Selectivity problems in the catalytic deoxygenation of benzoic acid

Article abstract of DOI:10.1006/jcat.1997.1665

The oxides of Mg, Pb, Cr, La, Pr, Yb, Mn, Fe, Co, Zn, Zr, and Ce were studied as catalysts for the reactions of benzoic acid in the temperature range 523-723 K. All experiments were performed in a continuous flow apparatus, at a total pressure of about one bar, with hydrogen as the reductant. The oxides studied could be subdivided into several groups according to their product pattern. The groups indicate the prevailing mechanism. The two basic mechanisms are: (i) Mars and van Krevelen (selective to benzaldehyde) and (ii) radical-like decomposition of benzoate. The highest yield of benzaldehyde can be obtained with oxides of high metal-oxygen bond strength.

Full text of DOI:10.1006/jcat.1997.1665

JOURNAL OF CATALYSIS 169, 13–21 (1997)
ARTICLE NO. CA971665
Selectivity Problems in the Catalytic Deoxygenation of Benzoic Acid
�
,1
Y. Sakata, C. A. van Tol-Koutstaal,† and V. Ponec‡
�
Faculty of Engineering, Yamaguchi University, Tokawadai UBE, Yamaguchi 755, Japan; †DSM Research BV, P.O. Box 18, 6160 MD Geleen, The
Netherlands; and ‡Leiden Institute of Chemistry, Gorlaeus Laboratories, Leiden University, P.O. Box 9502, 2300 RA Leiden, The Netherlands
Received July 30, 1996; revised November 5, 1996; accepted February 10, 1997
A glance at the patent literature (1–11) would reveal that
The oxides of Mg, Pb, Cr, La, Pr, Yb, Mn, Fe, Co, Zn, Zr, and a large portion of the elements of the periodic table has
Ce were studied as catalysts for the reactions of benzoic acid in the already been suggested for use in their oxidic form as cata-
temperature range 523–723 K. All experiments were performed in
a continuous flow apparatus, at a total pressure of about one bar,
with hydrogen as the reductant. The oxides studied could be sub-
divided into several groups according to their product pattern. The
groups indicate the prevailing mechanism. The two basic mecha-
components, when a multicomponent catalyst is used.
nisms are: (i) Mars and van Krevelen (selective to benzaldehyde)
and (ii) radical-like decomposition of benzoate. The highest yield
lysts (or catalysts’ component) for selective deoxygenation.
Among them are alkaliearth, transition metal, and rare-
earth oxides. From the patent literature the differences be-
tween the oxides are unclear as is the role of the individual
As far as aliphatic carboxylic acids are concerned, much
more is known about the mechanism and the role of various
oxides as catalysts or components of catalysts. This infor-
mation will be used in the discussion (12, 13).
of benzaldehyde can be obtained with oxides of high metal–oxygen
bond strength.
�c 1997 Academic Press
This paper reports catalytic properties of various oxides
in the deoxygenation of benzoic acid. The main aim was to
determine whether all oxides behave uniformly or whether
different categories of oxides, each characterized by a cer-
tain, common mechanism, can be distinguished. This ap-
proach may perhaps lead to a more rational design of new
catalysts.
INTRODUCTION
Due to the increasing demand for aromatic aldehydes
used as fragrances and, for example, the synthesis of
(
pyrethroic insecticides), there is also increasing interest in
information on the selective production of aldehydes from
aromatic carboxylic acids. The direct hydrogenation of car-
boxylic acids by using suitable catalysts is a very convenient
method for synthesis of aldehydes. By this method, alde-
EXPERIMENTAL
hydes would be synthesized effectively without producing Methods and Data Collection
undesirable by-products that can pollute the environment.
All experiments reported in this paper were performed
Therefore, much attention has been paid to exploring new
catalysts with high activities and a selectivity to aldehydes,
suitable for industrial processes. These are usually oxidic
catalysts (1–11).
with a glass continuous-flow apparatus. The schematic view
of the apparatus is shown in Fig. 1. The whole apparatus is
kept at a temperature higher than 473 K by heating tapes
and infrared irradiation. Helium, when applied as the car-
rier gas, and H2 (as a reactant) enter the apparatus through
valves A and B, respectively. Gases are purified in the BTS
catalyst column (D) and the molecular sieve column (E)
and then passed through a vessel containing liquid benzoic
acid. The saturator temperature is maintained at around
A long list of available patents (not all are quoted here)
documents the growing interest in practical applications,
but an inspection of scientific and patent literature reveals
that only very little is known about the fundamentals of the
reactions and the function of the catalysts. It is, for example,
still impossible to predict which catalysts should be suitable
for which substrate (aliphatic vs aromatic acid; benzoic vs
substituted acids) and why. The mechanism and the factors
determining the selectivity, etc., have yet to be discovered.
