3-hydroxypropionate beside minoramounts of acrylate and glycerate. Indeed, the oxidation can be controlledby tuning the alkali amount and temperature, as reported in Table 5.

According to the reported data, a high selectivity at full conversion to thehigh value 3-hydroxypropionic acid can be achieved with an excess ofNaOH at mild temperature, 323K.

Further experiments were carried out for comparing the home-made0.3% Au/C catalyst with the reference catalyst 1.5% Au/TiO2 provided byWorld Gold Council,43 under the same conditions (Table 6).

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 1 298 98 19 15 38.5 traces

2 3 298 100 16 30.5 5 0

3 1 323 98 42 13 33 traces

4 3 323 100 79 9.5 11 0

5 3 353 100 74 18 6 0

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 1 298 37 8 23.5 0 0

7 3 298 25 7 18 0 0

8 1 323 94 50 37 8 0

9 3 323 97 53 32 12 0

10 3 353 97 11 21.5 6 0

Catalysis, 2010, 22, 279–317 | 287

It can be outlined that Au/C catalyst presents the best results, in termsof yields of 3-hydroxypropionic acid (79% for Au/C against 53% forAu/TiO2).

The reaction mechanism of allyl alcohol aerobic oxidation leading to theunexpected product is of hard interpretation. The progressive samplingduring the 24 h, in fact, does not help to find the first transformation of thereagent, because no transient species could be detected beside acrylic acidand glyceric acid, while glycerol was detected only in trace amounts (o1%).Further tests showed that glycerol could be transformed into glyceric acidunder the above reported conditions but neither acrylic acid, nor glycericacid and 3-hydroxypropeneoxide (glycidol) could produce 3-hydro-xypropionic acid in detectable amounts. Different pathways for the variousproducts during allyl alcohol oxidation should be guessed. Moreover, usingacrolein as a one pot reagent no 3-hydroxypropionic acid was recordedbeside the yellow solid material, probably a condensation product. The slowaddition of acrolein (1.1ml), in progressive small portions (ten microlitersevery 0, 5 h), to the reacting mixture resulted in the formation of 3-hydro-xypropionate (10% yield after 24 h reaction). This test suggests acrolein asa probable non detectable intermediate, undergoing a base catalyzedMichael-addition of water to give 3-hydroxypropanal, whose rapid oxi-dation would give the observed 3-hydroxypropionate (Scheme 7).

A specific catalyst, 0.3% Au/C, was prepared as in Section 2.2, usingglucose as a protecting agent, decreasing the pH of the gold colloid to 4 bymeans of HCl and calcining at 673K under H2 for 2 hours.41

2.6 Other alcohols

The conversion of aliphatic and aromatic alcohols to aldehydes underneutral conditions can be successfully performed by gold catalysis. Thefacile achievement of the corresponding carboxylates is reached in thepresence of alkali. Hutchings et al. have reported exciting results usingAu-Pd bimetallic catalysts, owing to a synergistic effect between themetals,44 while Corma et al. have demonstrated the synergistic effect be-tween Au nanoparticles and the supporting nanometric CeO2 materialunder solvent-free conditions.45 For many practical applications in organicsynthesis, gold catalysis applied to alcohols oxidation is relatively slow(TOF values around dozens or hundreds h� 1). To overcome the intrinsiclow activity of gold, Hutchings and co-workers studied the contribute of asecond metal, in particular palladium. Thus, Au-Pd on TiO2 was reportedto show an extraordinary enhancement compared to monometallic gold.Benzylic alcohol, under solventless conditions, could be oxidized to alde-hyde five times faster in the presence of Au-Pd/TiO2 than Au/TiO2 at 373Kand 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 wasreported for 1-phenylethanol oxidation, using Au-Pd/TiO2. Other supports,as Al2O3 or Fe2O3, were not producing catalysts as active and selective asAu-Pd/TiO2. A variety of alcohols were selectively oxidized in the absenceof solvent; however, when toluene or water were used as solvents, a generaldropping in activity of the catalyst could be observed.

288 | Catalysis, 2010, 22, 279–317

Corma et al. developed a peculiar catalyst by supporting Au on nanosizedCeO2. This catalyst appeared to be not only highly selective toward theoxidation of alcohols to carbonylic derivatives, but also very active oper-ating, without any solvent, at 353K and atmospheric pressure. Benzylic andcinnamyl alcohols were smoothly oxidized to aldehydes, whereas secondaryalcohols were transformed to ketones. Primary alcohols, such as 3-phenyl-1-propanol, however produced 3-phenylpropyl-3-phenylpropanoate in 83%selectivity at 73% conversion.

The detailed description of this wide and important research area hasbeen recently reported.3

2.7 Aminoalcohols

The doping effect of the amino-group on traditional metal catalysts may bethe cause of the lack of literature regarding the aerobic oxidation of ami-noalcohols. Gold has been shown to be a pleasant exception for oxidizingthis important class of aminoacids.26,46–49 In fact, we firstly discovered thatnanometric gold represents a much better catalytic system if compared topalladium and platinum under similar conditions (Table 7). Alkali increases

OH

Allyl alcohol

OHHO

OH

Glycerol

OH

O

HO

O

OH

Acrylic Acid

O

HAcrolein

OH

O

HO

OH

Glyceric acid

(traces)

3-hydroxypropionic acid

Scheme 7 Pathways during the oxidation of allyl alcohol.

Catalysis, 2010, 22, 279–317 | 289

the oxidation rate, despite the amino group already ensures the presence ofa basic solution.26

The material employed as the gold support is relevant for affecting thecatalytic performance, as it has been outlined also in other cases: alumina isa better supporting material for gold nanoparticles than carbon (Table 8).

An interesting problem of chemo-selectivity is present in the catalyticoxidation of aminoalcohols of formula R1R2N–(CH2)n–CH2OH. As high-lighted in former experiments on primary aminoalcohols26 and in morerecent results on tertiary amines,49 this oxidation can result into the cor-responding aminoacid as well as the N-oxide. The resulting product is de-termined by the nature of the nitrogen substituents, experimental conditionsand catalytic system.

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=343K, t=2h. b)

presence of alkali, [Substrate]=0.4M, substrate/metal=1000, substrate/NaOH =1, pO2=3

bar, T=343K, t=2h.

