Catalytic Oxidation
Catalytic Oxidation
Catalytic Oxidation
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Catalytic
Oxidation
Principles and Applications
Editors
R.A. Sheldon
Delft Univ. of Technology
R.A. van Santen
Eindhoven Univ. of Technology
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
; World Scientific
Copyright World Scientific Publishing Co.
Singapore'NewJerseyLondoirHongKong
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
Published by
World Scientific Publishing Co. Pte. Ltd.
P O Box 128, Farrer Road, Singapore 9128
USA office: Suite IB, 1060 Main Street, River Edge, NJ 07661
UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE
All rights reserved This book, or parts thereof, may not be reproduced in anyformor
by any means, electronic or mechanical, including photocopying, recording or any
information storage and retrieval system now known or to be invented, without written
permission from the Publisher.
For photocopying of material in this volume, please pay a copying fee through the
Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923 USA.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
ISBN 981-02-2186-X
Printed in Singapore.
Selective catalytic oxidation is a key technology for converting oil and natural gas-
derived feedstocks to a wide variety of bulk chemicals. Moreover, in the wake of increa
singly stringent environmental legislation, attention is being focused on the development of
'greener' processes, such as catalytic oxidation, for the manufacture of fine chemicals. A
thorough understanding of fundamental mechanistic pathways of catalytic oxidations is of
paramount importance for the improvement of existing processes and the development of
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
new ones. Hence, the substantial interest, in both industrial and academic laboratories, in the
mechanisms of oxidation catalysis,
Compared to hydrogenations, for example, which are mechanistically relatively
straightforward processes, oxidations are enormously complex. For starters, molecular
oxygen, in contrast to hydrogen, reacts with most organic molecules even in the absence of
a catalyst. Hence, a thorough knowledge of these free radical chain processes (autoxidati-
ons) is an essential ingredient for understanding oxidation catalysis.
Catalytic oxidations with molecular oxygen (dioxygen) can be divided into two
types: heterogeneous, gas phase and homogeneous, liquid phase processes. Researchers
generally affiliate themselves with one or the other group and the two tribes speak largely
different languages. Consequently, there is tittle cross-fertilization of ideas and, moreover, a
third type-heterogeneous catalysis in the liquid phase- tends to suffer from lack of attention
by either group.
A major aim of this book is to provide a sound mechanistic basis for understanding
catalytic oxidation processes, which should be useful to reseachers in the field, irrespective
of their tribal affiliation. It is based on a course on oxidation catalysis held in Rolduc, The
Netherlands, in June 1994, under the auspices of the Dutch Research School in Catalysis
(NIOK). The course was given by international authorities from industry and academia in
the fields of both gas and liquid phase oxidations. It was targeted mainly at postgraduate
research students wanting to acquaint themselves with the basic principles and industrial
applications of oxidation catalysis. Moreover, it was hoped that the course would foster a
synergistic cross-fertilization of concepts and ideas between the aforementioned groups.
The participants had backgrounds ranging from surface science to organic synthesis.
v
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
VI
autoxidations The intricate mechanism of the Amoco process for the catalytic oxidation of
p-xytene to terephthalic acid is discussed in detail. Chapter nine (Sheldon) continues with
an elaboration of the various types of heterogeneous catalysts for liquid phase oxidations.
Particular emphasis is placed on molecular sieve catalysts containing redox metal ions
incorporated in the framework or metal complexes encapsulated in the micropores (ship-in-
the-bottle catalysts).
Chapter ten (Moiseev) elucidates the mechanistic roles of a and re complexes in
palladium-catalyzed oxidations of olefins. The oxidative acetoxylation of ethylene and
propylene, to vinyl acetate and allyl acetate, respectively, catalyzed by giant palladium
clusters is discussed. The latter can be considered as bridging the gap between homogene
ous palladium complexes and a palladium metal surface. Chapter eleven (Sheldon) is
devoted to the application of catalytic oxidations in fine chemicals synthesis. The charade-
ristics of fine versus bulk chemical manufacture and catalytic oxidation versus oxygen
transfer are explained. It also includes a section on catalytic asymmetric oxidation. This
part of the book concludes with a chapter (van Veen) on selective electrochemical oxidati
ons, encompassing both ejtsiui and in-situ generation of metal oxidants. Attention is also
focused an new developments in electrode materials and solid polymer electrolyte cells.
The third part of the book, chapter thirteen (Mills, Harold and Lerou), returns to
the subject of heterogeneous gas-phase processes with a comprehensive overview of cataly
tic reactor technology. The advantages and limitations of existing and emerging catalytic
reactors are discussed.
The book contains more than 600 literature references and a thoroughy cross-
referenced index, Hopefully it will be widely used by aspiring researchers in this fascinating
and economically important field. Finally, the editors would like to express their sincere
thanks to their friends and colleagues who have contributed chapters to this book and
Frank Sheldon for preparing the index.
RASheldon
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Preface vii
Index 373
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
R.A. SHELDON
Laboratoryfor Organic Chemistry and Catalysis,
Delft University of Technology, Julianalaan 136,
2600 GA Delft, The Netherlands
ABSTRACT
Processes for the manufacture of industrial chemicals by catalytic oxidation of feedstocks
derivedfromoil or natural gas are reviewed Both heterogeneous, gas phase and homogeneous liquid
phase processes are discussed and different processes to particular products, such as acetic acid,
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
propylene oxide, phenol and cyclohexane, are compared. Recent trends in chemicals manufacture are
also outlined. Finally, thereaderis introduced to mechanistic aspects of metal oxidations.
1. Introduction
Several important liquid phase processes were developed during the 1950's and 1960's. Examples
include the Wacker process for ethylene oxidation to acetaldehyde, the Celanese process for n-
butane oxidation to acetic acid and the Amoco/Mid-Century process for the production of
terephthalic acidfromp-xylene (figure 1).
Although, as noted above, catalytic oxidation is the most favorable technology for the
manufacture of many industrial chemicals this is not always the case. The most important process
for the production of acetic acid, for example, is the rhodium-catalyzed carbonylation of methanol
developed by Monsanto9 (seefigure2). This process has the advantage of high selectivity (99%)
coupled with cheap raw materials (methanol and carbon monoxide).
Several important gas phase oxidation processes are outlined in figure 3. These processes
were developed in the 1950's or earlier, although process improvements are still being made, e.g.
in maleic anhydride manufacture from n-butane (see chapter 12).
The manufacture of methylmethacrylate (MMA) is an interesting case in point for
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
comparing different technologies (see figure 4). The classical process involves the methanolysis of
acetone cyanohydrin, has an atom utilization5*8 of 29%, and produces 2.5 kg. of ammonium
bisulfate per kg of MMA. More recently Atahi and Mitsubishi have developed alternative
processes 10 based on the gas phase, catalytic oxidation of isobutene to methacrolein analogous to
the oxidation of propylene to acrolein (see figure 3). Alternatively, aiethacrylic acid can also be
prepared in a two-step, oxidation/dehydrogenation of isobutyraldehyde. The latter is a byproduct
of propylene hydroformyhrtion to n-butyraldehyde. Routes have also been developed based on
ethylene carbonylation or hydroformylation10 (see figure 4). Finally, Shell workers n have
reported an elegant one-step synthesis of MMA by methoxycarbonyiation of methylacetylene, the
latter being available as a byproduct of naphtha Cracking.
1
ca. 1950 CH3CHO + O2 ^°ruCoII» CH3CO2H
6075 bar
O2H
CoD/HOAc
ca. 1960 CH3CH2CH2CH3 + O2 , CH3CO2H
150-225760 bar
PdCiyCuCh
H2C=CH2 + O2 ■- CH3CHO
10077 bar
Figure 1. Liquid phase Oxidation Processes.
CH3CO2H
O2, CC-.tRbl/Cflfel]
[CoD] or [MnJI] 150-225°
40-50 bar 150°/15 bar
C4 CH3CH2CH2CH3 CH3OH C,
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Celanese Monsanto
Sel. =75-80% Sel. =99%
Figure 2. Acetic Acid Manufacturing Processes.
CH3CH=CH2
400°C
O2
H2C=C«CN
NH 3
Sohio, 1959
Q
0C ^^ O^
V2Q5/P2Q5.
CH3CH2CH2CH3
Propylene oxide was traditionally produced via the chlorohydrin process. However, this
low atom utilization process produces about 2 kg of CaCl2 per kg of propylene oxide. The Arc©
process3,12, in contrast, utilizes an alkyl hydroperoxide, e.g. tert-butyl hydroperoxide, in the
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
J HO CN
i) ^x^ +HCN ► y^
CO,Me
MeOH
■*- / k . * NH 4 HS0 4
H,SQ4
Atom utilization = 29% ; 2.5 kg salt per kg MMA
CHO CO,H
—►MMA
catalyst ^ catalyst ^
gas phase gas phase
CHO CO,H C0 2 H
3
> -^"s. —•* - ^ \ . * » ^S^ +H.O
catalyst \ *
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
COaH
CO/H 2 H2CO
catalyst C O
J H ♦A.
4) H,C=CH,_^
oCHO
n
catalyst " 2. O 2
COzMe
5) C H , C = C H + CO + MeOH
catalyst
* A.
99% s e l .
CHLOROHYDRIN PROCESS
Cl
HOCI * JL ^OH
CATALYTIC EPOXIDATION
Arco process :
^ + o, . X -
OaH [Mo 1 ^ O . . O H
0 M e
OH \ ^
>><^ + MeOH ► ^X' + HaO
MTBE
Shell SMPO process ;
O.H
Ph^ ^ + wo» *•
■" P h ^ \
Ph
02H OH
+ +
PH^\ -^- * y*Q P h ^ -
OH
+H
Ph^"^ *" PrT^ *°
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
As noted above, phthalic acid (anhydride) is produced by gas phase oxidation of o-xylene
over a vanadium oxide catalyst. In contrast most other aromatic carboxylic acids are produced by
metal-catalyzed autoxidation of the corresponding toluene in the liquid phase (reaction 1).
Catalyst
ArCH3+r/202 1 ► ArC0 8 H + H 2 0 (1)
Traditionally many aromatic carboxylic were produced by nitric acid oxidation of the
corresponding toluene or by side chain chlorination followed by hydrolysis. Unfortunately, these
processes suffer from serious drawbacks: large amounts of inorganic salt-containing effluent and
chloro(nitro) compounds as impurities. Catalytic oxidation, on the other hand, is a high atom
utilization, low salt process with no chloro(nitro) compounds as impurities.
Benzoic acid, for example, is produced commercially by the cobalt-catalyzed autoxidation
of toluene (see chapter 8). The most important aromatic carboxylic acid is, however, terephthalic
acid (see Table 1). It is produced by autoxidation of p-xylene in acetic acid in the presence of a
cobalt catalyst and a promotor : either acetaldehyde (Eastman-Kodak and Toray processes) or
bromide ion (Amoco/MC process)13. The two processes are compared in Table 2. The
mechanism of the bromide ion promoted catalysis is discussed in chapter 9.
Cooxidation Bromide-mediated
(Eastman Kodak/ Toray) (Amoco-MC)
Catalyst Co(OAc)2 Co(OAc)2/Mn(OAc)2
Solvent HOAc HOAc
Promotor CH3CHO Br"
Temp. (°C) 100-140 195
Pressure (bar) 30 20
Conv./ Sel. (%) >95/ >95 >95/>95
Advantage Less corrosion Lower [catalyst]
Disadvantage Coproduct HOAc (0.21 %) Corrosive (Ti-lined reactor)
The two industrial routes to phenol are outlined in figure 6. The benzene-based cumene
process is more selective but this tends to be offset by the lower price of toluene and the lower
number of steps (two compared to three). Direct hydroxylation of benzene remains a potentially
attractive alternative but up till now yields and/or productivities are much too low to be
competitive.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
10 bar [Co 11 ]
C0 2 H
0^ 220°
Or
0 2 /H 2 0
90% sel.
11
1 bar [Cu ]
OH
0^ Or 80 - 90% sel.
[HI
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
OH O
Assahi Chemical has recently commercialized yet another variant involving the selective
Sgiflrogenation of benzene to cyclohexene over a ruthenium catalyst, followed by zeolite-catalyzed
Bpttration to cyclohexanol and subsequent dehydrogenation to cyclohexanone. This process has
tttie advantages of higher conversions and selectivities than cyciohexane autoxidation. All of these
(processes are ultimately derived from either benzene or toluene as the basic building block.
CATALYST
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Another industrial chemical that is derived from cyclohexanone is adipic acid. Traditional
processes involve the autoxidation of cyclohexane to a mixture of cyclohexanol and cyclohexanone
followed by nitric acid oxidation in the presence of a copper/vanadium catalyst (figure 9).
H,
NH. HNO, NH2OH • HX
[Pt] [Pd]
EXISTING ROUTE
AMMOXIMATION
a
NH3 + H a 0 2 +
However, these processes suffer from the disadvantages of low conversions and/or selectivities
and the inherent drawbacks of nitric acid oxidation (see earlier). On the other hand, direct
oxidation of cyclohexane to adipic acid is not competitive due to low selectivities. Hence, much
effort has been devoted in recent years to the development of alternative processes based on
the catalytic carbonylation of butadiene . In this context it should also be mentioned
that DuPont produces adiponitrile via nickel-catalyzed addition of two equivalents of HCN to
butadiene.
The current trend in chemicals manufacturing is towards processes that are more efficient
in their use of raw materials and energy and more environmentally benign. Catch phrases that are
widely heard in the industry are integrated waste management, zero emission plants and benign-
by-design. There is a marked trend away from waste remediation, i.e. end-of-pipe solutions,
towards the development of processes that do not generate waste in the first place, i.e. primary
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
pollution control. The emphasis is clearly on higher selectivities rather than yields. Furthermore,
catalytic oxidations are being widely applied in the manufacture of fine chemicals ' . Another
identifiable trend is towards the use of cheaper feedstocks which is exemplified by the current
drive to replace alkenes by alkanes, e.g. acrylonitrile by ammoxidation of propane rather than
propylene. In this context it is worth mentioning that there are several "dream1 reactions (eqns 2-
4) for which there is enormous commmercial potential if an efficient catalytic system could be
found.
O
1 / \
C H , C H = C H 2 + /,o 2 CH3CH — C H , (2)
The ground state of dioxygen is a triplet containing two unpaired electrons with parallel
spins. Hence, the direct reaction of 0 2 with singlet organic molecules to give singlet products is
a spin forbidden process with a very low rate. Fortunately for mankind these unfavorable kinetics
preclude the spontaneous combustion of living matter into carbon dioxide and water, a
thermodynamically very favorable process.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
One way of circumventing this energy barrier is via a free radical pathway. The reaction of
a singlet molecule with 3 0 2 (reaction 5) forming two doublets (free radicals) is a spin-allowed
process. It is, however, highly endothermic (up to SO kcal/mol) and is observed at moderate
temperatures only with very reactive molecules that form resonance stabilized radicals, e.g.
reduced flavins (reaction 6). This is a key step in the activation of dioxygen by flavin-dependent
oxygenases.
H H
OH O
A second way to overcome the spin conservation obstacle is via complexation of 0 2 with
a paramagnetic (transition) metal ion (reaction 7). The expectation that the resulting metal-
dioxygen complex could react selectively with hydrocarbons at moderate temperatures provided
the stimulus for extensive studies of dioxygen activation by metal complexes during the last three
decades19. The various oxygenated species that can result from the interaction of metalcomplexes
with dioxygen, and may play a role in subsequent reactions with organic substrates, are depicted
in figure 10.
M" 1
0
superoxo
Mn+1 o n+2^-0
^ O
M- peroxo peroxo
O
II ^ o
0
oxo dioxo hydroperoxo
Basically, three different types of oxidation with dioxygen can be delineated with respect
to the mechanism involved (figure 11). The first type involves the generation of chain-initiating
radicals via the metal-catalyzed decompostion of alkyl hydroperoxides. The latter are omnipresent
in hydrocarbons that have not been rigorously purified by, for example, column chromatography
over basic alumina. Since this always constitutes a background reaction in hydrocarbon oxidations
it has led to many erroneous interprations of results in studies of 'oxygen activation' by metal
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Liquid phase
M n ' 1 + R0 2 H ■ M n + R O . + HO-
M n + R0 2 H M n " 1 + R0 2 « + H +
Net : R0 2 H R O . + R O , . + H20
R0 2 • (RO.) + RH R. *R02H(ROH)
R. + O, R02.
2. Coordination catalysis
«) H2c =
11
CH 2 + Pd + H 2 0 CH,CHO + Pd° + 2H +
[Cu«J
Pd° + V 2 0 2 + 2H + P d " + H20
H
b) \ / ♦ Pd"
\ o
. - C= O + Pd + HaO
OH /
Gas phase (Mars - van Krevelen mechanism)
Mn = O+S » M
n
"2 + so
Mn"2 + 0 2 ——► 2 M n
= 0
In the third type, which pertains mainly to gas phase oxidations, an oxometal species
oxidizes the substrate and the reduced form is subsequently reoxidized by dioxygen. This is
generally referred to as the Mars van Krevelen mechanism. With regard to the sequence and type
of redox events it closily resembles the type 2 mechanism.
A fourth type of mechanism comprises catalytic oxygen transfer processses whereby a
substrate reacts with an oxygen donor, such as H 2 0 2 or ROOH, in the presence of a metal
catalyst (reaction 8).
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
catalyst
S + XOY —*. SO + XY (8)
References
Jerzy Haber
Institute of Catalysis and Surface Chemistry,
Polish Academy of Sciences, Krakow, Poland
ABSTRACT
Molecular oxygen contains in its ground state two unpaired electrons, which are localized on
the degenerate antibonding Jt2p orbitals, the ground state is thus a triplet Because of the rule
of spin conservation, reactions between this triplet oxygen and organic molecules which are in
the singlet state experience high activation energies. This symmetry barrier may be overcome
either by activating oxygen to the singlet state or by activating the organic molecule to make it
susceptible to the reaction with molecular or atomic ions. Of particular importance is the
nucleophilic attack by the oxide ion 02".
Dioxygen, adsorbed at oxide surfaces, is present mainly in form of superoxide ions. The
existence of peroxide ions has not been proved directly, however it cannot be excluded a priori
At higher temperatures both adsorbed superoxide and peroxide dioxygen species are unstable
and presumably decompose with the formation of the ionradicalspecies O". All three activated
oxygen forms - the neutral singlet O2 and the ionic O2" and O" species- are strongly
electrophilic reactants which attack the organic molecule in the region of its highest electron
17
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
18
density, i.e. the n bonds (Fig.l). At variance with their behaviour in the liquid phase, the
peroxy and epoxy complexes formed as the result of an electrophilic attack of C>2' or O"
species on the n bonds of hydrocarbon molecule at the surface of an oxide are intermediates
which lead under heterogeneous catalytic reaction conditions to the degradation of the carbon
skeleton. In the case of olefins saturated aldehydes are formed in the first stage which are
usually much more reactive than unsaturated aldehydes or anhydrides and at higher
temperatures undergo a rapid total oxidation.
The presence of O2" or O" at the surface of an oxide may be detected by different techniques
such as EPR or IR spectroscopy. The method which permits the quantitative determination of
the number of electrons transferred between the solid and the adsorbed layer is the
BtctropNMc OKygtn ( I I I I I I I
-C-C-C-C-OH
spodts
1TV I I
Or Oi. 0- - I I l I I I I 1
-c-c-c-o ♦ c-o
-c-c-c-c- 1 T 1
M
activation of
hydrocarbon
-rtn" —c-00 ♦ -c—c-o
I I
oxidation of
mttat to form /jj\ 'H X ,
oxidttottico V > * ^ -
I I I I
- C — C-C-C—
NucttophUie OKvgtii
sptcits
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
measurement of the changes of work function in the course of adsorption. Thus, when
changes of work fucntion due to exposure to oxygen are followed upon temperature variation
and the amount of oxygen adsorbed is simultaneously measured, the number of elecrons
localized per oxygen atom adsorbed may be determined and hence the type of oxygen species
residing at the surface may be found. Results of such experiments carried out with different
oxides are summarized in table 1 and compared with catalytic properties of these oxides. They
indicate that whenever electrophilic oxygen species O2" or 0~ are present at the surface, total
oxidation is observed in the course of the catalytic oxidation of hydrocarbons.
The second route of heterogeneous oxidation is the reaction with lattice oxide ion
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
19
O2". These ions have no oxidizing properties but are nucleophilic reactants which can be
inserted into the activated hydrocarbon molecule by a nucleophilic addition, to form an
oxygenated product. This reaction path starts with the activation of organic molecules, which
thus become prone to undergo an attack by nucleophiles, and consists of a series of
consecutive steps of hydrogen abstraction and nucleophilic oxygen addition, with each of these
steps requiring different active centres to be present at the catalyst surface. It should be
emphasized that it is the cations of the catalyst which act as oxidizing agents in some of
Table 1.
Oxygen species at surfaces of various oxides
the consecutive steps of the reaction sequence, forming the activated hydrocarbon species.
In subsequent steps these undergo a nucleophilic attack by lattice oxygen ions and the
oxygenated product is desorbed, leaving oxygen vacancies at the surface of the catalyts.
Such vacancies are then filled with oxygen from the gas phase, simultaneously reoxidizing
the reduced cations. It should be noted that incorporation of oxygen from the gas phase into
the oxide surface does not necessarily take place at the same site from which surface
oxygen is inserted into the hydrocarbon molecule, but may occur at a different site, oxygen ions
being then transported through the lattice. This mechanism may be represented by the
cycles shown in Fig. 2. In the case of complex hydrocarbon molecules, the nucleophilic
addition of oxygen may take place at different sites of the molecule. Specifically, it will take
place at that site which is most electropositive by appropriate bonding of the molecule at
the active centre of the catalyst.
Reactions of catalytic oxidation may be thus devided into two categories: (1) electrophilic
oxidation, proceeding through the activation of oxygen, and (2) nucleophilic oxidation, in
which activation of the hydrocarbon molecule is the first step, followed by consecutive steps of
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Reactant 2°2
into the reacting molecule (Table 2). The mildest electrophilic oxidation is the addition of
oxygen to the double bond resulting in the formation of epoxides, or the oxyhydration of the
double bond to form respective saturated ketones (e.g.propene -— acetone). A more
pronounced structural change is the fission of the double C-C bond, saturated aldehydes being
formedfromolefins (propene — acetaldehyde + formaldehyde), or the fission of the aromatic
ring resulting in the appearance of anhydrides. The final stage is total oxidation to CO2+H2O.
Along the nucleophilic route the first, smallest structural change is the abstraction of hydrogen
in the process of oxidative dehydrogenation of alkanes and alkenes to dienes, or the
dehydrodimerization and dehydrocyclization. Deeper structural changes are involved in
reactions in which a heteroatom is introduced into the hydrocarbon molecule by a nucleophilic
addition. This may be oxygen, sulphur, nitrogen, etc. Introduction of the first e.g. oxygen
results in the formation of aldehydes; the introduction of the second one - in the formation of
acids or anhydrides. In all these processes the carbon skeleton and the n electron system
remain unchanged.
As mentioned above the oxygen species (>2» O2", and O" generated on adsorption at the
surface of an oxide catalyst are responsible for electrophilic oxidation. Both reactants of this
reaction are located at the gas phase side of the gas/solid interface. Such catalytic reactions are
called extrafacial, or reactions without transfer (Fig.3a). On the other hand, the nucleophilic
oxidation is areactionbetween the adsorbed reactant and the oxide ion of the catalyst lattice,
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Table 2
Heterogeneous oxidation of hydrocarbons
which is transferred across the gas/solid interface. Therefore this type of catalytic reaction is
called interfacial (Fig.3b). Two kinds of interfacialreactionsmay be distinguished. One kind is
reactions which are considered to proceed in two steps: (1) reaction between the reactant and
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
22
the oxide to give oxygenated product and partially reduced catalyst and (2) reoxidation of the
reduced catalyst with gaseous oxygen to restore the catalyst to its initial state. Because in this
kind of reaction the reduction part and the oxidation part are considered to be separate steps, it
is called the redox mechanism. A mechanism of this type was postulated by Mars and van
Krevelen to explain the kinetics of the oxidation of aromatics on V2P5 catalysts, therefore it is
also known as the Mars-van Krevelen mechanism. The second kind of interfacial reactions
comprises those in which both reduction and oxidation steps are performed in one
transformation. They are thus called concerted or push-pull reactions.
CO CO, __n
I
' ' C—0M> COj
WTM
oxide
MBTJ
ox i d *
777777777 m™kC%
oxidt
1 1 - Rtdox mtchoni«m|
\ -- a : - <=o2
mm w$m wm?
oxid. oxidt mid*
INTERFACE
r - INTERFACE
OKidt oxidt
n-Push-puU mtchonism |
CO
I • GI /
'mm*oxidt oxidt j
WUHLt »*TERFACE
Oxidation of a hydrocarbon molecule with high regio- and/or stereoselectivity requires finding
a compromise between many opposing factors, among which the following are the most
important: --`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
which dehydrogenate to give saturated ketones. As an example propene is transformed into
REACTIONS BY
CARBOCATION
MECHANISM
R-£-C ■,£-£-
k "
MeOMeOMeOMe
MeOMeOMeOMe
9
c-
■
J&&- c-
R-C-OC-C-
I o
, ACIDIC OXIDE
BASIC OK IDE
MeOMeOMeOMe
ft jut JI
DATBNTI
OXIDATCNTO
KETONE sz
MeOMeOMeOMe
OXIDATION TO
LDEHYDE
JL_2.
MeOMeOMeOMeOMe
R-C-CTCTC-
— "fa**'-
^-^c-c-c-d-c-
MeOMeOMe
acetone in a process called oxyhydration. The n bond of the olefin instead of interacting with a
surface proton, may react with a transition metal cation showing properties of a Lewis acid site
and may form surfce it-complex.
When the basicity of surface oxide ions is high enough, they may perform a nucleophilic attack
on me hydrogen atom in the a position which results in the formation of an allylic species.This
may be bonded to the transition metal ion either side-on as a x-allyl, or end-on as o-allyl. They
show high mobility at the oxide surface and in equilibrium between these two forms exists.
The s-allyl may be an intermediate in oxidative coupling to form dienes and the x-allyl may
undergo a nucleophilic attack of surface oxide ion resulting in a surface alkoxide intermediate,
which desorbs as an aldehyde in the case of addition to the primary carbon atom or a ketone
when secondary carbon atom is involved (nucleophilic oxidation). Thus, the type of adsorbed
intermediate complex predetermines the regio selectivity. One could envisage a situation in
which active centres abstracting hydrogen from hydrocarbon molecules will be dispersed far
apart and the mobility of alkyl and allyl moieties will be restricted. In such circumstances,
instead of a nucleophilic addition of one of the surface oxide ions the nucleophilic attack by an
adjacent OH" group may be more probable, resulting in the formation of an alcohol This is the
case in homogeneous oxidation, in which bom alkyl or allyl moieties and OH groups are
bonded to the same metal atom acting as the active centre. Many oxides contain surface
vacancies generating F-centres which may play the role of sites activating oxygen molecules to
species of electrophilic character. These may perform an electrophilic atack on any of the
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
PROPERTIES OF OXYGEN
^ L
HYDROGEN ABSTRACTION
ELECTROPHILIC
PEROXO
* ■ & o-Q'
M-O-M-O-M I
ELECTROPHIL1C ON GROUP I'////////////}
IV.V.VI METALS IN HIGH ELECTROPHILIC
OXIDATION STATES
NUCLEOPHILIC ON GROUP
VIII METALS IN LOW
OXIOATION STATES
OXO
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
ELECTROPHILIC OR
W//////////A
Y//////A
NUCLEOPHILIC NUCLEOPHILIC AT ,u-oxo
charge transfer to the oxygen molecule, but not strong enough to cause the cleavage of the O-O
bond, electrophilic O2" species are generated which may be bound to the active site end-on
behaving as superoxo species, or side-on to form moieties of the peroxo type. EPR and IR
spectroscopies provide information about the existence, location and spatial orientation of
these species (Fig.6). When the redox potential of the active site has an appropriate value,
cleavage of the O-O bond may take place and highly reactive O" may also be formed. All these
oxygen species interact with the it-electron system of organic molecules and start the
electrophilic oxidation. It should be born in mind that in the case of transition metal oxides
electrophilic oxygen species may appear at the surfacee also in the absence of oxygen in the gas
phase, as the intermediates in the transfer of oxygen from the lattice into the gas phase in the
course of the dissociation of the solid (Fig.7) due to equilibriation of the nonstoichiometric
oxide with the gas phase (Fig.8), or in the process of its reduction by e.g., hydrocarbon
molecules. Thus, not only the gas phase but also the lattice of the oxide may serve as the
source of electrophilic oxygen.
• Mg2+orC62+
Fig.6. Different adsorbed oxygen species detected at the surface of CoO-MgO solid solution
RCUH
o v x = o - = o22- = 0;
&- M ^ d Mn* 02_ Mn* 0 e_ Mn* O2" Mn+ 0 2 -
«2- >»2- rt2- «2- K2-
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
> s
2 n 2
0 " M * O" M ^ D Mn* Q2" Mn* 02" Mn* 02"
M207 — M20,
Quantum chemical calculations show that when an oxygen molecule approaches a hydrocarbon
molecule containing jc-bonds, the attack is directed into this region of high electron density
(Fig.9). Oxygen in the ground state encounters a very high potential barrier, but if it is
activated by stretching the 0 - 0 bond, e.g., through a transfer of an electron from the catalyst
onto the antibonding orbital and formation of O2" species the reaction becomes facile. The
character of the
4EteVl
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
transition state depends on the mutual orientation of reacting molecules (Fig. 10). In butene
oxidation, an oxygen molecule may approach butene along the axis of the molecule or
perperdicularly to this axis. In the latter case the first step of the reaction with butene-1 is the
end-on addition of oxygen species to the C2 atom to form a superoxide-like intermediate. This
intermediate easily reconstructs and forms a bridge-like transition state, in which the oxygen
molecule is bonded to two carbons, Cl and C2, forming a peroxo bridge. In the final step of
the reaction the cleavage of the two bonds takes place resulting in the formation of a molecule
of formaldehyde and propionaldehyde. When butene-2 is taken as reactant, the electrophilic
attack is directed on the double bond between C2 and C3 atoms and two molecules of
acetaldehyde are formed as the result of the cleavage of the peroxo bridge.
In the case of the interaction of oxygen with benzene the overlap of matching HOMO and
LUMO orbitals may take place only when the oxygen molecule with its axis perpendicular to
the plane of the benzene ring approaches the ring sideways (Fig.ll). On such approach the
benzene ring opens, the K-electron density is no longer delocalized and the O-O bond becomes
more and more elongated (Fig.12). Further evolution of the system results in the extraction of
two carbon atoms of the benzene ring, together with their hydrogens, towards the oxygen
molecule so that the system splits into two parts: a four carbon residual fragment of the
benzene ring and a nearly cyclic C2H2O2 moiety of a composition equivalent to glyoxaL Both
are highly reactive
■jj- •£•
h
I
*-*-«—11 » 1 1 a
OO, "*JL .ii*, * *
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
&
M if
Fig. 10. Pathways of the interaction of oxygen molecule with activated butene molecule.
.0 :
0009 /
aiot ,™* 7 .
Vz-^Voo6o "^us**
"J2M
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Fig. 12. Changes of conformation in the course of reaction along pathway II from fig.l 1.
species and it may be expected that under the conditions of a heterogeneous catalytic reaction in
the presence of gas phase oxygen they will react further to form products of total oxidation. A
different picture is obtained when molecular oxygen approaches benzene ring along a pathway
perpendicular to its plane with the axis parallel to the ring diagonal. In such case the reaction
takes place only after activation of oxygen molecule to stretch the 0 - 0 distance and leads to the
formation of an intermediate containing a peroxo bridge over the benzene ring (Fig. 13) in
analogy to the attack on olefins where a peroxo bridge is formed between the n-bonded carbon
atoms. This intermediate transforms into hydroquinone, which then may react with the next
oxygen molecules to form finaly maleic anhydride (Fig. 14).
1480
^ - • C
ff
1485
b(r=!49A)
a(r=1207&)
1490 if
'r-\ O^H
1495
RCAl
Fig. 13. A. Total energy of the C g H ^ - ^ system as function of the reaction coordinate along
pathway HI from fig. 11.Initial values of RQ<): (a,d) - 1.20A, (b,e) - 1.49A, (C,f) - 3.9A.
Curves a,b,c refer to the system C5H5+O2, curves d,e,f to the system C^H^+O^". B.
Optimal geometries for R=1.4A (minimum on curve b), and 0.2A (minimum on curve c)
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Fig. 14. Mechanism of elementary steps in the oxidation of benzene to maleic anhydride
The second route of oxidation reactions starts with the activation of the organic molecule,
followed by the nucleophilic addition of oxide ion. The classical studies by Adams using
deuterated propene and olefins C4 to Cg, and of Sachtler and de Boer with C -labelled
propenes showed unequivocally that activation of the olefin molecule consists in abstraction of
a-hydrogen and formation of a symmetric allyl intermediate, activation being the rate
determining step. £ = £ £ + 2 E#AB
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
A A<8
S.A
' «a
/ «v
-3
I*
m /
-4 *~
f y?
-5 \
--^ //"
0
-4 A
-T
/ \.x
• -•
-9
/ "» ii'
-10
—<* ***
1.9 1.0 2.1 1.2 I.) :.<
Fig. 15. Total energy (curve i) and diatomic contributions of C-H (curve II) and 0-H (curve III)
interactions as well as charge on hydrogen atom (curve IV) as a function of the distance of
propene molecule from the plane of C0O5 complex. Dotted curves refer to allyl species.
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
32
The mechanism of this process has been revealed by quantum-chemical calculations of the
interaction of propene molecule, approaching the complex composed of e.g. cobalt ion
surrounded by five oxygen atoms in the octahedral coordination, the sixth site being occupied
by the incoming propene. Fig. 15 shows the total energy (curve I) and the diatomic energy
contributions of the C-H (curve II) and OH (curve III) bonds plotted as a function of the
distance of propene molecule from the plane of cobalt complex playing the role of an active
site. A minimum of the total energy function (curve I) corresponds to the formation of a stable
intermediate surface complex. When propene molecule is approaching the complex, the C-H
bond is being continuously destabilized (curve II) whereas the strength of the O-H bond
increases (curve III) indicating that hydrogen atom is moving from the C-H bond in the methyl
group of propene to form the O-H bond with oxygen of the active site. The dotted curve I
represents the changes of total energy of the system when not the whole propene molecule, but
only the allylic species formed after abstraction of hydrogen is being removed away from the
surface, one hydrogen atom remaining at the surface as the OH group. It may be seen that this
process is energetically much more favourable than removal of propene. It may be thus
concluded that on contacting propene with the surface of e.g. cobalt oxide or oxysalt, its
reactive chemisorption takes place, consisting in the formation of allyl species. It is noteworthy
that in the course of the approach of propene molecule to the active site the charge on hydrogen
atom remains practically constant and amounts to about 0.7 (curve IV in Fig. 15). This
indicates that the abstraction of hydrogen is accompanied by a very rapid redistribution of
electrons. It should be also emphasized that the movement of hydrogen begins already when
propene molecule is still quite far from the active site, the elementary catalytic transformation
being thus a concerted redistribution of electrons and rearrangment of nuclei (Fig. 16). We shall
underline this point on many occasions.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
The question may be now raised as to where the activation of hydrocarbon and die subsequent
nucleophilic addition of oxygen takes place at the surface of an oxide catalyst Because in
partial oxidations of propene and majority of other hydrocarbons it is the first step which is rate
determining, studies of these reactions cannot yield any information on the next step in which
addition of oxygen takes place. One of the ways by which insertion of oxygen could be
investigated in more details is to by-pass the first step generating the allyl radicals by some
other more efficient route. This was accomplished by using allyl iodide which readily
decomposes into allyl radicals. The first step of the selective oxidation of propene is thus
facilitated making possible the examination of conditions which are necessary for the insertion
of oxygen in the next step of the reaction.
Important conclusions could be drawn from the comparison of the behaviour of Bi203 and
M0O3 as components of the classical catalyst for oxidation and ammoxidation of propene, in
the reactions with propene and allyl iodide (Table 3).When allyl iodide was passed over M0O3
practically total conversion is observed with 98% selectivity to acrolein already at 310°C. At
the same conditions M0O3 is completely inactive with respect to propene. On contacting allyl
iodide with Bi203 total conversion at 310°C was also observed, in this case however 70% of
the product formed was 1,5-hexadiene, practically no acrolein being detected. 1,5-hexadiene as
the product was also obtained in the reaction of propene on Bi203 indicating that activation of
hydrocarbon takes place resulting in the formation of allyl species, but there are no sites to
insert oxygen and in the absence of the next step they simply dimerize to give 1,5-hexadiene.
These results clearly demonstrate that in the molybdate catalysts it is the Mo-O sublattice which
performs the insertion of oxygen into the organic molecule. M0O3 itself is inactive in propene
oxidation because no centres are available for efficient generation of allylic species. When
however such species are formed by some other route, their total conversion to acrolein at the
surface of M0O3 takes place. In bismuth molybdate catalysts BP + ions play the role of active
site where allyl species are generated by activation of propene.
Table 3.
Interaction of propene and allyl with Bi203, M0O3 and Bi2MoOg
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
One of the problems in studies of oxide catalysts is the difficulty in counting the number of
active sites at the surface of oxide crystallites, whereat the elementary catalytic transformations
take place. In particular no such estimate is available concerning the activation of hydrocarbon
molecules. In order to obtain some relevant information following experiments were
performed. As already mentioned it is now well established that in the case of molybdate
catalysts the activation of hydrocarbon molecules takes place at the cationic sites such as Bi 3 + ,
Co 2 + , N i 2 + etc of the Bi 2 (Mo0 4 ) 3 , C0M0O4, NiMo0 4 etc. catalysts respectively. Let us
now support the defined number of isolated cations at the surface of a carrier. We could
determine the turn-over frequency of the activation of hydrocarbon molecules if we had a
method to count the number of activated molecules, i.e. gernerated allyl species. At this point it
should be reminded that experiments with allyl compounds have shown that allyl radicals when
contacted with the surface of M0O3 in appropriate conditions (cf.Table 4)- pick up oxygen
and are totaly converted to acrolein. Oxygen from M0O3 m a v m u s s e r v e ** a p r o b e t 0 d e t e c t
activated propene molecules, their number being measured by determining the number of
acrolein molecules formed. Following these ideas isolated bismuth ions were supported at the
surface of M0O3 and their activity in the oxidation of propene was measured by pulse
technique as a function of their surface concentration, expressed as number of bismuth
atoms per surface molybdenum atom. Results are shown in Fig. 17. Yield of acrolein
observed when allyl iodide was introduced was constant and independent of bismuth
30 • .
ocroUin from C , H , 0
^ — ^ J O — 1 /
20 ■
I --`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
9*
Ul
> ocroltin from frtt
mixture of MoOj
10 and Bi,(MoOJ,
— , CO, from C ^ , 01 •
11
>, , i i
T
Fig. 17. Yield of acrolein and CO2 in oxidation of propene and allyl iodide as function of
coverage of M0O3 surface with Bi ions.
coverage confirming the assumption that once allyl radicals have been generated they
rapidly undergo a nucleophilic attack by oxide ions from M0O3 lattice. On
introducing the mixture of propene and oxygen the activity at low surface coverage
with bismuth increased proportionally to this coverage, the turn-over frequency per
bismuth ion being thus constant. In the conditions of experiment it amounted to
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
35
0.5 propene molecule per bismuth ion per pulse. At higher bismuth coverages the activity
levels off because once B P + ions formed a monolayer no more change of the number of active
sites could take place. The yield of acrolein observed at the plateau is similar to that observed
with Bi2(Mo04)3 phase. It is noteworthy that the amount of C 0 2 formed remains constant
which indicates that the stray reaction of total combustion is not due to consecutive oxidation of
acrolein, but proceeds at some other sites, resulting from the properties of M0O3 itself.
Results of these experiments clearly demonstrated that activation of hydrocarbon molecules
takes place at the active sites composed of coordinatively unsaturated transition metal cations
coordinated by oxide ions.
Oxidation of benzene and toluene on vanadium oxide monolayer catalysts have been subjects
of many studies, in which attempts to identify the reaction intermediates by IR spectroscopy
and to elucidate the mechanism of the reaction have been undertaken. These studies led to the
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
conclusion that in the case of toluene the reaction starts with the formation of the benzyl
intermediate, which interacts with surface lattice oxygen to form, consecutively, adsorbed
benzaldehyde and benzoic acid precursors. These may either desorb or may be further oxidized
to carbon oxides, or undergo degradation of the aromatic ring with the formation of maleic
anhydride and carbon oxides. In order to find the mechanism of the initial activation of the
molecule and specify the factors determining the choice of the pathway by the reacting system
quantum chemical calculations were carried out of the interactions developing on approach of
benzene and toluene molecule to a cluster composed of six vanadium-oxygen square pyramids
assumed to be a model of supported vanadium oxide monolayer catalyst.
SINDO method was used for calculations. Toluene was approached side-on with the ring
plane parallel to the plane of the cluster or end-on with the molecular axis perpendicular along a
trajectory perpendicular to the plane of six edge- and corner-linked vanadium-oxygen square
pyramids, which represent an element of the (010) plane of V2O5. The trajectory was chosen
to point either at vanadium ion or bridging oxygen ions. In all cases the plot of total energy vs
reaction coordinate showed a minimum, corresponding to the formation of an adsorbed
complex (Fig. 18). Side-on adsorption was much stronger, with strong interactions of all
carbon atoms of the ring and methyl group with oxygen atoms of the cluster, accompanied by
simultaneous weekening of C-C bonds (Fig. 19). This indicates that side-on adsorption leads to
complete destruction of the aromatic molecule and formation of coke or carbon oxides. End-on
adsorption is most facile on bridging oxygen, and on approaching the toluene molecule two
hydrogen atoms of the methyl group move simultaneously away to form finally OH groups
with oxygens of the cluster and the methine group becomes linked to the bridging oxygen,
forming the precursor of Ar-CHO (Fig.20). The bonds between vanadium and bridging
oxygen are considerably weekened,resulting in reconstruction of the catalyst surface which
enables a facile desorption of benzaldehyde. It may be thus concluded that selective oxidation
-5300
-5350
-5400
-5450
-5500
-5550
-5600 J i i ■ «
0.5 1 1J5 2 2.5 3
REACTION COORDINATE [A]
Fig. 18. Total energy as function of the distance of toluene molecule from the plane of V^O^Q
cluster for different orientations and adsorption sites.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Fig. 19. Changes (in %) of the diatomic energy contributions in side-on adsorbate complex
C^H^CI^-bridging oxygen site in respect to the isolated cluster and toluene molecule
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
37
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Fig.20. Changes (in %) of the diatomic energy contributions in end-on adsorbate complex
CgHjCf^-bridging oxygen site in respect to the isolated cluster and toluene molecule.
E[.V]
'03/0 r
ma
om
'0340
0350
'03S0
'0370
6. STRUCTURE SENSITIVITY
In recent years experimental evidence is mounting, which indicates that catalytic properties of
oxides depend on the surface structure. They vary when different crystal planes are being
exposed on changing the crystal habit, when imperfections in form of steps and kinks are
generated at the surface or when the particle size is varied. Except of simple molecules such as
CO or H2 all oxidation reactions are multistep processes. In the case of hydrocarbons
nucleophilic oxidation starts with activation of hydrocarbon molecule by abstraction of
hydrogen from the selected carbon atom, which becomes exposed to the nucleophilic addition
of cfi~. The consecutive steps of hydrogen abstraction and oxygen addition may be then
repeated to obtain selectively the more and more oxygenated molecule. Each of these steps may
require the presence of a different type of active sites, which may not be uniformly distributed
over the surface of catalyst crystallites but each given type of sites may be characteristic for a
particular crystal plane. One can expect that in the case of transition metal oxides with strongly
pronounced crystallographic anisotropy different properties of active sites are related to the
differences of the surface structure of various crystal faces which results in structure sensitivity
of oxidation reactions.
The most spectacular example of me strong influence of surface structure on the direction
of the oxidation reaction is the behaviour of two cuprous molybdates: CU2MO3O10 and
CugMo40j5 in the oxidation of butene-1. Both are composed of the same chemical elements
in the same valence state and differ only in the spatial arrangement of atoms. Yet they show
entirely different catalytic properties, as shown in Fig.22: CU2M03O1Q is active in the
isomerization and oxidative dehydrogenation, but no traces of oxygenated hydrocarbon
molecules are present in the products, whereas C u g M o ^ ^ mainly inserts oxygen into the
organic molecule to form crotonaldehyde. The most striking feature is the complete absence of
isomerization in the latter case.
A pronounced influence of geometry on the pathway of the catalytic reaction was revealed also
for the case of the oxidation of o-xylene on V2O5. Fig.23 shows the selectivity for phthalic
anhydride and the selectivity for products of total oxidation as a function of the textural factor
of V2O5 crystallites, which is expressed as the ratio of intensities of (001) to (110) reflections.
--`,```,,`,`,`,,,`,``,,,`````,,
Cu,Mo 4 Ou
CU]MO}0„
Inns-Buf-l-ene
c— e e i B e-
-i ■ ' t_
: a « i » T t
Number of pulses
Fig.22. Conversion and selectivities to different products as a function of the number of pulses
of butene-1 introduced on G^Nfc^Ojg and Cu^NfttyO^ catalysts at 643K.
70 G4SI0R.MACHEJ (19S2}
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
> ,
«^ *i Phthalic anhydride E
60
+70 „•
A y( /^cocoJ e
"§50 6.0 S
(V o
/ / V \ 2ET i100 50 §
40 4.0 "
(-
U/ , o\-< 80
ao
L^ttf
30 - /[ 70
60
Analysis of the voluminous patent literature concerning the selective oxidation of hydrocarbons
clearly indicates that all efficient catalysts for these processes are based on group V,VI and VII
transition metal oxides. The chemistry of these oxides e.g. oxides of vaanadium or
molibdenum, is dominated by the consequencies of the considerable extension of their d-
orbitals and positions of the d-electronredoxpotentials relative to the anion valence band edge.
As the result n-bonds with terminal oxygen atoms are formed and the cations become displaced
from the centre of the octahedron towards terminal oxygen atoms. The large displacement
polarizabiMes give rise to high relaxation energy, which compensates the increase in cation-
cation repulsion energy involved in the rearrangement of the ocatahedra from corner-linked to
edge linked structures and makes possible the phenomenon of crystallographic shear by
strongly stabilizing the shear planes. Crystallites of these transition metal oxides assume layer
structures. They exhibit crystal faces, at which all constituent atoms are chemically saturated
and only HOMOLUMO type interactions may operate between the surface and the adsorbed
molecules, and crystal faces composed of coordinatively unsaturated cations and anions,
generating considerable variations of the potential along the surface which may induce
polarization and heterolytic bond rupture in the adsorbed molecules. As an example Fig.24
shows schematically the two types of crystal planes in the case of V2O5 crystallites.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Fig.24. Orbital structure and excess surface charges at different crystal faces of V2O5.
Comparison with the results of catalytic studies (c.f.Fig.23) indicates that high selectivity for
phthalic anhydride is observed in the case of plate-like crystallites exposing mainly the (001)
faces with the V=0 groups sticking out of the surface. However, when the crystallites expose
to a considerable degree the (110) faces, at which the shear planes may be nucleated and whole
perpendicular layers of oxygen may be extracted, total oxidation becomes predominant reaction
pathway. A general conclusion may thus be formulated that in compounds of those transition
elements, in which the phenomenon of the displacement stabilization results in the strong
anisotropy of properties, differences in surface and catalytic properties of different crystal faces
may be encountered.
Foreign ions present at the surface as impurities or additives constitute point defects which
may play the role of new active sites or modify the properties of the existing ones by shifting
the defect equilibria of the oxide. Moreover, these ions may preferentially accumulate only at
certain crystal faces, e.g. charged ions will segregate to polar crystal surfaces, a phenomenon
which may be called structure sensitivity of deposition. Unravelling of the role of these
parameters in determining the rate of elementary steps of catalytic oxidation reactions is a great
chalenge for the science of catalysis in the future.
After nucleophilic addition of the surface oxide ion to the carbon atom of the hydrocarbon
molecule, resulting in the formation of a precursor of the oxygenated species, the latter is
desorbed generating a surface oxygen vacancy. As mentioned above, oxides of group IV-VII
transition metals show a strong tendency to annihilate the vacancies by the formation of shear
planes, which are nucleated at the surface with the simultaneous release of oxygen by the
crystal (Fig.25). As one of the possible explanations of the fact that selective catalysts for
partial oxidation are always based on group V-VII transition metal oxides a hypothesis was
advanced that this tendency is the driving force facilitating the desorption of the oxygenated
product (Fig.26). A facile and efficient route is thus provided for the addition of a nucleophilic
lattice oxygen into the hydrocarbon molecule. Little is however known about the mechanism of
the release of an oxide ion from the surface layer of the oxide into the gas phase and about the
parameters which determine the rate of this process. It may be hoped that with the further
development of surface science techniques it will be possible to answer many of these
intriguing questions.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
C3H,0
rf°V»
+C H
1X1X1 3 AO
As discussed above heterogeneous oxidation reactions are multistep processes which require
multifunctional catalysts. Therefore multicomponent oxide systems are usually used as
catalysts in form of heterogeneous mixtures or supported oxide monolayers. In both cases
strong synergistic effects are often observed, non-existent in solid solutions which indicates
that they may be related to the presence of interfaces. The origin of these effects is one of the
most fascinating questions of catalysis to be answered in future studies.
