You are on page 1of 50

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/239525018

p-Xylene Oxidation to Terephthalic Acid: A Literature Review Oriented toward


Process Optimization and Development

Article  in  Chemical Reviews · June 2013


DOI: 10.1021/cr300298j · Source: PubMed

CITATIONS READS

227 9,782

3 authors:

Rogério Tomás João Bordado


Avantium Technical University of Lisbon
3 PUBLICATIONS   309 CITATIONS    224 PUBLICATIONS   5,316 CITATIONS   

SEE PROFILE SEE PROFILE

Joao F Gomes
Instituto Politécnico de Lisboa
267 PUBLICATIONS   2,333 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

ACT | Projeto nº 122APJ/11 - Fumos de Soldadura - Avaliação de Nanopartículas Emitidas e a sua Influência na Saúde dos Trabalhadores Expostos. View project

Indoor air quality regulation through the usage of eco-efficient mortars: INDEEd - INDoor Eco Efficient View project

All content following this page was uploaded by Joao F Gomes on 11 May 2020.

The user has requested enhancement of the downloaded file.


Review

pubs.acs.org/CR

p‑Xylene Oxidation to Terephthalic Acid: A Literature Review


Oriented toward Process Optimization and Development
Rogério A. F. Tomás,† Joaõ C. M. Bordado,‡ and Joaõ F. P. Gomes*,‡,§

ARTLANT, Zona Industrial e Logística de Sines, Zona 2, Lote 2E1, Monte Feio, 7520-064 Sines, Portugal

Instituto de Biotecnologia e Bioengenharia/Instituto Superior Técnico (IBB), Universidade Técnica de Lisboa, Av. Rovisco Pais,
1049-001 Lisboa, Portugal
§
Instituto Superior de Engenharia de Lisboa (ISEL), Á rea Departamental de Engenharia Química, R. Conselheiro Emídio Navarro,
1959-007 Lisboa, Portugal

*
S Supporting Information

5.6.1. Acetic Acid Derivatives AE


5.6.2. Benzaldehyde and Benzoic Acid Deriv-
atives AF
5.6.3. Terephthalic Acid Derivatives AG
5.6.4. Phenol Derivatives AG
5.6.5. Biphenyl and Benzophenone Deriva-
tives AG
5.6.6. Anthraquinone and Fluorenone Deriva-
tives AH
5.6.7. Ester Derivatives AH
5.7. Catalyst Deactivation AH
5.7.1. Catalyst Regeneration AI
6. Operational Optimization Studies AI
CONTENTS 7. Recent Research Trends in Terephthalic Acid
Synthesis AL
1. Introduction A 7.1. Oxidation in Sub- and Supercritical Water AL
2. Polyester History B 7.2. Alternative Catalyst Systems and Cocatalysts AM
2.1. AMOCO Process B 7.2.1. Zirconium AN
3. Chemistry Behind the Process C 7.2.2. Carbon Dioxide Use as Co-oxidant AO
3.1. General Aspects of Hydrocarbon Oxidation C 7.2.3. N-Hydroxyimides AP
3.2. p-Xylene Autoxidation to Terephthalic Acid E 7.2.4. Guanidine AQ
3.3. Catalyst Influence in Autoxidation F 7.3. Oxidation in Ionic Liquids AR
3.3.1. Cobalt and Manganese Acetates As 7.4. Heterogeneous Systems AS
Oxidation Catalysts F 8. Conclusions AT
3.3.2. Cobalt/Manganese/Bromide Catalyst Associated Content AT
Systems I Supporting Information AT
4. Activity P Author Information AT
4.1. Temperature and Pressure Effect on Activity P Corresponding Author AT
4.2. Water Effect on Activity P Notes AT
4.3. Catalyst Concentration Effect on Activity R Biographies AU
4.4. Cobalt/Manganese and Bromine/Metals Ra- Acknowledgments AU
tios Effect on Activity R References AU
4.5. Agitation Effect on Activity S
4.6. Alkali Metal Effects on Activity S
5. Selectivity T
5.1. Carbon Oxides U
1. INTRODUCTION
5.1.1. Carbon Oxides: Solvent Burning U Terephthalic acid is an aromatic carboxylic acid core to
5.1.2. Carbon Oxides: Mechanism of Solvent polyester fibers production. A large majority of the production
Burning W of terephthalic acid is via aerobic catalytic oxidation of p-xylene
5.2. Methyl Bromide Y with air in acetic acid medium, catalyzed by cobalt, manganese,
5.3. Methane Y and bromide compounds, in a process commonly known as the
5.4. Methyl Acetate Z
5.5. Selectivity Assessment AA Received: July 25, 2012
5.6. Byproduct Chemistry AC

© XXXX American Chemical Society A dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

AMOCO process. This work aims to be a review of the


chemistry fundamentals and kinetics behind the process of
catalytic oxidation of p-xylene, providing insight into the most
important aspects relevant for industrial terephthalic acid
production operation and optimization and also on the most
recent research aiming at alternatives to the existing dominant
aerobic catalytic oxidation with the cobalt−manganese− terephthalate, which readily esterifies to dimethyl terephthalate.
bromine catalyst in acetic acid medium. The two oxidations could be combined so that p-xylene and
In the first section, a brief evolution of the industrial process methyl p-toluate were oxidized in the same vessel and so could
of terephthalic acid is given, followed by general aspects of be the two esterification reactions. Currently, direct liquid-
hydrocarbon oxidation, relevant to the chemistry behind the p- phase catalytic oxidation processes, consisting of homogeneous
xylene oxidation process and the difficulties inherent to the liquid-phase oxidation of p-xylene in a solvent with air in the
oxidation reaction that conditioned evolution of the catalysts presence of a transition-metal catalyst, is the dominant
used until discovery of the cobalt/manganese/bromine industrial process for terephthalic acid manufacture. There are
AMOCO catalyst, together with the respective kinetic variations of this process depending on the operating
mechanism. In the second section, factors that have influence conditions and catalysts used, but from these, the Mid-Century
the activity of the catalyst are discussed focusing on process process, also known as the AMOCO process, is the most
parameters that have more impact on industrial process yield widespread used technology.3
and their optimization. In the third section, selectivity aspects 2.1. AMOCO Process
are presented with particular emphasis on the byproducts and The drive behind the research effort to direct oxidation has
the direct influence of the catalyst composition and their been the relatively high resistance to oxidation of the p-toluic
formation. A brief review of available references on operational acid which was first formed. This obstacle was overcome by the
optimization studies is made in the fourth section. Finally, in discovery of bromide-controlled air oxidation in 1955 by the
the fifth section, recent research in p-xylene oxidation is Mid-Century Corp.4−7 and ICI, with the same patent
reviewed, presenting alternative catalysts, solvents, and application date. The Mid-Century process was bought and
promoters and also heterogeneous catalysis studies, which all developed by Standard Oil of Indiana, later AMOCO, with
aim to be alternatives to the presently used dominant some input from ICI. The process uses acetic acid as solvent,
technology. oxygen from compressed air as oxidant, a temperature of about
200 °C, and a combination of cobalt, manganese, and bromide
2. POLYESTER HISTORY ions as catalyst. Various salts of cobalt and manganese can be
Terephthalic acid history is closely related to polyester history. used, and the bromine source can be hydrobromic acid, sodium
Since the pioneering research of Wallace Carothers on bromide, or tetrabromoethane among others. The preferred
condensation of aliphatic dicarboxylic acids with diols to yield system is composed of cobalt and manganese acetates and
polyesters and the first granted patent in 1941 to J. R. hydrobromic acid as bromine source. If a bromoalkane is used
Whinfield and J. T. Dickson, PET manufacture acted as a such as tetrabromoethane, the molecule must solvolyze or be
driving force to terephthalic acid industrial process develop- oxidized readily to HBr. An ionic source of bromide is
ment.1 The resistance of the p-xylene molecule to conversion to necessary to achieve an active catalyst. A feed mixture of p-
terephthalic acid constituted from the beginning, an intrinsic xylene, acetic acid, and catalyst is continuously fed to the
barrier, around which the several technologies developed oxidation reactor. The feed mixture also contains water, a
themselves to bypass that difficulty. One of the earliest byproduct of the reaction. The reactor is operated at 175−225
commercially viable routes to terephthalic acid was through °C and 15−30 bar. Compressed air is fed to the reactor in
the explosive hazardous liquid-phase nitric acid p-xylene excess to provide measurable oxygen partial pressure and
oxidation using 30−40% diluted nitric acid as oxidant at achieve high p-xylene conversion. The reaction is highly
temperatures between 160 and 200 °C and pressures from 8.5 exothermic, releasing 2 × 108 J/kg of reacted p-xylene. Most
to 13.5 bar. Terephthalic acid precipitates are separated and of the terephthalic acid precipitates, as it is formed, due to its
purified in subsequent steps but contaminated with colored and low solubility in acetic acid. This yields a three-phase system:
color-forming impurities.2 To overcome this a diversion tactic solid terephthalic acid crystals, solvent with some dissolved
was used consisting of esterification with methanol to dimethyl terephthalic acid, and vapor consisting mostly of nitrogen,
terephthalate (DMT); successive crystallizations and distilla- acetic acid, water, and smaller amounts of oxygen and carbon
tions were required to yield a product with acceptable quality. oxides. The heat of reaction is removed by solvent evaporation.
For the first few years of PET production, the polymer was all Over 98% of the p-xylene is reacted, and the yield to
made by a DMT ester interchange route.1 The Dynamit− terephthalic acid is greater than 95 mol %. Small amounts of
Nobel process quickly replaced the rather unsatisfactory and p-xylene and acetic acid are lost, owing to complete oxidation
hazardous nitric oxidation route to DMT. p-Xylene was to carbon oxides, and impurities such as oxidation inter-
oxidized with air, in the absence of solvent in the temperature mediates are present in reactor effluent. The excellent yield and
range of 140−180 °C and pressure between 5−8 bar using a low solvent loss in a single reactor pass account for the near
cobalt catalyst, to p-toluic acid, which was esterified by universal selection of this technology for new plants. Oxidation
methanol to form methyl p-toluate, which was then oxidized of the methyl groups is approximately consecutive, with two
by air to monomethyl terephthalate in the same vessel, which in main intermediates p-toluic acid and 4-formylbenzoic acid or as
turn was esterified with methanol to make DMT. it is customarily referred to 4-carboxybenzaldehyde (4-CBA).
The success of this process was that even though p-toluic 4-CBA acid is troublesome, owing to its structural similarity
acid is very resistant to oxidation with a cobalt catalyst, its to terephthalic acid. It cocrystallizes with terephthalic acid and
methyl ester is not; it is rapidly converted to monomethyl becomes trapped and inaccessible for completion of oxidation.
B dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

to acetophenone and acetaldehyde to acetic acid with air in the


temperature range of 102−104 °C for periods of 24 h, catalyzed
by metals. Catalysts used were acetates of cobalt, nickel,
manganese, vanadium, cerium, iron, and chromium. In the
catalyzed oxidation of ethylbenzene yields of acetophenone
were higher than those Stephens11 obtained for the uncatalyzed
oxidation. The basis of catalytic oxidation−reduction action is
electron transfer from the reductant to the catalyst, followed by
The slurry is passed from the reactor to one or more surge electron transfer from the catalyst to the oxidant, which is faster
vessels where the pressure is reduced. Solid terephthalic acid is than direct electron transfer from the reductant to the oxidant.
then recovered by centrifugation or filtration, and the cake is A necessary, but not sufficient, requirement for such catalysis is
dried and stored prior to purification. This is typically referred that the metal must possess two distinct oxidation states, none
to as crude terephthalic acid (CTA) but is more than 99% pure. of them too stable when compared to the other. Oxidation and
Crude terephthalic acid is unsuitable as a feedstock for reduction of saturated molecules frequently results in
polyester, primarily owing to the 4-CBA impurity concen- generation of free radicals. Transition metals, such as Cu, Co,
tration, which acts as chain termination agent in PET and Mn, have the ability to catalyze these oxidations. Among
manufacture, and so goes through a purification process. In such reactions are the autoxidations of hydrocarbons and other
the purification process, crude terephthalic acid is dissolved in organic molecules, initially to hydroperoxides, which proceed
hot water and 4-CBA is reduced and recovered to p-toluic acid by free-radical chain mechanisms in which the most important
via catalytic hydrogenation over a palladium catalyst at about propagation steps are
250 °C and recycled to improve the process yield. There are
also other color-forming impurities and residual amounts of R• + O2 → RO2• (3)
catalyst metals and bromine. The AMOCO purification process
removes 4-CBA to less than 25 ppm and also gives a white RO2• + RH → ROOH + R• (4)
powder from the light yellow CTA feed, commonly referred to
overall: RH + O2 → ROOH (5)
as purified terephthalic acid (PTA). This process allowed
development of a route to polyester from PTA by direct Metals catalyze the oxidation, generating free radicals in
oxidation, which has since become more widely used than the chain decomposition of hydroperoxides according to the
process using DMT. Several other processes for manufacturing following equations.
TA have been patented, and some of them have been used
commercially, but these two remain the most important.1,3 ROOH + Co2 + → RO• + OH− + Co3 + (6)

3. CHEMISTRY BEHIND THE PROCESS ROOH + Co3 + → RO2• + H+ + Co2 + (7)

3.1. General Aspects of Hydrocarbon Oxidation overall: 2ROOH → RO• + RO2• + H 2O (8)
Oxidation is a widely used synthetic route to a large range of followed by
chemicals. Published reports of oxidation studies of alkyl
aromatics with molecular oxygen are from as early as 1912, RO• + RH → ROH + R• (9)
when Ciamician and Silber8 studied the light effect on oxidation The chain reaction is continued by the newly formed
of toluene, o-xylene, m-xylene, p-xylene, and p-cymene. These radical.16 Indeed, there is in the literature accumulated evidence
hydrocarbons were lit by sunlight for a year in contact with that hydrocarbon oxidation follows a free-radical chain
oxygen, obtaining the corresponding carboxylic acids as main mechanism.17 Autoxidation is any oxidation that occurs in the
products. Isophthalic and terephthalic acids were formed presence of air or oxygen forming peroxides and hydro-
besides m- and p-toluic acids from m-xylene and p-xylene, peroxides. A generic commonly accepted mechanism for
respectively. Stephens (1926)9,10 carried out oxidation of hydrocarbon autoxidation found in literature reviews is as
several aromatic hydrocarbons with oxygen at temperatures follows18−21
around 100 °C in the presence of dimmed light. Hydrocarbons
such as xylenes and ethylbenzene were allowed to be in contact initiation step
with oxygen for periods from 24 up to 60 days obtaining for RH + initiator → R• + H ‐ initiator (10)
xylenes monoaldehydes as the major oxidation product and
only a fraction of the corresponding monocarboxylic acids. propagation steps
Stephens also accounted for a mechanism in which hydro- R• + O2 → ROO• (11)
carbon undergoes stepwise oxidation with aldehydes and
ketones as intermediates, excluding alcohols,11and also the ROO• + RH → ROOH + R• (12)
inhibiting effect of water in oxidation.11,12 Evidence for
formation of hydroperoxides as intermediates was found only termination steps
in cyclohexene oxidation.13 Hartman and Seibert succeeded in R• + R′• → R−R′ (13)
peroxide isolation in tetralin oxidation in 1932.14 Oxidation of
hydrocarbon induced by light is not practical for industrial ROO• + R′OO• → ROOOOR′ (14)
purposes due to the long induction times usually involved.
Action of catalysts is therefore necessary. The combined action ROO• + R′• → ROOR′ (15)
of molecular oxygen and metal catalysts is more suitable for RH represents a hydrocarbon from which a hydrogen atom can
industrial use, since the oxidation reaction happens much faster. be abstracted, yielding the free radical R•. ROO• and R′OO•
King et al.15 studied the liquid-phase oxidation of ethylbenzene are peroxide radicals and ROOH a hydroperoxide.
C dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

In the initiation step, the initiator attacks the substrate R• and O2 decreases with increasing resonance stabilization
molecule to generate the chain carrier free radical. At lower energy of the radicals.23 Considering the equilibrium given by
temperatures, a long induction period precedes chain initiation eq 16, for free radicals with a higher resonance stabilization
if the hydrocarbon is very pure or contains any molecules that than the peroxy radical, the left-hand side of the equation has
act as antioxidants that trap or decompose peroxide radicals, increased stability compared to the right-hand side. This means
inhibiting chain reaction. This is observed, for example, in the that the energy difference between R• and O2 reactants and the
oxidation of impure benzaldehyde with oxygen.22 When impure peroxide radical product RO2• is small. The energy barrier that
benzaldehyde is allowed to react with oxygen, an induction the peroxide radical has to overcome to decompose back to R•
period is observed before any oxygen absorption occurs. After a and O2 is small. In a temperatures range from 300 to 700 K, Keq
single distillation procedure, both the induction period and the decreases from the nonresonance-stabilized methyl radical to
maximum rate were reduced considerably. Further distillations the resonance-stabilized benzyl radical. Thus, according to the
eliminate the induction period, and the maximum reaction rate equilibrium expression of eq 17 for Keq, [ROO•] will be smaller
is achieved immediately at the beginning of the reaction. In than [R•].24 Consequently, hydroperoxide formation rates, as
addition, impure benzaldehyde oxidation shows both an per eq 12 will be low. Uncatalyzed oxidation will not be feasible
induction period and a maximum rate greater than that for practical purposes. In the case of olefins oxidation,
observed for the pure substance, which is an indication of the considering eq 12, the dissociation energy of the α C−H
presence of both inhibitors and catalysts, the latter suggested to bond decreases with increasing substitutions at the double
be metallic ions. bond, i.e., olefins with higher substitution are more reactive
In propagation steps, after the highly reactive free radical is
formed it reacts with an oxygen molecule to form a peroxy
radical (eq 11) which, by abstracting a hydrogen atom from
some other molecule, yields a hydroperoxide (eq 12). The
hydroperoxide will decompose in a sequence of further
oxidations to yield molecules with carbonyl and carboxyl
groups. Propagation steps are the product-controlling steps in
the autoxidation sequence. It is suggested in the literature19 that Similarly, α-hydrogen atom reactivity in toluene, ethyl-
radicals of intermediate stability can enter in equilibrium with benzene, and cumene toward a peroxy radical is 0.08, 0.5, and
oxygen 1.0, which is in accordance with the increasing degree of
substitution at the α-carbon atom.19 In benzene derivatives, the
R• + O2 ⇄ ROO• (16)
nature and position of a substituting group in the ring also plays
The equilibrium constant Keq is given by a role in the molecule’s reactivity toward a peroxy radical. A
series of changes in reaction conditions will cause changes in
[ROO•] the rate of a chemical reaction or in the position of its chemical
Keq =
[R•][O2 ] (17) equilibrium. If the series of changes affects rates or equilibrium
of another chemical reaction in the same way minus the effect
This equilibrium is important whenever radicals react with of a given substrate, there will be a Gibbs free energy
oxygen in the presence of the parent hydrocarbon to yield the relationship, ΔG, between the two sets of effects. This
dialkyl peroxide in preference to the hydroperoxide relationship between free energy and equilibrium constants or
ROO• + R• → ROOR (18) reaction rates was first established for the ionization reaction of
benzoic acid derivatives with meta and para substituents to each
The equilibrium between the radical, oxygen, and peroxy other25
radical may give only a low concentration of the peroxy radical
in the presence of alkyl radical, favoring formation of the dialkyl A ⎛⎜ B1 ⎞
−RT ln K + RT ln K 0 = ΔG = 2⎝
+ B2 ⎟
peroxide and not the hydroperoxide. Assuming sequential d D ⎠ (21)
unimolecular kinetics for propagation steps, the propagation
rate, vP, is given by K is a rate or equilibrium constant for a substituted reactant, K0
is the corresponding quantity for the unsubstituted reactants,
vP = kP[ROO•][RH] (19) ΔG is the Gibbs free energy or its kinetic analogue, d is the
distance from the substituent to the reacting group, D is the
The terms in right brackets represent the species concentration, dielectric constant of the medium, and A, B1, and B2 are
and kP is the propagation rate constant. If vP is fast enough, i.e., constants independent of the temperature and solvent.
vP ≫ vT, where vT is the rate of termination, the chain reaction Equation 21 can be rearranged to eqs 22 and 23
is viable and products drop out of each cycle via the
hydroperoxide molecule ROOH. It is easily understood that a log K = log K 0 + σρ (22)
critical issue is then how fast that peroxy radical is formed and
or
reacts with other molecules, as per reaction steps given by eqs
11 and 12, to form the hydroperoxide molecule: if the peroxy log k = log k 0 + σρ (23)
radical stability is low, [ROO•] will be low and hydroperoxide
product formation will be low. The rate of propagation steps Equations 22 and 23 are two simplified forms of what is known
controls the outcome of the autoxidation sequence, with radical by the Hammett equation, where
stability playing a major role. Factors governing radical A
stabilization are the ability of delocalizing electrons through σ=−
2.303R (24)
resonance stabilization and electron availability of the carbon−
hydrogen bond being ruptured. The rate of interaction between and
D dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

1 ⎛⎜ B1 ⎞ the polar effect is that p-xylene is much more reactive than p-


ρ= + B2 ⎟
d T⎝ D
2 ⎠ (25) toluic acid toward a peroxy radical.
σ is a substituent constant dependent upon the substituent, and
ρ is a reaction constant dependent on the reaction medium and
temperature. With ρ values greater than 1, the reaction rate is
more sensitive to electronic effects than is the ionization of Exceptions to these findings exist. When electron-with-
benzoic acids. If 0 < ρ < 1, electron-withdrawing groups still drawing substituents increase the rate or when electron-
increase the rate constant but to a lesser extent than for benzoic donating substituents decrease the rate, when compared with
acids. If ρ is negative, electron-donating groups increase the the nonsubstituted molecule, a higher resonance effect can
rate. Small ρ values, 0 < ρ < 1, often mean that the mechanism explain an enhanced net aralkyl radical stabilization. Any
involves intermediates with little charge separation, such as reactions in which free radicals are consumed, resulting in
radicals. High ρ values, both positive and negative, are molecules that are not free radicals, are the chain termination
consistent with a mechanism in which charged intermediates steps (reactions steps given by eqs 13−15). These will stop the
are present.26,27 The Hammett equation is also applicable to chain and cause the chemistry to stop or be slow. Products
oxidation of alkylaryl hydrocarbons, where polar effects in free- formed through termination reactions may not be necessarily
radical reactions are present.28 Considering the reaction given the same or similar to hydrocarbon being oxidized or its
in eq 12 in the propagation step of the autoxidation, the hydroxyl, carbonyl, or carboxylic derivatives. Alcohols, ethers,
reactivity of a benzene derivative toward a peroxy radical can be or peroxides may be obtained and even hydrocarbons resulting
explained taking into account polar effects and the stability of from condensation of several hydrocarbon radicals. For
the radical formed. Considering cumene, the rate of reaction of instance, as mentioned above, when the peroxy radicals are
its para-substituted derivatives shows a pronounced polar effect, resonance stabilized, reaction with the parent hydrocarbon may
as can be seen in Figure 1. yield peroxides as per eq 18 as well as dimer through
condensation in eq 13. Ketones may also be obtained through
interaction between two primary or secondary peroxide radicals
via a cyclic transition state.19