HO

NH2

1% Au/C 5% Pd/C 5% Pt/C

Conversion (%) a) 3 0 0

Conversion (%) b) 20 0 0

H3C

OH

NH2

1% Au/C 5% Pd/C 5% Pt/C

Conversion (%) a) 22 0 0

Conversion (%) b) 65 0 0

HO

NH2

1% Au/C 5% Pd/C 5% Pt/C

Conversion (%) a) 3 0 0

Conversion (%) b) 20 0 0

H3C

OH

NH2

1% Au/C 5% Pd/C 5% Pt/C

Conversion (%) a) 22 0 0

Conversion (%) b) 65 0 0

290 | Catalysis, 2010, 22, 279–317

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

These data underpin the observation that aliphatic amines cause thecatalytic deactivation of the traditional noble metals.

The same test carried out in the presence of NaOH at pH 10.8 resulted inworse performances with all the catalysts, except for Rh/C which arose itsactivity up to 33% (Fig. 2).

A general evaluation on the role of alkali in promoting or depressing thecatalyst activity is risky to be done, because of the limited number of testedsubstrates.26,49 The nature of substrates, however, has been outlined to beessential for affecting gold catalytic performance.

Table 9 Aerobic oxidation of 3-dimethylamino-1-propanol. Reaction con-

ditions: substrate 0.4 M, substrate/M=1000, pO2=2 atm, T=363K, t=24h.

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 100 100 0

1%Au/Al2O3 100 100 0

1%Au/TiO2 95 100 0

1%Rh/C 20 0 0

1%Pt/C 0 0 0

0.5%Au-0.5%Rh/C 33 100 0

0.5%Au-0.5%Pt/C 40 100 0

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=343K; t=2h.

Substrate Conversion %

H2NOH 23

OH

NH2 100

H2N OH 27

H2N OH

OH

32

Catalysis, 2010, 22, 279–317 | 291

3. Selective oxidation of carbohydrates

The worldwide availability of carbohydrates is attracting the interest in theiroxidative transformation. As a general trend, in the oxidation of aliphaticoxygenated compounds with supported gold particles the following order ofreactivity has been observed: aldehydesWprimary alcoholsWsecondary al-cohols; tertiary alcohols and carboxylic acids are almost inert under mod-erate conditions (up to 363K and 3 bar). In particular, the aerobic oxidationof aldehydes can be performed using water, organic solvents and solventfree conditions, also in the absence of alkali. We discovered that Au is ableto easily oxidize aldehydes in water solution and, differently from Pt, nodeactivation was observed on recycling. According to the expected trend,catalytic aldose oxidation occurs at the aldehydic group leading to carb-oxylic acid or carboxylates.

3.1 Glucose to sodium gluconate

Gluconic acid and gluconates are industrial intermediates widely employedin food chemistry, surfactants and cleansing agents. That is why the oxi-dation of glucose, a cheap renewable starting material, appears as acharming topic. Moreover, the present industrial production is only viafermentation by enzyme (Aspergillus niger mould), but the low productivityof this process has prompted the interest in finding eco-friendly technologiesbased on the use of oxygen in aqueous solution, under mild conditions byheterogeneous catalysis. Many efforts have been so far carried out: Pt-basedcatalysts, for example, allowed high conversion and good selectivity, butquickly deactivated because of leaching, self-poisoning and over-oxidation.These limits could be partly overcome by using bi- and tri-metallic catalystsand the promoting effect of Bi.50,51

0

20

40

60

80

100

Au/C Au/Al2O3 Au/TiO2 Rh/C Pt/C Au-Rh/C Au-Pt/C

Conversion Selectivity (N-oxide) Selectivity (Aminoacid)

%

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

292 | Catalysis, 2010, 22, 279–317

The first application of gold catalysis for transforming glucose into glu-conates was soon excellent in terms of activity, selectivity and durability. Acomparison among different Pd, Pt and Au catalysts, performed at 323Kand atmospheric pressure, demonstrate the great peculiarity of gold: whilepalladium and platinum catalysts led to selectivityo95%, Au resulted in aselectivity close to 100% at total conversion.52 Although superior to Pt andPd catalysts, in the first experiments gold was shown to be less efficient withrespect to enzymatic catalysis.

Encouraged by the promising achievements, efforts were addressed to thedeep improvement of the catalytic system, in order to find an alternative processto the biochemical route. A fundamental contribute was also played by themechanism studies (Section 3.1.1) which were carried out side by side. Startingfrom TOF of a few hundred h� 1 units, we reached the exceptional value closeto sixty thousand units,17 which is similar to the behaviour of enzymatic cata-lysis. Following a detailed study on the kinetics of enzymatic oxidation,53 wecomparedHyderase (fromAmano Enzyme Co., U.K.), a biological preparationcontaining glucose oxidase and catalase as active components and flavine-ad-enine dinucleotide (FAD) as the rate controlling factor (1.3� 10� 6mol g� 1),with the most efficient gold catalyst, 0.5% Au/C prepared as in Section 2.2,containing metal particles of 3.6nm mainly at the surface.

Kinetic data of glucose oxidation to gluconate were recorded using a glassreactor interfaced to an automatic titration device equipped with NaOH asa reagent. Magnetic stirring and high speed turbine were alternatively usedduring the tests. A careful control of various parameters such as pH, tem-perature, glucose concentration and stirring speed led to the comparativeresults reported in Scheme 8.53

Hyderase Au/C

[Glucose] 1 M 3 M

Catalyst/Glucose 56

pH 9.55-7

T 303 K 323 K

Stirring 900 rpm 39000 rpm

Spec. Activity 218 h-1145 h-1

Productivity 122 kg m-3 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 turnoverfrequency of 600 000 h� 1 was obtained, being this value undoubtedly betterthan the efficiency of active external gold atoms in the inorganic catalyst,calculated as 90 000 h� 1. However, taking into account the lower FADconcentration in the enzymatic extract and the potentiality of gold, whichallows a threefold higher glucose concentration, the final productivity issuperior with gold by using a similar amount of total catalyst.

