Role of cobalt molecular sieves in the liquid-phase oxidation of cyclohexane to adipic acid

Article abstract of DOI:10.1039/a801245h

The catalytic activity of CoAPO-5 in the oxidation of cyclohexane to adipic acid has been re-examined in order to specify the actual role of the solid. It was shown that the catalytic activity in acetic acid was ascribable to the leaching of cobalt. The activity of CoAPO-5 was comparable to that of cobalt(II) acetate at very low concentration. Surprisingly, a maximum of activity was found at 0.17 mmol dm^-3, which corresponds to the concentration of cobalt leached from the solid in the reaction. Such a concentration is far below the concentration usually used for homogeneous catalysis. Variations in adipic acid selectivity, which followed the variations in activity, were interpreted by changes in mechanism.

Full text of DOI:10.1039/a801245h

View Article Online / Journal Homepage / Table of Contents for this issue
Role of cobalt molecular sieves in the liquid-phase oxidation of
cyclohexane to adipic acid
Ines Belkhir,a Alain Germain,a¤* Francois Fajulaa and Eric Facheb
a L aboratoire de Materiaux Catalytiques et Catalyse en Chimie Organique, UMR-CNRS 5618,
ENSCM, 8, Rue de lÏEcole Normale, 34296 Montpellier Cedex 5, France
b Rhone-Poulenc Industrialisation, CRIT -Carrieres, 85, Avenue de Freres Perret, BP 62, 69192
Saint-Fons Cedex, France
The catalytic activity of CoAPO-5 in the oxidation of cyclohexane to adipic acid has been re-examined in order to specify the
actual role of the solid. It was shown that the catalytic activity in acetic acid was ascribable to the leaching of cobalt. The activity
of CoAPO-5 was comparable to that of cobalt(II) acetate at very low concentration. Surprisingly, a maximum of activity was
found at 0.17 mmol dm~3, which corresponds to the concentration of cobalt leached from the solid in the reaction. Such a
concentration is far below the concentration usually used for homogeneous catalysis. Variations in adipic acid selectivity, which
followed the variations in activity, were interpreted by changes in mechanism.
The selective catalytic oxidation by air is the most important
and economical process for the conversion of hydrocarbon
feedstocks to bulk or Ðne chemicals. When the products are
not volatile enough, for example carboxylic acids, the reaction
must be carried out in liquid phase. Under such reaction con-
ditions, homogeneous catalysis using soluble transition metal
salts is the only technology which had actually been devel-
oped until now.1 The direct oxidation of cyclohexane to
adipic acid is a typical example: homogeneous catalysis by
cobalt acetate in acetic acid yields adipic acid in good selec-
tivity (88% at 21% conversion),2,3 whereas heterogeneous
catalysis, using CoAPO molecular sieves, produces adipic acid
in low selectivity.4,5 The yield of adipic acid is particularly low
in the absence of acetic acid which is a solvent of cyclohexane,
adipic acid and other products.6,7 The use of acetone as
solvent did not give satisfactory results.8 The hydrophilicity of
such a solid, which does not favour the adsorption of the sub-
strate nor the desorption of the products, could be responsible
for this failure. Previously, in order to eliminate this disadvan-
tage, we have used cobalt containing dealuminated zeolites
beta.9 However, we have demonstrated that the observed
