Selective aerobic oxidation of para-xylene in sub- and supercritical water. Part 1. Comparison with ortho-xylene and the role of the catalyst

Article abstract of DOI:10.1039/c1gc15137a

The selective, continuous, aerobic oxidations of para-xylene (pX) and ortho-xylene (oX) were performed in an identical fashion in supercritical water. The xylenes were oxidized without a catalyst and with hydrobromic acid, cobalt(ii) and manganese(ii) bromide catalysts. The conversions and yields to phthalic acid (OA) from oX were always significantly higher than those for terephthalic acid (TA) from pX. The formation of CO₂ was significantly higher for pX than oX despite the higher conversions to oX. These results are unexpected because the literature teaches that thermal and catalytic decarboxylation is much higher for OA than TA. The superior yields from oX are consistent with a lower steady-state concentration of hydroxyl radicals, OH due to the internal, concerted attack of the peroxides with the oX methyl group. This mechanism forms the phthalide directly from o-tolualdehyde (oTOL) which is consistent with the observation that ortho-toluic acid (OTA) is much lower in oX than para-toluic acid, PTA, in pX oxidation. This mechanism also lowers the steady-state concentration of aromatic acids consistent with the observed lower benzoic acid and CO₂ yields. Overall, the results suggest that the metal catalysts can play more than one role, thereby opening up the opportunity for discovering new catalytic synergies which are explored in our next paper, Part 2 of this series.

Full text of DOI:10.1039/c1gc15137a

View Online / Journal Homepage / Table of Contents for this issue
Dynamic Article Links
Green Chemistry
Cite this: Green Chem., 2011, 13, 2389
www.rsc.org/greenchem
PAPER
Selective aerobic oxidation of para-xylene in sub- and supercritical water.
Part 1. Comparison with ortho-xylene and the role of the catalyst
a
a,b
a
a
b
Eduardo P e´ rez,* Joan Fraga-Dubreuil, Eduardo Garc ´ı a-Verdugo, Paul A. Hamley, W. Barry Thomas,
b
c
a
Duncan Housley, Walt Partenheimer* and Martyn Poliakoff*
Received 3rd February 2011, Accepted 18th May 2011
DOI: 10.1039/c1gc15137a
The selective, continuous, aerobic oxidations of para-xylene (pX) and ortho-xylene (oX) were
performed in an identical fashion in supercritical water. The xylenes were oxidized without a
catalyst and with hydrobromic acid, cobalt(II) and manganese(II) bromide catalysts. The
conversions and yields to phthalic acid (OA) from oX were always significantly higher than those
for terephthalic acid (TA) from pX. The formation of CO was significantly higher for pX than oX
2
despite the higher conversions to oX. These results are unexpected because the literature teaches
that thermal and catalytic decarboxylation is much higher for OA than TA. The superior yields
∑
from oX are consistent with a lower steady-state concentration of hydroxyl radicals, OH due to
the internal, concerted attack of the peroxides with the oX methyl group. This mechanism forms
the phthalide directly from o-tolualdehyde (oTOL) which is consistent with the observation that
ortho-toluic acid (OTA) is much lower in oX than para-toluic acid, PTA, in pX oxidation. This
mechanism also lowers the steady-state concentration of aromatic acids consistent with the
observed lower benzoic acid and CO yields. Overall, the results suggest that the metal catalysts
2
can play more than one role, thereby opening up the opportunity for discovering new catalytic
synergies which are explored in our next paper, Part 2 of this series.
Introduction
In this paper, we compare the results of oxidizing ortho-
and para-xylene, oX and pX. The differences in the product
distribution between the two xylenes suggest differences in the
mechanisms of the two oxidations and provide new insights
into the role of the catalyst and the mechanisms of burn of the
This is the first of two papers in which we report studies on
the feasibility of producing terephthalic acid, TA, by aerobic
oxidation of p-xylene, pX, in high temperature water, HTW, or
supercritical water, SCW. Our overall aim has been to develop
a better understanding of the reaction and to discover new
catalysts and better reaction conditions to give higher yields
6
aromatics. The second paper, describes our successful attempts
to find better catalysts and their optimization.
Research activity has been growing in the field of SCW (T
c
=
of TA and less burning of pX to CO
2
.
◦
3
74 C, p
c
= 221 bar) or more generally in high-temperature
Previously, we have reported that (a) high yields of TA can
be obtained by oxidation of pX in SCW using a continuous
water because of the unusual properties of H
conditions. A wide range of chemical reactions have been
2
O under these
7
1,2
reactor, (b) more generally, aromatic acids can be obtained
studied where SCW plays the role of a solvent and/or reactant
3
from oxidation of alkylaromatic compounds, (c) the simul-
taneous oxidation of mixtures of xylenes is feasible and (d)
increased Brønsted acidity, increased bromide concentration,
and the addition of aromatic acids can all substantially reduce
the precipitation of manganese oxides that can occur when
7–12
and/or catalyst.
Research on the homogeneously catalysed
4
oxidation of methylaromatics in HTW and SCW was first
13
1–4,14–21
reported by Holliday et al. and subsequently by others.
The aerobic oxidation of pX to TA is carried out indus-
trially using acetic acid as the solvent. Despite its extraordi-
nary efficiency (99% selectivity, >95% yield), the process still
5
MnBr is used as a catalyst.