One of the spectacular examples are multicomponent molybdate catalysts for oxidation and
ammoxidation of propene, some of them containing 10 or more components. X-ray
examination indicates that they are composed of three basic phases: Bi x Mo y O z ,
M3 + 2(MO04>3 and M 2+ Mo04, where M^ + and M 2 + are trivalent and divalent transition
metal ions. The most commonly used is the system based on I^CMoO^j, F e - ^ C M o O ^
and C0M0O4. Following data were obtained from the measurements of their behaviour in the
oxidation of propene at 320°C:
M 2 + a M 3 + b Bi x Mo y O z 95.7 69.7
Bi 2 (Mo04) 3 90.3 7.4
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
It may be seen that addition of trivalent and divalent cations to the bismuth molybdate system,
which is a selective catalyst, increases the catalytic activity by an order of magnitude without
much influencing the selectivity. Apparently the mechanism of the reaction remains the same,
but the rate determining step is accelerated.
As already discussed, in bismuth molybdate catalysts for oxidation of olefins the bismuth
ions play the role of sites activating the olefin molecule by abstraction of hydrogen and
formation of the allyl species, whereas the molybdate sublattice is responsible for the
nucleophilic addition of oxygen. It has been argued that introduction of the redox pair
Fe^ + /Pe^ + promotes oxygen and electron transfer. However, it is well established that
activation of the hydrocarbon molecule is the rate determining step of the reaction and it is not
clear how the redox pair interferes in this step.
Basing on the in situ studies of XRD and Mossbauer spectra it was shown (Fig.27) that in
the conditions of the catalytic reaction Fe^ + ions in Fe2(Mo04)3 are partially reduced to Fe^+
ions and nuclei of Fe^+MoC^ are formed. The latter serve as active sites for binding (>>
molecules and reducing them to O 2 " ions which are then transported through the defected
Fe2<Mo04)3 phase to replenish the active sites at the surface of Bi2(Mo04>3, reduced during
the abstraction of hydrogen from the olefin and subsequent addition of oxygen. The role of
C0M0O4 consists in stabilizing the isomorphous FeMoC>4 nuclei.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
500 7D0
TEMPERATURE. *K
Fig.27. Dependence of the isomeric shift on the Fe(III)molybdate in the Mossbauer resonance
spectrum (curves 1-3) and dependence of the rate of propene oxidation on temperature (curve
IV). o-l,o-2,A-3,A-4.
Further studies are needed to confirm this hypothesis and to unravel the mechanism of
elementary steps involved in such complex performance of the multicomponent oxide catalysts
and the role of individual components.
MoO] OMo MMo 0 * t o M|Mo ftfcto FcSb UP 2>Fe S^>2 B l O j TaOl (QllSkSriOt
(Cultt 143
■UOl t a TJO
•not uo
tm
«* i.ii 0J>* -0.1* 17 .0
M r oj*
HP tit OJX •OjOS, -r.i 2«,1
M b U» 3».«
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
M0fe « i
o>u» 1.12
MM* aii V6 2.*
** T.9-
Crtta. ait
MoO] ■0J1 1» tSt J.40 T.lt MJ 10-Z IS 10.1 I4.J4 23.4. 21.4. 13.09
CH»
It should be however borne in mind that on heating the oxide mixtures spreading of one of the
oxides over surfaces of the others may take place. Simple considerations of equilibrium
conditions at the interface between two solid phases and the gas phase show that when the
energy of cohesion of the clusters of supported oxide is smaller than the energy of adhesion of
this oxide to the support spontaneous spreading of the former over the surface of the latter will
take place as manifested in the phenomenon of wetting of one solid by another one. As a
result, there is always a tendency of oxide surfaces to become covered by a thin layer of other
components present in the mixture. The formation of such overlayer, too thin to be detected by
many of the standard experimental techniques, may profoundly modify the catalytic properties.
Therefore the phenomenon of wetting in oxide systems is of paramount importance for
preparation of catalysts. Little data concerning the surface free energy of oxides are available at
present and practically no information existsrelatingto the mechanism of surface migration.
The behaviour of oxide monolayers and three dimensional clusters, which may be
considered as colloidal particles, deposited at the surface of an oxide support, will be
controlled to a large extent by the surface free energy relations at the interfaces between the
support, the clusters of the monolayer and the gas phase. As the energy of cohesion of e.g.
V2O5 is smaller than its energy of adhesion to anatase or alumina, it is wetting these
supports and spreads over their surface. This is illustrated in Fig.30, in which the degree of
coverage of an anatase support with vanadium oxide monolayer is plotted as function of time
of calcination of a mechanical mixture of V2O5 and anatase at 450°C. In the same figure the
results are also plotted in the coordinates q vs. t. The linear dependence of parabolic
coordinates indicates that the kinetics of thermal spreading is diffusion controlled.The model of
this phenomenon is shown in Fig.31. When V* + -0 clusters are reduced to V^ + -0 clusters, the
energy of the cohesion increases to such an extent that it becomes greater than that of its
adhesion to the support and the monolayer of vanadium oxides shrinks and coalesces into three
dimensional particles. Thus, exposure of the vanadium oxide monolayer catalysts to alternating
oxidation and reduction cycles will entail dispersion and shrinking of the monolayer. These
processes may be followed by measuring the ir-spectra of appropriate probe molecules.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
0 10 20 TIMEh
Fig.30. Changes of surface coverage of T1O2 with VO x monolayer as function of the time of
heating of mechanical mixture of V2O5+TIO2.
f / / / / / /
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
at the surface of alumina, their acid strength depending on the coverage with vanadia and
degree of reduction of the latter. The mechanism of transformations of the vanadia monolayer
are summarized in Fig.33.
I 1 1 —
100 p
"1235
50
"0 5 10
c v , Vnm" 2
Ample experimental evidence accumulated in recent years indicates that oxide surfaces are in
dynamic interactions with the gas phase. The oxide system may respond to the change of
composition of the reacting catalytic mixture in three ways:
- defect equillibria at the oxide surface or in the whole bulk may be shifted and the change of
concentration of the given type of active sites involved in the catalytic transformation may
cause the change of catalytic properties;
- when the concentration of defects at the oxide surface surpasses certain critical value,
ordering of defects or formation of a new bidimentsional surface phase may occur resulting
often in a dramatic change of catalytic properties;
- when redox mechanism operates in the catalytic reaction, the ratio of the rates of catalyst
reduction and its reoxidation may be different for various oxide phases and hysteresis of the
dependence of catalytic properties on the composition of the gas phase may appear, these
properties being then strongly influenced by the type of pretreatment
R,rmAtW25cm-'l
TNH3 RjrmC
0 H - , 0 I12»-1260ciff'! 0 H. o
oxidized V/ NH. V/
surface V j OH .V
/ \ ! I / \
Al-0-Al-O -Al-O-At-0—Al-0
reduction
OH NH,
0-V-0-\At-0~AI-0-Al
Form C I
11250-1260 cm1)
0 2 ads
t«200*C
R>rmC J Rjrm B
il230-l250cm-»)M280crn t.
Fig.33. Transformations of V2O5 monolayer on reduction and reoxidation.
A general conclusion may be thus formulated that heterogeneous catalytic systems should not
be treated as two phase systems, but should be regarded as composed of three parts: gas and
solid phases and the surface region extending on both sides of the gas/solid interface (Fig.34).
On the side of the solid the surface free energy and the energy of interaction with adsorbed
species may cause the enrichments of the surface layer of the solid with the constituents of the
lattice (atoms of the solute in case of solid solutions, point defects in nonstoichiometric
compounds etc) or may result in the reconstruction or formation of two-dimensional surface
phases. On the side of the gas phase the species in the adsorbed layer may aggregate to form
two-dimensional liquid or may undergo a long-range ordering This surface region is not
autonomous but is in continuous interaction with the solid on one side and with the gas phase
on the other side. Its structure and properties may be thus modified either by changing the
composition of the gas phase or by altering the properties of the solid.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
o Q
O
O
O
O
c o?o!o
• D O
O
c
O
o SURFACE
"ENRICHMENT,
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
ADSORPTION
(BOMENSONAL BIDIMENSDNAL
SURFACE PHASES
GAS, LIQUID OR
ORDER LAYERS)
SURFACE REGION
Further reading
A.Bielanski, J.Haber, Oxygen in Catalysis, Marcel Dekker Inc., New York 1991.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
ON METALLIC OXIDES
JACQUES C. VEDRINE
Institut de Recherches sw la Catalyse, CNRS
2, avenue Albert Einstein F-69626 VILLEURBANNE Cidex, France
ABSTRACT
As a complement of the general presentation by J. Haber in this book, this article describes
general features of oxidation reactions in heterogeneous catalysis on metallic oxides.
Emphasis is placed on the Mars and van Krevelen catalytic mechanism and on its
implication in oxidation catalysis. The concept of active ales as ensembles of MOJJ species
of limited size is devebpped and examplified. The dynamical aspect of an oxide surface
during catalysis including the wetting phenomenon of an oxide spreading over another
oxide (multi component-type catalysts) or on a support is described. Such properties are
shown to greatly depend on the preparation procedure and on the activation and catalytic
reaction conditions and emphasis is placed on the importance of such parameters for the
catalyst performances in oxidation reactions.
1. Introduction
Oxidation catalytic reactions are of prime importance at an industrial level since they
correspond to a huge market. For instance in the US in 1991, 31.2% of the catalytic
production of major organic chemicals corresponded to oxidation catalytic processes (18.1%
heterogeneous, 13.1% homogeneous) and 17.8% to oxychlorination'- The market corresponds
to 20 billions US $ in the USA and world wide such numbers have roughly to be multiplied by
a factor of 2.5.
The concepts of oxidation began with Lavoisier's disproving of the phlogeston theory in
1773. One usually defines two groups of reactions namely homolytic and heterolytic. The first
type involves radicals formed by homolytic cleavage of interatomic bonds. The second type
involves an active oxygen compound or a metal ion which oxidizes the starting material in a
two electrons transfer reaction. The reduced oxidizing agent must be reoxidized in a second
step. One distinguishes several types of oxygen species as described in the previous paper of
this course by J. Haber.
Some important industrial processes and some great intermediates in industrial chemistry
are given in tables 1 and 2, respectively.
Majority of the catalysts correspond to metallic oxides with V or Mo as one of the key
elements. Some metals (mainly Ag for ethylene epoxidation), noble metals (as Pt, Pd for total
oxidation, etc) zeolites (Titano silicalite TS-1 from ENI for phenol oxidation) and
heteropolyoxometallates (e.g. H4PMO11VO40 for isobutene oxidation to methacrolein) may
also be used.
In majority of cases catalytic properties in oxidation reactions involve a redox mechanism
between reactant molecules and surface active sites as represented in the scheme in fig. 1 as
suggested by Mars and van Krevelen in 1953. Such a scheme necessitates a catalyst which
contains a redox couple as for instance transition metal ions and which exhibits high electrical
conductivity to favour electron transfer and at last which has a high lattice oxygen anion
mobility within the material to insure the reoxjdation of the reduced catalyst.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
53
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
54
From this feature arises the idea that the active sites are not isolated ions as in the
Taylor's model but rather an ensemble of ions in a kind of "inorganic oxide cluster". A
molecular concept of the active sites has then to be defined. Several examples are chosen
below to show how such a concept may be valid in oxidation reaction.
hydrocarbon ll/2 Q 2
y
oxygenate
Table 1: Some industrial processes for the formation of oxygenates by gas-solid heterogeneous reactions
Ag or Ag/support
propylene acrolein Bi Mo Co Fe K oxide 350-400 90-95
(multi component-type)
acrolein acrylic acid V2O5/M0O3 400-450 90-95
propylene acrylonitrile V 2 05/S!>205/Al203
benzene maleic anhydride VPO 400-450 70-75
butane maleic anhydride VPO 400-450 60-65
cyclohexane caprolactane Pd/Al 2 0 3
O-xylene phthalic anhydride V205/Ti02 350-550 75-80
S02 S0 3 V205/Si02 420 95
ethylene ethylene oxide Ag/Al 2 0 3 200-300 70-75
(10-30 atm)
ethylene vinyl chloride CuCl2,MgCl2>C 230
ethylene vinyl acetate Pd-Cu 175-200 91-95
(2-10 atm)
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
55
Table 2: Some great intermediates produced by heterogeneous oxidation catalysis and their more important
final products
HC^ \
CH3 - CH2 - CH2 - CH3 + 7(02") -» II 0 +4H20+14e-
HC / (maleic anhydride)
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
CH 3 ^O CH 2 ,0
"'^CH-C'' +0 2 " -» *C-CX +H 2 0 + 2e"
/ \ / \
CH 3 OH CH 3 OH
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
is a concerted action with homolytic breaking of metal - oxygen bond which transfers one
electron to the metal. Without any hypothesis about the nature of the metal oxygen bond one
can write with formation of a II alkyi complex (as usually admitted) or a chalky! complex.
Depending on the nature, oxidation state of the metal ion and its environment
(coordination structure), the metal oxygen bonds may be more or less polarised and therefore
the oxygen ion may exhibit electrophilic or nucleophilic properties. One may distinguished
three extreme cases:
8+ 8- 8- 8+
a: M = O (nucleophilic) b_:M = 0 and £: M - O (electrophilic).
In the first case reaction of protonation or deprotonation will be favoured, according to
M = 0 + H+-»M+-OH.
In the second case with non or weak polarisation, homolytic concerted reactions will be
favoured as allylic dehydrogenation of oiefins, according to the following scheme 1.
CH 2 = C H - C H 3
First hydrogen abstraction
on transition metal cation
■+•
CH 2 - CH - CH 2 + CH 2 - CH - CH 2 djmsrjzajion CH 2 = CH - CH 2 - CH 2 - CH - CH 2
_ _ .
02-Me+n02- 02-Me+no2-
First oxygen insertion on Mo O
polyhedra and second hydrogen
abstraction
H
I
CH2 = CH-CH 2 >CH2 = CH-CO
02-Mo+6o2- 02-M0+4D
| Second oxygen insertion on
I acid - base centers
CH2 = C H - C - 0 >CH2 = CH-COOHand
02-Mo+°o2- 02-Mo+^D
Scheme 1: Mechanism of propene oxidation into hexadine or acrolein or acrylic acid according to Man and
van Krevelen scheme
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
57
In the third case the electrophilic character of oxygen ion allows it to proceed to a direct
attack of a double bond or of an aromatic ring according to: ^^
0 H+
One may then obtain acetone from propene or even a double bond rupture giving
acetaldehyde and formaldehyde from propene or the anhydride from an aromatic ring. Final
stage could be total oxidation.
Obviously other oxygen species may exist as electrophilic surface species 0", superoxo
(electrophilic) M - O - 0°, peroxo (electrophilic) 0 - 0 " (02") and oxo (nucleophilic) as
O O
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Such a structure sensitivity is also described by J. Haber in his paper (preceding article)
and more recently on CujO surface for propene oxidation13. This aspect is partJcukiiy
important since it shows that by modifying preparation conditions of the metallic oxides one
may develop some faces rather than others and then one may obtain some modifications in
selectivities. A concept of epitaxial fitting of an oxide on a support developping specific
cristalline faces was developped by P. Courtine on V2O5 / T1O2 catalyst14. This concept is
interesting but was not confirmed later one.
Table 3: Structure sensitivity of the different faces of M0O3 crystals in define oxidation at 380°C (frontref.8).
Relative selectivity
Reactant Olefine Products Basal Side Apical
(010) (100) (101), (loi)
Propene Acrolein 0.06 2.3 0.7
CO, CO, 1 0 0
But - 1 - ene Butadiene 3 9.3 2
CO, CO? 1 0 0
Iso butene Methacrolein 0 0.6 0.1
acetone 0.06 0 0.06
CO, CO, 0 1 0
'C010J C*i°]
Figure 2: Cross section view of aMo03 (100) and (120) planes (projection of the lattice on the (001) plane)
(from ref. 12).
4 Vanadyl pyrophosphate15
Such a catalyst is well known for the oxidation of n-butane into maleic anhydride^. The
preparation necessitates the formation of VOPO4, O.5H2O as a precursor synthesized in an
aqueous or better in an organic medium and its activation in a flow of 1 to 2% butane in air at
the reaction temperature (ca 380°C).
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Here too the preparation and the activation of the samples appeared to be particularly
important to obtain a good catalyst. In all cases whatever the catalysts being good or
exceptionally good the (VO)2P207 phase (V 4+ cations) as a main constituent was detected by
X Ray Diffraction and by in situ Laser Raman spectroscopy in addition to small amounts of
some VOPO4 phases (V 5+ cations) as an, P, Y or 8 17 " 19 . Moreover VOPO4 pure phases
were observed to be more or less active and selective20. It turned out that the presence of
some V 5 + cations on the V 4 + catalyst surface of (VO)2P207 was necessary although an
excess (particularly when some of VOPO4 phases were present) was detrimental. Moreover
the catalyst surface is richer in P than the bulk by a ratio of about 2, presumably to protect the
active sites to be too much oxidized21.
Many studies have been performed showing several features:
- the precursor has a low activity and no selectivity into maleic anhydride. Maleic
anhydride formation only started when (VO)2?207 was formed as shown by in situ Laser
Raman22 and Xray diffraction23 in situ studies;
- well dispersed V 5 + ions on (VO)2P207 phase were identified by NMR studies of spin
echo mapping24 and were shown to arise from VOPO4 phase spreading over the (VO)2P207
phase playing the role of a support2";
Mr-?^ Me
H H
.'*♦ " > f?
1 1
0 Pv
'|\
j CH3 CH3
CH
H H
' nk °
_^°\P°
o^l\4<N) o^l H/^O 1
\ /
1 -H*0
\ ?* _ 0 /
oO " A L o 1
Figure 3: Proposed mechanism for the first step of the n-butanc oxidation over QJQypi&l (fromref16)
n-butane adsorbs on the free Lewis site and reacts with lattice oxygen.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
rn rrr
1 1 1 I f non equilibrium
/••* /•» ri
o O 0 *\0 r*
0 rk
0 y
7rr fff ftT J77" IrT
Figure 4: The two routes for butane oxidation in a ratle-type mechanism: (a) olefinic and (b) alkoxide.
- 0.05
C4 \ 0.002 ,^ ^v «,, ,
011
\ I 0.008 > x f 0 . 0 9 0.13
q3:::m.,nii,-m,,nii,,, --`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
d^::H3^m::t2KEfef""
* o / o
^ ¥ ^^a>>«
Figure 6: Schematicrepresentationof the surface structure of one polytype of (VOfyfyOl- "H* arrows
represent the possible pathwaysforfacile exchange of surface bound oxygen, either monoatomic or
diatomic, between the active sites. The "site-isolation" due to the diffusion barrier posed by the
pyrophosphate groups is clearly shown by these arrows (from ref. 28).
Figure 7: Model of adsorption of butane on (100) (VO2P2O7. Molecule I=butane; encircled area: cluster of
sites involved in MA and H2O formation. Molecules H-rV*=butene; various configurations of
adsorption leading to different products by reaction with oxygen (from ref. 30).
local defects which may then isolate better the domains I was describing above. Such a
conclusion is coherent with a work by Overbeek et afil in which VPO catalyst was deposited
in a support. It was then shown that the catalytic properties were very satisfactory even if
(VO)2P207 could not be detected by X Ray diffraction, presumably because of its too high
dispersion, i.e. its too small domains. Recall that XRD pattern can be obtained only if the
lattice contains at least above 10 unit cells in continuous without defect in between. For lower
ensemble sizes broadening of XRD peaks is known to occur even beyond detection for very
small ensembles. The presence of defects along chains obviously gives rise to the same
phenomenon. This is also a phenomenon frequently met in heterogeneous catalysis. Too well
organized i.e. well cristallized samples are not as good catalysts as defectuous catalysts. Note
that the characterization of such defects is particularly difficult. Such a characterization may be
possible by studying the magnetic susceptibility of the (VO)2P207 catalyst versus the
temperature. Such a susceptibility is very sensitive to the spin pairing of the V4"1" ions in the
double chains of the structure and changes with the presence of V 4 + defect densities^.
Such a reaction is important as a first, step to form methyl methacrylate monomer used
for altuglass or plexiglass formation by polymerisation.
The present process consists in preparing acetone cyanhydride by reacting HCN with
acetone. By acting H2SO4 (CH3COHCN CH3) one gets CH2 - C(GH3> - COONH2, H2SO4
which by action of CH3OH gives CH2 - C(CH3)COOCH3 + NH4HSO4. Such a process not
only uses environment unfriendly reactants as HCN and H2SO4 but also results in large
amount of ammonium sulfate (2 kg per kg ofacid formed)*
Other ways to prepare methacrylic acid consits on oxidising methacrolein (synthesized
by partial oxidation of isobutene for instance by hetero polyoxometallates) (C4 process) or by
propene carbonylation on HF medium (C3 process) or by hydroformylation of ethylene giving
propionaldehyde followed by action of formaldehyde giving methacrolein (C2 process).
The way under study here consists in oxidative dehydrogenation of isobutyric acid into
methacrylic acid by either iron phosphate catalysts using large excess of water (I2H2O per
IBA molecule) or heteropolyoxometallates such as H4PM04VO40 which are less stable but
necessitate much less water 1 to 2 H2O molecules per IB A molecule.
In the case of iron phosphate several phases have been studied 35 " 38 and it was shown
that the active phases consist of hydroxy phosphate with iron ions at two oxidation states +2
and +3 as summarized below.
It also appears that the best catalysts are composed of trimers (schematised below) of
edge sharing FeO^ octahedra separated one from the other by a vacancy and bonded to the
following chain by PO4 tetrahedra.
* see the notion of atom utilization proposed by R. A. Sheldon one of his papers in this book.
Copyright World Scientific Publishing Co.
--`,```,,`,`,`
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
65
F'Kttrc 9: Variations of the selectivity in methacrylic acid (%) as a function of the size of the (FeO^n clusters.
The numbers refer to samples in Table 4 (from ref. 39).
Fiffljre 10: Scheme of the redox couple and hydroxylation extent suggested for the isobutyric acid oxidation as:
H20 O2
Fe23+Fe2+(P207)2 <—> F e 2 3 + Fe 2 * (PC^OIfy > Fe 2 + X *+ F e ^ * (PC^OH)^ ( P 0 4 ) x
420°C
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
barbosaliie
Such polymers of FeO^ octahedra also exist in other inorganic compounds. For
instance they exist in ilvaite (CaFe^ + Fe^ + 2Si20700H) where silicate layers replace phosphate
anions and Fe0 6 octahedra form ribbons. It is interesting to note that such material is active
and selective for the reaction although less than the previous hydroxyphosphates40- This is
presumably due to the presence of Ca cations in the structure and of silicate counter anion
whose basicity in the sense of Pearson is different from that of phosphate anion.
The main idea we can remember from this study is that inorganic clusters of iron
octahedra with iron at two oxidation states are active for the reaction studied and that one has
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
to consider the active sites as these clusters, preferentially as ensemble of two trimers but other
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
67
sizes (dimers, tetramers, pentamers...) are also active and selective but to a lesser extent. The
iron oxidation state is changing during the reaction in a similar way that the y4+/v5+ redox
couple which was identified by spin echo mapping on VPO catalysts (vide supra §4). This
again shows that metallic oxides have to be considered with a dynamical view during the
oxidation reaction.
6 Heteropolyoxometallates 41 " 42
Such materials are constituted from a central atom X attached tetrahedrally by an
isopolyanion M x O y n " with:
M = Mo(VI),W(VI),V(V)
X - P(V), Si(TV), Bi(m), As(V) giving an heteropolyanion HPAn".
The negative charges are compensated by cations as protons, alkaline, alkaline earth or
transition metal ions. The main structures are those of Keggin (XMj2 O40)11" and Dawson
(X2 Mi g 062)P'- The M cation can be substituted partly by other cations as V leading to
compounds as H4 PMO] 1 V\ O40. Moreover the protons can be exchanged by metallic cation
as V 0 2 + , Cu 2 + , Mn 2 + etc. It follows that such materials may be used in oxidation reactions as
summarized below in table 5 and developed in more details in ref. 41,42 and in one of R. Sheldon
chapters in this book.
7 Zeolitic materials
Such materials are mainly known for acid type reactions such as fluidised bed cracking
catalysis (FCC), alkylation of aromatics, methanol conversion to hydrocarbon mixture in the
gasoline range (MTG process) or to low molecular weight olefins as C2 or C3 (MTO process),
etc. The size and shape of the cavities and channels influence the acidity strength, the acid sites
density and thus the catalytic properties.
In the case of oxidation reactions in gaseous phase zeolitic matrices usually are not
useful because the residence time of reactant molecules within the cavities is too long and
favors total oxidation.
At variance ENI company has developed a titanosilicate zeolite which is non acidic, has
a ZSM-5 structure (tridimensional framework with channels 0.52 x 0.56 nm in size) and is
working in liquid phase with H2O2 as the oxidant reactant. It is used for phenol oxidation to
anthraquinone and catechol and for many other reactions as alkanes oxidation or
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
68
hydroxylation. Many other reactions were studied on such catalysts designated TS-1 (Titano
silicate) and on other systems as VS-1, CrS-1 (with ZSM-5 structure or other structures and
and V or Cr elements substituting Ti), CrAPO-5 (El APO or MeAPO molecular sieves
developped by Union Carbide in the 1980*8 (E. Flanigen) with different known or unknown
structure and different element or metal at framework lattice positions), etc. More details are
given in the lecture #3 by R.A. Sheldon in this book.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
B. Delmon et al could even establish a relative scale of oxygen acceptor or donor properties
for many metallic oxide and then predict the catalytic properties of any mixed oxide 45 .
Another possibility was proposed long time ago by Matsuura, Schuit and other
scientists considering the cherry-like structure of a mixture of oxides. In such a model the
active phase as bismuth molybdate for propene oxidation to acrolein lies on the surface of the
other oxide as in a cherry. Such a model was reconsidered recently by O. Legendre et al 4 6 ,
J.C. Vedrine et al47» 4 8 . It was shown for instance that yBi2MoO$ phase catalyst which is not
as active and selective for the propene partial oxidation reaction than OCB12M03O12 phase
catalyse contains in fact excess of bismuth oxide which is not selective on the surface (XPS
study) 50 . By mixing with M0O3 t n e catalyst performances were very much exhanced. This
was assigned to the reaction of M0O3 with excess surface bismuth leading to active bismuth
molybdate.
A more striking example corresponds to a mechanical mixture of a solid solution
Fe(Co)Mo04 and Bi2(Mo04)3 47 » 4 8 . For Fe(Co)Mo04 solid solution Mflssbauer, Xray
diffraction,DTA and electrical conductivity studies showed 51 " 53 that when Fe was introduced
under reducing conditions in C0HM0O4, F e " ions were dissolved but also some FcPI ions.
The excess positive charge resulting from F e ^ ion dissolution is compensated by free electron
resulting in a large increase in electrical conductivity (multiplied by 3 orders of magnitude).
Moreover it was shown that the temperature of the a -> p phase transition of C0M0O4 varied
with Fe (total) and Fe 111 contents. This was shown to be important in catalysis, the 0 phase
being more efficient and the reaction temperature occuring in the same temperature domain
was usually increased with catalyst ageing. One can thus realize how F e ^ content could be
important for catalytic behaviour and catalyst life time.
When a mechanical mixture of large crystallites of Bi2(Mo04>3 phase and small
crystallites of Fe(Co)Mo04 was prepared and placed in a reactor for propene oxidation to
acrolein the following properties were observed4'»4&:
- the activity was increased with time on stream and at steady state reached a
multiplication factor of 200 to 300 with respect to that of the starting oxides. This is a huge
effect.
- The selectivity into acrolein increased also from 30% for Fe(Co)Mo04 to more than
95% as shown in fig. 11. A detailed study using XPS and EDX-STEM analysis clearly showed
that:
(i) before catalytic testing one has a mechanical mixture i.e the presence of grains,
closeby but with no overlapping of those of the solid solution Fe(Co)Mo04 and of
Bi2(Mo04)3. The EDX elemental analysis was found to correspond to the chemical analysis;
(ii) after catalytic testing (steady state) one still detects large crystallites of
Bi2(MoC>4)3 (several urn in size) but also Bi2(Mo04)3 spread over the small crystallites of
Fe(Co)Mo04 (50-150 nm in size). The latter ones are obviously the actors of the catalytic
reaction since because of their small size the surface accessible to the reactants is much larger
than for large Bi2(Mo04)3 crystallites. The active catalyst is then composed of the active
phase Bi2(Mo04>3 spread on the surface of the high electrical conductor solid solution
Fe(Co)Mo04 crystallites.A question still remains. Why does Bi2(Mo04)3 supported on
Fe(Co)MoC>4 so active? Is it a question of better dispersion of the active phase as suggested by
our experiments? Is it a change is the active site on Bi2(Mo04)3 phase itself due an epitaxial
contact with the support^?
An important observation is that during the catalytic reaction in the presence of
reactants and products the bismuth molybdate has spread over the other oxide in a process of
"wetting".
Such a wetting phenomenon was proposed by J. Haber et al^5 and was shown
previously to also depend on the more or less reducing atmosphere over the catalyst, on the
presence of water in the feed and to be more or less reversible.
300
IUU " ♦ f
*—v
90-
80-
/i"
70-
1 V B
60-
50-
40-
30-
20 -^
10 -j
OH —1 1 —
0 20 80 100
0 20 40 60 80 100
Feo.67Coo.33Mo04 Bi2Mo30i2
Feo.67Coo.33Mo04 wt % Bi2Mo30i2
Figure 11: Changes in acrolein selectivity and in its rate of formation at 380°C at steady state for
Feo.33Coo.67Mo04 + Bi2Mo3Oi2 mechanical mixture versus the bismuth molybdate
concentration (from ref. 47). aj. a phase, In p phase.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
OH
Sil +Mo0 4 2 "+2H f
OH
+ H20 :MO.
s
\
—Si
as in a silicomolybdic acid compound. >
Catalytic properties were studied for two reactions namely isopropanol conversion and
propene partial oxidation. The first reaction is a test reaction which allows to characterize
acidic, basic or redox properties of a catalyst. One gets dehydration to propene or di-
isopropylether for acid catalyst, acetone for basic catalyst in absence of air and acetone and
water for redox type catalyst in presence of air. The experimental results at 100°C clearly show
that at low Mo loadings acidic features are favored while redox features are favored at higher
loadings. This indicates that monomeric M0O4 species are acidic (presumably as in
silicomolybdic acid) while polymeric species exhibit redox properties (see fig. 12).
#
1
1 2- ^+propene
/ * * ▼
1- / ..'• y acetone
/
y
A' i ' * r- I '
0 1 2 3 4 5 6 7 8 9 10 11
Mo loading (wt %)
Figure 12: Isopropanol conversion at 100°C on M0O3/S1O2 catalysts versus Mo loading (from ref. 57) in air.
~ 1-6-
I 1.4. r-\
\ propanal
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
i 1.2.
\
§ < \
\
10.8 \
\
\
*■ 0.6.1
Figure 13: Propene oxidation to acrolein (nucleophilic attack) and propanaldehyde (electrophilic attack) at
400°C on Mo03/Si02 catalysts versus Mo loading (from ref. 57).
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
72
The second reaction studied is propene oxidation at 380°C (see fig. 13). Weak activity
was observed for low Mo loading. At higher Mo loading propanal was the major product while
acrolein was also observed. At very high Mo loadings and for M0O3 one gets almost
exclusively acrolein. Propanal is known to stem from propene by an electrophilic attack while
acrolein corresponds rather to a nucleophilic attack. These results indicate that monomeric
M0O42" species does not oxidize propene in our conditions while polymeric (polymolybdate)
species exhibit redox properties, with O species being rather electrophilic. Molybdenum oxide
exhibits redox properties with O species being rather nucleophilic. One can thus realise how
the size of the active site is important in oxidation catalysis. This is typical of structure
sensitive reaction.
9.2 Vanadium oxide on TiO2 support
Such a catalyst is well known for several reactions, such as O-xylene oxidation to
phthalic anhydride and selective catalytic reduction (SCR) of NO by ammonia. The anatase
form appears to be better than the rutile fom of TK>2. Such a catalyst with 1 and 8 wt%
V2O5 /anatase was prepared by Rhone-Poulenc (S * 10 m 2 g*1) for an exercise of
characterization by 25 different european laboratories. All results are assembled in a special
issue of Catalysis today published in May 1994 vol. 20 n°l 5 8 . Surface vanadium species were
observed to exist in three different forms: monomeric VO43" species, polymeric vanadate
species and V2O5 crystallites.
Catalytic properties were studied for two reactions, namely isopropanol conversion and
O-xylene oxidation^, 60 for both catalysts and both reactions the catalytic properties were
found to be similar indicating that only part of vanadium species, presumably dispersed
vanadium oxide species were active i.e. that V2O5 crystallites are inactive in our reaction
conditions. Chemical treatments by H2SO4 (IN) or ammonia solution at room temperature or
isobutanol at 80°C were carried out in order to dissolve preferentially the V2O5 crystallites
without dissolving well dispersed vanadium species. The catalyst then obtained had much
lower V content namely: 0.12wt% V2O5 from 1 (Vi) and 8 (Vg) wt% samples after NH3
etching, 0.24 wt% after H2SO4 etching and 0.21 and 0.57 wt% after isobutanol etching for V\
and Vg samples respectively. V\ and Vg samples were observed both to yield 74% phthalic
anhydride, 23.5% COx, at 96% conversion at 350°C while in similar conditions NH3 etched
samples exhibited 6% conversion only, with.22% to tolualdehyde and 76% COx selectivities.
Bulk V2O5 was observed to exhibit also low activity and low selectivity in phthalic anhydride
and high selectivity in COx even when similar surface areas of samples were used.
For isopropanol conversion at 2O0°C both Vj and Vg catalysts gave the same
selectivities (50% propene, 50% acetone) and conversion. For isobutanol etched samples only
propene was observed but with the same intrinsic rate (thus the same conversion also since
surface area values did not change) as for W\ and Vg samples.
The following conclusions could be drawn considering that a polyvanadate species
occupies a ca circular zone of a diameter of 0.38 nm per V atom (i.e. 0.165 nm2) and an
isolated monovanadate species 0.66 nm per V atom (i.e. 0.43 mrfl):
(i) Monomeric V04 3 + " species exhibit acidic character with OH groups
(Bronsted acidity) and result in propene formation for isopropanol conversion and in total
oxidation for O-xylene oxidation. Maximum coverage equals 0.43 wt% V2O5.
(ii) Polyvanadate surface species exhibit redox properties, namely give rise to
acetone for isopropanol conversion and phthalic anhydride for o-xylene oxidation.
(iii) V2O5 crystallites exhibit low activity for O-xylene conversion and high
selectivity in total oxidation.
the size of the deposited oxide species (monomeric, polymeric, bulk-type) results in different
catalytic properties.
Vanadium oxide has been deposited on several supports as silica, alumina, titania but
also zirconia, niobia, zirconium hydroxyphosphate, etc. Its reductibility as described in the
general features (§1) was studied by reduction by hydrogen at 400°C by J. Haber et al^1. It
was observed that V on Ti02 and Y.AI2O3 was reduced rather fast and one reached an O/N
atomic ratio near to 1 for Ti02 and 0.65 for AI2O3. At variance for silica the reduction was
much slower and an O/V ratio of 0.57 was obtained. Such differences were interpreted by the
authors as due to different species on the surface: mainly monomeric VO4 3 " species for Ti02,
dimeric species V2®1*~ f° r 1^2®3 ^ V2O5 crystallites for silica. Even if such a tendancy
is correct, I am of the opinion that the reality is much more complex.
The size of the polyvanadate species is known in solution particularly from UV spectra
data. For instance, it corresponds to V3O93-, V40i2^" with tetrahedrally coordinated V at
pH = 7 and 4.5 respectively and Vio028^" o r V I Q 0 2 8 ^ " with octahedrally coordinated V at
pH = 2.5. Deposition of such polyvanadates of different sizes was performed carefully on y
.AI2O3 support*^ and their initial structure despicted above were shown, to remain stable even
after calcination in flowing air at 500°C and to be only partly modified after catalytic reaction
of oxidative dehydrogenation of propane in the 350 to 450°C range. Moreover in such cases
the selectivity towards propene was observed to be the same at the same conversion level,
indicating that in this size domain of polyvanadate species the catalytic selectivity was not
changed, only the activity was observed to increase with vanadium loading.
Ethane oxidative dehydrogenation at 500/550°C has been studied for V2O5/S1O263
and V2O5/AI2O3 64 catalyst as a function of V content. It was observed that the activity was
much higher for alumina support with respect to silica support. Moreover it remained weak at
low V content and increased sharply and linearly with V content for a loading between 2 and
10 wt% V2O5 and more slowly above 10%. It is then suggested that polyvanadate species may
be as dimeric species as suggested by J. Haber et al 61 , are the active sites. Note also that the
selectivity into ethylene is higher at lower V content for the same ethane conversion level
(fig. 14).
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
_| 1_
2 4 6 8
Conversion (%)
Figure 14: Variation of ethene selectivity in ethane conversion at 550°C versus conversion at different V
loading
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
74
This indicates that smaller clusters are more selective in ethane oxidative
dehydrogenation. This feature has been observed to hold true for other supported oxides as
O2O3 on different supports as SK>2, AI2O3, zirconium hydroxyphosphate°*» ^ , etc. It is a
new example of structure sensitivity showing how the size of a deposited "inorganic duster"
influences the catalytic selectivity. The reducibility and reoxidability of V2O5/S1O2 and
V2O5/AI2O3 catalysts are studied in ref. 63 and 64 using inicrocatariinetry technique.
In a more general analysis, I. Wachs et al 6 '" 6 9 have studied the conversion of
methanol at 230°C. Such a reaction as proposed by J.M. Tatibouet et al70» 7l from Paris VI
university gives formaldehyde on redox sites,, dimethyl ether on acid sites and CO, CO2 on
basic sites. This is a test reaction to be compared to isopropanol conversion described above.
We have reported in table 6 the TON values obtained by the authors 68 for different supported
oxides with 1 wt% loading of the active pfcse corresponding to leas than the theoretical
monolayer coverage. ;
It clearly appears that the properties of supported oxides depend on both the nature of
the oxide support (a ratio of 10 3 in TON values were observed) and the nature of the surface
metal oxide (a ratio of 10 in TON values was observed) but remains about constant with the
metal oxide loading at least at relatively low loading (< 10 wt%) 68 .
labkiL. Turn over number values in (scconde)"1 for methanol reaction at 230°C on lwt % metallic oxide
deposited on several supports (from ref. 67).
10 Conclusions
Some general conclusions may be drawn from this general presentation:
(i) Oxidation reactions in gas-phase heterogeneous catalysis usually proceed via
Mars and van Krevelen mechanism i.e. involve lattice oxygen ions. Such ions exhibit an
electrophilic or a nucleophilic character and therefore present different catalytic properties
since the electrophilic oxygen interacts with a double bond or an aromatic ring while the
nucleophilic oxygen interacts with a C-H bond in a of the double bond or of an aromatic ring.
(ii) Oxidation reactions are structure sensitive and therefore greatly depend on the
local and surface structure of the oxide catalysts. Parameters such as reducibility and
reoxidability features of the oxides are very important for catalytic reactions.
(iii) Active sites for oxidation reactions appear to be "inorganic ensembles"81 of
metallic oxide atoms whose size greatly influences the catalytic properties. In some examples
the number of atoms constituting the active sites could be established. For instance double
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
trimers of edge sharing FeO^ octahedra are particularly active and selective for oxidative
dehydrogenation of isobutyric acid to methacrylic acid on iron hydroxyphosphates; ensembles
of four dimers of VO$ octahedra are suggested to be the active sites for butane oxidation to
maleic anhydride in vanadium pyrophosphate catalysts. Usually monomeric species as M o O ^ -
or VO4 3 " exhibit acidic feature and no partial oxidation properties. At variance low size
polymeric species exhibit better selectivity for many partial oxidation reactions than large size
species or bulk-type oxide.
(i v) Oxidation catalysts have to be considered with a dynamical view under
reaction conditions. This is related to the Mars and van Krevelen mechanism and also to the
mobility of the oxide lattice. This latter phenomenon results from the wetting effect particularly
for multicomposent and supported oxide catalysts. It follows that for many catalysts a certain
time on stream is necessary before the catalyst reaches its steady state. It is frequent that in an
industrial plant a steady state is reached only after one or two hundreds hours, the catalysts
lasting several years before being replaced. It is then proposed than the active catalyst for
multicomponent materials is composed of the active phase spread over the surface of the other
(s) oxide in a sherry-like morphology as suggested long time ago by scientists as Schuit et al
from the Netherlands or Matsuura from Japan. For more simple catalysts as doped vanadyl
pyrophosphates used for butane oxidation to maleic anhydride the right size of the active sites
(e.g. tetramers of vanadyl dimers) is monitored by the reactants in catalytic reaction conditions
leading to the right V * + / v 4 + ratio, by the preparation procedure to change the material
morphology (the (100) face of (VO)2?207 being developped), and by the adequate addition of
additive elements which regulate the site size and V^ + A r ^ + ion ratio. The view of an oxidation
catalyst as dynamical under catalytic reaction conditions is essential for our understanding of
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
its functioning.
References
45. L. T. Weng and B. Delmon, Appl. Catal. A 81 (1992) 141 and references therein.
46. O. Legendre, Ph. Jaeger and J. P. Brunelle, in New Developments in Selective Oxidation
by Heterogeneous Catalysis, ed. P. Ruiz and B. Delmon, Stud, in Surf. Sci. and Catal. 72,
Elsevier, Amsterdam (1992) 387.
47. J. M. Millet, H. Ponceblanc, G Coudurier, J. M. Hermann and J. C. Vedrine, J. Catal. 142
(1993)381.
48. H. Ponceblanc, J. M. Millet, G. Coudurier and J. C. Vedrine, in Catalytic Selective
Oxidation, ed. S. T. Oyama and J. W. Hightower, ACS Symp. ser. 523, Washington
(1993) 262.
49. D. Carson, G. Forissier, G. Coudurier and J. C. Vedrine, J. Chem. Soc, Faraday Trans. 1
79 (1983) 1617.
50. M. El Jamal, M. Forissier, G. Coudurier and J. C. Vedrine, in Prodeed 9th Inter,. Cong,
on Catal., ed. M. J. Phillipps and M. Ternan, Chem. Institute of Canada, Ottawa (1988)
1617.
51. H. Ponceblanc, J. M. Millet, G. Coudurier, O. Legendre and J. C. Vedrine, J. Phys. Chem.
96(1992)9462.
52. H. Ponceblanc, J. M. Millet, G. Thomas, J. M. Herrmann and J. C. Vedrine, J. Phys.
Chem. 96(1992)9466.
53. B. Benaichouba, P. Bussiere and J. C. Vedrine, Hyperfine Inter. 69 (1991) 739.
54. P. Courtine and A. Vejux, C. R. Acad. Sci. Paris 286 (1978) 135.
55. J. Haber, E. Mielczarska and W. Turek, Z. Phys. Chem. Neue Folge 144 (1985) 69.
56. M. de Boer, Ph D Thesis, University of Utrecht (1992).
57. T. C. Liu, M. Forissier, G. Coudurier and J. C. Vedrine, J. Chem. Soc, Faraday Trans. I
85(1989)1607.
58. Eurocat Oxide, ed. J. C. Vedrine, Catal. Today 20 (1) (1994).
59. B. Grzybowska-Swierkosz, G. Coudurier, J. C. Vedrine and I. Gressel, Ibid p. 165.
60. G. Golinelli, F. Trifiro, M. Baerns, F. Majunke, M. Messori, B. Grzybowska, J. Czekaj,
B. Majka, C. Dias and M. Portela, Ibid p. 153.
61. J. Haber, A. Kozlowska and R. Kozlowski, J. Catal. 102 (1986) 52.
62. J. G Eon, R. Olier and J. C. Volta, J. Catal. 145 (1994) 318.
63. J. Le Bars, A. Auroux and J. C. Vedrine in New Developments in Selective Oxidation by
Heterogeneous Catalysis, ed. P. Ruiz and B. Delmon, Stud, in Surf. Sci. and Catal.. ser.
72, Elsevier, Amsterdam (1992) 181.
64. J. Le Bars, J. C. Vedrine, A. Auroux, B. Pommier and G M. Pajonk, J. Phys. Chem. 96
(1992)2217.
65. M. Loukah, G. Coudurier and J. C. Vedrine in New Developments in Selective Oxidation
by Heterogeneous Catalysis, ed. P. Ruiz and B. Delmon, Stud, in Surf. Sci. and Catal.
ser. 72, Elsevier, Amsterdam (1992) 111.
66. M. Loukah, Ph D Thesis University of Rabat (1992).
67. I. E. Wachs, G. Deo, M. A. Vuurman, H. Hu, D. S. Kim and J. M. Jehng, J. Molec. Catal.
82(1993)443.
68. G. Deo and I. E. Wachs, J. Catal. 146 (1994) 323.
69. D. S. Kim, J. M. Tatibouet and I. E. Wachs, J. Catal. 136 (1992) 209.
70. J. M. Tatibouet and J. E. Germain, J. Catal. 72 (1981) 375.
71. C. Louis, J. M. Tatibouet and M. Che, J. Catal. 109 (1988) 354.
72. N. Nakuta, K. Tohji and Y. Ugadawa, J. Phys. Chem. 92 (1988) 2853.
73. C. T. J. Mensch, J. A. R. van Veen, van Wingerden and M. P. van Dijk, J. Phys. Chem. 92
(1988)4988.
74. J. A. van Veen and P. A. - J. M. Hendricks, Polyhedron 5 (1986) 75.
75. T. Tanaka, H. Yamishita, R. Tsuchitani, T. Funabiki and S. Yoshida, J. Chem. Soc.,
Faraday Trans. I 84 (1988) 2987.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
76. J. Wong, F. W. Lytlc, R. P. Mcssmcr and D. H. Maylottc, Phys. Rev. B30 (1984) 5569.
77. R. Kozlowski, R. T. Pettifer and J. M. Thomas, J. Phys. Chcm. 87 (1983) 5176.
78. G. Bergeret, P. Gallezot, K. V. R. Chary, B. Rama Rao and V. S. Subrahmanyam, Appl.
Catal.40(1988)191.
79. B. Moraweck, in Les Techniques physiques d'itude des Catafyseurs, ed. B. Imelik and
J. C. Vedrinc, Technip, Paris (1988) 587.
80. B. Moraweck in Catalyst Characterization, Physical Techniquesfor Solid Materials, ed.
B. Imdik and J. C Vedrinc, Plenum, New York (1994) 377.
81. J. C. Vedrine, J. M. Millet and J. C. Volta, Faraday Discuss. Chem. Soc. 87 (1989) 207.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
ABSTRACT
The reaction mechanisms of two transition metal catalysed reactions are
discussed: epoxidation of ethylene production and vinylacetate. In addition short
reference will be made to methanol and CO oxidation. The surface reactions are
related to the corresponding reactions in organometallic complexes. Also a
relation between surface science model studies and surface reactivity will be
made. Elementary surface reaction steps on surfaces will be highlighted;
subsequently the main mechanistic issues in oxygen CH and OH bond activation
will be described.
1. Introduction
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
2. Reaction mechanisms
11. Ethylene epoxidation by silver (3)
The epoxidation reaction probeeds at 250 *C over a Ag catalyst. The
selectivity of the catalyst strongly depends on catalyst composition and the
presence of chlorine containing hydrocarbons in the gas phase (~ppm).
The Ag particles are supported on a wide porous a-Al203 support and the
catalyst is promoted by addition of alkali, especially Cs.
The kinetics of the reaction consists of two parallel reactions and a
consecutive reaction:
(1)
x >c~c< Lj + o2
C0 2 + H 2 0
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
t ,8+
.Ag Ag Ag
O O
302
H H
-^ 2C0 2 + 2H 2 0
H H
O
Ag — Ag — Ag— Ag Ag—Ag
Hv_ .yH
>H>V<H
/Ags/Ag\/Afl\
Cl Cl
Figure 1
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
H (2)
CH3OH + O^ -> O^ + CH3O.*
1/2 0 3
insertion
Figure 2
Ag=0
(3)
After the acetate insertion step, the ethyl acetate intermediate undergoes 0
C-H cleavage to vinyl acetate. Bond breaking of this CH bond is facilitated by the
presence of the near polar C-O ester bond.
The other important steps are reductive elimination of acetic acid and
oxidation of Pd° to Pd2+. Except for the reoxidation step of Pd° (in acetoxylation
Cu+1 is not applied), the other elementary steps are well known in organometallic
chemistry. The reduced Pd forms metal clusters as in a metallic catalyst. Oxygen
can dissociate on these clusters. The non-selective total oxidation of ethylene to
acetic acid or combustion of acetic acid probably occurs on these metallic Pd
particles. However, according to Moiseev (see chapter 10) the Pd surface also
plays a role in the vinyl acetate reaction step.
2.4. CO oxidation
The recombination of adsorbed O atoms and CO to giwt C0 2 occurs
readily on most transition metals. The slow step is the dissociation of adsorbing
0 2 into adsorbed oxygen atoms. At low temperatures the reaction order is
negative in CO, implying that CO is Major Adsorbed Reaction Intermediate
(MARI). This high surface coverage suppresses 0 2 dissociation, for which at least
two neighbouring vacant sites are needed.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
of CH bond breaking in propylene versus oxygen insertion into the C-C x bond.