The effect of inhibitors (antioxidants) on hydrocarbon


Figure 1. ρσ plot for the peroxy radical attack on substituted cumenes. oxidation, with influence also in initiation steps as mentioned
kx and kH are the rates of reaction of the peroxy radicals with a above, also is a factor in reaction termination. This effect will
substituted cumene and cumene, respectively. σ constants from ref 29. depend upon whether a steady supply of free radicals (chain
Adapted from ref 28.
initiators) is being generated in sufficient number to overcome
the chain-breaking action of the inhibitor. The view that the
action of inhibitors is a relative rather than an absolute effect is
From Figure 1, it can be seen that electron-withdrawing substantiated by experimental evidence. For example, com-
substituents, like nitro or cyano, decrease the rate when pounds, such as phenols or alcohols, which also extend the
compared with nonsubstituted cumene and an electron- initiation inhibition period may have little effect on a reaction
donating substituent, like tert-butyl, has the opposite effect on which has reached a steady state.18
rate. Oxidation is favored by increasing electron density at the 3.2. p-Xylene Autoxidation to Terephthalic Acid
reaction site. This has been interpreted in terms of resonance In p-xylene oxidation, disregarding any byproducts formed in
stabilization in the transition state for the reaction between the secondary reactions, methyl groups are oxidized as per the
peroxy radical and the hydrocarbon carbon−hydrogen bond.19 sequence of eq 2.3 The mechanism of oxidation is also through
free radicals as presented in eqs 11−15, and it is in that
mechanism where the limitations lie. Considering attack of an
initiator on a p-xylene molecule, the p-methyl benzyl radical
Polar effects have their influence in resonance structure I, would be obtained as per eq 10.
which is stabilized by electron-donating substituents and
destabilized by electron-withdrawing substituents in R′.
Resonance structure II explains the effects of the stability of
the ROO• radical upon the reactivity of a carbon−hydrogen
bond. Therefore, if both polar and radical stability effects are
present, the reactivity of a carbon−hydrogen bond toward a This radical has the ability to become stabilized through
peroxy radical can be very high. One practical consequence of resonance.
E dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

3.3. Catalyst Influence in Autoxidation


In the presence of small amounts of transition metals the
mechanism given by eqs 11−15 of oxidation is strongly
modified. The metal performs four important functions.16−21
• First, some transition metals participate in the initiation
However, the product of radical interaction with O2, as per eq
steps, abstracting hydrogen atoms from hydrocarbons,
11, the peroxy p-methyl benzyl radical, is not able to achieve
forming free-radical species, and enhancing the rate.
such kind of stabilization. The carbon atom to which the
oxygen atom is bonded has all four possible valences occupied, • Second, metals react quickly with the primary formed
and so the free electron in the peroxy oxygen atom is isolated peroxides via the Haber−Weiss cycle.30,31,18,21
from the ring double-bond-conjugated system, and thus, the
radical cannot form any resonance hybrid, achieving a lower
stabilization. The equilibrium is therefore more favorable

toward the reactants in eq 11 and the peroxy radical


concentration is low, which according to eqs 17 and 19 causes
the propagation rate to be low. As mentioned above, in the
oxidation sequence in eq 2, an even slower step is oxidation of
p-toluic acid (eq 27). Consider the mechanism given by eqs
10−15: in the oxidation of the first methyl group of the p-
xylene molecule, the p-methyl benzyl radical is resonance
stabilized; in the oxidation of the second methyl group, in the
p-carboxyl benzyl radical, the COOH group is an electron-
withdrawing group, in opposition to the electron-donating
methyl group in the first oxidation. This will cause the electrons
of the ring to be attracted toward the COOH group, making
them less available to stabilize the CH2• group through The last step of this scheme, given by eq 37, shows reduction
resonance structures. of the metal catalyst to its initial oxidation state, showing that
decomposition is self-sustained with only small amounts of
metal.
• Third, metals also react very rapidly and selectively with
peracids, formed from benzaldehydes, and also with
higher yields.21
• Fourth, transition-metal complexes have the ability to
coordinate to radicals, such as the peroxy p-methyl
The carboxyl group makes p-carboxyl benzyl radical benzyl, trapping them, preventing them from reacting in
stabilization difficult. The methyl group carbon−hydrogen termination steps that decompose the radical and
bond of p-toluic acid is then strengthened. This is true also in decrease the rate, and driving them forward to
catalyzed oxidations. Oxidation of a methyl group to a products.16
carboxylic group causes the σpara substituent constant to 3.3.1. Cobalt and Manganese Acetates As Oxidation
increase from −0.1 to 0.42, making oxidation of the second Catalysts. Homogeneous transition-metal-catalyzed reactions
methyl group 4.9 times slower. Because p-xylene has the double are widely used in Industrial Chemistry. Hydrogenation,
of oxidizable hydrogen atoms of p-toluic acid, p-xylene is 2 × isomerization, polymerization, carbonylation, epoxydation, and
4.9 = 9.8 times more reactive.21 The deactivating effect of the oxidation are some examples of industrial reactions catalyzed by
carboxylic group is bypassed by converting the acid to a methyl transition metals.32,33 Cobalt and manganese are metals used in
ester as is done in the Dynamit−Nobel process. Ester groups some homogeneous-catalyzed processes. Manganese acetyla-
are electron donating and do not deactivate the ring as cetonate and manganese acetate as well as cobalt bromide,
carboxylic groups do. As a consequence, methyl p-toluate cobalt oleate, cobalt naphthenoate, and cobalt acetate are some
oxidizes to monomethyl terephthalate up to terephthalic acid. examples of the catalysts used. In oxidation, in particular,
Autoxidation of p-xylene to terephthalic has thus two great manganese and cobalt acetates are very important cata-
major constrains: greater stability of the p-methyl benzyl radical lysts.2,34,35 Morimoto and Ogata proposed the following
toward the p-methyl peroxybenzyl radical and p-toluic acid mechanism for substituted toluene oxidation catalyzed by
resistance to oxidation due to the instability of the p-carboxyl cobalt acetate in acetic acid.36
benzyl radical. These constrains make autoxidation a process
too slow for commercial processes. Air oxidation in commercial PhCH3 + Co3 + → PhCH 2• + Co2 + + H+ (38)
processes requires the use of catalysts to achieve higher rates
and yields. PhCH 2• + O2 → PhCH 2O2• (39)

F dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX


Chemical Reviews Review

Figure 2. Product concentration variation over reaction time: symbols, observed; line, calculated. Adapted from ref 43.

PhCH 2O2• + Co2 + → PhCHO + Co3 + + OH− (40)


K′
2Co(III) ⇄ Co(III)2 (49)
PhCHO + Co3 + → PhCO• + Co2 + + H+ (41) K″

Co(III)2 + RH ⇄ {RH+AcO− Co(III)} + Co(II) (50)
PhCO + O2 → PhCO3• (42)
{RH+AcO− Co(III)} → R• + Co(III) + AcOH (51)
PhCO3• + PhCH3 → PhCO3H + PhCH 2• (43)
where AcO− and AcOH represent the acetate anion and acetic
PhCO3• •
+ PhCHO → PhCO + PhCO3H (44) acid, respectively. During hydrocarbon oxidations in the
presence of oxygen the ratio of cobaltic ion to total cobalt
PhCO3H + Co2 + → PhCO2• + OH− + Co3 + (45) tends to reach a steady state value close to 50%. When
cobaltous ion is added, cobaltic ion is regenerated to reach 50%
PhCO2• + PhCH3 → PhCO2 H + PhCH 2• (46) again. This has been an argument to suggest that a dimer
comprising both cobaltic and cobaltous ions may also be an
PhCO2• + PhCHO → PhCO• + PhCO2 H (47) active catalytic species. As mentioned above, cobaltous ion has
been found to have an inhibiting effect; thus, the oxidation rate
PhCHO, PhCO2H, and PhCO3H represent benzaldehyde, is expected to decrease unless catalyst regeneration occurs. A
benzoic acid, and perbenzoic acid, respectively, and PhCO•, possible concerted mechanism for consumption and regener-
PhCO2•, and PhCO3• represent their respective radicals. ation for Co(III) is given by37
PhCH2• and PhCH2O2• represent the benzyl radical and
peroxy benzyl radicals, respectively. Generation of free radicals
can occur either by an electron-transfer mechanism in which an
electron is transferred from the aromatic hydrocarbon to the
metal catalyst with proton loss or hydrogen atom abstraction
from peroxy radicals, which is a propagation step. An induction
period may be observed or not depending on the amount of
cobaltic species (Co3+) present in the catalyst. It is observed
The same type of concerted mechanism may account for the
that it is suppressed by addition of benzaldehyde, which has the
activity of manganese as cocatalyst for cobalt
effect of inducing oxidation of part of the catalyst from the
cobaltous (Co2+) to the cobaltic state (Co3+).36,37 Evidence of RH + O2 + Mn(III)− Co(II)
the absence of an induction period for aromatic hydrocarbons
can be found in the literature.38−40 The induction period is due → ROO− + Mn(II)− Co(III) + H+ (53)
to the low concentration of cobaltic ion and can be understood
as the time required for the inhibiting cobaltous ion to be Mn(II)− Co(III) ⇄ Mn(III)− Co(II) (54)
oxidized to cobaltic ion until its concentration is high enough to
initiate hydrogen abstraction from the hydrocarbon as per eq Ichikawa et al.43 proposed a similar mechanism to the one
38.41 This is confirmed by some experimentally established given by eqs 38−46, excluding the reactions given by eqs 44
hydrocarbon oxidation rate laws which are directly proportional and 46, hydrogen abstraction from the aldehyde. The assumed
to [Co 3+ ] and [RH] and inversely proportional to reaction steps are the same as those in eq 2. Their comparison
[Co2+].37−40,42 of the obtained results with the calculations made from the
assumed kinetic mechanism shows good results, as shown in
d[O2 ] [RH][Co3 +]2 Figure 2.
− ∝k In terms of catalyst structure, mixtures of several similar
dt [Co2 +] (48)
multinuclear complexes have been proposed,44 the most
The rate law, second order dependent on [Co3+],38−40,42 common being the proposed catalyst structure of a binuclear
accounts for the postulated pre-equilibrium where a cobaltic μ-dihydroxy complex for both cobaltic and cobaltous acetates,
dimer, (Co(III)−Co(III)), is formed, with this dimer being the Figure 3.43
active species for hydrocarbon hydrogen abstraction in the early Hendriks et al.45 proposed a mechanism where the species
stage of the reaction abstracting a hydrogen atom from the hydrocarbon molecule
G dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

an aldehyde and from acid-catalyzed reduction with Co2+


ions.
• In the Hendriks et al. mechanism Co2+ ion appears to be
more active in generating more nonradical products,
especially in peracid decomposition, and the Co3+ ion
Figure 3. Proposed structure for the catalyst in Teijin process. also appears to be more active, generating both
Adapted from ref 43. nonradicals and radicals from aldehydes.
Hendriks et al.45 found that the oxidation rate of toluene by
does not involve formation of a dimer, but initiation made by Co(III) in anaerobic and aerobic conditions follows the same
Co3+ involves formation of an intermediate radical−cation. rate equations, and the rate constants under different
k1(CH3CO2 H) conditions differ slightly. The product distribution, on the

Co3 + + PhCH3 XoooooooooooooooY {(PhCH3+ )(CH3CO2−)} other hand, was found to be considerably different: in anaerobic
k −1(CH3CO2 H)
conditions products were mainly benzyl acetates and
+ Co2 + + H+ (55) benzaldehydes derived from the several toluene derivatives; in
aerobic conditions products were mainly benzaldehydes and
Co3 + + {(PhCH3+•)(CH3CO2−)} benzoic acid. The finding that hydrocarbon oxidation by
benzylperoxy radicals is not a reaction of any significance led to
⇄ Co3 +{(PhCH3+•)(CH3CO2−)} (56) the conclusion that aerobic oxidation is not a radical chain
reaction, in opposition to the typical free-radical chain
Co3 +{(PhCH3+•)(CH3CO2−)} mechanism proposed by Morimoto and Ogata. A similar
→ Co3 + + PhCH 2• + CH3CO2 H (57) conclusion was obtained by Heiba et al.,46 who found that the
cobaltic-catalyzed tertiary alkyl benzene oxidation rate was
CH3CO2 H/H 2O slower than that of primary alkyl benzene. This is in opposition
Co3 + + PhCH 2• ⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯→ Co2 + + PhCH 2OCOCH3
with what was mentioned above for the α-hydrogen atom
/PhCH 2OH + H+ (58) relative reactivity of 0.08, 0.5, and 1.0 for toluene, ethylbenzene,
and cumene in free-radical autoxidations.19 The σ plot data
PhCH 2• + O2 → PhCH 2O2• (59) from Hendriks et al.45for the relation between the hydrogen
abstraction rate from a substituted toluene by Co3+ and the
Co3 + + PhCH 2O2• → Co2 + + PhCHO + 0.5O2 (60) Hammett substituent constants are shown in Figure 4.
Co3 + + PhCHO → Co2 + + PhCO• + H+ (61)
H 2O
Co3 + + PhCO• ⎯⎯⎯→ Co2 + + PhCO2 H + H+ (62)

PhCO• + O2 → PhCO3• (63)


H+
Co2 + + PhCO3• ⎯→
⎯ Co3 + + PhCO3H (64)

PhCO3• + PhCHO → PhCO3H + PhCO •


(65)
+
2H
2Co2 + + PhCO3H ⎯⎯⎯→ 2Co3 + + PhCO2 H + H 2O (66)
Some of the main differences between the two proposed Figure 4. ρσ plot for hydrogen abstraction by Co3+ on substituted
mechanisms are as follows. toluenes. kx1 and kH are the reaction rates for substituted toluene and
• The Morimoto and Ogata mechanism explains benzylic toluene, respectively. σ+ constants from ref 47 Adapted from ref 45.
radical formation by direct hydrogen abstraction by the
cobaltic ion, and in the mechanism proposed by The ρ value found for this series plot is −1.97, a large
Hendriks et al. a radical−cation is formed. The negative ρ. Like in the case of autoxidation of cumene
Morimoto and Ogata mechanism explains aldehyde derivatives previously mentioned, a ρ negative value means that
formation from Co2+ oxidation and peroxide radical electron-donating groups, like the methyl group, increase the
reduction, whereas Hendriks et al. explains aldehyde rate constant and electron-withdrawing groups, like the carboxy
formation through Co3+ reduction and peroxide radical group, decrease the rate constant. However, unlike the case of
oxidation. autoxidation, ρ has a large value, meaning that the in the
• Acids in the Morimoto and Ogata mechanism are formed transition state positively charged intermediates develop, which
from hydrogen atom abstraction from a hydrocarbon is inconsistent with a free-radical mechanism26,46,49,52. In
molecule by the PhCO2• radical, and the Hendriks et al. addition, from Figure 4 it can be seen that p-toluic acid (p-
mechanism explains it by PhCO• oxidation in the carboxy toluene) is less reactive toward oxidation than toluene,
presence of water by reduction of a Co3+ ion and even in the presence of a catalyst. Kashima and Kamiya40 also
hydroperacid decomposition with Co2+ ions. observed that the apparent chain length of p-toluic acid
• Peracids in the Morimoto and Ogata mechanism are oxidation by cobaltic acetate in acetic acid is much shorter
formed exclusively from PhCO3• hydrogen abstraction compared to that of p-xylene. This is related with the increased
from a hydrocarbon molecule, and in the Hendriks et al. oxidation potential arising from the carboxylic group presence,
mechanism it is from PhCO3• hydrogen abstraction from causing electron transfer from the molecule to the catalyst
H dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

Figure 5. Data fitting for product concentration variation over reaction time according to the Hendriks et al. mechanism: symbols, observed; line,
calculated. Adapted from ref 45.

complex to be harder. This feature has been of great as a solvent that no terephthalic acid yield was observed. They
importance in the industrial oxidation of p-xylene to also concluded that Co and Mn action in the Co/Mn/Br
terephthalic acid. The resistance of p-toluic acid toward catalyst was synergistic, other halide salts of Co and Mn gave
oxidation accounts for some reported need for use of larger low terephthalic acid yields, and finally acetic acid was the most
amounts of catalyst in industrial processes, compared with what stable solvent. The main discovery was the ability of bromine to
is usually employed in other autoxidations.48,43 Despite the increase the rate of oxidation.7 Bromine and bromine and
differences in the mechanisms, the oxidation rate in the cobalt show activity toward hydrocarbon oxidation, but the
Hendriks et al. mechanism, Figure 5, follows the same most successful combination has proved to be combination of
proportionality as the one found in the Morimoto and Ogata bromine, cobalt, and manganese, the core catalyst of the
mechanism, given by eq 48, and the fit for periodical variation AMOCO MC method.
of each compound, like the one obtained by Ichikawa et al.43 3.3.2.1. Catalysis by Bromine. Bromine compounds, in
also shows good results. Both mechanisms therefore explain the particular, hydrogen bromide, are known hydrocarbon oxidants,
product formation sequence given by eq 2. according to the mechanism54−56
The catalyst structure, as proposed by Hendriks et al., in
acetic medium is a six-ligand mononuclear complex, rather than HBr + O2 → Br • + HO2• (68)
a dimer structure, based on magnetic susceptibility measure-
RCH3 + Br• → RCH 2• + HBr (69)
ments. 49 Cobaltous acetate can be represented as
CoII(OAc−)2(AcOH)4 and cobaltic acetate by RCH 2• + O2 → RCH 2OO• (70)
CoIII(OAc−)3(AcOH)3. In solution the acetic acid ligands can
be progressively exchanged by water, aldehydes, phenol, and k
carboxylic acids. Octahedral geometry is the usual geometry for RCH 2OO• + HBr → RCH 2OOH + Br • (71)
d-block six-coordinated metal complexes, such as cobalt The peroxide, on loss of a water molecule, yields an aldehyde
complexes, Figure 6.50 (or ketone). Further oxidation yields a carboxylic acid.
Recent research has shown that in acetic acid solution two
forms of mononuclear catalytically active cobaltic complexes
may coexist in equilibrium. Reduction of cobaltic species is
accompanied by oxidation of the solvent and is given by the
following mechanism51
Coα 3 + ⇄ Coβ 3 + → Co2 + (67)
3.3.2. Cobalt/Manganese/Bromide Catalyst Systems.
The resistance of p-toluic acid toward oxidation was overcome
with a successful catalyst breakthrough obtained by Saffer and
Barker 4−6 almost simultaneously with McIntyre and
O’Neill.52,53 The first attempts to increase rates in aromatic
hydrocarbon oxidation were made with manganese bromide
playing the role of catalyst in p-cumene oxidation in acetic acid
medium. The overall oxidation yield was found to be no better
than that obtained using other manganese salts, but practically
no intermediates were found. They also tried all the elements in
the periodic table and found Co and Mn to give the highest
terephthalic acid yields. They then varied the anion of Mn and 3.3.2.2. Catalysis by Cobalt and Bromine. Ravens57 studied
discovered that bromide increased both the yield and the rate the oxidation of p-toluic acid by cobalt and manganese
of oxidation. They then switched to p-xylene as the reagent and bromides in acetic acid. They found a rate law given by
found that Mn/Br gave a 77% terephthalic acid yield while
d[O2 ]
Mn(II) acetate gave a very low yield. Then they established that − ∝ k[Co2 +][NaBr]1/2 [O2 ]1/2
Co/Br gave a 79% yield and that if benzene replaced acetic acid dt (77)

I dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX


Chemical Reviews Review

The reaction rate has a linear dependence on cobalt acetate catalysis in solvent media by either cobaltous bromide or a
concentration at constant sodium bromide concentration and a mixture of a cobaltous salt and inorganic bromide as catalyst. At
square root dependence on sodium bromide concentration the end of the first stage, all ionic bromide disappeared from
with constant cobalt concentration dependence. The rate solution and the bromine is present in the form of p-
dependence on [NaBr]1/2 holds only up to a bromide:cobalt bromomethyl benzoic acid. The second stage is catalyzed by a
ratio smaller than 2, where an inflection point occurs, and cobalt salt and p-bromomethyl benzoic acid with no bromide
beyond it, large deviations occur due to structural changes in ions in solution. The role of bromine in the catalytic oxidation
the cobaltous ion complex. The small initial rate in the absence of hydrocarbons is explained as to be other than that of reaction
of one or the other ion shows that the combined action of both initiation by hydrogen abstraction as postulated by Ravens,
catalysts has a beneficial effect on the reaction rate. The Bawn, and Wright, since formation of hydrobromic acid from
bromide salts as per eq 78 proceeds to a very low extent due to
the low equilibrium constant. In fact, the long induction periods
observed in cobalt-bromide-catalyzed oxidations lead to the
hypothesis that its role is to catalyze propagation steps.59−63
Figure 7 shows the oxygen absorption rate for p-xylene
oxidation catalyzed by cobalt dibromide, CoBr2, and cobalt
decanoate, CoDe2, and sodium bromide.