2
22
has some drawbacks. The crude TA contains ~0.3% of 4-
carboxybenzaldehyde (4CBA) that must be removed via a costly
and energy-consuming process. Another negative factor is loss
of ca. 0.05 kg of acetic acid per kg of TA manufactured due
a
University of Nottingham, Nottingham, UK NG7 2RD. E-mail:
martyn.poliakoff@nottingham.ac.uk, eduardo.velilla@nottingham.ac.uk;
Fax: +44 115 951 3058; Tel: +44 115 951 3520
23
b
to combustion. Moreover, solvent degradation generates the
INVISTA Textiles UK Limited, Wilton, Cleveland, UK TS10 4RF
c
greenhouse gas CO2 and the ozone-depleting agent CH Br.
DuPont (retired), Consultant, Current address 7253 SW 26th Ave.,
3
Portland, OR, 97219, USA. E-mail: wpartenheimer@msn.com
Finally, re-use of the solvent necessitates an energy-consuming
This journal is © The Royal Society of Chemistry 2011
Green Chem., 2011, 13, 2389–2396 | 2389

View Online
distillation step to remove the H
2
O formed as a by-product (2
where [A] is the molar concentration of the acid, [X] is the
molar concentration of pX or oX if the conversion had been
moles per TA mole produced). Replacing the acetic acid by water
could overcome these drawbacks as well as giving more efficient
heat recovery due to the higher oxidation temperatures utilised
0. i.e. allowing for dilution. [X] = (F
total are the flow rates of the xylene and the total flow after
quench; r and M are the density and molar mass of the
xylene, respectively. CO yield was calculated from the carbonate
concentration obtained from titration by eqn (2), with [BA]
X
r
X
)/(Ftotal
M
X
), where F ,
X
F
22
in HTW and SCW. Water is therefore a serious candidate for
replacing acetic acid due to the combination of economic and
environmental benefits that it offers.
X
X
2
Currently, a potential major economic inconvenience of the
water-based method is the undesired total oxidation (burn) of
pX and decarboxylation of the aromatic acids. Minimizing burn
and decarboxylation in HTW is therefore an important goal
in making this new methodology commercially viable for TA
manufacture. This paper reports differences in yields, selectivity
and burn of xylene between oX and pX oxidation in SCW, which
provide evidence for the mechanisms that cause total oxidation.
The metal/bromide catalyzed autoxidation of pX and oX as
being subtracted to avoid counting the CO
from decarboxylation.
2
molecules arising
2
−
[
CO3 ]−[BA]
Y CO2 =100
(2)
8[X]
Finally we define “burn” as YCO /Y
A
; that is the number of
2
moles of xylene totally oxidized per mole of dicarboxylic acid
produced. Hence burn is a measure of the amount of xylene that
is lost through combustion.
24
well as p- and o-toluic acid have been reviewed and summarized
but there is only a small amount of literature which directly
2
5–27
compares results from oX and pX even in acetic acid.
4,5
Oxidation of oX in SCW has been reported previously, but
Results
without any data for formation of CO
catalyst stability in the oxidation of oX in SCW and, although
we have noticed differences between the isomers of xylene,
2
. Recently, we studied
6
Our strategy has been to run the oxidation of pX and oX in
the same reactor and in an identical fashion so that genuine
similarities and differences can be identified and directly com-
pared. The following catalysts were used: four concentrations of
2,4
this paper is the first time that we have studied these differences
systematically.
HBr, CoBr
2
, and MnBr
2
by itself or with either HBr or benzoic
5
acid added to increase the solubility of the catalyst. Control
experiments were run without any added catalyst.
Method
3
Fig. 2 summarizes the yields of the dicarboxylic acids and
A simplified schematic of our tubular continuous flow reactor is
CO
have the effect of increasing the yield of diacid compared to the
control experiments. Also, under these conditions, CoBr is a
2
. It can be seen that, for both oX and pX, all of the catalysts
shown in Fig. 1. The reactions were carried out at a temperature
◦
of 380 C and a pressure of 230 bar. More details of the
experimental equipment and procedure can be found in the
2
6
much poorer catalyst than MnBr for the selective oxidation of
Experimental section and in the second paper of this series.
2
both xylenes.
However, it is the differences in the oxidation of these two
apparently similar substrates that are particularly striking. The
yields of o-phthalic acid, OA, are always higher than those of
TA, with and without the presence of catalyst. That is surprising
in view of the fact that OA is known to deactivate the Co/Mn/Br
catalyst in acetic acid, due to catalytic decarboxylation and lower
29
yields are observed for oX than pX. Fig. 2 also shows that the
CO yield in SCW is always higher when using pX, except for
2
the uncatalyzed reaction where it is similar to oX. However, even
in the uncatalysed reaction, the OA yield is much higher than
that of pX; so the burn will be much lower for oX. This suggests
Fig. 1 Simplified scheme of the continuous reactor used in this work.
O
2
is generated from aqueous H
2
O
2
by high temperature decomposition
/H O, pure xylene, and an aqueous
3
that the intrinsic amount of CO being generated by pX is always
2
in a pre-heater, PH. The mixture O
2
2
higher than that of oX (even for the uncatalyzed reaction).