At low oxygen coverage this reaction will compete with propylene oligomerization,
at higher oxygen coverages consecutive reaction of acrolein will occur that will
lead to total combustion.
On the less reactive Ag surfaces it appears that with propylene oxygen
insertion into the C-C x bond competes with methyl CH activation. The CH bond
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
transition metals acetate decomposition will occur upon heating17. At lower
temperature CO and O,* formation occurs. At higher temperature the acetate
tends to decarboxylate. Coadsorbed Oxygen tends to suppress the low temperature
decomposition path because of site blocking.
Only atomic oxygen adsorbed to silver gives epoxide formation by reaction
with ethylene. On Pd or Rh aldehydes or ketones win be formed. In the presence
of excess oxygen the main reaction will be total oxidation. The analogous
homogeneous reaction of ethylene on Pd2+ is the Wacker reaction.
The steps in the homogeneous phase that lead to the aldehyde formation
are:
CH 2 OH
HV ^/ H CH, (7)
Pd2+___ OH Pd 2+ Pd-H
• O R p«
R
H3C-C< +"N C = C / T> r^C—^C-
OH H ' ^H H
^~°- Y *
H+ H \f'
(8)
.R' ,R'
H3C—C
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY
OH H
C—C<
x
0 ' H H'
H
- > * H3C-CI
OH V
Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
89
In reaction scheme eq.8 the proton is the catalyst that intermediates oxygen
insertion from acetic peroxide into ethylene. Epoxidation occurs via intermediate
formation of the electron deficient species II.
Also epoxide formation may occur upon reaction of ethylene with Mn0 3+
stabilized in a porphyrin18 or a T10+ species as present in T120319. Again
illustrating the need for electron deficient oxygen.
Two factors contribute to the uniqueness of Ag. On Ag the CH bond will
not be readily activated, whereas on the transition metal with a partially filled
d-valence electron band CH bond activation will readily occur. In addition at high
surface coverage the oxygen-metal bond energy has weakened, so that the weakly
exothermic epoxidation reaction can occur.
Ethylene when adsorbed to a transition metal surface can have the x or
di-a adsorbed state:
H
>C=C<H H^C==C/HH
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
i / \
i / \
M M M
x adsorbed ethylene di-a adsorbed ethylene
M
i x M M
intermediate III,
metallocycle
4. Summary
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
References
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
ABSTRACT
This chapter focuses on the direct catalytic partial oxidation of small alkanes and
NH3 to produce chemicals. Processes such as these are now the heart of the chemical
industry, and the discovery of new processes will be essential to utilize inexpensive
hydrocarbon feedstocks in chemical synthesis.
The discussion centers around four industrial processes: NH3 oxidation to
HNO3, the ammoxidation of CH4 to produce HCN, the direct oxidation of CH4 to
produce syngas, and the oxidative dehydrogenation of C2H5 to produce ethylene. All of
these processes operate at or above atmospheric pressure over Pt or Rh catalysts with
contact times of approximately 10"3 sec.
These reaction systems are very complex because they are so fast that mass
transfer effects dominate the rates and so exothermic that the reaction temperature rises to
800 to 1000'C in the very short residence time (10"3 sec) over the catalyst A third
complication in direct oxidation processes is the possibility of homogeneous reaction
which leads to poor selectivity, flames and explosions.
These processes can only be managed and understood by consideration of mass
and heat transfer along with kinetics. We consider the principles by which these
processes operate and the factors which must be considered in designing new direct partial
oxidation processes.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
This subject goes back to the origins of chemistry and catalytic reactions. In the
early 19 th century, Sir Humphrey Davy and his assistant Michael Faraday were interested
in the problem of explosions in coal mines. It was found that when air was pumped into
mines so miners could breathe, explosions occurred frequendy and unpredictably. The
cause of the explosions was of course methane combustion,
a very fast and very exothermic reaction which proceeds homogeneously as a free radical
chain reaction which produces flames and explosions.
People were interested in design of miner's lamps and flame arrestors. At about the
same time platinum metal had been discovered, and Davy and Faraday discovered that a Pt
wire could be made to glow spontaneously with no homogeneous flame when held in a
mixture of CH4 + air or H2 + air if the wire was preheated to ignite the surface reaction1.
This was the first recorded adiabatic catalytic reactor occurring on a single wire. The
93
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
94
principle of operation of this reaction, written down 100 years later, is that a reacting steady
state exists in which the heat of reaction of the catalytic combustion reaction, -192
kcal/mole, is sufficient to heat the wire to a temperature where reaction is self sustaining
and the process is autothermal2. This is the principle of operation of the autothermal
catalytic reactor, the major reactor type used for carrying out catalytic oxidation reactions.
Thefirstcommercially viable process was discovered by Wilhelm Ostwald in
Leipzig who used the same principle but now with a woven mesh of Pt gauze across a
tube3"5. He found that the reaction
could be made to proceed with >90% efficiency (selectivity) over Pt with a negligible
contribution from the competing reaction
i
which is even more exothermic and has a more favorable equilibrium constant Nitric acid
is then easily produced by further homogeneous oxidation and hydradon of NO. Thus the
use of catalysts was established for obtaining highly selective production of a desired
product, and the use of the heat of oxidation processes to provide internally die heat
necessary to heat the catalyst to the desired operating temperature.
Nitric acid plants quickly became an essential part Of the world's technology for
fertilizer and explosives manufacturing. Current plants operate under similar conditions,
now using Pt-10% Rh gauze catalysts and pressures up to 10 atm3. A related and even
earlier process was the manufacture of sulfuric acid, made by catalytic oxidation of SO2 to
SO3 followed by hydration to H2SO4,
S01+±02->SOs->H2S04 (4)
again a catalytic process first carried out over Pt However, the process was soon switched
to the much cheaper V2O5 catalyst which is used for sulfuric acid production today3. The
latter is one of thefirstapplication of metal oxides in catalytic oxidation. We shall confine
this discussion to noble metals, but oxide catalysts also have wide applications in
oxidation, giving different product selectiyities and being much cheaper than noble metals.
The mechanism of oxide catalysis is much different than on noble metals, noble metals
working by dissociative chemisorption of reactants and oxides by reduction-oxidation
using lattice oxygen.
There is a natural and very distinct division of oxidation processes between
fuel-lean and fuel-rich situations. As indicated in Figure 1, for oxidation of any
hydrocarbon in fuel-lean environments, reaction selectivities of hydrocarbons
(homogeneously or heterogeneously) favdr total oxidation to CO2 and H2O. This is the
regime of total combustion which finds wide application in incineration of volatile organic
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
fuel/0 2
Figure 1: Schematic plot of the selectivities expected versus fuel to oxygen ratio for the oxidation of a
hydrocarbon over a suitable catalyst Total combustion occurs in excess oxygen, syngas in excess
fuel, and chemicals possibly in large excess fuel.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
reactors are depicted in Figure 2. A pack of several layers of Pt-10% Rh gauze, as used
commercially in HCN synthesis and NH3 oxidation4, is the thin limit of monolithic type
catalysts. In later sections, we describe hydrocarbon oxidation over noble metal coated
foam monoliths. Like gauze packs, these ceramic supports provide a tortuous path for the
reacting gases and lead to high rates of mass transfer. For either gauze or monolithic
catalysts, the reaction zone is usually insulated as shown in Figure 2 to better approximate
adiabatic operation.
Reaction engineering principles must be understood in considering most chemical
processes, but this is especially true of oxidation processes. The student is referred to
Chemical Engineering texts such as those by Levenspiel6 or Fogler7 for thorough
consideration of reaction engineering issues.
We shall center our discussion around four important industrial reactions for the
production of commodity chemicals. These are listed in Table 1.
This is one of the workhorse reactions in the chemical industry3*5. Nitric acid is a
Gauze Catalyst
Figure 2: Simplified diagram of gauze or monolithic catalyticreactors.The gauze pack or noble metal
coated ceramic foam monolith is placed in the reactor tube between two inert ceramic monoliths
which act as radiation shields and reduce the heat loss in the axial directionfromthe glowing gauze
or monolith. The reaction zone is also externally insulated to reduce the radial heat loss and better
approximate adiabatic operation.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
major industrial acid which is a strong oxidizing agent which produces no liquid residues.
It was developed early in the 20th century to prepare nitrates for fertilizers and explosives.
The process was developed by Ostwald and has seen only minor modifications over the
past 90 years.
The reactor is a Pt-10% Rh woven gauze formed into 10-50 layers forming a
catalyst "bed" several millimeters thick and several feet in diameter. In typical operation
10% NH3 in air at 200#C is flowed over this catalyst at up to 10 atm pressure at flow
velocities of several meters/second. Exothermic reactions heat the catalyst gauze to 800'C,
and under typical operating conditions >90% of the NH3 is reacted with over 90%
converted to NO which if further oxidized homogeneously to NO2 and N2O5 which is then
hydrated to HNO3.
As noted previously, the dominant competing reactions in this process can be
written as
and
Andrussow found in the 1950's that the products in the Ostwald process for nitric
acid could be completely changed by adding methane and removing some of the oxygen to
form not NO but HCN in the overall reaction3*4
NH 3 ^ NO CH 4 -%> CO
No COo
N, HCN COo
Q*1^ C
2H4
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
n n *
CO — - C 0 2
Figure 3: Schematic diagrams for he reaction networksforNH3 oxidation. CH4 oxidation, CH4
ammoxidation, and C2H4 oxidation. The networks exhibit the series and parallel nature of these
partial oxidation reactions. The desired product in each of the systems is highlighted in bold type.
temperature, but the heat of reaction is somewhat higher to yield a gauze temperature of
1100 - C, and the process usually operates just above atmospheric pressure.
HCN is a key intermediate in Nylon 66 and in methyl methacrylate, and typical
units produce several hundred tons per day using a 1mm thick gauze reactor several feet in
diameter3'5. Both nitric acid and HCN synthesis reactors operate at extremely short contact
times; gas velocities are approximately 1 m/sec, and the catalyst is just over 1 mm thick, so
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
the contact time is approximately 10*3 sec. We shall see that similar short contact times are
also essential in formation of synthesis gas and olefins by direct oxidation.
CO + 2H2->CH3OH (9)
over Cu/ZnO catalysts and to produce synthetic diesel fuel by Fischer-Tropsch synthesis
This is a highly endothermic reaction and the reaction must be carried out in a tube furnace
with heat supplied from the outside of the tubes to heat each tube to 900#C to attain a
favorable equilibrium in this reaction and obtain adequate kinetics. Typically this process is
operated at 900*C at pressures of 30 atm with residence times of several seconds.
In a later section, we discuss the direct oxidation of methane to form syngas. The
combustion reaction (eq. 1) is in competition with syngas production as shown in Figure 2,
but this process shows promise when compared to the endothermic steam reforming
reaction (eq. 11).
The steam cracking of alkanes to olefins is also a very old process which is quite
inefficient Olefins are made by homogeneous pyrolysis of alkanes in a hot tube furnace10.
With ethane as feed the dominant reaction Svould be
Equilibrium and kinetics require that the reactor be maintained at temperatures greater than
~850*C, and the reaction is run in empty tubes with a contact time of approximately one
second. Since these reactions involve gas phase free radicals, many product species are
formed, ranging from methane to benzene. One of the products is solid carbon, and a large
excess of steam is added with the alkane feed to reduce (but not eliminate) carbon
formation. Since carbon formation is inevitable, this carbon must be periodically burned
out of the tubes.
Olefins are the major building block of organic chemicals in the chemical industry,
and steam cracking is the primary process by which they are formed currently. However,
this process is very expensive because of the heat which must be supplied to operate the
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
tube furnace, high separation costs of the complex product mix, and the down time because
of carbon formation.
Thus, the reactions listed in Table 1 represent two fairly old oxidation processes
which operate quite successfully, and two old endothermic processes which are in need of
improvement if the technologies using their products as feedstocks are to be improved.
The intent of this discussion is to describe how the first two operate and to show
that it appears possible to replace the endothermic and slow steam reforming and steam
cracking processes by direct oxidation processes which offer the promise of much smaller
and more efficient reactors.
It is our contention that this procedure almost never works for oxidation processes,
and that the new paradigm in designing oxidation reaction processes should be simply
small large KL }
reactor ^ reactor *
Bench scale reactors are frequently worthless in designing a practical reactor process
because they fail to capture too many features which are essential in designing the large
reactor.
The basic problem is that oxidation reactions are very fast and exothermic so that
most oxidation processes are (1) mass transfer limited and (2) heat transfer limited. The
former means that it is essentially impossible to measure reaction kinetics accurately: in
most experiments one measures only mass transfer rates because these rate coefficients are
smaller than reaction rate coefficients. The only way to avoid this problem is to dilute the
reactants or to lower the temperature sufficiently to slow the reaction. However, this
inevitably results in a regime of conditions where rates are not very meaningful with respect
to the operating regime of an industrial reactor.
The second problem is heat transfer effects and temperature nonuniformity in the
reactor. In a packed bed microreactor it is essentially impossible to maintain the
temperature precisely uniform because exothermic reactions heat the catalyst, particularly
near the center of the bed where rates then accelerate and generate even more heat This
obscures rate measurements, and only by severe dilution or lowering temperature can these
effects be minimized.
Another consequence of exothermic reactions is the possibility of ignition of the
reaction, the phenomenon of the glowing wire. As one begins to heat a reactor, the rate is
negligible until the catalyst reaches a temperature where the rate suddenly accelerates, the
temperature rises rapidly, and the reaction is suddenly nearly at completion. Thus
oxidation reactors exhibit "lightoff in which multiple steady states occur6'7. This is in fact
beneficial in oxidation reactors: the reactor is said to become "autothermal" with the heat of
reaction providing the heat to sustain the reactor operating adiabatically. Most industrial
oxidation reactors operate autothermally and nearly adiabatically. This is a key to their
successful operation: fast and exothermic reactions supply their own heat and thus avoid
the necessity of external process heat as must be added in steam reforming to produce
syngas (eq. 11) and steam cracking to produce olefins (eq. 12).
4. Reactor Simulation
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
must be included, but packed bed reactors can usually be modeled as nearly plug flow
reactors
dC;
u ^ = Ivijri (15)
and
uCp^*Z(-AHiri)-Q(t,zl (16)
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
z (total length L). Equation 16 is the energy balance on temperature T for heat capacity Cp
and AHj the heat of reaction i. Both gauze and monolithic reactors can be modeled as plug
flow reactors.
In afluidizedbed reactor or in any reactor at sufficiently low pressures that
diffusion provides rapid mixing (for example, UHV experiments), the mass balance is
described as a stirred tank reactor,
In the stirred tank reactor the residence time x is defined as the reactor volume divided by
the volumetric flow rate. In the plugflowreactor, the residence time is L/u, and in any
case we are very interested in comparing conversion and selectivity versus residence time x
for different feed and reactor parameters.
We shall not of course be able to develop these equations in this presentation which
are discussed in the texts by Levenspiel6 and Fogler7, but the student should be aware of
the governing mass and energy balance equations in interpreting experiments and in
predicting how experiments suggest operating conditions for an industrial process.
As we indicated previously, this reaction involves the reaction of CH4, NH3, and
O2 (eq. 8) over a Pt-10% Rh surface to yield 70% HCN. We have examined the reactions
NH3->JN2+JH2 (20)
(right) formation versus PcH4 for the same conditions as used in the upper panels. The
calculations use Langmuir-Hinsheiwood kinetics as listed in eq. 22-23.
which suggests that the mechanism is one of the bimolecular reaction (eq. 19) competing
with the unimolecular decomposition of NH3 (eq. 20). It is seen that for pure NH3 the
only reaction is formation of N2 (eq. 20) and that, as the CH4 pressure is increased, the
HCN increases linearly with PCH4 (the rate is first order in CH4 in this regime). Then near
a 1/1 ratio, both the HCN and N2 begin to decrease rapidly (the rate is proportional to
PCH4"4)- It is seen that me maximum selectivity to HCN
r
%CN= "^r =0.9, (21)
and that this occurs when PCH4 = PNH3- This experiment shows that these low pressure
experiments do produce predominantly HfcN.
We next attempted to fit the data to a Langmuir-Hinshelwood type rate
expressions14. These expressions are the ones usually used to fit experimental data in
surface reactions. The model assumes that (1) adsorbates are confined to a monolayer, (2)
all adsorbates are competitive, and (3) that adsorbate properties are independent of the
coverages. These are highly simplifying assumptions, but they are found to give
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
K P
_ I CH4?NH3 m
HCN ( )
~ WCH* 4
P
NH3
and
r = y*», 3)
P
NH3
The derivation of these equations is too involved to consider here, but they basically come
from the requirement that both reactions require 4 vacant sites, and the denominators come
from assuming the rate proportional to (1-ZOj)4
In all of these expressions the K's are groupings of reaction and equilibrium rate
coefficients. It is seen in Figure 4 that these expressions give an excellent fit to the data
over a wide range of pressures and pressure ratios. They suggest (but do not prove) that
we have the correct mechanism.
These kinetics show that in excess CH4 the rates are strongly blocked by some
species (as the 4th power of its coverage), and we initially thought this species was carbon
from methane decomposition in excess CH4.
We next used ultrahigh vacuum (UHV) experiments on Pt and Rh single crystals to
try to identify the species responsible for the slowing down of NH3 decomposition to allow
HCN to form15*16. We did this by temperature programmed desorption (TPD), a very
versatile and simple technique in which one heats the surface with adsorbates and follows
the desorbing produces with a mass spectrometer.
However, we cannot examine this reaction in UHV because neither CH4 or NH3
will react at the low temperatures (<300K) necessary to attain significant adsorption at 10"8
Torr. However, we were able so simulate the HCN synthesis reaction, not by using a
mixture of CH4 and NH3, but instead by using methyl amine, CH3NH2. This molecule
contains the C-N single bond and we can use it to determine the decomposition of C-H,
N-H, and C-N bonds. Dissociation of the C-N bond would not lead to HCN, but
dehydrogenation would yield HCN
In fact when we adsorb a monolayer of methyl amine on Pt, the only products we
observe are H2, HCN, and cyanogen (C2N2). Cyanogen is the product which must form
if H has evaporated to leave only CN which must dimerize to evaporate. The other
surprising feature of these experiments is that C2N2 does not desorb in UHV until
1000 K. In other words, the CN species is very strongly adsorbed on Pt (more strongly
than O atoms or any other common adsorbates).
Therefore we are led to the conclusion that the HCN reaction occurs, not because
carbon blocks NH3 decomposition but because the surface nitrile CN species blocks the
surface against NH3 decomposition and allows HCN to form.
We finally used rate expressions such as those described in section 4 to attempt to
model the performance of the atmospheric pressure reactor14. The flow of gases through
the gauze reactor should be nearly plug flow ( a "tube" several mm long and with a
diameter equal to the distance between gauze wires which is 0.1 mm), so the equations to
be solved are the plug flow mass balances (eq. 16) with now the concentrations expressed
in partial pressures. We solved the approximately 10 equations (one for each species) with
in initial conditions corresponding to the partial pressures of the feed gases. We solved for
the exit concentrations and compared these results to those observed in the experimental
reactor. We observed very good agreement between model and predicted concentrations
for all species, especially with the variation with feed conditions, a strong test of the model.
Thus we conclude that we can describe the performance of the HCN reactor
exceedingly well. Further, we believe we understand the mechanism in terms of an
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
oxidative dehydrogenation of CH4 and NH3 which produces CN that blocks the surface
against excessive NH3 decomposition.
One of the problems in interpreting reactions is the multiple products which are
formed and the multiple reactions which occur. We must handle many reactions to describe
processes of practical interest
We next consider the simplest example of a catalytic oxidation process which is the
oxidation of H2
on noble metals. Although* this process his little industrial importance, it is a prototype of
more realistic reactions. Furthermore, this reaction is a subset of all oxidation reactions
involving H and O species.
We know more about H2 and O2 adsorption properties than almost any species
except CO, and therefore many of the elementary steps are available from the surface
science literature. We stress, however, that we know of no experiments in which the
kinetics of this reaction have been measured except under UHV conditions; in all other
experiments only the rates of adsorption or mass transfer rates of H2 or O2 can be
measured because this reaction is almost invariable mass transfer limited.
The elementary steps in this reaction are probably
H2<+2HS, (27)
02<->20s, (28)
Os+Hs++OH5, (29)
OHs+Hs~H2Os, (30)
H&t+HiQ, (31)
20Hs^H2ps+H5. (32)
All of these steps are reversible, so the process must involve 12 distinct elementary steps.
Here we see the problem with this simplest of oxidation reactions: it must involve at
least 12 elementary steps. All of the adsorption and desorption steps are well characterized
in the literature. However, the surface reactions steps, eq. 29,30, and 32, were not well
known until recently. They all require setae spectroscopic characterization of the surface
coverage of species to measure the kinetics of these steps.
We have recently used laser induced fluorescence (UF) of desorbing OH (in the
gas phase) and laser induced desorption of product water (measured on a microsecond time
scale) to characterize the elementary steps of the surface reaction steps17*18.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
These results are shown in Table 2. Listed are the reaction steps of the above
sequence (eq. 27-32) and the preexponential factor and activation energy of each reaction.
We assume that adsorption steps are of the form
raj=ka(T)Pj(l-Ze) (33)
rdj=kdjej, (34)
r*=kRejOr. OS)
vertical scale is energy in kilocalories per mole while the horizontal axis is the "reaction
coordinate" over which molecules pass in going from H2 + O2 at the left to H2O on the
right The gaseous reactants and product have different energies by the enthalpy of the
reaction or the heat of formation of water whiqh is -57 keal/mole.
It is evident that this is a "downhill" or exothermic process as are all oxidation
processes. The largest barrier on Pt is the step OHs + H s -» H20 s (eq. 30), which has a
OH»+iHaw
H^IV OH<*+H«)
\
if.
OHoo+H(«)
Potential
Energy
(kcal/mol)
m ^ * Ow
—J-2M
HaOw
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
2Hd)+0<g)
b) Hj + 0 2 on Pt
OH^ + i H ^
Potential
Energy
(kcal/mol)
H20W
Figure 5: Potential energy diagrams for H2 oxidation on Rh (upper) and Pt (lower) surfaces. The vertical
axis is potential energy in kcal/mol with respect to gas phase hydrogen and oxygen atoms at
298 K. The horizontal axis is the reaction coordinate.
7. Hydrocarbon Oxidation
The next simplest fuel after H2 is CH4. We now consider the direct oxidation of
CH4 over Rh and Pt coated monolithic catalysts to form syngas 19*20
as compared to the current industrial process of steam reforming CH4 over Ni catalysts to
form syngas (eq. 11).
The symbols in Figure 6 represent experimental data for methane oxidation over a
l.Oi , I.OI-
CH 4 /0 2
Figure 6: CO selectivity, H2 selectivity, CH4 conversion, and reaction temperature for CH4 oxidation in
air as a function of the fuel to oxygen ratio over a 80 ppi 9.8 wt.% Rh foam monolith (closed
symbols, solid lines) and a 50 ppi 11.6 wt.% Pt foam monolith (open symbols, dashed lines).
The symbols are experimental data points and the curves are model predictions. In all cases, the
CH4 and air mixture was premixed and heated to 460*C.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
CH4->Cs+4Hs, (38)
Cs+Os->COs, (39)
C02s->COs+Os, (42)
plus the steps in the H2 + O2 reaction listed in Table 2. This is a total of 19 steps. The
elementary steps in the surface reaction
co+\o2->co2 (43)
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
are well known on Pt and Rh from the surface science literature, although there are in fact
many disagreements between experimental determinations of many of these parameters.
Table 3 lists the rate coefficients for these reactions in methane oxidation21. The
notations and units are the same as for H2 oxidation, and the entries for this reaction are not
repeated here. We have used these parameters the develop a computer simulation of
methane oxidation over noble metal coated monoliths. Theresultsof this simulation are
shown as the curves in Figure 6. Obviously, the model does an excellent job of predicting
the data when kinetics, heat transfer, and mass transfer are all considered.
We can use these activation energies to construct a potential energy surface over
which these molecules must pass in creating gaseous H2, H2O, CO, and CQ2- This is
shown in Figure 7. As with H2+O2, the gaseous species energies are obtained from
thermodynamic data. Note that the right hand portion of these curves are simply taken
from Figure 5 forH2 + 02-
It is evident that the energy surfaces are very similar on Pt and Rh. The major
difference comes in the H2 + O2 portion of the diagram, and this explains in fact why Rh is
a superior catalyst to Pt in producing H2. The barrier to form H2 compared to H2O is
lower on Rh, while the barrier to form H2O is lower on Pt Restated, the adsorbed H s
atoms on Rh have a lower barrier to dimerize
2HS->H2 (44)
rather than wait for reaction with adsorbed oxygen which forms OHs and then H2O which
quickly desorbs.
Oxidation reactions become even more complex when we consider the oxidation of
larger hydrocarbons. Ethane is the next simplest hydrocarbon, but it can undergo several
oxidation reaction including combustion
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
CH + 2
(a)Pt 4(0 \/ °«t)
f2HKg)
W /**«•
Energy •**•
(kc«l/i»ok) 4I
^
^+20»t5fe2OHw+2Hw \«*
A2H2O<0
|&I«J
2HaOfc
(b)Rh
Potential
Energy -Mi
(kcal/mok)
CO^A , 4 ,
|2H20(|)
■ftTJ/
Figure 7: Potential energy diagrams for CH4 oxidation on Pt (upper) and Rh (lower) surfaces. The vertical
axis is potential energy in kcal/mol with respea to gas phase methane and oxygen at 298 K. The
horizontal axis is the reaction coordinate.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Since all of these reactions (eq. 45-47) are exothermic, they can provide the heat for
endothermic reactions to take place including the thermal dehydrogenation of ethane to form
ethylene (eq. 12).
We have examined the partial oxidation and oxidative dehydrogenation of ethane
over Pt and Rh coated monolith 22 in an autothermal reactor operating at ~1000*C near
atmospheric pressure with residence time % of 1 to 10 milliseconds. Figure 8 shows
Rh
Si 80
1 I SyCO
I •"
•
L
^T CO
—■
-E= i , i -i 1
Total Hydrocufaom
J.
1 I I I
Rh
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
selectivities and ethane conversion as a function of the reactant composition over both Pt
and Rh catalysts.
For these plots, the conversion and selectivities have been defined on a carbon atom
or hydrogen atom basis. This implicitly accounts for any mole number change due to
reaction. For CO, CO2, and C2H4 the selectivity is defined as (assuming CO, CO2, and
C2H4 are the only products containing carbon)
C2H6-+C2H5s + Hs (50)
C2H5s->C2H4S + Hs (51)
C2H4s<r>C2H4 (53)
plus all of the steps in the H2 + O2 reaction listed in Table 2 as well as all of the additional
steps in CH4 oxidation listed in Table 3 except the decomposition of CH4 (eq. 34)22. A
reaction network describing these probable reaction paths is shown in Figure 9. The rate
expressions for these additional steps (eq. 50-53) are less well known and a mechanistic
model is still in preparation.
8. Summary
We have discussed several catalytic oxidation process, some old, some new. These
processes all involve very fast, very exothermic reactions. In all cases very high
selectivities to the desired products are obtained at nearly complete reactant conversion in
C
2H6 +
°5 — O^ + C j H s j ^ C ^ ■*" C 2 H 4 (on Pt)
C2H4
-ctH ^C4H8
C2H6
^ » C,H
'4"10
CA; Cs+ H s - H
2
•CO;
(onRh)
CO
os
CO,
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Cs (onPd)
■*CIL
H
•*- H 2 0
Figure 9: Proposed surface reactions in ethane oxidation At therightare indicated gaseous species produced.
9. References
1. Bielanski, A., and Haber, J. Oxygen in Catalysis; Marcel Dekker: New York,
1991; Vol. 43.
2. Rideal, E. K., and Taylor, H. S. Catalysis in Theory and Praxis; Macmillan:
London, 1926.
3. Satterfield, G. N. Heterogeneous Catalysis in Industrial Practice; 2 ed.; McGraw-
Hill, Inc.: New York, 1991.
4. Twigg, M. V. Catalyst Handbook; Wolfe Publishing, Ltd.: London, 1989, pp 470-
489.
5. Honti, J. D. In The Nitrogen Industry, 1976; pp 381-400.
6. Levenspiel, O. Chemical Reaction Engineering; 2 ed; John Wiley and Sons: New
York, 1972.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
G.B. MARIN
ABSTRACT
The oxidative coupling of methane is discussed in terms of
reaction pathways and catalyst development. The importance
of the interplay between chemical kinetics and mass transport
phenomena and its consequences for selectivity towards ethane
and ethene is highlighted.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
1. Introduction
Steam reforming aimed at the production of synthesis gas is the most important
commercial process for methane conversion into chemicals and liquid fuels. The
conversion of methanol to gasoline (MTG) or oleflns (MTO) or the Fischer-Tropsch
syntheses to hydrocarbons, such as the Shell Middle Distillate Synthesis (SMDS) use
synthesis gas as feedstock. Another important route towards chemicals consists of the
hydroformylation or the carbonylation of olefins. In addition to these indirect processes,
the direct methane pyrolysis towards ethene and ethyne is also possible, but temperatures
higher than 1500 K are required11,73. The introduction of dioxygen as reactant, however,
makes the direct conversion thennodynamically feasible at much lower temperatures.
Thus, recent research efforts were focussed on oxidative routes for direct methane
conversion to liquid fuels or chemicals, in particular on the partial oxidation of methane
to Cj hydrocarbons, more often referred to as the oxidative coupling of methane.
At the temperatures and pressures under which the catalytic coupling of methane
is usually carried out, coupling also occurs in the absence of catalyst. Hence, the
oxidative coupling of methane in the absence of catalyst has been addressed in several
papers2,5'23,35'5^55,6*92. The selectivity to JC2 products amounts to approximately 50 % at
methane conversions lower than 15 % depending on the partial pressures of methane
and oxygen. The selectivity for products is defined as the amount of moles of methane
converted into Q products per mole of methane converted. Selectivity not only
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
determines the efficiency with which the carbon of the feedstock is used, but also the
reaction heat. Selectivities to ethene exceeding 65% are necessary to maintain the
reaction heat below 1200 kJ per mole of ethene produced. The selectivity to CO is
higher than that to CO* in contrast to what is observed for the catalytic reaction. The
higher CO selectivity can be an advantage, since CO is a more valuable chemical than
C0 2 . A series of patents assigned to the BP Company16*40 claims processes consisting of
the simultaneous oxidation to Q hydrocarbons and synthesis gas, CO and H2. A
disadvantage of the methane coupling in the absence of catalyst consists of the lower
reaction rates. At atmospheric pressure, a ten to hundred fold higher space time is
needed to reach the same conversion as in the presence of catalyst However, this can
be compensated for by carrying out the reaction at elevated pressure. It follows from
Figure 1 that in the absence of catalyst from 400 kPa on the space-time yield for Q
products, defined as the numbers of moles produced per unit reactor volume per second,
rises to a level comparable to that encountered in commercial refining operations.
Weisz91 noted that most space-time yields in the petroleum refining and petrochemical
industry fall within a window of 1 to 10 mol m^^,"3 s"1. The QH* space-time yield in
an ethane steam cracking coil, for example, amounts to 12 mol m s"1 based on the
radiation section. The lower limit of the window corresponds to uneconomical high
investment costs, while the upper limit corresponds to limitations by physical transport
of mass and/or heat. In a highly exothermic process such as the oxidative coupling of
methane the latter is limiting. Figure 2 shows the selectivity to the major products as a
function of the oxygen conversion. From oxygen conversions of 80% on the product
selectivities no longer depend on the total pressure.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Dpagation/Branching: T, <y. T
CH4 + HO,' •• C H , ' * Hp s 1 1 14
H^O t + M + 2 OH'* M 1 1 13
Termination:
A number of studies has been devoted to the modeling of the coupling in the
absence of catalyst3,22,23,37,55,92. Reaction networks were set up on the basis of elementary
free-radical steps. The Arrhenius parameters were selected from data bases in the
literature which originate mainly from combustion kinetics82,90. Chen et al.22,23 estimated
the Arrhenius parameters for the most important reactions in a model consisting of 39
elementary reactions by the regression of experimental data covering a wide range of
conditions. The full lines drawn in Figures 1 and 2 are calculated with this model. Figure
4 shows the corresponding typical calculated concentration profiles of the important
radicals along the axis of the laboratory] reactor used by Chen et al 23 . The concentrations
increase initially and reach a maximum corresponding roughly to the maximum in the
axial temperature profile. The radical concentrations are spread over almost 5 orders of
magnitude, with the methyl and hydrogenperoxy radicals being the most and the
hydrogen atom being the least abundant. The hydrogenperoxy radical is present in high
concentrations due to its inactive nature. It is the precursor of hydrogenperoxide, H 2 0 2 ,
the most important chain-branching agent. The hydrogen atoms, together with the
hydroxyl radical are important chain carriers, especially through the hydrogen abstraction
from the methane molecule. The methyl radical is also an important chain carrier not
only in the chain shown in Figure 3 but also for the dehydrogenation of ethane to
ethene.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
p = 400 kPa, T ^ * 1227 K, CH^O^, » 5, V / F c ^ - 0,858 10 2 m3 s mol 1
3. Catalyst development
3.7. Criteria
3.2 Performance
A large variety of inorganic oxides has been tested as catalyst for the oxidative
coupling of methane in the cofeed mode. They can be classified into three major
categories: reducible multivalent metal oxides, alkali and alkaline earth metal oxides, and
rare earth metal oxides.
Table 1. Optimal conversions, selectivities and space-time yields observed at atmospheric pressure*
a. Balance N2 or He.
b. Not reported explicitly.
c. This data was taken after 216 ks in operation under atmospheric pressure at 1123 K, pot s 20 kPa,
CH4=80 kPa, and W / F ^ l . 7 kgs mol"1, when the catalyst performance was at a steady state,from54
ks on. Ten times as much sintered a-Al203 as the catalyst weight were used to dilute the catalyst bed.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
With the exception of one of the Li/MgO catalysts, the data reported in Table 1
correspond to observations performed shortly after the start-up of the reaction. Catalyst
deactivation can occur, however. This is the case for the reducible oxides and the
Li/MgO catalysts. The deactivation of PbO/a-Al203 becomes important after 90 ks and
was attributed to the volatization of Pb and sintering. The loss of C\ causes the selectivity
over Li-Mn02 to decreases sharply after 7 ks without affecting the conversion. The loss
of l i causes deactivation of Li/MgO. However, it follows from Table 1 that a steady-
state behaviour can be reached after 54 ks with a sufficiently high space-time yield.
Promotion by Sn allows to maintain a constant performance during at least 250 ks . The
performance of Na/CaO can be maintained during at least 36 ks. The deactivation
behaviour of Li-Sm203 has not been reported, but similar catalysts such as Sm203 Ca-
Sm203, and Na-Sm203, are stable during at least 216 ks45. It can be expected that li-
Sm203 would show a similar performance. No tests on the deactivation of CaO-La^
nor of the layered Ti0 2 have been reported to our knowledge. A constant performance
during 180 ks was reported for the perovskite listed in Table 1.
4. Catalytic routes
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
It is widely accepted that the primary function of a catalyst for the oxidative
coupling of methane consists of the generation of methyl radicals through the abstraction
of a hydrogen atom. The possibility that subsequent coupling occurs via combination of
two methyl radicals in the gas phase has been demonstrated by isotopic exchange
experiments59'60 and indirectly from the measurement of the methyl radical
concentration in the gas phase1819'28'29. A methyl radical concentration corresponding to
0.006 % of the methane fed was observed over a Li/MgO catalyst28. Note that the
maximum methyl radical concentration in the absence of catalyst shown in Figure 4 is
of the same order of magnitude, which is in line with the similar space time yields
observed in both cases. According to Lunsford54, the coupling of gas-phase methyl radical
contributes to the (^ production for at least 28 % on Na/CaO, 40 % on Li/MgO, and
70 % on L a ^ . These values are a lower limit, since the calculations were based on the
measurement of radical concentrations in the bulk of the gas phase, i.e., beyond the film
surrounding the catalyst pellets. McCarty57 has shown that important methyl radicals
concentration profiles can develop inside the pores of the pellets and in the film.
o2 + □ + io)~ - 4 o; i
o; + c # 4 - OH; + c#3 4
20jy; - o)~ + D + # 2 o 4
Figure 5. The production of methyl radicals on a Li/MgO catalyst [Ito et al., 1985].
<r. stoichiomctric number.
The nature of the active sites on the catalyst is still a matter of debate. In the closed
sequence in Figure 5, the active site is assumed to be the non fully reduced Os" based on
ESR measurements41. Alternative reoxfilation mechanisms have been proposed by Sinev
and Bychkov79. *
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
5. Catalyst properties
From the previous section it can be understood why the correlation of the catalyst
performance and its physico-chemical properties has been studied in particular with
respect to surface basicity, if ethene is looked at as a Lewis base, and solid electrical
conductivity, if the regeneration of active sites is considered to be important. There are
strong indications that basic solids are more appropriate catalysts than acidic solids8. A
positive correlation between basicity and performance was found for PbO catalysts on
different supports17 and CaO-based catalysts10. However, no correlation was observed for
the CaO-L^C^ catalysts, which lead to the conclusion that a certain amount of surface
acidity is also required to obtain a maximum C / -hydrocarbon selectivity for this type of
catalyst7. Another important feature of a solid metal oxide relevant to its performance
in the oxidative coupling of methane is the oxygen ion conductivity within the solid, or
oxygen mobility. Several studies have correlated this property to conversion and
selectivity increase with the oxygen mobility7,67. Two techniques have been employed to
measure oxygen mobility: electrical conductivity measurements and steady-state isotopic
transient kinetic analysis (SSITKA). Baerns and coworkers have shown a positive
correlation between the selectivity and the oxygen ion conductivity for Ce02-CaO
catalysts and promoted L a ^ catalysts. A similar conclusion was reached by Peil et al.67
for MgO-based catalysts and Sm203. Dubois and Cameron30 concluded after a survey of
a large number of catalysts that active and selective catalysts are usually high-
temperature p-type semiconductors under normal operating oxygen partial pressures. The
electrical conductivity not only depends upon the mobility of the charge carriers but also
upon their concentration. Dopants effects, e.g., the replacement of Mg2+ by Li+, can be
explained in these terms66.
Power law rate equations have been derived for several catalysts, e.g. PbO39,
Na/MgO42 and Li/MgO74'86. Such rate equations allow to investigate the effect of the
reactor configuration and operation conditions on the conversions and selectivities that
can be obtained in the industrial methane coupling process76 but not to obtain further
understanding of the functioning of the catalysts.
Rate equations taking into account the interaction of the reactants with the catalyst
through Eley-Rideal or Langmuir-Hinshelwood steps have also been derived for Sm203
63
, Na/MgO42, Li/Sn/MgO44, Na/CaO52 and Li/MgO85'86. In these kinetic models, the
role of the gas phase was restricted to mere coupling of methyl radicals or absent.
Table 2. Kinetic models for the catalytic coupling of methane based on elementary steps
A EJLH /kJmol-
0 2 «■ 2» * 2 0 » (1) 2.5 1Q-3 -11.8
CHt-0* -~CH2* +OH» (2) 1.9 10* 123.1
a
C7H9 - O ♦ - CjH,' * OH * (3) 1.4 10 108.1
C-tfx - 0 ♦ - C / ^ * OH ♦ (4) 1.4 10* 127.8
\ OH* -*H26 * O* - ♦ (5) 6.9 10' 152.7
CH3« + O ♦ ~ CHzO * (6) 3.6 10* 0.0
2 CH30 ♦ + 5 O ♦ - 2 C0 2 * 3 H 2 0 * 7 • (7) OS 0.0
CO + 0 ♦ — COz * • (8) 1.9 10* 93.7
COj ♦ • * C O r » (9) 5.0 10"* -108.4
4 H02» - 3 0 2 + 2 H 2 0 (10) 1.0 10"2 0.0
Units: m3fl mol'1 (1,9), m3, m\ s ' (2,3,4,6,8), mol nr3p s"' (5), m\ m\ s ' (10).
Reactions 2, 3 and 4 are the main source of radicals. Reactions 1 and 5 regenerate
the oxygen species which are involved in the production of radicals. At the investigated
conditions they are close to equilibrium. The catalyst is also involved in non-selective
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
routes. Tong and Lunsford83 showed that methyl radicals are oxidized by feeding methyl
radicals to a bed with Li/MgO, see reactions 6,7 and 8. Clearly reaction 7 is not
elementary, but it occurs on a much smaller time scale than the others and in particular
than reaction 6. Hence, the detailed reaction path through which it occurs is not relevant.
Van der Wiele et al.86 showed that carbon monoxide was relatively rapidly converted to
carbon oxide over Li/MgO by comparing experiments with and without catalyst and by
CO oxidation experiments. It is well known that addition of C0 2 lowers the reaction rate
of the oxidative coupling of methane44'88. This is taken into account by reaction 9. Using
the above catalytic reactions together with gas-phase reactions without considering the
heterogeneous termination, see reaction 10, did not allow an adequate description of the
experimental data. Especially the calculated selectivities at high space times were too
low. This can be attributed to an overestimation of the importance of the gas-phase
reactions occurring in both the pores of the catalyst and in the interstitial phase. Tulenin
et al.84 suggested that MgO does not only produce but also quenches radicals. This
radical quenching has a inhibiting effect on the gas-phase reactions. Therefore the
heterogeneous termination reaction 10 suggested by Cheaney et al.21 was added to the
reaction network. The hydrogenperoxy radical is one of the most important chain carriers
in the gas-phase reactions.
Table 4. Conditions covered by the kinetic model of Couwenberg* for a Sn/Li/MgO catalyst
T /K 973-1023
P . /kPa 110-150
CH^Ojo / mol mor1 2-12
CJHJCHAO / mol mor1 0 - 0.1
COJ/C^IO / mol mor1 0-0.2
/ kg 3 mor1 1 -10
1% 2.5 - 15
/% 15-75
V
Another important question is related to the source of the methyl radicals. If the
chain branching as shown in Figure 3 can be neglected, the major part of the methyl
radicals originates from steps involving the catalyst. Shi et al.78 concluded that at 1 bar
without feed dilution, 973 K and a reactor inlet methane to oxygen ratio of two, the
methane consumption in the gas phasei amounted to 15% of that at the surface of a
Li/MgO catalyst Zanthoff and Baerns93 estimated that on a Pb/Al 2 0 3 catalyst about half
of the converted methane is activated in the gas-phase at 1 bar, 1020 K and a feed
stream composition of CH4:02:N2 ■ 10:1*4.
The ultimate Q yield achievable was estimated by Shi et al.78. An ideal catalyst
which would only activate methane, but not ethane or ethene was considered. The Cj
yield was calculated as a function of the reactants partial pressure PCB+POM- A* a
reactant partial pressure of 100 kPa without dilution, a Q yield of 26% was obtained, in
agreement with the maximum yield observed experimentally. This yield limit was caused
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
In the preceding section only chemical phenomena were discussed, i.e. no attention
was given to the possible effects of transport limitations on the obtainable yields for Q
products. Such effects have indeed been observed33 and discussed by means of models
taking into account both chemical and transport phenomena during reactor
simulations33,72. Depending upon the details of the reaction network increasing pellet
diameters can either lead to a higher or a lower selectivity for Q products33,71. The above
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
131
The kinetic network given in Table 3 was used to calculate the concentration
profiles in a laboratory fixed bed reactor operated at the conditions summarized in Table
5. Strong concentration gradients for reactive
intermediates, e.g. methyl radicals and hydrogen- ™° 5: Conditions used during the
peroxy radicals, are developed without significant simulation (Couwenberg etal«).
gradients for the stable molecules, such as methane,
T 998 K
oxygen, ethane, and ethene. This is shown in figures
P 135 kPa
6 and 7 where the intraparticle and interstitial CH4/O2i0 4.0
concentration of the methyl and the hydrogen-peroxy Fo 1.5 10"4 mol s' 1
radicals are plotted versus the axial reactor position Wc 0.375 10"4 9
and the pellet coordinate. In these figures the zero dp 2 10-4 m
on the pellet-coordinate axis represents the centre of
the catalyst pellet, the gas-solid interface is located
at 1.0 KT4 m, and the space between 1.0 10"4 m and 1.5 1&4 m corresponds to the
interstitial gas phase. The origin of the axial position axis represents the reactor inlet.
The strong concentration gradient for the radicals is caused by the lower time scale for
reaction than for transport by diffusion. This is in contrast to oxygen of which the time
scale for diffusion is lower than the time scale for reaction and, hence, the concentration
gradient can be neglected. In the centre of the pellet, the methyl-radical concentration
is high due to their high catalytic production rate. Near the interface the concentration
decreases strongly, due to diffusion into the interstitial phase. In the interstitial phase the
profile is almost flat because the diffusivity is much higher, but the concentration is
approximately one order of magnitude lower than inside the catalyst pellet.
The intraparticle concentration of the surface-terminated hydrogen-peroxy radical is
much lower than the interstitial concentration because of the high rate of the
heterogeneous termination reaction.
The effects of the diffusion limitations of the reactive intermediates on the selectivity
were investigated by changing the pellet diameters. Based on the calculated
concentration profile of the methyl radical an increase in selectivity is expected, because
the volume where the high methyl radical concentration exists is larger when bigger
pellets are applied. This would favour their coupling towards ethane, since this is a
second order reaction, all other steps being first order with respect to the methyl radical
concentration. The results of these calculations presented in figure 8, however, show a
decrease in the C ^ selectivity with increasing pellet sizes. This is caused by the
important contribution of the heterogeneous termination reaction on the termination of
the branched chain reactions in the interstitial phase. Increasing the pellet diameter leads
to a decrease of the ratio between the gas-solid interfatial surface area and the volume
of the interstitial phase. This results In a lower contribution of the heterogeneous
termination reaction and thus to higher concentrations of the bydrogen-peroxy radical
in the interstitial phase. The higher interstitial H0 2 * concentration results in a higher
rate of the gas-phase reactions in the interstitial phase and thus in lower selectivities.
This was verified by removing the heterogeneous termination from the kinetic
network. The results of the calculations without this reaction are shown in figure 9. It can
be seen that the selectivity indeed increases with increasing pellet size. Another
significant effect of neglecting the heterogeneous termination is the much lower
selectivity, as a result of the increased importance of the non-selective gas-phase
reactions. The calculated methane conversion in figures 8 and 9 is constant with
increasing pellet diameter, because the intraparticle concentration gradient for the
molecules involved in the kinetic network can be neglected. The rate of methane
consumption is mainly determined by the rate of heterogeneous methyl radical
generation which is only depending on the concentrations of methane, oxygen and carbon
dioxide.
65i
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
* L ^
ft2
r \
f" \
O 93 [
5o' ' ' ' ' ■ '
OOOOO 0.0001 00002 0.0003 O0004 OJOOOO 0.0001 OO002 0.0003 O0004
FigureftCalculated selectivity versus the pellet Figure 9; Calculated selectivity versus the pellet
diameter using the kinetic network with a diameter using the kinetic network without a
heterogeneous termination. Conditions see heterogeneous termination. Conditions see
Table 5 (Couwenberg etal"). Table 5 (Couwenberg etal.9*).
8. Conclusions
reactions towards CO are detrimental for the selectivity at which methane is converted
into ethane and ethene. One of the major functions of the catalyst consists in increasing
the methyl radical concentration and, hence, the selectivity of methane conversion
without increasing the concentrations of the chain carriers towards CO. Only limited
success has been met so far. None of the presently developed catalysts allows to avoid
side reactions of the reaction intermediates or consecutive reactions of the reaction
products limiting the obtained yields for ethane and ethene to 20%. The kinetics of the
occurring reactions are rather well understood and can be adequally described, however,
despite the complexity of the involved chemistry and of the interplay between chemical
kinetics and transport phenomena. This should provide ways for further improvement of
the process.
9. Acknowledgement
The financial support by the Commission of the European Communities in the
framework of the Joule programme, subprogramme Energy from Fossil Sources,
Hydrocarbons, Contract No. JOUF-0044-C, is acknowledged.
References
1. Y. Amenomiya, V.I. Birss, M. Goledzinowski, J. Galuszka and A.R. Sanger, Catal.
Review-Science and Engineering, 32 (1990) p. 163.
2. J.R. Anderson, Applied Catalysis, 47 (1990), p. 177.
3. Z.S. Andrianova, A.N. Ivanova, P.E. Matkovskii and G.P. Startseva, Kinetics and
Catalysis, 34 (1993) p.344.
4. L.M. Aparicio, S.A. Rossini, D.G. Sanfilippo, J.E. Rekoske, A.A. Trevino, J.A.
Dumesic, Ind. Eng. Chem. Res., 30 (1991) p.94.
5. K. Asami, K. Omata, K. Fujimoto, and H. Tominaga, Energy Fuels, 2 (1988) p.574.
6. R.C. Baetzold, and G.A. Somorjai, Journal of Catalysis, 45 (1976) p.94.
7. M. Baerns, Oxidative Coupling of Methane for the Utilization of Natural Gas. Paper
presented at Nato-Asi, London, Ontario, Canada (1991).
8. M. Baerns, Basic solids as catalysts for the oxidative coupling of methane. In:
Methane conversion by oxidative processes. Fundamental and engineering aspects.
(Edited by E.E. Wolf), pages 382-402. Van Nostrand Reinhold, New York (1992).
9. M. Baerns and J.R.H. Ross, Catalytic chemistry of methane conversion. In:
Perspective in Catalysis (edited by J.M. Thomas and K.I. Zamaraev), the IUPAC
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
monograph, (1991).