Figure 6. Octahedral representations of CoII(OAc−)2(H2O)4 (1) and


CoIII(OAc−)3(H2O)3 (2).

dependence of the reaction rate on [Co2+] and [NaBr] found


by independent variation of the catalyst component concen-
trations is further substantiated by the rate dependence on
[CoBr2]3/2, which may also be expressed as [Co2+][NaBr]1/2.
The dependence of the reaction rate on [NaBr]1/2 and [O2]1/2
indicates that both oxygen and sodium bromide are involved in
the initiation step and that chain termination is by a
bimolecular radical−radical reaction. The initiation step by
reaction between HBr and O2, as per reaction in eq 68, arises
from pre-establishment of the equilibrium Figure 7. Oxygen absorption curves in the autoxidation of p-xylene in
acetic acid catalyzed with 10−2 M CoBr2 and with a mixture of 10−2 M
NaBr + CH3CO2 H ⇄ HBr + CH3CO2 Na (78) CoDe2 and 2 × 10−2 M NaBr at 80 °C. Adapted from ref 62.
The proposed mechanism for p-toluic acid oxidation is given
by Cobalt dibromide shows an initial higher rate, but it
decreases after approximately 30 min, showing that catalyst is
PhCH3 + Br • → PhCH 2• + HBr (79) becoming deactivated. The rate of the mixture containing
• • CoDe2 and NaBr is kept constant over a large period. Higher
PhCH 2 + O2 → PhCH 2OO (80) cobalt dibromide concentrations show pronounced induction
periods, which, on addition of sodium bromide, are eliminated.
PhCH 2OO• + Co2 + → PhCHO + OH− + Co3 + (81) This shows that cobalt dibromide is not the active catalyst, but
cobalt acetate bromide, formed by eq 86, is
Co3 + + HBr → Co2 + + H+ + Br • (82)
CoBr2 + NaCH3COO → CoBrCH3COO + NaBr (86)
2PhCH 2OO• → PhCH 2CH 2Ph + 2O2 (83)
In addition, increasing the ratio between sodium bromide and
PhCH 2OO• + PhCH 2• → PhCH 2OOCH 2Ph (84) cobalt increases the rate from a ratio of 0.5 to 1.0 from which
the rate is kept constant, as Figure 8 shows. At a ratio of 1.0, the
2PhCH 2• → PhCH 2CH 2Ph (85) concentration of free HBr is very low and most of the bromide
where Ph stands for is complexed with cobalt. This confirms that the oxidation rate
increase is due to catalysis by cobalt acetate bromide and not by
HBr.62
The catalytic mechanism in the propagation step is given
by62
An important aspect of this mechanism is that p-toluic acid
RH
does not enter in its propagation, reaction propagation being ⎯ R•
ROOH + (Co)2 → radical ⎯→ (87)
through the electron-transfer reaction between a cobaltous ion
and peroxide radical, as given by eq 81. This absence of p-toluic • •
R + O2 → RO2 (88)
acid is accounted for in agreement with the rate law equation
• •
given by eq 77. The role of cobalt is that of decomposing RO2 + RH → ROOH + R (89)
peroxide radical and regenerating the mechanism initiating
species, Br•, through eq 82. Bawn and Wright58 also studied p- RO2• + Co2 +BrH → products + (Co2 + + Br•) (90)
toluic acid oxidation by cobalt bromide in acetic acid medium
and divided the oxidation in two stages. The first stage involved (Co2 + + Br•) + RH → (Co2 +BrH) + R• (91)

J dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX


Chemical Reviews Review

hydrocarbon + dioxygen
catalyst = metal(s)/bromide
⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯→ oxygenate + water
solvent (94)

More than 251 alkylaromatic compounds have been oxidized


using the method described in eq 94.21 The most common
example is oxidation of alkylbenzenes to aromatic carboxylic
acids

Figure 8. Steady state rate of oxidation of p-xylene in acetic acid as a


function of the NaBr/Co molar ratio. Adapted from ref 62. When bromine and manganese compounds are added to a
reaction catalyzed by cobalt compounds, creating a Co/Mn/Br
RO2• + RO2• → products + O2 (92)
catalysts system, five important changes result, indicating that
new catalytic pathways are available.21
(Co)2 represents a dimer cobalt species as postulated above for • Catalyst becomes 16 times more active.
cobalt acetate catalysis,38−40,42 and Co2+BrH represents the • Formation of CO2 and CO from secondary reactions
cobalt acetate bromide, the active species in this catalytic decreases by a factor of 10, indicating a more selective
scheme reaction, accompanied also by a higher activity.
• The steady state concentration of cobaltic ion is reduced
to only 0.6%.
• The maximum temperature of oxidation achievable is
higher than the 150 °C usual for cobalt-catalyzed
oxidations with comparable yields.
• Long induction times are absent, except for highly
purified hydrocarbons.
Although Co/Mn/Br catalyst type is most common in the
Replacement of approximately 20% of cobalt by manganese
AMOCO MC method, there are other catalyst systems with
increases by a factor of 5 the rate of oxidation, showing that a bromide and with one, two, or three metals, Table 1.
strong catalytic synergy exists, due to the ability of manganese The most common catalyst is Co/Mn/Br followed by Co/Br
to react with peroxy radicals. This synergy exists up to a liming and Mn/Br. Cobalt is present in the majority of the catalysts
extent, from where manganese deactivation starts to occur, due to its known ability to rapidly decompose hydroperoxides
Figure 9. via the Haber−Weiss cycle,30,31 the primary products of the
3.3.2.3. Catalyst Breakthrough: AMOCO MC Method. The oxidation process. The new available kinetic pathways for a Co/
corollary of the experiments with inclusion of bromine was the Mn/Br MC catalyst type in an aromatic hydrocarbon oxidation
can be depicted as shown in Figure 10.
conclusion that the critical combination of components
In simple terms, bromine acts as promoter, initiating
required to achieve high efficiency in the oxidation of p-xylene hydrogen abstraction from the hydrocarbon and generating
to terephthalic acid was a source of bromine and a mixture of bromide. Bromide ion is oxidized by Mn3+, yielding Mn2+. Mn2+
cobalt and manganese salts.7 This is the basis of the AMOCO is oxidized back to Mn3+ by Co3+, yielding Co2+, which, by
Mid-Century (MC) method peroxide decomposition, will be oxidized back to Co3+.

Figure 9. Mixing effect of cobalt with manganese in the rate of oxidation of p-xylene in acetic acid. Adapted from ref 62.

K dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX


Chemical Reviews Review

Table 1. Examples of Catalyst Systems Using Bromine and with n = 1 and 2, denoting monomer or dimer, and the square
One, Two, or Three Metalsa brackets denote the ligands in the inner coordination sphere of
the metal, M. Figure 11a and 11b shows suggested structures
one metal Co/Br Mn/Br Co/Br/ Co/Br/Cl
acetate for the left-hand side structures with n = 1 and 2, respectively.
V/Br Cu/Br Ce/Br
two metals Co/Mn/Br Co/Ce/Br Co/Cu/Br Co/Mn/Br/
acetate
Co/Mn/Br/ Mn/Si/Br Mn/Cu/Br Ni/Mn/Br
Cl
Co/Ca/Br Mn/Fe/Br Co/Te/Br Co/Ni/Br
Co/U/Br
three Co/Mn/Zr/ Co/Mn/Ce/ Co/Ce/Zn/ Co/Ce/Zr/Br;
metals Br Br Br
Ni/Mn/Zr/ Co/Mn/Hf/ Co/Mn/Ru/ Co/Mn/Cr/Br Figure 11. Suggested structures for Co/Mn acetate catalyst in
Br Br Br;
anhydrous acetic acid mixtures. M = Co(II), Co(III), Mn(II), Mn(III).
Co/Mn/Fe/ Co/Mn/Mo/
Br Br Adapted from ref 65.
a
Adapted from ref 21.
Addition of water to acetic acid results in acetic acid ligands
The nature of the Co/Mn/Br catalyst in the oxidation of p- being exchanged with aqua ligands
xylene can be very diverse and complex.64 The reaction mixture
[M(HOAc)m (H 2O)m − 4 (OAc)2 ]n
may contain, in addition to Co(OAc)2·4H2O and Mn-
(OAc)2·4H2O other mononuclear species like Co(OAc)Br ⇄ {[M(HOAc)m (H 2O)m − 5 (OAc)]n (OAc)} (97)
and Co(OAc)3, multinuclear species like Co3(O)(OAc)x and
Mn3(O)(OAc)x, and heteromultinuclear complexes like with m ranging from 1 to 4 and decreasing with increasing
Co2Mn(O)(OAc)x or CoMn2(O)(OAc)x. Mononuclear com- water concentration. Figure 12a and 12b shows the ligand
plexes are the predominant species in the initial stages of distribution around the metal in acetic acid/water mixtures.
oxidation: they show the ability to abstract a hydrogen atom At water concentration close to 10%, which is common in
from p-xylene. Some of these species like cobalt acetate industrial processes, the predominant structure has four aqua
bromide, Co(OAc)Br, are very short lived due to their high and two acetate ligands. The proposed structure is shown in
reactivity toward p-xylene, arising from their very high redox Figure 13.
activities. Cluster heteromultinuclear complexes, found in the Addition of hydrobromic acid in anhydrous acetic acid results
later stages, probably play an important role in the oxidations of in acid−base neutralization between HBr and acetate ions,
both p-toluic and 4-carboxybenzaldehyde. Cobalt acetate leading, predominately, to bromine inner-sphere coordination
bromide, Co(OAc)Br, which as previously mentioned is [M(HOAc)4 (OAc)2 ]n + {[M(HOAc)5 (OAc)](OAc)}n
described to be a powerful and active species in aromatic
hydrocarbon oxidation, is often studied in anhydrous acetic + HBr ⇄ [M(HOAc)4 (OAc)(Br)]n
acid.65 In industrial oxidation of p-xylene, the reaction medium
contains 5−15% of water to make the solvent less aggressive + {[M(HOAc)4 (OAc)](Br)}n + HOAc
toward the materials used in the equipment and associated (98)
pipework. In anhydrous acetic acid the predominant species can Most of the bromide is coordinated in anhydrous acetic but
be described through the right-shifted equilibrium rapidly decreases as water concentration increases. If HBr is
+HOAc added with more than 5% water, as is the case for most
[M(HOAc)4 (OAc)2 ]n XoooooooY {[M(HOAc)5 (OAc)](OAc)}n industrial processes, little bromide is coordinated and the
(96) predominant species is the ion-paired bromide salt

Figure 10. Catalytic pathway for MC oxidation of p-toluic acid. Adapted from ref 21.

L dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX


Chemical Reviews Review

Figure 12. Distribution of ligands around cobalt(II) (a) and manganese(II) (b) as a function of water concentration in acetic acid. Adapted from ref
65.

RO2 + [Co(II)− Mn(II)]


H+
XooY ROOH + [Co(III)− Mn(II)] (102)
Peracids react preferentially with metal dimers
RO2 H + [Co(II)− Co(II)]
Figure 13. Suggested structures for Co/Mn acetate catalyst in an H+
acetic acid solution with water concentration of 10%. M = Co(II), XooY ROH + [Co(III)− Co(III)] (103)
Co(III), Mn(II), Mn(III). Adapted from.65.
RO2 H + [Co(II)− Mn(II)]
{[M(HOAc)m (H 2O)m − 5 (OAc)n ]n (OAc)} + HBr H+
XooY ROH + [Co(III)− Mn(III)] (104)
⇄ {[M(HOAc)m (H 2O)m − 5 (OAc)]n (Br)} + HOAc
Ligands were not represented for simplicity.
(99)
3.3.2.3.1. Kinetic Mechanism of the AMOCO MC Method.
The proposed structure for a Co/Mn/Br catalyst in acetic acid In the literature there are several studies describing p-xylene
solution in the presence of water is shown in Figure 14. oxidation, but these were performed at conditions different
from those adopted in industrial processes or the models were
based on empirical data and not a kinetic mecha-
nism.43,45,61,66−68 Recently, Wang et al.69 proposed a stepwise
mechanism through intermediates formed as per eq 105 to
explain the oxidation of p-xylene catalyzed by cobalt and
manganese acetates and hydrobromic acid in acetic acid at
conditions close to those used in an industrial process.

Figure 14. Suggested structures for Co/Mn/Br catalyst in an acetic


acid mixture solution. M = Co(II), Co(III), Mn(II), Mn(III). Adapted
from ref 65.

Bromide is hydrogen bonded to one of the coordinated aqua


ligands.
The homogeneous Co/Mn/Br catalyst is thus not a single
catalyst but a mixture in which several catalytically active
species can coexist. In this sense a catalyst is any structure made The combined effect of Co and Mn is to decompose
when its metals are oxidized by oxidants like peroxy radicals or peroxides to yield the oxygenate product. Mechanism initiation
peracids. In such a mixture the oxidants may react with either a is done by hydrogen abstraction from the hydrocarbon by
monomer or a dimer bromine radicals generated through bromide ion oxidation by
H+ the metals
RO2• + [Co(II)] XooY ROOH + [Co(III)] (100)
Mn 2 + + Co3 + → Co2 + + Mn 3 + (106)

RO2 + [Co(II)− Co(II)] Co3 + + Br − → Co2 + + Br • (107)
H+
XooY ROOH + [Co(III)− Co(II)] (101) Mn 3 + + Br − → Mn 2 + + Br • (108)

M dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX


Chemical Reviews Review

3.3.2.3.1.1. Oxidation of p-Xylene to p-Tolualdehyde (PX Br• + HOOCPhCH 2O2• → HOOCPhCH 2O2 Br (127)
to TALD).
3.3.2.3.1.4. Oxidation of 4-Carboxybenzaldehyde to
Br• + CH3PhCH3 → Br − + CH3PhCH 2• + H+ (109) Terephthalic Acid (4-CBA to TA).
CH3PhCH 2• + O2 → CH3PhCH 2O2• (110) Br• + HOOCPhCHO → Br − + HOOCPhCO• + H+
(128)
CH3PhCH 2O2• + Co2 + + H+ HOOCPhCO + O2 →•
HOOCPhCO3• (129)
3+
→ CH3PhCHO + Co + H 2O (111)
HOOCPhCO3• + Co2 + + H+
CH3PhCH 2O2• + Mn 2 + + H+ → HOOCPhCOOOH + Co3 + (130)
3+
→ CH3PhCHO + Mn + H 2O (112) HOOCPhCO3• + Mn 2 + + H+
Br• + CH3PhCH 2• → CH3PhCH 2Br (113) → HOOCPhCOOOH + Mn 3 + (131)

Br• + CH3PhCH 2O2• → CH3PhCH 2O2 Br (114) HOOCPhCOOOH + HOOCPhCHO


Electron-transfer mechanism initiation between metal and → 2HOOCPhCOOH (132)
hydrocarbon producing a radical cation by reactions given in
eqs 55−57, more common in the absence of bromine, was not Br• + HOOCPhCO• → HOOCPhCOBr (133)
considered. Attack of peroxide radical on hydrocarbon and •
Br + HOOCPhCO3• → HOOCPhCO3Br (134)
solvent molecules was neglected. Also, hydroperoxide for-
mation and decomposition reactions by Co2+ and Mn2+ via the The reaction rates for the most important intermediates, PX,
Haber−Weiss cycle were not considered. The only termination TALD, PT, and 4-CBA, are given by
reactions considered were those between the bromine radical
d[PX]
and the free radicals (eqs 113, 114, 120, 121, 126, 127, 133, and = −r1 = ke.99[PX][Br •]
134). dt (135)
3.3.2.3.1.2. Oxidation of p-Tolualdehyde to p-Toluic Acid d[TALD]
(TALD to PT). = r1 − r2 = r1 − [TALD](ke.105[Br •]
dt
Br• + CH3PhCHO → Br − + CH3PhCO• + H+ (115) + ke.109[CH3PhCOOOH]) (136)

CH3PhCO + O2 → CH3PhCO3• (116) d[PT]
= r2 − r3 = r2 − ke112[PT][Br •]
CH3PhCO3• + Co 2+ +
+ H → CH3PhCOOOH + Co 3+ dt (137)
(117) d[4‐CBA]
• 2+ + 3+
= r3 − r4 = r3 − [4‐CBA](ke.118[Br•]
CH3PhCO2 + Mn + H → CH3PhCOOOH + Mn dt
(118) + ke.122[CH3PhCOOOH]) (138)
CH3PhCOOOH + CH3PhCHO → 2CH3PhCOOH Applying the steady state approximation to all radical and
(119) peracids and considering that the total concentration of
Br• + CH3PhCO• → CH3PhCOBr (120)
radicals, [I•], is approximately equal to a constant, ε, i.e.
4
Br• + CH3PhCO3• → CH3PhCO3Br (121) [I•] = ∑ ki[I•] = ∑ fi ci[Br•] = ε
Product p-toluic acid is yielded through reaction between j=1 (139)
aldehyde and peracid by the Baeyer−Villiger70 oxidation (eq then the general rate law is given by
119).
3.3.2.3.1.3. Oxidation of p-Toluic Acid to 4-Carboxyben- kici
ri = kici[Br •] =
zaldehyde (PT to 4-CBA). λ1c1 + λ 2c 2 + λ3c3 + λ4c4 + ε (140)
Br• + HOOCPhCH3 → Br − + HOOCPhCH 2• + H+ λi are rate constants from the reaction mechanism, and c1, c2, c3,
(122) and c4 are PX, TALD, PT, and 4-CBA concentrations,
respectively. Equation 140 shows that the rate follows a direct
HOOCPhCH 2• + O2 → HOOCPhCH 2O2• (123) proportionality to [Br•], a critical factor to initiate the oxidation
reaction. Bromine radical concentration depends on inter-
HOOCPhCH 2O2• + Co2 + + H+ mediate product concentrations. As these decrease, [Br•]
→ HOOCPhCHO + Co3 + + H 2O (124) increases, increasing the reaction rate. When the intermediate
concentration is very low when compared with ε, ∑j4= 1 λjcj ≪
HOOCPhCH 2O2• + Mn 2 + + H+ ε, the rate is first order to the intermediate, which is found in
several other kinetic mechanisms.37−40,42 The rate law is also
→ HOOCPhCHO + Mn 3 + + H 2O (125) independent of oxygen concentration, which in industrial
processes is fed to the oxidation reactor in excess to achieve
Br• + HOOCPhCH 2• → HOOCPhCH 2Br (126) high conversion of the limiting reactant p-xylene and also of the
N dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

Figure 15. Evolution of reactants concentration as a function of time: (a) batch experiment at 160 °C and 14 atm; (b) semicontinuous experiment at
160 °C and 18 atm. Adapted from ref 69.

intermediates. Figure 15 shows intermediate concentration d[TALD]


evolution as a function of time. = k1[PX] − k 2[TALD]
dt (142)
The results show good agreement between the data from the
batch and semicontinuous experiment and the concentration d[PT]
= k 2[TALD] − k 3[PT]
predicted from eq 140. Figure 15a shows the typical behavior of dt (143)
consecutive reactions and that the methyl groups are oxidized
consecutively, in agreement with the sequence given by eq 105. d[4‐CBA]
= k 3[PT] − k4[4‐CBA]
The concentration of p-toluic acid tends to build-up in the dt (144)
experiments, showing the typical difficulty in oxidizing the On integration with the initial conditions, t = 0, [PX] = [PX]0,
second methyl group of p-xylene, as confirmed by Figure 15b. [TALD] = [TALD]0, [PT] = [PT]0, and [4-CBA] = [4-CBA]0,
The above mechanism and rate law proposed by Wang requires rate laws are given by
regression from experimental data of 9 parameters. The number
of these parameters increases as the number of elementary [PX]
= e−k1t
reactions considered in the mechanism increase. This inevitably [PX]0 (145)
increases the complexity of the mathematical model as it
requires more and more kinetic parameters determined from [TALD] k1
= (e−k1t − e−k 2t )
experimental data fitting regression. Several authors have [PX]0 k 2 − k1 (146)
adopted simplifications.67,68 Sun et al.71 established a radical
chain mechanism that assumes that all participating peroxide [PT] k1k 2e−k1t k1k 2e−k 2t
free radicals have the same reactivity for removing an α- = −
[PX]0 (k 3 − k1)(k 2 − k1) (k 3 − k 2)(k 2 − k1)
hydrogen from any hydrocarbon intermediate substrate shown
in eq 105. It is also assumed that all substrates have the same k1k 2e−k3t
+
initiation rate constants. This assumption in particular is (k 3 − k 2)(k 3 − k1) (147)
cumbersome: historically, development of chemical process
technology in the process of terephthalic acid from p-xylene [4‐CBA] k1k 2k 2(e−k1t − e−k4t )
oxidation encountered resistance in the oxidation of p-toluic =
[PX]0 (k 3 − k1)(k 2 − k1)(k4 − k1)
acid and differences in reactivity have been largely accounted
for, as easily seen from a σ-constant plot. Finally, in Sun’s k1k 2k 2(e−k 2t − e−k4t )
mechanism, differences in the termination rate constants are −
(k 3 − k 2)(k 2 − k1)(k4 − k 2)
small and, therefore, negligible. Nevertheless, the first-order
kinetic model only relies on six rate constants, and the data fit k1k 2k 2(e−k3t − e−k4t )
+
of concentration changes of reactants for different catalyst (k 3 − k 2)(k 3 − k1)(k4 − k 3) (148)
concentrations shows, like in the case of Wang’s mechanism,
good agreement. The ultimate simplification of the kinetics of [TA] = [PX]0 − [PX] − [TALD] − [PT] − [4‐CBA]
p-xylene oxidation is to consider a lumped scheme, as per eq (149)
105, with no initiation, propagation, and termination 72
Li and Li adopted a lumped scheme for characterization of
elementary reactions as in a “classical” reaction chain p-xylene oxidation catalyzed by Co(OAc)2/Mn(OAc)2 in the
mechanism, considering only the most relevant intermediates presence of CoBr2 and MnBr2 at temperatures of 100, 110, and
in the overall reaction pathway. The kinetics can be explained 120 °C and oxygen pressures of 2 and 10 atm. They obtained a
in terms of classical kinetics of a consecutive reaction involving good fit of eqs 145−149 to their experimental results, showing
four steps. Considering the simplified eq 105 scheme, reaction that assuming pseudo-first-order reaction rates for each step of
rates are given by p-xylene oxidation is a reasonable simplification of the reaction
mechanism. Wang et al.73 adopted the same lumped kinetic
d[PX] scheme for the oxidation catalyzed with Co(OAc)2/Mn(OAc)2
= −k1[PX]
dt (141) and HBr in acetic acid in the temperature range from 188 to
O dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

197 °C and oxygen partial pressures from 12 to 40 KPa. The It can be seen that as temperature increases, the conversion
rate law is given by rate of 4-carboxybenzaldehyde to terephthalic acid also
increases. Figure 17 shows the trend of 4-carboxybenzaldehyde
Cj
ri = kj concentration.
4
(∑i = 1 diCi + θ )βj (150)

with β and θ being kinetic parameters. The agreement between


experimental data and model prediction based on the lumped
scheme is satisfactory, being detailed enough to characterize the
distribution of the most important species concentration with
different reaction conditions. Lumped kinetic schemes show
enough detail to give time function concentrations of the main
intermediate species; more detailed mechanisms account for
other issues of great importance in industrial processes like
formation of byproduct and solvent burning into carbon oxides.