Scheme 1 shows the various intermediates expected for the
oxidation of pX and oX, and their sequence. The sequence
of formation of intermediates for pX in acetic acid is well
solution of the catalyst are continuously fed into the reactor by means
of HPLC pumps. After the reactor, the mixture is quenched with a
cold solution of NaOH that neutralizes any CO to form carbonate. An
2
average residence time of 5.8 s is calculated from the reactor volume, the
total flow rate and the density of pure water, obtained from the NIST
30
established as well as their relative reactivities and our previous
28
1–5
database.
work has shown that the sequence is likely to be the same in
SCW. Table 1 summarizes the observed selectivity of the main
The concentrations of dicarboxylic acids and the intermedi-
ates in the product stream were analysed by HPLC. The yield of
intermediates as well as the yields of diacid and CO and the
overall burn. Together, Fig. 2 and Table 1 indicate the following:
1. HBr catalyses both the formation of diacids and yield
2
dicarboxylic acid, Y , was then calculated by eqn (1).
A
of CO
formation of OA than of TA (see also Fig. 3) and it increases
the yield of CO from pX more than from oX.
2
. However, HBr is far more effective at catalyzing the
[
[
A]
X]
Y A =100
(1)
2
2
390 | Green Chem., 2011, 13, 2389–2396
This journal is © The Royal Society of Chemistry 2011

View Online
◦
Fig. 2 Summary of the yields of dicarboxylic acid (OA or TA) and CO
2
obtained from the oxidation of [ꢀ] oX and [ꢁ] pX in SCW at 380 C and 230
bar with a number of different catalysts. For detail, see Table 1. The data clearly show that under all circumstances oX produces more dicarboxylic
a
b
acid than does pX, and almost always less CO
2
than pX. Note that the % scale for CO
2
is expanded compared to the diacid. [HBr] = 2.57 mM.
c
d
[
HBr] = 5.13 mM. [HBr] = 10.3 mM. [HBr] = 20.5 mM.
◦
Table 1 Comparison of the yields and selectivities of the autoxidation of pX (in bold) and oX (italics) in SCW at 380 C and 230 bar
Yield, mol (%)
Selectivity (%)
Entry Catalyst species [Metal]/mM [Br]/mM CO TA or OA Burn TA or OA Toluic acid CBA HMBA or phthalide TOL BA
2
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
None
HBr
HBr
HBr
HBr
0.0
0.00
2.57
5.13
10.3
20.5
5.13
5.13
7.70
5.13
14.3 0.6
14.6 16.0
26.6 3.7
11.8 36.2
29.8 8.0
13.7 44.2
32.4 15.8
15.2 58.6
33.7 19.1
23.4 55.6
14.8 6.3
12.4 33.9
18.4 36.1
13.4 55.0
14.3 22.7
11.1 56.7
17.5 35.6
13.1 52.4
23.8 8.4
73.3
1.8
65.1
2.2
52.2
2.9
27.3
0.7
21.0
0.0
56.4
3.3
34.9
3.2
44.9
4.3
4.7
20.1 28.1
3.9 2.0
11.0 23.7
1.8
9.9
1.2
4.3
0.0
0.6
0.0
0.6
0.0
1.3
0.0
3.0
0.8
0.0
0.0
0.0
0.0
0.0
0.0
1.9
2.6
5.7
2.7
7.2
2.9
9.8
3.4
7.4
2.4
2.8
3.2
6.0
2.1
3.7
1.5
6.3
1.4
0.91 46.3
7.19 19.1
0.33 60.3
3.73 30.8
0.31 69.1
2.05 59.9
0.26 94.1
1.76 68.1
0.42 97.6
2.35 19.4
0.37 58.1
0.51 51.3
0.24 75.2
0.63 39.5
0.20 79.9
0.49 48.7
0.25 75.0
0.0
0.0
7.1
6.6
2.2
0.7
0.2
0.0
2.0
18.5
0.2
1.1
0.0
0.0
0.0
0.0
0
1
2
3
4
5
6
7
8
CoBr
MnBr
MnBr
MnBr
2
2.57
2.57
2.57
2.57
15.8 2.6
15.2 19.3
2
2
2
6.3
7.0
8.1
3.3
7.0
5.0
1.4
12.5
3.8
a
+ HBr
11.0
1.8
b
+ BA
36.2
2.6
16.0
a
b
MnBr
2
+ HBr in 1 : 1 molar ratio; ratio Br : metals is 3 : 1. BA concentration is 45.5 mM.
2
. MnBr
TA and of OA, which is to be expected since most of the
studies of pX oxidation in SCW have focused on MnBr as the
2
is a good catalyst for the formation both of
4. Under the same catalytic conditions, the proportion
of intermediates with two methyl groups totally or partially
oxidized is always higher for oX than pX; i.e. by inspection
of Table 1 one can see that the sum of selectivities for the diacid,
CBA, and 4HMBA (or phthalide) is higher than those of TOL
and the toluic acid for the oxidation of oX.
2
1
–5,14–17,20–21
catalyst.
However, addition of HBr to MnBr
2
reduces
the yield of TA compared to that obtained with MnBr
2
alone.