3.1.1 Kinetics and models. Kinetic investigations were performed on theselective liquid phase oxidation of glucose, using carbon supported goldparticles54 or unsupported colloidal gold particles55 as the catalysts. Thefirst study was carried out by Claus and co-workers: they proposed aLangmuir-Hinshelwood model, considering the surface oxidation reactionas the limiting factor of the whole reaction rate, while adsorption of sub-strate and desorption of the product were regarded as fast steps. TheAuthors finally suggested a dehydrogenation mechanism converting glucoseto sodium gluconate, supposing water as the reduction product of dioxygenand detecting a scarce effect of glucose concentration on the reactionrate. In the second paper, the application of ‘‘naked’’ gold particles as acatalyst led to detect hydrogen peroxide-instead of water-as the reductionproduct of O2.

Kinetic tests at low glucose concentrations (o0.1 M) recorded a firstreaction 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 hy-drated form on gold, was then suggested because of the optimum fittingwith the experimental evidence. As a consequence, a rate equation which fitsboth the first-order with respect to oxygen and the decreasing order withrespect to glucose was found.55

In order to compare the enzymatic catalysis with the gold catalytic sys-tem, a kinetic investigation of glucose catalysed by the above citedHyderasewas performed under similar conditions applied for the gold catalysis.According to measurements of initial rate as a function of initial glucoseconcentration were recorded. A Michaelis-Menten mechanism resulted tobe coherent with the experimental data.56 It is worth noting that both goldcatalysis and enzymatic catalysis are able to promote the oxidation ofglucose with the same stoichiometry where the 2 electrons reduction ofmolecular oxygen produces hydrogen peroxide as the transient by-product.Nevertheless, enzymatic and inorganic catalysis apply different strategicmechanisms. Regarding the enzymatic system, the rate determining step isthe oxidation of the substrate by the enzyme, which is transformed into thereduced form according to a faster step and showing a zero order with re-spect to dioxygen. The oxidation of glucose by dioxygen dissolved in waterrepresents the rate determining step of the gold catalytic process, with a firstorder dependence of the reaction rate on pO2.

The corresponding rate determining steps interestingly recorded similaractivation energies (47.0 kJ/mol for gold and 49.6 kJ/mol for enzymaticcatalysis).55,56

294 | Catalysis, 2010, 22, 279–317

Scheme 9 visualizes the proposed molecular mechanism of glucose oxi-dation on a gold nanoparticle.57

This mechanism suggests that the presence of alkali is fundamental foractivating gold particles.52 Kinetic experiments confirmed that there is nodifference in the activity of unsupported and supported gold nanoparticles,except for a better stability of the supported gold catalyst in the time, thusexcluding any role of the support in promoting the oxidation.57

The key role of the particle size in affecting gold catalytic activity is aleit motiv also characterizing glucose oxidation. The sudden loss of activityof particles larger than 10 nm was recorded using unsupported colloidalparticles (10� 4M Au, 0.38M glucose, 303K under oxygen at atmosphericpressure), due to a quasi-linear correlation between the initial specific molaractivity (moles of reacted glucose/ mole total gold� h) and the inverse of themean diameter in the range 2–7 nm.17

A correlation between surface structure and catalytic behaviour in solidmaterials is of strategic importance for producing quick and clean industrialreactions. This has prompted to derive a geometrical model for describingthe morphological properties of two catalysts made of carbon supportedgold particles, prepared as in Section 2.2, having a known size distributioncentred at 3.30 nm and 7.89 nm respectively.58 For this purpose, the pro-gressive poisoning effect of different molecules on these catalysts, performedduring the aerobic oxidation of glucose, has been used as a diagnostic tool.The observed deactivation trend follows the order thiocyanateWcyanideEcysteineWthiourea and each of them obeys an exponential law. Thekinetics of catalyst deactivation has been interpreted by considering the

R

O

H

HO−

R

OH

H

R

OH

H

O

Au-

O2 slow

R

H

OH

O OR

O

O−

H2O2

R

OH

H

O−

AuAu

degradation

R

O

H

+ O−

O−

+ Au R

OH

H

O

Au−

HAu

+ AuAu+

Scheme 9 Molecular mechanism of aerobic glucose oxidation with gold nanoparticles.

Catalysis, 2010, 22, 279–317 | 295

contribute of electronic factors which overlap the space shielding of activesites, due to long range poison-catalyst interaction influencing the entiremetal particle.

The consequent insight in the aerobic oxidation of glucose suggested amolecular model for electronic interactions in gold nanoparticles: correl-ating the nature of the molecules, which caused a consistent poisoning ef-fect, and considering the promoting effect of OH� , we found that thedioxygen reduction step is differently influenced by soft and hard-nucleo-philes. In conclusion, it has been underlined how the competition of thepoisoning molecule with the reagents can be discussed considering twoextreme cases: for a hard nucleophile, no back-donation from metal to theLewis base is expected, leaving in the reacting solution the original or ahigher catalytic effect, as in the case of OH� . Regarding a nucleophile, N,showing p back-bonding ability, the removal of the electron density fromthe metal inhibits dioxygen reduction thus decreasing the catalytic propertyof the entire gold particle (Schemes 9, 10).

3.2 Free gluconic acid

Many efforts have been devoted to find a one-pot synthesis of free gluconicacid, because the present method-from calcium gluconate and sulfuric acid-leads to large amounts of CaSO4 as a by-product. Despite all the attempts,following either enzymatic or inorganic routes, no interesting results were sofar recorded due to the inhibition of the catalytic systems at low pH values.Focusing on this challenging topic, multi-component catalysts resultedmore active than monometallic gold particles. In particular, an importantsynergistic effect between gold and platinum was observed (Fig. 3).

The most promising AuþPt combination was further optimized, leadingto a quite active catalyst for alkali-free oxidation of glucose containing goldand platinum in the ratio 2:1 (w/w). The synergistic effect was detected in aseries of experiments which compare colloidal catalysts and supportedcatalysts.59 A direct synthesis of gluconic acid by aerobic oxidation ofglucose seems to be possible with gold-based catalysts, starting from theencouraging TOF values around 1000 h� 1.

4. Selective oxidation of hydrocarbons

In order to highlight the versatility of heterogeneous gold catalysis, someapplications of unsubstituted hydrocarbon oxidation are considered.