4. ACTIVITY Figure 17. Concentration of 4-carboxybenzaldehyde vs time at


In the AMOCO process, oxidation of p-xylene progresses different reaction temperatures. Adapted from ref 69.
nearly to completion: over 98% of the p-xylene is reacted and
the yield to terephthalic acid is greater than 95 mol %. This At all times, the concentration of 4-carboxybenzaldehyde is
presents a problem when monitoring reaction activity in a smaller at greater temperatures. Temperature has the effect of
continuous reactor. A major change in activity does not result decreasing the amount of 4-carboxybenzaldehyde as it
in an easily measured decrease in the level of reactant, as p- increases. In an industrial reactor, temperature is indirectly
xylene is only present in the reactor contents in small amounts controlled by pressure, showing a direct proportionality
as it quickly reacts with oxygen. Monitoring an intermediate between both physical properties.
concentration can be a useful method to monitor the activity. 4.2. Water Effect on Activity
Among the several intermediates, considering only those given
The stoichiometric reaction of p-xylene oxidation with oxygen
by eq 105, p-toluic acid and 4-carboxybenzaldehyde can both
yields two water molecules. Water is produced mainly as a
be, in a first approach, chosen to monitor activity. As expected,
byproduct in the main reaction but also in other chemical
p-toluic acid is present at the highest concentration among
reactions such as oxidation to carbon oxides of both p-xylene
intermediates in the reaction medium due to its known
and acetic acid. Water is separated from acetic acid by means of
resistance in oxidizing the second methyl group of p-xylene.
fractional or azeotropic distillation. In addition, the water
Because it is very soluble in the acetic acid solvent, very little
present in the reaction medium may also come as part of the
precipitates with terephthalic acid. In the purification process, it
feed streams to the reactor, as an impurity or contamination.
is removed from terephthalic acid, because of its solubility in
Concerning the reaction rate, water has a net decreasing
water. Historically, p-toluic acid has not been the major concern
effect,21,48,66 as seen from Figure 18b.
when assessing the AMOCO process terephthalic acid quality.
From Figure 18a, as water concentration increases, relative
Because 4-carboxybenzaldehyde is less soluble in acetic acid
rate decreases, showing that water concentration has a
than p-toluic acid and it has a structure similar to terephthalic
decreasing effect on activity. The decreasing effect is due to a
acid, part of it coprecipitates with terephthalic acid and is
progressive substitution of acetate ligands coordinated to
present as an impurity in the CTA. Therefore, it presents itself
cobalt,48 which occurs as water in the medium increases,65 as
as the preferred intermediate to monitor the activity, even
seen from Figure 19a and 19b. However, the effect of water on
though p-toluic acid can also be used to assess it.
activity may not be so direct and is, and it has, in fact, a subtle
4.1. Temperature and Pressure Effect on Activity effect, which needs to be balanced and accounted for. Water
An Arrhenius plot of the rate constant for the oxidation may have a positive effect on activity, as74 found for peroxide
reaction of 4-carboxybenzaldehyde to terephthalic acid is shown initiation of hydrocarbons like tetralin, cumene, cyclohexene,
in Figure 16. and methylcyclohexane. The protons of the water molecule
interact with hydroperoxide molecules, favoring splitting of the
O−O bond, yielding more radicals. This effect is only visible at
small quantities of water, water having an inhibiting effect at
higher quantities. Partenheimer75 compared the effect of water
in the oxidation of 4-chlorotoluene in anhydrous acetic acid and
acetic acid containing 5% water. The initial rate of oxidation is
higher in anhydrous acetic acid than the rate with 5% water, but
at longer reaction times, the effect is the opposite.
Water may have an oxidation-promoting effect due mainly to
the interaction with catalyst. During oxidation, benzylic
bromides are formed, which are known to be catalytically
inactive forms of bromine. In fact, in anhydrous acetic acid the
Figure 16. Temperature dependence of the rate constant for the initial rates are high, but as benzylic bromides are formed, the
oxidation reaction of 4-carboxybenzaldehyde to terephthalic acid. Data catalyst loses bromine, becoming deactivated, which explains
from ref 69. the decreasing rates of oxidation at longer reaction times. In
P dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

Figure 18. Decreasing effect of water in p-xylene oxidation catalyzed by cobalt acetate: (a) data from ref 48; (b) reaction temperature 120 °C and air
pressure of 10.3 bar. Adapted from ref 21.

Figure 19. Effect of water on (a) Co/Mn/Br-catalyzed oxidation of 4-chlorotoluene and p-toluic acid; (b) molecular oxygen uptake during Co/Mn/
Br-catalyzed oxidation of 4-chlorotoluene. Adapted from ref 75.

acetic acid with 5% water, the initial rates are lower, due to inhibitor conversion, and less oxidation of Mn(II) to Mn(IV),
water inhibition, but at higher reaction times, water hydrolyzes leading to less MnO2 formation and faster Br− oxidation, are
the benzylic bromide, releasing it to regenerate the catalyst, favored at lower water concentrations; hydrolysis of benzylic
increasing the rate. bromide, less formation of aryl formate and inhibiting phenol,
less carboxylic acid metal precipitation, and higher reactivity of
peroxy radicals due to acetic acid H bonding are favored at
higher water concentrations.75 Wang et al.76 found that water
has the effect of decreasing the rates of the oxidation reactions
in the first methyl group and increasing the rates of the
The balanced effect of water may be summarized as follows: oxidation reactions of the second methyl group. Overall, there
reactivity of peroxy radicals of aldehydes and alcohols, catalyst is an optimum water content in the reaction media above which

Figure 20. Reaction rate constant kj as a function of water content and batch reaction time as a function of initial water. Data from ref 76.

Q dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX


Chemical Reviews Review

Figure 21. Experimental and simulated 4-carboxybenzaldehyde concentration at different catalyst concentrations at 191 °C and 40 bar oxygen partial
pressure: scatter, experimental; line, data fitting. Br/metals ratio = 0.7. Adapted from ref 73.

the global reaction rate is decreased, hence activity, and below consistent with the rate law given by eq 48. Figure 21 shows the
which the rate is increased. time evolution of 4-carboxybenzaldehyde concentration in p-
The reason can be explained in terms of changes in the xylene air oxidation in acetic acid with different concentrations
catalyst structure. In anhydrous acetic acid most of the bromide of Co/Mn/Br catalyst.
ions are coordinated to the metal, but this rapidly decreases as Similar observations resulted for p-xylene, p-tolualdehyde,
water increases. When bromide is coordinated directly to the and p-toluic acid. In all experiments, the total amount of
metal, electron transfer from the Br− ligand to the Mn3+ catalyst varied but the Co/Mn/Br ratio was kept constant,
metallic complex center occurs easily. At higher water keeping the catalyst composition identical. Drawing a vertical
concentrations, as aqua ligands progressively substitute bromide line intercepting the 4 curves at a given time, it is easily seen
ligands, the latter are in the outer coordination sphere of the that 4-carboxybenzaldehyde and the other main intermediates
complex, coordinated through hydrogen bonds to aqua ligands concentration decrease as catalyst concentration increases,
(see Figure 14). This increases the resistance of electron showing that the activity of the catalyst has increased, Figure
transfer from Br− to the metal cation, as per eqs 107 and 108, 22.
decreasing the reaction rate. Also, with increasing number of
aqua ligands approach of the organic intermediates and
transient species from entering them metallic inner coordina-
tion sphere of the catalyst is prevented.77 On the other hand,
the oxidation potential of metals is also changed. For the redox
pair Co3+/Co2+, the reduction potential decreases from 1.9 V in
acetic acid to 1.84 V in water, making reduction of the cobaltic
ion to cobaltous ion easier, thus enhancing the reaction rate as
water increases. For oxidation of p-xylene and p-tolualdehyde,
the control steps are Mn3+ reduction by Br− as per eq 108 and
hydrogen abstraction by eqs 109 and 115, respectively.
Figure 22. Intermediate product concentration variation with total
Addition of water increases the resistance of electron transfer catalyst concentration. Catalyst composition kept constant. Adapted
from bromide ion to the metal cation, slowing down the from ref 73.
oxidation rate. However, for oxidation of p-toluic acid and 4-
carboxybenzaldehyde oxidation, for the abstraction step, the
first methyl has been oxidized to the electron-withdrawing 4.4. Cobalt/Manganese and Bromine/Metals Ratios Effect
carboxy group. The electron-withdrawing effect decreases the on Activity
reaction rates of the step given by eq 108 and hydrogen The cobalt/manganese ratio has an overall strong effect on the
abstraction steps given by eqs 122 and 128, respectively. The rate of each main step in the oxidation of p-xylene, but the
carboxyl group makes radical stabilization difficult to occur (see different steps of eq 105 show different sensitivities.78 The two
eq 32), and the control step of both p-toluic acid and 4- last steps, oxidation of p-toluic acid and 4-carboxybenzaldehyde,
carboxybenzaldehyde oxidation processes shifts to generation show a higher rate of sensitivity toward the [Co]/[Co + Mn]
of Co3+, controlled by the reaction given by eq 109 and ratio when compared with the rather insensitive two first steps.
propagation steps given by eqs 123−125 and eqs 129−131, The rate constant shows an increasing trend with increasing
respectively. Addition of water decreases the redox potential of ratio, up to an inflection maximum, from where it starts to
Co2+/Co3+ and Mn2+/Mn3+ and makes the oxidation of Co2+ decrease, showing that a minor quantity of manganese metal is
by Mn3+ as per eq 106 and reactions in the propagation steps required. Because the electron-transfer rate between catalyst
easier, resulting in oxidation rate enhancement. When water and electron depends on the catalyst cluster structure, catalysis
content exceeds the optimal value wH2O, the resistance of activation of the cluster is different at different [Co]/[Co +
electron transfer becomes the control step once again. Overall, Mn] ratios. The optimum ratio, the highest achieved rate at a
the effect of water on rate is a competition between an given ratio, shows a decreasing trend with increasing temper-
inhibiting coordination effect and an electrochemical promot- ature, as shown by Figure 23.
ing effect. At lower temperatures, the optimum ratio is nearly close to 1,
meaning that cobalt is the dominant metal species is the
4.3. Catalyst Concentration Effect on Activity catalyst, which accounts for its well-known higher catalytic
Activity normally increases with catalyst concentration up to a activity when compared with manganese. As temperature rises,
maximum value and then decreases.21 This observation is its effect on promoting activity compensates for the lower
R dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

is required even at low amounts to catalyze the reaction


properly, and its absence has to be compensated with larger
amounts of catalyst. Similar trends on both ratios were found
by Kamiya and co-workers,62 as shown in Figures 8 and 9.
4.5. Agitation Effect on Activity
Brill48 reported that the oxidation reaction of p-xylene appears
to be independent of rapid stirring, although with very slow
agitation both the rate and the yields are strongly affected. As
one may expect, a regime of poor mixing will lead to a decrease
in the contact between the oxygen molecule and the
hydrocarbon radicals.
4.6. Alkali Metal Effects on Activity
Figure 23. Temperature dependence trend of the optimum [Co]/[Co
+ Mn] ratio. Adapted from ref 78. Apart from hydrogen bromide, other sources of active bromine
possible in the AMOCO MC include sodium bromide and
manganese catalytic activity, allowing its content in the catalyst potassium bromate.21 Use of sodium bromide as a substitute for
to increase. This trend is consistent with the observation made hydrogen bromide is desirable in order to minimize corrosion
by Cheng and co-workers, that the [Co]/[Co + Mn] ratio of equipment and pipework. When hydrogen bromide is used
effect on the variation is much more remarkable at lower as the source of active bromine, sodium hydroxide is added to
temperatures. Figure 2479 shows a plot of the variation of the control corrosion in several points of the plant and necessarily
ends in the recycled solvent to the reactor feed lines. Line
flushing between vessels is also made with sodium hydroxide
whenever a line choke occurs. It is therefore important to assess
sodium influence in the p-xylene oxidation chemistry. Jhung
and co-workers,80−82 studied the effect of alkali metal addition
to p-xylene oxidation media by establishing a comparison of the
acceleration and deceleration of oxygen uptake relative to a
baseline with no alkali metal addition.
As Figure 25 shows, deceleration occurs in the early stage of
the reaction and acceleration occurs in the later stage and both
increase with the alkali metal concentration, in the studied case,
potassium. In addition, the deceleration in the beginning of the
reaction is of the same magnitude as the acceleration in the
later phase of it. Also, Jhung and co-workers found that the
Figure 24. Catalyst concentration as a function of the bromine/metals
at constant activity. Adapted from ref 79. effect of potassium addition is higher when the manganese and
bromine contents in the catalyst are low. In fact, the Co/Mn/Br
catalyst system is activated the most when alkali metal addition
total catalyst concentration as a function of the bromine/metals equalizes the atomic concentration of bromine, which,
ratio to keep activity constant, in this case, the same 4- according to Jhung, means the existence of an optimum for
carboxybenzaldehyde in product. It can be seen catalyst its concentration. The acceleration and deceleration effect is
function decreases with decreasing bromine/metals over a due to the alkali metal and not to the basicity of the
ratio of 0.3; increasing the ratio has little effect on activity. companying anion and explained by the influence in the
However, below 0.3, the amount of catalyst required to hold coordination of bromine to the catalyst metals throughout the
the activity constant increases sharply. This observation is Co−Mn−Br oxidation−reduction catalytic cycle shown in
consistent with the known Br role in the mechanism: bromine Figure 10. The alkali metal influence on the catalytic cycle

Figure 25. Difference in oxygen uptake variation with addition of alkali metals throughout the oxidation reaction of p-xylene. Adapted from ref 81.

S dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX


Chemical Reviews Review

Figure 26. Influence of the alkali metal in the Co−Mn−Br catalytic cycle. Adapted from ref 80.

appears to be independent of the functional groups of the 5. SELECTIVITY


molecule being oxidized, which suggest that the interaction is The selectivity of a reaction may be defined as the fraction of
with the catalyst and not with the organic substrate. Figure 26 the starting material that is converted to the desired product. In
shows the influence of the alkali metal in the Co−Mn−Br a broader view, selectivity can be defined between the desired
catalytic cycle. chemistry and the undesired chemistry. Within the scope of p-
In the early stage of the reaction, rate deceleration is caused xylene catalytic oxidation in acetic acid media, one can consider
by interaction of the alkali metal with the bromine ion, binding not only the main and parallel reactions in which p-xylene is
it and diverting it from the reduction of Mn3+ ion in the consumed yielding other products than terephthalic acid but
catalytic cycle. With the progress of the reaction are formed also reactions that consume acetic acid. Degradation of p-xylene
to carbon oxides and to aromatic byproducts and acetic acid
other substances; catalyst poisoning species that also interact
degradation also to carbon oxides, methyl bromide, methane,
with bromine, slowing the reaction. In a later stage of the and methyl acetate are reactions that contribute to an overall
reaction, as the alkali metal progressively interacts with bromine economic yield decrease of a the process. Like in activity
and the poisonous species, the interaction between bromine monitoring, for selectivity evaluation, 4-carboxybenzaldehyde
and the poisonous species will decrease and rate will increase can be used to establish a baseline to assess the competitive
again. chemistry in p-xylene oxidation. In the AMOCO process,
T dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

conversion of p-xylene to terephthalic acid is nearly 100%


complete, so 4-carboxybenzaldehyde as a percentage of
terephthalic acid product is normalized against the p-xylene
feed rate, thus being a good selectivity indication. This allows
byproduct formation also to be normalized against the feed rate
of p-xylene and be used as a comparative measure against 4-
carboxybenzaldehyde as a percentage of terephthalic acid
product in selectivity assessment.
5.1. Carbon Oxides
Complete oxidation of a hydrocarbon yields water and carbon
dioxide. As seen from eq 105, progressive and partial oxidation Figure 27. Carbon oxides formation trend with increasing catalyst
of a methyl group in organic molecules yields a carboxylic acid concentration during p-xylene oxidation catalyzed by a Co/Mn/Br
group. If further oxidation is occurs then decarboxylation of the catalyst at 197 °C. Adapted from ref 79.
molecule occurs, yielding carbon monoxide and carbon dioxide
as products, COx. Decarboxylation occurs not only in the observation is consistent with a need for a balance between
aromatic acids, either intermediate or product, but also in the bromine and metals: metals in their trivalent oxidation state
acetic acid solvent. In acidic medium, acid-catalyzed decarbox- react with bromide ions to form bromide atoms, which will
ylation reactions are dependent on factors such as the ionic abstract hydrogen atoms from methyl groups to yield radicals;
strength and substituents on the aromatic ring.83 The however, competitive degradation of carboxylic groups of both
mechanism is given by product and solvent molecules to carbon oxides is favored by a
high concentration of metals, in particular cobalt, which has a
well-known ability to decarboxylate acids.49,84
Co3 + + CH3CO2 H ⇄ Co2 + + CH3CO2• + H+ (154)

In dilute acid solution, ipso protonation is the rate- Co3 + + CH3CO2•


determining step, whereas in highly acidic solutions, the rate- CH3CO2 H
⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯→ Co2 + + CO2 + CH3CO2 CH3 + H+ (155)
determining step is aromatic carbocation decarboxylation.
Aromatic acid decarboxylation can also occur by a free-radical Thus, as a result of these two competitive reactions, one can
pathway, catalyzed by metals and radicals, though hydrogen expect an optimum Br/metals ratio to achieve greater catalyst
abstraction from the carboxylic group of the molecule. selectivity with respect to carbon oxides formation.
5.1.1. Carbon Oxides: Solvent Burning. Of particular
importance in industrial manufacture of terephthalic acid is
acetic acid solvent degradation to carbon oxides, as this is one
of the greatest parcels of operational costs. Not only does
addition of manganese into the catalyst feed promote oxidation
of the hydrocarbon due to the synergistic effect with cobalt, as
shown in Figure 10, but also it has been found that it decreases
the content of carbon oxides in the oxidation off-gas. 14C
isotopic experiments performed by Kenigsberg et al.85 lead to
the conclusion that acetic acid degradation into carbon oxides,
i.e., solvent burning, has a dependence on the manganese to
cobalt ratio. In addition, they also found the following.
The benzoyloxyl radicals formed rapidly decarboxylate to • Carbon dioxides are formed from both the methyl and
yield aryl radicals. The aryl radicals can then react with several the carboxylic groups of the acetic acid molecule.
other molecules to yield other undesired byproducts, • CO2 is formed in quantities higher than CO: from 3 to
decreasing the overall selectivity. Regarding decarbonylation, 26 times more.
very little carbon monoxide, generally less than 0.01%, is • Cobalt has a stronger ability to form COx from the
formed during the reaction of benzaldehydes described, which carboxylic group.
is indicative that decarbonylation is not important in aromatic • Manganese has a stronger ability to form COx from the
aldehyde autoxidations.77 It may then be expected that most of methyl group.
the CO comes from decarbonylation of acetic acid and not the • Radicals from the main reaction also decompose acetic
benzaldehydes. As may be expected, the effect of catalyst acid, yielding methyl radicals and CO, CO2, formic acid,
increase in the oxidation is to increase COx formation, as shown and formaldehyde, Figure 29.
by Figure 27. There is a minimum, located toward low [Mn]/[Mn + Co]
The effect of catalyst composition in terms of bromine/ ratios, where COx formation can be optimized: low cobalt
metals ratio is shown in Figure 28. content in the catalyst caused carbon oxides formation to
The total amount of carbon oxides moles per feed mole of p- increase, but the same will be expected to happen if the cobalt
xylene shows a minimum for a molar bromine/metals ratio content becomes too high, Figure 30.
close to 0.3. As the ratio falls below the 0.3 minimum In Figure 30, the minimum COx formation corresponds to a
production of carbon oxides increases rapidly. Above the high cobalt content, with a general trend of COx formation
minimum formation of COx shows a less steep trend. This increasing with increasing manganese. However, Cheng78 and
U dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

Figure 28. Carbon oxides formation trend with increasing bromine/metals molar ratio during p-xylene oxidation catalyzed by a Co/Mn/Br catalyst
at 197 °C. Adapted from ref 79.

Figure 29. Carbon oxides degree of decomposition trend dependence with [Mn]/[Mn + Co] ratio for 14CH3COOH and CH314COOH. Adapted
from ref 85.

Figure 30. Total COx formation trend dependence with [Mn]/[Mn + Figure 31. Carbon oxides generation trend dependence with [Co]/
Co] ratio for 14CH3COOH and CH314COOH. Adapted from ref 85. [Co + Mn]. Adapted from ref 78.

Li et al. showed that the formation rate constants of CO2 and


co-workers found the opposite trend at similar catalyst CO vary with process conditions that normally also affect p-
concentration magnitudes and temperature ranges, as seen xylene oxidation. CO2 and CO formation rate constants
from Figure 31. The main difference in the catalyst feed was the increase with increasing [Co]/[Mn] ratio and decrease with
source of bromine. Kenigsberg used NaBr, whereas Cheng used [Br]/[Co] ratio and decreasing temperature. Water concen-
HBr as the bromine source. Yet, the bromine to metals ratio tration has a promoting effect on COx formation rate constants
used by Kenigsberg was almost three times higher, which up to 10−15%, from which it begins to have an inhibiting effect.
suggests that bromine is the differentiating factor, most likely Increasing temperature, air flow rate, and total metals
due to changes in catalyst structure. concentration also increases COx formation rates.86 In later
V dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

Figure 32. Formation rates of (a) CO2 in batch oxidation experiment, (b) CO in batch oxidation experiment, (c) CO2 in semicontinuous oxidation
experiment, and (d) CO in semicontinuous oxidation experiment. Acetic acid/p-xylene weight ratios: (□) 20/1, (○) 10/1, (△) 5/1, (▽) 3/1; ()
model fitting. Adapted from ref 87.

work, Cheng et al. suggested that rates of solvent burning and 1


rCO2 = rCO
the main reaction are related and have a competitive 1−η x (158)
relationship between them.87 The observed rates of COx
formation during p-xylene oxidation in batch and semi- η
rCO = rCO
continuous experiments are shown in Figure 32a and 32b. 1−η x (159)
Data fitting is given by
and η being within the 0.244−0.252 range.
0.791NyCO 5.1.2. Carbon Oxides: Mechanism of Solvent Burning.
rCO2 = 2
Like the main reaction, solvent burning follows a free-radical
Vr{1 − (yO + yCO + yCO )} (156) oxidation mechanism. Throughout the p-xylene oxidation
2 2
several radicals are formed which can abstract hydrogen
atoms from other reactant, intermediate, product, or solvent
0.791NyCO molecules. Among these radicals, RCO• and RCOO• radicals
rCO =
Vr{1 − (yO + yCO + yCO )} will undergo decarbonylation and decarboxylation yielding CO
2 2 (157)
and CO2, respectively
where rCO2 and rCO are the CO2 and CO formation rates, RCO• + R′H → RCHO + R′• (160)
respectively, N is the air flow rate, Vr is the reaction volume,
and yO2, yCO2, and yCO are, respectively, the oxygen, carbon RCO• + O2 → RCOOO• (161)
dioxide, and carbon oxide molar fractions in the reactor off-gas. RCOO• + R′H → RCOOH + R′• (162)
The behavior of the batch experimental curves shows a camel
back like shape: an initial sharp increase of COx to a first peak, RCO• → R• + CO (163)
from which it decreases to a minimum; from this minimum the
rate increases again to a new peak, from which it drops at the RCOO• → R• + CO2 (164)
end of the reaction. In the semicontinuous curves the COx where R represents any hydrocarbon radical and R′H
generation rate increases sharply at the beginning of the represents any molecule with an abstractable hydrogen atom
reaction, after which it shows a relatively stable decreasing including acetic acid
plateau throughout most of the reaction time. At the end of the
reaction the COx formation rate drops sharply when the p- I• + CH3COOH → HI + CH3COO• (165)
xylene feed stops. The fact that CO and CO2 formation rates
show the same behavior either in batch or in semicontinuous I• + CH3COOH → HI + •CH 2COOH (166)
experiments leads to the conclusion that the two rates are •
proportional, i.e., rCO2 = ηrCO, with CH 2COOH + O2 → CO2 , CO, H 2O (167)

W dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX


Chemical Reviews Review

Figure 33. Comparison between solvent burning and the main reaction on (a) batch experiment and (b) semicontinuous experiment. Adapted from
ref 87.