3
. Benzoic acid (BA) always forms during the autoxidation
of xylenes. The selectivity to BA was higher for pX than for
oX in seven of the nine different catalyst solutions used in this
paper. A maximum value of 9.8% is seen in entry 7 of Table 1.
The ratio of BA to diacid yield is greater for pX oxidation than
5. The observed amounts of phthalide are high, reaching
almost 30% in some cases. This strongly suggests that the second
pathway to OA is significant, see reaction 11 of Scheme 1.
6. The selectivity to o-toluic acid is always significantly lower
than that to p-toluic acid. However the selectivity to p-toluic acid
decreases markedly as the concentration of HBr is increased; see
oX oxidation. This is analogous to burn, defined as CO
ratio. See Fig. 3.
2
:diacid
This journal is © The Royal Society of Chemistry 2011
Green Chem., 2011, 13, 2389–2396 | 2391

View Online
Scheme 1 Suggested sequence of formation of intermediates during
the aerobic oxidation of (a) pX to TA and (b) oX to OA.
Table1. HBr also reduces the burn, even though the total yield
of CO increases; see Fig. 3.
2
Discussion
Considering the industrial importance of avoiding decarboxy-
lation and burn in the oxidation of methyl aromatics in general
and of pX in particular, the main question that needs to be
resolved is: what are the mechanisms mainly responsible for these
undesirable phenomena? The differences between the oxidations
of oX and pX can give us a clue. The higher yields of OA than
TA as well as the lower burn, point to factors which might favour
oX oxidation over pX and these factors are probably related to
an inhibition in SCW of the decarboxylation and burn of oX
compared to pX.
Fig. 3 Comparison of uncatalyzed and HBr catalyzed autoxidation
of pX (black symbols) and oX (white symbols) as a function of the
concentration of HBr. (ꢁ,ꢀ) Diacid yield, (᭹,᭺) toluic acid yield,
∑
The hydroxyl radical, OH , can be easily generated by thermal
∑
∑
(᭡,ꢀ) burn, (᭜,᭛) ratio YBA/YDiAcid
.
decomposition of the organic peroxides, ROOH → RO + OH ,
at several points in the oxidation pathways from xylene to
diacid. The hydroxyl radical is a powerful oxidant that reacts
unselectively with organic substrates. It prefers to react with the
∑
Table 2 Rate constants for the reaction of OH with selected species in
31
aqueous solution at room temperature
Reacting
species
Rate constant/L
mol s
31,32
benzylic ring rather than the methyl group of xylenes
which
-1 -1
Reaction product
pH
leads to formation of phenol, a strong anti-oxidant, followed by
31
Br-
Co(II)
Mn(II)
pX
BrOH-
Co(III)OH
Mn(III)OH
HOPh(CH )
3 2
HOPh(CH3)2
10
its total destruction to carbon dioxide. Elegant pulse radiolysis
1
7
6.7
~7
~7
1.1 ¥ 10
5
∑
8 ¥ 10
experiments on SCW have shown that OH radicals react with
7
9
9
2.9 ¥ 10
7.0 ¥ 10
6.7 ¥ 10
33,34
aromatics at close to diffusion controlled rates.
Therefore the
∑
lifetime of OH is far too short to be detected in our experiments.
Nevertheless many of the observations described in this paper
can be rationalized by effects that decrease the concentration
of the unselective hydroxyl radical and by the behavior of the
Mn(II) catalyst.
oX
∑
∑
Br + PhCH
3
→ HBr + PhCH
2
(4)
-
∑
Thus the reaction of Br can decrease the concentration of OH
and hence increase the overall rate of reaction, the yield to the
desired diacid, and decrease the rate of CO formation.
2
Rationalization of the HBr catalyzed reactions
One of the factors leading to the increased selectivity in the
Oxidation of oX
∑
-
presence HBr is that OH can react very rapidly with Br , see
∑
One of our key observations has been the relatively large
amounts of phthalide that are formed in the oxidation of
oX. We suggest that this is the result of a second pathway to
the formation of OA in SCW, see reaction 11 of Scheme 1. In
particular, we propose that oTOL can be converted directly to
Table 2, to give Br which can then propagate the reaction
∑
selectively unlike the indiscriminate reaction of OH with the
aromatic ring.
-
∑
∑
-
Br + OH → Br + OH
(3)
2
392 | Green Chem., 2011, 13, 2389–2396
This journal is © The Royal Society of Chemistry 2011

View Online
∑
-
phthalide via reaction 11 in Scheme 1. In detail, we believe that
reaction 11 could proceed by the peroxyacid of oX reacting
with the adjacent methyl group in a concerted fashion to form
phthalide, as shown in Scheme 2.
Mn(II) + OH → Mn(III) + OH
(9)
Assuming that the relative redox potentials are the same at
80 C as they are in ambient water, reaction (9) should
be thermodynamically favoured since the standard reduction
potential of OH /OH is 2.7 V in ambient water compared to
.5 V for Mn(III)/Mn(II).
The MnBr catalyst gives higher yields of both OA and TA
than does CoBr as well as lower burn. One reason for this
may be the fact that Co(II) reacts with the OH radical 36 times
slower than Mn(II), see Table 2. The higher activity of MnBr
has been reported for pX oxidation in HTW and SCW but the
comparison has not previously been made for oX.