H

RO

OH

O

OAu

N

σ π

Scheme 10

296 | Catalysis, 2010, 22, 279–317

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

4.1 Propene epoxidation

All the efforts to produce propene epoxide (PO) commercially by directoxidation of propene, similarly to the silver promoted synthesis of etheneoxide, have been so far vain and the largest amount of PO is still manu-factured by chlorohydrine process (49%) and the indirect hydroperoxideprocesses.42 The progressive request for eco-friendly processes, which canexclude chlorine dependence and huge amounts of undesired by-products,has prompted basic research to find alternative catalytic routes. In thiscontext, gold has been shown to be a very promising catalyst. In fact, thepioneering Haruta’s work has reported that supported gold catalyst is ableto promote the gas phase epoxidation of propene by O2 in the presence ofH2.

60 The behaviour of gold is unique, as shown by comparing differentmetals dispersed on titania (M=Au, Pt and Pd) under moderate conditions(298–353K) when equimolecular amounts of H2 and O2 are reacted withC3H6. Only Au produces propene oxide (PO), while Pd and Pt promotemainly the hydrogenation of C3H6 to C3H8 and the formation of smallamounts of acetone and carbon dioxide. All the experiments leading to thetotal selectivity to PO, underlined the strategic role of TiO2 as the support.However, the low conversion needed to reach high selectivity represents aweak point. As a consequence, most of the subsequent investigations weredevoted to improve PO yields meanwhile maintaining a high selectivity.

Au Pt Pd Rh Au-Pt Au-Pd Au-Rh

0

50

100

150

200

250

300

Conversion %

TOF (h-1)

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

Catalysis, 2010, 22, 279–317 | 297

Insights on gold catalysts supported on non-porous and mesoporous tita-nia-silica allowed further progress in PO productivity, also indicating theeffect of preparation conditions and pre-treatments on their activity andstability. A series of Au/TS-1 catalysts with different gold and titaniumcontents was examined at 413–473K at a space velocity of 7000mL(h gcat)

� 1. A catalyst prepared with a Si/Ti=36 (atomic ratio) and a goldloading of 0.05 wt% produced 116 gPO(h kgcat)

� 1 at 473K, which was thehighest rate at that time reported for a TS-1-based catalyst with no de-activation during 40 h. Catalysts prepared with lower titanium and goldcontents resulted in very active catalysts, up to 350 gPO(h gAu)

� 1 at 473Kfor 0.01 wt%Au/TS-1 (Si/Ti=500), indicative of a more efficient use of goldand titanium for the epoxidation reaction. The low gold loading coupledwith non-detectable gold particles in TEM micrographs suggested that, inthese 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 afourfold increase in Au/TS-1 catalysts. The higher gold amount producedcatalysts allowing quite high conversions of propene (5–10%) with accept-able selectivity (75–85%), at 473K and a space velocity of 7000ml(h g cat)

� 1. The related productivity resulted in 134 gPO (hkg 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 representedby the acetoxylation to vinylacetate (VAM), the latter being the monomericunit for the production of polyvinylacetate (PVAc). Pd-Au bimetallic silica-supported catalyst, promoted by potassium acetate, is the well-knownsystem commercially applied for the production of the monomer (VAM).3

The relevant added value of vinyl acetate monomer has attracted theinterest of both academic and industrial research groups, thus leading to anumber of patents. Researchers at Celanese International Corporation wereparticularly active in this area describing in details successful preparationmethods63: the general procedure, similar to the deposition-precipitationtechnique, provides the active metals on the surface firstly as water-insol-uble compounds which are reduced by a second step to the metallic form.

A shell-impregnated catalyst, Pd-Au on a silica support, was also de-scribed for the synthesis of vinyl acetate which allowed a selectivity over90%.64

4.3 Oxidation of higher alkenes

Supported gold catalysts have been employed in the aerobic oxidation ofother alkenes without using a second reagent, H2, as a sacrificial reductantor CH3COOH as the acetilating reagent. Hutchings et al.65,66 and otherresearch groups67–69 found that alkene oxidation fairly proceeds by addinga catalytic amount of peroxides (either hydrogen peroxide or tert-butylhy-droperoxide) as an oxygen chain initiator. These works fairly emphasizehow much selectivity and conversion are dependent on substrate, catalystand experimental conditions. The oxidation of cyclohexene, styrene, stil-benes 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 WorldGold 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 1,2,3,5-tetramethylbenzene solvent (TMB). Moreover, a promoting effect of bis-muth on Au/C catalyst led to 98% selectivity of a valuable mixture ofproducts.65

Styrene could be converted by aerobic oxidation into epoxide with a lowselectivity (29%), by using either a mixture of 1,2,4,5-tetramethylbenzene(TMB)/1,4-dimethylbenzene (DMB) or hexafluorobenzene as a solventand 1% Au/C as a catalyst. However, the major oxidation product wasbenzaldehyde with a selectivity around 46% for both solvents, while acet-ophenone 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 Yinget al.68 in styrene oxidation in the presence of gold on mesoporous alumina,obtaining 70% selectivity to epoxide along with a consistent production ofbenzaldehyde.

Similarly, nanosized-gold deposited on TiO2 by deposition-precipitationmethod was shown to be an active and selective catalyst (around 50% se-lectivity) for the epoxidation of styrene by TBHP. Furthermore, more exoticoxides as supporting materials for the same reaction were investigated, suchas gallium, indium and thallium oxides, thus revealing a pretty good per-formance of Au/Tl2O3 (around 60% selectivity).67

Cis-stilbene oxidation with dioxygen using a 1%Au/Graphite catalyst ledto the corresponding epoxide, with cis : trans ratio depending on the solventbut always in favour of the trans conformation: 74% selectivity to trans-stilbene epoxide at 48% conversion was found by means of i-propylbenzeneas a solvent.65

The oxidation of cis-cyclooctene was investigated using 1% Au/C cata-lysts both in the presence and absence of solvents by Hutchings et al.65 Thebest 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 solvent81% epoxide at 8% conversion was obtained.