10. S. Becker and M. Baerns, Journal of Catalysis, 128 (1991) p.512.
11. F.G. Billaud, C.P. Gueret, F. Baronnet and J. Weill, Ind. Eng. Chem. Res., 31 (1992)
p.2748.
12. M. Bistofli, G. Fornasari, M. Molinari, S. Palmery, M. Dente and E. Ranzi, Chem.
Eng. Sc, 47 (1992) p.2647.
13. H. Borchert, Z.L. Zhang and M. Baerns, The effect of oxygen ion conductivity of
catalysts for their performance in the oxidative coupling of methane. Paper presented
at the American Chemical Society's 1992 National Meeting in San Francisco (1991).
14. M. Boudart, Kinetics of chemical processes; Prentice-Hall; Englewood Cliffs, New
Jersey (1968).
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
134
15. M. Bowker, LB. Parker and K.C. Waugh, Applied Catalysis, 14 (1985) p.lOl.
16. J.H. Bropy, Conversion process. EP 0178855 A2 (1985).
17. W. Bytyn and M Baerns, Applied Catalysis, 28 (1986) p.199.
18. K.D. Campbell, E. Morales and J.H. Lunsford, Journal of American Chemical
Society, 109 (1988) p.7900.
19. K.D. Campbell and J.H. Lunsford, Journal of Physical Chemistry, 92 (1988) p;5792.
20. JA.S.P. Carreiro and M. Baerns, Journal of Catalysis, 117 (1989) p.396.
21. D.E. Cheaney, D.A. Davies, A. Davis, D.E. Hoare, J. Protheroe and A.D. Walsh,
Effects of surfaces on combustion of methane and mode of interaction of antiknocks
containing metals. 7th Symposium on Combustion, Butterworths Scientific
Publications, London, (1959) p.183.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
22. Q. Chen, J.H.BJ. Hoebink and G.B. Marin, Ind. Eng. Chem. Res., 30 (1991) p.2088.
23. Q. Chen, P. Couwenberg and G.B. Marin, AIChEJ, 40 (1994) p.521.
24. Q. Chen, P. Couwenberg and G.B. Marin, Catalysis Today, 21 (1994) p.309.
25. P. Chu and M.E. Landis, Selective oxidative coupling. US patent 4,914,252 (1990).
26. P. Couwenberg, PhD Thesis, Eindhoven University of Technology (1995).
27. K.M. Dooley and J.R.H. ROSS, Appl. Catal., 90 (1990) p.159.
28. DJ. Driscoll and J.H. Lunsford, Journal of Physical Chemistry, 89 (1985) p.4415.
29. DJ. Driscoll, W. Martir, J.-X. Wang and J.H. Lunsford, Journal of American
Chemical Society, 107 (1985) p.58.
30. J.-L. Dubois and C.J. Cameron, Applied Catalysis, 67 (1990) p.49.
31. J.H. Edwards, K.T. Do, RJ. Tyler, The Oxco process for natural gas conversion via
methane oxidative coupling. Proc. Symp. Natural gas conversion, Oslo, (1990) p.489.
32. M.G. Evans and M. Polanyi, Faraday Society, 32 (1936) p.1333.
33. G. Follmer, L. Lehmann and M. Baerns, Catalysis Today, 4 (1989) p.323.
34. O. Forlani, M. Lupieri, V. Piccoli, S. Rossini, D. Sanfilippo, J.A. Dumesic, L.A.
Aparicio, J.A. Rekoske and A.A. Trevino, New developments in selective oxidation
(edited by G. Centi and F. Trifiro), (1989) p.417. Elsevier, New York.
35. J.W.M.H. Geerts, J.H.B J. Hoebink and K. van der Wiele, Ethylenefromnatural gas.
Proven and new technology. Paper presented at the American Chemical Society's
1990 National Meeting in Bosten (1990a).
36. J.W.M.H. Geerts, J.M.N. van Kasteren and K. van der Wiele, Catalysis Today, 4
(1989) p.453.
37. J.W.M.H. Geerts, Q. Chen, J.M.N. Van Kasteren and K. van der Wiele, Catalysis
Today, 6 (1990) p.519.
38. D.M. Ginter, E. Magni, G.A. Somorjai, H. Heinemann, Catal. Lett., 16 (1992) p.197.
39. W. Hinsen, W. Bytyn and M. Baerns, Oxidative dehydrogenation and coupling of
methane. Proceedings of the 8th international congress on Catalysis, 3 (1984) p.581.
40. MJ. Howard, The homogeneous partial oxidation of methane-containing parafflnic
hydrocarbon. EP 0302665 Al (1989)1
41. T. Ito, J.-X. Wang, C-H. Lin and J.H. Lunsford, Journal of American Chemical
Society, 107 (1985) p.5062.
42. E. Iwamatsu and K.-I. Aika, Journal of Catalysis, 117 (1989) p.416.
43. G.E. Keller and M.M. Bhasin, Journal of Catalysis, 107 (1989) p.9.
44. SJ. Korf, Catalysts for the oxidative coupling of methane. Ph.D. Dissertation,
University of Twente, NL (1990).
45. S.J. Korf, J.A. Roos, J.M. Diphoorn, R.HJ. Veehof, J.G. Van Ommen and J.R.H.
Ross, Applied Catalysis, 52 (1989b) p.119.
46. O.V. Krylov, Kinetics and Catalysis, 34 (1993) p.l.
47. O.V. Krylov, M.U. Kislyuk, B.R. Shub, A.A. Gezalov, N.D. Maksimova and Yu.N.
Rufov, Kinetika i Kataliz, 13 (1972) p.598.
48. J.A. Labinger, Catal. Lett., 1 (1988) p.371.
49. J.A. Labinger and K.C. Ott, Journal of Physical Chemistry, 91 (1987) p.2682.
50. G.S. Lane and E.E. Wolf, Journal of Catalysis, 113 (1988) p.144.
51. J.S. Lee and S.T. Oyama, Catal.Review Science and Engineering, 30 (1988) p.249.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
52. L. Lehmann and M. Baerns, J. Catal., 135 (1992) p.467.
53. J.H. Lunsford, Catalysis Today, 6 (1990) p.235.
54. J.H. Lunsford, Methane conversion by oxidative processes. Fundamental and
engineering aspects, (edited by E.E. Wolf), (1990) p.3. Van Nostrand Reinhold, New
York.
55. J.C. Mackie, Catal.Review Science and Engineering, 33 (1991) p.169.
56. J.G. Mc Carty, Methane conversion by oxidative processes. Fundamental and
engineering aspects, (edited by E.E. Wolf) (1990) p.321. Van Nostrand Reinhold,
New York.
57. J.G. Mc Carty, A.B. Mc Ewen and M.A. Quinlan, New developments in selective
oxidation (edited by G. Centi and F. Trifiro) (1990) p.405. Elsevier, New York.
58. D.F. Mc Millen and D.M. Golden, Annual review physical chemistry, 33 (1982)
p.493.
59. C.A. Mims, R.B. Hall, K.D. Rose and G.R. Myers, Catalysis letters, 2 (1989) p.361.
60. P.F. Nelson, C.A. Lukey and N.W. Cant, Journal of Physical Chemistry, 92 (1988)
p.6176.
61. T. Norby and A.G. Andersen, Applied Catalysis, 71 (1991) p.89.
62. O.T. Osager, R. Lodeng, P. Soraher, H. Anundskaas and B. Helleborg, Catal. Today,
4 (1989) p.355.
63. K. Otsuka and K. Jinno, Inorg. Chim. Acta, 121 (1986) p.237.
64. K. Otsuka, Q. Liu, M. Hatano and A. Morikawa, Chemistry Letters (1986a) p.467.
65. K. Otsuka, Q. Liu, M. Hatano and A. Morikawa, Chemistry Letters (1986b) p.903.
66. D. Papageorgiou, D. Vamvouka, D. Boudouvas, and X.E. Verykios, Catalysis Today,
13 (1992) p.391.
67. K. Peil, G. Marcelin and J.G. Goodwin, Methane conversion by oxidative processes.
Fundamental and engineering aspects, (edited by E.E. Wolf) (1992) p.3. Van
Nostrand Reinhold, New York.
68. P. Pereira, S.H. Lee, G.A. Somorjai and B. Heinemann, Catal. Lett., 6 (1990) p.255.
69. M.G. Poirier, A.R. Sanger, and K.J. Smith, The Canadian Journal of Chemical
Engineering, 69 (1991) p.1027.
70. J. Rasko, P. Pereira, G.A. Somorjai and H. Heinemann, Catal. Lett., 9 (1991) p.395.
71. S.C. Reyes, C.P. Kelkar and E. Iglesia, Catalysis Letters, 19 (1993a) p.167.
72. S.C. Reyes, E. Iglesia and C.P. Kelkar, Chem. Eng. Sc, 48 (1993b) p.2643.
73. O.A. Rokstad, O. Olsvik, B. Jenssen and A. Holmen, Novel Production Methods for
Ethylene, Light Hydrocarbons and Aromatics (L.F. Albright, L. Crynes, S. Nowak,
eds.) Marcel Dekker, New York (1992) p.259.
74. J.A. Roos, S.J. Korf, R.HJ. Veehof, J.G. Van Ommen and J.R.H. Ross, Applied
Catalysis, 52 (1989) p.131.
75. J. Sanchez-Marcano, C. Mirodatos, E.E. Wolf and G.A. Martin, Catalysis Today, 13
(1992)p.227.
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
136
76. J.M. Santamaria, E.E. Miro and E.E. Wolf, Ind. Eng. Chem. Res., 30 (1991) p.1157.
77. N. Semenov, Some problems in chemical kinetics and reactivity. Translated from
russian by Boudart, M.; Princeton University Press; Princeton, New Jersey (1958).
78. C. Shi, M. Hatano and J.H. Lunsford, Catalysis Today, 13 (1992) p.191.
79. M.Yu. Sinev and V.Yu. Bychkov, Kinetics and Catalysis, 34 (1993) p.272.
80. J.A. Sofranko, J.J. Leonard and GA. Jones, Journal of Catalysis, 103 (1987) p302.
81. G.A. Somorjai, PhiLTrans.R.Soc. London A318 (1986) p.81.
82. W. Tsang and R.F. Hampson, Journal of Physical Chemistry, Reference Data, 15
(1986) p.1087.
83. Y. Tong and J.H. Lunsford, J. Am. Chem. Soc, 113 (1991) p.4741.
84. Y. Tulenin, A. Kadushin, V. Selezpev, A. Shestakov and V. Korchak, Catalysis
Today, 13 (1992) p329.
85. W.-Y. Tung and L.L* Lobban* I&BC Research, 31 (1992) p.1621.
86. K. van der Wide, J.W.M.H. Geerts and J.M.N. van Kasteren, Methane conversion
by oxidative processes. Fundamental and engineering aspects, (edited by EJE. Wolf),
pages 259-319, Van Nostrand Reinhold, New York (1992).
87. W. Vermeieren, I.D.M.L. Lenotte, J.A. Martens and P.A. Jacobs, Perovskite-type
complex oxides as catalysts for the oxidative coupling of methane. In: Natural gas
conversion, (edited by A. Holmen, K.-J. Jens and S. Kolboe) (1^1) p33. Elsevier,
Amsterdam, The Netherlands.
88. D. Wang, M. Xu, C Shi and J.H. Lunsford, Catalysis Letters, 18 (1993) p323.
89. J. Warnatz, Ber. Bunsenges. Phys. Chem., 87 (1983) p.1008.
90. J. Warnatz, Rate coefficients in the C/H/O system. In: Combustion Chemistry
(Edited by Gardiner, W.C.), Springer Verlag, New York (1984).
91. P.B. Weisz, CHEMTECH, (1982) p.424.
92. H. Zanthoff and M. Baerns, Ind. Eng. Chem. Res., 29 (1990) p2.
93. H. Zanthoff and M. Baerns, Combined kinetics of catalytic and non catalytic
reactions in the oxidative coupling of methane. Paper presented at the American
Chemical Society's 1992 National Meeting in San Francisco (1991).
94. V.P. Zhdanov, J. Pavlicek and Z. Knor, CataLReviews-Science and Engineering, 30
(1988)p.501.
95. P.M. Couwenberg, Q. Chen and G.B. Marin, Plenum, submitted (1994).
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
ABSTRACT
The principles and present-day embodiments of fuel cells are discussed. Nearly
all cells are hydrogen/oxygen ones, where the hydrogen fuel is usually obtained on-
site from the reforming of methane or methanol. There exists a tension between
the promise of high efficiency in the conversion of chemical into electrical energy
and of very low emissions of noxious compounds, and the enormous difficulty of
manufacturing the fuel cells cost-effectively. After three decennia of widespread
effort to adapt the fuel cell to terrestrial applications, it is still too early to say
whether their large-scale introduction will prove to be viable.
1. Introduction
engine, which is subject to the Carnot principle, the available energy can, in principle,
be completely transformed into electrical energy. Thermodynamicaily we have:
AG° = -nFE° (3)
137
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
138
Fuel in Oxidant in
where AG° is the standard free energy change of the cell reaction, n the number of
electrons involved (two for hydrogen), F the Faraday constant (96500 C mol -1 ), and
E° the standard cell voltage. It is because of this promise of a very high conversion
efficiency that people have been coming back to the fuel cell. Nowadays, the promise
of very low production of environmentally objectionable compounds, e.g. of zero
NOJC emission, is seen as being quite as important, with Japan leading the way. Both
these aspects, in fact, played their part in assuring fuel cells a prominent place in the
energy scenario's discussed at the UNCED \ Rio de Janeiro, 1992. Fuel-cell systems
are also characterized by very low noise levels.
The electrolyte within the cell can be of various types. In low-temperature fuel
cells one employs aqueous acid or base, either as such or immobilized in a matrix,
or a solid polymer membrane. At intermediate temperatures one goes for a molten
salt, and for high-temperature (about 1000°C) applications a ceramic membrane
is used. Indeed, most fuel cells are characterized by the type of electrolyte upon
which they are based. In this contribution we will discuss, however briefly, alkaline
fuel cells (AFC), phosphoric-acid fuel cells (PAFC), polymer electrolyte fuel cells
(SPFC), direct methanol fuel cells (DMFC), the one exception here because it is
distinguished by the fuel it converts, molten-carbonate fuel cells (MCFC), and solid-
oxide fuel cells (SOFC). The low-temperature cells, being easier to engineer, are
closest to full commercialization, but the higher-temperature ones promise higher
overall efficiencies.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Cell voltage
A
Mixed potential formation because of side-reactions etc.
I
1
13
O
TJ l_
CD o
+J
c 0)
U Q. o Usable voltage
O 0 Voltage drop in
o ■o 0 electrolyte
£ 3
<U I/)
O
2, f
\f
Mixed potential formation because - 1 Cell current
of side-reactions etc.
Fig. 2. Schematic representation of the processes leading to loss of cell voltage when current is drawn.
The measured efficiency is always lower than the calculated one, for a variety
of reasons. To begin with, even at open-circuit conditions the measured cell voltage
is often, certainly in the case of low-temperature cells, less than the thermodynamic
value due to, for example, side-reactions or the presence of impurities leading to
the formation of a mixed potential. When current is drawn, "polarization" occurs
at both electrodes (see Fig. 2) as the electrode reactions are not infinitely fast, i.e.
it takes some energy to make them go at an appreciable rate. Polarization can be
caused by obstruction of electron transfer between the electrode and the reacting
molecule (charge-transfer polarization), by limitations in mass transfer (concentra
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Before discussing the individual types of FC, some historical background will
be providedfirst.The whole chapter is largely based on the material that can be found
in refs. [2-6].
The fuel cell has a rather long history, relatively speaking. Electrochemistry
began in earnest only around 1800, and already in 1839 thefirstfuel cell (hydrogen/
oxygen with Pt electrodes) was described by Sir William Grove (who went on to
become Lord Chief Justice, for which, in fact, he was knighted...). At the time, how
ever, it was not looked upon as a power generator, but as one of the devices showing
the interconversion of the various "forces", and indeed it was somewhat lost sight
of after the establishment of the first law of thermodynamics (although Grove did
emphasize the need for a "notable surface of action", meaning a high area of contact
between the electrolyte, the gaseous reactant, and the electrocatalytic conductor, cf.
above). The former aspect, however, was much canvassed at the turn of the century,
with the emphasis shifting from hydrogen to a much more practical fuel like coal, and
this was mainly due (i) to the great improvements in the electrical industry, and (ii)
to the conversion efficiency of coal into electricity being well below 10% at the time.
Despite the hopes of Ostwald and Jacques ("Think of a smokeless London!") that the
20th Century would become the Age of Electrochemical Combustion, the fuel cell
did not make it, defeated for the moment by the low reactivity of common fossil fuels
and by the emergence of the internal combustion engine.
The story restarts with Sir Francis Bacon (who was knighted for his fuel-
cell work), who from 1933 onwards pioneered the cells that were eventually to be
employed in the Apollo flights (1960s), after a concentrated development effort by
the Pratt & Whitney division of United Technologies, Inc. These were H2/O2 fuel
cells, employing an alkaline electrolyte (KOH), sintered Ni being the anode, and
porous lithiated NiO the cathode. It is fair to say, that space travel has played a very
important role in keeping fuel cells aHve, and indeed H 2 /0 2 cells are still being used
in the Challenger flights (and cf. below). Large-scale terrestrial applications have
still to materialize, although they have been predicted to be imminent for some time
now.
In the late 1950s, early 1960s the first polymeric acidic membrane fuel cell
(H2/O2, Pt-based electrodes) was developed by General Electric, and applied in
die Gemini flights. This type of cell was abandoned for a time in favour of the
phosphoric-acid cell (PAFC), but has recently resurfaced. The PAFC was developed
in an attempt to convert carbonaceous fuels directly, but again these turned out to
be too unreactive despite the relatively high temperatures that can be reached with
phosphoric acid (around 200°C), and so one has returned to hydrogen here as well.
The use of polytetrafluoroethylene (PTFE, "Tfeflon") as a wet-proofing agent was very
important, in that it allows almost all of the high-surface-area Pt in an electrode to be
actually contributing to the activity.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
141
At about the same time, work on the molten-carbonate fuel cell was started by
Ketelaar en Broers in the Netherlands. They employed Ni-based electrodes similar
to Bacon's. Not much later, the solid-oxide fuel cell was introduced, where the doped
zirconia electrolyte was adapted from the "Nernst glower" of 1900. Here again,
hydrogen is the ftiel of choice.
Still in the 1960s, the direct methanol fuel cell came to the fore as an attractive
candidate for such low-power applications as vehicle traction. Methanol was widely
seen as the best compromise between reactivity, handleability and cost. Such oil
majors as Exxon and Shell were involved in this effort for a time, but the promise has
still to be redeemed.
We will skip the last two decades, and turn instead to a slightly more de
tailed description of the various types of fuel cell in their more or less present-day
embodiments. Apart from the DMFC, all FCs are effectively H2/O2 cells. It should
be realized that they have only about 1 V/cell, so that stacks have to be built to
obtain useful voltages, but the technologies involved will not be discussed. It is worth
pointing out, however, that this modularity makes FC systems very flexible.
(a) when using air, one has to scrub it to remove the C0 2 , which would otherwise
carbonate the electrolyte with deleterious effect on the cell performance, (b) car
bonaceous fuels cannot be used for the same reason, CO2 being the product of the
electrooxidation reaction, and (c) AFCs are not very compact (limited power density).
At present, several applications are envisaged for the AFC, apart from their
continued use in space travel by UTC. An example is Elenco NV's study of their
suitability for transportation purposes (e.g., in city buses), with circulating electrolyte
and with the hydrogen fuel being stored on board in liquid form. A similar objective
is pursued by Siemens, but in this case submarines are the primary target. Also in the
Catalyst particle^
Catalyst agglomerate
ELECTROLYTE
the reactant gas with a concomitant increase in diffusion polarization, nor too small,
which would lead to important ohmic losses (the electrode should not be too thick
for the same reason). Finally, the Ifeflon content should be finely tuned, in that too
much will make the whole electrode too hydrophobic, preventing thefloodingof the
agglomerates, and too little will compromise the gas-diffusion characteristics and the
structural integrity. These problems can all be satisfactorily solved. (Often an extra
porous Ifeflon layer is applied at the gas side of an electrode to prevent the electrolyte
from entering the gas compartment).
The Grove cell featured aqueous sulfuric acid as the electrolyte, which limits
the operating temperature to about 90°C. Since, ceteris paribus, higher temperatures
mean higher efficiencies, the switch was later made to phosphoric acid, which can be
used at temperatures of up to 210°C. At the latter temperature it actually exists as
pyrophosphoric acid, H4P2O7.
Both anode and cathode are carbon-supported Pt-based wet-proofed Tfeflon-
bonded electrodes [about 0.5 mg Pt/cm2 (geometric)], while the electrolyte is immo
bilized in a SiC matrix. The carbon support is preferably a high-surface-area (HSA)
graphite, HSA since the Pt dispersion increases (Pt particle size decreases) with in
creasing carbon surface area, and graphite because it is more corrosion resistant than
other types of carbon. Alloying of the Pt is frequently practiced, not only to improve
its activity, especially for oxygen reduction (see below), but also for stability reasons.
Ternary alloys such as Pt/Co/Cr and Pt/Co/Ga have been mentioned for the cathode;
one still has metal leaching in these cases, but the Pt particle growth can be largely
suppressed. Alloys like PtPd can be used in the anode; here again Pt dispersion is
better maintained, and the poison resistance (CO, H2S) is increased.
As mentioned in the introduction, the H2 feed for a fuel-cell system is often
generated through the steam-reforming of, e.g., methane:
CH4 + H2O -► CO + 3H 2 (4)
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
CO + H 2 0 -* CO2 + H2 (5)
For transportable and vehicular type systems, one often prefers methanol as
the primary fuel to reform:
CH3OH -» CO + 2H 2 (6)
CO+H2Q -► C0 2 H-H 2 (7)
In both cases the hydrogen is liable to contain some CO, and the technical
success, such as it is, of the PAFC system is due to the fact that, although the
efficiencies are relatively modest (about 40-45% for FC plus ancillaries), it can cope
much better with a few percent CO in the anode feed than the other low-temperature
FCs.
The phosphoric-acid utility power concept was developed by United Technolo
gies Corp. (UTC), later International Fuel Cells Corp. (DFC), in the US, leading to
the 4.8 MW project in Manhattan. The power plant never came on stream, however,
due to several unexpected difficulties largely unconnected with the fuel-cell stacks
themselves. The technology wasfinallylicensed to Ibkyo Electric Power Co., which
resulted in the successful demonstration of an 11 MW utility plant, with an overall ef
ficiency of around 40% (the fuel is natural gas). At the present time, however, PAFCs
are expected to be mainly used in such areas as hospitals, paper mills, municipal waste
treatment plants and chlor-alkali production.
Some attention is also being paid to PAFC application in the transportation
sector, with (reformed) methanol as the fuel, but this would appear to have to take
the backseat vis-a-vis the polymeric membrane FCs, to which we will now turn.
to produce a more intimate contact between the catalyst and the electrolyte. The
MEA is usually bonded together by heating the components under pressure to
a temperature at which the polymer softens. Techniques have been developed to
emplace the Pt particles exactly there where they are wanted, i.e. at the three-phase
boundary, so that high-performance cells do not need more than 0.2 to 0.3 mg Pt/cm2.
The vehicle-traction application is under intense scrutiny, for example, at Bal-
lard Technologies (who have teamed up with Dow as the membrane manufacturer),
Los Alamos National Laboratory, and Siemens. The operating pressure is envisaged
to be somewhat elevated, 2-5 atm, operating temperatures being of the order of
70-100°C. Many of the present automotive programmes envision using methanol
reformed on board of the vehicle as fuel, which requires (i) efficient removal of
CO, and (ii) fast load-following of the reformer. Another area requiring attention is
the cost (and, indeed, the performance) of the polymer electrolytes. The water/heat
management, though not easy, seems to be under control. The major challenge left is
whether the PEMFCs can be designed and mass-produced cost-effectively.
step is the dual-site dissociation of the hydrogen molecule. Although this is a non-
electrochemical step, the reaction rate is still a function of the potential, because the
dihydrogen oxidation reaction is self-poisoned by adsorbed hydrogen atoms, Had, and
the hydrogen atom adsorption isotherm is a function of the polarization. At open
circuit, the platinum surface is nearly completely covered with Had, and the coverage
decreases approximately linearly with the polarization to reach zero at about 0.3 V,
where therefore the oxidation current is at its maximum.
The H2 oxidation reaction over Pt is very fast indeed, and the only major
complaint here is the sensitivity of Pt for CO. Especially in the case of the PEMFCs,
which work at a much lower temperature than the PAFCs, this is a problem: the fuel
should be essentially free of CO (<5 ppm). This can in principle be achieved with a
selective oxidizer (e.g. 1% Pt/alumina at 125°C with 0 2 /CO about 2/1), if the starting
CO level is not too high (a few percent at most). There is as yet no alternative to Pt as
an anode catalyst; WC has been a candidate for some time because it is insensitive to
CO, but its activity could not be increased to acceptable levels.
Thefirsttwo steps in the electrochemical reduction of dioxygen over Pt in acid
are:
0 2 + H + + e + Pt = P t - 0 2 H (8)
P t - 0 2 H + Pt = P t - O + P t - O H (9)
after which the adsorbed O and OH species are further reduced to H 2 0. The
Fig. 4. Specific activity for oxygen reduction vs. electrocatatyst nearest neighbor distance; 100%
H 3 PO 4 ,200°C 8 .
H 2 0 -► - O H + H+ + e (11)
The general view is that it is the activation of water, step (11), coupled with the
strong adsorption of COHad (or its product, GOad), that is the difficulty here. Many
elements and compounds have been evaluated as possible promoters of the activity
of Pt, the most promising system being PtRu supported on carbon. The action of Ru
is generally discussed in terms of its accounting for reaction (11) — the bifunctional
mechanism — but there are also indications that it modifies the bonding between
Pt and the methanolic residue (ligand effect). In addition, more complete oxidation
of the methanol is observed, thus avoiding the formation of formic acid, and hence
methylformiate. The surface coverage of Pt should be relatively low, because the
adsorption of methanol requires quite a large ensemble of Pt atoms, which would not
be available were the Ru coverage to increase much beyond 0.1 monolayer.
With both the MeOH oxidation and the 0 2 reduction being on the very slow
side, commercialization of the DMFC has always looked a long way off, but the advent
of the new generation of PEMFCs has increased hope again. Indeed it seems that it is
not so much the slowness of the oxidation reaction that is limiting the possibilities of
the PEMDMFC, but the migration of MeOH through the membrane to the cathode,
where it interferes with the reduction of dioxygen. So, this is one problem to solve,
and one should also have a close look at the stability of the PtRu electrode under
operating conditions.
ture fuel cells, because of the higher quality of the waste heat. Industry's high
heat-to-electricity utilization limits the advantages of FCs in any case. Nevertheless,
molten-carbonate and solid-oxide fuel cells are considered to be competitive in base-
load applications, given a lifetime of at least 40,000 hours. Application in ships and
submarines is also envisaged, however.
Of the higher-temperature fuel cells the MCFC is technically the most ad
vanced. Its standard fuel is steam-reformed methane (natural gas) or naphtha. But
a coal gasifler + FC can also be an interesting alternative (e.g., to pulverized coal
combustion), and at the UNCED conference FCs have been discussed in terms of
biomass as the primary fuel *. The electrode reactions are:
anode: H2 + COf" -> H 2 0 + C0 2 + 2e (13)
(CO + COf- -* 2C0 2 + 2e)
cathode: 0 2 + 2C0 2 + 4e -► 2CO^~" (14)
The anode catalyst is Ni, the cathode one is NiO, while the electrolyte consists
of a Li2C03/K2C03 eutectic melt immobilized in a LiA102 tile. The cell is operated
at about 650°C. Under these conditions the presence of CO in the H2 stream is
not problematic. Both the cathodic and anodic reactions are fast, i.e. on a par with
the oxidation of H2 over Pt in acid solution (note that C0 2 needs to be cofed with
0 2 to the cathode). Both electrodes are porous with the pores (partially) filled with
electrolyte — improved performance, therefore, depends o.a. on the possibility of
reducing the electrolyte resistance and of producing a large internal surface area.
The development of MCFCs have only just reached the demonstration phase.
Problem areas include: (i) cathode (NiO) dissolution during cell operation — can be
counteracted by doping the cathode, adding elements to the electrolyte, and lowering
the C0 2 partial pressure, (ii) creep; electrolyte loss and migration phenomena;
corrosion—the seriousness of these phenomena depends on the seal material and its
design, and on the material selection and surfacefinishfor the other cell components,
and (iii) compatibility of cell and stack component materials. Single-cell lifetimes
have reached the 20,000 hours mark, and people are now studying the behaviour of
!
large stacks (100 kW units).
A hot topic here is so-called "internal" reforming, meaning that the (endother-
mic) methane reforming reaction is carried out inside the anode compartment (at
650°C), instead of externally (at 800^), see Fig. 5. Getting the reaction going at
650°C, however, is an as yet unsolved problem, and the "direct*1 variant suffers also
from catalyst deactivation due to electrolyte components evaporating through the
anode. Realization of the internal reforming option would boost the efficiency of the
MCFC to 60-^5%.
The solid-oxide fuel cell has the highest operating temperature of all, viz.
approximately 1000°C. Because of this high temperature, internal reforming presents
no problem, which is good for system simplicity. Power generation efficiency is
reasonably high at 55% or so. The all solid-state cells usually show highly stable
performance. It isin fact the SOFC that has aroused most industrial interest.
The clear technology leader is Westinghouse Electric Corp., who started out in
this field in the mid-Seventies, but there are many other important players (in North
America, Japan, Europe). The development of SOFCs is a purely ceramic affair,
and the SOFC fabricator maxim's is reputed to be: "They who control the materials,
control the technology". Barriers to progress in the larger-scale manufacture are: (i)
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
• ••• • • •
CH4+H20
H 2 CO ♦
♦ i
♦
C0 2 H 2 0
H 2 CO i ♦
^ j C 0 2 H20
IZ
C02+H20 **
ANODE AN00E C0 2 +- H20
CATHODE C OCTHOOfc
— ♦ ♦ I
o 2 co 2 ^ 1 °2 C02
-TAirCCg) Air(02)+C02
EXTERNAL REFORMING
650°C
H 2 + C0
H 2 CO ♦ t
Reforming
K 8 0 0 °C
f I
AWODE
C0 2 H20
MATRIX C 0 3 t
A C0 2 + H 2 0
catalyst
CATHODE
t ♦
05 C0 P
CH4+H20 I J Air(0 2 ) + C 0 2
assuring that a cell is functioning well (e.g. no cracks in the electrolyte) for a large
majority of cells produced, (ii) achieving powder starting materials consistency, and
(hi) correct matching of the thermal expansion coefficients of the various cell (and
stack) components, to prevent failure after thermal cycling.
The electrode reactions are as simple as you can get:
anode: H2 + 0 2 _ H20+2e (15)
2
cathode: C>2 + 4e 20 " (16)
The anode is a porous Ni-Cermet, fabricated as a mixture of yttria-zirconia
and NiO which is converted (reduced) to the conductive cermet in situ within the cell.
The electrolyte is an impervious yttria-stabilised zirconia (at least 94% of theoretical
density), which is a stable, and reasonably good, oxygen-ion conductor, and a good
insulator, at 1000°C. The cathode is a perovskite, e.g. (La,Sr)Mn03. It performs
better than, for example, Pt, because in the latter case only surface diffusion of Oad
to the electrocatalyst/electrolyte interface can take place (a rate-limiting step), while
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
in the former also bulk diffusion occurs, thus increasing the flux of O to where the
action is. Another advantage of perovskites is that they form a rather stable interface
with the zirconia electrolyte (insensitive to the O2 partial pressure), in contrast with
Pt and such good O^/e conductors as the new high-temperature superconductors.
The interconnect between individual cells is often also a perovskite, viz. a Ca,
Mg, or Sr-doped LaCr03, but metal alloys are also being studied, as they possess
the advantages of much higher electrical and thermal conductivities. Their thermal
expansion coefficients have to be watched, however, and they have of course to satisfy
the requirements of mechanical stability, resistance to corrosion, etc.
In spite of all the progress made over the years, the projected costs for
electricity generation (utility application) remain about two times higher than those
for a conventional system.
6. Concluding remark
Fuel-cell technology has the reputation of being "eternally promising". That is,
for some time now the large-scale introduction of FCs of all types has been predicted
to be imminent, without reality, as is its wont, conforming to this prediction. However,
given the amount of work being undertaken and the results being achieved, we may
yet see it come true soon.
References
1. T.B. Johansson, H. Kelly, A.K.N. Reddy and R.H. Williams (Eds.), Renewable Energy-Sources for
Fuels and Electricity (Earthscan/Island Press, 1993) (input to the UNCED process).
2. W.J. Albery, Electrode Kinetics (Clarendon Press, Oxford, 1975).
3. A.J. Appleby, C.K. Dyer, P.T. Moseley and D.A.J. Rand, Proceedings of the Third Grove Fuel Cell
Symposium, J. Power Sources 49 (1994).
4. L.J.MJ. Blomen and M.N. Mugerwa (Eds.), Fuel Cell Systems (Plenum Press, New York, NY,
1993).
5. D.G. Lovering, Proceedings of the Grove Anniversary Fuel Cell Symposium, J. Power Sources 29
(1990).
6. W. Vielstich (Ed.), Proceedings of a Discussion Meeting on "Fuel cells and their Applications", Ben
Bunsenges. Phys. Chem 94, No. 9 (1990).
7. J. Giner and C. Hunter,/. Electrochem. Soc. 116 (1969) 1124.
8. V Jalan, and E.J. Taylor, /. Electrochem. Soc. 130 (1983) 2299.
9. J.A.R. Van Veen and J.E van Baar, Rev. Inorg. Chem. 4 (1982) 293.
10. K. Kishida, in: Proceedings of a Discussion Meeting on "Fuel cells and their Applications"
(W. Vielstich, Ed), Ber. Bunsenges. Phys, Chem. 94, No. 9 (1990)941.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
R.A. SHELDON
Laboratoryfor Organic Chemistry and Catalysis,
Delft University of Technology, Julianalaan 136
2628 BL Delft, The Netherlands
ABSTRACT
The free radical chain mechanism of liquid phase oxidations with dioxygen is reviewed.
The intricate mechanism of the oxidation of substituted toluenes to the corresponding
carboxylic acids with the Co/Mn/Br catalyst (Amoco/MC system) is discussed in detail.
1. Introduction
Many liquid phase oxidations of organic substrates with dioxygen are known as
autoxidations because they are subject to autocatalysis by the initial products of
oxidation, alkyl hydroperoxides . The pioneering work of Backstrom demonstrated
that these processes are radical chain reactions. Criegee made an important contribution
in 1939 when he showed that the primary product of the oxidation of cyclohexene by
dioxygen is the allylic hydroperoxide (reaction 1). Subsequently it was recognized that
the controlled autoxidation of hydrocarbons can be a useful method for preparing
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
oxygenated derivatives. This led to the development, by Hock and Lang in 1944 of
what is still the most important industrial process for the production of phenol and
acetone, by acid-catalyzed Criegee rearrangement of cumene hydroperoxide derived
from the autoxidation of cumene (reaction 2).
o -—6 (D
OH
+
-*" O < C H 3>* C O (2)
151
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
152
Liquid phase autoxidations proceed via a free radical chain mechanism described
by the general scheme shown in reactions 3-8.
Initiation
R,
ln 2 _*, 2 In* (3)
Propagation
R# + 0 2 ► ROa# (5)
K
R0 2 « + RH — ► ROaH + R# (6)
Termination
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
R#+02 — _*, R02R (7)
2k t
R02. + R02. ► RO4R
Alkylperoxy radicals play vital roles in both the propagation and termination
steps. The primary products are alkyl hydroperoxides and in some cases, e.g. cumene
hydroperoxide, they may be isolated in high yields. At oxygen pressures used in practice
(1 bar or higher) chain termination proceeds exclusively via the mutual destruction of
two alkylperoxy radicals (reaction 8) The predicted rate equation is given by equation 9.
A* A*
- *,PWI A w
2k- <*>
Substrate y^k^xlO3
(M-1/z s'1/2)
2,3-Dimethyl-2-butene 3.2
Cyclohexene 2.3
1-Octene 0.06
Cumene 1.5
Ethylbenzene 0.21
Toluene 0.01
p-Xylene 0.05
Benzaldehyde 290
1 Benzyl alcohol 0.85 1
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Table 2. Initiatorsforautoxidatkms*.
I Initiator E Temp. °C
a ,
(kcal.mor1) for ti/5 = 1 hr 1
I HO-OH 48 "" 1
| t-BuO-OH 42 — I
I t-BuO-OBu-t 37 150 1
1 t-BuO-0 2 CPh 34 125 1
I PhC0 2 -0 2 CPh 30 94
I CH3CO2-O2CCH3 30 85
I i-PrC(CN)N=N(CN)Pr-i 30 85
I t-BuON=NOBu-t 28 60
I t-BuO-OoCCOo-OBu-t 25.5 40 |
Reaction of the alkyl radical (R) with dioxygen is, in most cases, diffusion
controlled (i.e. k 2 > 10 9 M"1 s"1) and the rate-controlling step in autoxidations is
hydrogen transfer from the substrate to the alkylperoxy radical (reaction 6). The rate
constants (k_) for this reaction can be roughly correlated with its exothermicity.
Oxidations are favorable when the bond that is formed (ROO-H) is at least as strong as
that which is broken (R-H). The ROO-H bond is about 90 kcal mol"1 which is larger
than that for benzylic, allylic and aldehydic C-H bonds (see Table 3).
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Propagation rate constants are also dependent on the nature of the attacking
alkylperoxy radical. Table 4 compares the propagation rates for the reaction of various
substrates with its own alkylperoxy radical with that of the reaction of tert-butylperoxy
radicals (cross propagation) with the same substrates.
Table 4. Rate constants per labile hydrogen for reactions of substrates with their own peroxy radical ( k j and
with tert-butylperoxy (k^) at 30 °Ca.
Substrate
V V
1 1
flvr s- ) (M"1 s"1)
1-Octene 0.5 0.084 6.0
Cyclohexene 1.5 0.80 1.9
Toluene 0.08 0.012 6.7
Ethylbenzene 0.65 0.10 6.5
I Cumene 0.18 0.22 0.9
Tetralin 1.6 0.5 3.2 j
1 Benzyl alcohol 2.4 0.065 37.0 1
1 Benzyl acetate 2.3 0.0075 307
1 Benzyl bromide 0.6 0.006 100 1
1 Benzaldehyde 33,000 0.85 40,000 1
At reasonable oxygen pressures chain termination occurs exclusively via the self-
reaction of two alkylperoxy radicals (reaction 8). The overall rate of autoxidation is
governed by both the rate of propagation (kp) and termination (kj) as given by eq. 9.
Examination of the rate constants in Table 5 reveals that the lower rates of autoxidation
at primary and secondary C-H bonds compared to their tertiary counterparts are due not
only to the lower reactivity of the C-H bonds (see Table 3) in the former but also to the
|R02- I—^r~\CMT1'*'1)
H0 2 - 8 x 105
RCH202- 107
R2CH02- 106
Kco/ io 3 1
a. Data takenfromref. 5.
R2C ) O
-► R2CO + H2CHOH + Oa (10)
H * O
\
CHR„
R 0 2 + A r O H - R0 2 H + ArO- (11)
3. Olefin Autoxidation
In the autoxidation of olefins chain propagation can occur via the usual
abstraction mechanism (reaction 15) or via the addition of the alkylperoxy radical to the
double bond (reaction 16). Addition can be followed by unimolecular decomposition of
the p-alkylperoxyalkyl radical (reaction 17) affording epoxide and an alkoxy radical or
by its reaction with oxygen to give polyperoxides (reaction 18).
abstraction I
RO.H +—C — C = C C (15) _„/
1
' /
RO.» + —C — C = C (
H
RO — C — C • (16)
addition
O
«» RO« + \/ C/ —\ C ^/ (17)
RO,—C — C •
I I
•>• R0 2 —C—C—O a « , etc. (18)
aS=S=SaSaSSa=S=SS&aBBS= sassssa^ssaBssan
I Olefin Abstraction Addition 1
1 (%) (%)
1 Propylene 50 50
| 1-Hexene 68 32
1 Cyclohexene 95
1 Cyclooctene 30 70
Isobutene 17 83
2-Butene 38 62
1 1-Butene 73 27
1 a-Methylstyrene 0 100 1
asssssssssaassssssBsss
4. Aldehyde Autoxidation
Initiation •
5. Cooxidations
example, in the cooxidation of aldehydes and olefins, the acylperoxy radical and the
peroxy acid are utilized for the epoxidation of the olefin:
I I v / \ /
RCOa — C — C # -+> RCO,# + ) C — C ( (23)
o
1
/ \
RCQ3H+)c = c ( ► RC02H+)c—c( (24)
The cooxidation affords much higher yields of epoxides than those obtained in
the autoxidation of the olefin alone, since acylperoxy radicals have a much more
favorable addition/abstraction ratio compared to alkylperoxy radicals.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
phase oxidations, in contrast, generally afford carbonyl compounds and or
dehydrogenation products. This difference is not due to changes in the fundamental
mechanism, but rather to the availability of different pathways for further reaction of the
alkylperoxy radicals under gas-phase conditions, i.e. high temperature and low substrate
concentration. An illustrative example is the autoxidation of isobutane which has been
extensively studied ' and is of considerable importance. The liquid phase
autoxidation of isobutane at 125 °C affords a mixture of approximately 75% tert-butyl
hydroperoxide (TBHP), 21% tert-butanol (TBA), 2% acetone and 1% isobutyl
derivatives12. The kinetic chain lengths are rather long and TBHP and TBA are formed
via the classical autoxidation mechanism (reactions 25-29).
In the gas phase at 155 °C and at relatively low pressures, in contrast, there is
insufficient isobutane to sustain chain propagation (reaction 26). Consequently, the self-
reaction of tert-butylperoxy radicals produces tert-butoxy radicals (reaction 28). Since
there is also insufficient isobutane to sustain reaction 29 most of the tert-butoxy radicals
undergo unimolecular fragmentation to acetone and methyl radicals (reaction 30). The
latter react with oxygen to form methylperoxy radicals, which are more reactive in chain
termination than tert-butylperoxy radicals. As a result, kinetic chain lengths are short and
the principal products are acetone, methanol and TBA:
Me + 0 2 - Me0 2 - (31)
Interestingly, there is a smooth transition from the liquid to gas phase reaction.
Thus, an increase in the isobutane pressure in the gas phase oxidation at 100 °C leads to
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
an increase in the yield of TBHP at the expense of TBA, acetone and methanol. At low
rates of initiation and 13 bar isobutane a 92% yield of TBHP can be obtained. Similarly,
dilution of the liquid phase reaction at 100 °C with the inert solvent, carbon
tetrachloride, leads to a simulation of gas phase conditions and the yields of TBA and
acetone increase at the expense of TBHP.
At even higher temperatures (300 °C) in the gas phase, the major primary
product becomes isobutene, formed via reaction 34. Although reaction 25 is faster than
reaction 34 at all temperatures, it is reversible at high temperatures, whereas reaction 34
is not.
The scheme outlined above for isobutane applies to all autoxidations. In the
liquid phase at relatively low temperatures, the kinetic chain lengths are long and the
major products are hydroperoxides. In the gas phase, kinetic chain lengths are short and
the major products are carbonyl compounds resulting from thermal fragmentation of
intermediate alkylperoxy radicals.
These differences between gas and liquid phase conditions also have implications
for oxidations in the presence of metal catalysts. In the liquid phase it is difficult to
compete with the ubiquitous free radical chain autoxidation. In the gas phase, in
contrast, kinetic chain lenghts are short and intermediate radicals can react via alternative
pathways with the metal catalyst. This partly explains why Mars van Krevelen type
mechanisms (see Chapter 1) are favored in the gas phase.
Variable valence metals, such as cobalt, manganese, copper and iron, catalyze
liquid phase autoxidations by promoting the homolytic decomposition of alkyl
hydroperoxides into chain initiating radicals via reactions 35 and 36.
Net reaction:
Co^Co 111
2R02H - R O + R0 2 - + H 2 0 (37)
Since alkylperoxy radicals are strong oxidants they are also capable of oxidizing
the reduced form of the catalyst (reaction 38). In this case the metal ion is acting as an
inhibitor. Hence, transition metal ions, especially in media of low polarity such as neat
hydrocarbons, often behave as autoxidation catalysts at low concentrations and inhibitors
at high concentrations. This phenomenon is referred to as catalyst-inhibitor
conversion 13 . It manifests itself in the long induction periods often observed in
metal-catalyzed autoxidation in nonpolar media.
Co n + R0 2 - - R02Com (38)
Low [Co]
KCL = kp[RH]/[2ktRi]I/a (39)
High [Co]
KCL = kpIRHl/lqtCo11] (40)
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Catalyst inhibitor conversion is observed when the chain length becomes less
than unity, i.e. [Co11] > kpfRHJ/lq. In practice an abrupt transition from catalysis to
inhibition is generally observed for hydrocarbon autoxidations. In the cobalt-catalyzed
autoxidation of neat tetralin at 65 °C, for example, an abrupt transition from rapid
reaction to inhibition was observed at a catalyst concentration of approximately 0.1 M.
Catalyst-inhibitor conversion can be circumvented by adding an alkyl
hydroperoxide such that [R02H] > [Co] or by carrying out the reaction in polar media,
such as acetic acid (see later). The use of polar solvents also prevents catalyst
deactivation by precipitation as insoluble carboxylate saltsfromnonpolar media.
CO.H
(41)
[Co(OAc) 2 ]
t M
3 C02H
I
6
I
°2 (42)
►
Co(OAc) 2 /Mn(OAc) . $
1 /NaBr or NH 4 Br I
CH 3 C02H
HOAc solvent
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
The exact conditions of the two processes depicted in eqns 41 and 42 are quite
different. In order to explain the need for such different conditions, in particular the
necessity of the combination Co/Mn/Br* in HOAc in the Amoco/MC process (reaction
42) we need to delve more deeply into the mechanisms of these fascinating processes.
If one compares the oxidizabilities of toluene, ethylbenzene and cumene one sees
that the much lower reactivity of tolufcne is largely due to the much higher rate of
termination of the primary benzyloxy radicals (Table 7). The much higher oxidizability of
benzaldehyde, on the other hand, is largely due to a much higher propagation rate.
1 Substrate 2ktxl0'6
(MTV 1 ) (M-vy
(ivr,/2s"1/2)
PhCH3 0.24 300 0.014(1)
PhCH2CH3 1.3 40 0.21 (15)
PhCH(CH3)2 0.18 0.015 1.5(107)
1 PhCHO a 12,000 1760 290 (21,000) |
l^sssssssssssssssssss;
a. Measured at 0 °C.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
ArCH 33 + uo 111
T Co ► [ArCHJ Coo 1 1
LAfUM3j t• + u (43)
ArCH 2 0 2 # + Co 11 ► [ArCH— 0 - ^ O C o m ]
In the third type, typified by the Amoco/MC terephthalic acid process (reaction
42), lower concentrations of cobalt(II) are employed, in acetic acid solvent, and bromide
ion and manganese are added as cocatalysts. The second and third type processes are
generally employed with substrates that are more difficult to oxidize.
The metal-catalyzed autoxidation of substituted toluenes can be conveniently
divided into two stages. In the initial stage the toluene is oxidized to the corresponding
benzaldehyde as discussed above. Since the benzaldehyde is much more reactive than the
substrate the reaction soon enters a second stage in which the aldehyde undergoes rapid
autoxidation to give the corresponding aromatic percarboxylic acids as key
intermediates. Interestingly, Jones showed that the latter, in contrast to alkyl
hydroperoxides, oxidize Mn and Co via a heterolytic mechanism in acetic acid
solution, affording the u-oxo dimer of Mn111 or Co , respectively (reaction 47).
In the oxidation of p-xylene the first methyl group undergoes rapid autoxidation to
afford p-toluic acid (reaction 48). The second methyl group is, however, deactivated
by the electron-withdrawing carboxyl group, and further oxidation to terephthalic acid
(reaction 49) is much slower (toluene is 26 times as reactive as p-toluic acid). It is not
surprising, therefore, that the autoxidation of p-xylene to terephthalic acid proved to be a
difficult proposition17.
(48)
(49)
C02H
Two types of processes are used for the industrial oxidation of p-xylene to
terephthalic acid (see Chapter 1). In the Eastman Kodak/Toray process a cosubstrate
(acetaldehyde or methylethyl ketone) is used in combination with high concentrations of
cobalt(III) acetate in acetic acid solvent. The mechanism involves direct reaction of
cobalt(m) with the p-xylene substrate as outlined in reactions 43-47. The Amoco/MC
process, on the other hand, employs low concentrations of a catalyst cocktail comprising
cobalt(II), manganese(II) and bromide ion. As noted by Partenheimer15 this technology
has been successfully applied to the oxidation of about 270 different substrates. Some
examples of aromatic di- and tricarboxylic acids that are commercially produced using
the Amoco/MC process are shown in Figure 1.
Because of its industrial importance the mechanism of the Amoco/MC process
has been extensively studied, notably by Jones16'18 and by Partenheimer15'1*21. A question
which inevitably arises in this context is: why the combination Co/Mn/Br in acetic acid?