Table 1 shows that addition of HBr to the MnBr catalyst
◦
3
∑
-
31
36
1
2
2
Scheme 2 Suggested pathway from oTOL to phthalide.
2
1
6
In this pathway, the formation of o-toluic acid from oTOL
(reaction 2, Scheme 1b) can be skipped by direct formation of
2
decreases the TA yield but slightly increases the OA yield.
This may be the result of competing effects; addition of HBr
increases the concentration of Br , which might accelerate
phthalide from oTOL. This could explain why less o-toluic acid
is observed in SCW from oxidation of oX than p-toluic acid
∑
-
from pX. Furthermore, the intramolecular transfer of OH in
∑
selective oxidation, but also lowers the pH which could promote
Scheme 2 avoids formation of free OH and therefore reduces
∑
24
burn.
the steady-state concentration of OH in oX oxidations. This
pathway could also occur with the primary hydroperoxide of
the oX in an analogous way. Since transfer of OH also oxidizes
∑
Mechanisms for decarboxylation and burn of aromatic acids
the adjacent methyl group, it increases the overall reaction rate
Savage and coworkers have reported the relative rates of thermal
while simultaneously reducing unwanted burn via unselective
decarboxylation in water at various temperatures as OA > TA >
∑
attack by OH on the C–H bonds of the aromatic ring. Since no
37
BA. But even for OA, the process is slow compared to the
analogue of Scheme 2 is possible for pX, more ring attack would
be expected in pX reactions. Thus, especially in the uncatalysed
reactions of pX, one would expect (a) a lower yield of TA and (b)
residence times in our reactor, e.g. 73% of OA decarboxylates
◦
to BA in 1 h at 300 C. Therefore, thermal decarboxylation is
unlikely to be responsible for the large amounts of CO
in our reactions, particularly because the yield of BA is much
lower than that of CO . On the other hand, if the peroxy acid
intermediates discussed above do not react via intramolecular
2
observed
a higher rate of CO formation due to ring degradation, precisely
2
what is observed experimentally.
2
The steps in Scheme 2 occur in the later stages of the oxidation
and there must be an additional initiation step of the uncatalyzed
reaction. Since thermal cracking of aliphatic hydrocarbons starts
∑
∑
transfer of OH , they will not only liberate OH but will also
open up pathways leading to decarboxylation and ultimately
◦
35
◦
at ~400 C, there should be sufficient thermal energy at 380 C
to dissociate a benzylic C–H bond of xylene in preference to
the stronger bond of the aromatic carbons, reaction (5). This
preferential dissociation will lead to selective autoxidation of a
methylaromatic even without a catalyst.
38
burn, for both oX and pX, see Scheme 3.
There is also the possibility that the metal ions could catalyse
the decarboxylation (Scheme 3, reaction 4). We have previously
pointed out similarities between the mechanism of the metal
3
catalyst in acetic acid and SCW. It is known that, in acetic
Initiation step:
acid, transition metals have a role as redox catalysts recycling
Br(-1) to Br(0), Scheme 4. Metals with a weak ligand-field
environment also have sufficient energy for decarboxylation of
aromatic carboxylic acids e.g. Co(III), Mn(III), Ce(IV), Ag(II),
∑
∑
PhCH
3
→ PhCH
2
(benzylic radical) + H
(5)
(6)
Propagation steps:
39–43
∑
∑
Cu(II).
PhCH
PhCH
2
+ O
2
→ PhCH
2
OO (peroxy radical)
Catalytic decarboxylation may occur when a carboxylic acid
enters the coordination sphere of the metal while it is in its
lower oxidation state. The metal is oxidized in the redox cycle
in Scheme 4 but then the acid decarboxylates releasing a proton
and deactivates the catalyst by reducing the metal ion back
to its lower oxidation state without oxidation of Br(-1). In
∑
2
OO + PhCH
3
→ PhCH
2
OOH (benzylic
∑
(
7)
hydroperoxide) + PhCH
2
Rationalization of results from metal-catalyzed oxidation of pX
and oX
an oxidative environment, as well as CO
2
, the products of
decarboxylation are phenols and quinones which are also known
It can be seen from Table 1 that, during the oxidation of both
44
to strongly retard the rate of reaction. However, the fact that
there is such a marked difference in burn between oX and pX in
the absence of catalyst or with HBr, i.e. without metals present,
xylenes, the MnBr
2
catalyst gives substantially higher diacid
-
yields than does HBr, at the same total Br concentration. One
effect of Mn(II) is to reduce the steady state concentration of
∑
∑
suggests that reactions of OH are more significant than metal-
OH by reacting selectively with the benzylic hydroperoxides:
catalyzed decarboxylation and burn in SCW.
We have included CO in because we detected CO in a number
Mn(II) + CH
3
PhCH
CH
2
OOH → Mn(III)(OH) +
(
8)
∑
3
PhCH O
2
of experiments. However, the amount was low compared to CO ,
2
And also by the oxidation of Mn(II) by the OH radical, see
Table 2:
<10%, and there did not appear to be any significant difference
between the relative amounts of CO generated from oX and pX.