This study will concentrate mainly on the last of the above-
mentioned areas.
4
50 K. The partial pressure of benzoic acid, fed into the
reactor, is controlled by a condenser (G) which is kept at
a desired (lower) temperature. The reactor H is typically
filled with ca 0.3 g of catalyst. The unreacted reactants and
products are condensed in a trap tube (K) at liquid nitrogen
temperature and can be used for analysis.
The product mixture isanalyzed bya gasliquid chromato-
graph (GLC). The sample of the product mixture (ca 5 ml) is
1
To whom all correspondence should be addressed.
13
0021-9517/97 $25.00
Copyright �c 1997 by Academic Press
All rights of reproduction in any form reserved.

�
1
4
SAKATA, VAN TOL-KOUTSTAAL, AND PONEC
2
nesium hydroxide at 815 K. La2O3 (surface area 3 m /g) was
prepared by calcining La(NO3)3 (Merck) in air at 975 K
2
for 10 h. �-Mn3O4 (surface area of �Mn3O4 was 26 m /g),
2
2
MnO2, ZrO2 (26 m /g), and CeO2 (50 m /g) were pre-
pared by calcining the corresponding hydroxides, obtained
by precipitation of Mn(NO3)2, ZrO(NO3)2 (Aldrich), and
Ce(NO3)3 (Merck) with NH4OH at pH = 9 and calcination
in air at 400 K for �-Mn3O4 and 773 K for the other ox-
2
2
ides for 10 h. PbO (Aldrich) (1 m /g), Fe2O3 (7 m /g sur-
2
face area), Pr6O11, Yb2O3 (Fluka), Cr2O3 (3 m /g), CO3O4
10 m /g), and ZnO (5 m /g) (British Drug House) were
FIG. 1. Schematic view of the flow apparatus used in this work: A, He; purchased. ZnMn O (23 m /g) was prepared by calcining
2
2
(
2
2
4
B, H2; C, mass flow controller; D, a column with BTS catalyst; E, a column
with a molecular sieve; F, saturator with benzoic acid; G, condenser for
benzoic acid; H, reactor; K, second condenser; L, flow meter at the exit of
the second condenser; M, infrared lamp for heating; S, the position of the
sampling port.
the coprecipitate of corresponding hydroxides, obtained by
the same procedure as, for example, Mn3O4, at 400 K. The
oxides were kept in situ at 523 or 723 K, under He flow,
for 1 h before the reaction was initiated. The crystallo-
graphic structure and phase purity was measured by XRD
for all oxides, where this was relevant (ZrO2, Mn-oxides,
Co3O4, CeO2). Values of the surface area are shown above
to indicate, rather the dispersion of oxides before the re-
action than the number of active sites under the running
reaction.
taken as gas from the sampling port (S) by a special syringe
that is kept at 473 K during sample collection and injection
into the gas chromatograph. The sample of the condensed
phase in the condenser (K) can be analyzed by GLC as
well. A HP 5890 II GC, equipped with flame ionization de-
tector, was used for the analysis; the data were collected and
evaluated by a HP-Chem.Station. Separation of the com-
pounds in the product mixture is achieved by means of a
RESULTS
Results obtained with various oxides, i.e., the conversion
and selectivity to the main products, both as a function of
temperature, are shown in Figs. 2 to 6. As we shall see later,
there are reasons for subdividing the oxides in this way.
Different oxides produce a different spectrum of products,
and this gives some indication of the prevailing mechanism
and the factors controlling the selectivity. In addition to the
benzaldehyde production, the production of toluene and
benzene (the ratio of these two) is particularly important.
2
5 m wide-bore CP-Sil 13 column (Chrompack, NL).
In a standard experiment, benzoicacid wasadded at 35Pa
and a flow rate of 28 ml/min; pure hydrogen was added to
give a total pressure of one bar.
The conversion � is defined as
�
A
� ABA
BA
�
=
.
[1]
�
A
BA
The selectivity to a component “i” is calculated in terms of
the carbon atom balance using the following equation:
Group 1
The results obtained with Group-1 oxides are shown in
Fig. 2.
Ai
Si = P
.