4.4 Oxidation of alkanes

The activation of C–H bond in the selective oxidation with dioxygen by goldcatalysis has appeared to be a promising reality. Research has particularlyfocused on the synthesis of cyclohexanone and cyclohexanol, because it is amatter of growing interest for chemical industry. These oxidation products,in fact, are fundamental intermediates for making e-caprolactam and adipicacid, thus leading to nylon-6 and nylon-6,6 manufacture beside less im-portant applications as stabilizers, homogenizers for soaps and syntheticdetergent emulsions, and as solvents for lacquers and varnishes. Zhao andco-workers70 first applied gold catalysis in the activation of cyclohexane:Au/ZSM-5 and Au/MCM-41 favoured a selectivity around 90% and con-versions of 10-15% at 423K, even though a loss in both activity andselectivity 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 satisfactoryone-pot oxidation of cyclohexane65,70–72: Au/graphite without any solvent,but using a halogenated benzene as an additive, led to 92% selectivity(cyclohexanoneþ cyclohexanol) at low conversion (1%).65 Higher con-versions (20–30%) and selectivity (95%) were achieved by Zhu et al.71 withgold on mesoporous silica catalysts, a clearly better result compared to thecurrent industrial process leading to 70–85% selectivity at 4% conversionand 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. Gold catalysis in material science

5.1 Conducting polymers: polyaniline and polypyrrole

Since the first preparation of the highly conducting polyacetylene (PA) in1977, much effort has been focused on the synthesis of other organic con-ducting polymers such as polyaniline (PANI) (Fig. 4), polypyrrole (PPy)(Fig. 5) and polythiophene (PTh) and their applications in devices com-bining optical, electrochemical and conducting properties, owing to theirgreat versatility.73,74 PANI, in particular, is unique because of its tuneableconductivity being connected to the degree of acid-doping (pH) and oxi-dation state of the material. Equal numbers of oxidized and reduced units(emeraldine form), with one proton doping every two units, guarantee op-timum conductivity of the polymer.

Recently, great attention has been put on the morphology of PANInanostructure, as nanofibers and nanotubes. In general, PANI nano-structures are achieved by a ‘‘template synthesis’’ route using zeolitechannels, track-etched polycarbonate and nanosized alumina membrane astemplates, which address the growth of the nanostructures.75 Even thoughthis method allows the perfect control of the length and diameter of theproducts, by template selection, in most applications the template must beremoved, requiring additional workup and causing disorder or modificationof the micro/nanostructures. It has been demonstrated that the morphologyand chemical properties of PANI are closely associated with the preparationmethod and many synthetic procedures have been experimented.76 Gener-ally, aniline polymerization is performed through oxidative coupling ofaniline or its dimer, N-(4-aminophenyl)aniline, using oxidants such as am-monium persulfate (APS), K2Cr2O7, KIO3. The most common oxidant usedfor the preparation of conducting polymers is APS, but its inorganic by-product (ammonium sulfate) represents a limit for further applications.77

The use of other metals in a high oxidation state does not overcome thisdrawback.75–79 A recent application of chloroauric acid (HAuCl4), asan oxidant for the polymerization of aniline, resulted to be effective forachieving nanofibers, nanotubes,80 and nanoballs.73 The introductionof metal species can deeply influence the electronic and chemical propertiesof the polymer. Chattopadhyay and co-workers reported a new route forsynthesizing Au–PANI composites based on the use of H2O2 as thebifunctional reagent for the reduction of HAuCl4 and the oxidation ofaniline, leading to the formation of interesting PANI–gold composites.78

300 | Catalysis, 2010, 22, 279–317

NH

HN

HN

HN

HN

HN

HN

HN

2

Leu

coem

eral

dine

NH

NN

NH

NH

NH

NH

N

Em

eral

dine

Con

duct

ing

Em

eral

dine

NH

NH

NH

NH

N+

N+

N+

N

HH

H

H

Cl-

Cl-

Cl-

Pern

i gra

nilin

e

NH

NN

NN

NN

N

Fig.4

OxidationstatesofPANI

Catalysis, 2010, 22, 279–317 | 301

Metal particles can be introduced in the polymeric framework also in twoseparate steps.81 Mallick et al.82 have described further preparation methodsand applications of gold–polyaniline composites. Nanoparticles of pre-formed metallic gold have not yet been used as a catalyst in the oxidativepolymerization of aniline, but Bicak and Karagoz performed the synthesis ofemeraldine base from aniline and gaseous oxygen with Cu(II) as the cata-lyst.83 The use for organic solvents and a soluble catalyst, however, makesthis route far to be a real eco-friendly method, even though the achievedyields were interesting. Also the application of H2O2 for aniline polymer-ization appears to be attractive, mainly for large-scale applications, becauseits reduction product, H2O, improves recycling of the reagent. A limit,however, is the slow reaction rate which is prompting research to findsuitable catalytic systems for improving the kinetics:77 Sivakumar andGedanken have successful applied ultrasonic irradiation for achieving con-ductive polyaniline84 and Au–polyaniline composites have been produced inthe presence of chloroauric acid.78,84 The exciting performance of goldnanoparticles as the catalyst in the oxidative polymerization of pyrrole85

(Section 5.4), has suggested the investigation on this catalytic method foroxidising aniline. While gold was shown to be almost inert towards theaerobic oxidation of aniline, polyaniline could be synthesized using hydro-gen peroxide as the oxidant under mild conditions.86 As pointed out before,the request of a simple and clean preparation method is required for par-ticular technological applications, such as changeable conducting ma-terials,87 electronic displays,88 electrode materials,89 molecular electroniccircuit elements,90 restoration of data,91 detectors92 and biochemical analy-sis.93 In fact, the electrical conductivity of PPy, due to the electrons hoppingalong and across the polymer chains with conjugating bonds,94,95 is par-ticularly sensitive to residues of reagents and organic solvents disturbing theco-planarity between interchains.96 Similar considerations must be taken inconsideration for supporting new catalytic procedures for the synthesis ofconducting 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, de-pending on the conditions of the reaction, but the mechanism of their for-mation has not yet been elucidated. When aniline is oxidized in an acidicaqueous medium with ammonium peroxydisulfate, a PANI precipitate isproduced. The blue pernigraniline form, present during the polymerization,converts into the green protonated emeraldine at the end of the

N

H

N

H

N

H

N

Hn

Fig. 5 Polypyrrole

302 | Catalysis, 2010, 22, 279–317

polymerization. The reaction is exothermic and leads, besides PANI for-mation, to sulfuric acid as a by-product. The progress of polymerization canbe followed in situ by recording either the temperature or the pH. It has beenrecently reported104 that the mechanism of aniline oxidation with ammo-nium peroxydisulfate in aqueous solution of strong (sulfuric) or weak(acetic) acids is substantially different. In sulfuric acid solution a granularPANI was produced; in acetic acid solution, on the contrary, PANI nano-tubes were obtained. It has been demonstrated that aniline polymerizationproceeds well even in water, without any added acid, when ammoniumperoxydisulfate was used as an oxidant. The sulfuric acid produced by thedecomposition of peroxydisulfate, in fact, gradually provides the necessaryacidity and the final PANI is protonated with this acid. Nevertheless, theconductivity of PANI prepared with this simple method is rather low.