catalytic activity was due to a partial dissolution of cobalt in
acetic acid, even if this metal was included into the framework
of the zeolite. Furthermore, Sheldon has recently determined
that the allylic oxidation of a-pinene with hydroperoxides, in
the presence of chromium substituted molecular sieves
(CrAPO and Cr-ZSM-5), was also due to leached chromium,
though a minute amount of metal was dissolved.10 These
results prompted us to reconsider the reality of the heter-
ogeneous catalysis by CoAPO-5.
CoAPO-5, a crystalline cobalt aluminophosphate, contain-
ing 3.0 wt% of cobalt, was obtained from the Laboratoire des
Materiaux Mineraux in Mulhouse (France). The solid was
pretreated by temperature-programmed heating in a nitrogen
Ñow (heating rate \ 1 K min~1) in successive 2 h steps at 473,
673 and 873 K. Then a calcination under Ñowing dry air at
873 K was carried out for 6 h.
Cobalt containing zeolites were obtained and activated as
previously described.9
Procedure
Cyclohexane oxidation was carried out in a 300 cm3 titanium,
semi-batch, mechanically stirred Parr-type reactor. A typical
procedure is described in detail for an experiment at 383 K
and 21 bar total pressure. The reaction feed consisted of cyclo-
hexane (45 cm3; 690 mmol), glacial acetic acid (68 cm3), cata-
lyst (0.5 to 3 g) and acetaldehyde (0.24 g; 5 mmol) used as
promoter. The autoclave was brought to the operating tem-
perature and pressure, then held there under a constant Ñow
of 20 dm3 h~1 of oxygen and nitrogen (10/90). The reaction
rate was determined by the instantaneous oxygen consump-
tion which was monitored by following the oxygen concentra-
tion and the Ñow rate in the output. After an initiation period
(ca. 0.5 h), the oxygen consumption became stabilized for
hours until the cyclohexane conversion reached about 25%.
Then, the amount of water produced caused demixing of the
reaction mixture which stopped the oxidation. Most of the
time the reaction was stopped at a particular time or at a
particular conversion rate, by stopping the gas Ñow and agita-
tion. The reactor was cooled and depressurized, and the
product mixture was removed.
The aim of the present work is to investigate the nature of
the catalysis involved in the oxidation of cyclohexane into
adipic acid in the presence of CoAPO-5 in acetic acid, and to
compare this catalysis to the homogeneous catalysis by
cobalt(II) acetate.
The reaction products were analysed using a Hewlett
Packard gas chromatograph equipped with a Carbowax 52
CB polar capillary column and a Ñame ionization detector
assembled with a Shimadzu programmed and computerized
Chromatopac CR6A. An aliquot of the reaction mixture (2 g)
was esteriÐed with methanol (15 cm3) in the presence of 2
drops of concentrated H SO in order to obtain the acids in
Experimental
2
4
the ester form. The reaction products consisted of adipic, glu-
taric succinic and 6-hydroxycaproic acids, cyclohexanone,
cyclohexanol and butyrolactone. In the reaction medium,
hydroxy groups are present as acetate.
Materials
Cobalt(II) acetate tetrahydrate (Aldrich), acetic acid purex and
cyclohexane for analysis (SDS) were used as received.
Recycled catalyst was obtained by Ðltration of the reaction
medium, washing of the solid by water until neutral pH, fol-
lowed by calcination at 773 K.
¤ E-mail: germain=cit.enscm.fr
J. Chem. Soc., Faraday T rans., 1998, 94(12), 1761È1764
1761