[2]
Ai � ABA
MgO. The conversion increases with increasing reaction
temperature, but is low (below 20% ) in the whole temper-
ature range up to 723 K. In the experiment represented
in Fig. 2 the reaction was first carried out below 573 K for
more than 10 h. Thereafter the reaction temperature was in-
creased to 598 K, i.e., above the melting point of magnesium
benzoate (melting of the catalyst was visible), and monitor-
ing of the reaction was continued. This suggests that MgO
probably reacted with benzoic acid to form magnesium-
benzoate, which has a low melting point.
In both these equations, A is the integrated peak intensity
of the FID corresponding to the gas mixture component,
indicated by the index, the index BA indicates benzoic acid,
and A is the signal of the feed.
In a standard experiment, the reaction was started at
�
BA
5
23 K, and the temperature was increased in steps of 25 K,
up to 723 K. The products were analyzed every 30 min, and
when the steady state of the reaction was reached at a given
temperature, the reaction temperature was increased.
The selectivity to benzaldehyde (BALD) which is the
desired product, is very low (below 35% ). It is interesting
Catalysts
Severaloxidesused in thisstudywere prepared in our lab- that at low conversions, as much as 40% phenol (8-OH)
oratory but commercially available oxides were also used. can be formed. Another important feature is that benzene
2
MgO (surface area 91 m /g) was prepared by calcining mag- is formed while the production of toluene (8-CH3) is low.

CATALYTIC DEOXYGENATION OF BENZOIC ACID
15
FIG. 2. Reaction profiles of benzoic acid under H2 over MgO, PbO, Cr2O3, and La2O3: � , conversion; ᭜, benzene; ᭡, toluene; � , benzaldehyde;
, benzylalcohol; ᭝, phenol; ᭹, benzphenon.
᭛
The rather high selectivity to acetophenone is remarkable.
Cr2O3. This oxide exhibited a very low activity in the
This product can be considered as an analogue of ketones whole temperature range. The product formation was seen
formed from aliphatic acids (e.g., acetone from acetic acid). above 548 K. At very low conversion, there was some selec-
Ketonization of aliphatic acid is known to proceed by two tivity to benzaldehyde (ca 80% at around 623 K). Again, the
different mechanisms (14–16) of which only one, not requir- production of benzene is clearly higher than that of toluene.
ing the alpha-hydrogen, would be possible here. We shall In the low temperature range, a relatively rather high for-
return to this point in the “Discussion.”
mation of products of destructive reactions, such as low
PbO. The conversion of benzoic acid reached a maximum molecular weight hydrocarbons and alcohols (not shown in
value at 673 K. This value is rather high (ca 90% ). The se- the figure), was observed (absolutely, in small amounts).
lectivity to the most desired product (BALD) was rather
La2O3. The conversion to the various products was very
high (80–90% ), too. Benzene was produced but the pro- low, below 600 K, but a value of more than 95% at 723 K
duction of toluene was very low. The selectivity to benzene was reached. As can be seen in the corresponding figure,
increased while the selectivity to BALD as well as the total the selectivity to benzaldehyde does not vary with increas-
conversion decreased above 673 K. The latter is probably ing temperatures in a monotonic way. The dip in SBALD is
caused by a partial reduction of the oxide under the running accompanied by a marked increase in benzene selectivity.
reaction.
The production of toluene was, as in all cases mentioned

1
6
SAKATA, VAN TOL-KOUTSTAAL, AND PONEC
up to now, lower than that of benzene. Pr6O11 and Yb2O3 but the selectivity of toluene increased markedly above
behaved similarly as La2O3.
673 K.
�
-Mn3O4. The behavior was similar to that of MnO2, but
Group 2
the corresponding curves shifted slightly to a lower tem-
perature. The conversion of benzoic acid showed a non-
monotonic variation with the temperature, as can be seen
in Fig. 3 at 650 K. It achieved a high value (around 95% )
at 700 K. The selectivity to benzaldehyde remained high,
being more than 90% in the reaction temperature range
from 550 to 673 K; thereafter it decreased. The selectivities
The results obtained with Group-2 oxides are shown in
Fig. 3. This group is formed by different manganese based
oxides. Two of the these oxides (�-Mn3O4 and ZnMn2O4)
are A(II)B(III)2O4 spinels, with A(II) being either Mn or
Zn.