In order to answer this question we need to examine the role of the various catalyst
components.
Partenheimer showed15 that when a solution of cobalt(D) acetate in acetic acid at
113 DC was treated with dioxygen about 1% of the cobalt was converted to the trivalent
state. In the presence of a substituted toluene two reactions are possible: formation of a
benzyl radical via one-electron oxidation of the substrate or decarboxylation of the
acetate ligand (reactions 50 and 51).
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
CO2H
CO2H
w —* 10
H
3C\ ^ \ /CH3 HO2C. ^ / \ /CO2H
CH3 CO2H
CO2H
HO2C
Co I I ! OAc
CH3.+ C o n + C 0 2 (51)
In the presence of bromide ion the slow one-electron transfer oxidation of the
ArCH3 substrate is replaced by the rapid one-electron oxidation of bromide ion by
cobalt(III), affording a bromine atom (reaction 52). The latter, ©r rather its adduct with
bromide ion, Br2~, acts as the chain transfer agent in the reaction with the substrate
(reaction 54).
diffusion
Bv + Br- ► Br 2 - (53)
controlled
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Br2- + ArCH3 ► ArCH 2 * + Br" + HBr (54)
In order to provide an insight into the nature of the catalytic species Jones
investigated the reaction of cobalt(II) acetate with m-chloroperbenzoic acid in 95%
aqueous acetic acid at 0 °C The composition of this mixture corresponds reasonably
well with that which is formed during ArCH3 autoxidation. He found that Co(OAc)2
was instantaneously oxidized to a n-oxocobalt(III) dimer. The latter was a very active
catalyst and was denoted as C o n i a (see Figure 2). Within a few minutes at 25 °C this
apple green complex reacted with a molecule of water to form an olive green,
hydroxyl-bridged dimer. The latter was much less reactive and was denoted as Co s .
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
On standing for several days at 25 °C the hydroxyl-bridged dimer reacted with Co11 to
form a n-oxo mixed trimer of C o m and Co" denoted as C o m c , which had previously
been identified by Ziolkowski and coworkers .
2 Co"(OAc)2
pink
Instantaneous Y ArCO.H
Arou,M
(retarded by H 2 0)
f < — _ » - ArCO f H
Few minutes, 25 °C
(retarded by H 2 0)
H
Co
" C s^c°w Co"
H
olive green
Several days, 25 °C
3h, 60 °C
| jo-
bo" " Co"
Co"
Co"
reactive than Co fflc (Table 8). Furthermore, the rate of conversion of C o m a to Co1118
was much higher than the rate of reaction of Co111* with ArCH3. In other words, in the
absence of Br" or Mn n the cobalt species that reacts with ArCH3 cannot be Co .
Coffla Co018 Co 1 * 5 1
Br" 30,000 4 1
Mn11 6,000 4 1
IHQQQ 700 ; 5 1 1
a. Data taken from ref. 18.
Table 9 shows the relative rates of reaction of ArC0 3 H with the various catalyst
components in 90% aqueous acetic acid at 25 °C. Thermal homolytic decomposition is
negligible under these conditions. The relative rates of reaction of ArC0 3 H with Co11,
Br" and Mn11 are 3900:4.7: l 2 1 . This is not what one would predict from the decreasing
order of reduction potentials: Br" > Mn D > Co n , i.e. these reactions are kinetically rather
than thermodynamically controlled. In practice this means that in a mixture containing
roughly equal amounts of Co11, Mn" and Br- together with ArC0 3 H more than 99% of
the latter will preferentially react with the Co .
SSSSSSSSSSSSI^SBSSSSSSSSBSSSSSaSESSSSSSSSiHaaHHH
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
1 Reaction Rel. rate 1
ArC03H + Co n - Co m a 3900
AT
I ArCO^H -» ArCOy+HO- IP"4 1
a. Data taken from ref. 21.
Once the active C o m a catalyst has been formed by peracid oxidation of Co its
fate is determined by the relative rates of its reaction with the other species present in the
reaction mixture, i.e. Mn , Br" and substrate compared to its rearrangement to Co .
As can be seen from the relevant data (Table 10) by far the most favorable reaction is
oxidation of Mn11 to Mn111 which is 940 times faster than conversion of C o m a to C o n i s
under these conditions.
Co ffla + Br_ - Br 23 84 1
Co ffla - Co IIIs 25 1
In other words, in the mixture containing C o m a , Mn11, Br" and p-xylene more
than 90% of the Co IIIa reacts with Mn11 to afford M n m and there is negligible reaction
of C o n i a with the substrate.
Based on this detailed kinetic analysis of the individual steps we are now able to
provide an interpretation of the synergistic effect of the Co/Mn/Br" catalyst cocktail.
With cobalt alone in acetic acid the reaction of cobalt(III) with p-toluic acid is much too
slow. Bromide alone is rapidly oxidized by ArCC^H to afford hypobromite, by a
heterolytic mechanism. In the presence of cobalt, bromide ion is oxidized to chain- --`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Reaction tuftnin)
CoraOAc - Co n + C H 3 + C 0 2 14
Co m + Mnn - Co n + Mn m <0.2
a. Data takenfromref 21
The use of the Mn/Co/Br" system allows for higher reaction temperatures and
lower catalyst concentrations than the bromide-free processes. The only disadvantage is
the corrosive nature of the bromide-containing system which necessitates the use of
expensive, titanium-lined reactors The complete mechanism of the autoxidation of
substituted toluenes in the presence of the Co/Mn/Br" catalyst, which must surely be
considered a work of art, is depicted in Figure 3.
ArCO.H ArCO.H
Co(OAc),
i
Co(OAc)/
slow
->Co(OAc),'
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Mn(OAc), Mn(OAc)a
Br" Bf
}
ArCH •
T
ArCH,
The Amoco/MC catalyst system has also been applied to other types of
alkylaromatic oxidations, e.g. the oxidation of m-phenoxyethylbenzene to the
pharmaceutical intermediate, m-phenoxyacetophenone (reaction 57)2 .
CH
*CH3 0,,HOAc ^ ^ C 0 C H
3
(57)
Con/Mnn/NaBr
1 0 0 : 8 0 : 1
OPh OPh
84% yield
Indeed, one cannot help but wonder if the full potential of this fascinating and
elegant technology has yet been realized.
Catalytic autoxidations of hydrocarbons in the liquid phase have been around for
several decades. Nevertheless, they still constitute a very useful technology for the
synthesis of a variety of products. Their utility is likely to be further broadened in the
future as they are applied to the manufacture of fine chemicals as replacements for
classical oxidations with stoichiometric inorganic reagents. Finally, it cannot be
emphasized enough that free radical chain autoxidation is always occurring as a
background reaction in any system comprising metal catalysts, dioxygen and
hydrocarbon substrates in the liquid phase. Consequently, it is difficult to design
conditions in the liquid phase in which alternative pathways, involving dioxygen
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
References
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Oyama and J. W. Hightower, ACS Symp. Ser. 523 (1993) p. 81.
21. W. Partenheimer and R. K. Gipe, ACS Symp. Div. Petrol. Chem. Preprints, ACS
Meeting, Washington, D C , Aug. 1992, pp. 1098-1104.
22. T. Szymanska-Buzar and J. J Ziokowski, J. Mol. Caki. 5 (1979) 341.
23. Jap. Pat. 5867640 (1983) to Nippon Kayaku; CA 99(198*) 382046.
R.A. SHELDON
Laboratoryfor Organic Chemistry and Catalysis,
Delft University of Technology, Julianalaan 136
2628 BL Delft, The Netherlands
ABSTRACT
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
The various types of heterogeneous catalysts for liquid phase oxidations are reviewed. In
recent years there is a marked trend towards the use of molecular sieve catalysts,
containing redox metal ions incorporated in the framework or metal complexes
encapsulated in the micropores, instead of the more traditional metals, metal ions or metal
oxides on amorphous supports. These redox molecular sieves and ship-in-the-bottle
complexes have many features in common with redox enzymes.
1. Introduction
STOICHIOMETRIC CATALYTIC
HOMOGENEOUS HETEROGENEOUS
175
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
176
In this context it is worth noting, as was already discussed in Chapter 8., that the
chance of observing selective, non-classical oxidation of organic substrates with
dioxygen is, generally speaking, greater in the gas phase than in the liquid phase. In the
gas phase substrate concentrations are much lower and radical chain oxidation is less
favorable. Consequently, reactive intermediates, e.g. alkyl radicals, may react with metal
oxidants at the catalyst surface leading tb a Mars-van Krevelen type mechanism (Figure
2). ;
Liquid phase
initiation
RH — ► R.
R« + o2 ► R0 2 #
RQ2# + RH —► R0 2 H + R#
S +M=0 - ► SO + M
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
r
2 M + 02 -i ► 2 M=0
v
III
PM I V MIVP
PM = 0 + PM
P = porphyrinato
v
M = Fe , Mn
(1)
IV
Ti = 0 (2)
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
/ ^
IV
/sio a
OH
+
s3 0^ — 0^
Active catalyst:
\ \
^Si — O H . - S i - —o
1. TiCI 4 / \ OH
/ a^
0 Tl
o 2. HaO
^Si—OH
V —0
/ \
OH
\
^.Sl O
A
•»•
/ V
-H20 O fl = 0
\ /
Si-0
Heterogeneous catalysts for liquid phase oxidations can be divided into four basic
types: (a) supported metals (e.g. Pt/C)t (b) supported metal ions and complexes, e.g.
metal ions on ion exchange resins and metal ion exchanged zeolites, (c) supported
oxometal (oxidic) catalysts, e.g. Ti^/SiC^ and redox molecular sieves (see later) and (d)
metal complexes encapsulated in zeolites and related molecular sieves, so-called ship-in-
the-bottle complexes.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Pt
RCH2OH -► RCHO + H0 (3)
Pt
2 H2 + 0 2 -► 2 H20 (4)
The exact role of the base in these reactions is still not clear. It is generally
thought to be necessary to remove strongly absorbed carboxylic acids, the ultimate
products of these reactions, from the catalyst surface.
Platinum and palladium-based catalysts have been widely used > in the oxidative
dehydrogenation of vicinal diols, hydroxy acids and carbohydrates. Some examples are
shown in reactions 5-7.
0 2 ; NaOH
CH3CH(OH)CH2OH CH 3 COC0 2 H (5)
Pt-Pb/C
or Pt-Bi/C
0 2 ; NaOH
CH,CH(OH)CO.H CH3COC02H (6)
Pt-Pb/C
OH
0 2 ; NaOH
OH -^" HO
HO HO Pd-Bi/C
OH
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
One reaction that has been the focus of much attention is the catalytic oxidation
of D-glucose to D-gluconate. Palladium catalysts exhibit high selectivity to gluconate but
the catalyst is rapidly poisoned by dioxygen. However, the simple deposition of bismuth
onto palladium-on-charcoal affords a catalyst with excellent activity, stability and
selectivity (> 99%)7.
Notwithstanding the enormous effort that has been devoted to noble
metal-catalyzed oxidative dehydrogenations few of these processes have been reduced to
industrial practice, largely due to the above mentioned problem of catalyst poisoning.
OH
Bu* JL Bu*
o
0 2 ; H20
•=— r-»
(p)-CoP*Ts
Transition metal ions supported on ion exchange resins have also been used to
catalyze a variety of oxygen transfer reactions with hydrogen peroxide and alkyl
hydroperoxides ' . From a practical viewpoint the crucial question is whether these
catalysts retain the metal over a long period, i.e. whether or not they are subject to
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
leaching.
Transition metal ions can also be immobilized by ion exchange with zeolites ' .
However, the same reservations apply to metal ion-exchanged zeolites as to the metal
ions supported on ion-exchange resips discussed above, i.e. their practical utility is
crucially dependent on their long term stability towards leaching. In this context it is
worth noting that redox molecular sieve catalysts (see later) are probably less prone to
leaching than traditional supported metal catalysts.
NaOH
C0 2 Na
Pb/Ru or Bi/Ru * L ^ CCO.Na
02l (9)
Q
25-95 °C ; 2 bar
81-87% yield
pH > 13
Although the experimental conditions closely resemble those of the noble metal-
catalyzed oxidative dehydrogenations (see section 4) the active oxidant in the ruthenium
pyrochlore system is clearly a high-valent oxoruthenium species, i.e. it is a
heterogeneous oxidic type catalyst.
7. Redox Molecular Sieves - Unique Solid Catalysts for Liquid Phase Oxidations
to inorganic surfaces, e.g. via Si-O-M bonds, often leads to catalysts that are susceptible
to solvolysis by polar molecules, e.g. water, diols and carboxylic acids, present in
oxidation reaction mixtures. This leads inevitably to leaching of the catalyst from the
surface to form homogeneous systems. Moreover, such catalysts are often deactivated
by strong coordination of polar molecules, such as water, thus preventing diffusion of
hydrophobic hydrocarbon substrates to the active site.
One approach to isolating redox (oxo)metal species in stable inorganic matrices
is via incorporation in the framework of a molecular sieve (zeolite, silicalite,
aluminophosphate, etc.) We coned the generic name redox molecular sieves 1,14 ' 5 to
describe such materials, which have many features in common with redox enzymes and
several advantages compared to conventional supported catalysts. Unlike amorphous
materials they possess a regular microenvironment having highly homogeneous internal
structures with well-defined cavities and channels. They also exhibit enhanced stability
towards leaching. A possible explanation is that confinement of the active catalytic
species on the curved internal surface of a molecular sieve renders it less accessible to
solvolysis by polar molecules. Furthermore, confinement of the active site in channels
and cavities of molecular dimensions imparts redox molecular sieves with the additional
feature, in common with enzymes, of shape selectivity. Another feature that redox
molecular sieves share with enzymes is the possibility of creating a hydrophobic or
hydrophilic environment around the active site, by a suitable choice of molecular sieve.
Silicalite, for example, contains hydrophobic micropores while aluminophosphates are
hydrophilic. This can lead to more pronounced solvent effects than are observed with
conventional supported catalysts. The molecular sieve can be considered as a second
solvent that extracts the substrate out of the bulk solvent. The efficiency of this process
will be governed by the relative hydrophobidty/hydrophilicity of the substrate, product,
solvent and the micropores of the mdecular sieve. This offers the possibility of 'fine
tuning* the size and hydrophobic/hydrophilic character of the redox cavity to create
'tailor made1 oxidation catalysts that may be truly regarded as 'mineral enzymes'.
Various types of molecular sieves are, in principle, amenable to framework
substitution by transition metal ions. The basic building blocks are tetrahedral Si0 4 ,
A10 4 and PO4 units (Figure 4). Silicalites are examples of all-silica molecular sieves,
combination of Si0 4 with A10 4 leads to zeolites, A10 4 with P 0 4 to aluminophosphates
(APOs), and the combination of all three building blocks to sOiaialuminophosphates
(SAPOs).
Silicalites and aluminophosphates are electroneutral materials devoid of cation
exchange properties. In zeolites, on the other hand, the extra charge on silicon (4*)
compared to aluminium ,(3*) has to be balanced with a proton (or a cation), which
confers ion exchange properties on these materials. Moreover, the proton forms of
zeolites are very strong acids.
Molecular sieves are synthesized16 by allowing the appropriate sol gel to
crystallize, usually at temperatures of around 175 °C, in the presence of a template
(so-called structure directing agent) which is generally an amine or a
tetraalkylammonium salt. Subsequent calcination of the crystalline material at about 500
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
SAPOS
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
REDOX I REDOX I
REDOX APOS
ZEOLITES J SILICALITES
- VAPO Ti - Al - B TS - 1
- CrAPO Ti - ZSM-5 , etc. VS - 1
- CoAPO, etc. CrS - 1, etc.
Selected examples of molecular sieves, with various pore shapes and sizes, are
collected in Table 1. Examples of transition metal ions that have been incorporated into
the various structures are also shown. Molecular sieves can have channels that are
disposed in a three dimensional array or that are uni- or bidirectional. ZSM-5, for
example, has a three dimensional array of intersecting channels while AlP0 4 -5 and
A1PO4-H have unidirectional channels that do not intersect. The dimensionality of
molecular sieves can be of importance with regard to the accessibility, to the substrate,
of the active catalytic site.
snaaBEBBKS
The first example of a redox molecular sieve was the titanium(IV)silicalite (TS-1)
catalyst developed by Enichem workers17"20. This truly remarkable catalyst mediates a
variety of synthetically useful oxidations with 30% aqueous hydrogen peroxide (Figure
5).
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
R2CO
OH ^ OH
+ OH
oc 0
H202
catalyst
^ ^ OH H O ^ ^
■ ' SS^SSSSSSSSSSSSSSSSS
[Pt] [Pd]
Existing route
Ammoximation
cf
Figure 6. Two routes to cyclohexanone oxime.
NH, + H,0, +
a' - c"~
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
186
The TS-1 catalyst exhibits some quite remarkable activities and selectivities. For
example, relatively unreactive olefins, such as ethylene, propylene and allyl chloride, are
epoxidized efficiently^1 under very mild conditions in methanol as solvent (Table 3). As
a result of the shape selective properties of the catalyst larger, more reactive olefins such
as cyclohexene, are epoxidized only slowly, if at all
From a mechanistic viewpoint it is noteworthy that the TS-1 catalyst contains the
same chemical elements, in roughly the same proportions (2% Ti), as the Shell
Ti rv /Si0 2 catalyst discussed earlier. Yet these two catalysts exhibit strikingly different
and complementary catalytic behavidr. TrVSi0 2 catalyzes epoxidations with alkyl
hydroperoxides, such as TBHP, but is uneffective with H 2 0 2 while TS-1 is effective
with aqueous H 2 0 2 but not with TBHP. The novel activities of both catalysts are
assumed to accruefromthe site-isolation of monomeric Tr* centres, which are probably
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
packing of molecules in crystals can lead to reaction rates that are higher than the
corresponding ones in solution.
The mechanism of TS-1 catalyzed epoxidation of olefins most likely involves
oxygen transfer from a coordinated tert-butylperoxy group to the double bond of the
olefin (Figure 7). This could be facilitated by a silanoxy ligand (mechanism a) or by a
coordinated molecule of methanol solvent, or silanol ligand (mechanism h).
n
SiO T V SiO
** Ti = 0 + SiOH
SiO
o
SiO °Si
\ I
Ti
/ \
SiO OH
SiO ?Si O T H
SiO OSi
X
T! /y ■•« "^
/ \
SiO I 0-L_0 SiO OR
r OH O
+ L\
Molecular graphics studies24 show that although the cavity of TS-1 can
accommodate a molecule of TBHP severe steric resUictions are imposed on the
approach of an olefin to the coordinated tert-butylperoxo group. Thus, TS-1 catalyzes
the formation of TBHP from tert-butanol and H 2 02 but it does not catalyze
epoxidations with TBHP. Similarly, the lack of reactivity observed with cyclohexene is
presumably a result of severe steric constraints imposed on the transition state for
oxygen transfer.
In short, TS-1 is an excellent catalyst for oxidations of relatively small substrates
with aq. H 2 0 2 . It is not effective, however, with larger substrates and/or typical alkyl
hydroperoxide oxidants, such as TBHP. Consequently, there is considerable interest in
the incorporation of titanium into larger pore molecular sieves. Corma and coworkers
reported the incorporation of titanium into zeolite beta (7.6 x 6.4 A). They showed that
the resulting Ti-Al-P, in contrast to TS-1, catalyzed the oxidation of both 1-hexene and
cyclohexene with aq. H 2 0 2 at roughly the same rate. However, the major product was
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
not the epoxide but the glycol monomethyl ether, presumably formed byringopening of
the epoxide with the methanol solvent (see Table 4). The latter reaction is probably
catalyzed by the Brensted acid sites associated with the zeoUte framework.
Table 4. Ti-Al-p catalyzed oxidation of olefins with 35% aq. H 2 0 2 in methanol at 25 °Ca.
H,Oa / ° \
R1CH=CHR2 ► R 1 CH—CHR*
catalyst
(I)
MeOH
► R 1 CH—CHR 2
catalyst I I <n>
OH OMe
BBSBSBS 1 1 ==aaas=s=s=SBB
I Catalyst Product sel. 1
Olefin H 2 0 2 conv.
(%) (%) I
W W
1-Hexene TS-1 98 96 4
Ti-Al-p 80 12 80
When the Brensted acid sites of the Ti-Al-P were neutralized by treatment with
an aqueous solution of an alkali metal acetate prior to use this led2" to a dramatic
increase in the epoxide selectivity in the oxidation of 1-octene with aq. H 2 0 2 (Table 5).
Table 5. Effect of alkali metal exchange on selectivity of Ti-Al-p catalyzed epoxidation of l-octenea.
O
35% H a 0 2 / \
C6H13CH = C H S C,H13CH — C H 2
catalyst
MeOH solvent
40°/100min
MeOH
C6H13CH(OH)CH2OMe
catalyst
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
190
iSSSSSSSSSESSSXSSSSa&SSSSS — — * ■ * - — : '
TBHP
Ti-MCM-41 (10)
CH.CL ; 40°C ;
2 a
' Conv. 30%
Sel. 90%
OH
Bu Bu' Bu*
30% H 2 O t
(11)
catalyst
J
30% H 2 0 2
+* \( (12)
catalyst ; *<
acetone
Following the seminal studies of the Enichem group on the TS-1 catalyst, several
groups have reported the incorporation of other redox metals into the silicalite
framework, e.g. zirconium , tin and chromium32. ZrS-1 catalyzed thei hydroxylation
1
of benzene and phenol with aq. H 2 0 2 but activities and selectivities were significantly
lower than with TS-1 . SnS-1, on the other hand, reportedly^ catalyzes the
hydroxylation of phenol with aq. H 2 0 2 with activities only marginally lower than that of
TS-1.
CrS-1 exhibits different types of catalytic activity to the above mentioned
catalysts. For example, it catalyzes the selective oxidative cleavage of oleflns with aq.
H2O2 in acetonitrile . Methyl acrylate and methyl methacrylate were converted to the
methyl esters of glyoxylic acid (reaction 13) and pyruvic acid (reaction 14), respectively.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
H20a;[CrS-1l
H2C=CC02CH3 — ► CHOCO a CH 3 (13)
CH3CN ; 40oC C o n y - 9g o /o
Sel. 90%
Sel. 80%
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
H
HO OP I ft
H
2° I -OP D O °
O \ || oOP
w x
2 *. " v i ^
Cr ^ ► *Cr ^
500 °C y^l >w
p= 6
O
// <•) (b)
As shown in Table 7 both CrAPO-5 and CrS-1 are active catalysts for the
decomposition of cyclohexyl hydroperoxide (CHHP) in cyclohexane at 70 °C, the
highest selectivity for cyclohexanone (86%) being observed with CrAPO-5. Other
MeAPOs and silicalites gave both lower activities and selectivities.
Table 7. Decomposition of cyclohexyl hydroperoxide (CHHP) over various redox molecular sieves*.
TS-1
1 None
0
0
—
- 1 :1
a. CHHP (2.9 mmol) in cyclohexane (12 ml) stirred with the catalyst (0.029 mmol
metal) for 5 h at 70 ° C (data takenfromref. 35).
0 ,5>
\Jy< ^ - V ^
An intermolecular mechanism via further oxidation of initially produced cyclohexanol
with CHHP can only be a minor pathway since reaction of cyclohexanol with
cyclohexene-3-hydroperoxide or TBHP afforded only about 6% cyclohexanone under
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
,1 „ ,
TBHP
PhCH(R)OH ^ PhCOR (16)
CrAPO-5
PhCI ; 85°C
^ Sel. (%)
CH, 96
C2H6 100
OH
rr
u
TBHP
rr°
u (17)
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
CrAPO-5
PhCI;85°C Sel. 85%
Conv. 72%
^ ^OH
TBHP JY
U
T
CrAPO-5
PhCI ; 85°C VT (18)
y^\
A
Sel. 94%
Conv. 62%
The CrAPO-5 catalyst was stable and recyclable, the reused catalyst being even
more active than the fresh one in the oxidation of a-methylbenzyl alcohol. Moreover,
treatment of the catalyst with sodium acetate prior to use led to a catalyst that was both
more active and selective (see Table 9) .
As in the case of alkyl hydroperoxide decomposition (see earlier) evidence for
the reaction taking place inside the micropores was obtained by using the bulky
triphenylmethyl hydroperoxide as the oxidant in the oxidation of a-methylbenzyl
alcohol . Hardly any reaction was observed with CrAPO-5 while homogeneous
Cr(acac) 3 , in contrast, afforded roughly the same conversion with triphenylmethyl
hydroperoxide as with TBHP.
Table 9. Catalyst recycle in the CrAPO-5 catalyzed oxidation of a-methylbenzyl alcohol with TBHP in PhCl
at 100 °Ca.
I1l 2 85
96
100
96
65
79
95
93
97 89 83 86 1
y
a. Data taken from ref. 38.
CH2R2
TBHP
R* (19)
CrAPO-5
PhCl ; 80°C
TBHP / CrAPO-5
ArCH2R ArCOR
PhCl/80°C/16h
If """ " ""' " " "
OOH
o 0 2 ; 115°C
CrAPO-5
at 3% cyclohexane conversion
In short, CrAPO-5 appears to form the basis for a variety of clean oxidation
processes in the liquid phase, whereby the catalyst is readily recovered and recycled.
OH
TBHP
Cr-PILC
- ^X ^
82% yield
(21)
CH2CI2 ; 25°C
CHO
(22)
"OH
Cr-PILC 84% yield
CH 2 CI 2 ; 25°C
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
the iron can be introduced as ferrocene49 or as a carbonyl complex47. In this way the
presence of unchelated metal ion in the zeolite cages is largely avoided.
FePc encapsulated in zeolite Y catalyzed the oxidation of cyclohexane with
TBHP with a turnover of 6000 compared to 25 with homogeneous FePc, indicating that
occlusion of the catalyst in the zeolite enhances its stability towards oxidative
degradation49.
More recently, Jacobs and coworkers have described the use of zeolite
encapsulated manganese(II) bipyridine (bipy) complexes, e.g. cis-Mn(bipy) -Y, as
catalysts for the oxidation of olefins with 30% aq. H 2 02 in acetone at ambient
temperature (Table 12). At low conversions the epoxide and diol were the major
products. At complete conversion, on the other hand, the dicarboxylic acids, resulting
from oxidative cleavage of the double bond, were the only products observed with
cyclohexene and cyclododecene.
Table 12. cis-Mn(bipy)22+-Y catalyzed oxidations of olefins with 30% aq. H 2 0 2 in acetone at 20 °Ca.
References
27. T. Sato, J. Dakka and R. A. Sheldon, J. Chem. Soc.f Chem. Commun., (1994)
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
201
1887.
28. A. Corma, M. T. Navarro and J. Perez-Pariente, J. Chem. Soc., Chem. Commun.,
(1994) 147.
29. P. T. Tanev, M. Chibwe and T. J. Pinnavaia, Nature, 368(1994)321
30. M. K. Dongare, P. Singh, P. P. Moghe and P. Ratnasamy, Zeolites 11 (1991) 690.
31. N. Kishor Mai, V. Ramaswamy, S. Ganapathy and AV. Ramaswamy, J. Chem.
Soc., Chem. Commun., (1994) 1933
32. J. S. T. Mambrim, E. J.S. Vichi, H. O. Pastore, C. U. Davanzo, H. Vargas, E.
Silva and O. Nakamura, J. Chem. Soc., Chem. Commun., (1991) 922.
33. M. Kawai and T. Kyoura, Jap. Patents 0358,954 and 0356,439 (1991) to Mitsui
Toatsu Chemicals, CA 115 (1991) 48863d and 48864e
34. S. T. Wilson, B. M. Lok, C. A. Messina, T. R. Cannan and E. M. Flanigen, J. Am.
Chem. Soc. 106 (1984) 6092.
35. J. D. Chen, J. Dakka and R. A. Sheldon, Appl. Catal. A: General 108 (1994) LI.
36. J. D. Chen, J. Dakka, E. Neeleman and R. A. Sheldon, J. Chem. Soc., Chem.
Commun., (1993) 1379.
37. J. D. Chen, M. J. Haanepen, J. H. C. van Hooff and R. A. Sheldon, Stud. Surf.
Sci. Catal. 84 (1994)973.
38. R. A. Sheldon, J. D. Chen, J. Dakka and E. Neeleman, in Zeolites and
Microporous Crystals, eds. T.Hattori and T.Yashima (Elsevier, Amsterdam,
1994)407.
39. J. D. Chen, H. E. B. Lempers and R. A. Sheldon, Colloids and Surfaces,
submitted for publication.
40. F. Figueras, Catal. Rev. Sci. Eng. 30 (1988)457.
41. B. M. Choudary, A. Durga Prasad, V. Bhuma and V. Swapna, J. Org. Chem. 57
(1992)5841.
42. B. M. Choudary, A. Durga Prasad and V. L. K. Valli, Tetrahedron Lett. 31 (1990)
5785.
43. B. M. Choudary, V. L. K. Valli and A. Durga Prasad, J. Chem. Soc., Chem.
Commun., (1990) 721.
44. M. S. Rigutto and H. van Bekkum, J.Mol.Catal., 81(1993)77
45. B. M. Choudary, V. L. K. Valli and A. Durga Prasad, J. Chem. Soc., Chem.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
i
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
ILYA I. MOISEEV
N.S.Kurhakov Institute of General & Inorganic Chemistry, Russian Academy of Sciences,
Leninsky Prosp. 31, 117907 Moscow GSP-1, Russia
ABSTRACT
Oxidation of alkenes catalyzed with palladium compounds represents a typical
example of oiganometallic catalysis involving transition metal organyl formation.
Ethylene oxidations to produce acetaldehyde and vinyl acetate are being widely used
on large scale. In this paper the experimental data relevant to the mechanistic aspects
of redox reactions between Pd(II) salts and alkenes are discussed. The latter are
compared with those related to the oxidation reactions catalyzed with Pd-561 clusters
modeling the active sites of supported Pd metal catalysts. By using solubility
measurements and kinetic methods, all stages of the alkene oxidation reactions with
Pd(II) salts have been characterized quantitatively. The mechanism of 71-complexes
and a-organyls transformation was elucidated by the study of oxidation reactions in
nonaqueous hydroxyl-containing solvents. The reactions of p-hydroxyethylpalladium
halogenides, their esters and ethers generated in situ by reacting palladium (II) salt with
corresponding organylmercuric halogenides were used to investigate the pathways of n
and a-bonded Pd-organyl reactions. All the data obtained including isotopic solvent
and substrate effects are in a good agreement with a mechanism involving the
formation ofrc-complexesand its transformations involving an 71-a-rearrangement to
form palladium (II) a-organyl derivative and its oxidative decomposition giving rise to
the observed product of alkene oxidation and the reduced form of palladium.
Giant cationic palladium clusters approximated as Pd56iL6()OAcigo
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
2CuCl2' + 2H+ + 1/2 0 2 * 2CuCl2 + H 2 0
Pdd 4 ~ CuCl2
C 2 H 4 + 1/2 0 2 ► CH3CHO + 52.26 kcal/mol (1)
C
nH2n +PdC1
42- — C
n»2nPdCl3" + C1
~ (5)
C
n H 2n + PdC1
4 2 -+ H
2° ^ ^ C
nH2nPdC12OH2 + 2C1
" (6)
*The perchlorate anion selected for compensation of the concentrational changes of the
chloride ion has appeientiy the least tendency of entering the inner coordination sphere of
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
The study of the olefin oxidation under conditions when the palladium
salt is consumed to form Pd metal and there is an increase in the concentration
of both hydrogen and chloride ions is connected with a number of
inconveniences. For this reason, in order to simplify the observed relationships
as much as possible the first papers on the kinetics of this reaction 10-13 w e r e
devoted to the investigation of the oxidation of olefins by PdCL^- in the
presence of p-benzoquinone as oxidizing agent of the reduced form of
palladium. In contrast with such oxidants as cupric chloride, the addition of p-
benzoquinone to the solution does not cause any changes in the concentration
of the hydrogen and chloride ions. The concentration of the olefin at a given
moment during the course of the reaction can be followed through the changes
in the oxidation-reduction potential of the quinone-hydroquinone system. Under
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
The reaction was found to strictly obey the first order with respect to the
ethylene concentration^"^. The observed first order rate constant (k\) is
proportional to the concentration of PdCLj^- and a reciprocal to the hydrogen
ion concentration and to the square of the chloride ion concentration. The rate
of the reaction decreases by 4.02±0.15 times on replacing water by the deuterium
oxide.
On the basis of the aforementioned data the author, Vargaftik and
Syrkinl3 proposed in 1963 a reaction mechanism which accounted for all
available facts concerning this reaction (Eq.(7.1)-(7.6)):
* We do not discuss the kinetics of olefin oxidation in the presence of both palladium
and copper saltsl^"!^ or Pd and vanadium oxides^.
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
209
found that the oxidation rate of propylene, but-1-ene, cis- and trans-but-2-enes
in oxidant-free solutions obeys the same law. Henry's conclusions as to the
sequence and nature of the kinetic steps of the reaction of olefins with PdCL^-
and his concepts on the mechanism of decomposition of the c-bonded organo-
palladium compound are very similar to those presented i n ^ .
However, already at this early stage of our investigation there were
indications that Eq. (9) does not quite satisfactorily describe the reaction
kinetics. In particular, it was observed that the reaction rate grows faster with
increasing concentration of palladium salt than could be expected on the basis
of mechanism (7).
The study of this phenomenon led to the conclusion^ that in the interval
of PdCL^- concentrations from 0.02 to 0.2 M the kinetic data obey the Eq.
(10):
[ P d C l ^ ] [ C n H 2 J t w [PdCi;-]2[CnH2n] ( x1 0 )
W = K| + 2r + KJI + 35 '
[H 3 o ][cr] [H 3 o ][cr]
At PdCl|2- concentration 0.02 M contribution of the first term is 75%
and that of the second one is only 25%. For 0.2 M PdCL^- it is vice versa: 25
and 75%. Nevertheless in both cases (and also at intennediate concentrations of
PdCLj2-) the ki - 1/[H30 + ] plots pass through the origin. The same is also true
for other olefins.
The validity of the Eq. (10) is exemplified by Fig.l
The presence of a second term leads to the conclusion, that reactions of
olefins with P d C ^ - proceed along two parallel routes. We believe that stages
(7.1) and (7.2) and possibly (7.3) of scheme (7) are common to both routes.
Further transformations of the it-complex C n H2 n PdCl20H" can proceed either
without participation of other ions (first route) or with participation of a
palladium acido complex (second route). In the latter case the following
reactions should proceed alongside with stages (7.4) and (7.5):
Q 2-
Cl^ XC1^ | x 0 H
Pdci42-+ c n H 2n Pda 2 oH- ^ = ^ ^Pd^ ^M + cr
(I)
OH v (11)
a ci cr ci c n H 2 n oH
a ■ ^ a ^ jz\~~ CL ^a2"
Pd Pd + H20 ► ^Pd-Pd +
Cl" x c i " X C n H 2n OH Cl C1
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
+ CnH2nO + H3O+ + CI"
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
210
somewhat more favorable and occur more easily than the outer-spheric ones. In
amine media, a more basic solution, the amination of an coordinated alkene
proceeds as trans-addition like the expected one by24,25 the Eq. (12). If the first
and second routes are suppressed by high chloride ion concentrations, in the
presence of Cu(II) salts or nitrate ion, glycoi esters are formed instead of
acetaldehyde. This reaction supposedly involves intermediate complex formation
between Pd(II) and the oxidant used, e.g. Cu(II) salt or NO3" ion26. These
reactions have been shown to involve trans-attack of the nucleophile on the
coordinated alkene27-30.
1.4. Mechanism of oxidative reductive decomposition of the n-complexes
The reactions of olefins with palladium (II) salts in water and in
nonaqueous solvents have a number of common features. Both in water and in
the nonaqueous solvents decomposition of the n-complex results in Pd metal
and the product of the organic ligand oxidation. An easily detectable genetic
relationship exists between the products of the reaction of ethylene in a variety
of solvents (vinyl ester and ethylidene diacetate in acetic acid5,31-33( a c e tal hi
alcohol^'33, acetaldehyde in water3>4,6).
These facts lead to the conclusion that the carbonyl compound in water
and the olefin oxidation products in nonaqueous media have precursors of
similar structure. Of course, both the formation of these intermediates and their
further transformations are not completely equivalent in different media. This
section is devoted to certain aspects of the mechanisms of formation and
decomposition of these hypothetical intermediates.
1.4.1. a-Bonded Organo-palladium Compounds in the Oxidation of Olefins
The formation of acetaldehyde by hydrolytic decomposition of the anion
of the Zeise's salt was once considered34 as a support for the ethylidene
structure of the organic ligand in this complex:
Cl3Pt2--CH-CH3 + ► Cl3Pt-=CH-CH3
It is not clear, however, why only the oxocomplex can undergo such
reaction and what is the function of the bonded hydroxyl. Moreover, if
conversion of the 71- to the carbene complex is an equilibrium reaction as it was
proposed by Aguilo^ then palladium (II) catalyzed isomerization of the olefin
should take place. Actually the isomerization is catalyzed by complexes of low
valence palladium. No isomerization was observed in PdCl2 solutions containing
oxidants, where palladium (II) 7t-complexes including C a H2 n PdCl20H" are
formed, and Pd(I) and Pd(O) are practically absent35-3o_ Henry has found
C2D4 and C2H4 to react with equal rates^. Hence, if the above discussed
mechanism is considered to be valid, it must be assumed that 7t-complex to
carbene-complex conversion is the slowest stage of the over-all reaction and,
moreover, proceeds without an isotopic effect. At present there is no actual
evidence confirming this mechanism. --`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Cl CnH2n
The data on the equilibrium of complex formation and kinetic studies of
ethylene oxidation by palladium acetate in acetic acid led to the conclusion that
in this medium the reaction proceeds by inner-spheric 7t-a-transformation of the
complex C2H4Pd(OAc)2HOAc. The reaction rate depends only on the
concentration of this complex and does not depend on the concentrations of
free AcONa.
Oxymercuration, the most completely studied reaction of oxymetallation,
with few exceptions, proceeds via trans-additional .38. On the contrary,
oxypalladation in the cases when this reaction is inner-spheric should typify cis-
addition. In analyzing the reactions for such differences as polarities and strength
of the metal-oxygen and metal-carbon bonds in the Pd(II) and Hg(II)
complexes, nucleophilic activities of the migrating ligands in these complexes,
etc., should be taken into account. The absence of pertinent data precludes any
detailed discussion. However, two facts should be mentioned which perhaps
could shed some light on this problem.
In contrast with comparatively stable palladium complexes, 7t-complexes
Copyright World Scientific Publishing Co.
of under
Provided by IHS Markit mercury are
license with unstable because of the lower
WILEY ability- 7 -of
Licensee=FLUOR Hg(II)
Mexico, to back-donation.
D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
213
In reaction (13) the C=C n-bond is replaced by a stronger C-0 bond but
the Pd atom now has only one ligand instead of two. Owing to this, reaction
(13) should have a highly negative heat effect absolute value being probably
several times larger than the activation energy of this elementary act*. Obviously
in order to compensate the energy loss associated with the loss of one ligand
some bonds should arise in the resultant complex which are not shown in the
structure of the o-carbocomplex in Eq. (13); for example, "hydrogen bond" (II)
or Pd-0 bond (III):
A >HN
CloPd Cl2~Pd CHOH CHOH
X \ /
CH2 CH2
II III
At present no data are available either to support or invalidate these
assumption.
However, one cannot exclude the possibility that it-o-isomerization is not
monomolecuiar but proceeds with participation of solvent molecules which enter
the inner coordination sphere of palladium and occupy the fourth coordination
site of the organopalladium compound. For the reaction in water the following
scheme appears to be plausible.
Cl Cl Cl #
H20 H20 H20
OH-Pd — C H 2 Pd— CH 9
I CH9 HO' IC ' H 2 HO-|-CH2
l 2
Cl Cl Cl
IV IVa
(H 2 0)Cl 2 Pd-CH 2 CH 2 OH
* The heat effect of reaction (13) was estimated to be about 80-100 kcal/mol. The
activation enthalpy of ethvlene oxidation in water (for the first rout) is 16-19 kcal/mol^' 13 ; in
acetic acid 17 kcal/mol^4. The activation enthalpy of the elementary act under discussion
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
probably does not differ from the observed values by more than 5-10 kcal/mol.
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
214
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
For binuclear complexes (second route) the Jt-o-isomerization can
proceed also without participation of the solvent. The palladium atom bonded
with olefin increases its coordination number to 5 through the two bridged
chlorine atoms*. In the result o-bonded organopalladium compound both
palladium atoms have usual coordination number 4 (Eq. (11)).
The greater size of the H20Cl2Pd group in comparison with that of the
hydroxyl favors Markovnikov's rule addition in the oxydation of a-olefins in
water1*. Palladium acetate in acetic acid reacts with olefins also mainly by
Markovnikov's rule48. Here also the size of the inserting acido-complex group is
greater than that of the acetoxy group. However, the olefin oxidation by
palladium chloride in acetic acid is considerably less selective^,50_ X J ^ c a n be
easily understood if one takes into account that the size of the acido-complex
group with two chlorine atoms is much less than that of the Pd(OAc)2HOAc
group, and, therefore, in reaction with palladium chloride steric control
apparently plays a minor role.
* The interaction between palladium atoms in the binuclear complex via bridged
chlorines is likely to increase the acceptor ability of the n-complexed palladium atom and,
Copyright World Scientific Publishing Co.
therefore, facititate the it-o-isomerization.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
215
rci in)
^\^\ Y HO
(H 2 0)ClPd—'CH2-CH-OH 2 ► PdCl- (aq )+CH 3 CHO+Cl-+H30+
In the light of the above mechanism the absence of glycols among the
products of the reaction of olefins and palladium salts in aqueous solution can
be regarded as indication that heterolysis of the Pd-C bond and the hydride shift
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
proceed faster than reaction of the organopalladium compound with the solvent
molecule, resulting in its alkylation. It is possible that the 1,2-hydride shift in
this case occurs not only in the field of two carbon atoms but also under
influence of the palladium atom as Henry postulated^.
In the reaction of olefins in alcohol or acetic acid, elimination of R+,
when R is CH3, C2H5 or Ac in the course of decomposition of the
organopalladium compound is hindered because of the lesser tendency of alkyls
and acyls than hydrogen to exist in the form of positively charged ions. Perhaps
in these solvents one could isolate the kinetic stage of carbonium ion formation:
complexes with the same acido-ligands. Correspondingly changes also may take
place in the rates of formation and decomposition of oigano metallic
compounds. In the case of the Tl(III) and Hg(II) glycol formation becomes an
important reaction path56-58#
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
217
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
palladium metal or Pd(II) 63 * 64 .
Catalytically active solutions were found to be formed via reduction of
Pd(OAc)2, in the presence of the ligands L (L=l,10-phenanthroline, 3,3"-
dicarboxy-2,2"-bipyridine, 2,2"-biquinoline, etc.), with various reductants, the
most convenient being H2^>64 Catalytic activity for oxidation of olefins, as a
function of total content of a "mild" base L, exhibited an extreme patter.
In order to elucidate the nature of these compounds, the stoichiometry of
the reactions leading to their formation, as well as their compositions and
structures, were studied.
The primary isolated product of palladium (II) acetate reduction with H2
in AcOH solution (containing 0.5 mol of phen per Pd atom) was an X-ray-
amorphous substance with Pd4phen(OCOCH3)9 as the simplest compositional
formula, as ascertained by elemental analysis^. However, as was found by
volumetric measurement, H2 consumption was 1.3 ± 0.05 mol per Pd (II) atom,
corresponding to the stoicheiometric equation:
4Pd 3 (OCOCH 3 )6 + 3phen + 15H2 = 3Pd 4 phen(OCOCH 3 ) 2 H4 +
+ I8CH3COOH
The suggestion that the complex 1 is a hydride is supported by ^H NMR
spectra65,66
The metal core size for 1 molecules found from TEM and SAXS data led
to conclusion that the value of n in the idealized [Pd4phen(OAc)2H4]n formula
as approximately 100, supposing the packing density of Pd atoms in 1 core to be
nearly the same as that for bulk palladium metal.
2.2. Palladimn-561 cluster (2)
When reacted with O2, compound 1 lost hydride atoms to form H2O. In
this process only a small portion of the palladium atoms was oxidized to Pd(II),
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
218
The nearly identical patterns of the interplanar distances found for 2 and
for Pd metal might indicate the coincidence of the main symmetry features for
the 2 metal skeleton with those of the lattice for bulky Pd metal. However this
suggestion contradicts the EXAFS data, which definitely indicate icosahedral
packing of Pd atoms in 2 (see below). In this situation it seems more reasonable
to assume that the similarity in the ED patterns of 2 and metallic Pd results
from the destruction of 2 molecules to produce particles of metallic Pd under
the influence of the electron beam. A destructive influence of the electron beam
upon cluster molecules exposed in the electron microscope is indeed known to
occur under the conditions of TEM and ED experiments. For example, the loss
of cluster ligands, agglomeration and beam damage during TEM studies have
been observed for Au55(PPh3)i2Cl66^»6^ and
69
[Ni38Pt6(CO)48H](NMe3CH2Ph)5 clusters at high electron beam intensities
and long experiment times. In TEM experiments performed at low beam
intensities, neither agglomeration nor beam damage of the shape of palladium
clusters was found upon variation of the exposure time from 10 s to 10 min6^.
However, total or partial loss of the cluster ligands cannot be excluded, even at
the low beam intensities used. Ligand loss may result in a relaxation of the initial
icosahedral metal skeleton into a f.cc. one (e.g. a cubooctahedral).
In the absence of the agglomeration of palladium particles, relaxation of
the metal skeleton may be expected not to affect noticeably the size of the metal
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
molecules. From the half-width of diffraction rings, the size of the particles
responsible for the diffraction pattern was found to be ca. 25 A, in agreement
with TEM and SAXS data.
The consistency of the data on metal skeleton size obtained with various
techniques (Table 3) seems to confirm the lack of considerable destruction of the
metal skeleton or agglomeration of Pd clusters in the TEM and ED
experiments*^.The size distribution of 2 cores was found to be monomodal.
The packing of Pd atoms in 2 was elucidated with EXAFS
spectroscopy^,70, The four intense peaks corresponding to the four shortest
Pd-Pd distances have been found (Table 4). The set of Pd-Pd distances obtained
is seen to be consistent with the icosahedral packing of Pd atoms in the metal
skeleton of the cluster 2 (the interatomic distance ratios expected for the four
nearest neighbor atoms of the icosahedral skeleton are 1:1.2:1.4:1.6) and to
deviate notably from the patterns of distances expected for f.c.c. and h.c.p.
packing.
Table 3. Metal skeleton size (d) of 2 molecules as evaluated by various techniques*^
1 Method d(A)
TEM 26 ± 3.5
ED -25
SAXS 20 ± 5
TABLE 4. The four shortest Pd-Pd distances in the metal skeleton of 2 found from
EXAFS data, compared with those distances expected for different packings of the Pd atoms
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
The idealized formula Pd5(>]phen(>o(OCOCH3)i80 see nis to correspond
to some average size and composition of this cluster, rather than to a certain
fixed size and composition. In other words, the existence of distributions in the
size and composition of various particles of 2, around the average value of d «
25 A and the idealized composition Pd56iphen6o(OCOCH3)i80 should be
anticipated.
HREM studies^ of clusters, approximate formula Pd56iL6o(OAc)igo
(L=Dipy, Phen), have confirmed the conclusions^**)*) concerning with
monomodal size distribution of the cluster cores. Besides Pd species with f.c.c.
metal core, larger Pd particles (8 nm) exhibiting multiple twinning and
evidencing for a distorted icosahedral structure were observed.
The observed by TEM metal cores of the particles of 2 are indeed
characterized by a distribution over the values of their diameters, rather than by
a single value of the diameter. This suggests that, in fact, various particles of the
isolated substance 2 contain different numbers of palladium atoms in their cores.
The situation here seems to resemble to some extent that for organic polymers,
in which various molecules contain different numbers of monomer units. Taking
into consideration that the average energy of Pd-Pd bonds is rather small (40-60
kJ moH), the icosahedral metal clusters with d * 25 A can be expected to have
some defects ("caps", "nichas" , etc.) which perhaps are not detected by TEM or
HREM and other techniques used to characterize the cluster.
^—i~i—-^ ML
•AS • 1 1 ,J m i , > \ (OAc)
Wmffit-K®
6 / 7~''";f»5 C;X |§)
6) .r^Sfa%QQ@R«r/j i
T@..
HREM studies. The difference is presumably due to the ligand shell of the
clusters, which is invisible to electron microscopy^.
Cluster 2 can be precipitated from aqueous solutions by adding salts such
as NaX or KX (X" = Cl", Br, CIO4-, HSO4-, PF6~, AsF6"), with substitution
of OAc anions by X" anions. Upon treatment of 2 with KPF6, a cluster soluble
in CH3CN, with idealized formula Vd%\phtn^QO^Q(VF^Q9 3, was obtained75
Pd561phen6o(OAc)180 + 60 PF6~+ 60 H 2 0 "♦
► Pd56iPhcn6006o(PF6)60 + 120. AcOH + 60 AcO" (14)
The shape of the metal skeleton of cluster 3, as a TEM micrograph
shows, is almost spherical, 28 ± 5 A in diameter, consistent with SAXS data.