The use of other oxidants, such as oxygen and hydrogen peroxide, in thepresence 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; onadding a gold catalyst (Aniline: Au=100–1000) the insoluble green polymer‘‘emeraldine’’ is formed. Purification of the product can be achieved byextraction with 1-methyl-2-pirrolidone and evaporation of the solventunder vacuum.

Among many catalytic systems, as colloidal gold, gold supported on car-bon and gold supported on TiO2, this latter produced the highest yield86 Theaniline/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=1.23V). As H2O2 presents a higher value of E1=1.78V, onecould expect this oxidant to be effective in affording the conductive polymer.Actually, while no aerobic oxidative polymerisation of aniline was observedusing different catalytic systems with O2 at 3 bar and room temperature, amodest catalytic effect (typically 4–5% yield) was detected when a smallamount of colloidal gold (Au:Aniline=0.001) was used as a catalyst incombination with the H2O2 reagent. No product was isolated without anycatalyst, and no benefit was recorded from using H2O2 in excess (Table 10).

However, the PANI yield could be increased by increasing the goldamount in the range Au:Aniline=0.001–0.004 (molar ratio) reaching theasymptotic value of 27% in 24h-tests (Fig. 6).

The humble catalytic life of gold ‘‘naked’’ particles in oxidation reactionscould be the cause for the limited conversion of aniline to PANI. The

Table 10 Oxidative polymerization of aniline in the pres-

ence colloidal gold and different H2O2 amounts

Test Au:Aniline H2O2:Aniline Yield%

1 0 1 0

2 0.001 1 4.8

3 0.001 2 4.8

4 0.001 4 4.0

Catalysis, 2010, 22, 279–317 | 303

oxidative polymerisation of aniline was consequently experimented also inthe presence of supported gold catalyst (0.5% Au/C, 1% Au/TiO2), whichwas proven to be more stable in former catalytic applications.17 Theachievements, reported in Table 11, show improved performances.

In particular, the high activity of Au/TiO2 can be ascribed to a strongcontribution of the TiO2 support. Differently from unloaded carbon, whichwas inert in aniline polymerisation by H2O2, P25 titania catalyzed thepartial oxidation of aniline to soluble dark oligomers. However, no solidmaterial was formed. The products obtained in all of the preparations wereidentified as ES (emeraldine salt) according to the FT-IR.81 UV–vis,74 andXRD spectra75 and84 reported in Figs. 7–9.

The morphology of the products, investigated by transmission electronmicroscopy (TEM) and scanning electron microscopy (SEM), revealedemeraldine in form of nanospheres of 44–160 nm (Fig. 10A) alternating withmicrometric rods (Fig. 10B and Fig. 11).

Concerning the morphological properties, similar nanospheres were ob-tained for the Au/C and TiO2 catalysts. In the case of Au/TiO2, however, thenanospheres were assembled in a cluster-like organization, which was absentin the product derived using the carbon-supported catalyst. The conductivityof 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 wasdetermined with a standard conductivity cell (CON–H Material Mates).

In conclusion, it has been shown that conductive PANI, mainly in formof nanospheres, can be easily fabricated from aniline by H2O2 oxidation inthe presence of gold nanoparticles as a catalyst. The conductivity values,correlated with the bulk resistance of this polymer, are similar to thoseobtained through other polymerisation methods.84,106

0

5

10

15

20

25

30

0 0.005 0.01 0.015 0.02

Au/Aniline

Yie

ld %

Fig. 6 Dependence of aniline polymerization by the amount of colloidal gold.

Table 11 Polymerization of aniline by supported gold catalysts.

(Aniline:Au=1000)

Test Catalyst Aniline/H2O2 Yield%

1 0.5% Au/C 1 11.4

2 1% Au/TiO2 1 70.1

304 | Catalysis, 2010, 22, 279–317

5.4 PANI-based composites

Conducting polymers have been proven to be suitable host matrices fordispersing metallic particles. Conducting polymer composites with metalnanoparticles allow a facile flow of electronic charges through the polymermatrix during electrochemical processes. Through a suitable combination ofconducting polymer and metal nanoparticles, novel promising electrodes

400 600 8000.0

0.5

A

Wavelength (nm)

Fig. 8 UV-vis spectrum of emeraldine salt in NMP as the solvent.

4000 3500 3000 2500 2000 1500 1000 500

10

15

20

25

30

35

40

45

%T

Wavenumber (cm-1)

Fig. 7 FT-IR spectrum of emeraldine salt in KBr pellet.

Catalysis, 2010, 22, 279–317 | 305

could be generated with higher surface areas and enhanced electrocatalyticactivities, particularly interesting in the fuel cell technology.

5.4.1 Synthesis of PANI/MCM-41. PANI is insoluble in commonsolvents and impossible to be molten. These drawbacks can be overcome bymodifying the polymer with the incorporation of inorganic materials.As PANI/inorganic nanocomposites combine the advantages of PANIand inorganic nanoparticles, extensive research has been carried out in thisfield. Silica nanoparticles, in particular, have received great attention be-cause of their unique properties and wide applications. Primarily, the mainobjective is to keep the conducting polymer in a stable colloidal form.

Fig. 10 TEM images of PANI synthesized with (A) (Bar=200nm), H2O2 and (B) (Bar=1mm),‘‘naked’’ gold nanoparticles.

20 40 60 80

100

200

300

400

500

600

700

Lin

(Cou

nts)

2-Theta-Scale

a

b

Fig. 9 X-ray diffraction pattern of emeraldine salt.