87
Figure 34. Relation of terephthalic acid selectivity and rC̅ Ox in (a) batch experiment and (b) semicontinuous experiment. Adapted from ref .

CH3COO• → CH3• + CO2 (168) reaction and the main reaction is shown in Figure 33 for both
batch and semicontinuous experiments.
I• represents any radical, like the very strong oxidizing radical The results suggest an interstimulative relationship between
OH• or PhCH2OO• but also cobalt or manganese in their both reactions. Considering the scheme in eq 105, the rate of
trivalent oxidation state, originating from the thermal CO2 formation shows an accompanying behavior to both the
decomposition of the respective acetates, which is found to
rates of formation of all intermediates, rall, and formation rates
be the most important.88 Consumption of I in the burning
reaction thus shows a competition relationship between the of p-toluic acid and terephthalic acid, r2 + r4, suggesting a
main reaction and the solvent burning reaction. In the latter, proportionality relation. While Call increases, the opportunity
the methyl radical is of particular importance as it yields some that the reactant and intermediates are attacked by the active
the most important byproducts in the MC-AMOCO industrial free radicals and metal ions increases too and the opportunity
process, such as methyl bromide, methane, and methyl acetate for solvent being attacked decreases reversely. Therefore, the
rate of the burning side reaction, rCO2, will decrease faintly with
CH3• + Br• → CH3Br (169) increasing concentrations of reactant and intermediates, Call.
For batch and semicontinuous oxidations, the rate of CO2
CH3• + RH → CH4 + R• (170) formation is, according to Cheng, given by

CH3• + CH3COO− → CH3COOCH3 (171) α1(r1 + r2)


rCO2 = 4
+ α2(r1 + r2)
(∑i = 1 Ci + ε) (173)
CH3• + O2 → CO2 , CO, H 2O (172)

Comparison between the CO2 generation rate and rates of with α1 and α2 being empirical model parameters and rCO2 =
intermediate reaction as per eq 105 for the solvent burning ηrCO.
X dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

Figure 35. Methyl bromide formation trend at a constant 4-carboxybenzaldehyde with increasing bromine/metals molar ratio during p-xylene
oxidation catalyzed by a Co/Mn/Br catalyst at 197 °C. Adapted from ref 79.

Figure 36. Comparison of methane formation with CO2 and CO formation with increasing metal concentration at constant air flow. Adapted from
ref 84.

From the relationship between the main reaction and solvent decrease, as it formed one undesired product, and of activity
burning, the following COx formation−terephthalic acid decrease. Methyl bromide, a gas in industrial process operating
selectivity relation arises as depicted in Figure 34. conditions, is partially lost in the off-gas process and treated as
In batch oxidation, the burning side reaction rC̅ Ox decreased an effluent. The higher the formation of methyl bromide, the
with the increase of p-xylene/acetic acid mass ratio but lower the amount of bromine present in the catalyst, which has
terephthalic acid selectivity increased. In semicontinuous the effect of decreasing its activity and selectivity.
oxidation, the burning side reaction rate rC̅ Ox increased with Figure 35 indicates that there is an optimum Br/metals ratio
which yields a minimum of methyl bromide at a constant 4-
the increase of p-xylene feed rate but terephthalic acid carboxybenzaldehyde in the product. At low Br/metals ratios,
selectivity remained relatively constant. In semicontinuous the catalyst is not active enough and so a higher catalyst
oxidation, rC̅ Ox is lower than that in batch oxidation and concentration is required to keep 4-carboxybenzaldehyde
terephthalic acid selectivity is slightly higher. One conclusion is constant. However, high absolute catalyst concentrations lead
that the operation mode affects the p-xylene oxidation process, to decarboxylation of solvent and other carboxylic acids present
and therefore, in an industrial continuous reactor there is an in the reactor. This yields more methyl radicals, which will yield
optimal operation mode in which low burning side reaction and methyl bromide. At high Br/metals ratios, the high
high terephthalic acid selectivity can be achieved.87 concentration of bromine favors formation of methyl bromide
5.2. Methyl Bromide as it reacts competitively with the methyl radical, diverting it
from other reactions. Optimal conditions for low methyl
As mentioned above, methyl bromide is formed in side bromide formation and high catalyst selectivity are a balance
reactions during oxidation of p-xylene. Its formation is the between other factors, which includes catalyst concentration
result of a combination of methyl radicals with bromine, as per and composition.
the reaction in eq 169. Methyl radicals come from degradation
of solvent or other aromatic molecules and bromine from 5.3. Methane
catalyst: therefore, methyl radical formation decreases the As one of the byproducts formed from decarboxylation
overall yield and is an example of both catalyst selectivity degradation, methane is formed in eq 170. Methane is insoluble
Y dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

Figure 37. Comparison of methane formation with carbon dioxide formation with increasing air flow/metal ratio. Adapted from ref 84.

in the reaction media and present in the reaction off-gas. As •


CH 2CO2 H + O2
expected, increasing catalyst concentration causes more
decarboxylation to occur and more methyl radicals will be → CO, CO2 , CH3OH, CH 2O, HCOOH (177)
formed, Figure 36. These will abstract hydrogen atoms from
Both methyl radical and methanol can react with acetic acid
other molecules yielding methane. Abstraction of hydrogen
to yield methyl acetate
from methyl groups of p-xylene is favored by the resonance
stabilization of the radical, as per eq 30. CH3• + Mn + 1 + CH3COOH
Less methane, relative to carbon oxide, is seen as the air
flow/metals ratio increases, as shown in Figure 37, and the ratio → CH3COOCH3 + Mn + + H+ (178)
of carbon dioxide/methane increases. As the air flow/metals
ratio decreases, the oxidation media becomes increasingly CH3OH + CH3COOH → CH3COOCH3 + H 2O (179)
starved of oxygen and methyl radicals tend to preferentially
When the metal catalyst is in the form of acetates, the metal
abstract hydrogen atoms from other molecules to form
in its higher oxidation state can oxidize the acetate ligand to
methane as per eq 170 rather than forming carbon oxide as
carbon dioxide and methyl acetate49
per eq 172. A competitive behavior appears to exist. In addition,
one may expect that increasing the degree of agitation within an 2Co(CH3COO)3 → 2Co(CH3COO)2 + CO2
industrial reactor would also contribute to formation of
methane the same way increasing the air flow/metals ratio. + CH3COOCH3 (180)
5.4. Methyl Acetate Methyl acetate formation depends on the process conditions
Methyl acetate is one of the main byproducts in the that lead to acetic acid degradation like water concentration,
terephthalic acid process.66 Its formation can be described as temperature, p-xylene oxidation rate, catalyst concentration,
oxidative decarboxylation of acetic acid89 and composition, as shown in Figures 38 and 39.
The role of water in decreasing methyl acetate formation is
2CH3COOH + 0.5O2 → CH3COOCH3 + H 2O + CO2 due to ligand substitution in the catalyst, decreasing its
(174) activity49,77 by destabilizing the highest oxidation states of the
metals responsible for the decarboxylation reaction and also
Methyl acetate formation in industrial production of due to its role as reactant in methyl acetate hydrolysis to acetic
terephthalic acid is of great importance as it is formed mainly acid and methanol, the latter undergoing oxidation to carbon
by acetic acid degradation, causing solvent loss, which is oxides.
controlled to a certain extent by recovering it in the process and
recycling it to the oxidation reactor, where it is reconverted to CH3COOCH3 + H 2O → CH3COOH + CH3OH (181)
acetic acid by hydrolysis.66,90,91 Decarboxylation is initiated by
peroxide interaction with the carboxylic or methyl groups of Just like the activity behavior of catalyst with water
acetic acid, followed by degradation of the resulting radicals, concentration, methyl acetate formation does not always
yielding carbon monoxide, carbon dioxide, methanol, form- decrease with increasing water. Roffia found the water
aldehyde, and formic acid.85,88,89 Methanol results from concentration optimum to be within the range 12−16%. At
reaction between a methyl radical and water. Oxidation of lower water concentrations, methyl acetate formation is high
methanol initially yields formaldehyde, which upon further mainly due to increased catalyst activity, decomposing solvent
oxidation yields formic acid. Formic acid can esterify with molecules. The increase of p-xylene rate in an oxidation reactor
methanol to yield methyl formate.88 is proportional to its feed rate. Increasing the amount of p-
xylene comparatively increases its competition with acetic acid
CH3CO2 H + ROO• → ROOH + CH3• + CO2 (175) solvent molecules for reacting with catalyst and peroxides,
decreasing methyl acetate formation. An optimized p-xylene/
solvent ratio will contribute to less methyl acetate formation.
CH3CO2 H + ROO• → ROOH + •CH 2CO2 H (176) Like in formation of carbon oxides, increasing the total catalyst
Z dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

Figure 39. Methyl acetate formation trend with varying bromine to


metals ratio. Adapted from ref 89.

bromine/metals ratio decreases COx formation, meaning less


methyl radicals and hence less methyl acetate. Also, bromide
ions, by coordinating to the catalyst metals, replace acetate ions
in the electron-transfer process between ligand and metal in its
high oxidation state, hindering RCOO• acyloxy radicals
formation and promoting formation of bromine radicals,
which are promoters of the main reaction synergistic cycle in
detriment of the methyl acetate byproduct formation secondary
reaction.89
Increasing the relative amount of acetate ion in catalyst
composition also has the effect of increasing methyl acetate
formation.92 As seen above,65 as the catalyst is more acetate
ligand rich, i.e., less coordinated aqua ligands, its activity
increases, yielding more carbon oxides and methyl acetate.
5.5. Selectivity Assessment
As mentioned above, for selectivity evaluation during p-xylene
oxidation, monitoring 4-carboxybenzaldehyde in terephthalic
acid product is a valuable tool. The optimization goal is to have
a high yield of terephthalic acid with a low content of 4-
carboxybenzaldehyde and simultaneously low burning to
carbon oxides. Figure 41 shows CO2 formation vs 4-
carboxybenzaldehyde content in terephthalic acid product at
different temperatures.
The shape of the curves shows competition behavior
between 4-carboxybenzaldehyde content in terephthalic acid
product and unwanted degradation of p-xylene and acetic acid
to carbon dioxide. It can be seen that for all the curves carbon
Figure 38. Methyl acetate formation trends with variation of water dioxide formation increases as 4-carboxybenzaldehyde de-
concentration. (a) Three levels of p-xylene oxidation rate (mol kg−1 creases. Driving 4-carboxybenzaldehyde to low levels requires
h−1); Co2+, 0.022%; 220 °C; stirring speed, 500 rpm. (b) Three levels use of a higher catalyst concentration. This causes carbon
of Co2+ concentration (% wt); 220 °C; p-xylene oxidation rate, 3.8 oxides formation to be higher, as shown in Figures 28 and 36,
stirring speed, 500 rpm. (c) Three levels of temperature; Co2+, in particular, if higher cobalt concentrations are used, as shown
0.022%; p-xylene oxidation rate, 3.8; stirring speed, 500 rpm. Adapted in Figures 30 and 31. As an alternative, for achieving a lower 4-
from ref 89. carboxybenzaldehyde concentration, more severe oxidation
conditions, i.e., higher temperatures are required. This will
concentration also leads to increased methyl acetate formation. cause more carbon dioxide to be formed, as also confirmed
Association of these two trends is explained easily by the when looking at Figure 42, which shows the effect of
mechanism of methyl acetate formation which requires methyl temperature on 4-carboxybenzaldehyde and carbon oxides at
radicals, as per eq 178, which in turn are formed during solvent constant catalyst composition.
decomposition through eq 175. The effect of increasing Typically, for a constant level of burn, an optimal catalyst
temperature increases methyl acetate formation, especially at composition to obtain a minimum 4-carboxybenzaldehyde
low water concentrations. This is explained by the higher exists.97 The trade-off is to achieve terephthalic acid product
catalyst activity at higher temperatures, leading to more methyl with low 4-carboxybenzaldehyde content suitable for PET
acetate formation. Methyl acetate formation also depends on manufacture and simultaneously reduce solvent burning, which
the catalyst composition. As shown in Figure 39, methyl acetate has a detrimental effect on the manufacturing process
formation tends to decrease with increasing bromine/metals economical reliability. The cobalt to manganese ratio and the
ratio. As seen previously, to a certain extent decreasing the bromine to metals ratio are essential 4-carboxybenzaldehyde
AA dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

Figure 40. Carbon oxides and methyl acetate formation trends with varying catalyst acetate to metals ratio during p-xylene and pseudocumene
oxidation. Adapted from ref 92.

Figure 41. Carbon dioxide formation vs 4-carboxybenzaldehyde present in terephthalic acid product at different temperatures. Black curves on the
left vertical axis, and gray curves on the right vertical axis. Adapted from refs 93−95.

Figure 42. Effect of temperature on 4-carboxybenzaldehyde and Figure 43. 4-Carboxybenzaldehyde in terephthalic acid product as a
carbon oxides at constant catalyst composition. Scatter: data points at function of the Co/(Co + Mn) ratio trends at different relative levels
190, 175, and 170 °C. Solid line: data trend. Adapted from ref 96. of burning. Adapted from ref 97.

optimization parameters. Figure 43 shows trends of 4- by an increase in carbon oxides formation, as cobalt is attacking
carboxybenzaldehyde at different levels of burning with the molecules’ carboxylic group, in particular, that of acetic acid.
changing metals ratio in the catalyst. The minimum of 4- Figure 44 shows a similar behavior when varying the
carboxybenzaldehyde occurs at higher cobalt concentrations, bromine/metals ratio at different levels of burning. Decreasing
showing its higher selectivity when compared with manganese. the bromine/metals ratio causes 4-carboxybenzaldehyde
However, a further increase causes a steep increase in 4- concentration to increase, because the catalyst loses activity,
carboxybenzaldehyde, which means that the catalyst is losing its resembling a bromine-free metal catalyst. Increasing the ratio
selectivity: the Br/Mn/Co synergy cycle (Figure 10) is being too much causes the catalyst to be poor in metals and not to be
broken. In addition, the loss of selectivity is also accompanied able to catalyze peroxide decomposition reactions properly.
AB dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

increases as Br/(Co + Mn) decreases. This shows that the


catalyst selectivity toward product terephthalic acid worsens as
the bromine to metals ratio decreases, forming more carbon
oxides. The reason behind this is that catalyst activity decreases
as bromine content decreases. To keep 2,6-dicarboxyfluorenone
concentration constant a catalyst compensation is necessary. In
proportion, cobalt content will necessarily increase, making the
catalyst more selective to decarboxylation reactions, thus
forming more carbon oxides.
Figure 49 shows the carbon oxides formation trend when the
inverse of 4-carboxybenzaldehyde concentration in product
terephthalic acid changes at different sodium concentrations
and constant catalyst feed.
Figure 44. 4-Carboxybenzaldehyde in terephthalic acid product as a As seen in Figures 41 and 48, the inverse relationship
function of the Br/(Co + Mn) ratio trends at different relative levels of
burning. Adapted from ref 97. between 4-carboxybenzaldehyde and carbon oxides is also
maintained in the presence of sodium. From Figure 49, despite
the scarcity of experimental data, a trend arises: for constant 4-
Other substances formed in the oxidation process, such as carboxybenzaldehyde concentration in product terephthalic
2,6-dicarboxyfluorenone or 2, 6-DCF, which have a detrimental acid, as the sodium concentration increases, carbon oxides
effect on terephthalic acid color and hence in its quality, may formation tends also to increase, which means that higher
also be used for selectivity monitoring.98 Figure 45 shows the sodium concentrations tend to worsen the selectivity. This
2,6-dicarboxyfluorenone concentration and carbon oxides selectivity decrease may be related with decreased activity in the
formation trends with changing catalyst composition. presence of alkali metals in the early stage of the reaction as
Increasing COx formation with increasing the cobalt to observed by Jhung and co-workers.80−82 Overall, 4-carbox-
manganese ratio and with the relative increase of the bromine ybenzaldehyde, 2,6-dicarboxyfluorenone, and carbon oxides all
to metals ratio is a consequence of greater catalyst activity, have an unfavorable effect on terephthalic acid manufacture,
leading to higher decomposition of carboxylic groups of the either by having a quality specification decrease effect as high 4-
molecules within the reaction media. Decreasing the 2,6- carboxybenzaldehyde and 2,6-dicarboxyfluorenone do or by
dicarboxyfluorenone concentration requires both a more active accounting for process economic losses as the burning of
and a more concentrated catalyst, with progressively greater solvent does. Driving process conditions to less solvent burn
cobalt/manganese ratios and bromine contents, either in a leads to undesired byproducts increase and vice versa. In
smaller range as shown by Figure 45b or in a wider ratios range addition, carbon oxides formation is preceded by degradation
as shown by Figure 46. not only of solvent but also of carboxylic groups of other
Like in the case of 4-carboxybenzaldehyde, 2,6-dicarboxy- molecules in the reaction media, which yields several free
fluorenone and carbon oxides show an antagonistic competitive radicals. This causes not only solvent and terephthalic acid loss,
behavior. Figure 47 shows that cobalt-rich catalyst composi- but the combination of these radical also leads to formation of
tions, i.e., increasingly higher Co/Mn, lead to progressively less other undesired byproducts, decreasing the overall selectivity
2,6-dicarboxyfluorenone at the expense of more COx formation. and efficiency of the process and product quality.
Figure 48 shows the effect of bromine to metals ratio on
5.6. Byproduct Chemistry
selectivity at constant 2,6-dicarboxyfluorenone concentration.
The data shows that for a given catalyst composition, The extent of selectivity of the oxidation of p-xylene to
allowing the Br/(Co + Mn) to vary to maintain a constant 2,6- terephthalic acid will determine the amounts of byproducts
dicarboxyfluorenone concentration, carbon oxides formation formed during the reaction. Byproducts formation arises both

Figure 45. (a) 2,6-Dicarboxyfluorenone concentration trend in reaction effluent slurry, and (b) carbon oxides formation trend with simultaneously
changing Br/(Co + Mn) and Co/Mn ratios: cobalt held constant; bromine and manganese varying. Adapted from ref 98.

AC dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX


Chemical Reviews Review

Figure 46. 2,6-Dicarboxyfluorenone concentration trend in reaction effluent slurry with changing Co/Mn and Br/(Co + Mn) ratios over a wide
range. Adapted from ref 98.

Figure 47. Carbon oxides vs 2,6-dicarboxyfluorenone concentration trend (black line read on left vertical and low horizontal axis), and 2,6-
dicarboxyfluorenone and carbon oxides formation trends with changing Co/Mn ratio (dark gray and light gray lines read on right vertical and upper
horizontal axis). Adapted from ref 98.

Figure 48. Carbon oxides vs 2,6-dicarboxyfluorenone concentration at different Br/(Co + Mn) ratios. Adapted from ref 98.

from degradation of the molecules and subsequent combina- As seen previously, acetic acid degradation leads to formation
tions among them to form several more products. of the methyl radical, which then leads to formation of carbon
AD dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

Figure 49. Carbon oxides formation with changing 4-carboxybenzaldehyde concentration in product terephthalic acid at different sodium
concentrations. Dashed gray lines represent trend extrapolations. Adapted from ref 99.

oxides, methane, formic acid, formaldehyde, methanol, methyl oxidation media and their concentration tends to build up,
bromide, and methyl acetate. Decarboxylation of the carboxylic unless a process purge is made. It is therefore important to
group of any other molecule, including both terephthalic acid know what the main byproducts formed are.102
and its preceding intermediates, also leads to formation of 5.6.1. Acetic Acid Derivatives. Decomposition of acetic
methyl radical with subsequent formation of those products acid and acetate ligands by radicals or metals yields several
and also to formation of other radicals, which upon further byproducts via the CH3• , •CH2COOH, and CH3COO•
recombination can yield more complex molecules. In addition, radicals, as shown previously by eqs 160−167 and 175−181,
the impurities present in p-xylene and in the solvent in Figure 51.
commercial manufacture of terephthalic acid are also a source
of molecules that can undergo several transformations like full
or partial oxidations and subsequent decarboxylation to yield
other byproducts. For instance, p-xylene may contain impurities
such as toluene, o-xylene, m-xylene, pseudocumene, 1,2,3-
trimethylbenzene, and 1,3,5-trimethylbenzene. The corre-
sponding carboxylic acids resulting from oxidation of these
impurities, i.e., benzoic acid, o-phthalic acid, isophthalic acid,
trimellitic acid, hemimellitic acid, and trimesic acid, are then
found in the oxidation mother liquor.100 Formation of alkyl
radicals throughout the oxidation reaction can also cause
catalyst decomposition and deactivation. Considering reaction Figure 51. Main byproducts formed from decomposition of acetic
of the methyl radical with bromine to yield methyl bromide as acid.
per eq 169 is the simpler form of bromoalkane formation. This
last secondary section is of particular importance as it consumes
bromine from the catalyst, deactivating it. Some of the several
byproduct pathways are shown in Figure 50. Sun et al. attributes formation of the acetic acid hydro-
Aside from yield decrease, byproducts represent a quality peroxide intermediate HOOCH2CO2H as the source of CO,
issue for PET manufacture, as many of them decrease CO2, formaldehyde, and formic acid.103 As mentioned above,
polymerization rate and, being monofunctional, act as chain acetic acid degradation together with aromatic acid decarbox-
termination agents, causing the polymer to have an undesired ylation and decarboxylation are the main sources of CO and
molecular weight and improper color and optical properties.101 CO2. Formation followed by coupling of two •CH2COOH
Because solvent is recycled, they are always present the radicals is related also to formation of succinic acid102
2•CH 2COOH → HO2 CCH 2CH 2CO2 H (182)
Succinic acid can be brominated, yielding bromosuccinic
acid, which in turn can lose a HBr molecule to yield maleic acid
and fumaric acid isomers. When manganese in the trivalent
oxidation state attacks the methyl group of acetic acid
generating the carboxymethyl radical, glycolic acid can result
from reaction with methane in the presence of water. Glycolic
acid can then either react with acetic acid with loss of one water
molecule to yield acetoxyacetic acid or be oxidized to glyoxylic
acid. The latter can yield oxalic acid upon reaction with a water
molecule. All these substances upon decarboxylation and
Figure 50. Byproduct chemistry pathways in the oxidation of p-xylene decarbonylation will yield, respectively, CO2 and CO in the
in acetic acid by a Co/Mn/Br catalyst. off-gas.88
AE dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

This electrophilic radical can add to p-xylene and other byproduct chemistry as they yield considerably diverse radicals
aromatic molecules which on combination with themselves, in nonmainstream
chemistry free-radical termination reactions, yield progressively
more complex products. Formation of benzoic acid derivatives
is preceded by decarbonylation and decarboxylation reactions
of carbonyl and carboxylic groups, respectively, and formation
of a phenyl radical. Phenyl radicals can abstract hydrogen from
other hydrocarbon and form benzoic acid, react with bromine
Methyl radicals formed from acetic acid decomposition can from the catalyst to form p-bromo benzoic acid, and be
also react with other aromatic molecules and radicals forming oxidized to p-hydroxy benzoic acid.79,102
more byproducts via methyl bromide. They can add to an
aromatic ring in Friedel−Crafts-type reactions,104,105 promoted
by excessive solvent degradation and poor gas−liquid oxygen
transfer and catalyzed by metals, such as iron, from corrosion of
equipment. An example of this reaction is formation of
trimellitic acid

5.6.2. Benzaldehyde and Benzoic Acid Derivatives.


Considering p-tolualdehyde and p-toluic acid to be, respec-
tively, benzaldehyde and benzoic acid derivatives and 4-
carboxybenzaldehye a derivative of either one of them, their
reactions, other than those leading to terephthalic acid, also
account for formation of byproducts. Figure 52 shows some of Figure 53. Benzoic acid derivatives byproducts formed from
the byproducts formed from benzaldehyde reactions. decarbonylation and decarboxylation reactions. Adapted from ref 102.
Together with the common free-radical hydrogen abstraction
reaction, decarboxylation reactions play a major role in

Figure 52. Benzaldehyde byproduct chemistry; X = H, CH3. Adapted from ref 77.