MnO2. The conversion of benzoic acid was low below to other products were low; only the selectivity to toluene
00 K. It increased with increasing temperature up to 648 K. increased markedly above 673 K, similarly as with MnO2.
6
However, it decreased again at 673 K and achieved a very
2 4
ZnMn O . The conversion of benzoic acid increased
high value at still higher temperatures. The selectivity to monotonically with increasing reaction temperature and
benzaldehyde increased with increasing reaction temper- reached a very high value above 648 K. This tendency is
ature to around 95% , and then, it decreased as temper- slightly different from the tendency of manganes oxides
ature rose above 673 K. The selectivity to other prod- mentioned above. The selectivity to benzaldehyde was over
ucts, such as benzene and toluene, were low below 673 K, 95% between 573 K and 648 K. However, above 648 K, the
FIG. 3. Reaction profiles of benzoic acid under H2 over MnO2, �-Mn3O4, and ZnMn2O4: � , conversion; ᭜, benzene; ᭡, toluene; � , benzaldehyde.

CATALYTIC DEOXYGENATION OF BENZOIC ACID
17
deoxygenation of benzoic acid proceeded further by the reached 90% above 648 K. The selectivity to benzalde-
formation of toluene. The selectivity to toluene increased hyde was high below 623 K, where, however, the conversion
with increasing of reaction temperature, while that to ben- was low (see Fig. 4). The selectivity to toluene increased
zaldehyde decreased. The main difference in the catalytic while that to benzaldehyde decreased above 623 K. The
behavior as compared to �-Mn3O4 and MnO2 was that the selectivity to benzene increased with increasing tempera-
conversion increased monotonically with increasing in the ture. It is clear that catalytic behavior changes markedly
reaction temperature. It is noteworthy that with all the cata- at reaction temperatures above 623 K. It appears that the
lysts in this group, the production of toluene was always change is somehow related to the reduction of Fe2O3 by
clearlyhigher than that ofbenzene (in contrast to the Group the reaction mixture. Therefore, Fe2O3 was prereduced
1
catalysts, for example).
at 673 K and converted by that into a Fe3O4/400 sam-
ple (see Fig. 4). Prereduction clearly increased the conver-
sion which reached a value of more than 95% already at
Group 3
The results obtained with the Group-3 oxides are shown
in Fig. 4. The oxides are Co3O4, Fe2O3, and Fe3O4, the latter
prepared by reducing Fe2O3 with hydrogen at 673 K for 2 h.
5
98 K. The selectivity to benzaldehyde can be very high
at low conversion, but it decreases with increasing conver-
sion. The selectivity to toluene increased again as the se-
lectivity to benzaldehyde decreased. The selectivity to ben-
Fe2O3 and Fe3O4. The conversion of benzoic acid was zene increased with temperature but was lower than for
rather low at reaction temperatures below 623 K, but toluene.
FIG. 4. Reaction profiles of benzoic acid under H2 over Fe2O3, Fe3O4, and CO3O4: � , conversion; ᭜, benzene; ᭡, toluene; � , benzaldehyde; F,
methane.

1
8
SAKATA, VAN TOL-KOUTSTAAL, AND PONEC
FIG. 5. Reaction profiles of benzoic acid under H2 over ZnO and ZrO2: � , conversion; ᭜, benzene; ᭡, toluene; � , benzaldehyde; ᭛, benzylalcohol.
Co3O4. The conversion is low below 548 K, but reached 648 K. The selectivity to benzaldehyde decreased gradu-
a value of more than 90% at 598 K. The selectivity to ben- ally with increasing reaction temperature, while the selec-
zaldehyde is ca 90% , but only at 548 K when the conversion tivity to other products, such as benzyl alcohol, toluene, and
is still very low. The selectivity to methane is very high at benzene, gradually increased at temperatures above 648 K.
reaction temperatures over 573 K (in this temperature re- Selectivity to toluene increased at 723 K.
gion one observes also decomposition to methane of the
ZrO2. The catalytic behaviour was nearly the same as that
adsorbed benzoic acid), while the selectivity to other prod- of ZnO (see Fig. 5). This oxide was a very good catalyst, as
ucts is low. However, formation of toluene (as a possible also reported in the patent literature (18). The conversion
intermediate product) is clearly observable.
above 90% can be achieved with selectivity to benzalde-
hyde higher than 90% at 673 K. The main by-products are
toluene and benzene and their formation increases with
increasing temperatures. The selectivity to toluene is al-
ways higher than that to benzene (note the difference with
Group-2 oxides). Let us add that ZrO2 is very similar to
TiO2 and HfO2.
CeO2. The results obtained with CeO2 are presented in
Fig. 6. The conversion could be as high as 95% at 648 K. The
selectivity to benzaldehyde was more than 90% up to 648 K.