Molecular clusters containing 4-38 palladium atoms are known to be
diamagnetic73>76# Unlike these, clusters 1-3, with several hundred Pd atoms in
their cores, reveal metal-like properties such as temperature independent
paramagnetism (Fig. 3)77- The values of the specific magnetic susceptibility at
300 K (x g 3 0 0 = 1.0 xlO"6 CGSU for 2 and Xg* 00 ^ 0.8 xlO"6 CGSU for 3) are
close to those for palladium particles supported on SiC>2 (dispersion 0.2-0.5,
6
Xg 300= (0.8 ± 0.2) xlO- CGSU)77.
$14
<> o
2 '^^Jt^o^^^
0,1.0
X
0.6
0.2
100 200 • 300 K
Pig. 3. Specific magnetic susceptibility, Xg, of clusters 2 (O) and 8 (A) as a function of
temperature.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
223
distances was obtained from the g(R) function for cluster 3, which was
interpreted in terms of a model, assuming that the product of the reaction 2 with
KPFg is similar to the initial cluster 2. Within this model, the EXAFS data
should be treated as evidence of a more complicated arrangement of Pd atoms in
3 than in 2.
The ligand substitution process (see Eq.14) is assumed to include other
anions besides PFg". Perhaps the reaction is a hydrolysis resulting in the
appearance of the anions OH" and O^" at the cluster surface, accompanying the
lidand substitution in aqueous solutions at pH 5- 6. Unlike large PF6" and OAc~
anions, the small OH" and O^- anions can perhaps be bound directly to Pd
atoms, rather than being located in the outer sphere of the cluster.
The lack of an absorption in the 3600 cmfl region (which is characteristic
of OH groups in the IR spectrum of 3) makes the 0^~ anion a more likely
hypothetical additional ligand.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
The oxygen ligands may also be inserted between two outer (i.e. the 4th
and 5th) icosahedral layers of Pd atoms. Both coordination of O atoms at the
surface of the skeleton of 3 and their insertion between the outer layers of the
skeleton are expected to lengthen the Pd-Pd distances in the outer layer. This
can explain the appearance in the g(R) function of cluster 3 of additional 4.45 A
distance that does not fit the idealized icosahedral structure.
Thus, cluster 3 is inferred to be icosahedral, with the inner layers of its
core being packed on the same way as those in cluster 2 (Pd-Pd distances are
2.55, 3.05, 3.60 and 4.05 A), and outer- layer Pd atoms (ca. 50% of total Pd)
located at perturbed distances.
The giant clusters obtained eventually serve as a bridge between ordinary
molecular clusters and colloidal metals. The sizes of the metal skeletons of the
giant clusters exceed notably those of large molecular clusters such as
Pd3g(CO)28PEt3)i2 ^ and Au55(PPh3)i2Cl6 ^> being close to lower sizes of
colloidal metal particles. Unlike the latter, the giant clusters have a distinct
ligand environment with a definite stoichiometry inherent in molecular clusters.
However, a set of more or less imperfect metal polyhedra with a certain size
distribution, rather than a single perfect polyhedron as in molecular clusters,
seems to arise when the number of metal atoms amounts to several hundred.
"Nevertheless, the idealized formulae based on the data mentioned above are
useful as models characterizing the average size and composition of the giant
clusters and providing an understanding of their catalytic properties.
When obtaining vinyl and allyl acetates with solid catalysts under oxygen
+ acetic acid vapor at 430-470 K^>^,62 ) oxidation of metallic palladium to
form Pd(II) may, in principle, occur. However, the thermal instability of Pd(II)
acetate complexes at temperatures above 373 K, as well as the data on selectivity
of the reactions of alkenyl ester formation in systems including Pd(II),
contradicts the hypothesis of the heterogeneous process via reactions (18) and
(19).
In the case of Pd(OAc)2, interaction with propylene in liquid AcOH, a
mixture of allyl, isopropenyl and n-propenyl acetates is formed83. The allyl
acetate yield decreases with increasing temperature up to 380 K. and aqueous
contents of the solution up to 1-3 %, at the expense of acetone yield increase.
Over Pd metal catalysts, the noted side products are not actually formed, and
almost no decrease in the yield of allyl acetate is observed upon a temperature
increase from 420 to 460 K and addition of water up to 10%64.
In the case of ethylene, both in heterogeneous and homogeneous systems,
the same ester, vinyl acetate, is formed. However, in Pd(II) systems, even small
(0.2-1.0 %) quantities of H2O resultin the appearance of acetaldehyde, whereas
in the case of Pd metal catalysts, vinyl acetate is the sole product of the
oxidation of ethylene even at significantly higher water contents.59-62,84
For oxidation of alkyl arenes in the presence of carboxylic acids,
substantial differences in selectivity of homogeneous Pd(II) and heterogeneous
Pd metal catalysts are also observed. For example, under the action of Pd(II),
toluene is converted mainly into bitolyls, which are the products of oxidative
coupling of aromatic rings. Over metallic Pd in the presence of AcOH, oxidative
carboxylation of the methyl group becomes the main route of toluene oxidation.
PI1CH3 + l / 2 0 2 + AcOH * PI1CH2OAC + H2O
The above data suggest that vinyl and allyl acetate can be formed by two
different routes depending on the Pd-atom oxidation state in the catalyst used to
perform the reaction.
Only vinyl and allyl acetates were formed by ethylene and propylene
oxidations, respectively, in solutions of clusters 2 and 3 containing up to 10%
H2065. The fact that carbonyl compounds are absent from the products seems
to exclude the possibility that the reactions occur via oxidation of some of the
cluster Pd atoms to Pd(II) and subsequent reduction of Pd(II) by olefins, i.e. via
reactions (18) and (19). Therefore, clusters 2 and 3 may be regarded as good
models of solid catalysts for which, in contrast to homogeneous Pd(II) systems,
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
20' /
10 ■
0 I 2 0 5 10 IS
(a) lo'l^Hd.M (b) IO'ICJHJ.M
Fig. 4 The observed rate constant, * 0 b„ for formation of vinyl acetate (a) and allyl
acetate (b), as functions of the ethylene and propylene concentrations. 333 K, solvent
MeCN-AcOH mixture, [AcOH] - 1.59 M, [Oj] - 4.2 X 1<T3 M, [3] - 1.25 X 10"4 M.
Cl"l CjH,
t.{
2
0 05 1 1.3 o 2 4 6
(•) [AcOHJ, M
C,H4 as <w* S
c2
E
1
•^ 1
' >r al
■s
acs
0 2 4 8 0 OS 1
(b) l/[Ac0Hl M-'
Fig. B ■. The values of *,*, u a function of AcOH concentrations for ethylene (») and
propylene (0) oxidations; (a) the same daU in Michaelis-Menten coordinates; (b) 333 K,
solutions in MeCN-AcOH mixtures, [ 0 , ] - 4.2 X NT* M; for (») [CH4] - 2.16 X 10^ M;
[3] - 1.09 x lO"" M; for (0) [CjH«] - 9.0 X 10"2 M, [3] - 1.25 X 10"^ M.
c
I
'c 15
■g 4
r 0
3 l^jirCS^* 04
2
5
0.2
1
0 1 2 3 4 0 1 , 23 3 4
M IO- /t0jl.M-'
(«) 10 I°il, (b)
Fig.6. (a)Dependence of k ^ on the O2 concentrations for ethylene under P=0.2 atm and
propylene under P=0.5 atm.;(b) the Michaelis-Menten anamorphosis for ethylene oxidation.
333 K, solvent MeCN-AcOH; [C 2 H4]=2.16X10-2M, [AcOH]=1.59 M, [3]=1.09X10" 4 M;
[C3H6]=0.162 M, [AcOH]=1.75 M, [3]=1.25X10-4M
Thus, experimental values for the reaction rate may be described by the
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
where [Oust] and [01] stand for the concentrations of a giant cluster and
an olefin, respectively. Note that the reaction between the coordinated CnH2n>
O2 and AcOH species characterized by the effective rate constant k, is expected
to be in fact a complex one, consisting of several elementary steps.
From Table 5 it is seen that the rate constants, k, for ethylene and
propylene oxidations in solutions of both clusters differ less than three-fold. The
Michaelis constants Kj, KTT and K m for both olefins are also of the same order
of magnitude, the largest variation upon the change of the olefin being observed
for the constant Kj. The Michaelis-Menten character of the kinetics of reactions
(15) and (16) suggests the product formation to be preceded by stages of
reversible coordination of olefin, O2 and AcOH molecules with the cluster. A
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
assumption of the absence of such influence seems quite reasonable for a big
coordination centre such as a giant cluster with several hundred atoms of
palladium.
A smaller value of the stability constant for the u-complex of the giant
cluster with propylene as compared to ethylene may be explained by the size of
the propylene molecule exceeding that of ethylene. The surface of the cluster
core is expected to be substantially screened by the bulky molecules of the
phenanthroline ligand. Examination of the idealized structure of cluster 2 (Fig.
2) using molecular models shows that only ca. 20 palladium atoms on the
surface of the metal skeleton of the cluster are sterically accessible for
coordination of the olefin molecules. It may happen that because of steric
hindrance the more bulky C3H6 molecules will be coordinated at these sites less
strongly that the smaller C2H4.
For real molecules of giant clusters, with the structure probably deviating
somewhat from the perfect polyhedron, the number of accessible coordination
sites may deviate from that for the idealized model. The number of these sites
can be determined experimentally using poisoning techniques, i.e. by carrying
out the oxidation reaction in the presence of some ligands that can be strongly
bound to those palladium atoms on the surface of the metal skeleton which are
active in catalysis (Fig.7).
1.0
aaa
*
joe
CM
02
0 X> 40 60 ''tO
[L] £ /[Cl U st] ?
Fig. 7. The rates of the oxidative acetoxylation for propylene and ethylene, in MeCN-
AcOH solution of cluster 3 as a function of the concentrations of ligand-inhibitors: (•)
CjHsSH; (v) I3; (0) EtjNCSjNa; (•) KSCN| (A) phen; (•) PP n3 .
The data on the influence of the poisons on the rates of C2H4 and C3H6
oxidative acetoxylation in solutions of cluster 3 showed that the bulky ligands
that coordinate to Pd(II) atoms, e.g. PH13 and phen, actually have no effect on
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`--
the rate of the olefin oxidation, apparently because they cannot be coordinated
on the sites suitable for olefin, O2 and AcOH binding in the course of the
reaction. The smaller ligands, e.g. C2H5SH and the thiocyanide anion,
efficiently suppress the catalytic activity of the cluster. For complete inhibition of
C2H4
Copyright World Scientific oxidation,
Publishing Co. it is necessary to introduce ca. 50 ligand molecules to the
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
231
H2C=CH H V
-Pd-Pd-••Pd-
H C—CH—CH3 slow
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
2 H2C/ I %
CH2 H
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
-Pd-Pd-Pd-
No reproduction or networking permitted without license from IHS
-Pd-Pd---Pd-
Not for Resale, 11/16/2017 06:42:59 MST
233
Cn***!
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Acknowledgments
I am indebted to many co-workers and collaborators whose names appear
in the reference list below and whose contributions have been essential to the
development of our research in the field.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
References
1. RASheldon and J.K.Kochi, Metal-Catalyzed Oxidations of organic
Compounds, Acad. Press, New Jork (1981), 423.
2. F.C.Phillips, Z. Anorg. Chem., 6 (1894), 213.
3. J.Smidt, W.Hafner, RJira, J.Sedimeier, R.Sieber, R.Ruttinger, and H.Kojer,
Angew. Chem,, 71 (1959), 176.
4. RJira, and W.Freiesleben, in Oranometallic Reactions, Ed. by E.Becker and
M. Tsutsui, Wiley&Sons, 3 (1972), 5.
5. I.I.Moiseev, M.N.Vargaftik, and Ya.K.Syrkin, Doklady Acad. Nauk SSSR,
133 (1960), 377 (in Russ.).
6. I.I.Moiseev, M.N.Vargaftik, and Ya.K.Syrkin, Doklady Acad. Nauk SSSR,
130 (1960), 820 (in Russ.).
7. J.Chatt, Nature, 165 (1950), 859.
8. J.Chatt, and R.G.Wilkins, /. Chem. Soc. (1952), 2622.
9. I.I.Moiseev, M.N.Vargaftik, and Ya.K.Syrkin, Doklady Acad. Nauk SSSR,
152 (1963), 147 (in Russ.).
10. I.I.Moiseev, Amer.Chem.Soc, Div.Petrol.Chem., 14 (1969), B49.
11. M.N.Vargaftik, I.I.Moiseev, and Ya.K.Syrkin, Doklady Acad. Nauk SSSR,
139 (1961), 1396 (in Russ.).
12. M.N.Vargaftik, I.I.Moiseev, and Ya.K.Syrkin, Doklady Acad. Nauk SSSR,
147 (1962), 399 (in Russ.).
13. I.I.Moiseev, M.N.Vargaftik, and Ya.K.Syrkin, Doklady Acad. Nauk SSSR,
153 (1963), 140 (in Russ.).
14. (a) P.M.Henry, J.Amer.Chem.Soc, 86 (1964), 3246; (b) P.M.Henry
Palladium Catalyzed Oxidation of Hydrocarbons, Reidel, Dordrecht (1980).
15. A.Aguilo, Adv. Orgonometal. Chem., 5 (1967), 321.
16. T.Dozono, and T.Shiba, Bull. Japan. Petr. Inst, 5 (1965), 8.
17. K.I.Matveev, I.F.Bukhtoyarov, N.N.Schul"tz, and O.A.Emel"yanova, --`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
(1986), 253 (in Russ.); (b) V.P. Zagorodnikov, M.N. Vargaftik, D.I.
Copyright World Scientific Publishing Co.
Kotchubey,
Provided by IHS Markit under V.A.Likholobov, V.N. Kolomiychuk,
license with WILEY A.N.D.F.Naumochkin,
Licensee=FLUOR - 7 - Mexico, A.L. Gerardo
Mexico/2110503118, User=Villalobos,
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
238
Chuvilin, V.M.Novotortsev, O.G. Ellert and I.I. Moiseev, Izv. Akad. Nauk
SSSR, Ser. Rhim., (1989), 849 (in Russ.).
76. E.G. Mednikov, N.K. Eremenko, Yu.L. Slovokhotov and Yu.T. Struchkov,
J.Chem. Soc. Chem. Commun., (1987), 218.
77. S. Ladas, B.R.A. Dalla Betta and M. Boudart, /. Catal, 53 (1978), 356.
78. A.N. Naumochkin and D.I. Kotchubey, Nucl. Instruments and Methods in
Physical Research, A261 (1987), 163.
79. G. Schmid, R. Pfeil, R. Boese, F. Bandermann, S. Meyer, G.H.M. Calis
and J.W.A. van der Velden, Chem. Ber.,\\A (1981), 3634.
80. V.P. Zagorodnikov and M.N. Vaigaftik, Izv. Akad. Nauk SSSR, Ser. Rhim.,
(1985), 2652 (in Russ.).
81. P.LPasichnyk, M.KStarchevsky, Yu.A.Pazdersky, M.N. Vaigaftik, LI
Moiseev, Mendeleev Commun. (1994), 1.
82. M.K.Starchevsky, S.L.Gladyi, YuAPazdersky, M.N. Vaigaftik, I.I Moiseev,
to be published.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
83. S. Winstein, J. McCaskie, H.-B. Lee and P.M. Henry, /. Am. Chem. Soc,
98 (1976), 6913.
84. M. Tamura and T. Yasui, Shokubai, 21 (1979), 54.
85. M.K. Starchevsky, M.N. Vaigaftik and LI Moiseev, Kinet. Ratal, 20 (1979),
1163 (in Russ.).
86. I.P. Stolarov, M.N. Vaigaftik and LI. Moiseev, Rinet. Ratal, 28 (1987),
1359 (in Russ.).
87. V.N. Kondrat'ev, (ed.), Chemical Bond Energies, Ionisation Potentials and
Electron Affinities, Nauka, Moscow (1974), 39 (in Russ.).
88. LI. Moiseev, Advances in Science and Technology, Ser: Kinetics and
Catalysts, VINITI, Moscow, Vol. 13 (1984), 147 (in Russ.).
89. V.P. Zagorodnikov and M.N. Vaigaftik, Rinet. Ratal, 27 (1986), 851 (in
Russ.).
90. A. Farkas and L. Farkas, /. Am. Chem. Soc, 60 (1938), 22.
91. E.M. Stuve and R.J. Madix, /. Phys. Chem., 89 (1985), 105 43. I.P.
92. Stolarov,M.N. Vaigaftik, O.M. Nefedov and LI. Moiseev, Izv. Akad. Nauk
SSSR, Ser. Rhim. (1983), 1455 (in Russ.).
93. LI. Moiseev, A.P. Belov, V.A. Igoshin and Ya.K. Syrkin, Dokl. Akad. Nauk
SSSR 173 (1967), 863 (in Russ.).
R. A. SHELDON
Laboratory for Organic Chemistry and Catalysis,
Delft University of Technology, Julianalaan 136,
2628 BL Delft, The Netherlands
ABSTRACT
The characteristics of fine versus bulk chemicals manufacture and catalytic oxidation versus catalytic
oxygen transfer are explained. The applications of a variety of catalytic systems in organic synthesis,
including catalytic asymmetric oxidations, are reviewed.
Although they share many common features there are several basic differences between
fine and bulk chemicals manufacture that can have an important bearing on process selection. Fine
chemicals are often complex, multifunctional molecules with low volatility and limited thermal
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
NH2
Fe MnQ2>
HC1 H 2 S0 4
Fe HC1
vJ^W
02 [H+] ^
catalyst -2Me2CcT
O2H
Classical > 10 kg salts per kg HQ
Catalytic < 1 kg salts per kg (HQ light)
stability. This necessitates reaction in the liquid phase at moderate temperatures. In addition, many
of the desired transformations involve chemo-, regio- or stereoselectivity. Processing tends to be
multipurpose and batch-wise, rather than dedicated and continuous as in bulk chemicals
production. This means that not only raw materials costs but also simplicity of operation and
multipurpose character of the installations are important economic considerations. Furthermore,
fine chemicals cannot, generally speaking, bear the costs of the extensive research program
characteristic of the development of a proprietory catalyst for a large volume chemical.
Consequently, one may have to be content with a catalyst that is perhaps not optimal but is readily
available.
As noted above the use of gas phase oxidation is often precluded in fine chemicals
synthesis. Nevertheless, where feasible it will often be the method of choice and several examples
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
^ 3 CQjH
Oleum
NO2 N0 2
37 kg salts / kg
1. Fe/HCl
l.Qz
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
2. aq. HC1
reflux 2. H+(-3Me2CO)
S
X
X=Br, Cl
of fine chemicals production via gas phase oxidation are known. A prime example is the BASF
process for the manufacture of citral, an intermediate in the synthesis of fragrance chemicals and
vitamin A. The traditional process involves a five step process, a key step of which is a
stoichiometric oxidation with MnC>2
The first step in the BASF route (figure 4) involves acid-catalyzed condensation of the
inexpensive raw materials, isobutene and formaldehyde, to give 2-methyl-l-butene-4-ol
(isoprenol). A key step in the process is the chemoselective gas phase oxidation of the latter over
a supported silver catalyst to yield the corresponding unsaturated aldehyde (figure 4). In order to
obtain a high selectivity (95%) the reaction is carried out continuously in a short fixed bed reactor
at 500°C with very short (0.001 sec.) residence times. The catalyst is the same as that used in the
manufacture of formaldehyde by gas phase oxidation of methanol This process is a perfect
example of the commercial production of a fine chemical via gas phase oxidation with dioxygen.
AT HC1
Cu8Cl2
^B-pinene myrcene
l.NaOAc
Cl 2.NaOH OH
MnO,
CHO citral
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
>
+ H2C0 1 'OH
0 2 ; 500 °C
Ag/Si0 8
OH 1 CHO
^ ^ ^ O ^ W ^ v ^— ^ ^ ^ N ^ W CHO
Figure 4. BASF Process for Citral Nfanufacture.
Another example of the use of gas phase oxidation for fine chemicals production is the
Lonza process5 for nicotinic acid manufacture in which the key step is the ammoxidation of beta-
picoline 3-cyanopyridine (reaction 1). Gas phase ammoxidation has also been applied to the
production of other (hetero) aromatic nitriles, e.g. 2-cyanopyrazine (reaction 2) 6 and 2,6-
dichloro-benzonitrile (reaction 3). The former is the precurser of 2-amidopiperazine, an
antituberculostatic, and the latter is an agrochemical intermediate.
(D
350 °C
(2)
^<^y Sb/V/Mn oxide ^Njj^
CH 3 CN
(3)
In bulk chemicals manufacture the choice of oxidant is largely restricted, for economic
reasons, to dioxygen. In contrast to catalytic reactions with H 2 and CO catalytic oxidations with
0 2 are complicated by the fact that dioxygen reacts with organic substrates even in the absence of
a catalyst. In the liquid phase this blank reaction involves the formation of hydroperoxide
intermediates via a free radical chain process (see chapter 8). A major problem associated with
autoxidations is that they are largely indiscriminate, i.e. they exhibit poor chemo- and
regioselectivities. Moreover, primary oxidation products such as alcohols and aldehydes are
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
generally more susceptible to autoxidation than the hydrocarbon substrates, thus necessitating low
conversions and recycling of large quantities of substrate. Consequently, liquid phase autoxidation
is synthetically useful only with relatively simple substrates containing one reactive position, e.g.
the oxidation of substituted toluenes to the corresponding carboxylic acids (see chapter 8). The
economics of fine chemicals manufacture, in contrast, allow for a broader choice of oxidant.
Indeed, even though it is more expensive than dioxygen, hydrogen peroxide is often the oxidant of
choice in fine chemicals synthesis because of its simplicity of operation. An example of its
commercial use is in the synthesis of a mixture of catechol and hydroquinone by the hydroxylation
of phenol (reaction 4) using, for example, the TS-1 catalyst (see chapter 10). In this case,
synthesis via the autoxidation of o-diisopropylbenzene is not feasible due to competing
intramolecular processes.
OH OH OH
0H+ w
6 ♦•*■ - ^ dr 6
OH
1 1
Reaction 4 is one example of a general type of oxidation process, referred to as catalytic oxygen
transfer,7 that is described by the general equation 5.
catalyst
S+X-O-Y ► SO + XY (5)
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
- Cheap oxidant (0 2 ) - Relatively cheap oxidants - Broad scope
- No effluent problem (e.g.H 2 0 2 ,NaOCl)
- Environmentally acceptable Disadvantages:
Disadvantages: - Broad scope - Expensive oxidants
- Limited scope (K 2 Cr 2 0 7 , KMn0 4 ,
(petrochemicals) Pb(OAc)4, etc.
attractive, much used alternative. It should be noted, however, that in metal-catalyzed oxidations
with peroxidic oxidants one still has to cope with the problem of competing homolytic and
heterolytic pathways.
- Attractive alternative with broad scope in - Only sporadically applied because of broad
fine chemicals scope of hydrogenation with H 2 |
Virtually all of the transition metals and several main group elements are known to
catalyze oxygen transfer processes. > A variety of single oxygen donors can be used (Table 3).
Next to price and ease of handling two important considerations which influence the choice of
oxidant are the nature of the coproduct and the percentage weight of available oxygen. The
former is obviously important in the context of environmental accepability and the latter bears \
directly on the productivity (kg product per unit reactor volume per unit time).
With these criteria in mind it is readily apparent that hydrogen peroxide is a choice
oxidant, its coproduct being water. In principle, it contains 47% active oxygen but in practice it is
generally used as a 30-35% aqueous solution which translates to 15% active oxygen.
The coproduct from organic oxidants, such as TBHP and amine oxides, is readily
recycled via reaction with hydrogen peroxide. The overall process produces water as the
coproduct, but requires one extra chemical step compared to the corresponding reaction with
hydrogen peroxide. With inorganic oxygen donors environmental considerations are relative.
Sodium chloride and potassium bisulfate are obviously preferable to heavy metal (Pb, Cr, Mn, etc)
salts. Generally speaking, inorganic oxidants are more difficult to recycle, in an economic manner,
than organic ones. Indeed, the ease of recycling may govern the choice of oxidant, e.g. NaOBr
may be preferred to NaOCl because NaBr can, in principle, be reoxidized with H 2 0 2 .
Heterolytic oxygen transfer processes can be divided into two categories based on the
nature of the active oxidant: an oxometal or a peroxometal species (figure 5). Generally speaking,
catalysis by early transition metals (Mo, W, V, Ti, etc) involves high-valent peroxometal
complexes whereas later transition metals (Mn, Fe, Ru), particularly first row elements (e.g. Cr),
mediate oxygen transfer via oxometal species. Some elements (e.g. vanadium) can, depending on
the substrate, operate via either mechanism. Although the pathways outlined in figure 5 pertain to
peroxidic reagents analogous schemes, involving M=0 or MOX (X=C10,10 4 , R3N, etc) species,
can be envisaged for other oxygen donors.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
HX
-> M - 0 2 R > MOR + SO
PEROXOMETAL PATHWAY
MX + R02H
OXOMETAL PATHWAY
S
ROH M-O > MX + SO
7. Olefln Oxidations
The most important example of a catalytic oxygen transfer process is undoubtedly the
metal-catalyzed epoxidation of olefins with alkyl hydroperoxides.8"11 The epoxidation of
propylene with TBHP or ethylbenzene hydroperoxide (EBHP), for example, accounts for more
than one million tons of annual, worldwide production of propylene oxide (reaction 6).
catalyst
CH3CH=CH2 + ROaH CH,CH — C H ' + ROH (6)
R = (CH3)3C- or PhCH(CH3)-
VI (Arco)
Catalyst : Homogeneous : Mo
Heterogeneous : Ti 1V / SI0 2 (Shell)
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
M —OR + RO t H + C==C
/*> RO A H
OR
>o> ir &,
i
R
I 1
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
o
M —OR + C—C M —OR + C—C ++ ROH
EPOXIDATION
OH
■^Xl ^
8. Alcohol Oxidations
HO
rr OH
TBHP
VO(acac) 2
HO
f^lf
O
(rel. 13) (7)
98% yield
TBHP
M l (ref. 14) (8)
CeIV/NAFK
98% yield
TBHP
RCH2OH■ ■—■■■- « ^pi RCHO (ref. 15) (»)
ZrO(acac) 2
85 - 95% yield
TBHP
RCH(OH)CN ► RCOCN (ref. 16) (10)
72 - 99% yield
In some cases the use of different oxygen donors with the same metal catalyst can lead to
dramatically different results, e.g.17
H202
C=C—C— (11)
OH
I [TiO(acac) 2 ]
c = c—c—
I
H \ O OH
TBHP / \ |
-► C —C — C — (12)
I
H
A possible explanation is that in the presence of water (i.e. with H2O2) epoxidation of the double
bond is seriously hampered.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
The alcohol oxidations outlined above can involve peroxometal or oxometal mechanisms.
Which of the two is operating can quite easily be ascertained by carrying out the reaction, with a
stoichiometric amount of the catalyst, in the absence of the oxygen donor. Oxidation is observed
under these conditions only with catalysts operating via an oxometal mechanism, e.g. vanadium:
H
RO ° >
\
(RO) 2 VOH+ C = 0 (13)
/
RO O
\
(RO)2H ; CHOH
-H 2 0
At this point it is also worth noting that TBHP has several advantages compared to other
peroxidic reagents. It has high thermal stability and is safer to handle than H2O2 or CH3CO3H. It
is non-corrosive and unreactive to most functional groups in the absence of catalysts. It is readily
soluble in nonpolar solvents, e.g. hydrocarbons, and reactions are carried out under neutral
conditions. Furthermore, the coproduct tert-butanol is readily removed by distillation.
Consequently, metal-catalyzed oxidations with TBHP have found wide applications in organic
synthesis1"3'*'9
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
OAc
-NH
30% AcOOH
►
NaOAc/HOAc/EtOAc
i (14)
94% yield
OSiMe2Bu* OSIMe,Bu
*2 fc
30% AcOOH
(15)
NaOAc/HOAc/EtOAc
NH
[5% Ru/C] ; RT O
(1'R , 3R , 4R)
99% yield
>99% e e
As noted earlier the oxidant of choice in the fine chemicals industry is often 30% aq.
H 2 0 2 Unfortunately, H 2 0 2 (in common with many other useful oxygen donors such as NaOCl)
is insoluble in many common organic solvents. One way of circumventing this problem is by the
application of phase transfer catalysis. This usually involves the transfer of the primary oxidant
(e.g. GO", S20g2") or the catalyst (e.g. Ru0 4 ", HM0O5"), as an anion of a quaternary ammonium
salt, to the organic phase. It has been used, for example, in the ruthenium-catalyzed cleavage of
olefins (reaction 16^)19 and the oxidation of substituted toluenes to the corresponding carboxylic
acids (reaction 17) 20 with NaOCl asthe primary oxidant.
NaOCl NaOH
CH 3 (CH 2 ) 1 2 CH=CH 2 ! *» CH3(CH2)12C02Na (16)
[Ru0 2 /Bu 4 NBr]
CH2CI2/HaO 100% yield
NaOCl , NaOH
ArCH 8 , ^ ArCOaNa (17)
[RuCI3/Bu4NBr]
CICH 2 CH 2 CI/H 2 0
98% yield
25°C , pH = 9
Probably even more interesting, from an industrial viewpoint, is the finding21 that
catalytic autoxidations of substituted toluenes can be improved by the application of phase
transfer catalysis. Autoxidations that normally employ a cobalt acetate/bromide catalyst in acetic
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
acid can be carried out with the neat hydrocarbon by employing a tetraalkylammonium
(phosphonium) bromide in combination with cobalt chloride. High selectivities were observed at
high substrate conversions.
0 2 ;[CoCL/QBr]
" ^ (18)
130-170 °C
ca. 10 bar
X = H , o - Me , m - Me , p - Me , p - Br ,
p - N0 2 , p - Ph , o - Ph , p - MeO
Q =
( io H 2i) 2 ( c H 3 ) 2 N ; (C6H13)4P , etc.
C
The first example of a successful catalytic epoxidation of olefins with H 2 0 2 under phase
transfer conditions was reported by Venturello and coworkers :
>=<♦«,<>, "• | W
W' o x
l V-VC
/ ~ \ C
(19)
90% yield
H 2 0 2 /t-BuOH
-#» (20)
QaPW12O40
a
so«c
OH
OH
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
N+-(CH2)18CH3
Heteropolyacids (HPAs) and their salts are polyoxo compounds incorporating anions
(heteropolyanions) having metal-oxygen octahedra (M0 6 ) as the basic structural unit. They
contain one or more heteroatoms, such as Si, Ge, P or As, that are usually located at the centre of
the anion The M 0 6 octahedra are linked together to form a thermally stable and compact
structure for the heteropolyanion. One of the most common types comprises the so-called Keggin
anions, XM^M 12-11O40 T n e associated cations may be protons or metal ions. Despite their
rather awesome formulae they are easily synthesized by acidification of aqueous solutions
containing the heteroelement and the appropriate mixture of alkali metal molybdate, tungstate and
vanadate.
HP As are both strong Bronsted acids and multielectron oxidants, i.e. they are potential
bifunctional catalysts. They are soluble in oxygen-containing organic solvents which means they
can be regarded as soluble oxides'. Furthermore, the heteropolyanion constitutes a multielectron
ligand that can stabilize reactive high-valent oxometal species, i.e. transition metal-substituted
HP As (polyoxometallates) can be regarded as oxidatively resistant analogues of
metalloporphyrins.
In the past, HPAs have been widely applied in heterogeneous gas phase reactions29 where
oxidations proceed via Mars-van Krevelen type mechanisms (see chapter 1 ). More recently they
have been increasingly applied to liquid phase oxidations . As noted in the preceding section
Mo-and W-based HPAs are used in combination with H 2 02 as the primary oxidant, under phase
transfer conditions, for a variety of oxidative transformations. However, a word of caution is in
order: in the presence of H 2 0 2 heteropolyanions may undergo degradation to less condensed
structures (see section 10).
Some HP As, such as H3PMoVI12_II V v n O 4 0 (PMoV-n) are strong oxidants in their own
right and catalyze the oxidation of organic substrates with dioxygen as the primary oxidant. For
example, 2-methylnaphthalene is oxidized to 2-methyl-l,4-naphthoquinone (menadione) by
dioxygen (3-8 bar) in the presence of 0,02-0.2 M PMoV-n in a HOAc-H 2 0-H 2 S0 4 solution at
120-140°C (reaction 21). Tradionally menadione (vitamin K3) was produced by oxidation of 2-
methylnaphthalene with stoichiometric amounts of chromium trioxide, a process that produces
18kg of chromium-containing solid waste per kg of product.
Oy
PMoV-n
+ 3 0, ► + H 2 0 (21)
H 2 0/HOAc
120-140 °C
3-8 bar 0
Conv. 78%
Sel. 82%
OH
[Fe-V]
+ MeOH (22)
gas phase
Sel. >90%
OH
[PMoV-n]
+ o„ ► H20 (23)
liquid phase
Sel. 83%
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
The same system has been used for the oxidation of 2,3,6-trimethylphenol to the
corresponding para-benzoquinone ° (reaction 24), which is an intermediate in the synthesis of
vitamin E, and for the oxidative coupling of 2,6-dialkylphenols to diphenoquinones (reaction
25).
OH 0
||
o s
+
o2
[PMoV-n]
aq.HOAc
►
v
A £ *H-°
V
(24)
O
Conv. 100%
Sel. 86%
OH
^ RR [PMoV-n] R/ r ^ /=(
R
D +
o2
H20
0 = { \ = { f = 0
/ \
* 2H20 (25)
25-50 °C
R R
1-5 bar
R = Me ,t-Bu 100% yield
Evidence has been presented26 to support a mechanism for these phenol oxidations in
which V(>2 + , formed via dissociation of the PMoV-n, is the active oxidant. The heteropolyanion
is required, however, to facilitate reoxidation of vanadium (IV) by dioxygen.
PMoV-n has also been used in combination with palladium(II) and dioxygen for the
Wacker-type oxidation of olefins to carbonyl compounds, e.g. ethylene to
acetaldehyde26,31(reaction 26). In the classical Wacker process catalyst a copper chloride
cocatalyst is used to mediate the reoxidation of the reduced palladium to palladium(II). In this
alternative process the heteropolyanion performs this task. A major advantage of the alternative
process is that it eliminates >99% of the formation of chlorinated byproducts.
Pd n /PMoV-n
H2C = C H 2 + 1 / 2 0 2 -** CH8CHO (26)
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
PMoV-2 and related HPAs have also been extensively applied by Bregeault and
coworkers to the oxidative cleavage of vicanal diols (reaction 27) and ketones3 (reaction 28).
The high selectivities observed at high conversions, coupled with the mild reaction conditions and
inexpensive oxidant, would seem to make this method highly attractive for use in fine chemicals
manufacture.
a" OH
0 2 ; [PMoV-2]
EtOH ; 75 °C
f^^C02Et
k^C02Et
90% selectivity
62% conversion
(27)
0 2 ; [PMoV-2]
20 °C / 1 bar 6
90% selectivity
(28)
96% conversion
From the above selected examples it is apparent that HPA-catalyzed oxidations with clean
oxidants, such as 0 2 and H 2 02, have considerable synthetic potential. They may be considered as
a bridge between heterogeneous gas-phase oxidations and liquid phase homogeneous oxidations.
It should be pointed out, however, that the heterogeneous redox molecular sieve catalysts such as
titanium silicalite (TS-1) and related catalysts (see chapter 10) catalyze many of the same
reactions. The latter catalysts have the advantage of high stability and ease of recovery and
recycling.
One of the most challenging goals in catalysis is the design of simple abiological catalysts
that can achieve high levels of enantioselectivity. With this goal in mind much effort has been
devoted to the development of relatively simple metal catalyst/chiral ligand/oxygen donor
combinations capable of mimicking nature's selective and versatile oxidation catalysts, the
monooxygenases. Perhaps the most well-known of these is the titanium(IV)/dialkyltartrate/TBHP
system developed by Sharpless and coworkers . This system is very effective for the catalytic
asymmetric epoxidation of allylic alcohols (reaction 29).
TBHP R1 Rs
V */ OH (29)
[Ti(OPr')4/L'] R2 O
CH 2 CI, 20 °C
mol. sieve
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
The secondary interaction of the hydroxyl group in the substrate with the catalyst is crucial
for achieving high enantioselectivities and the Sharpless reagent is not effective with simple,
unfunctionalized olefins Some examples of the high enantiomeric excesses (ee) obtained with
various allylic alcohols are shown in Table 4.
Substrate Yield
a. 5 mol % catalyst
More recently, Jacobson35 and Katsuki36 have independently developed manganese (IE)
complexes of chiral Schiffs bases as catalysts for the enantioselective epoxidation of
unfunctionalized olefins using NaOCl or PhlO as the oxygen donor. Subsequentfinetuning of the
chiral ligand structure afforded highly effective catalysts (e.g. complexes A and B) for the
enantioselective epoxidation (reaction 30) of a range of olefins (see Table 5)37
From the viewpoint of commercial applications the stability of these relatively expensive
chiral ligands towards oxidative conditions is of crucial importance. Reported37 catalyst turnovers
are of the order of 25-100 and this needs to be increased by a factor of 10 or even 100 to obtain
attractive economics.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
R1 o
K aq. NaOCI
CH2CI2 ; 0 *C
catalyst (0.5 m %)
R2 R^
(30)
o/
R t . B uHD- ro
Bu* t-Bu
A (R,R) B (S,S)
R = CH 3 , t-Bu
The Sharpless group 3 8 , 3 9 has also developed a highly effective system for the asymmetric
vicinal dihydroxylation of olefins (reaction 31). The system comprises an O s 0 4 catalyst
incombination with dihydroquinine or dihydroquinidine esters or related chiral ligands and N-
methylmorpholine-N-oxide (NMO) or potassium ferricyanide, K 3 Fe(CN) 6 , as the primary
oxidant
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Ph J A 0.04 84 92
0.01 80 88
0.04 67 88
^s
0.02 87 98
OO
Ph COaEt
0.15 63 94
B 0.08 67 97
In this case secondary interaction with a coordinating functional group is not essential and
the reaction is successful with a broad range of olefin substrates. The efficacy of the Sharpless
asymmetric epoxidation (AE) and asymmetric dihydroxylation (AD) methods are compared in
Table 6. It is readily apparent that asymmetric dihydroxylation has a much broader scope.
R
' R3
[OsQ4/L'] OH
R2
70-95% yield
20-95% ee
or
MeO
pMe
R = Cl NMO = N-methylmorpholine
- N-oxide
M + L* M - L* (32)
i
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Table 6. Comparisn of asymmetric epoxidation (AE) and asymmetric dihydroxylation (AD) of various olefms."
80% ee >95% ee
P K ^ ^ O H
.. OH
>95% ee 30-50% ee
Ph ^^^ ^ ^
P H ^ ^ ^ X
X - OAc , OCH,Ph , N, , Cl >95% ee NR
OCH,
>95% ee NR
Ph ^ ^ OCH3
>95% ee NR
Ph ^^ OCH3
0
„,/vA NR . >95% ee NR
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
R 1 HO
V-< OH
R - ^
,0s
5 ^ I ^0 XI H20, L*
0
/ 0 s ^
I 0
high "O"
enantioselectivity
low
enantioselectivity
VsW"
R
O^l «' -R
electrocatalytic version has been developed in which the enantiomerically pure diols are formed
from the olefin, water and electricity, with hydrogen gas as the only coproduct .
This is an appropriate note on which to end our discussions of the application of clean
catalytic technologies to the synthesis of fine chemicals. It is obviously an area eith a golden
future.
References
ABSTRACT
At the present time, electrochemical reactions do not constitute an important part
of industrial organic syntheses. Nevertheless, the future of organic electrosyntheses
is usually considered to be very bright, especially in the production of fine and spe
cialist chemicals. Reasons for this optimism, which include environmental benignity,
are discussed, as are the problematic aspects — expensive separation operations, for
example.
The following subjects are addressed: (i) electrochemical regeneration of highly
selective redox couples whose stoichiometric application is ruled out for economic
and/or environmental reasons — regeneration can be carried out ex situ, in situ, or
throughfixingthe couple onto an electrode — the synthesis of quinones being among
the successful applications; (ii) oxygen-transfer anodes, such as Pb02-epoxidation is
an important topic here, as it is in (i); (iii) hydrogen abstraction anodes, such as
NiOOH, whose action is very similar to the chemical reagent nickel peroxide, an
important application of which is the conversion of a sorbose to its corresponding
acid — a step in vitamin C synthesis; (iv) Pt anodes, which are o.a. applied in
the classical Kolbe synthesis; (v) heterogeneous catalysis of consecutive reactions
following the initial charge-transfer step.
Finally, attention will be paid to new developments in electrode materials —
polymer-modified electrodes, for example — and to the recent and promising devel
opment of solid polymer electrolyte cells.
1. Introduction
highly selective redox systems whose stoichiometric application is ruled out for eco
nomic or environmental reasons. First, however, let us summarise the pros and cons
of electrochemical synthesis routes in general.
From the point of view of process engineering, electrochemical reactions offer
a number of advantages:
- they are usually easily controlled by means of additional parameters such as
current density and charge;
- they take place under mild reaction conditions (low temperatures, atmospheric
pressure) which often implies that relatively inexpensive equipment can be
used;
- they are environmentally benign in that they do not, in general, entail any waste
air or waste water problems. This aspect will become even more important in
the future. Competition, in this case, will come, of course, from catalytic
processes.
Offsetting these advantages are a number of problem areas:
- electroorganic syntheses require, in general, special reactors;
- because such syntheses are phase-boundary reactions, the associated prob
lems may be expected to occur; e.g. electrode deactivation, poisoning and/or
corrosion;
- the necessity of using electrolytesfrequentlyresults in expensive separation
operations during work up of the electrolysis mixtures. It is, in fact, this aspect
that renders many electroorganic syntheses uneconomical;
- there is very little possibility of extrapolating the experience gained with one
process to the scaling up of new reactions.
It is also to be noted that small processes, such as those producing fine
and specialist chemicals, do not usually warrant the development of specific cells,
electrode materials, or membranes. Therefore, the process designer will have to select
the cell and its components from those already available but probably optimized for
other purposes. On the bright side, a number of cells are now commercially available.
Firms active in thisfieldinclude ICI, Electrocell AB, Reilly Thr and Chemical, Steetley
Engineering and Electrocatalytic. This reflects the increasing confidence that in the
fine-chemicals area, electrochemical processes have an important role to play. Water
and methanol are preferred solvents in that they are cheap, give rise to solutions of
high conductivity and clean counterelectrode reactions (H2 evolution in the case of
electrooxidations). However, they are not always the ideal medium for the organic
reaction. Where anodic and cathodic processes interfere with each other, a membrane
is used to separate the two electrodes (adding to the resistance of the cell).
As to electrode selection, in the first instance one chooses from a restricted
range of readily available materials, e.g. steel, Pb0 2 , Pt (usually as a coating on Ti),
diverse forms of C (carbon black, graphite, glassy carbon, etc.) or Ru02-coated Ti02.
The choice of the right one is quite empirical as it is thus far veiy difficult, if not
impossible, to predict the success of aln electrode material or to define its lifetime
without extended studies under realistic process conditions. Accelerated testing is
rarely satisfactory except to indicate catastrophic failure.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
The present subject can not profitably be discussed from the point of view of
electrocatalysis: although adsorbed species are, every now and then, considered to be
the key, very little is really known in this area (cf. Section 2.3). The electrical double
layer which exists at the electrode/solution interface, on the other hand, would appear
to play only a minor role in determining the rates of electron transfer.
What follows is essentially based on several recent review papers *~8 and aims
to give an impression of what is electrochemically possible in the area of selective
oxidation. The emphasis is on industrial application and, whenever possible, on the
electrocatalytic aspects. We will first discuss some aspects of the use of regenerable
redox couples and then focus our attention on oxygen-transfer anodes and on Pt
and C anodes. Finally, some recent developments in electrode materials and the
emergence of solid-polymer electrolyte cells will be discussed.
2. Selected topics
Membrane
cell
Reactions carried
out solid to solid,
chromatography
Pb0 2 style
anode
The spent redox reagent is then passed through an activated carbon bed to remove
traces of organic material which would otherwise lead to loss of current efficiency
(increase in the relative rate of oxygen evolution, the principal side reaction) and
the Cr3"1" solution is recycled to the cell. This technique has, however, become less
important since the development of catalytic methods (Cr03, liquid phase, 50-150°C;
or Iron vanadate, gas phase, 34G-390°C9).
The oxidation of naphthalenes to yield naphthoquinones (used, for example,
as a pulping additive) would be technically interesting. In addition to chromate re
generation, Ce(lV) oxidation and regeneration were, in particular, investigated. The
best results (naphthoquinone selectivity: >95%, current efficiency for the regenera
tion: 95%) were obtained using an ex-cell process and cocatalysts (Ag + , Co 2+ ) in the
electrochemical regeneration. Because of the poor solubility of the cerium salts, very
large reaction volumes are required: it is necessary to apply and regenerate at least
1001 of electrolyte solution per kg of naphthoquinone produced 4. This has triggered
a search for methods to increase the solubility which resulted in the development of
a Ce(lV) procedure based on aqueous methanesulfonic add at WR Grace16a. This
technology has been scaled up by HyflroQuebec to the 100 ton/year scale and it is
to be licensed to Ikysung Enterprises Co (Ikiwan) for the manufacture of a quinone
intermediate for a dyestuff16b.
The oxidation of benzene to benzoquinone, an important target in industrial
chemistry, has also been attempted but the results obtained with indirect processes
were inferior to those obtained by direct electrochemical oxidation (see Section 2.2).
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
The core of the new synthesis is the in-cell regeneration of the chemical oxidant
Mn(OAc)3 in the presence of catalytic amounts of Cu(II) salts. However, although
the starting materials butadiene and acetic acid are inexpensive, the cost of the
process is still high due to high conductive salt concentrations coupled with low
concentrations of the desired product (2-4%) and considerable corrosion problems.
Thus, the work-up procedure is rather expensive 4.
Ethylene can be oxidized to acetaldehyde in high yields similar to the Wacker
process if electrogenerated palladium(II) is used as catalyst. In this way the copper(II)
catalyzed air oxidation of palladium(O) is replaced by its electrooxidation2.
As an interesting detail, we note here that total oxidation of organic waste —
to CO and CO2 — has been recently reported to take place when using Ag 2+ /Ag + as
the in-cell redox reagent29.
Thus far, we have discussed only cationic redox couples. However, one can also
apply anionic ones, of which Br~/Br2 is the best known example. It is, for example,
applied in the oxidation of furans and aldoses.
Technically important is the electrochemical methoxylation of furans to yield
the corresponding dimethoxydihydrofurans. It is now carried out industrially by BASF
and Otsuka:
/T~\ CH OH-NoBr J==\
0 C anode *~ CH30 ^ 0 ^ 0 C H 3
Conversion: 8 0 %
Selectivity: 96%>
This very elegant reaction was also used on the laboratory scale for the synthesis of
biocides, cyclopentenones and prostaglandin intermediates. Although the bromide
ion plays an essential role here, the mechanism is not as yet completely elucidated.
However, bromine and brominated furans are probably important intermediates in
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
HgO-NaBr
Pt anode
Selectivity: 8 0 %
Current efficiency: 5 0 %
There is, however, no industrial application for this process. It has proven im
possible to suppress the formation of the 1,2-dihalopropane byproduct and, because
of the low concentrations of desired product (2-4% of PO in the electrolyte) and the
presence of numerous byproducts, the work-up procedure is complicated. The process
is thus rendered uneconomical. However, an alternative which combines chloralkali
electrolysis with the chlorohydrin process is still being pursued 4. The electrosynthesis
of hypochlorites has been studied in detail by Olin and Steetley Engineering13.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
The indirect electrochemical epoxidation of C3-C5 olefins with the help of an-
odically generated [(Py)2Ag(III)-—0]~ species has been studied at Shell14. Although
high current yields and efficiencies were observed, it is, at present, not industrially
applied. The principle of the indirect electrochemical epoxidation can also be applied
to complicated structures. For example, ICI utilised this reaction for the preparation
of intermediates for the synthesis of fungicides:
Y-/>-C=CH H20-PMF-MBP ^ Y H f ~ V Q
x-/ CH *
2
\=/ 1 \ = / ra
R R
X,Y: F, Cl Conversion: >90%>
R: atkyl, H Selectivity: > 9 0 %
Current efficiency: about 40°/o
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Fig. 2. "Swiss-roll" cell — arrangement of the electrodes: 1 = steel net cathode; 2,4 = polypropylene
net as insulating separator; 3 = nickel net anode; 5 = current feeder.
rate-determining step; (d) further oxidation of the radical formed in step (c):
hazard from explosion due to the mixing of the hydrogen gas formed at the cathode
with the oxygen gas evolved as a by-product at the anode 6«14. It is said that Merck is
using the reaction in the industrial production of vitamin C.
The nickel oxide electrode is widely applied in laboratory electrosynthe-
ses 3>4'17'18. It should also be mentioned that the anodic oxidation of aliphatic amines
to nitriles at nickel (hydr)oxide electrodes may be economically preferable to con
ventional chemical processes. It is worth noting that very similar chemistry can be
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
achieved at oxide-covered silver, cobalt and copper, and monel anodes in alkaline
solutions. An increase in electrode stability appears to be achievable in the case of
cobalt/nickel spinel electrodes7.
Attempts to develop anodes that act by heterogeneous redox catalysis and are
stable in acid media have met with little! success to date, despite the efforts of Beck et
al.8'19. Ceramic metal oxide electrodes have attracted the most attention since they
ideally combine the porosity of the oxide layers and the high reactivity of surface
groups. High turn-over numbers can usually be achieved, but the operating life of the
electrode is limited by leakage of redox metal into the acid solution.