306 | Catalysis, 2010, 22, 279–317

Armes et al.107–110 succeeded in incorporating silica nanoparticles into thecore of PANI. Aniline was polymerized by ammonium peroxydisulfate in thepresence of silica colloids at low concentration of monomer and oxidant.This technique allows to slow down the rate and degree of polymerizationand promote the polymerization on the colloidal surface rather than into thebulk. Effectively, the outer PANI layer of PANI/SiO2 colloids becomessoluble to some extent. Porous silica possesses high surface area and the sizeof pores can be tuned from 2 to 10 nm in a narrow diameter distribution bychanging the experimental conditions. PANI synthesis inside porous silicachannels has attracted more attention. Wu and Bein111 fabricated a con-ductive filament of PANI in MCM-41, by using MCM-41 as the host whichwas contacted with aniline gas (not beneficial to the environment) at 313Kfor 24 h, thus obtaining significant conductivity. PANI/MCM-41 compositewas synthesized by chemical oxidative polymerization of aniline on thesurface of MCM-41 in the presence of HCl. It was shown that PANI waspresent not only on the surface of MCM-41 but also inside the pores.Moreover, the higher HCl concentration, the higher conductivity of thecomposite, because of a wider delocalization of the resulting emeraldine salts.

5.4.2 Preparation of PANI nanoparticles by Fe3O4. Colloidal particleswith magnetic dipole moments are able to self-assemble into flexiblechains. In particular, they can form chains with threefold junctions resultingfrom a delicate balance between the dipolar interactions. These chainscan be thought as a ‘‘living polymer’’ whose length and organizationare determined not by the reaction conditions, but by a thermodynamicbalance of forces. Water ‘‘soluble’’ Fe3O4 nanoparticles, with a coatingof polyethylene glycol nonylphenyl ether (NP5) and cyclodextrin (CD),were synthesized and used as templates for the preparation of PANI

Fig. 11 SEM images of PANI synthesized with H2O2 and ‘‘naked’’ gold nanoparticles(Bar=500 nm)

Catalysis, 2010, 22, 279–317 | 307

nanostructures.112 NP5 was used to stabilize the magnetic nanoparticles byformation of a suitable surface coating and CD for improving their misci-bility with water. Y-junction PANI nanorods and nanotubes have beensynthesized by the use of in situ self-assembled magnetic nanoparticles astemplates and pH control of the reaction system. It has been found that aninitial pH of the reaction system around 8–10 favours the formation ofnanorods, whereas a starting pH range of 5–6 leads to the formation ofnanotubes. The morphology of Y-junction PANI nanostructures dependson the reaction conditions such as on aniline concentration and the presenceof an organic solvent.

5.4.3 Preparation of PANI nanoparticles by Pt. The incorporation of Ptalone or Pt with secondary metals including Ru, Os, Sn and Mo into PANIhas been primarily focused on exploiting the catalytic activity of themetals.113 For example, polymer supported Pt has been used in the catalyticoxidation of methanol, formic acid and hydrazine, in addition to the re-duction of dioxygen. The reaction of methanol at Pt surfaces has been thefocus of numerous studies. Laborde et al.114 have studied the oxidation ofmethanol with a platinum modified polyaniline electrodes in acidic medium.The results indicate that methanol oxidation involves the direct formationof CO2 through reactive intermediates.

The synthesis of polyaniline/platinum composites has been achievedusing controlled electrochemical reduction of PtCl2�6 and PtCl2�4 . Theelectrochemical and chemical syntheses of PANI/Pt by reduction of PtCl2�6and PtCl2�4 produce morphologically different composite materials.

5.4.4 Preparation of PANI nanoparticles by CuSO4. A simple, one potand in situ chemical synthesis route for the preparation of polyaniline usescupric sulfate as an oxidizing agent. The reaction has been conducted inmethanol and the cupric sulfate has been added drop by drop. The size ofthe copper nanoparticles varied in the range 2–5 nm. The formationmechanism of micro or nanostructured polymer is not yet clear. In the firststep, we can assume that an oligomeric form of aniline can be formed due tothe presence of a suitable oxidizing agent. The oligomerized aniline thusproduced may act as a nucleation centre, which catalyses the oxidation ofthe remaining monomers present in the solution. The oxidizing agent getsreduced and forms copper nanoparticles that lead to the formation of ametal-polymer composite material, which could represent an importantadvantage for further applications. This method involving copper has beenextended to gold and palladium systems.74

5.4.5 Preparation of PANI-gold composites. A simple route for poly-aniline-gold composites synthesis uses tetrachloroauric acid (HAuCl4) asthe oxidizing agent, forming metallic gold nanoparticles (10–50 nm) at theend of oxidation. The reaction is conducted in toluene with a phase-transfercatalyst. Other works report the same reaction conducted in 1M HClaqueous73 or in a solution of D-camphor-10-sulfonic acid.115

A new method of synthesis of an Au nanoparticle-conducting polyanilinecomposite has been proposed using H2O2 both for the reduction of HAuCl4and 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 oxidationby using gold catalysis.85 Stirring pyrrole (Py) in acidic (HCl) aqueoussolution in air at room temperature (292–295K) and in the presence ofcolloidal gold a slow oxidation takes place producing a dark product. Theobserved yields, up to 75%, are depending on the total gold amount in therange Py:Au=1000–10000, as reported in Table 12.

An acceleration of the reaction can be obtained by carrying out the re-action under pure oxygen as reported in Table 13 thus allowing a higheryield (82%) of polypyrrole in less time (24 h).

A great improvement of pyrrole polymerisation has been observed byusing H2O2 instead of O2 as the oxidant.

In this case we must outline that pyrrole plus HCl, dissolved in waterunder nitrogen atmosphere, undergoes a slow oxidative polymerisationalso in the absence of catalyst, producing polypyrrole in 57% yield in 24 h(Table 14). The addition of gold improves the polymerisation which can beimplemented to 99% of the product.

Fig. 12 visualizes the catalytic effect of gold in polypyrrole formation withvarious oxidative agents. Fig. 13 reports the kinetics of polymeric material

Table 12 Aerobic polymerization of pyrrole with air.

Reaction time 3 days

Test Py/Au (molar ratio) Yield%

1 No gold 0

2 10000 42

3 5000 50

4 1000 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 8 No gold 0

6 8 10000 0

7 8 1000 1

8 16 No gold 5

9 16 10000 31

10 16 1000 34

11 24 No gold 10

12 24 10000 50

13 24 5000 60

14 24 1000 82

15 72 No gold 12

16 72 10000 61

17 72 5000 72

18 72 1000 99

Catalysis, 2010, 22, 279–317 | 309

formation, during pyrrole polymerisation by dioxygen in the presence ofdifferent quantities of gold.