It decreased markedly above 648 K. The main by-products
were benzene and toluene, and their selectivities increased
with increasing reaction temperature above 673 K. The se-
lectivity to benzene is higher than that to toluene, and al-
though CeO2 is otherwise very similar to the other oxides
of Group 4, it is different in this respect.
Group 4
The results obtained with these oxides are shown in the
Figs. 5 and 6.
ZnO. The conversion of benzoic acid reached a value
of more than 90% at 673 K. The selectivity to benzalde-
hyde was high, ca 98% , at reaction temperatures below
DISCUSSION
Before discussing the catalytic behavior of oxides as
reported in this paper, information on the reactions of
carboxylic acids from the literature will be summarised:
(
i) Aliphatic acids react in hydrogen-free or hydrogen-lean
mixtures to ketones; the existence of two mechanisms of
ketonization have been proven (14–16): one involves the
�
-hydrogen of the hydrocarbon chain, and the other can
be characterized as a radical-like decomposition of car-
boxylates. Obviously, benzoic acid cannot react by the first
FIG. 6. Reaction profiles of benzoic acid under H2 over CeO2: � , con-
version; ᭜, benzene; ᭡, toluene; � , benzaldehyde; ᭛, benzylalcohol.

CATALYTIC DEOXYGENATION OF BENZOIC ACID
19
mechanism, since for that the presence of �-hydrogen in the zaldehyde, and ketones were produced along with a small
molecule is required and the potential products of the de- amount of toluene. This indicates that the reaction with the
composition of the surface benzoate (or bulk benzoate, as in reaction mixtures simulates the thermal decomposition and
the case with magnesium) are benzene (de-carboxylation), hydrogenolysis of benzoate. Hydrogen is needed to form
phenol, benzophenone, and some secondary products. It is benzaldehyde and benzene. However, low production of
not clear whether some benzaldehyde may also be a prod- toluene and benzylalcohol shows that stepwise consecutive
uct of such a reaction network. (ii) Selective deoxygenation hydrogenation hardly takes place on these surfaces.
of aliphatic acids to the corresponding aldehydes is proba-
bly a reaction of the Mars and van Krevelen (Kr o¨ ger) type at low temperatures. It is known that phenol is formed in
17, 18). It means that an oxidic catalyst offers an oxygen the presence of water and oxygen (27–30) and that water
These may be the mechanisms by which phenol is formed
(
vacancy to the adsorbate, one of the oxygens of the (ben- is produced upon the formation of the benzoate, benzalde-
zoic) acid enters it, and subsequently one carbon–oxygen hyde, and other species.
bond is broken. Hydrogen, from the OH groups in the sur-
Different contributions by different mechanisms can be
roundings of the fragment, hydrogenates the fragment to an observed with Groups 2 and 3 oxides. These oxides can form
aldehyde. The deposited oxygen is subsequently removed nonstoichiometric oxides, and they tolerate active oxygen
by hydrogen from the reaction mixture. There are several vacancies on their surfaces. It has been established that both
factors which strongly suggest that such a mechanism is in the selective deoxygenation of nitrobenzene to nitrosoben-
operation (13, 19): the effect of hydrogen/acid ratio, the zene (22) and the selective reduction of aliphatic carboxylic
effect of the prereduction of the oxide on its activity and acids to the corresponding aldehyde can take place on
selectivity and the existence of an “optimal” metal–oxygen Fe3O4 (31). With this catalyst it has been confirmed that
bond strength, corresponding to the highest activity in the the oxygen vacancies in the surface play an important role
deoxygenation. The IR spectra of adsorbed acetic acid also in the selective reduction of aliphatic acids (13, 19, 31), and
support this idea (20). For chemically related reactions, the the overall mechanism of aliphatic acid reactions is similar
deoxygenation of nitrobenzene to nitrosobenzene (and the to mechanism (ii) described above with benzoic acid. The
reversed reaction), evidence for the Mars and van Kreve- catalytic behavior of MnO2 was nearly the same as that of
len mechanism exists, based on the results of experiments �-Mn3O4. The latter oxide showed better catalytic perfor-
with isotopically labeled molecules (21). Other results also mance at lower reaction temperatures. MnO2 must prob-
support this conclusion (22).
ably be first prereduced to become active in deoxygena-
In this paper we divide the oxide catalysts into four tion (see a similar situation in (22)). None of these oxides,
groups according to their catalytic behavior in the deoxy- however, was catalytically stable. The catalytic behaviour of
genation of benzoic acid, whereby the most principal differ- both oxideschanged dramaticallybetween 648K and 673K.