Recently, reports have appeared on the high surface-area ternary ruthenates,
A2RU2O7 (A = Pb, Bi), that are effective in the electrocatalytic oxidation of or
ganic substrates in strongly alkaline media, e.g. sec. alcohols to ketones and alkenes
and vicinal diols to carboxylic acids20* e.g. Rx—CH=CH—R2 -» Ri—COOH +
R2-COOH.
It could be shown21 that activity for the latter reaction is associated with the
presence of a Ru(V) surface state, capable of being oxidised to Ru(VI) at higher
potentials. Substrates are oxidized by Ru(VI), which is continually regenerated if the
potential is kept high enough. This is another case of heterogeneous redox catalysis.
As in the case discussed above, operating life is limited by leakage of the active
species:
Ru(V) = Ru(VI) + e
Ru(VI) + S = Ru(V) + products
side reaction: Ru(VI) = (RuOj"")^
It appears, that reoxidation of Ru(V) can also be effected by molecular 0 2 , enabling
the oxidation reaction to be carried out in a trickle-bed reactor 22.
There are a number of anode reactions that involve the introduction of oxygen
into the electroactive species and that are almost exclusive to lead dioxide as an anode
material. These include both inorganic and organic reactions. The most extensive
use of Pb0 2 is as an anode for the oxidation of chlorate to perchlorate. It is also
used for the conversion of chromium(IH) to chromic acid (cf. Section 2.1.1) and of
manganese(II) to permanganate where Pb0 2 generally gives better selectivity than
Pt/Ti (although the latter can be improved by using silver(II) ions as cocataiyst2).
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
277
Some examples of the use of lead dioxide for the oxidation of organic compounds will
be given below. Typically, the current efficiencies here are 50-100% (some oxygen is
also evolved) but no more than a few percent at electron-transfer electrodes such as
platinum.
Although it is not implied that all reactions at Pb0 2 occur by such mechanisms
(nor that they cannot occur elsewhere) it is tempting to propose that it is possible for
the lead dioxide surface to transfer oxygen atoms to appropriate acceptors, i.e.7
Pb0 2 + X - R = 0 = X - R + PbO
PbO + H 2 0 = Pb0 2 + 2e + 2H+
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
further testing for every new process.
An electroorganic reaction that is carried out preferably with lead dioxide
anodes is the oxidation of benzene rings to quinones (cf. the indirect method discussed
above). Considerable efforts were made to develop the electrochemical oxidation
of benzene to p-benzoquinone to the industrial scale thus forming a basis for a
new hydroquinone process. The electrochemical oxidation of benzene in aqueous
emulsions containing sulfuric acid using divided cells and Pb0 2 anodes forms p-
benzoquinone The product can then be reduced cathodically at a lead electrode to
yield hydroquinone in a paired synthesis.
The process has, so far been unable to compete successfully with the H202
oxidation of phenol or the Hock Process (air oxidation of p-diisopropylbenzene9).
The reasons for this are the poor benzene conversion, the low quinone concentration
in the organic phase (<6%), the complicated work-up procedure (again!) and the
unsatisfactory electrode lifetimes. Current densities have to be kept low since, other
wise, quinone will be consumed by further oxidation to maleic acid. Further losses in
current yield are then experienced because an increasing percentage of the current
goes into oxygen evolution. Some improvements appear to have been made by Dow
through the use of a special porous electrode made of Pb0 2 and PTFE 4'23.
This type of reaction can also be used for the production of substituted
quinones and hydroquinones. For example, BASF has developed two lab processes
for the synthesis of trimethyl-benzoquinone and trimethylhydroquinone. The latter
is required for the synthesis of vitamin E. It is noted, in passing, that the synthe
sis of (substituted) /?-benzoquinone can be approached differently (Hoechst4). By
oxidation in methanol solution at glassy carbon electrodes and in the presence of
tetraalkylammonium fluorides as conductive salts, p-benzoquinone tetramethyl ketal
is formed. This can be converted to p-benzoquinone in a simple procedure while
methanol can be recycled.
Whereas the indirect process of the production of propylene oxide has not
advanced beyond the stage of experimental production, the direct electrosynthe-
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
278
H2 H2S 4
C\ °" ° > C \
Yield: 8 0 %
Current efficiency: 67°/o
Very recently, Johnson et al.26 have described efforts to improve the rate of
oxygen-transfer reactions at lead dioxide by including catalytic amounts of group Ilia
and Va metals, especially Bi, in the oxide coating. The idea is to incorporate a material
with a low oxygen-evolution overpotential in the Pb0 2 surface in such a way that the
catalytic sites are well separated from each other. They will, therefore, generate the
active 0(H)-radi$al species with a, nonetheless, relatively low yield of dioxygen. Thus,
DMSO can be smoothly oxidized to dimethylsulfone on Pb0 2 : Bi, but not on Pb0 2
itself. As is the case with heterogeneous redox electrodes, however, stability of the
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
R—R
(a) ^ R-
R—Cof ~^r R-
-C02
-e (b)
were electrolysed instead. This Brown-Walker version of the Kolbe reaction is, as yet,
the most important onefroman industrial point of view.
Classical Kolbe synthesis is carried out at platinum anodes but various forms
of carbon can be used as well. Depending on which species is strongly adsorbed,
the carboxylate anion (Pt, vitreous carbon) or the alkyl cation (graphite), the Kolbe
reaction yields different products: e.g. ethane resp. methylacetate in the electrolysis
of potassium acetate, pathways (a) resp. (b) in Fig. 31»7'24'25. It is worth noting that
strong adsorption of either carboxylate or alkyl species is exactly why the Kolbe
reaction works at all: otherwise we would only observe the (under the prevailing
conditions thermodynamically strongly favoured) evolution of dioxygen.
The anodic oxidation of adipic half esters to the corresponding sebacic acid
diesters has been studied in extensive detail:
ft \-NHCOCOOH
CH3CN - C H 2 C l 2 - (C 2 H 5 ) 4 NOH
o~
At 2F/mole:
:C = 0
H 0 H H - C 0 0 H
C H Conode" HO-QK-CHO
OH
R:H, 0CH3, or OCjHs
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
self-coupling at C but not at Pt anodes.
There are many other examples of reactions that show a strong dependence
on electrode material. This is believed to be caused by the effects of adsorption, but
we will limit ourselves to only two of them7. The hydroxylation of tetrahydrofuran in
aqueous acid
+ Ha 2e_
O ° - —~ C X OH
+ 2HT
gives yields that vary strongly with the anode. On Pt, for example, the selectivity
is 95% and current efficiency is 70% under conditions where it may be shown that
THF adsorbs strongly on the platinum oxide surface. On Pb0 2 , on the other hand,
the major product is butyrolactone. In the acetoxylation of mesitylene, very different
ratios of nuclear to side-chain acetoxylation were observed for different electrodes:
from 3.6 at Au to 23 at graphite.
CH3 9H3 CH2QAC
-2e~ + OAc" ►
C H ^ X ^ ^ C H ;3
It was pointed out above (Section 2.3.1) that the Kolbe electrolysis is successful
only because strong reactant adsorption suppresses the evolution of oxygen. A similar
effect has been noted in the anodic conversion of cyclohexene in the presence of
chloride ions (example cited by Beck28). At high potentials, the formation of 3-
chloro-cyclohexene, rather than the evolution of Cl2,is observed.
In fuel-cell electrode reactions, of course, adsorption of fuel molecules and
the activation of water molecules are of prime importance. However, for a discussion
of this topic the reader is referred to the appropriate chapter.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
The anodic oxidation of olefins in the presence of nucleophiles, such as
CH3OH or CH3COOH is, in principle, a reaction of great industrial interest since
it permits allyl oxidation as well as C—C coupling4. Nevertheless, it is almost never
used in industry today. This is essentially because the selectivities are frequently
poor4. Over the past few years, the reaction principle has been used in synthesis
problems in the area of fine chemicals. For example, the anodic methoxylation of
citronellol is a key step in a new rose oxide synthesis by Sumitomo. Also, Kuraray
used the addition of anodically generated radicals of 1,3-dicarbonyl compounds for
C—C coupling in the preparation of intermediates for j8-blockers.
BASF developed a process for the preparation of 2,5-dimethoxy-2,5-dihydro-
furan from butene-l,4-diol, which is an alternative to the anodic methoxylation of
furan (see above). The anodic acetoxylation of olefinic terpenes was used for the syn
thesis of new fragrances and for the intermediates of canthaxanthin. These reactions,
with the exception of that by BASF, have not progressed beyond the laboratory stage 4.
Some examples of the anodic functionalization of aromatics have already been
presented above. Here we simply note that the electrochemical oxidation of substi
tuted phenols has been used on the laboratory scale for the production of specialties
such as antioxidants. Another reaction of interest is the nuclear acyloxylation of aro
matics, opening up a new synthetic route to phenols. Much work has been done on
the synthesis of naphthyl acetate. The principal problem with this reaction was the
large amount of conductive salts which had to be separated and recycled. Another
problem was the formation of methylnaphthalenes and their acetoxylation products
by Kolbe electrolysis of the solvent acetic acid. The disadvantages have been sub
stantially overcome by using distillable conductive salts and conductive polymers as
electrodes:
OAc Conversion of naphthalene: 20-50°/©
^ ^ JL Selectivity for naphthyl acetate: 7 0 - 8 5 %
HOAC-(CH3)3NHOAC | IT |1 Current efficiency: 45-65°/o
C-poiypropyiene anode k s ^ ^ J ^ s . J^ /3-naphthyl acetate fraction: 3 - 5 %
Undivided cell
RH —► RHS - R
CN
+ +
-H -H —e
CN" '
^RCN
~©- AcOH
Optimum reaction conditions for thefirst,acetoxylation, step vary with X; some of the
tricks used include: use of quaternary ammonium salts as supporting electrolytes, ad
dition of metal salts (e.g. Co and Cu(OAc)2), application of phase-transfer conditions.
If defined amounts of water are added to the electrolyte, the anodic acetoxylation
yields the corresponding aldehydes with very good selectivities.
Using methanol instead of acetic acid makes the reaction go through the
dimethyl acetal stage. If the phenolic group is protected, it is also possible to obtain
p-hydroxybenzaldehyde derivatives. Ibluene itself, however, cannot be oxidized to
benzaldehyde dimethyl acetal under similar conditions: its oxidation to benzaldehyde
can, on the other hand, be effected via indirect electrosynthesis using Ce4+.
Anodic cyanations which can occur at the aromatic nucleus and on the side
chain, and which can result both in nitriles and in isocyanides, are very effective meth
ods for carbon-carbon and carbon-nitrogen bond formations. This is schematically
shown in Fig. 4.
Some examples of the anodic oxidation of heterocyclic compounds have al
ready been presented above. Anodic oxidation of aliphatic ethers, both cyclic and
non-cyclic, is preferably carried out on glassy carbon electrodes (higher current effi
ciencies). In methanol and with (CH3)4NS04CH3 as conducting salt, it is possible to
convert THF into 2-methoxy THF and dioxane in methoxydioxane in 80% yields with
current efficiencies of 70-75%.
In the anodic oxidation of sulfur compounds, industrial work has been concen
trating on the search for alternative processes for the production of tetraalkylthiuram
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
disulfides4:
C H C H
3^ n 3^ II II /CH 3 Conversion: 1 0 - 2 5 %
^N-^-SNa J i ^ ^ \ 4 - 5 - S - L N 7 Selectivity: 9 5 %
CH3"^ Pt anode C H / ^CH 3 Current efficiency: 88%
Their synthesis can also be carried out as a two-phase electrolysis using ammonium
salts. Sulfenamides can be produced by oxidizing tetraalkylthiuram disulfides in the
presence of amines:
s s s
II || R 1 ,NH/DMF/LiCl0 4 II ,
R9N-C-S-S-C-NR2 ► R2N-C-SNR'
* * Pt anode * *
The last reaction type of this miscellany concerns indirect organic oxidation
involving an electrogenerated superoxide anion radical^ Molecular oxygen can be
electroreduced in aprotic media to superoxide ion, 02". The superoxide ion can
abstract protons from even very weakly acidic substances and, thus, can act as an
electrogenerated base. In the presence of H 2 0 it decomposes rapidly. In a recent
application of this method32, 4-(Di-/i-propylsulfamyl) toluene is converted in DMF
to an important drug, 4-(Di-n-propylsulfamyl)benzoic acid. The advantage of the 0{
method here is that the sulfamido group remains entirely intact.
It is to be emphasized that the above is only a small selection from the existing
literature and some reaction types, such as electrochemical halogenation, although
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
not unimportant industrially, have not been discussed at all. For further information,
we direct the reader to some excellent monographs28,33""35.
ever, polymer-coated electrodes have not fared much better and, indeed, it appears
that experimental procedures for asymmetric electrochemical reactions at modified
electrodes are, as yet, not straightforward enough to allow a general application
(reproducibility problems).
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
285
CH3
CH3CN-Lia-Et4NBF4 f^^l ( ^ V C l
Conversion: about 5 0 %
■ u u ii Selectivity monochlorotoluene:99%
Cl
The use of graphite anodes modified with cyclodextrins allows the increase of the
p/o ratio to above 4 and, at the same time, the use of aqueous electrolytes (NaCl,
HC1-H 2 0).
CO + 2ROH -► R O - C ( 0 ) - O R + 2H+ + 2e
In homogeneous systems, various transition-metal porphyrins have been used
to epoxidize olefins or hydroxylate alkanes. In these reactions there is stoichiometric
conversion of the added O-donor. Recently, some papers have appeared that describe
possible ways to carry out the above reactions electrochemically.
The electrochemical epoxidation of olefins with Mn meso-tetraphenylpor-
phyrin catalyst and H2O2 generation at polymer-coated electrodes has been reported
by Nishihara et al.42 and is schematically shown in Fig. 5. At high concentration of
olefin and porphyrin, the electrocatalytic reaction runs at nearly 100% current effi
ciency for production of cyclooctene oxide. Porphyrin stability remains a problem,
however. Electrocatalytic hydroxylation of alkanes with iron 2,6-difluorotetraphenyl-
porphyrin in dichloromethane, with water from added hydrated fluoride salt providing
the oxygen source, is reported in Ref. 44. A very interesting electrocatalytic oxygena-
tion of several hydrocarbons has been accomplished with a manganese porphyrin/
periodate system under conditions of phase-transfer catalysis. Epoxides with up to 90%
yields were obtained. Unactivated alkanes were hydroxylated, giving yields between 25
and 77%. The periodate was regenerated electrolytically in the aqueous phase45. At
the time, this system was said to possess potentialities for larger-scale applications.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Scheme 1
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
3. Concluding remarks
References
1. H. Wendt, Electrocatalysis in organic electrochemistry, Electrochim. Acta 29 (1984) 1513.
2. E. Steckhan, Organic syntheses with electrochemically regenerable redox systems, Topics in
Current Chemistry 143 (1987) 1.
3. H.-J. Schafer, Oxidation of organic compounds at the nickel hydroxide electrode, Topics in
Current Chemistry 143 (1987) 101.
4. D. Degner, Organic electrosyntheses in industry, Topics in Current Chemistry 148 (1988) 1.
5. A. Merz, Chemically modified electrodes, Topics in Current Chemistry 152 (1990) 49.
6. A. Pletcher and EC. Walsh, Industrial Electrochemistry 2nd ed. (Chapman and Hall, 1990).
7. A.M. Couper, D. Pletcher and EC. Walsh, Electrode materials for electrosynthesis, Chem. Rev.
90 (1990) 837.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
8. F. Beck, B. Wermeckes and E. Zimmer, Kann eine Elektrode die Rolle eines Redoxsystems in
der Losung ubernehmen?, Dechema-Monographien, Band 112 (VCM Verlagsgesellschaft, 1988)
p. 257.
9. K. Weissermel and H.-J. Arpe, Industrielle Organische Ckemie (Verlag Chemie, 1976).
10. K. Otsuka et al., Electrochim Ada 34 (1989) 1485; Chem. Lett. (1990) 509; Nature 345 (1990)
697.
11. Catatytica Highlights, 16,No. 3 (1990) 3.
12. E Frusteri, A. Iannibello, A. Parmaliana, A. Cannizarro and N. Giordani, in: New Developments
in selective Oxidation (G. Centi and F. THfiro, Eds.) (Elsevier 1990) p. 733; A. Parmaliana et al.,
paper presented at the 11th NAM Catalysis Soc., Dearborn, May 1989
13. D. Hughes, The Chemical Engineer, September (1987) p. 17.
14. J.M. van der Eijk, Th.J. Peters, N. de Wit and H.A. Colijn, CataL Today 3 (1988) 259.
15. K.-H. Grosse Brinkhaus, E. Steckhan and D. Degner, Tetrahedron 42 (1986) 553.
16. (a) R.P. Kreh, R.M. Spotnitz and J.T. Lundquist, /. Org. Chem. 54 (1989) 1526;
(b) Electrochem. Processing—A Quarterly Commentary, ICI Appl. Electrochem., August 1993.
17. P. Verbragge,J.de Waal and D.Sopher,feP 199413 (1987).
18. J.Kaulen and HJ.Schafer, Tetrahedron J8 (1982) 3299.
19. H. Schulz and F. Beck,^**. Chem 97 (1985) 1047 (Int. Ed. Engl. 24 ~ 1049).
20. H.H. Horowitz et al, US Patent 4203871 (1980), Eur. Pat. Appl. 813.0574.8 (1971), US Patent
4434031 (1984); H.H. Horowitz, H.S. Horowitz and J.M. Longo, in: Electrocatafysis, Proceedings,
VoL 82-2 (The Electrochemical Society, 1982) p. 285; H.S. Horowitz, J.M. Longo and H.H.
Horowtz,/. Electrochem Soc. 130 (1983) 1851; H.S. Horowitz, J.M. Longo, H.H. Horowitz and
J.T. Lewandowski,^4C5 Symp. Ser. 279 (1985) 143.
21. J.A.R. van Veen, J.M. van der Eijk, R. de Ruiter and S. Huizinga, Electrochim. Acta 33 (1988)
51.
22. T.R.Fe\thoust,J.Am. Chem. Soc. 109(1987)7566.
23. D. Danly, in: Emerging Opportunities for Electroorganic Processes (Marcel Dekker, 1984) p. 229.
24. H.-J.Schafer,Ref.5,p.91.
25. MRJ.BrennanandR.Brettle,/. Chem Soc.,Perkin Trans.I(1973)257.
26. D.C. Johnson, J.A. Polta, T.Z. Polta, G.C. Neuburger, J. Johnson, A.P.-C Tang, I.-H. Yeo and J.
Bam, J. Chem Soc, Faraday Trans. 782 (1986) 1081; I.-H. Yeo and D.C. Johnson,/. Electrochem
Soc 134 (1987) 1973; H. Chang, Ph.D. Thesis, Iowa State University (1989); I.-H. Yeo, S. Kim,
R. Jacobson and D.C. Johnson,/. Electrochem Soc. 136 (1989) 1395.
27. J.M. BobbitMCS Symp. Ser. 390 (1989) 176.
28. F. Beck, Elektroorganische Chemie (VCH Verlagsgesellschaft, 1974).
29. D.F. Steele, D. Richardson, J.D. Campbell, D.R. Craig and J.D. Quinn, Transl Chem E 68
(1990) 115.
30. D.K. Kyriacou and D.A. Jannakoudakis, Electrocatafysis for Organic Synthesis (John Wiley and
Sons, 1986).
31. New Horizons in Electrochemical Science and Technology (National Academy Press, 1986).
32. M. Michman and M. Weiss, Ref. 12, p. 667.
33. S. Torii, Electroorganic Synthesis. Part 1: Oxidations (VCH Verlagsgesellschaft, 1985).
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
34. K. Yoshida, Electrooxidation in Organic Cfiemistry (John Wiley and Sons, 1984).
35. M.M. Baizer and H. Lund, Organic Electrochemistry, 2nd ed. (Marcel Dekker, 1985).
36. J. Heinze, Ref. 5, p. 1.
37. J.A.R. van Veen and J. F. van Baar, Rev. Inorg. Chem. 4 (1982) 293.
38. B. van Wingerden, J.A.R. van Veen and C.T.J. Mensch, /. Chem Soc, Faraday Trans. 184 (1988)
65; J.A.R. van Veen, H.A. Colijn and J.F. van Baar, Electrochim. Acta 33 (1988) 801.
39. D. Wohrle et al., Adv. Pofym ScL 50 (1983) 45; /. Mol CataL 21 (1983) 255; Angew. MakromoL
Chem 117 (1983) 103;/. Electrochem Soc. 132 (1985) 2144; MakromoL Chem, MacromoL Symp.
8(1987)195.
40. A. Bettelheim, B.A. White and R.W. Murray, /. Electroanal Chem 217 (1987) 271.
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
289
41. J.E van Baar, J.A.R. van Veen and N. de Wit, Electrochim Acta 27 (1982) 57; J.F. van Baar,
J.A.R. van Veen, J.M. van der Eijk and N. de Wit, Electrochim Acta 27 (1982) 1315.
42. H. Nishihara, K. Pressprich, R.W. Murray and J.P. Collman, Inorg. Chem. 29 (1990) 1000.
43. M.S. Freund, J.A. Labinger, N.S. Lewis and J.E. Bercan, /. Moke. Catal. 87 (1994) L l l .
44. A. Nanthakumar and H.M. Goff, /. Am. Chem. Soc. Ill (1990) 4047.
45. J.T. Groves and T.J. Murry, Am. Chem. Soc. Abstracts of Papers, 186th Meeting, April 1984.
46. R.R. A6zic,Adv. Electrochem Electrochem Eng. 113 (1984) 159.
47. M.M.P. Janssen and J. Moolhuysen, /. Catal 46 (1977) 289; Electrochim. Acta 21 (1976) 861,
869; B.D. McNicol, Electrocatafysis, Specialist Periodical Reviews: Catalysis, Vol. 2 (Royal Society
of Chemistry, 1978) Ch. 10.
48. T. Iwasita, EC. Nart and W. Vielstich, Ber. Bunsenges. Phys. Chem 94 (1990) 1030.
49. S. Motoo, in: The Chemistry and Physics of Electrocatafysis, Proceedings, Vol. 84-12 (The Electro
chemical Society, 1984) p. 331.
50. G. Kokkinidis et al., Electrochim Acta 30 (1985) 493; Electrochim Acta 30 (1985) 1611; /.
Electroanal. Chem. 257 (1988) 239; Electrochim Acta 34 (1989) 803; /. Electroanal Chem. 192
(1985) 375.
51. I.T. Bae, X. Xing, C.C. Liu and E. Yeager. /. Electroanal Chem 284 (1990) 335.
52. S.J. Langer and J.C. Card, /. Mol Catal. 42 (1987) 331; see also: T.D. Iran, I. Londner and S.H.
Langer, Electrochim. Acta 38 (1993) 221.
53. V.D. Neff, /. Electrochem Soc. 125 (1978) 886; D. Ellis, M. Eckhoff and V.D. Neff, /. Phys. Chem.
85 (1981) 1225; K Itaya et al., /. Appl Phys. 53 (1982) 804; /. Am Chem. Soc. 104 (1982) 4767;
K. Itaya, 1. Uchida and V.D. Neff,y4cc. Chem Res. 19 (1986) 162.
54. C.G. Murray, R.G. Nowak and D.R. Rolison, /. Electroanal Chem 164 (1984) 205; RK. Ghosh
and A.J. Bard, /. Am Chem Soc. 105 (1983) 5691; RK. Ghosh, A.W.-H. Mau and A.J. Bard, /.
Electroanal. Chem. 169 (1984) 315; H.Y. Liu and EC. Anson, U. Electroanal. Chem. 184 (1985)
411.
55. B. Keita and L. Nadjo, /. Electroanal. Chem. 230 (1987) 267; B. Keita, L. Nadjo and J.M.
Saveant, /. Electroanal Chem. 243 (1988) 267.
56. RJ. Kulesza and L.R. Faulkner, /. Am. Chem. Soc. 110 (1988) 4905; /. Electroanal. Chem. 259
(1989) 81.
57. RJ. Kulesza and L.R. Faulkner, /. Electroanal Chem. 248 (1988) 305.
58. C M . Castro-Acuna, F.R. Fan and A J. Bard, /. Electroanal. Chem 234 (1987) 347; WE. Rudzein-
ski and D. Root, /. Electroanal Chem. 243 (1988) 367.
59. K.E. Creasy and B.R. Shaw, Electrochim. Acta 33 (1988) 551; B.R. Shaw et al., /. Electrochem
Soc. 135 (1988) 869; D.R. Rolison, Chem. Rev. 90 (1990) 867.
60. K. Prater, /. Power Sources 29 (1990) 239.
61. P.M. Bersier, L. Carlsson and J. Bersier, Electrochemistry for a better environment, Topics in
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
ABSTRACT
1. Introduction
Oxidation processes using air and oxygen are generally used for the synthesis of
various inorganic or organic chemicals having utility as intermediates or final products.
These processes can be conducted in either the liquid phase or vapor phase, depending
upon such factors as the reactant volatility, thermal stablity of the reactants and
products, specific reaction rate, and overall process economics. Examples of inorganic
chemicals produced include nitric acid, sulfuric acid, hydrogen cyanide and the
oxychlorination of HCl. Typical organic chemicals that are commercially manufactured
using vapor-phase oxidation processes include ethylene oxide, acrolein and acrylic acid,
methacrolein, methacrylic acid, maleic anhydride, and phthalic anhydride. The products
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
and reactor technology where the yield of the desired product is maximized. Optimal
yields of the desired product can only be obtained if the product does not undergo
thermal decomposition in the given reaction environment and in any downstream unit
operations associated with product recovery. However, many products of selective
oxidation reactions are highly unstable since they are based upon anhydrides or epoxides
and may contain reactive organic functional groups, such as aldehydes or acids. Some
newer reactor concepts, such as membrane reactors, involve simultaneous reaction and
removal of die reaction product and have potential for significant improvement in product
yield due to isolation of the product in a more favorable reaction environment.
Notwithstanding the inherent hazards of oxidation processes, only 2% of chemical
plant accidents occurred in oxidation process, as shown in Table I.
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
.293
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Formaldehyde Methanol/air 99 94
Ethylene oxide Ethylene/oxygen 15 80
1,2-Dichloroethane Ethylene/air/HCl 95+ 95+
1,2-Dichloroethane Ethylene/axygen/HCl 95+ 97+
Acrolein Propylene/air 90+ 80-85
Acrylic acid Acrolein/air 95+ 90-95
Acrylonitrile Propylene/air/NH3 99+ 7 3-77
Methacrolein Isobutene/air 97+ 85-90
Methaciylic acid Methacroiein/air 70-75 80-90
Maleic anhydride Benzene/air 98 75
Maleic anhydride n-Butane/air 75-90 67-72
Phthalic anhydride Naphthaleoe/air 99+ 84
Phthalic anhydride o-Xylene/air 99+ 79
Table 3 lists some of the key monomers that are used as chemical intermediates
and form the basis for a large variety of engineering plastics and polymer. A comparison
between the reactants used in most existing processes and newer reactants from recent
patent and open literature publications is also provided. This comparison clearly points
to the direction where future oxidation process development is being emphasized. The
trend is obviously to replace more expensive olefin feedstocks by cheaper alkanes. This
effort was actually initiated at least ten years ago and has continued. Recent examples are
the synthesis of acrylonitrile from propane versus propylene, and the synthesis of
methaciylic acid from isobutane versus isobutylene.
existing and emerging processes is an active area of research, only those processes that
have achieved commercial status will be discussed. Details on reactor modeling are not
included for brevity, but will be presented elsewhere4.
2. Fixed-Bed Reactors
2.1. Introduction
Compared to other types of vapor-phase catalytic reactors, fixed-bed reactors
have received the most attention from the perspective of application and reaction
engineering analysis. For this reason, the level of understanding for this configuration is
perhaps the greatest when compared to other reactor types. This is primarily due to the
development of a hierarchy of fixed-bed reactor models. A combination 6f experimental
data on intrinsic and apparent reaction kinetics, engineering correlations for transport
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
296
coefficients, and detailed reactor modeling allows rational design and scale-up procedures
to be followed. Even though the knowledge base in fixed-bed reactor engineering is very
good, potential pitfalls inreactordesign or analysis of performance are still present.
Figure 1 lists some of the key issues must be considered when modeling fixed-bed
reactors for vapor-phase catalytic systems.5 A wide range of length scales must be
traversed in a realistic model of the fixed-bed reactor. The individual catalyst sites
represent the microscale where the catalytic reactions occur. Reaction kinetic
measurements using state-of-the-art laboratory reactors and information derived from
catalyst characterization instrumentation provide the basis for development of kinetic
models and identification of kinetic parameters based upon a sequence of elementary
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
steps.
Reaction Kinetics
Catalyst Site
Microscale 'r
Interfacial intraparticle
AC AT? AC AT?
Catalyst Particle
t
Type of Convective Mass
and Heat Transfer
• Plug Flow?
Macroscale Reactor
• -i- Axial or Radial Dispersion?
• Nonuniform Velocity
+ Radial Dispersion?
The collection of catalytic sites represent the next greater dimension on the
microscale since the characteristic dimension now corresponds to the shape factor for the
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
297
catalyst particle. Here, both intraparticle and interparticle transport processes and their
interactions with local intrinsic kinetics must be considered. Moreover, a realistic
description of the catalyst particle morphology, such as the distribution of pore sizes, is
needed.
The next larger length scale is the bed of catalyst particles and corresponds to the
macroscale level. Here, transport processes that occur on characteristic dimensions of
the reactor, such as the reactor diameter and overall length of the fixed-bed are the focus.
This includes a description of deviations of the gas flow pattern from ideal plug-flow, and
deviations of the temperature within the catalyst particles and in the gas bulk from an
ideal isothermal condition.
As illustrated in Figure 2, mathematical models for fixed-bed reactors are typically
classified as being either the pseudo-homogeneous or heterogeneous type. Pseudo-
homogeneous models do not explicitly distinguish between the fluid and solid phases
during formulation of the mass, energy balances and momentum balances. On the
contrary, heterogeneous models explicitly account for transport-kinetic interactions for
both the gas and solid phases. The advantage of the heterogeneous models is their ability
to describe intraparticle and interparticle transport-kinetic interactions on a single particle
level. Heterogeneous models should be used for reactions with moderate to fast rates and
moderate to high heats-of-reaction.
models explicitly account for radial gradients in the differential material and energy
balances.
The least complex class of reactor* used for gas-phase catalytic oxidation reactions
are adiabatic fixed-bed reactors. The single-stage or single-bed adiabatic fixed-bed reactor
has the simplest design. The number of processes mat employ adiabatic reactors for
oxidation processes is limited. Reactions carried out in the single-stage adiabatic reactor
include: (1) the oxidation of ammonia to nitric oxide over a Pt/Rh gauze, which is the key
step in the manufacture of nitric acid, and (2) the oxidation of a mixture of methane and
ammonia to HCN, which is an important intermediate needed in the synthesis of several
polymers, over a Pt/Rh gauze. Also included in the adiabatic reactor class is the multi
stage design, even though heat is exchanged between adjacent stages. Reactions carried
out in this multi-reactor system include: (1) the oxidation of SO2 to SO3 on vanadium
pentoxide catalyst, which is the key step in the manufacture of sulfuric acid, and (2) the
silver-catalyzed oxidation of methanol to formaldehyde. Finally, the most recently
developed reactor type in the adiabatic class is the reverse-flow reactor. This reactor has
been used in the former Soviet Union to oxidize SO2 to SO3.
Unlike catalytic reactors with heat exchange within the reactor, the distinguishing
feature of the adiabatic reactor is that the heat generated by an exothermic reaction goes
unchecked. The steady-state temperature profile that results is a monotonically
increasing function of distance from the inlet. The temperature increase asymptotically
approaches the adiabatic temperature rise. Temperatures can exceed the adiabatic
temperature rise in the unsteady state adiabatic reverse-flow reactor, as described in
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
section 2.3.
Because of the potential severity of the heat effects, oxidation reactions carried
out in adiabatic reactors form a special sub-class. Not only must the catalyst tolerate a
temperature rise of several hundreds of degrees, but the desired product selectivity must
also not be negatively impacted by the mgh temperature. If these two requirements are
met, then one can take advantage of the high rate of reaction at the high temperature
without significant penalty. Moreover, the relative simplicity of the adiabatic reactors,
compared to their cooled counterparts, can be exploited in design, scale-up, and operation.
Some typical adiabatic reactors are shown in Figures 3 and 4. These include the
simple catalyst bed 6 , the disk or shallow-bed reactor 6 - 7 .and the radial-flow reactor 7 .
The shallow-bed reactor is used in fast reactions
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY
where complete conversion can be
Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
299
achieved in a very short contact time. The shallow-bed affords minimal pressure drop,
although flow distribution can be an imposing problem7. The radial-flow reactor is used
to reduce pressure drop as well.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Particles for
Uniform Flow
Distribution
Catalyst
Particles
Catalyst
Dump
Flange
Output
-•—Ammonia +
Air
if \~\
Pd/Au
Getter
Support
Pt/Rh Gauze
Random k| M
1M
w\
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Pack - Catalyst
Catalyst Screen
Quench
hi
M
■*— Nitrogen Oxide
^,K
Shallow-bed Reactors Radial-flow Reactor
Figure 4. Three single-stage adiabatic fixed-bed reactor types.
Ammonia oxidation can also form nitrous oxide (N2O) in small amounts. The
desired NO producing reaction is extremely fast since the contact time is about one
millisecond and is limited by gas-solid mass transfer. For this reason, the gas linear
velocity and flow uniformity are critical issues. The yield to NO is an increasing function
of temperature and a slight decreasing function of total pressure. At atmospheric
pressure and 850 °C, the NO yield is 98%, while yields of ca. 96% are obtained at 8 atm
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
301
and 900 °C. The loss of precious metal catalyst becomes more important at higher
temperatures.
Air O H
O Condensate
$L
Steam
a Cold
Wat*
#f-A-4
I "
The high pressure process reaction operates at 110 psig and offers reduced
equipment size at the expense of slightly lower NO selectivity. In the split-pressure
process, the ammonia conversion is carried out at an intermediate pressure of about 30
psig, while the absorption is carried out at higher pressure. The NO that is formed in the
ammonia converter oxidizes further to NO2:
The primary reaction is reversible and highly endothermic, while the second reaction is
clearly undesired as it represents a yield loss. The methane and ammonia are reacted
using a very high mass velocity through ceramic tubes packed with a Pt/Rh gauze. The
contact time is on the order of one millisecond. As in ammonia oxidation, gas-solid mass
transfer limits the overall rate. So, the reactor must be designed to provide for a spatially
uniform, high-rate gas flow. The necessary energy is provided by burning fuel, such as
methane, on the shell side of the non-porous ceramic tubes. The main disadvantage of the
Degussa process is the degradation of the tubes at the high reaction temperatures
(1000-1100 °C).
In the Andrussow process, oxygen is present in the feed gas. The main and side
reactions are
CH4 + 2 0 2 -» C 0 2 + 2 H 2 0 (9)
The main role of oxygen is to oxidize methane to provide the necessary energy for the
HCN formation.
There are three primary points of distinction in comparing the Andrussow
process to the Degussa process. First, the Andrussow converter is an adiabatic reactor,
which considerably simplifies the reactor (design. Second, since oxygen is fed with the
ammonia and methane, and additional side reactions occur that reduce the overall yield.
As shown above, these include the oxidation of ammonia to nitrogen and the hydrolysis
of HCN to ammonia and CO. To minimize the latter reaction, the reactor effluent must
be rapidly cooled. Third, an additional constraint is the need to operate outside the zone
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
offlammability.
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
303
2.2-2. Reactor Performance and Modeling. As described earlier, modeling and analysis
of a single adiabatic fixed-bed reactor requires knowledge of the catalytic kinetics,
intraparticle and interparticle transport processes, and the macroscopic flow and
transport phenomena5. The modeling of adiabatic reactors is more straightforward than
nonadiabatic reactors because the complexities introduced by intrareactor heat exchange
are avoided. Modeling of multi-stage adiabatic reactors is much more difficult because of
the thermal coupling that occurs between the stages.
Scale-up of the single-stage adiabatic reactor from laboratory-scale is simple in
principle. For example, suppose the production rate in a laboratory-scale adiabatic
reactor with a cross-sectional area for flow of A (m 2 ) and bed-depth of d (m) is equal to P
(moles/hr). Moreover, suppose that complete conversion of the limiting reactant is
achieved in the laboratory reactor. Suppose further that the desired commercial-scale
production rate is 10 4 • P. Then the commercial-scale reactor should have the same bed
depth as the laboratory reactor but a cross-sectional area equal to 100 • A. This area
ensures that the same linear velocity, and hence contact time, is maintained in both
reactors. Moreover, for fast reactions external transport processes are typically rate-
limiting. By maintaining a constant velocity in the lab and pilot-scale reactors, the
external transport rates are held constant as well. Factors that could complicate this
simple scale-up procedure include non-adiabatic operation in the laboratory reactor, or
flow maldistribution in the commercial-scale reactor.
The conversion and product distribution data from a laboratory reactor study of
Pt-catalyzed methanol oxidation are shown in Figure 6. These data display the important
requirement that a reaction system must maintain a highly desired product selectivity at
high temperature, if it is to be carried out adiabatically. As the bed temperature is
increased for a feed that is rich in methanol, the formaldehyde selectivity actually
increases from less than 50% at 250 °C to over 90% at 600 °C. At reduced methanol to
oxygen feed ratios, the selectivity does not exhibit such favorable features. It is for this
reason that more than one adiabatic reactor can be used to avoid the need to feed all the
required oxygen to attain a desired methanol conversion per pass to a single reactor. This
operational strategy is discussed in the next section.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
2.2-3. Multistage Adiabatic Reactor. There are two classes of reaction systems for
which the single-stage adiabatic reactor is incapable of satisfying the conversion and/or
selectivity demands. The first class is reversible exothermic reactions. Multiple stages
with interstage cooling are required for these reactions in order to achieve an acceptable
conversion level with a reasonable reactor volume. The classical oxidation example is SO2
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
304
to SO3, which is described in more detail below. The second class is hydrocarbon partial
oxidation reaction systems in which the desired product selectivity is sufficiently
sensitive to temperature and oxygen concentration. More specifically, yield losses to
carbon oxides, which result from sequential or parallel side reactions, undermine the goal
of achieving high, intermediate partial oxidation product selectivity at a reasonable
hydrocarbon conversion per pass. Most hydrocarbon partial oxidations have this feature,
two examples of which include the silver-catalyzed oxidations of methanol to
formaldehyde and the catalytic oxidation of monomethylformamide to methyl isocyanate.
60
■- ^ Total
f^J\* -n— 6
^"CH20
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
50 L
h
40 h
£ 30h
|- /
(0 ^^^* " ^ ^ H 2
c
o 1
5 20 1r #1# CO(x10)
>
c A
o
o
10 h-
r~ /
HCOOCH^X
\ y •
—*■"■*■* c o 2
1 L-*L.1 V . 1 -f ^ l _L I I
200 400 600 800
Temperature (°C)
The key attribute of the first reaction is that it is both exothermic and reversible. As
shown in the conversion versus temperature plot in Figure 7, low temperature favors
1.0
Bed 4 Bed 3
Equilibrium
i
0.8 \— Steam super-
/ / /* *
Sg>
heaters S&
O A& Bed 2
^
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
(0
o x
I °-6
"55
k.
>
I Boiler #2
/
y
Bed 1
C
o
o
c 0.4 [—
CO
o
«
0.2
0.0 I
400 500 600
Temperature, °C
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
particular one shown has feed-effluent heat exchangers and cold-shot interstage heating.
More elaborate multi-stage reactors have been developed due to environmental pressures
to reduce sulfur emissions in sulfuric acid plants. One such process is shown in Figure 9.
The final stage of the reactor in this process is fed with a stream in which a fraction of the
SO3 product has been recovered. This enables the final stage to push the SO2 conversion
closer to completion.
Standard (;.,,„.„.
Cuoic Feet Scaraur
of Air (Dry Basisi
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Drying
Pump Tank
4 Purs
2.020,000 BTU in Cooling Water when Producing 33% H2SO4
*©-
41 -%J Catalyst
Silver 1
S-»-
cw H2l ^ 4 1 6sS
Formaldehyde
®—r
Methanol
reed
Methanol Recycle
Product
Although not shown in Figure 10, the oxygen addition is distributed between two
adiabatic reactors in some cases. In the first stage, the feed oxygen is completely
converted and most of the methanol converts to formaldehyde. Additional air and cooling
is carried out between the first and second reactors. The feed temperature and feed
compositions to each stage must be selected to maximize the formaldehyde yield and
methanol conversion. Moreover, recall that high temperatures in this system are
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
beneficial (see Figure 6). This means that the oxygen feed concentration and feed
temperature to each reactor must be sufficiently high to ensure light-off of the catalyst.
On the other hand, too high an oxygen concentration or feed temperature may mean a
reduction in formaldehyde selectivity or sintering of the catalyst.
The inherent advantage of the silver-catalyzed, adiabatic reactor formaldehyde
process over the metal oxide catalyzed, cooled reactor counterpart is the smaller process
vessel sizes. The smaller volumes are possible because temperature control is not critical
and a hydrocarbon-rich feed mixture is used. Thus, a high diluent flow rate is not needed.
The disadvantage is that methanol must be recovered and recycled. The alternative
process is discussed in section three.
2.2-5. Reactor Performance and Modeling. In multi-stage adiabatic reactor systems, heat
exchange is accomplished with a standard heat exchanger or with a "cold shot" of one of
the reactants. The reactor design can be an imposing optimization problem that involves
a large number of manipulated variables. The simplest objective function consists of two
terms: (1) the revenue generated by the production of desired product, and (2) the
annualized cost of the reactor system and catalyst. The difference between these is a
measure of the net profit. Obviously, this profit value is a liberal estimate given that
there are other costs in the overall process. Solution for the optimal set of variables
requires a detailed description of the multi-reactor and heat exchanger system.13'14
For the partial oxidation reaction system in particular, the multi-stage reactor
optimization problem is especially challenging given that multiple steady-states and
parametric sensitivity are the rule in exothermic reaction systems. In fact, to our
knowledge, this problem has not yet been solved in the open literature. Multi-stage
adiabatic reactors are impractical for these reaction systems if the number of stages is too
large. Moreover, radial temperature gradients may become too large from the standpoint
of catalyst durability or desired product selectivity. In these cases, the logical design
choice is a cooled multi-tubular reactor. This class of reactors is discussed in section
three.
Consider the oxidation of SO2 as a representative example of a single exothermic
reversible reaction. In this situation, the interstage cooling is accomplished with standard
heat exchangers. To achieve complete conversion in a single stage, the reaction would
have to be performed at a low temperature. However, the required reactor volume would
be prohibitively large. Instead, several reactors in series are used in practice that have
progressively decreasing feed and reactor temperatures. The feed temperatures to the
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
first and second stages are sufficiently high to ensure that a reasonable conversion is
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
310
achieved. The final stage has a sufficiently low temperature so that remaining reactant is
converted. Additional measures may be needed to push the conversion to 100%; one
measure is shown below.
n l-D
Adiabatic
Reactor
H B
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Valve Pairs
A B
Flow upward through reactor: Open Closed
Flow downward through reactor: Closed Open
temperature front would move out of the reactor after a certain time, thereafter, the
reaction would be extinguished. To prevent extinction, the direction of flow through the
fixed bed is periodically reversed with the help of valves. As a result, the portion of
packing which has cooled down is heated up again by the combustion zone moving in the
opposite direction. Hence, the two end regions of the packing act as regenerative heat
exchangers.
After a considerable number of flow reversals, a periodically steady state is
established in the fixed bed in which, in one-half period, the zone of reaction moves a
distance upward, whereas, in the next half period, the zone moves the same distance
downward. ..."
The ability to capture the hot spot within the bed by flow reversal relies on the
large difference in the characteristic time for convective massflowand conductive energy
transport. Flow switching can be easily accomplished on a time period that is much
shorter than the characteristic time of transit of a creeping hot spot to traverse the entire
length of the reactor.
The reverse-flow reactor can be used to carry out the complete combustion of
pollutants contained in air streams and to carry out reversible exothermic reactions. For
the former application, the efficient exchange of energy enables the complete combustion
of pollutants in small concentrations in air streams with low feed temperature. For the
latter application, the pseudo steady-state temperature profile that is established within
the bed is close to the optimal profile for achieving high conversion in an equilibrium
limited situation. It is this application that has relevance for oxidation to useful
chemicals.
Figure 12 shows representative temperature profiles within a bed of vanadium
oxide particles during flow reversal for SO2 oxidation to SO3.15 Profile one was measured
during flow from left to right through the bed, while profiles two, three, and four during
flow from right to left. The profiles reveal the very large temperature rise that can be
sustained (ca. 400 °C) compared to the adiabatic temperature rise (ca. 70-80 °C) for the
1.7% SO2 mixture in air.
3. Multi-tubular Reactors
3.1. Introduction
A larger fraction of selective oxidation reactions require finer temperature control
than the staged adiabatic reactor types can provide. In these reaction systems, the
selectivity is a decreasing function of temperature. This implies that the activation energy
of the desired reaction is exceeded by that of one or more of the non-selective reactions.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
In such cases, ineffective control can lead to hot spots and a subsequent loss in desired
product selectivity. The multi-tubular catalytic reactor consists of a bank of tubes
arranged in parallel that are immersed in a vessel through which a heat transfer fluid, such
as Dowtherm®, steam, or molten salt, is circulated. The schematic in Figure 13 does not
display the intricacy of the state-of-the-art multi-tubular reactor, which may contain up
to 25,000 individual tubes. The multi-tubular arrangement provides sufficient heat
transfer area and reduces the effective radial heat transfer distance. A large number of
parallel reactors are sometimes used instead of a single unit to prevent hot spot formation
and desired product losses.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Feed Gas
Heat
Carrier
Heat
Carrier
Some of the key attributes of the multi-tubular fixed-bed are provided in Table 4.
Shown also for comparison are the corresponding attributes of the fluidized bed, which is
an alternative reactor type for carrying out oxidation reactions that will be described in
section four.
The choice between the multi-tubular and alternative reactor types must be
considered on a case-by-case basis, taking all of these issues into account. Additional
considerations on reactor selection are discussed in section four.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
3.2.2. Ethylene to Ethylene Oxide. Ethylene oxide capacity is on the order of 10 million
tons worldwide because of it's use in the production of many important chemicals,
including ethylene glycol, ethanolamines, and polymers containing the hydroxyethyl
group. Ethylene oxidation to ethylene oxide is a prime example of the need for a rational
catalyst and reactor design to maximize yield of a partial oxidation product.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Undesired reactions are primarily the parallel and consecutive complete combustion
reactions:
C2H4 + 3 0 2 -> 2 C 0 2 + 2 H 2 0 (17)
t Tail Gas
I ^ /^\ Process
Water
^¥rS
Air
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Formaldehyde
Product
®s
Methanol
Feed
Vaporizer Reactor Absorber
All the reactions are exothermic. The heat of reaction of the complete oxidation is 1419
kJ per mole of ethylene. Moreover, the activation energies of the undesired reactions are
higher than that of the desired reaction. This temperature sensitivity of the ethylene
oxide selectivity requires very good temperature control. In order to suppress the
consecutive oxidation of ethylene oxide, ethylene conversions per pass are low,
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
316
Carbon dioxide and acetaldehyde are the primary carbon-containing by-products. The
commercial bismuth-molybdate catalyst has several promoters, that result in an acrolein
selectivity as high as 85% at a propylene conversion of about 90%. 20 Given the
exothermic nature of the reaction system, a multi-tubular reactor is used to minimize
temperature nonuniformities and the resultant loss in acrolein selectivity.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Reactor
Figure 15. Process flowsheet for n-butane oxidation to maleic anhydride using
multitubular reactors.21
r n Carb
Carburetor
Salt
Air Preheater
Air
~Ll
Q
Air
6 O-Xylene
Preheater
Cooler After- f
cooler
h Switch
Condensers
p j Blower
O-Xylene
Reactor
*TL Raw PA to
Purification
Gas
O-Xylene Cooler
Pump
Figure 16. Process flowsheet for o-xylene oxidation to phthalic anhydride using
multitubular reactors.22
reduce tube diameter, and (5) dilute the catalyst bed, which corresponds to activity
profiling The first three measures are the simplest to employ, but adversely affect the
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
320
process economics. For example, an increased carrier gas flow rate means that the volume
of the reactor and downstream units increases. Moreover, the separation demands are
increased. An increase in the bed thermal conductivity can be accomplished by using a
different support material or by diluting the catalyst with catalytically inert particles that
have a higher conductivity. The final two measures are considered in more detail.
reducing the activity by a factor of two over the first 20% of the bed length is not
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
321
100
i 1.0
1 0.8
> 0.6
>
o 0.4
..J 1 J J
Reactor Length Reactor Length
Figure 18. Catalyst activity profiling to reduce the magnitude of hot spots in a
fixed-bed reactor.23 Figure 18a (left). Straightforward step-function activity
profiling. Figure 18b (right). More elaborate saw-tooth type of activity profiling.