This journal is © The Royal Society of Chemistry 2011
Green Chem., 2011, 13, 2389–2396 | 2393

View Online
are quite different from those observed in SCW. Deliberate
spiking of selected aromatics into a Co/Mn/Br autoxidation
in acetic acid has demonstrated that OA undergoes catalytic
decarboxylation much more readily than TA and this results
29
in a much stronger deactivation. It has been reported that
the initial rate of oxidation for oX is greater than pX in the
Co/Mn/Br catalyzed oxidation in acetic acid but the rates
26
reverse themselves in the second half of the reaction, when
high concentration of OA strongly deactivates the reaction. In
the Co/Br catalyzed oxidation of ortho- and para-toluic acid
in acetic acid, at low initial toluic acid concentrations, the
rate of oxidation of o-toluic is faster than p-toluic acid but,
25
at higher concentrations, the opposite is observed. This is
expected since higher concentration of OA form at higher con-
centration of o-toluic acid, hence more strongly deactivating the
system.
Finally, in the cobalt catalyzed oxidation of oX and pX in
acetic acid, the reaction abruptly stops when OA starts to form
27
while with pX the reaction continues to produce TA. The
steady-state concentration of Co(III) is much higher in a cobalt
catalyzed oxidation than a Co/Br oxidation; hence catalytic
decarboxylation is more prominent accounting for the abrupt
end of the oxidation when OA starts to form.
Thus in acetic acid, OA has a much stronger negative influence
on the reaction rate than TA in acetic acid but, in SCW,
the opposite is observed. Catalytic decarboxylation seems to
be an important mechanism for oX oxidation in acetic acid;
however, there must be an inhibition to this phenomenon
in SCW. One can rationalize these differences by postulating
that OA is the predominant species in acetic acid while in
SCW, phthalic anhydride predominates which cannot undergo
catalytic decarboxylation.
This hypothesis seems reasonable in view of the properties of
acetic acid and water at high temperature because the dielectric
7
constant of water decreases to values typical of an organic
solvent. Therefore ionic and polar solutes are disfavored versus
Scheme 3 Formation of benzoic acid and carbon dioxide. R = –CH
3
or
–
COOH.
neutral and non-polar compounds. There are some examples
45,46
of SCW acting as a dehydrating medium,
that would favour
the formation of the cyclic condensation product such as the
46
reaction of 1,4-butanediol to tetrahydrofuran. Furthermore,
entropy may favour the dehydration of OA to phthalic anhydride
at these elevated temperatures. Finally, carboxylic acids form
strong H-bonds between them so one can think that aromatic
acids are well solvated in acetic acid. By contrast, the number
7
of hydrogen bonds decreases in SCW, possibly favoring the
existence of the anhydride. This could have a significant indus-
trial importance because it would allow the direct formation of
phthalic anhydride, skipping the dehydration step.
Experimental
Scheme 4 Catalytic decarboxylation in metal/bromide autoxidations.
Reactions were carried out in a continuous rig similar to the
one reported in reference 1 and shown in Fig. 4. The reactor
39–43
There is some evidence
that the carboxylic acid is in the coordination
1
sphere of the metal during the decarboxylation.
was made of Hastelloy C276 pipe inch external diameter and
4
0
.46 cm inner diameter. The length of the reactor was 34 cm. The
-
1
total volumetric flow rate through the rig was 12 mL min . At
Comparison between oxidation in acetic acid and SCW
◦
380 C the residence time in the reactor was 5.8 s. The density of
28
Large differences in yields and intermediates are found when oX
pure water is used for these calculations. All the components
25–27
and pX are oxidized in acetic acid.
However these differences
are fed into the reactor via reciprocating pumps.
2
394 | Green Chem., 2011, 13, 2389–2396
This journal is © The Royal Society of Chemistry 2011

View Online
Phenol concentrations are very low because it is oxidized much
more rapidly than pX. Benzene is present in extremely small
amounts because its formation would require the sequential
decarboxylation of terephthalic acid to benzoic acid and then of
benzoic acid to benzene.
2
-
CO
3
concentration was measured by titration of the sample
-
with 0.2 N HCl. The first titre accounts for the excess OH
and the second for the protonation of CO
2
-
-
3
to HCO . The
3
difference between these two titres was used to calculate the
carbonate concentration. Further details for the experimental
apparatus and can be found in the ESI of the second paper in
6
this series.
Conclusions
Fig. 4 Configuration of the continuous aerobic oxidation reactor. P1–
pressure transducers. T2–5: log temperatures. T1, T6 and T8: control
temperatures.
3
This paper reports the unexpected observation that oX is more
easily oxidized to OA in supercritical water than pX is converted
to TA. In addition, the burn to CO is less for oX. We attribute
2
these differences to the presence of ortho methyl groups in oX,
which facilitates an internal transformation of the peroxide
directly to the phthalide thereby reducing the steady-state
concentration of OH radical as well as lowering the steady-state
concentration of the intermediate aromatic acids. Elucidating
these differences has provided insights into the mechanisms of
burn and decarboxylation. We have also shown that the presence
ofacatalysthas asignificantimpact onthe yields andselectivities
of both the desired and the undesired products. The fact that the
metal can play more than one role opens up the possibility of
synergistic effects in catalysts containing more than one metal.