AF dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX


Chemical Reviews Review

In the decarbonylation process, when a carboxy benzoyl catalyst becomes progressively deactivated and the Br/(Co +
radical is formed it can react with methyl radicals to form p- Mn) ratio decreases, formation of insoluble and product
carboxy acetophenone blackening Mn(IV) compounds, like MnO2, also occurs.76,106
5.6.3. Terephthalic Acid Derivatives. Terephthalic acid
can also undergo decarboxylation and bromination reactions to
form some of the compounds shown in Figure 55. Evidence of
addition of a methyl radical to the benzene ring is given when
formation of 2-methyl terephthalic acid and trimellitic acid is
found in quantities larger than expected if all the
pseudocumene found as a p-xylene contaminant would have
p-Xylene oxidation intermediates can undergo bromination its methyl groups oxidized partially or totally to carboxylic
catalyzed by equipment corrosion metals groups.102
Trimellitic acid can react with manganese and form
manganese trimellitate, which is moderately insoluble in acetic
acid/water mixtures and contributes to depletion of manganese
from the liquid medium

Another possible route for benzyl bromide formation is


through direct interaction of the alkyl aromatic molecule with
the bromide ion present in the coordination sphere of the
catalyst75
5.6.4. Phenol Derivatives. The importance of phenols in
p-xylene oxidation is related to the fact that they are known to
be effective oxidation inhibitors.107,108 Decarbonylation or
Phenol derivatives can react to form brominated compounds, decarboxylation followed by oxidation of phenyl radicals
Figure 54. represents the intermediate stage in byproduct phenol
synthesis. Bromination then yields substances like 2,4-
dibromophenol and 2,4,6-tribromophenol. 2-Hydroxy-4-meth-
ylbenzaldehyde can be either the result of solvolysis of a
brominated compound or a product of oxidation of
trialkylbenzene formed as a byproduct or present as an
impurity in p-xylene, Figure 56.102

Figure 56. Some of the phenol derivatives formed in p-xylene


oxidation. Adapted from ref 102.

Figure 54. Products resulting from bromination of phenol byproduct 5.6.5. Biphenyl and Benzophenone Derivatives.
derivative from benzoic acid. Adapted from ref 102. Coupling of aryl radicals, a termination reaction, yields biphenyl
derivatives. The presence of these compounds in the mother
Bromination reactions not only represent formation of liquor is evidence of the occurrence of decarbonylation and
undesired products but also have the consequence of decarboxylation of aromatic acids and aldehydes. Coupling
deactivation of the catalyst. As benzylic bromides are formed, between phenyl and an aroyl radical is also a termination
byproduct cascades are started due to conversion of a much reaction that yields benzophenones. Formation of an aroyl
more Co/Mn/Br active catalyst to a less active and less radical depends on the reaction conditions. If conditions are
selective Co/Mn catalyst, Br/(Co + Mn) decreases, which is such that the aroyl radical does not have enough oxygen to
also responsible for coloration of the product.76 In addition, as form a peroxy radical, it will attack a substrate molecule to yield

Figure 55. Some of the terephthalic acid derivatives formed in p-xylene oxidation. Adapted from ref 102.

AG dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX


Chemical Reviews Review

Figure 57. Formation routes of biphenyl and benzophenone derivatives during p-xylene oxidation: X, Y = H, CH3, CHO, COOH, OH; m, n = 0, 1,
2. Adapted from ref 102.

a benzophenone.101 Other route to benzophenones is through 5.7. Catalyst Deactivation


alkylation of an aromatic with a benzyl bromide derivative via A catalyst is usually defined as a substance which increases the
Friedel−Crafts alkylation. This yields a biphenylmethane, rate of a reaction without being consumed and with its
which upon oxidation yields a benzophenone compound, properties indefinitely unchanged. However, this is not an
Figure 57.102 absolutely true, as a catalyst undergoes chemical and physical
5.6.6. Anthraquinone and Fluorenone Derivatives. degrading modifications which alter its activity and selectivity.
Cyclization of o-carboxy benzophenones is the most likely path This degradation of catalytic properties is often referred to as
to yield anthraquinone derivatives. Oxidation of dihydroan- catalyst deactivation and despite being more common in
thracene derivatives is also a plausible route. Note that the heterogeneous catalysis also occurs in homogeneous catalysts,
dihydroanthracene may be the reduction product of an mainly due to the high reactivity of the active species in the
anthraquinone recycled to the process. Fluorenones, in metal complex.113 With a Co/Mn/Br catalyst, deactivation is
particular, 2,6-dicarboxyfluorenone, have a detrimental effect due mainly to reactions of bromine with organic substrate
on terephthalic acid color.98 Fluorenone can be formed by two molecules and precipitation of insoluble manganese com-
distinct paths. One route may start with decarboxylation of an pounds. The bromination reaction as a form of catalyst
o-carboxy benzophenone, yielding an o-benzoylphenyl radical, deactivation is closely related to formation of byproducts.
which upon electron bonding completes the cyclization.102,109 Coupling between a bromine atom and a methyl radical, as per
The other route is via a biphenyl with a methyl group in the eq 169, yields methyl bromide which being volatile at normal
ortho position. Oxidation of the methyl group yields a reaction conditions is removed together with the reaction off-
carboxylic or a carbonyl group, which after formation of an gas and is taken this way from the reaction media. Formation of
aroyl radical is converted to a ketone group via cyclization. This benzylic bromides also accounts for catalyst deactivation
later route has been employed successfully to synthesize 2,6- despite remaining in the reaction media. Benzylic bromides,
dicarboxyfluorenone via oxidation of 2,4,5-trimethylbiphenyl to as a source of bromine to the Co/Mn/Br catalyst, have little or
2,4,5-tricarboxybiphenyl, Figure 58.110 no catalytic activity in the oxidation of alkyl aromatics.75
5.6.7. Ester Derivatives. p-Toluic acid and terephthalic Addition of a benzylic bromide to the oxidation media
acid esterification products are also byproducts found in p- comprising a Co/Mn catalyst and an alkyl aromatic has a
xylene oxidation mother liquor. The routes to esterification negligible effect on the rate of oxidation, Figure 59.
may involve other byproducts as intermediates such as benzoyl When bromine is lost from the catalyst one passes from a
peroxides, phenols, and aroyloxy radicals. Esterification of p- high active Co/Mn/Br catalyst to a much less active Co/Mn
toluic acid and terephthalic acid with 4-(hydroxymethyl)- catalyst. Bromine losses, in particular, to methyl bromide, make
benzoic acid, a minor intermediate found in terephthalic the Br/(Co + Mn) ratio decrease. This leads, according to
acid,111 is also one possible route for ester formation. Benzyl Figures 28 and 35, when at the optimum, to carbon oxides and
acetate esters formation is also important as the main reaction methyl bromide formation increases.
is carried out in acetic acid media. Examples of esters formed in Considering the catalytic cycle given in Figure 10, the redox
p-xylene oxidation in acetic acid together with other byproducts reactions between manganese and bromine cease and the
of the families mentioned above are given in Table 2. concentration of Mn3+ ion tends to reach a steady state. The
AH dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

Figure 58. Formation routes of anthraquinone and fluorenone derivatives during p-xylene oxidation: X, Y = H, CH3, CHO, COOH, CH2OH; m, n =
0, 1, 2. Adapted from ref 102.

implication is that because water is present in the oxidation organic bromide to the ionic bromide form, causing the
media, disproportionation of Mn3+ can occur reaction rate to increase significantly.75
2Mn 3 + → Mn 2 + + Mn 4 + (189)

Water is known to increase the proportion of Mn4+


formation in acetic acid/water solutions, and more Mn4+ is In addition, because in the oxidation media water is present,
formed as the water concentration increases, Table 3.114 hydrolysis of benzylic bromide can also occur.This reaction is
From Mn4+, MnO2 is formed, a product coloring substance another possible route for formation of benzylic alcohols.
and also insoluble, which by removing manganese from the
reaction media also contributes to catalyst deactivation.
Manganese trimellitate, as seen previously, has the same effect
on the catalyst system. Bromine loss to byproducts, together
with manganese precipitation of insoluble compounds, results
in a cyclic cascade of events that ends ultimately in a low 6. OPERATIONAL OPTIMIZATION STUDIES
catalyst activity, Figure 60. Optimization of an operating plant deals with reducing costs
5.7.1. Catalyst Regeneration. The quickest way to while maintaining product quality. In a commercial terephthalic
regenerate a bromine-depleted Co/Mn/Br catalyst is through acid plant to achieve this objective the variables are many and
addition of hydrobromic acid, but other forms are possible. to better manipulate them understanding of the chemistry
Benzylic bromides can be reconverted, recovering bromine in behind the process is fundamental. The main goal is to
its ionic form, thus regenerating the catalyst. Addition of minimize 4-carboxybenzaldehyde in the final product, keeping
sodium chloride leads to displacement of the covalent bonded operational costs within controlled economical limits while
AI dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

Table 2. High Molecular Weight Byproducts Found in p-Xylene Oxidation Solvent101,102,112

AJ dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX


Chemical Reviews Review

Table 2. continued

Table 3. Effect of Water in the Calculated Equilibrium


Values of Mn4+ from the Mn3+ Disproportionation in Acetic
Acid/Water Mixturesa
water (vol %) water (M) Kd (×106) Mn as Mn(IV) (%)
1 0.56 7.6 2.7
4.7 2.61 8 2.8
9.1 5.06 30 5.2
16 8.89 800 22
a
Data from ref 114.
Figure 59. Effect of different bromine sources on the O2 uptake rate
during Co/Mn/Br-catalyzed oxidation of pseudocumene. Adapted
from ref 75.
concentrations, water concentration in the mother liquor, rate,
impurities, etc.
maximizing production rate. In a simplified manner, this is 4‐CBA = ϕ([Mn], [Co], [Br], vent O2 %, H 2O%,
usually a trade-off between product quality, expressed in 4-CBA
content, and production cost, expressed in terms of solvent p‐xylene feed rate, total feed rate,
burning, i.e., COx in reactor off-gas, Figure 61. residence time, temperature, ...) (192)
In a more sophisticated model, solvent burning by itself is
not the only cost factor but parcels like catalyst concentration, Construction of models based on process variables allows us
power, utilities, etc., should also be taken into account and 4- to predict 4-CBA in the product and establish which process
CBA is dependent on many process variables, like catalyst variables have greater influence in its variation. Selection of
AK dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

7. RECENT RESEARCH TRENDS IN TEREPHTHALIC


ACID SYNTHESIS
7.1. Oxidation in Sub- and Supercritical Water
Water is an excellent environmentally friendly solvent. Because
it is nontoxic, readily available, inexpensive, and benign to the
environment, it has been the subject of much research as
solvent media for oxidation of p-xylene.123 More recently,
research has turned to sub- and supercritical water as oxidation
media.124,125 Sub- and supercritical water (Tc = 647.3 K; Pc =
221.2 bar]126) are interesting reaction media because their
properties vary considerably as temperature increases. As the
temperatures rises up to the critical temperature, some of the
physical properties of water progressively approach those of
Figure 60. Schematic cyclic cascade of events leading to catalyst organic solvents. Figure 62 shows the dielectric constant,
deactivation.

Figure 61. Trade-off between product quality and production cost.

these variables allows their proper manipulation within a


product quality and cost policy.
Han et al.115,116 uses Multivariate Statistical Process Models
to build 4-CBA prediction models based on regression of
process variable data. Liu et al.117 use a Fuzzy Neural Network
with a reduced number of variables, and Zhang et al.118 use a
Fuzzy Support Vector Regression Model built from 10 process
variables of industrial data to predict 4-CBA concentration. Mu
et al.119,120 using a set of 10 process variables developed a
genetic algorithm which optimizes operation profit either by
manipulating product quality, i.e., 4-CBA and process
throughput, or by operating the process variables sequence in
the proper operating order. Hong et al.121 present a Fuzzy
Adaptive Immune Algorithm. Gujarathi and Babu122 present a
multiobjective optimization algorithm based on effect of six
decision variables identified, namely, catalyst concentration,
vent oxygen content, % water in solvent, feed xylene rate,
temperature of reactor, and total feed rate, also to achieve the
same effect of allowing the industrial operator to make the
desirable improvement in the process design decisions.
It is a difficult and time-consuming method to choose an
appropriate set of operation conditions for PTA production by
empirical means alone. These models, reflecting the same
individual trends observed individually from experimental
observations, provide a useful way to manipulate one variable
individually or several variables in a given sequence, within
preset limits, to aim for a desired 4-CBA concentration and
economical profit with a comfortable confidence margin to Figure 62. Properties of water as a function of temperature at 25 bar:
avoid unnecessary economical cost caused by poor trial and (a) dielectric constant; (b) terephthalic acid solubility; (c) ion product.
error industrial experiments. Adapted from ref 125.

AL dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX


Chemical Reviews Review

Table 4. Yields of p-Xylene Oxidation to Terephthalic Acid in Sub- and Supercritical Water
media T (°C) catalyst max yield (%) 4-CBA yield (%) oxidant ref
subcritical 300 MnBr2 49 ± 8 0.03 ± 0.04 H2O2 127
supercritical 380 MnBr2 57 ± 15 H2O2 128
supercritical 400 MnBr2 >90 ∼0 H2O2 129
supercritical 380 MnBr2 95 <0.1 H2O2 130,131
subcritical 300 MnBr2 80 0.6 O2 132
supercritical 380 HBr 19.1 0.06 O2 133
supercritical 380 CoBr2 6.3 5.13 O2 133
supercritical 380 MnBr2/HBr 22.7 4.66 O2 133
supercritical 380 MnBr2/benzoic acid 35.6 5.12 O2 133
supercritical 380 CuBr2 55.6 0.77 O2 124
supercritical 380 CuBr2/ NiBr2 59 ∼0 O2 124
supercritical 380 Cu/Co/Br 60 O2 124
supercritical 380 Cu/Co/NH4/Br 70.5 O2 124

Figure 63. (a) Rate of p-xylene consumption and (b) degree of a Co/Mn/Br catalyst activation as a function of the Zr/Co ratio with air oxidation at
atmospheric pressure at 95 °C. Zr used in the form of zirconium acetate solution. Adapted from ref 137.

terephthalic acid solubility, and ionic product of water as a high yield of terephthalic acid in supercritical water is
function of temperature. The disruption of hydrogen bonding achievable and comparable to that achieved in commercial
causes the dielectric constant to be very low when compared to processes with comparable selectivity, in particular, with low 4-
that at room temperature, making supercritical water a suitable carboxybenzaldehyde content. Despite being a promising
solvent for organic substances, as can be seen from the sharp process, the major drawback of p-xylene oxidation in super-
increase of terephthalic acid solubility. This means that the critical water is the pressure and temperature required to
oxidation media for carrying out the reaction and product achieve supercritical conditions, which are extremely high when
recovery from the solvent can be easily adjusted and achieved compared to those used in the AMOCO process, making the
by adjusting the temperature. oxidation process very hazardous.
Table 4 shows results of p-xylene oxidation to terephthalic 7.2. Alternative Catalyst Systems and Cocatalysts
acid carried out in sub- and supercritical water.
Manganese bromide is the preferred catalyst in the majority Improving the overall yield of a process with formation of less
of the experiments reported in the literature as it yields some of byproducts due to secondary and decomposition reactions in
the best results, some of them compared with the 95% yield terephthalic acid manufacture has been the driving force of
found in the AMOCO process with the Co/Mn/Br catalyst. many research efforts for alternative and more selective
HBr, HBr/MnBr2, CoBr2, MnBr2/benzoic acid, CuBr2, and catalysts. Use of titanium lining in the equipment due to the
CuBr2/NiBr2 catalysts have also been used but with lower presence of HBr in the catalyst led to research of low bromine
yields.127−134 Use of hydrogen peroxide as oxidant is related catalysts. Saha and Espenson135 report the use of a low-
with experimental difficulties in controlling a compressed bromide-containing catalyst with simultaneous use of Zn-
oxygen feed to the reactor; when fed at the experimental (OAc)2 and strong acid such as trifluoroacetic acid or
conditions, hydrogen peroxide decomposes yielding oxygen. A heptafluorobutyric acid, yielding the inorganic bromophilic
AM dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

Figure 64. (a) Effect of zirconium addition on aldehyde/(alcohol + acetate) ratio yield as a function of the Co/Mn/Br catalyst concentration at
atmospheric pressure and 95 °C. Zr/Co = 0.17. (b) Correlation of degree of activation caused by zirconium with aldehyde/(alcohol + acetate) ratio.
Aldehyde/(alcohol + acetate) ratio is at a p-xylene conversion of 30%. Adapted from ref 137.

species ZnBr+, which minimizes bromine losses to the is dehydrated directly to p-tolualdehyde by the catalyst metals,
catalytically inactive benzylic bromides. The activating effect making the reaction rate much faster. The same principle was
is due to their protonation strength more that the nature of considered for the p-toluic acid oxidation mechanism to 4-
their anion.136 Li and Li72 completely removed HBr from the carboxybenzyl alcohol. Addition of Zr and subsequent
catalyst and achieved a maximum yield of 93.5% of terephthalic activation of the catalyst bypasses the chemistry via the benzyl
acid at 100 °C and oxygen pressure of 10 atm using CoBr2 and alcohols.
MnBr2 as catalyst. Figure 64a shows the effect on zirconium addition on the
7.2.1. Zirconium. Table 1 reports several MC-type catalysts aldehyde/(alcohol + acetate) ratio yield as a function of p-
using other metals than or in addition to cobalt and manganese. xylene conversion. Note that in this ratio acetate refers to 4-
Use of zirconium as cocatalyst has revealed an interesting methylbenzyl acetate, formed from esterification between 4-
potential for p-xylene oxidation, because when added to a Co/ methylbenzyl alcohol and the acetic acid solvent.
Mn/Br catalyst it increases its activity and selectivity.114,137 When zirconium activates a Co/Mn/Br catalyst, the ratio
Figure 63 shows the rate of p-xylene consumption and aldehyde/(alcohol + acetate) increases with conversion of p-
degree of the Co/Mn/Br catalyst activation as a function of Zr/ xylene. In contrast, at a cobalt concentration of 0.02 M, as
Co ratio with air oxidation at atmospheric pressure at 95 °C. Figure 64a shows, no activation occurs, and for a constant p-
At higher cobalt concentrations zirconium does not have an xylene conversion the ratio aldehyde/(alcohol + acetate) suffers
activating effect as cocatalyst, but at lower cobalt concentrations no increase. This is more evident observing Figure 64b, which
zirconium increases the catalyst concentration as the Zr/Co shows that, in general, the greater the degree of activation the
ratio increases with a more pronounced effect at lower Zr/Co greater the aldehyde/(alcohol + acetate) ratio. Figures 63 and
ratios. The degree of activation, measured by the ratio between 64 show that the activating effect of zirconium does not occur
the rate constant with a Zr/Co/Mn/Br catalyst and the rate at the highest cobalt concentration but at intermediate
constant with a Co/Mn/Br catalyst, i.e., kZr/Co/Mn/Br/k Co/Mn/Br, concentrations instead. This is an indication that adding
at the same Co/Mn/Br catalyst concentration also reflects the zirconium to a Co/Mn/Br catalyst allows one to decrease the
activating effect of the Zr cocatalyst and occurs at an optimum cobalt concentration while maintaining the same catalyst
cobalt concentration. activity and increased selectivity. The effect of zirconium is
In addition to increasing activity, zirconium also increases the due to its high Lewis acidity in the promotion of hydroperoxide
catalyst selectivity. The strong activating effect of Zr favors decomposition via a dehydration reaction: Zr(IV) > Co(III) >
direct formation of p-tolualdehyde from reaction p-xylene with Mn(III) > Co(II) > Mn (II). During the catalytic cycles shown
the oxygen molecule in 1:1 stoichiometry in relation to in Figure 10, cobalt and manganese are present mainly in their
formation of 4-methylbenzyl alcohol in 1:0.5 stoichiometry. divalent forms, and hence, the promotion effect of their
Note that according to Partenheimer137 formation of p- trivalent forms is small. Addition of Zr(IV) to Co/Mn/Br will
tolualdehyde from p-xylene oxidation may involve the be expected to increase the rate by promoting the benzylic
intermediate 4-methylbenzyl alcohol, which upon further peroxide to the benzaldehyde, as zirconium is present as
oxidation yields p-tolualdehyde. However, the reactivity of 4- Zr(IV). Zr(IV) will have stronger coordination with the
methylbenzyl alcohol is, according to the same author, 15 times peroxide, weakening the O−O bond, decomposing it to
higher than that of p-xylene and 5 times higher than the aldehyde and water, and thus increasing the aldehyde/(alcohol
reactivity of p-tolualdehyde .This step was not considered by + acetate) ratio, Figure 65.137
Wang et al.69 in their proposed kinetic mechanism, as they did Zirconium addition to Co/Mn/Br allows a decrease in the
not detect 4-methylbenzyl alcohol, mainly because their study cobalt concentration in the catalyst, but it should be always
was carried out at a higher temperature and high catalyst balanced for optimization purposes. It may not be desirable as
concentration, and so the primary hydroperoxide from p-xylene cobalt provides other selective pathways, such as avoiding
AN dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