ence is between Group 1 on one side and Groups 2, 3, and 4 In this temperature region the oxides are probably overre-
on the other. The differences emerging from a comparison duced and a high concentration ofoxygen vacanciesrenders
of the results in Figs. 2 to 6 are most likely a consequence of a loss of both oxygen atoms; subsequently toluene is formed.
different contributions of various mechanisms of benzoic It is not a consecutive hydrogenation (for example, no de-
acid conversion. Within each group of oxides, the relative pendence on the flow rate, as observed with 8NO2, where
contributions by the “basic” mechanisms, seem to be simi- overreduction leads to aniline formation), but overhydro-
lar. Taking the points (i) and (ii) into account, the activity genation during one sojourn of the molecule on the surface.
and selectivity patterns, shown in Figs. 2 to 6, can be ratio-
nalized as follows.
In an attempt to stabilize the Mn oxide against overre-
duction a zinc manganese spinel was prepared and the re-
The oxides belonging to Group 1, such as MgO and the action monitored with it. As shown in Fig. 2, the conversion
lanthanide oxides, generally show basic properties. With reached a value higher than 90% with a selectivity to ben-
the basic magnesium and rare-earth oxides the prevailing zaldehyde of ca 97% . However, above 673 K, the selectivity
mechanism is the decomposition of the benzoate. Benzoate to toluene increased.
species have been observed by means of infrared spec-
troscopy(23). The layered structure (24–26) ofcarboxylates oxidation–reduction property of the active center and a suf-
makes an intermolecular transfer of groups possible. ficient production ofactivated hydrogen are the two keyfac-
In a previous paper (31), it was established that an
To support this we monitored the decomposition of ben- torsin the selective reduction ofaliphaticcarboxylicacidsto
zoic acid, preadsorbed on La2O3, in pure H2 atmosphere. the corresponding aldehydes. It has also been reported that
Decomposition of the benzoate started at 598 K, the max- highly reduced Fe2O3, which exposes the surface of Fe3O4
imum amount of products was obtained at 648 K and al- mixed with Fe metal (31) or a platinized Fe3O4 (19), shows
most all preadsorbed species decomposed or desorbed at a good catalytic performance in the selective reduction of
7
23 K. The distribution of desorbed products was similar acetic acid to acetaldehyde. In this study we used the same
to that observed with the reaction mixture: benzene, ben- Fe3O4 catalyst for the reduction of benzoic acid. However,

2
0
SAKATA, VAN TOL-KOUTSTAAL, AND PONEC
at higher temperatures, the production of toluene was high, ducibility of the oxide, is responsible for the beneficial cata-
as with the reduced manganese oxides. lytic properties in the reaction. However, it is observed that
Obviously, the steady state degree of cobalt oxide reduc- the selectivity to both benzene and toluene increases with
tion is too high (most likley, metallic cobalt is present at increasing temperature, whereby the selectivity to benzene
higher temperatures). The catalyst contains then too many is higher than that to toluene. It seems that, in the behavior
sites on which oxygen can be deposited and the metallic of CeO2, the properties of the Group 4-oxides are combined
cobalt is a good catalyst for hydrogenolysis up to methane. with those of basic oxides (Group 1). The reaction may
The reducibility of transition metal oxides, discussed up proceed by the prevailing mechanism of type (i) at lower
to this point, follows the order Mn < Fe < Co, which is the reaction temperatures and both mechanisms (i.e., types (i)
order of decreasing selectivity to benzaldehyde, a molecule and (ii)) operate simultaneously at higher temperatures. A
which is formed by a selective loss of only one oxygen atom number of other oxides, not mentioned explicitly up to now,
of the benzoic acid. Because of the very poor catalytic also show a behaviour which can be considered as arising
performance of cobalt oxide in the desired reaction, this from a combination of the mechanisms (i) and (ii) above.