Dioxygen was used as primary oxidant. An aqueous H
solution was passed through a coiled pre-heater at supercritical
temperature where it decomposed into a homogeneous mixture
2
O
2
of O
residence time in the pre-heater was long enough to achieve
total decomposition of H . The catalysts were introduced in
2
and SCW before being driven to the reactor. The
2
O
2
aqueous solution. The conditions were those optimized and
previously reported by us for the oxidation of p-xylene to
1
terephthalic acid, except that the catalyst concentrations were
2,3
lower than those used previously. This was to ensure that the
6
catalyst was unsaturated. A 7.7 mM catalyst solution was fed
6
-
1
We investigate such synergies in our second paper which reports
at 4 mL min , thus giving a catalyst concentration of 2.6 mM
in the reactor. Concentration of H in the feed solution was
% vol. and the flowrate was 8 mL min . The organic flowrate
how the reaction can be improved dramatically by finding a more
effective catalyst, thereby bringing the possibility of a greener
process for oxidizing pX one step closer to realisation.
2
O
2
-
1
2
-
1
was 0.06 mL min , that is 0.5% w/w in the reactor. At the
end of the reactor, the liquor was quenched with a solution of
1
M NaOH, to neutralize CO
2
in the form of carbonate and
Acknowledgements
keep it in solution. NaOH also prevents precipitation of TA in
-
1
We thank EPRSC, the University of Nottingham and INVISTA
for financial support. We thank Graham Aird, Stuart Coote, Lu-
cinda Dudd, Annette Matthews, Ian Pearson, John Runnacles,
Peter Fields, Richard Wilson and Mark Guyler for their help.
pX oxidations. The quench solution flowrate was 3.5 mL min .
All the chemicals were purchased from Aldrich Ltd and used
without further purification.
CAUTION: This type of oxidation is potentially extremely
hazardous, and must be approached with care and a thorough
safety assessment must be made. The apparatus is regularly
hydrostatically pressure tested at room temperature. For safety
reasons a start-up procedure must be followed: (1) water is
pumped at 230 bar and room temperature at the desired
flowrates; (2) the heaters are turned on; (3) once the operating
temperature had been reached, the reactants are pumped.
Several samples were collected for sequential periods of 5 min
and analysed. Three samples taken when the system reaches
steady state were analyzed and the results were averaged.
Analysis of the products was done by HPLC. A Waters Xterra
References
1
P. A. Hamley, T. Ilkenhans, J. M. Webster, E. Garc ´ı a-Verdugo, E.
Vernardou, M. J. Clarke, R. Auerbach, W. B. Thomas, K. Whiston
and M. Poliakoff, Green Chem., 2002, 4, 235–238.
2
J. Fraga-Dubreuil, E. Garc ´ı a-Verdugo, P. A. Hamley, E. M. Vaquero,
L. M. Dudd, I. Pearson, D. Housley, W. Partenheimer, W. B. Thomas,
K. Whiston and M. Poliakoff, Green Chem., 2007, 9, 1238–1245.
3 E. Garc ´ı a-Verdugo, E. Vernardou, W. B. Thomas, K. Whiston, W.
Partenheimer, P. A. Hamley and M. Poliakoff, Adv. Synth. Catal.,
2
004, 346, 307–316.
4
E. Garcia-Verdugo, J. Fraga-Dubreuil, P. A. Hamley, W. B. Thomas,
K. Whiston and M. Poliakoff, Green Chem., 2005, 7, 294–300.
◦
reverse phase C18 column, maintained at 37 C, was used (flow
5 J. Fraga-Dubreuil, E. Garc ´ı a-Verdugo, P. A. Hamley, E. P e´ rez,
I. Pearson, W. B. Thomas, D. Housley, W. Partenheimer and M.
Poliakoff, Adv. Synth. Catal., 2009, 351, 1866–1876.
-
1
rate 0.7 mL min , run time 15 min; UV detection at 230 nm).
Solvents acetonitrile (ACN) and CH CO Na/CH CO H buffer
3
2
3
2
6
7
E. P e´ rez, J. Fraga-Dubreuil, L. Dudd, E. Garcia-Verdugo, P. A.
Hamley, M. L. Thomas, C. Yan, W. B. Thomas, D. Housley, W.
Partenheimer and M. Poliakoff, Selective aerobic oxidation of para-
xylene in sub- and supercritical water. Part 2. The discovery of better
catalysts, Green Chem., DOI: 10.1039/c1gc15138j.
were used. The method was as follows: isocratic method (16.7%
ACN) for the first 4 min; 4–8 min, gradient method (16.7 to 40%
ACN); back to isocratic method (16.7% ACN) for the last 7 min.
The analysis for phenols and benzene was not carried out
routinely because they are present only in very small amounts.
H. Weing a¨ rtner and E. U. Franck, Angew. Chem., Int. Ed., 2005, 44,
2672–2692.
This journal is © The Royal Society of Chemistry 2011
Green Chem., 2011, 13, 2389–2396 | 2395

View Online
8
9
A. Kruse and E. Dinjus, J. Supercrit. Fluids, 2007, 39, 362–380.
J. Fraga-Dubreuil and M. Poliakoff, Pure Appl. Chem., 2006, 78,
27 W. F. Brill, Ind. Eng. Chem., 1960, 52, 837–840.
28 http://webbook.nist.gov/chemistry/.