causing an explosion or fire, and a higher concentration of O2


could be employed in the absence of an inert gas diluent such
as nitrogen air. The beneficial effect of CO2 as co-oxidant may
be explained by three mechanistic interpretations, independ-
ently or by their combination.140
• The presence of CO2 may have a suppressive effect in
Figure 65. Suggested mechanism of metal-catalyzed dehydration of decarboxylation reactions of intermediates and in
benzoyl peroxide. X = CH3 or COOH. Adapted from ref 137. product, which would improve the overall yield and
appear to enhance the oxidation rate. This effect has also
been used to explain the inhibiting role of CO2 as an
addition of peroxy and alkoxy radicals through their inert gas in acetic acid decomposition.144,145
decomposition, reducing the concentration of hydroxyl radicals
from benzyl hydroperoxide thermal decomposition, and • Improvements in oxidation rate and product selectivity
decreasing formation of phenyl formate, one of the possible might arise from increased solubility of substrate and
products of the Baeyer−Villiger reaction. Phenyl formate is intermediates and, particularly, increased solubility of
undesirable because not only it is a decrease of terephthalic acid oxygen. The effect is due to solvent expansion caused by
yield but also with its hydrolysis it yields its phenol antioxidant CO2, more notorious as CO2 pressure is increased
derivative.107,108 In addition, phenol can be further oxidized, further,145 even at a temperature of 160 °C, a low
yielding the undesirable, color-forming quinone product. temperature when compared to those normally em-
7.2.2. Carbon Dioxide Use as Co-oxidant. Carbon
ployed in commercial processes.
dioxide is one of the byproducts formed during p-xylene
oxidation and is formed from degradation of both products and • The increased solubility of oxygen in the presence of
acetic acid solvent. CO2 has been given a new role as co-oxidant CO2 may indeed account for formation of a catalytically
in homogeneous catalytic oxidation of aromatic hydro- active cyclic peroxocarbonate species, CO42−, on the
carbons.138−143
Figure 66 shows the beneficial effect of CO2 when it is
introduced in the gaseous feed stream.
The beneficial effect of CO2 as co-oxidant is clear, as an
increase in carboxylic acid in the product and increase in
oxygen consumption can be observed, showing that the activity
and selectivity toward the carboxylic acid is enhanced, even at
milder conditions. In addition to Co/Mn/HBr catalyst systems,
CO2 also has beneficial effects as co-oxidant in catalyst systems
of the type Co/Mn/HBr/M and Co/Mn/HBr/M−M′, where
M stands for an alkali metal or alkaline earth metal and M′
stands for a transition metal, including Zr, Hf, and Mo and
lanthanides, such as Ce, Pr, and Sm.140 Oxidation of
intermediates, such as p-toluic acid, the rate-determining step, Figure 67. Postulated formation mechanism and form of the
shows a similar behavior, which suggests that the presence of peroxocarbonate species. Oxygen atoms in bold represent those
from the original O2 molecule. M = Co or Mn. Adapted from ref 140.
CO2 yields a product with less intermediate impurities, which in
turn may reduce the demand in the purification step in a
commercial AMOCO process. With the use of CO2, a higher metal center, via a synergistic interaction between the O2
concentration of O2 could be used in the reaction without and the CO2 molecules, as shown in Figure 67.

Figure 66. Effect of CO2 in the liquid-phase oxidation with a Co/Mn/Br catalyst of (a) m-xylene and (b) p-xylene in the presence of 147 ppm of
potassium. Adapted from ref 140.

AO dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX


Chemical Reviews Review

The key role of the peroxocarbonate species for oxidation of turn generate catalytically active Co(III) species, increasing its
alkyl aromatics such as p-xylene is not clear. It may lie in concentration, thus promoting oxidation of the aromatic
transferring an oxygen atom bonded to the metal in the hydrocarbon.37 More recently, the drive to remove corrosive
peroxocarbonate complex to the p-xylene molecule to form an bromine from the catalyst and from the process led to research
alcohol and then in assisting abstraction of hydrogen to form in transition-metal salt catalytic oxidation of aromatic hydro-
aldehyde, forming the p-toluic acid intermediate. The same carbons promoted by N-hydroxyphthalimide (NHPI) and
mechanism is repeated in oxidation of the other methyl group derivatives.146−149 Oxidation of p-xylene yields up to 84% of
of the p-toluic acid molecule to produce terephthalic acid. terephthalic yield with less than 1% 4-carboxybenzaldehyde
However, it may indirectly participate in formation of free yield when 20 mol % NHPI is used together with cobalt and
radicals by acceleration of the catalytic pathway shown in manganese acetates in acetic acid at 150 °C and at a pressure of
Figure 10, acting in particular in Co2+ oxidation acceleration, 30 atm for 3 h.147 Figure 69 shows the postulated mechanism
inducing an acceleration in the other steps, Figure 68.140,141 catalyzed by NHPI similar to the classic free-radical oxidation
mechanism. It involves initial in situ NHPI decomposition by a
free-radical chain initiator such as a peroxide compound or a
labile metal complex to the phthalimide N-oxyl (PINO) radical.
PINO radical abstracts a hydrogen atom from the hydrocarbon,
generating an alkyl radical which will react with the oxygen
molecule to yield a peroxy radical. The PINO radical is
regenerated back to NHPI through hydrocarbon hydrogen
abstraction. Peroxide radicals yield alkyl peroxide molecules,
which will be decomposed by the transition-metal complex to
the oxygenate product.
The mechanism above follows a classical free-radical chain
Figure 68. Postulated role of the peroxocarbonate species in the reaction, and when fast oxygen consumption occurs, both
conventional MC oxidation catalytic pathway. Adapted from ref 140. NHPI and the PINO radical decompose, decreasing the overall
oxidation rate of the substrate.148 This means that to keep a
7.2.3. N-Hydroxyimides. Organic promoters have been steady oxidation rate NHPI needs to be continuously added to
used in terephthalic acid synthesis and manufacture.48 In the the reaction medium. To reduce the amount of NHPI required,
Teijin process, methyl−ethyl ketone is used as a promoter,2,43 N-acetoxyphthalimide (NAPI) can effectively achieve a similar
with its principal function to generate peroxy radicals, which in yield on terephthalic acid (80%) with only 5 mol % together

Figure 69. Postulated mechanism for aerobic oxidation of an aromatic hydrocarbon promoted by NHPI in the presence of a transition-metal catalyst
complex. X = H, CH3, CO2H; LnCo: cobalt or transition-metal complex with n ligands L. Adapted from ref 149.

AP dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX


Chemical Reviews Review

with 0.5 mol % of both cobalt and manganese acetates. NAPI


can be prepared according to eq 193.

The time dependence of the aerobic oxidation of p-xylene


catalyzed by NHPI or NAPI combined with cobalt acetate is
shown in Figure 70.
From the figure it can be seen that p-xylene oxidation in the Figure 71. Resonance stabilization of (a) guanidine and (b) guanidine
presence of NAPI is slower and also that the yield in ion.
terephthalic acid is higher. It has been suggested that the
early stage of p-xylene oxidation NHPI is rapidly decomposed
to phthalimide and phthalic anhydride. NAPI is considered to stronger electron donor than water, guanidine can progressively
be resistant to this rapid decomposition when violent chain replace aqua and acetic ligands in the catalyst structure and
reactions take place, since it is gradually hydrolyzed to NHPI by bridge metal ions, as shown in Figure 73. The Y-shaped
water present in acetic acid as well as the water resulting from aromatic structure disperses the π electrons in the transition
oxidation.146,148 state of the complex, stabilizing it, and lowering the activation
7.2.4. Guanidine. Cheng and co-workers150 studied the energy of the electronic transfer reaction in the inner
effect of guanidine as promoter in Co/Mn/Br p-xylene coordination, effectively enhancing the rate. This means that
oxidation. The high basicity of guanidine (pKa = 13.6) and addition of guanidine to the catalyst will make structures II and
the exceptional “Y-aromacity” resonance stabilization of the 6 π III in Figure 72 more active than structure I.
electrons of guanidine and of its conjugated acid ion are Coordination of guanidine to the inner coordination sphere
responsible for its thermodynamic stability, Figure 71.151 is limited, and thus, high outer-sphere formation of the ion pair
Figure 72 shows the effect of guanidine addition to the Co/ guanidine ion−acetate ion, {Gua+OAc−}, will occur. This ion
Mn/Br catalyst in the rate constants on stepwise oxidation of p- pair will go through ion exchange reactions with HBr
xylene to terephthalic acid, as per eq 48, and on formation of molecules, either free or bonded to the catalyst, yielding the
carbon dioxides. ion pair {Gua+Br−}. This will cause a depletion of Br− ions on
Concerning rate constants, guanidine appears to have an the catalyst coordination sphere (see Figure 14), decreasing its
inhibiting effect on the first two steps of the oxidation and a catalytic activity, as formation of bromine free radical via the
clear promoting effect on oxidation of the second methyl group, synergy cycle of Figure 10 is lessened. Concerning formation of
as happens with the influence of water in the catalyst activity, carbon oxides, the positive effect of guanidine is also due to an
which is an indication of guanidine influence on catalyst increase of catalyst inner-sphere electron-transfer reaction rates.
structure. In addition, a clear effect in carbon oxides formation As these electron-transfer rates increase stationary Co(III)
decreases is observed, indicating higher catalyst selectivity when concentration decreases. Cobaltic ions are responsible for
compared with the normal Co/Mn/Br catalyst. Being a decarboxylation reaction of acetic acid as per eqs 154 and 155.

Figure 70. Time-dependence concentration of p-xylene, p-toluic acid, and terephthalic acid in the aerobic oxidation of p-xylene catalyzed by (a)
NHPI and (b) NAPI combined with cobalt acetate. Adapted from ref 146.

AQ dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX


Chemical Reviews Review

Figure 72. Effect of guanidine composition on rate constants (black lines) and in COx generation (gray line). Adapted from ref 150.

fluids at room temperature. They have very low or no


measurable vapor pressure and thus at ambient temperature
are effectively nonvolatile. Because they consist of at least of
one anion and one cation, properties such as melting point,
viscosity, density, and hydrophobicity can be adjusted to a
particular purpose simply by changing the nature of one of the
components of the ion pairs. They are immiscible with some
alkane solvents and so can be used in two-phase systems. Their
composing anions are often weakly coordinating ligands, such
as BF4− and PF6−, and, hence, have the potential to be highly
polar yet noncoordinating solvents. They can be expected,
therefore, to have a strong rate-enhancing effect on reactions
involving cationic intermediates. Ionic liquids can dissolve a
wide range of organic inorganic and organometallic compounds
and also gases such as H2, CO, and O2, which makes them
suitable for catalysis of several reactions.152,153 Ionic liquids can
be used in catalysis as either catalyst or solvent or a
combination of these, as catalyst activator, or as cocatalyst for
reaction, as a catalyst ligand source, or as solvent alone for
several reactions, including aerobic oxidation reactions, as they
are stable to oxidation.154 Saleh155 patented a process to
produce aromatic aldehydes from carbonylation of suitable alkyl
aromatic compounds with pyridine or imidazole derivative ionic
liquids in a pressure range from 4 to 100 bar and temperature
range from 10 to 100 °C. The carbonylation reaction of toluene
in those experimental conditions can achieve a conversion of
66% with p-tolualdehyde selectivity up to 89.3%. Subsequent
oxidation of p-tolualdehyde with air in acetic acid with a Co/
Mn/Br catalyst provides an alternative route from toluene to
Figure 73. Suggested structures for the Co/Mn/Br/guanidine catalyst terephthalic acid. Howarth156 achieved oxidation of several
system. M, M′ = Co(II), Co(III), Mn(II), Mn(III). Adapted from ref aromatic aldehydes catalyzed by nickel acetylacetonate in
150. [bmim][PF6] at 60 °C and atmospheric pressure. In oxidation
of benzaldehyde, p-tolualdehyde, and terephthaldicarboxyalde-
hyde, yields of 66%, 63%, and 47% of the respective aromatic
In addition, acetic ligand substitution by guanidine also keeps acids, benzoic acid, p-toluic acid, and terephthalic acid, were
them away from the metal, avoiding their decarboxylation in achieved. Earle and Katdare157 patented a process in which
the inner sphere of the catalyst. In conclusion, an optimized toluene is oxidized with nitric acid and air with pyridine or
amount of guanidine promoter will increase the reaction rate imidazole derivatives ([bmim][NO3]) as the ionic liquid from
and decrease solvent burning.150 atmospheric pressure up to 100 bar in the ideal temperature
7.3. Oxidation in Ionic Liquids range from 100 to 120 °C. Benzoic acid yield was between 70%
Most of chemical reactions are performed in molecular and 90%. When p-xylene is used as feedstock in [bmim][OMs],
solvents. Recently, ionic liquids have emerged as alternative terephthalic acid yield is low, 4.8%, with a low selectivity,
solvents for several chemical reactions. Ionic liquids are often 11.5%, with the other products formed in the reaction, p-toluic
AR dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

acid, with 36.4% yield, and with a higher selectivity of 88.5%. attainable yields, and selectivity close to that used in the
The nature of the ionic liquid also has a decisive influence on conventional MC AMOCO process and are bromine-free
the outcome of a reaction, as Earle and co-workers158 have catalysts.
shown for oxidation of toluene with nitric acid, Figure 74. 7.4. Heterogeneous Systems
Like with research in homogeneous systems, research on
heterogeneous systems is also motivated by efforts to improve
yields, achieve higher selectivity, and replace of hazardous
chemicals of softening process conditions.
In their 1986 paper, Hronec and Hrabě160 review briefly
efforts of hydrocarbons liquid-phase oxidations catalyzed by
heterogeneous catalysts, including oxidation of p-xylene over
CoY and MnY zeolites, but with very low yields and oxidation
in only one methyl group. In the same paper, the results are
reported for oxidation of p-xylene in water medium catalyzed
by metal oxides such as lead, cerium chromium, silver nickel,
manganese, and silver oxides. These metal oxides show little or
no activity as catalysts for p-xylene oxidation to p-toluic acid or
terephthalic acid in the absence of p-toluic acid. Addition of p-
Figure 74. Influence of the ionic liquid nature on the outcome of the toluic acid increases activity, but only the cobalt and manganese
oxidation of toluene with nitric acid. Adapted from ref 158. oxides show activity to yield terephthalic acid. With the other
metal oxides p-xylene conversion ceases at 5−10%. These
Terephthalic acid can be obtained directly from oxidation of observations led the authors to the conclusion that the
p-xylene using NHPI/O2/HNO3 as oxidative system in catalytically active species are not heterogeneous metal oxides
[bmim][OMs] at 110 °C and 1 bar.159 The maximum p-xylene but the homogeneous toluate salts. The activity of cobalt and
conversion achieved was 98%, with a 96% terephthalic acid manganese oxides is strongly dependent on calcination
yield and 98% selectivity. The proposed mechanism is shown in temperatures and shows both a critical concentration above
Figure 75. which terephthalic acid yield is maximum and optimum water
NHPI and nitric acid react to yield the PINO radical, water, content. Only Co−Cr and Mn−Ce bimetallic catalysts show a
and NO2 radical. The PINO radical then abstracts a hydrogen higher activity than single metal catalyst, which is ratio
atom from the methyl group of p-xylene, initiating the chain dependent, as there are both antagonistic and synergistic
reaction. NO2 radical decomposes the peroxide radical, effects.
regenerating HNO3 and yielding the oxygenate product. Ionic Chavan and co-workers161 reported high activity and
liquids present themselves as a promising research field for selectivity toward terephthalic acid when p-xylene is oxidized
oxidation of alkyl aromatics to produce alternative routes to in acetic acid−water liquid medium at normal industrial process
aromatic carboxylic acids like terephthalic acid, with the temperature and pressures between 14 and 38 barg using as
important advantage of use of nonvolatile recyclable solvents, heterogeneous catalyst neat and μ3-oxo-bridged Co/Mn cluster

Figure 75. Proposed mechanism for oxidation of p-xylene to terephthalic acid by the NHPI/O2/HNO3 system in [bmim][OMs] at 110 °C and 1
bar. Adapted from ref 159.

AS dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX


Chemical Reviews Review

Figure 76. Proposed catalytic oxidation of p-toluic acid over CoMn2(O)−Y catalyst. Adapted from ref 161.

complexes encapsulated in Zeolite−Y. Using such catalyst, with constrains that may also constitute a barrier to a viable, full-
heteronuclear complexes, 100% conversions of p-xylene were scale commercial process.
achieved and distributions in the product being between 98.9%
and 99.4% for terephthalic acid and 0.01% or less for 4- 8. CONCLUSIONS
carboxybenzaldehyde. These results were, however, obtained at
38 barg, a much higher pressure that the 14−16 barg used in In this paper, p-xylene oxidation is reviewed with particular
commercial processes. In commercial processes 4-carboxyben- emphasis on the AMOCO MC method, giving a historical
zaldehyde is present in 0.1−0.5% in the crude product before evolution of the processes of terephthalic acid manufacture
the hydrogenation purification process. The reasoning behind from the early days’ difficulties inherent to the p-xylene
these results is in the low redox potentials shown by the cluster resistance toward oxidation to the breakthrough of the
complexes, indicating easier electron transfer between Co2+/ AMOCO MC cobalt/manganese/bromine catalyst system.
Co3+ redox states in the oxidation and reductions steps of the The main state of the art detailed kinetics are presented,
proposed mechanism of catalytic oxidation, as shown in Figure which are a useful tool for investigation of mechanisms of new
76. catalyst and byproducts, but a lumped kinetic mechanism
Due to the close dimensions of catalyst cluster and zeolite,
proves to account effectively for the main intermediates during
leaching is difficult and hence catalyst is stable and can be
p-xylene oxidation to terephthalic acid. The actual chemistry
separated and recycled. Washing the solid mixture with 0.1 N
NaOH is required to recover the carboxylic acids in the form of proves to be very complex, and the literature gives evidence
water-soluble salts, and the catalyst is recovered by filtration. that changes in catalyst composition cause changes in its
No plugging due to build-up in the catalyst is reported. Indeed, structure which account for different behavior in terms of
heterogeneous catalyst may be deactivated by solids which activity and selectivity, which in turn are important when it is
block the active sites. This was reported by Kim et al. in their desirable to correlate this change with industrial process control
studies of p-xylene oxidation using transition metals supported variables. Optimization of an industrial process is thus
SBA-15,162 by Das and Chakrabarty in the oxidation catalyzed intimately related with optimization of the catalyst. Notwith-
by Co(III)−oxo cubane clusters,163 and by Bastock et al. using standing its high yield and widespread use, the current
Envirocat EPAC as catalyst with a terephthalic acid yield of dominant process of terephthalic acid manufacture using the
only 37.5%.164 This effect constitutes a barrier when develop- AMOCO MC Method catalyst still has some handicaps related
ment of a commercial process is desired. with the aggressive nature of its components and required
More recently, patents have been filed claiming development
investment in specialized equipment. These handicaps are
of heterogeneous catalysts consisting of cobalt and manganese
heteronuclear complexes encapsulated in a zeolite and drives to the study of new catalyst, promoters, and solvents,
conceptual design of a process using solvent consisting of a some of which are presented in this paper, and are an object of
mixture of p-xylene and water.165−170 Oxidation is done in a interest. Further research efforts and process development are
series of four reactors which operate at progressively increasing necessary to shift from the current mainstream technology to
temperature from 256 to 300 °C and 50 bar absolute. more environmentally friendly and equally effective technolo-
Conversion of p-xylene in the four reactors is extremely low, gies.
1.5%, but the formed terephthalic acid remains in solution as
well as 4-carboxybenzaldehyde. In commercial processes, once ASSOCIATED CONTENT
formed terephthalic acid as 4-carboxybenzaldehyde both
precipitate, making their separation difficult and creating the *
S Supporting Information

need for a purification unit. Remaining in solution, 4- This material is available free of charge via the Internet at
carboxybenzaldehyde is converted to terephthalic acid; in http://pubs.acs.org.
addition, because both substances are in the liquid phase, no
blocking of the catalyst is expected. Process conditions are AUTHOR INFORMATION
maintained such that terephthalic acid is precipitated from
solution only when it leaves the reaction system, and the Corresponding Author
solvent is recycled to the reactors, being undersaturated *E-mail: jgomes@deq.isel.ipl.pt.
terephthalic acid. The terephthalic acid product may contain
Notes
less than 0.005 w/w % of 4-carboxybenzaldehyde. The low p-
xylene conversions and high temperatures and pressures are The authors declare no competing financial interest.
AT dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX
Chemical Reviews Review

Biographies Technical University of Lisbon. He worked for several years in


Portuguese National Laboratories, supporting industry and developing
R&D projects, and is now a Habilitated Coordinating Professor with
the Chemical Engineering Departmental Area at ISEL, Lisbon
Polytechnic, and a researcher of IBB.

ACKNOWLEDGMENTS
R.A.F.T. thanks ARTLANT PTA SA for all the training and
support given which allowed this paper to be written and also
Instituto Piaget−Pólo de Santo André for logistical support
during the literature research.