oxide was not studied any further. However, it can be
remarked that methane can be also formed from CO ex ides of Group 1 manifest mainly their basic properties;
benzaldehyde, etc. an easy formation of benzoate and subsequent reactions
At this point we can make the following conclusion. Ox-
Good catalytic performance and a highly selective reduc- to this formation. Group-2 oxides, the various oxides of
tion of benzoic acid to benzaldehyde was observed over manganese, show the (expected) redox properties. A high
ZrO2 and ZnO, which represent Group 4 (our number- conversion and a high selectivity to benzaldehyde can be
ing) oxides. A previous paper reported the catalytic per- achieved at rather low temperature. Oxides (Group 3)
formance of TiO2, ZrO2, and HfO2, with results similar to with redox properties, but a higher reducibility than man-
those reported here for ZrO2. Although these oxides are ganese oxide (Fe, Co), remove nonselectively both oxy-
relatively stable and have a relative high M–O bond en- gen atoms from the benzoic acid. Redox oxides (Group 4)
ergy, they all (TiO2, ZrO2, HfO2) can tolerate some oxygen with a high metal–oxygen bond strength (low reducibility)
vacancies in their surface (32). However, the concentration show a higher selectivity. The selectivity to benzaldehyde
of the vacancies remains low, these oxides are much less at higher conversions follows the order: Hf > Zr > Ti (33)>
reducible than the Mn, Fe, and Co oxides, which obviously Mn > Fe > Co.
suppresses toluene formation (33). In this study only ZrO2
was studied as a catalyst representative of the Group 4 ele-
ments, but the results are in good agreement with those of
Ref. (33).
CONCLUSIONS
The oxides can be divided into several categories, each
With the TiO2, ZrO2, and HfO2 oxides, the activity varied being characterized by the product pattern and/or prevail-
as Ti > Zr > Hf, while the selectivity followed the reversed ing mechanism. The categories are as follow:
order Hf > Zr > Ti (33). The order of the metal–oxygen
(
1) Group 1, for which the radical-like mechanism clearly
bond strength is also Hf > Zr > Ti. The main by-product
with these oxidesistoluene, and itsproduction could be sup-
pressed by water in the feed. The activity/selectivity pattern
and the effect of water strongly suggest that, when the sur-
face concentration of oxygen vacancies increases, toluene
formation (both oxygens are lost) increases, too, while that
of benzaldehyde decreases (only one oxygen is removed in
the latter reaction). When the vacancies are blocked by oxy-
gen from water, the selectivity to benzaldehyde increases.
Zinc oxide is very similar to this group oxides, although zinc
oxide is not a transition metal oxide.
prevails. Phenol is among the products (mainly those
formed at low temperature), and benzene formation is
higher than that of toluene.
(
2) Groups 2, 3, and 4 oxides all have some redox proper-
ties, and the selectivity of deoxygenation to benzaldehyde
is controlled by the concentration of oxygen vacancies. In
other words, the selectivity is determined by the degree of
reduction of the oxide in the steady state of the catalytic
reactions. Group 4 represents the best catalysts.
(
3) CeO2 also shows a very good catalytic performance
in the selective deoxygenation of benzoic acid, but both
the redox and the radical-like mechanisms operate here
simultaneously.
Cerium dioxide is slightly different from other oxides of
Group 4. Cerium dioxide is, with its chemical properties,
close to the series of lanthanide oxides, but as well as basic
properties it also shows oxidation–reduction properties
(
4) Oxides of Group 4 are most promising as catalysts for
selective deoxygenation of benzoic acid.
(
34–37). In particular, a nonstoichiometric oxide lattice
can easily be formed under a reducing atmosphere, and
the average valency of the oxide changes by changing the
atmospheric condition. A rather high metal–oxygen bond
ACKNOWLEDGMENTS
Y. Sakata is grateful to the Netherlands Organization for International
strength which is, nevertheless, accompanied bya certain re- Cooperation in Higher Education (NUFFIC) for their support during his

CATALYTIC DEOXYGENATION OF BENZOIC ACID
21
stay in The Netherlands. The financial support by SON/NWO and DSM
21. Grootendorst, E. J., Verbeek, Y., and Ponec, V., J. Catal. 157, 706
(
The Netherlands) was highly appreciated.
(1995).
2
2
2. Maltha, A., Favre, T. L. F., Kist, H. F., Zuur, A. P., and Ponec, V.,
J. Catal. 149, 364 (1994).
3. van Tol-Koutstaal, C. A., Dissertation thesis, Leiden University, 1995
REFERENCES
(
unpublished).
1
2
3
4
5
6
. Drossbach, O., and Johansson, A., U.S. Patent 2,018,350 (1935).
. Wada, A., Japan Kokai Patent 75-111,034 (1975).
. Strojny, E. J., U.S. Patent 4,328,373 (1982).
. Maki, T., Japan Kokai Patent 85-126,242 (1985).
. Gelbein, A. P., and Hansen, R., U.S. Patent 4,585,899 (1986).