4. Fluidized-Bed Reactors
4.1. Introduction
While fixed-bed reactors have been widely utilized for industrial oxidation and
ammoxidation processes with a significant degree of success, they have several non-ideal
characteristics that have prompted numerous investigations on development of alternate
reactor types and operating modes. The most notable non-ideal characteristics of fixed-
beds include: (1) non-isothermal temperature profiles in both the axial and radial direction
that can lead to local hot spots with an adverse effect on selectivity; (2) lean hydrocarbon
compositions must be used to satisfy the lower flammability limit and to prevent ignition;
(3) maintenance of a uniform gas distribution across the reactor cross section is sometimes
difficult, which can lead to local temperature excursions; (4) the catalyst utilization is
often less than ideal due to internal and external heat and mass transport limitations; (5)
the catalyst must be able to withstand both high mechanical and thermal stresses that are
induced during loading and normal reactor operation; (6) the catalyst must maintain
activity and selectivity over an extended time-on-stream so that the process economics of
catalyst replacement make fixed-bed operation economical; and (7) the mechanical design
of the fixed-bed tube sheet and shell is complicated, especially for large reactor systems
that contains tens-of-thousands of tubes.
Fluidized-bed reactors provide an attractive alternative to fixed-bed reactors for
industrial oxidation and ammoxidation reactions, and have been successfully developed
and applied to several key process applications. Their unique operating characteristics
include temperature uniformity, high rates of mass and heat transfer, a high degree of
catalyst utilization, and ability to operate in a number of hydrodynamic regimes, all of
which result in operating flexibility.24 The complex gas and solids mixing patterns and
hydrodynamic behavior that can occur in this reactor type provides a number of
challenges from a process and reaction engineering perspective. These difficulties, along
with some important issues related to the development of an attrition-resistant catalyst
that can withstand the rigors of gas-particle fluidization, provide a partial explanation
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Process / Type
McAfee / Batch
Houdry/Fixed Bed
Suspensoid / Liquid
Phase, Powder
TCC-Houdriflow /
Pills, Moving Bed
The demands placed upon the petroleum refining industry as a result of World
War II provided the impetus for development of more efficient processes with increased
production rates than that available with existing fixed-bed processes. Innovations to the
fixed-bed processes included the so-called alternating-bed reactors. In these systems, the
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
324
regenerator and reaction beds were connected in series, and the catalyst was continuously
moved from one reactor to the other. Superficial gas velocities less than 1 m/s were used
to minimize erosion of the reactor hardware and catalyst attrition. This led to the
observation that dense beds of catalyst powder could be maintained in place with
relatively small losses due to elutriation, even at superficial gas velocities, that were
orders of magnitude greater than the calculated settling velocity of the individual particles
that comprised the bed. It was observed that at these gas velocities the particles were
considerably agitated by gas bubbles that flowed upward through the mixed particles.
The measured pressure drop through this gas-solid suspension was equivalent to that
calculated using the weight of catalyst in the bed as the basis. Hence, the weight of the
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
catalyst bed was buoyed by the force induced by the upward flowing gas bubbles.
Industrial application of fluidized-bed technology was born from these observations.
The above efforts resulted in the development and operation of the first truly
fluidized-bed cracking process in 1942 at the Standard Oil Company of New Jersey's
Baton Rouge refinery. A schematic of an early fluidized-bed catalytic cracking reactor is
shown in Figure 20. Inspection of this figure shows that the two key elements of the
Cottrell
Precipitator
Recycle
Heat
(Exchanger
system are the reactor and regenerator which are connected so that fresh and regenerated
catalyst can be sent both to and from these two units. The vaporized oil was cracked in a
dense, fluidized bed of catalyst with solids residence times on the order of two to three
minutes. Additional details on the design and operation of these early units are given in
the above reference and those cited therein. Today, there are more than 250 commercial
fluidized catalytic cracking units in operation, with more than half of these located in the
United States.25 Most of these have been designed by either Kellog or UOP.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
ammoxidation are also manufactured influidized-bedreactors. A generalized schematic of
a fluidized bed reactor that might be used in a particular chemical processing application is
illustrated in Figure 21.
Conveying
gas vent
t
Product A Separator
gas
Bubbie Displacing
annular downflowing
shell of bed solids
Cyclone
*<-
1
^ -j^Twake ^ -V '
i 1.5 D- 1
Conveying ga:
In nearly all processes involving fluidization, the solid is transported from one
stage or processing step through pipelines that may contain various types of valves.
Upward transport of solids is usually achieved by contacting the solids with a flowing gas
stream with a sufficiently high velocity so that the gas-to-particle drag force exceeds the
downward drag force of the particles. Transport of solids in the downward direction is
performed by allowing the solids to settle through a pipeline. Examining Figure 21, the
process gas enters the plenum below a grid that supports the bed where it forms bubbles
that flow upward and fluidized the solids. When the bubbles reach the upper region of
the gas-solid fluidized zone, they burst into the freeboard or disengagement region and
entrain some particles. The exiting gas passes through a cyclone that separates the gas-
solid suspension into a product gas stream and a solids stream where the latter is returned
to the bed. If the catalyst particles undergo deactivation or require some type of post-
reaction treatment, they can be withdrawn through a standpipe and returned to the
fluidized bed via a pneumatic lift and separator. Once in the separator, the solids will
flow downward through the standpipe seal leg back into the reactor. Although several
variations on this particular schematic can be envisioned, most of these incorporate the
basic concepts illustrated in Figure 21.
Both the internal and external heat transfer resistances in these coils are smaller than those
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
327
obtained in either internal or external heat exchanger used in fixed-bed reactors, which
results in a more effective use of the heat transfer surface.
The ability to use hydrocarbon concentrations that are in the flammable region is
another attractive feature of fluidized-bed reactors over fixed-bed reactors.27 This is
accomplished by keeping the fuel and oxygen sources separate until they are contacted in
the fluidized bed of catalyst. The high heat capacity of the catalyst and the rapid heat
transfer between the gas and solid particles minimizes or prevents the formation and
movement of a stable flame front that is needed for ignition. In butane oxidation, for
example, the hydrocarbon concentration is ca. 4 mole % in air for certain fluidized-bed
processes, while fixed-bed processes contain less than 2 mole % of hydrocarbon in the
feed. 27 The use of a higher butane concentration results in a lower air compressor duty,
lower utility costs, and produces a more concentrated product stream. This, in turn,
reduces the size of the product gas cooler, reduces the amount of nitrogen that is recycled,
and reduces the size of the absorbers used for product recovery. This last point is
noteworthy, because it suggests that any comparisons of process technology involving
different reactor types must be performed using a model of the entire process, and not
just direct comparisons of reactor performance variables.
Charging and removal of the tableted catalyst used in fixed-bed reactors requires
that the catalyst have a high mechanical strength and that the characteristic pressure drop
versus gas-flow rate response function for each tube fall within a certain tolerance to
minimize preferential gas flow. By contrast, loading and unloading of the catalyst in a
fluidized-bed reactor is much easier and can be accomplished through well-known
pneumatic transport techniques However, the catalyst must possess sufficient attrition
resistance so that the process economics associated with attrition and elutriation of fines
is not adversely affected.
Catalyst costs associated with fluidized-bed processes can be more expensive than
their fixed-bed counterparts because of the additional steps involved in catalyst
manufacturing that are associated with imparting the required attrition resistance. For
applications where catalyst deactivation occurs, fluidized beds may be the only
economical option, since catalyst replacement is simpler than fixed-beds.
One significant complication of fluidized beds is generally associated with the
complexities associated with their gas-solid hydrodynamics and mixing characteristics.
The ideal mixing states of plug-flow of gas and perfectly backmixed solids are often not
readily achieved in fluidized beds. More importantly, these mixing states may not even
--`,```,,`,`,`,,,`,``,,,`````,,-`-`
be desirable for certain types of reaction networks where the selectivity to the desired
product may be adversely affected. In addition, achievement of a high degree of reactant
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
328
conversion may be difficult in a fluidized bed and may require the use of staging or
internals to reduce solids backmixing or minimize imperfect gas-solid contacting or gas
bypassing. By contrast, reaction engineering of fixed-bed reactors is better understood
and amenable for scale-up using intrinsic kinetic data and existing knowledge on transport
effects.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
corresponding company that developed the process technology is given in Figure 22.
4.4-1. Sohio-BP Acrylonitrile Process. A process flow diagram of the Sohio-BP process
for the manufacture of AN from the air oxidation of propylene and ammonia is shown in
Figure 23. The overall reaction is
The main reaction by-products are hydrogen cyanide, acetonitrile, and carbon oxides. The
reaction mechanism has been studied in detail, and involves a complex sequence of
oxidation-reduction reactions involving various metal oxides.20
The reaction is typically conducted using stoichiometric ratios of the reactants
with mean-residence times of the gas being on the order of a few seconds. The operating
conditions may vary, but can range between 20 to 200 kPa gauge (2.9 to 29 psig) total
pressure and 400 to 500 °C. The process is highly selective, and does not require
recycling to produce high AN yields of approximately 0.8 to 0.9 kilogram per kilogram of
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
propylene. The excess heat of reaction is recovered in the form of steam. Commercially
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
330
recoverable quantities of both hydrogen cyanide and acetonitrile (0.10 to 0.2 kg per kg of
propylene fed) are also produced.
Al A Water / 7 \ A A r\
HP Steam
BFW
Air
V VJ Y Water
V^ V^ ^Y Heavy Impurities
Ammonia
QPfOEyJene.
The bismuth-phosphomolybdate catalyst used in the early Sohio process has been
steadily improved and replaced over the years by more highly active and selective
catalysts. Sohio introduced Catalyst 21 in 1967, which was an antimony-uranium based
oxide. Catalyst 41, which was based upon ferrobismuth-phosphomolybdate, was
introduced in 1972, while Catalyst 49 was introduced in 1978. All of these give improved
process efficiency and a reduction in by-products.
In the process-flow diagram shown above in Figure 23, the fluid-bed reactor crude
effluent is cooled and scrubbed with water in a countercurrent absorber. The off-gas,
which consists primarily of nitrogen, is vented. The reaction products remain in the
absorber aqueous phase. By-product acetonitrile is removed from the absorber bottoms
by extractive distillation in the acetonitrile recovery column. In the next column, the
crude acrylonitrile and hydrogen cyanide are removed in the distillate, while water and
residual acetonitrile are removed in the column bottoms product. In the remaining two
columns, the hydrogen cyanide is separated from the wet acrylonitrile, the water content
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
of the product is then reduced, and any remaining nonvolatile impurities are removed in
the bottoms product.
The major by-products of this process, hydrogen cyanide and acetonitrile,
normally are incinerated as their supply often exceeds demand. Unused ammonia can be
recovered as ammonium sulfate and then disposed, but it is often vented to the
atmosphere. Aqueous wastes containing cyanides, sulfates, and various organic by
products must be incinerated, by deepwell injection, or be pre-treated for subsequent
biological waste treatment. Recent state and government regulations involving plant
emissions have resulted in major reductions and handling of these waste products.
4.4-2. Maleic Anhydride Processes. Maleic anhydride is an important raw material in the
manufacture of alkyd and polyester resins, surface coatings, lubricant additives,
plasticizers, copolymers, and agricultural chemicals. For this reason, a number of
industrial processes have been developed that are based upon fixed-bed reactors,
fluidized-bed reactors, and recirculating solids reactors. These are summarized in Figure
24 Those developed by Mitsubishi Chemicals, Alma/Alusuisse, and Sohio/UCB are
based upon classical fluidized-bed reactors and have been implemented on the commercial
Company Reactor
ALMA (Alusuisse) Fluidized Bed
Alusuisse Italia Fixed Bed
Amoco Fixed Bed
Mitsubishi Fluidized Bed
Monsanto Fixed Bed
Scientific Design Fixed Bed
Sohio-UCB Fluidized Bed
Dupont Recirculating Solids
scale. Another process developed by Badger and Denka has also been developed and
implemented on the pilot-scale, but has not yet been realized on a commercial scale.
Some noteworthy features of these processes are discussed individually below. 27
(i) Mitsubishi process. A process-flow diagram of the fluidized-bed maleic-
anhydride process developed by Mitsubishi Chemicals is shown in Figure 25. The
hydrocarbon source is the crude C4 fraction from naphtha crackers, so it represents a
departure from the more classical benzene-based processes and more recent processes
based upon n-butane. Although recent information is not available, the capacity of the
plant is at least 18,000 metric tons per year of purified maleic anhydride.
Steam.
Waste Gas
Combustor
Steam
Water
i\ Quench Tower
C 4 Fraction V Product
Pelletizers
A
Air
Bottoms
The fluidized-bed reactor is fed with the crude C4 fraction and air where the
hydrocarbon concentration is high enough that it exists in the explosive range (i.e.,
between 1.8 to 8.9 mole %). Because of the rapid gas-solids mixing and associated
hydrodynamics in the grid region, an ignition cannot occur, so that safe operation is
possible. The excess heat of reaction is removed by internal cooling coils that results in
the generation of high pressure (430 to 710 psig) steam. The hot reactor off-gas, which
contains the crude maleic anhydride, is quenched in a spray tower so that an aqueous
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
333
solution of crude maleic acid is produced. The off-gas from the spray tower overhead
contains a small fraction of reaction by-products and is incinerated in the waster gas
combustion chamber to generate additional high pressure steam.
Dehydration and purification of the crude maleic acid stream obtained from the
quench tower bottoms is performed by evaporation and distillation. The purified maleic
anhydride vapors are taken overhead and condensed for subsequent pelletization and
packaging.
A clear advantage of this process is the ability to utilize a relatively cheap C4
source versus a more expensive n-butane feedstock, and the generation of high pressure
steam that can be integrated into other near-by processes located on the same
manufacturing site. The attrition-resistant vanadium-phosphorus catalyst is
manufactured using classical spray-drying techniques with sufficient silica to impart
hardness
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
4 mole %
f Catayst
latter exceeds that of maleic anhydride (ca. 202 °C at 1 atm). This is claimed to minimize
the formation of unwanted by-products that are otherwise formed in aqueous recovery
processes involving the evaporation of aqueous maleic acid.
The vanadium-phosphorus catalyst is transformed into an attrition-resistant form
using a proprietary commercial-scale spray-drying process. It is claimed that the catalyst
experiences negligible losses due to activity and attrition so that economical operation can
be maintained. Both the catalyst and process are available for licensing, and a commercial
facility was recently started up in Europe.29
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
To Incinerator Dehydrator
Fluid-Bed
Reactor
Maleic Anhydride
Air Compressor
Source: G.S. Schaffel et al., Proc. Second World Congr. of Chem. Eng., Vol. 2 (1981) 4.
The front end of the process is similar to the previously described Mitsubishi
process, since it uses a classical fluidized bed and involves absorption of the crude maleic
anhydride product in water to form maleic acid. Dehydration of the maleic acid to maleic
anhydride is performed in a series of fractionation columns where an organic solvent, such
as xylene, is used as the organic entrainer. As in the previous fluidized-bed processes, a
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
335
catalyst having excellent attrition resistant and stable activity over an economical life is
claimed.
Water T
HP Steam u
BFW
Residue
Crude Maleic Anhydride
Butane
all of the issues given above are not specific to maleic anhydride processes, but generally
apply to any application where afluidized-bedreactor is being considered.
Phthalic Anhydride
Condenser
Tail Gas
Incinerator
Flu id-Bed
Reactor
Naphthalene
(Start) Q *|
m Crude P.A. r
*^e? Storage V^
^ >
%J*\C ¥1
(_D
Heat Treating X J Distillation Refined P.A.
Tank Column Storage
The catalyst that is elutriated in the reactor product gas is removed by filters, if
total catalyst retention is desired. Alternatively, cyclones are used for dust recovery
where a small catalyst loss is acceptable. A catalyst makeup rate of less than one
kilogram per 100 kilograms of feed is generally practiced to maintain constant activity.
Because catalyst losses by attrition are generally less than the makeup required for
activity maintenance, total fines retention with filters is not absolutely essential.
Yields are on the order of 98 kilogram of phthalic anhydride per 100 kilograms of
feed gas, which corresponds to ca. 85% selectivity and is comparable to other fluidized-
bed reactions. The actual dimensions of commercial-scale reactors can be quite significant.
Reactors more than six meters in diameter with height-to-diameter aspect ratios between
one to two are still in operation.
Pros Cons
Temperature Control Lack of Temperature Gradients
Heat Management Higher Pressure Drop
Continuity of Operation Catalyst Elutriation
Less Active Catalysts Catalyst Attrition
Deactivating Catalysts Incomplete Gas-Solid Contacting
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
according to the classification of Geldart, the transition from the fixed-bed regime to
successive regimes occurs as the superficial gas velocity is increased beyond the
minimization fluidization velocity Umf Typical fluidized catalytic-cracking riser reactors
with solids that belong to the Geldart powder group A can operate in the fast, fluidized-
bed regime with superficial gas velocities between 8 to 18 m/s. 32 This particular regime
occurs when the gas velocity is increased within the turbulent fluidization regime so that
the overall bed voidage increases and the top surface between the dense bed and the
freeboard region becomes less and less distinct. The gas velocity is large enough that
solids elutriate from the top of the reactor so that it becomes necessary to either add fresh
solids at the reactor inlet, or to capture the elutriated solids and recycle them to the
reactor, if the solids inventory is to be kept constant.
p £&*&
F*ed
Bed
Paniculate
Fluidization
of.
r\0 n
Bubbling
Regime
Slug
Flow
mm.
mm
mm
Turbulent
Regime
w Fast
Fluidization
Pneumatic
Transport
Aggregative Fluidization
classical bubbling bed to the transport bed reactor. It shows that the difference between
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
341
the mean-gas velocity and mean-solids velocity, or the so-called slip velocity, is greatest
for Type C circulating fluidized-bed reactors. The remaining reactor types, such as the
classical bubbling fluidized-bed, and the transport-bed reactor, result in gas-solid
hydrodynamics with lower slip velocities. Based upon these comparisons, the circulating
fluid bed would be the preferred mode of operation for systems where high rates of
interparticle mass transfer and rapid gas-solids mixing might lead to an overall
improvement on performance.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Ar1/3 = d p [pApg/^ 2 ] 1 / 3
The above discussion suggests that the advantages of operating a fluidized bed in
the fast-fluidization regime for gas-solid catalyzed reactions, relative to the other lower-
level hydrodynamic regimes, include: (1) higher gas throughput per unit cross-sectional
area of the reactor, (2) adjustable retention time of the catalyst, (3) the gas flow pattern
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
342
approaches plug-flow with negligible axial dispersion, (4) high rates of heat and mass
transfer between the gas and solids on a local level, (5) uniform or nearly uniform gas-
solid suspension temperature, (6) reduced tendency of the solids to cake together or
agglomerate, (7) possibility of staged addition of gas along the reaction zone, (8) addition
and removal of catalyst in the recycle leg is simplified, and (9) the catalyst can be exposed
to two different reaction environments. From the perspective of industrial oxidation and
ammoxidation reactions, these advantages can translate into superior process operation
and economics if the proper combination of kinetics, transport effects, and catalyst
performance can be identified or developed.
Increasing Expansion
Figure 33. Effect of mean gas velocity on the mean solids velocity for various
fluidized-bed operating regimes.33
reaction with butane, since the oxygen incorporated into the maleic anhydride is obtained
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
344
from the catalyst lattice. The ratio of the gas to solids flow rates, the riser inside pipe
diameter and overall length, the particle characteristics according to the Geldart
classification, the thermodynamic and transport properties of the reaction gas, and the
suspension recirculation rate must be carefully considered to ensure that the optimum
yield of maleic anhydride is obtained. A comparison between the gas velocities and solids
concentration for transport-bed and classical bubbling fluidized-bed modes of operation is
given in Table 6.
Regenerator |
rQ
Stripper I
&H
Standpipe~|
\ /
it T Inert}
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
M Butane
The reduced catalyst, once it enters the stripper, is contacted with inert gas to
reduce the concentration of any adsorbed maleic anhydride and other species on the
catalyst surface before being introduced into a conventional fluidized-bed regenerator that
is operated in the bubbling hydrodynamic regime. Here, the reduced catalyst is contacted
with air for several minutes so that any lattice oxygen removed during the reduction step
can be replenished. The heat of combustion and reaction is removed by internal cooling
coils with steam generation. The off-gases from the regenerator, which contain COx's and
nitrogen, are directed to a CO converter for generation of additional steam. Because the
catalyst is exposed to butane-rich and oxygen-rich reaction gases in the riser and
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
345
regenerator, respectively, the average oxidation state of the metal oxide is cycling at the
characteristic frequencies associated with the gas-solid contact time distribution in these
two reaction vessels.
a net upward velocity component. Vigorous exchange of gas and solids occurs at any
given local axial position along the riser height. The gas-phase flow pattern approaches
plug-flow, while the solids approaches plug-flow with some degree of internal recycle
before it escapes the riser section and enters the stripper and catalyst regenerator. Such a
description of the gas and solids flow patterns have been observed in gas and solids tracer
experiments, as well as in other non-invasive techniques, such as computer-aided
tomography and computer-aided radioactive particle tracking.37 Taken collectively, this
type of hydrodynamic behavior reduces local temperature gradients and is more readily
scaled up to larger vessels, since the production rate per unit volume of reactor is greater
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
than either fixed beds or fluidized beds without any recycle.
The use of a conventional bubbling; bed for the catalyst regeneration step permits
well-known techniques for design and scale-up of this section of the transport-bed
system to be applied. The hydrodynamics of the gas-solid mixing are such that high heat
transfer coefficients between the internal coils and gas-solid suspension can be obtained,
which translates into good temperature control.
Possible limitations with this reactor system include the usual complexities
associated with the design and operation of coupled reactor vessels and separators where
transpon of gas-solid suspensions are involved. These difficulties can be overcome with
fundamental process models for the reactor system that account for kinetics, transport
effects, and hydrodynamics, as well as a well-designed and proven process control
systems. In addition, an attrition-resistant catalyst that can maintain long life, high
activity and high selectivity within the process economic constraints is essential. Key to
the development of the DuPont riser process was the development of novel technology
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
347
for a catalyst that met these constraints that was proven in laboratory-scale and pilot-
plant performance evaluations. Given the special nature of this particular challenge, this
is discussed in detail below.
5
F.C.C. Catalyst ^m MO/PSA
35
Figure 36. Attrition-resistant butane oxidation catalysts.
a (left): conventional spray-dried catalyst; b(right): porous hardened shell catalyst.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Recirculating solids reactors have been used in two industrial oxidation and
ammoxidation processes that have already been commercialized, or are in the process of
being commercialized. These include: (1) the DuPont process for manufacture of
36
tetrahydrofuran , and (2) the Lummus process for production of certain aromatic
nitriles.40 In comparison to fluidized catalytic-cracking processes where over 250 units
are operational worldwide, 34 these two processes represent frontier applications of
recirculating solids reactor technology.
Most of the tetrahydrofuran and its derivatives that are commercially produced
today are based upon the production of 1,4-butanediol by the reaction of acetylene and
formaldehyde using the classical Reppe process. For specific reasons, such as acetylene
availability, raw material costs, and environmental considerations, this route is not as
attractive as the DuPont two-step process that occurs according to the following
reactions:
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
for PTMEG is in the manufacture of Spandex fibers and copolyester elastomers (COPE)
for products such as Lycra®, Hytrel®, and other consumer products. A 100 million lb/yr
plant based on maleic anhydride from n-butane is scheduled to start up in 1996 in
Asturias, Spain.36
Although most commercial maleic anhydride processes in operation today use
n-butane as the hydrocarbon source, other feedstocks, such as benzene, butene, and
butadiene, can be used. Prior to the 1960's, benzene was exclusively used as the raw
material for commercial MAN processes. Early in the 1970's, increases in the cost of
benzene lead to a raw material advantage for C4 hydrocarbons, especially n-butane, as the
preferred feedstock. Increasing environmental hazards associated with benzene, the lower
carbon efficiency of benzene when compared to n-butane since one-third of the benzene
carbon is lost due to combustion, and potentially higher MAN yields from n-butane, lead
to the commercial dominance of n-butane as the preferred feedstock. Most commercial
processes in operation today are based upon multitubular fixed-bed reactors with
n-butane as the feedstock. The largest single reactor is capable of producing about 40,000
tons/yr. The DuPont THF process represents the first commercial application of
transport-bed reactor technology to the area of selective oxidation.
Aromatic nitrites, such as benzonitrile, phthalonitrile, isophthalonitrile,
terephthalonitrile, and nicotinonitrile, are important chemical intermediates that were
most recently produced by conventional ammoxidation in fluidized-bed or fixed-bed
reactors using air, ammonia, and the hydrocarbon reactant in the presence of a suitable
metal oxide catalyst. Nicotinonitrile may be hydrolyzed to nicotinamide or nicotinic acid
or niacin, either of which may be used as a component of vitamin B complex.
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
350
CH, C* <37>
tartphthalonitrile
CN
JjjY +NH3 + (Cat)0X ► fiT
(38)
nicotinonitrite
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Transport
Bed
Reactor
Butane •
Figure 39 shows the maleic anhydride selectivity versus n-butane conversion data
extracted from the DuPont riser reactor system shown in Figure 34 with the patented
attrition-resistant VPO catalyst. These data show that a wide range of n-butane feed
compositions can be used in the absence of gas-phase oxygen without affecting the maleic
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
anhydride selectivity to any significant degree. The selectivity losses are due to the
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
352
formation of CO x that is associated with the non-selective oxygen species that is present
on the catalyst surface after re-oxidation in the fluidized-bed catalyst regenerator shown
earlier in Figure 34.
100 r Ts360'C
80
> 60
UJ BUTANE
40 IN FEED
• 1-2%
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
20 • 3-6-I.
• 12-50*
0 L
20 40 60 80 100
CONVERSION
Figure 39. Butane oxidation in a recirculating solids reactor at various butane feed
42
gas compositions.
|*60|-
o
0,40
20
0
20 40 60 80 100
Conversion
5.5-2. Lummus Aromatic Nitrile Process. A schematic of the process flow diagram for
the manufacture of aromatic nitriles by oxidative ammonlysis is shown in Figure 42. The
process can be divided into a reaction section, a product-recovery section, and an
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
ammonia-recovery section.
The reactor section is very similar to that used in the DuPont THF process, and
contains separate reactors for reduction of the metal oxide catalyst, stripping of the
catalyst to remove adsorbed organics, and regeneration of the metal oxide with air.
The reactor feed contains the organic reactant, recycle gas and organic
intermediates, and ammonia supplied from the recovery section and storage. The
partially vaporized feed is introduced to the reactor through the gas distributor along with
the liquid heel. The use of partial vaporization avoids the problem of fouling the feed
evaporator with residues having a low volatility. If any heavies are formed that deposit
on the catalyst, they can be removed in the fluidized-bed regenerator by total oxidation
with air.
100 p
.STRIPPED CATALYST
80h
? 60 CATALYST
NOT STRIPPED
4 0
« T = 360°C
20
0
20 40 60 80 100
Conversion
nitrogen, carbon oxides, and a small amount of hydrogen cyanide. The hot reactor effluent
is first cooled in a quench tower as a single-stage separation of the condensibles, e.g.,
organics and water vapor, from the non-condensibles such as nitrogen, carbon oxides, and
ammonia. Isolation of the desired aromatic nitrile may be performed by a variety of
combined unit operations and is dependent upon the particular nitrile being manufactured
owing to their different physical properties. Typically, the unit operations involved will
include crystallization, fractionation, washing, and centrifugation. Additional details are
available in the paper of Sze and Gelbein.40
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
355
Flue ♦
Gas
J L ^ Regenerator
CO2/HCN
To Regenerator
A
Veni_
Gases
Organic
Feed %.
Make Up Ammonia
T
Product Nitrite Waste Water
40
Figure 42. Lummus aromatic nitrile process.
The use of excess ammonia in the riser dictates that it must be recovered and
recycled for economical plant operation. This is performed using the classical approach
through ammonia absorption, solution stripping, CO2 absorption, and ammonia carbonate
dissociation.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
Table 7 gives a summary of the aromatic nitrile yields reported by Lummus40 that
are based upon operation of a pilot-plant. Overall yields of at least 75%, with most being
between 85 to 90%, are claimed.
It has been reported that one commercial process for the manufacture of
isophthalonitrile from /w-xylene is currently in operation.40
Heat Removal _ + +
Temperature Control - + +
Maximum Feed Cone, % 1.8-2.1 4 >10
Capital Investment - + ++
Selectivity + , - ++
Catalyst Attrition + -
Although the specifics associated with the above ratings are omitted, inspection of
the results indicates that the ranking for this particular application is riser reactor >
fluidized bed > fixed bed. This assumes that a suitable attrition-resistant catalyst can be
identified which has been successfully performed in the current application.39*40
5.7. Summary
Particular processes that are being commercialized, or are in the process of being
commercialized, include the DuPont tetrahydrofuran process and the Lummus aromatic-
nitrile process. The DuPont process combines a novel application of the recirculating
solids reactor to the selective oxidation of n-butane to maleic anhydride with a novel
invention for imparting catalyst attrition resistance. In the Lummus process,
ammoxidation of aromatics to nitriles is conducted in a fashion that is analogous to the
DuPont process, although precise details on the attrition resistant are lacking. The benefit
of separating the selective reaction that uses lattice oxygen from the catalyst from the
catalyst re-oxidation is also apparent in this application.
Reaction engineering of recirculating solids reactors for these particular
applications is perhaps more complicated than that for classical fixed-bed or fluidized-bed
reactors without catalyst recycle. Part of this is due to the lack of published data on the
oxidation-reduction kinetics and availability of fundamental models for the gas-solids
hydrodynamics and transport effects. These, along with other issues, provide the basis
for future fundamental and practical challenges associated with this technology.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
6. Emerging Reactor Types
Figure 43 shows four common catalytic membrane reactor types. The catalytic
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
A + B -> R (39)
A + R -» P + Q (40)
A + B -» P + Q (41)
where, without any loss in generality, A is oxygen, B is the hydrocarbon, R is the desired
selective product, and P and Q are undesired total oxidation products.
It is instructive from a catalytic membrane design standpoint to examine the
reaction structure for two limiting cases. 60 Under conditions in which component A is the
limiting reactant and the rate of the third reaction is negligible, the network has the
parallel structure given by
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
359
A -* R (42)
A -> P + Q (43)
1.CMR 2. CNMR
uwMMmaaaiiwjmaumam
3. PBCMR 4. PBMR
^W\SWV\W\V
j^^^^yy^g^yiw
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
nonpennselective perm selective permselective layer support catalyst particle
active layer active layer without activity
Under these conditions, the point selectivity to R, the ratio of the rates of the first and
second reactions at a local point (•) within the reactor, depends on the reaction orders
with respect to A in each reaction, given by nAi and nA2 for the first (selective) and
second (non-selective) reactions, respectively. If both nAi and nA2 are positive, then the
following rules apply to the impact of mass transport limitations with respect to A:
• If nAi > nA2 > then mass transport limitations are detrimental to the yield of R
• If nAi = i*A2, then mass transport limitations have no effect on the yield of R
• If nAi < nA2 . then mass transport limitations are beneficial to the yield of R
• For positive nAi and nA2 values, the overall rate of A consumption decreases
as the limitation due to mass transport increases
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
360
On the other hand, if component B is the limiting reactant in the reaction network above,
the reaction system has the consecutive structure given by
B -» R (44)
R -» P + Q (45)
Mass transport limitations with respect to component R in this case have a detrimental
effect on both the yield of R and on the overall rates of both reactions. That is, if the
characteristic time for R transport is of the same order, or longer than the characteristic
reaction time, then the extent of non-selective consumption of R can be significant.
These limiting case behaviors provide some guidance about how to tailor a
catalytic membrane to improve the selectivity to desired product R. For partial oxidation
reactions with kinetics that satisfy IIAI < n^2, the membrane should be designed with one
or more of the following properties:
The combination of properties (i) - (iv) implies a controlled supply of the oxidant (A)
through the support layer to the active layer. By feeding B from the active layer side, a
low effective A/B ratio is achieved in the reaction zone. A segregated feed condition is
best accomplished if the thin film is permselective to A. Another way to achieve a
segregated feed is to use a support layer with a sufficiently large thickness or sufficiently
small pores to limit the flux of A to the active layer. Both of these approaches are
described below.
The combination of properties (i) - (iii) and (v) means that the intermediate R that
is formed is selectively removed from the catalytic zone. The resulting reduction in the
concentration of R, and hence the rate of the non-selective reaction, increases the
selectivity of R. This concept has been advanced theoretically by Lund and
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
61 62
coworkers. »
Copyright World Scientific Publishing Co.
This approach is described below.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
361
Their concept is to employ a PBMR in order to maximize the yield of the desired
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
intermediate species R. The hypothetical PBMR consists of catalyst on the tube side
wherein reaction occurs. The key feature of the reactor is the membrane that is assumed
to be permselective to the intermediate R. As reaction occurs on the tube side, the desired
intermediate is selectively removed to the shell side where it is swept away by an inert
component. The model shows that for an intermediate range of conversion of main
reactant B, the yield of R is higher for the permselective compared to nonpermselective
situation.
These simulations provide a materials challenge: Develop an inorganic membrane
through which, for example, a partially oxidized hydrocarbon permeates at a higher rate
than the unoxidized hydrocarbon and oxygen. Membranes that come closest to meeting
this challenge are of the Knudsen variety. Selectivity ratios for R/B, for example, do not
exceed 3 for such membranes. Thus, while intriguing, the concept remains untested.
A second type of permselective membrane specifically applied to partial oxidation
reactions involve the selective permeation of oxygen through a solid electrolyte. Solid
electrolytes used for oxygen transport fall into two main categories. The first includes
mixtures of divalent and trivalent cations with tetravalent metals, an example being yttria-
stabilized zirconia (Y203-Zr02) or solid mixtures containing Bi2(>3 (e.g., (Bi203)o.85
(La2C>3)o.l5). The transport of oxygen proceeds through a lattice that has anion
vacancies created by the doping material. The second category is the mixed-conducting
(ionic and electronic) materials. One example is the family of perovskite materials,
denoted by the general formula ABO3, such as (Lai.xSrx)MnC>3 . d-63"65 The oxygen
transport in this case proceeds by a vacancy diffusion mechanism, with the driving force
being an imposed chemical potential gradient. Another example is YSZ doped with an
oxide of a metal that has both ion and electron conductivity, such as T1O2 or CeC^ 6 6
The concept of the dense oxide membrane reactor for partial oxidation is as
follows: Air is flowed on one side of a membrane device that consists of a macroporous
support and a dense oxide layer that can selectively permeate oxygen. Hydrocarbon is
flowed on the other side. The permeation of oxygen through the oxide lattice provides a
controlled supply of lattice oxygen to the other side where the catalyst and flowing
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
362
hydrocarbon are located. The catalyst is either the permsdective oxide layer itself or is
another material that is deposited in participate form on top of the membrane layer. The
key to the idea is that the lattice oxygen may be the more selective form of oxygen for
hydrocarbon partial oxidation reactions.67*68 On the other hand, gas phase oxygen and
adsorbed oxygen react with the hydrocarbon to form undesirable carbon oxides in addition
to the partially oxidized hydrocarbon. Thus, if the supply of lattice oxygen can be
properly balanced with its rate of consumption to the desired species, this approach
enables a sustained oxidation of the hydrocarbon without mixing of hydrocarbon and air.
substrates were in the form of flat circular plates with a diameter of 39 mm. Microprobe
analysis indicated that most of the vanadium was located within the Y-AI2O3 top layer.
Membrane layers were prepared in both supported and non-supported forms for studies
using either the membrane reactor configuration or in a packed-tube configuration. A
stagnation flow configuration was used to feed gas mixtures to both sides of the
asymmetric membrane.
The experimentally observed product distribution from integral operation of a
packed tube of A^Cb-VxOy catalyst indicated the following reaction network:
Both acetaldehyde and CO (and water) co-produced at low ethylene conversion. As the
ethylene conversion (i.e., temperature) was increased, acetaldehyde and CO were fully
oxidized to CO2 and water.
Membrane reactor tests consisted of two flow types. In the mixed feed (called
configuration A) experiments, a mixture of ethylene, oxygen, and helium with a prescribed
composition were fed to the active layer side of the membrane reactor, while pure helium
was fed to the support side. In the segregated feed (configuration B) experiments, a
mixture of ethylene and helium were fed to the active layer side, while a mixture of oxygen
and helium were fed to the support side. Comparisons of the two flow types were made
by fixing the overall feed gas composition while varying the furnace temperature over a
range sufficient to span a representative range of ethylene (or oxygen) conversions.
An analysis of the results from the mixed-flow experiments revealed a product
distribution similar to that obtained for the packed-tube. At low temperature, the major
products were CO and acetaldehyde. At high temperatures, the major product was CO2.
The product distribution from the segregated feed experiments was similar. Both CO and
acetaldehyde were produced at lower temperatures, giving way to CO2 at higher
temperatures. Interestingly, the maximum selectivity to acetaldehyde was approximately
65% in the segregated feed experiments. This was considerably higher than the
corresponding value in the mixed flow runs (ca. 20%). But, this maximum occurred at a
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
higher temperature, (ca. 310 °C), than that obtained for the mixed feed (ca. 220 °C). The
ability to partially oxidize ethylene to acetaldehyde depends on both the intrinsic
catalytic chemistry and kinetics and the interaction between the chemical and physical
rate processes. It is the latter interaction that is impacted by the use of reactant
segregation with this nonpermselective membrane reactor. The observed higher
acetaldehyde selectivity, at a fixed conversion level of ethylene and overall contact time,
demonstrates the principle of reactant segregation. By controlling the flux of oxygen to
the active layer, a higher ethylene-to-oxygen ratio is maintained. In effect, a diffusion
limitation (for oxygen) is created with the reactant segregation. The oxygen diffusion
limitation is beneficial to acetaldehyde selectivity, but is detrimental to overall rate. Thus,
while a higher catalyst temperature is needed to achieve the desired ethylene conversion, a
higher acetaldehyde selectivity is achieved. The same result might be obtained with the
conventional fixed-bed reactor by maintaining a high ethylene-to-oxygen ratio along the
entire reactor length.
REACTANT
RECYCLE \ SOUPS FEED
REACTOR
REACTANT STRIPPING
FEED UNIT
Consider the case of a single reaction A -> B. The catalytic reaction occurs on the
catalyst particles, yielding a mixture of A and B. In general, components that are more
strongly bound to the adsorbent particles move downward with the solid phase.
Components that bind weakly to the solids flow upward in the gas stream. In an
idealized situation, weakly-bound reactant A flows out the top of the bed, while more
strongly-bound reaction product B exits the bottom of the column with the solid. Thus,
reaction and separation are effected in a single unit. This concept has been predicted in
simulations by Viswanathan and Aris 69 and demonstrated in experiments by Takeuchi
and Uraguchi. 70 Other theoretical studies of the chromatographic reactor have
71 73
appeared. "
More intriguing is the ability of the chromatographic reactor to increase the
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
conversion in an equilibrium-limited reaction or the selectivity of an intermediate in a
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
365
consecutive reaction system.74 The former concept has been experimentally demonstrated
by Fish and Carr75 using the hydrogenation of 1,3,5-trimethylbenzene to 1,3,5-
trimethylcyclohexane on supported Pt as the test reaction system. In reference to the
latter, Takeuchi et al. 76 have shown that the selectivity of an intermediate product in a
first-order consecutive reaction system in a chromatographic reactor can be increased over
selectivity levels obtained in a conventional fixed-bed reactor. This concept has direct
relevance to selective oxidation reactions.
In their recent study, Tonkovich et al. 76 demonstrated the viability of the
simulated countercurrent moving-bed chromatographic reactor (SCMCR) in the methane
coupling reaction to ethane and ethylene. The SGMCR consists of a tandem of several
beds of stationary solids with a moving feed. This configuration "simulates" a moving
bed. In the methane coupling experiment the catalyst (S1112O3) and adsorbent (activated
charcoal) were contained in separated columns because of the vast difference in operating
temperatures (900 to 1100 K for the former and 373 K for the latter). Considerable
improvement in C2 yields and methane conversion were achieved over levels obtained in
fixed-bed or fluidized-bed reactors. More specifically, methane conversion exceeded 60%
and C2 yield of >50%. These figures compare to typical C2 yields of 20% in
conventional reactors. Three key operating advantages were achieved in the study.
First, suppression of C2 intermediate product oxidation to carbon oxides was achieved by
a chromatographic separation of the C2 species from the oxygen. By binding more
strongly to the charcoal, the C2 species were retained in the charcoal for a longer time
period than the methane or oxygen. Second, better contacting of the methane with the
catalyst enabled higher methane conversions. Third, the C2 product stream was free of
both oxygen and methane.
The results of Tonkovich et al. 76 bring to question whether other partial oxidation
reactions can be carried out in a chromatographic reactor with improved performance. It
would also be useful to compare the performance of the chromatographic reactor and
membrane reactor in carrying out a selective oxidation reaction as part of an effort to
identify the preferred reactor type.
--`,```,,`,`,`,,,`,``,,,`````,,-`
References
14. J. D. Paynter., J. S. Dranoff, and S. G Bankoff, Ind Eng. Chem. Process Des. Develop.
10(1971)244.
15. G. K. Boreskov and Y. S. Matros, Catal. Rev.Sci. Eng. 25 (1983) 551.
16. G. K. Boreskov and Y. S. Matros, in Recent Advances in the Engineering Analysis
of Chemically Reacting Systems, ed. L. K. Doraiswamy (Halsted Press, Wiley,
New York, 1984).
17. Y. S. Matros, A. Noskov, V. Chumachenko, and O. Goldman, Chem. Eng. Sci.
43(1988)2061.
18. G. Eigenberger and U.Meken, Int. Chem. Eng. 34 (1994)4.
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
A. W. Sleight, in Studies in Surface and Catalysis, Vol. 38, ed. J. W. Ward (Elsevier,
Amsterdam, 1987), pp. 645 - 654.
43. R. M. Contractor, U. S. Patent 4,668,802 assigned to DuPont, May 26,1987.
44. R. M. Contractor, U. S. Patent 5,021,588 assigned to DuPont, June 4,1991.
45. V. T. Zaspalis and A. J. Burggraaf, in Inorganic Membranes Synthesis,
Characteristics and Applications, ed. R.R. Bhave (Van Nostrand Reinhold, New York,
1991) Chapter 7, p. 177.
46. J. Armor, Appl. Catal. 49 (1989) 1.
47. P. J. Gellings and H. J. M. Bouwmeister, Catal. Today 12 (1992) 1.
48. H. P. Hsieh, Catal. Rev.Sci. Eng. 33 (1991) 1.
49. T. T. Tsotsis, R. G. Minet, A.M. Champagnie, and P. K. T. Liu, in Computer-Aided
Design of Catalysts, eds. E. R. Becker and C. J. Pereira (Marcel Dekker, New York
1993).
50. M. P. Harold, C. Lee, A. J. Burggraaf, K. Keizer, V. T. Zaspalis, and R. S. A. de
Lange, MRSBulletin, Vol. XDC (April 1994) 34.
51. H. J. Sloot, Ph.D. Thesis, University of Twente, Enschede, the Netherlands (1991).
52. H. J. Sloot, C. A. Smolders, W. P. M. van Swaaij, and G. F. Versteeg, AIChEJ. 38
(1992) 887.
53. V. T. Zaspalis, W. van Praag, K. Keizer, J. G. van Ommen, J. R. H. Ross, and
A. J. Burggraaf, Appl Catal. 74 (1991) 249.
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
369
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
and giant metal clusters are interesting in this context as they constitute intermediate
stages between monomeric (oxo)metal centers and metal (oxide) surfaces,
respectively.
Traditional interdisciplinary barriers separating heterogeneous, homogeneous
and biocataiysis are gradually disappearing. Hopefully this course has contributed to a
further stimulation of this process. In this context we note that a future course on this
subject should include some discussion of oxidation catalysis by metalloenzymes in
order to promote a further cross-fertilization of ideas.
Looking to the future, we note that the potentially attractive 'dream'
reactions, such as direct catalytic epoxidation of propylene with dioxygen and methane
to methanol conversion, remain extremely elusive goals. Nevertheless they will
continue to be the focus of considerable research effort in the future and there is still
a definite need for new approaches to the activation of dioxygen.
We expect that catalytic oxidations, particularly in the liquid phase, will find
broad applications in fine chemicals manufacture, using dioxygen, hydrogen peroxide
and alkyl hydroperoxides as primary oxidants. In this context redox molecular sieves
and related solid catalysts are likely to find widespread application in liquid phase
oxidations. We also expect that significant advances will be made in the area of
asymmetric catalytic oxidation using biomimetic catalysts or even (modified)
biocatalysts themselves.
Finally, a prerequisite for optimum performance in catalytic processes is a
proper choice of reactor configuration. Consequently, we predict further advances in
371
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
372
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
373
Copyright World Scientific Publishing Co.
Provided by IHS Markit under license with WILEY Licensee=FLUOR - 7 - Mexico, D.F. Mexico/2110503118, User=Villalobos, Gerardo
No reproduction or networking permitted without license from IHS Not for Resale, 11/16/2017 06:42:59 MST
374
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---
2,6-Di-tert-butylphenol, Ethene oxide,
oxidation of, 180,190 from ethylene,4,315,316
DuPont process, Extrafacial reactions, 20,22
for tetrahydrofuran production, 351-
353 Fermi level, 24
Fine chemicals,
Eastman-Kodak process, vs bulk manufacture, 239
for terephtalic acid, 7 Fixed bed reactors, 295-311
Electrodes, adiabatic, 298-311
schematic illustration, 143 modeling issues, 296
structure of teflon bonded, 142 Flavin-dependent oxygenases, 12
teflon bonded, 143 Fluidized-bed reactors, 322-339
Electrophilic oxidation, 17,19,21 vs transport-bed reactors, 356
mechanism, 24 Formaldehyde,
Electrophilic oxygen, 25,26 by methanol oxidation, 83
formation at oxide surface, 25,26 formation, 28
Electrosynthesis, from methanol, 307-309, 314
indirect, 269 Fuel cell,
Energy, alkaline, 141
of C-H and OH interaction, 31 direct methanol, 146
of the benzene + 02 system, 30 low-temperature, 141
Epoxidation, molten-carbonate, 147
mechanism, 88 phosphoric acid, 143
of ethylene, 84 schematic illustration, 13 8
silver catalyzed, 80-83 solid oxide, 148
Epoxidation with dioxygen, solid polymer electrolyte, 144
mechanism of, 80-83 Fungicides,
Ethane oxidation, synthesis, 274
elementary steps, 114
Ethane, Gas vs liquid phase oxidation, 160, 176, 240
dehydrogenation to ethene, 110-114 Glyoxal,
selectivity to ethene conversion, 73 oxidation to glyoxylic acid, 273
Ethene, Glyoyilic acid,
adsorbed state, 89 from glyoxal, 273
by dehydogenation of ethane, 110- from methyl acrylate, 191
114
kinetics of oxidation, 225-231 HCN, synthesis, 97-99, 301, 302
oxidation to vinylacetate, 83-85 mechanism and kinetics, 102-105
oxidation to ethylene oxide, 4 Heterogeneous vs homogeneous catalysis,
oxidation to acetic acid, 3 177, 178
Spill-over, 44 172
Spin conservation, Toray process,
in reactions of dioxygen, 12,17 for terephtalic acid, 7
Structure sensitivity, 38 Transition metal surfaces,
of Mo03, 59 reactivity, 86
of oxidation reactions, 58 Transport phenomena, 130
Sulfur trioxide, Transport-bed reactors,
from sulfur dioxide, 305,306 vsfluidized-bedreactors, 356
Sulfur dioxide, 2,3,6-Trimethylphenol,
to sulfur trioxide, 305,306 oxidation to the benzoquinone, 256
Supported metal catalysts, Triphenyteiethyl hydroperoxide,
oxidative dehydrogenation, 179,180 decomposition over chromium
Supported metal ions and complexes, 180 catalysts, 194
Supported oxometal catalysts, 181
Surfaces, reactivity of transition metal, 86 Vanadium oxide,
Swiss roll cell, 275 supported on Ti02,72
Synergy, Vanadium-pillared montmorillonite, 198
in multi component catalysts, 68 Vanadyl pyrophosphate, 59,60
in oxidation of isobutene, 44 adsorbtion of butane, 62
of catalytic properties in oxides, 42- mechanism of oxidation of n-butane
45 over, 60
Syngas, surface structure, 61
from methane, 109,110 Vinylacetate,
generation, 97,99 from ethylene oxidation, 83
Vitamin C,
Takasago process, 251 industrial production, 276
TerephthaMe acid, 3 Vitamin K3,
from p-xylene, 3,163 from 1-naphthol, 255
manufacture, 7
via Amoco process, 165-173 Wacker process, 13,67,83,256,272
Tetrahydro&ran, for acetaldehyde, 3
hydroxylation, 280 Wetting, of oxide surfaces, 46
Titanium silicalite, 10,68,184-187,244,248
mechanism of oxygen transfer, 187 o-Xylene,
oxidations catalyzed by, 184 oxidation to phthalic anhydride, 4,7,
Toluene, 38,39,317-319
automation to benzoic acid, 163 p-Xylene,
mechanism of oxidation, 37 oxidation to terephtalic acid, 3,7,
oxidation on vanadium oxide, 35 163
oxidation to benzoic acid, 7 via Amoco process, 165-173
oxidation to benzyl acetate, 224
selective oxidation, 35,36 Zeolitic materials,
Toluenes, as oxidation catalyst, 67
catalytic autoxidations, 163
Co/Mn/Br catalyzed autoxidations,
--`,```,,`,`,`,,,`,``,,,`````,,-`-`,,`,,`,`,,`---