1
971–1982.
29 W. Partenheimer and J. A. Kaduk, Proceedings of the 4th Inter-
national Symposium on Activation of Dioxygen and Homogeneous
Catalytic Oxidation, New York, 1991.
30 W. Partenheimer, J. Mol. Catal. A: Chem., 2003, 206, 105–119.
31 G. V. Buxton, C. L. Greenstock, W. P. Helman and A. B. Ross, J. Phys.
Chem. Ref. Data, 1988, 17, 513–886.
32 R. A. Sheldon and J. K. Kochi, Metal Catalyzed Oxidations of
Organic Compounds, Academic Press, New York, p. 330.
33 J. Feng, S. N. V. K. Aki, J. E. Chateauneuf and J. F. Brennecke, J. Am.
Chem. Soc., 2002, 124, 6304–6311.
34 J. Feng, S. N. V. K. Aki, J. E. Chateauneuf and J. F. Brennecke,
J. Phys. Chem. A, 2003, 107, 11043–11048.
1
1
0 P. E. Savage, Chem. Rev., 1999, 99, 603–621.
1 D. Br o¨ ll, C. Kaul, A. Kr a¨ mer, P. Krammer, T. Richter, M. Jung, H.
Vogel and P. Zehner, Angew. Chem., Int. Ed., 1999, 38, 2998–3014.
2 M. D. Bermejo and M. J. Cocero, AIChE J., 2006, 52, 3933–3951.
3 R. L. Holliday, B. Y. M. Jong and J. W. Kolis, J. Supercrit. Fluids,
1
1
1
998, 12, 255–260.
4 P. E. Savage, J. B. Dunn and J. Yu, Combust. Sci. Technol., 2006, 178,
43–365.
5 J. B. Dunn and P. E. Savage, Environ. Sci. Technol., 2005, 39, 5427–
436.
6 J. B. Dunn and P. E. Savage, Ind. Eng. Chem. Res., 2002, 41, 4460–
465.
7 J. B. Dunn, D. I. Urquhart and P. E. Savage, Adv. Synth. Catal., 2002,
44, 385–394.
8 Y.-L. Kim, S. J. Chung, J.-D. Kim, J. S. Lim, Y-W. Lee and S.-C. Yi,
React. Kinet. Catal. Lett., 2002, 77, 35–43.
9 Y.-L. Kim, J.-D. Kim, J. S. Lim, Y.-W. Lee and S.-C. Yi, Ind. Eng.
Chem. Res., 2002, 41, 5576–5583.
1
1
1
1
1
1
4
5
4
35 K. Weissermel and H. J. Arpe, Industrial Organic Chemistry, Verlag
Chemie, Weinheim, New York, 3rd edn, 1997, p. 61.
36 Handbook of Chemistry and Physics, ed. R. C. Weast, The Chemical
Rubber Company, 67th edn, p. D-151.
3
37 J. B. Dunn, M. L. Burns, S. E. Hunter and P. E. Savage, J. Supercrit.
Fluids, 2003, 27, 263–274.
38 L. N. Mander and C. M. Williams, Tetrahedron, 2003, 59, 1105–
1136.
2
2
2
0 M. Osada and P. E. Savage, AIChE J., 2009, 55, 710–716.
1 J. B. Dunn and P. E. Savage, Green Chem., 2003, 5, 649–655.
2 W. Partenheimer and M. Poliakoff in Handbook of Green Chemistry –
Green Catalysis: Homogeneous Catalysis, ed. P. T. Anastas and
R. H. Crabtree, Wiley-VCH, Weinhein, 2009, vol 1, ch. 12, pp. 375–
39 S. S. Lande and J. K. Kochi, J. Am. Chem. Soc., 1968, 90, 5196–5207.
40 J. M. Anderson and J. K. Kochi, J. Am. Chem. Soc., 1970, 92, 2450–
2460.
41 J. D. Bacha and J. K. Kochi, J. Org. Chem., 1968, 33, 83–93.
42 J. M. Anderson and J. K. Kochi, J. Org. Chem., 1970, 35, 986–989.
43 R. A. Sheldon and J. K. Kochi, J. Am. Chem. Soc., 1968, 90, 6688–
6698.
3
97.
2
3 Process Economic Programme Report 9F, Terephthalic Acid,
September 2005, SRI Consulting, Menlo Park, California, 94025.
4 W. Partenheimer, Catal. Today, 1995, 23, 69–158.
5 Y. Kamiya and T. Ushiba, Bull. Chem. Soc. Jpn., 1975, 48, 1563–1566.
6 S. H. Jhung, K. H. Lee and Y. S. Park, Bull. Korean Chem. Soc., 2002,
2
2
2
44 W. Partenheimer, Adv. Synth. Catal., 2005, 347, 580–590.
45 X. D. Xu, C. P. De Almeida and M. J. Antal, Ind. Eng. Chem. Res.,
1991, 30, 1478–1485.
2
3, 59–64.
46 T. Richter and H. Vogel, Chem. Eng. Technol., 2001, 24, 340–343.
2
396 | Green Chem., 2011, 13, 2389–2396
This journal is © The Royal Society of Chemistry 2011