Using pure oxygen gas, a slow auto-oxidation produced a modest yield ofinsoluble polymer (12%) after 3 days. On adding colloidal gold, a strongcatalytic effect was detected, as the polymeric material was formed withalmost total yield (99%) after 3 days (Fig. 13).

Using hydrogen peroxide as the oxidant, a consistent polymerization ofpyrrole was observed also in the absence of catalyst. In fact, the reactionoccurred with 57% yield with respect to pyrrole in 24 h. However, also inthis case, a catalytic contribute of gold was demonstrated because the yieldincreased up to 99% by adding a small quantity of colloidal metal (Fig. 12).

A concert of different analytical techniques allowed the characterizationof the obtained polypyrrole structure.116–119 Independently from the syn-thetic method, all the prepared polymers show a similar IR spectrum whichis represented in Fig. 14.

Fig. 15 shows typical X-ray diffraction patterns of PPy synthesized in thepresence and in the absence of gold nanoparticles.120

The TEM images of the various polymeric materials are reported inFigs. 16 A–E. The comparison among the different products highlights howmuch the morphology is affected by the nature of the oxidising reagent. Byusing H2O2, the quick polymerization led to a partially reticulated structure

0

20

40

60

80

100

120

0 0.0002 0.0004 0.0006 0.0008 0.001 0.0012

Au/PY

O2, 24h

H2O2, 24h

O2, 3 days

Air, 3days

Yie

ld %

Fig. 12 Gold effect in the polymerization of pyrrole by air, O2 and H2O2.

Table 14 Catalytic polymerization of pyrrole using

H2O2 as the oxidant. Reaction time 24 h

Test Py/Au (molar ratio) Yield%

19 No gold 57

20 10000 66

21 5000 90

22 1000 99

310 | Catalysis, 2010, 22, 279–317

without gold (Fig. 16A) and to amorphous material using gold catalysis(Fig. 16B).

More ordered structures were achieved by the non-catalyzed polymer-ization with gaseous oxygen (Figs. 16C, D). Abundant peculiar thin squareswere observed in the high yield polymerization catalysed by gold (Fig. 16E).This latter structure is unusual also in the context of conventional poly-merizations of pyrrole and could be interesting for tailor-made compositeapplications.

The conductivity of the synthesized polypyrrole materials is similar tothat obtained in conventional chemical polymerization using stoichiometric

0

20

40

60

80

100

0 8 16 24 32 40 48 56 64 72

Reaction time (h)

no gold

Py/Au=10000

Py/Au=1000

Yie

ld %

Fig. 13 Kinetic data of pyrrole polymerization under O2.

3200 2400 1600 80010

20

30

40

%T

Wavenumber (cm-1)

Fig. 14 FT-IR spectrum of a PPy sample in KBr.

Catalysis, 2010, 22, 279–317 | 311

reagents and it was determined with the standard conductivity cell (CON–HMaterial Mates) also used for polyaniline (Section 5.2).121,122 The s values,determined for various preparations and measured in samples of differentthickness under different voltages (Table 15), ranged from 2.7� 10� 4 to

20 40 60 80

100

200

300

400

500

600

700

Lin

(Cou

nts)

2-Theta-Scale

a

b

Fig. 15 X-ray diffraction patterns of PPy synthesized in the absence (a) and in the presence (b)of gold nanoparticles

A) By H2O2, no Au B) By H2O2, Py/Au=1000.

C) By air,no Au. D) By O2,no Au E) By O2,Py/Au=5000

200 nm 100 nm

200 nm 100 nm 200 nm

Fig. 16 TEM images of different polypyrrole samples.

312 | Catalysis, 2010, 22, 279–317

5� 10� 3 S� cm� 1. These conductivities are typical for polymers obtainedby 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 comparisonof s values is reported in Table 16, showing how the wide range in theconductivities (from 10� 10 to 102 S� cm� 1) of the differently synthesizedpolymers allows a wide choice of application of these materials, accordingto the required conductivity.

6. Conclusions

The last twenty years represent a milestone in the catalytic selective oxi-dation of organic compounds under eco-friendly conditions and this state offerment goes on. In this context, a key role has been played by gold, whosecatalytic properties are rapidly flowing into applications rich in promisingresults. Starting from liquid phase processes carried out under mild con-ditions, gold catalysis is fast extending to gas phase processes at highertemperature with comparable perspectives. The focus of the extraordinarypeculiarities of gold in catalysis is often represented by its highly dispersednanometric particles.

A novel scenario is now emerging: the application of gold in materialscience, particularly promising in the facile polymerization of aniline andpyrrole via environmentally friendly routes. Even though methods andtechniques have still to be refined, the progress in oxidation reactions bygold and its successful versatility will contribute to the definitive decline ofthe so-called ‘‘stoichiometric oxidants’’, so far used in organic synthesis butproducing undesired wastes.

Catalysis appears to be the powerful tool for making environmentallyfriendly processes a wide reality.

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) 2.46� 10� 3 5.88� 10� 4 4.80� 10� 4

3 (1.10) 5.09� 10� 3 1.38� 10� 3 3.84� 10� 3

4 (0.63) 8.94� 10� 4 1.39� 10� 3 4.09� 10� 3

18 (0.64) 8.47� 10� 4 9.97� 10� 4 7.23� 10� 4

19 (0.60) 1.33� 10� 3 6.17� 10� 4 3.34� 10� 4

20 (0.67) 4.81� 10� 3 2.70� 10� 3 4.85� 10� 3

21 (0.64) 9.00� 10� 4 6.04� 10� 4 4.56� 10� 4

22 (0.68) 2.74� 10� 4 5.53� 10� 4 4.97� 10� 4

Table 16 Comparison among different conductivity data

Research groups Conductivity (S� cm� 1)

Rossi et al.85 2.7� 10� 4� 5.1� 10� 3

Chao and March116 5.0� 10� 3� 2.8

Toshima and Tayanagis125 10� 2

Izumi and Toshima124 10� 10� 2.1� 10� 2

Diaz and Fanazawa117 10–100

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 Sudzucker, 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. Dimi-

tratos, 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. Com-

mun., 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. Moller, 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