REFERENCES
(1) McIntyre, J. E. The Historical Development of Polyesters. In
Modern Polyesters: Chemistry and Technology of Polyesters and
Rogério A. F. Tomás holds a B.Sc. degree in Chemical Engineering Copolymers; Scheirs, J., Long, T. E., Eds.; John Wiley & Sons, Ltd.:
from the Technical University of Lisbon and a M.Sc. degree in Chichester, 2003.
Management and Industrial Strategy also from the Technical (2) Raghavendrachar, P.; Ramachandran, S. Ind. Eng. Chem. Res.
1992, 31, 453.
University of Lisbon. He was responsible for start-up of the new
(3) Sheehan, R. J. Terephthalic Acid, Dimethyl Terephthalate and
producing PTA plant from Artlant in Sines, Portugal, and is now Isophthalic Acid. Ullmann’s Encyclopedia of Industrial Chemistry; Wiley-
responsible for design and start-up of a pilot plant in The Netherlands. VCH Verlag GmbH & Co.: Weinheim, 2005; Vol. 35, pp 639−651.
He is also studying for his Ph.D. degree on optimization of the PTA (4) Saffer, A.; Barker, R. S. U.S. Patent 2833816, May 6, 1958.
production process. (5) Saffer, A.; Barker, R. S. GB Patent 807091, January 7, 1959.
(6) Saffer, A.; Barker, R. S. U.S. Patent 3089906, May 14, 1963.
(7) Landau, R.; Saffer, A. Chem. Eng. Prog. 1968, 64 (10), 20.
(8) Ciamician, G.; Silber, P. Ber. Dtsch. Chem. Ges. 1912, 45 (1), 38.
(9) Stephens, H. N. J. Am. Chem. Soc. 1926, 48 (7), 1824.
(10) Stephens, H. N. J. Am. Chem. Soc. 1926, 48 (11), 2920.
(11) Stephens, H. N. J. Am. Chem. Soc. 1928, 50 (9), 2523.
(12) Stephens, H. N. J. Am. Chem. Soc. 1928, 50 (1), 186.
(13) Stephens, H. N. J. Am. Chem. Soc. 1928, 50 (2), 568.
(14) Hartman, M.; Seibert, M. Helv. Chim. Acta 1932, 15 (1), 1390.
(15) King, E. P.; Swann, S., Jr.; Keyes, D. B. Ind. Eng. Chem. 1929, 21
(12), 1227.
(16) Halpern, J. Discuss. Faraday Soc. 1968, 46, 7.
(17) A General Discussion on Oxidation. Trans. Faraday Soc. 1946,
42, 99.
(18) Frank, C. E. Chem. Rev. 1950, 46 (1), 155.
(19) Russell, G. A., J. Chem. Educ., 1959, 36(3), 111.
(20) Mayo, F. R. Acc. Chem. Res. 1968, 1 (7), 193.
João C. M. Bordado holds a B.Sc. degree in Chemical Engineering and (21) Partenheimer, W. Catal. Today 1995, 23 (2), 69.
a M.Sc. degree in Chemical Engineering and was awarded his Ph.D. (22) Mulcahy, M. F. R.; Watt, I. C. Proc. R. Soc. London, Ser. A 1953,
degree in Chemical Engineering, all from the Technical University of 216, 10.
Lisbon. After several years working in industry on polymer plants and (23) Bolland, J. L. Q. Rev. Chem. Soc. 1949, 3, 1.
developing R&D projects, he is now Full Professor with the Chemical (24) Benson, S. W. J. Am. Chem. Soc. 1965, 87 (5), 972.
Engineering Department of IST, Technical University of Lisbon, and a (25) Hammett, L. P. J. Am. Chem. Soc. 1937, 59 (1), 96.
researcher of IBB. (26) Lowry, T. H., Richardson, K. S. Mechanism and Theory in
Organic Chemistry, 3rd ed.; Harper Collins: New York, 1987.
(27) Hammett, L. P. J. Am. Chem. Soc. 1937, 59 (1), 96.
(28) Russell, G. A. J. Am. Chem. Soc. 1956, 78 (5), 1047.
(29) Jaffé, H. H. Chem. Rev. 1953, 53 (2), 145.
(30) Haber, F.; Weiss, J. Die Naturwiss 1932, 20, 948.
(31) Haber, F.; Weiss, J. Proc. R. Soc. London, Ser. A 1934, 147, 332.
(32) Hagen, J. Industrial Catalysis. A Practical Approach; Wiley-VCH
Verlag GmbH & Co.: Weinheim, 2006.
(33) van Leeuwen, P. Homogeneous Catalysis. Understanding the Art,
2nd ed.; Kluwer Academic Publishers: Dordrecht, 2004.
(34) Donaldson, J. D.; Beyersmann, D. Cobalt and Cobalt
Compounds. Ullmann’s Encyclopedia of Industrial Chemistry; Wiley-
VCH Verlag GmbH & Co.: Weinheim, 2005; Vol. 35, pp 639−651.
(35) Reidies, A. H. Manganese Compounds. Ullmann’s Encyclopedia
of Industrial Chemistry; Wiley-VCH Verlag GmbH & Co.: Weinheim,
2005; Vol. 35, pp 639−651.
(36) Morimoto, T.; Ogata, Y. J. Chem. Soc. (B) 1967, 62.
João F. P. Gomes holds a B.Sc. degree in Chemical Engineering and (37) Hanotier, J.; Hanotier-Bridoux, M. J. Mol. Catal. 1981, 12, 133.
was awarded his Ph.D. degree in Chemical Engineering, both from the (38) Kamiya, Y.; Kashima, M. J. Catal. 1972, 25 (3), 326.

AU dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX


Chemical Reviews Review

(39) Kamiya, Y.; Kashima, M. Bull. Chem. Soc. Jpn. 1973, 46 (3), 905. (82) Jhung, S. H. Bull. Korean Chem. Soc. 2002, 23 (3), 503.
(40) Kashima, M.; Kamiya, Y. Bull. Chem. Soc. Jpn. 1974, 47 (2), 481. (83) Britt, P. F.; Buchanan, A. C., III; Eskay, T. P.; Mungall, W. S.
(41) Márta, F.; Boga, E.; Matók, M. Discuss. Faraday Soc. 1968, 46, Mechanistic Investigations into the Decarboxylation of Aromatic
173. Carboxylic Acids. Invited Paper for Presentation at the American
(42) Scott, E. J. Y.; Chester, A. W. J. Phys. Chem. 1972, 76 (11), 1520. Chemical Society National Meeting, New Orleans, Aug 22−26, 1999.
(43) Ichikawa, Y.; Yamashita, G.; Tokashiki, M.; Yamaji, T. Ind. Eng. (84) Partenheimer, W. J. Mol. Catal. 1991, 67 (1), 36.
Chem. 1970, 62 (4), 38. (85) Kenigsberg, T.; Ariko, N.; Agabekov, V. Energy Convers. Manage.
(44) Sharp, J. A.; White, A. G. J. Chem. Soc. 1952, 110. 1995, 36 (6−9), 677.
(45) Hendriks, C. F.; van Beek, H. C. A.; Heertjes, P. M. Ind. Eng. (86) Li, D.; Jin, H.; Shi, G. Chin. J. Proc. Eng. 2006, 6 (4), 539.
Chem. Prod. Res. Dev. 1978, 17 (3), 256. (87) Cheng, Y.; Peng, G.; Wang, L.; Li, X. Chin. J. Chem. Eng. 2009,
(46) Heiba, E. I.; Dessau, R. M.; Koehl, W. J., Jr. J. Am. Chem. Soc. 17 (2), 181.
1969, 24 (91), 6830. (88) Partenheimer, W. Appl. Catal., A: Gen. 2011, 409−410, 48−54.
(47) Ritchie, C. D., Sager, W. F. An Examination of Structure- (89) Roffia, P.; Calini, P.; Tonti, S. Ind. Eng. Chem. Res. 1988, 27 (5),
Reactivity Relationships. In Progress in Physical Organic Chemistry; 765.
Streitwieser, A., Jr., Taft, R. G., Cohen, S. G., Eds.; John Wiley & Sons: (90) Palmer, D. A.; Larson, K. D.; Fjare, K. A. U.S. Patent 5113015,
Weinheim, 1963; Vol. 2, pp 323−400. May 12, 1992.
(48) Brill, W. F. Ind. Eng. Chem. 1960, 52 (10), 837. (91) Michel, R. E.; Rudolph, R. G. U.S. Patent 5142097, Aug 25,
(49) Hendriks, C. F.; van Beek, H. C. A.; Heertjes, P. M. Ind. Eng. 1992.
Chem. Prod. Res. Dev. 1979, 18 (1), 43. (92) Partenheimer, W.; Graziano, D. J. U.S. Patent 5081290, Jan 14,
(50) Shriver, D. F., Atkins, P. W., Langford, C. H. Inorganic 1992.
Chemistry, 2nd ed.; Oxford University Press: Oxford, 1994. (93) Kimura, T.; Hashizume, H.; Izumisawa, Y. U.S. Patent 4051178,
(51) Jean, L.; Bernard, L. Can. J. Chem. Eng. 2010, 88, 67. Sept 27, 1977.
(52) McIntyre, J. E. U.S. Patent 2907792, Oct 6, 1959. (94) Shigeyasu, M.; Kusano, N. U.S. Patent 4160108, July, 3, 1979.
(53) McIntyre, J. E.; O’Neill, W. A. GB Patent 833438, Apr 27, 1960. (95) Seko, M.; Miyake, T.; Takeuchi, H.; Tanouchi, M. U.S. Patent
(54) Rust, F. F.; Vaughan, W. E. Ind. Eng. Chem. 1949, 41 (11), 2595. 4230882, Oct, 28, 1980.
(55) Bell, E. R.; Irish, G. E.; Raley, J. H.; Rust, F. F.; Vaughan, W. E. (96) Hashizume, H.; Komaya, T.; Fukuda, K. U.S. Patent 5763648,
Ind. Eng. Chem. 1949, 41 (11), 2609. June 9, 1998.
(56) Barnett, B.; Bell, E. R.; Dickey, F. H.; F., F.; Vaughan, W. E. Ind. (97) Scott, L. S.; Sommers, R. W. U.S. Patent 158738, June 19, 1979.
Eng. Chem. 1949, 41 (11), 2612. (98) Sumner, C. E., Jr.; Hembre, R. T.; Lange, D.; Lavoie, G. G.;
(57) Ravens, D. A. S. Trans. Faraday Soc. 1959, 55, 1768. Tennant, B. A.; Floyd, T. R.; Davenport, B. W.; Compton, D. B.; Bays,
(58) Bawn, C. E. H.; Wright, T. K. Discuss. Faraday Soc. 1968, 46,
J. N. U.S. Patent 2006/0205977 A1, Sept 14, 2006.
164. (99) Lavoie, G. G.; Hembre, R. T.; Sumner, C. E., Jr.; Bays, J., N., B.
(59) Hay, A. S.; Blanchard, H. S. Can. J. Chem. 1965, 43, 1306.
W.; Compton, Tennant, B. A.; Davenport, B. W.; Lange, D.; Floyd, T.
(60) Kamiya, Y. Tetrahedron 1966, 22 (7), 2029.
(61) Kamiya, Y. J. Catal. 1974, 33 (3), 480. R. U.S. Patent 2006/0205975 A1, Sept 14, 2006.
(100) Huang, H.; Wei, M.; Lin, Y.; Lu, P. J. Chromatogr. A 2009,
(62) Kamiya, Y.; Nakajima, T.; Sakoda, K. Bull. Chem. Soc. Jpn. 1966,
39 (10), 2211. 1216 (12), 2560.
(63) Sakota, K.; Kamiya, Y.; Ohta, N. Bull. Chem. Soc. Jpn. 1968, 41 (101) Allen, N. S.; Edge, M.; Daniels, J.; Royall, D. Polym. Degrad.
(3), 641. Stab. 1998, 62 (2), 373.
(64) Chavan, S. A.; Halligudi, S. B.; Srinivas, D.; Ratnasamy, P. J. Mol. (102) Roffia, P.; Calini, P.; Motta, L. Ind. Eng. Chem. Prod. Res. Dev.
Catal. A: Chem. 2000, 161 (1−2), 49. 1984, 23 (5), 629.
(65) Partenheimer, W. J. Mol. Catal. A: Chem. 2001, 174 (1−2), 29. (103) Sun, W.; An, M.; Zhong, W.; Zhao, L. Int. J. Chem. Kin. 2012,
(66) Srivastava, K. K.; Chandalia, S. B. J. Chem. Technol. Biotechnol. 44 (5), 277.
1981, 31 (1), 609. (104) Friedel, C.; Crafts, J. M. C. R. Acad. Sci. 1877, 84, 1392.
(67) Cao, G.; Servida, A.; Pisu, M.; Morbidelli, M. AIChE J. 1994, 40 (105) Friedel, C.; Crafts, J. M. C. R. Acad. Sci. 1877, 84, 1450.
(7), 1156. (106) Jhung, S. H.; Park, Y. Bull. Korean Chem. Soc. 2002, 23 (3),
(68) Cao, G.; Pisu, M.; Morbidelli, M. Chem. Eng. Sci. 1994, 49 369.
(24B), 5775. (107) Ingold, K. U. Chem. Rev. 1961, 61 (6), 563.
(69) Wang, Q.; Cheng, Y.; Wang, L.; Li, X. Ind. Eng. Chem. Res. 2007, (108) Denisov, E. T.; Khudyakov, I. V. Chem. Rev. 1987, 87 (6),
46 (26), 8980. 1313.
(70) Baeyer, A.; Villiger, V. Ber. Dtsch. Chem. Ges. 1899, 32 (3), 3625. (109) Russell, J.; Thomson, R. H. J. Chem. Soc. 1962, 3379.
(71) Sun, W.; Pan, Y.; Zhao, L.; Zhou, X. Chem. Eng. Technol. 2008, (110) Warner, K. F.; Bachrach, A.; Rehman, A.; Schnatter, W. F. K.;
31 (11), 1402. Mitra, A.; Shimanskas, C. J. Chem. Res., Synop. 1998, 12, 814.
(72) Li, K.; Li, S. Appl. Catal., A: Gen. 2008, 340 (2), 271. (111) Moraes, M. L. L.; Rubim, J. C.; Realpozo, R. R.; Tavares, M. F.
(73) Wang, Q.; Li, X.; Wang, L.; Cheng, Y.; Xie, G. Ind. Eng. Chem. M. J. Braz. Chem. Soc. 2004, 15 (3), 400.
Res. 2005, 44 (2), 261. (112) James, D. E. U.S. Patent 4782181, Nov 1, 1998.
(74) Csányi, L. J.; Jáky, K. J. Catal. 1993, 141 (2), 721. (113) Guisnet, M.; Cerqueira, H. S.; Figueiredo, J. L.; Ramôa Ribeiro,
(75) Partenheimer, W. Adv. Synth. Catal. 2004, 346 (2−3), 297. F. Desactivação e Regeneração de Catalisadores; Fundaçaõ Calouste
(76) Wang, Q.; Li, X.; Wang, L.; Cheng, Y.; Xie, G. Ind. Eng. Chem. Gulbenkian: Lisbon, 2008.
Res. 2005, 44 (13), 4518. (114) Partenheimer, W. J. Mol. Catal. A: Chem. 2003, 206 (1−2),
(77) Partenheimer, W. Adv. Synth. Catal. 2005, 347 (4), 580. 131.
(78) Cheng, Y.; Li, X.; Wang, L.; Wang, Q. Ind. Eng. Chem. Res. 2006, (115) Han, I.; Kim, M.; Han, C. Theor. Appl. Chem. Eng. 2002, 8 (2),
45 (12), 4156. 2989.
(79) Broeker, J. L.; Partenheimer, W.; Rosen, B. I. U.S. patent (116) Han, I.; Kim, M.; Lee, C.; Cha, W.; Ham, B.; Jeong, J.; Lee, H.;
5453538, Sept 26, 1995. Chung, C.; Han, C. Korean J. Chem. Eng. 2003, 20 (6), 977.
(80) Jhung, S. H.; Lee, K. H.; Park, Y. Appl. Catal. A: Gen. 2002, 230 (117) Liu, R.; Su, H.; Mu, S.; Jia, T.; Chen, W.; Chu, J. Chin. J. Chem.
(1−2), 31. Eng. 2004, 12 (2), 234.
(81) Jhung, S. H.; Lee, K. H.; Park, Y. Bull. Korean Chem. Soc. 2002, (118) Zhan, Y.; Su, H.; Liu, R.; Chu, J. Chin. J. Chem. Eng. 2005, 13
23 (1), 59. (5), 642.

AV dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX


Chemical Reviews Review

(119) Mu, S.; Su, H.; Gu, Y.; Chu, J. Chin. J. Chem. Eng. 2003, 11 (5), (156) Howarth, J. Tetrahedron Lett. 2000, 41 (34), 6627.
536. (157) Earle, M. J.; Katdare, S. P. WO Patent 02/30862 A1, Oct 5,
(120) Mu, S.; Su, H.; Gu, Y.; Chu, J. Comput. Chem. Eng. 2004, 28 2001.
(11), 2219. (158) Earle, M. J.; Katdare, S. P.; Seddon, K. R. Org. Lett. 2004, 6 (5),
(121) Hong, H.; Wenli, D.; Feng, Q.; Weimin, Z. Ind. Eng. Chem. Res. 707.
2010, 49 (12), 5683. (159) Yvari, I.; Karimi, E. Synth. Commun. 2009, 39 (19), 3420.
(122) Gujarathi, A. M.; Babu, B. V. Mater. Manuf. Processes 20069, 48 (160) Hronec, M.; Hrabě, Z. Ind. Eng. Chem. Prod. Res. Dev. 1986, 25
(12), 5883. (2), 257.
(123) Hronec, M.; Ilavský, J. Ind. & Eng. Chem. Prod. Res. Dev. 1982, (161) Chavan, S. A.; Srinivas, D.; Ratnasamy, P. J. Catal. 2001, 204
21 (3), 455. (2), 409.
(124) Fraga-Dubreuil, J.; Poliakoff, M. Pure Appl. Chem. 2006, 78 (162) Kim, Y.; Son, Y.; Ryu, J.; Yang, H.; Jun, K.; Park, S. J. Korean
(11), 1971. Ind. Eng. Chem. 2004, 15 (1), 131.
(125) Savage, P. E. J. Supercrit. Fluids 2009, 47 (3), 407. (163) Das, B. K.; Chakrabarty, R. J. Chem. Sci. 2011, 123 (2), 163.
(126) Reid, R. C.; Prausnitz, J. M.; Poling, B. E. The Properties of (164) Bastock, T. W.; Clark, J. H.; Martin, K.; Trenbirth, B. W. Green
Gases & Liquids, 4th ed.; McGraw-Hill Book Co.: New York, 1987. Chem. 2002, 4, 615.
(127) Dunn, J. B.; Savage, P. E. Ind. Eng. Chem. Res. 2002, 41 (18), (165) De Meyer, H.; Sijben, J. WO Patent 2006068471A1, June 6,
4460. 2006.
(128) Dunn, J. B.; Urquhart, D. I.; Savage, P. E. Adv. Synth. Catal. (166) Sijben, J.; De Meyer, H. WO Patent 2006068472A1, June 6,
2002, 344 (3−4), 385. 2006.
(129) Hamley, P. A.; Ilkenhans, T.; Webster, J. M.; Garcia-Verdugo, (167) Sijben, J.; De Meyer, H. U.S. Patent 2009/0088586A1, April 2,
E.; Venardou, E.; Clarke, M. J.; Auerbach, R.; Thomas, W. B.; Whiston, 2009.
K.; Poliakoff, M. Green Chem. 2002, 4 (3), 235. (168) Sijben, J.; De Meyer, H. U.S. Patent 2009/0062563A1, March
(130) Garcia-Verdugo, E.; Venardou, E.; Thomas, W. B.; Whiston, 5, 2009.
K.; Partenheimer, W.; Hamley, P. A.; Poliakoff, M. Adv. Synth. Catal. (169) Sijben, J.; De Meyer, H. U.S. Patent 8188309B2, March 29,
2004, 346 (2−3), 307. 2012.
(131) Garcia-Verdugo, E.; Fraga-Dubreuil, J.; Hamley, P. A.; Thomas, (170) De Meyer, H.; Sijben, J. U.S. Patent 8187992B2, March 29,
W. B.; Whiston, K.; Poliakoff, M. Green Chem. 2005, 7 (5), 294. 2012.
(132) Osada, M.; Savage, P. E. AIChE J. 2009, 55 (3), 710.
(133) Pérez, E.; Fraga-Dubreuil, J.; García-Verdugo, E.; Hamley, P.
A.; Thomas, W. B.; Housley, D.; Partenheimer, W.; Poliakoff, M. Green
Chem. 2011, 13 (9), 2389.
(134) Pérez, E.; Fraga-Dubreuil, J.; García-Verdugo, E.; Hamley, P.
A.; Thomas, M. L.; Yan, C.; Thomas, W. B.; Housley, D.;
Partenheimer, W.; Poliakoff, M. Green Chem. 2011, 13 (9), 2397.
(135) Saha, B.; Espenson, J. H. J. Mol. Catal., A: Chem/ 2007, 271
(1−2), 1.
(136) Hanotier, J.; Hanotier-Bridoux, M.; de Radzitzky, P. J. Chem.
Soc., Perkin Trans. 1973, 2 (4), 381.
(137) Partenheimer, W. J. Mol. Catal. A: Chem. 2003, 206 (1−2),
105.
(138) Jhung, S.; Lee, K.; Park, Y.; Yoo, J. S. U.S. Patent 6180822 B1,
Jan, 30, 2001.
(139) Jhung, S.; Park, Y.; Lee, K.; Yoo, J. S.; Chae, J. U.S. Patent
6194607 B1, Feb, 27, 2001.
(140) Yoo, J. S.; Jung, S.; Lee, K.; Park, Y. Appl. Catal., A: Gen. 2002,
223 (1−2), 239.
(141) Burri, D. R.; Jun, K.; Yoo, J. S.; Lee, C. W.; Park, S. Catal. Lett.
2002, 81 (3−4), 169.
(142) Baek, S.; Roh, H.; Chavan, S. A.; Choi, M.; Jun, K.; Park, S.;
Yoo, J. S.; Kim, K. Appl. Catal., A: Gen. 2003, 244 (1), 19.
(143) Park, S.; Yoo, J. S. Stud. Surf. Sci. Catal. 2004, 153, 303.
(144) Zuo, X.; Subramaniam, B.; Busch, D. H. Ind. Eng. Chem. Res.
2008, 47 (3), 546.
(145) Zuo, X.; Niu, F.; Snavely, K.; Subramaniam, B.; Busch, D. H.
Green Chem. 2010, 12 (2), 260.
(146) Ishii, Y.; Sakaguchi, S.; Iwahama, T. Adv. Synth. Catal. 2001,
343 (5), 393.
(147) Tashiro, Y.; Iwahama, T.; Satoshi, S.; Ishii, Y. Adv. Synth. Catal.
2001, 343 (2), 220.
(148) Saha, B.; Koshino, N.; Espenson, J. H. J. Phys. Chem. A 2004,
108 (3), 425.
(149) Ishii, Y.; Sakaguchi, S. Catal. Today 2006, 117 (1−3), 105.
(150) Cheng, Y.; Li, X.; Wang, Q.; Wang, L. Ind. Eng. Chem. Res.
2005, 44 (20), 7756.
(151) Gund, P.; Guanidine. J. Chem. Educ. 1972, 49 (2), 100.
(152) Earl, M. J.; Seddon, K. R. Pure Appl. Chem. 2000, 72 (7), 1391.
(153) Sheldon, R. Chem. Commun. 2001, 23, 2399.
(154) Welton, T. Coord. Chem. Rev. 2004, 248 (21−24), 2459.
(155) Saleh, R. Y. U.S. Patent 6320083 B1, Nov 20, 2001.

AW dx.doi.org/10.1021/cr300298j | Chem. Rev. XXXX, XXX, XXX−XXX

View publication stats

You might also like