24. van Niekerk, J. N., and Schoening, F. R. L., Acta Crystal. 6, 227 (1953).
25. Agterberg, F. P. W., Driessen, W. L., Reedijk, J., Oevering, H., and
Buijs, W., in “New Developments in Selective Oxidation II” (V.
Cort e´ s-Corber a´ n and S. Vic Bell o´ n, Eds.), p. 639, Elsevier, Amster-
dam, 1994.
. Holy, N. L., Gelbein, A. P., and Hansen, R., U.S. Patent 4,585,900 26. Campbell, G. C., Reibenspies, J. H., and Haw, J. F., Inorg. Chem. 30,
1986). 171 (1991).
. van Geem, P. C., and Janssen, L. H. W., Europ. Patent 02 900 96 B1 27. Kaeding, W. W., Lindblom, R. O., and Temple, R. G., Ind. Eng. Chem.
(
7
(
1991); John, C. S., Europ. Patient 02 178 718 (1986).
53, 805 (1961).
8
9
. Maki, T., and Yokoyama, T., Europ. Patent 01 509 61 B1 (1986).
. Wambach, L., Irgang, M., and Fischer, M., Europ. Patent 03 048 53 A2 29. Kaeding, W. W., J. Org. Chem. 26, 3144 (1961).
28. Kaeding, W. W., and Shulgin, A. T., J. Org. Chem. 27, 3551 (1962).
(
1989).
30. Miki, J., Asamima, M., Tachibana, Y., and Shikada, T., J. Catal. 151,
323 (1995).
1
1
1
0. Hargis, D. C., U.S. Patent 4,950,799 (1990).
1. Joentgen, W., and Fiege, H., Europ. Patent 04 140 65 A2 (1991).
2. Pestman, R., Duijne, A., Pieterse, J. A. Z., and Ponec, V., J. Molec.
Catal. 103, 175 (1991).
31. Grootendorst, E. J., Pestman, R., Koster, R. M., and Ponec, V., J. Catal.
148, 261 (1994).
32. Henrich, V. E., and Cox, P. A., in “The Surface Science of Metal Ox-
ides,” Cambridge Univ. Press, Cambridge, 1993.
33. Koutstaal, C. A., and Ponec, V., in “Science and Technology in Catal-
ysis, Proc. Int.-Conf. Catal., Tokyo 1994,” p. 105, Kodansha, Tokyo;
Elsevier, Amsterdam, 1995.
34. Fierro, J. L. G., Soria, J., Sanz, J., and Rojo, J. M., J. Solid State Chem.
66, 154 (1987).
35. Laachir, A., Perrichon, V., Bari, A., Lamotte, J., Catherine, E.,
Lavalley, J. C., Fallah, J. E., Hilaire, L., le Nomand, F., Quemere, E.,
Sauvior, G. N., and Touret, O., J. Chem. Soc. Faraday Trans. 87, 1601
(1991).
36. Bernal, S., Calvino, J. J., Cifredo, G. A., Catica, J. M., Perez Omit, J. A.,
and Pintado, J. M., J. Chem. Soc. Faraday Trans. 89, 3499 (1993).
37. Perrichon, V., Laachir, A., Bergeret, G., Frety, R., Tournayam, L., and
Touret, O., J. Chem. Soc. Faraday Trans. 90, 773 (1994).
1
1
1
1
3. Pestman, R., Koster, R. M., and Ponec, V., Recl. Trav. Chim. Pays-Bas
1
13, 426 (1994).
4. Sugiyama, S., Sato, K., Yamasaki, S., Kawashiro, K., and Hayashi, H.,
Catal. Lett. 14, 127 (1992).
5. Yakerson, V. I., Fedorovskaya, E. A., Klyachko-Gurvich, A. L., and
Rubinstein, A. M., Kinet. Katal. 2, 907 (1961).
6. Yakerson, V. I., Fedorovskaya, E. A., Klyachko-Gurvich, A. L., and
Rubinstein, A. M., Izv. Akad. Nauk. USSR, Otd. Khim. Nauk. 1527
(
1961).
1
1
1
7. Mars, P., and van Krevelen, D. W., Chem. Eng. Sci. 3, 41 (1954).
8. Kr o¨ ger, C., Z. Anorg. Alg. Chemie 206, 289 (1932).
9. Pestman, R., Koster, R. M., Pieterse, J. A. Z., and Ponec, V., J. Catal.,
to appear.
2
0. Pei, Z.-F., and Ponec, V., Appl. Surf. Sci. 103, 171 (1996).