View Article Online
Filtrates of CoAPO were obtained by treating 1È2 g of
CoAPO-5 in 75 cm3 acetic acid at reÑux overnight. After cen-
trifugation, the desired amount of Ðltrate was fed into the
reaction system.
Cobalt, aluminium and phosphorus in solution were deter-
mined by atomic absorption analysis (SCA-CNRS, Solaize,
France).
cobalt from CoAPO-5 is very small, its e†ect on the conver-
sion of cyclohexane should be very limitedÏ,4 our results
showed that the catalytic activity of the CoAPO-5 catalyst
was essentially due to leached cobalt and are in agreement
with our previous work concerning the catalysis by cobalt
containing zeolites.9
According to experiment 5, very signiÐcant activity was
obtained though the amount of dissolved cobalt was very low
(0.145 mmol dm~3). Such a concentration was far below the
concentrations in cobalt(II) acetate usually required for the
observation of homogeneous catalytic activity ([2.5 mmol
dm~3).2 Thus, all happened as if the partial dissolution of
CoAPO-5 would produce cobalt species, di†erent and much
more reactive than cobalt(II) acetate, in a way “superactiveÏ.
Results and Discussion
Reaction without catalyst
In the presence of an initiator (hydroperoxide, aldehyde or
ketone) and using acetic acid as solvent, the aerial oxidation
of cyclohexane could occur without a catalyst.2 The Ðrst step
of the reaction was the formation of the cyclo-
hexylhydroperoxide which was then converted to cyclo-
hexanol and cyclohexanone. Further oxidation led to the
cleavage of the cyclohexyl ring. Under our standard condi-
tions using acetaldehyde as initiator, the conversion was low,
as well as the adipic acid selectivity (see Table 1, experiment 1).
Activity of cobalt(II) acetate at low concentrations
In order to check this assumption and quantify the possible
increase in the catalytic activity, the study of the activity of
cobalt(II) acetate at low concentrations (\1 mmol dm~3) has
been undertaken. The results are presented in Table 3. DeÐ-
nite catalytic activity was observed at these low levels of con-
centration. In addition, the activity did not change
monotonically with concentration: a sharp maximum was
observed around 0.17 mmol dm~3. Below this concentration
limit the oxidation rate was an increasing function of cobalt
concentration, and above it the reaction rate was decreasing
until reaching the level of the uncatalysed reaction. This
behaviour can be identiÐed with the so-called “catalystÈ
inhibitor conversionÏ phenomenon,11h14 already known for
some autooxidations catalysed by transition metal salts in low
polar media. With the exception of our previous work per-
formed using manganese acetate or manganese containing
zeolites,15 such a phenomenon has never been observed to
date in acetic acid.16a
Action of CoAPO-5
The addition of CoAPO-5 led to a slight increase in reaction
rate, cyclohexane conversion and adipic acid selectivity, as
emphasized in Table 1. However, it is noteworthy that the
activity proved independent of the amount of added solid
(experiments 1 and 2), as has already been pointed out in the
literature.4 Leaching of cobalt was observed together with the
dissolution of some traces of aluminium and phosphorus. The
dissolved cobalt concentrations are given in Table 1. Never-
theless, a recycled catalyst preserved 90% of its initial activity
(experiment 3). It can also be noticed that the activity is not
directly related to the amount of dissolved cobalt.
E†ect of leached cobalt
Comparison between cobalt molecular sieves and cobalt(II)
acetate catalysis
In order to determine the nature of the catalysis occuring in
this system, the CoAPO-5 catalyst was treated overnight
under reÑux in acetic acid and the Ðltrate, after centrifugation,
was fed into the reaction medium. This operation was repeat-
ed several times and the results are reported in Table 2. The
Ðltrates exhibited e†ective but variable catalytic activities,
often higher than the ones of the solid. The activity level was
not uniformly dependent on cobalt concentration. Unlike the
assertion of Lin and Weng: “Since the amount of dissolved
According to this Ðnding, the behaviour of cobalt containing
molecular sieves must be revised. For that, the activity of
CoAPO-5 and its Ðltrates were compared, in Fig. 1, to that of
cobalt(II) acetate. In order to generalize the comparison, the
activity of Ðltrates of cobalt containing zeolites beta, which
were studied previously9 were included in this Ðgure. It is
noteworthy that the activity of all the catalytic systems was
similar to that of cobalt(II) acetate for the same concentration
Table 1 Catalytic activity of CoAPO in the oxidation of cyclohexanea
cobalt in CoAPO/ dissolved Co/ reaction rateb/ cyclohexane
adipic acid
selectivityc (%)
catalyst
mmol
mmol dm~3 mmol min~1
conversionc (%)
0 none
0
0
0.36
0.44
0.46
0.42
6.6
10.3
9.7
16
19
18
18
1 CoAPO
2 CoAPO
3 CoAPOd
0.18
0.36
0.35
0.009
0.080
0.017
9.3
a Cyclohexane: 690 mmol; acetaldehyde: 5 mmol; acetic acid: 68 cm3; N /O : 90/10; pressure: 21 bar; Ñow: 20 dm3 h~1; 110 ¡C. b Rate of
2
2
oxygen consumption determined after 2 h of reaction. c After 3 h of reaction. d Recycled catalyst.
Table 2 Catalyst activity of CoAPO Ðltrates in the oxidation of cyclohexanea
dissolved Coc/
mmol dm~3
reaction rateb/
mmol min~1
cyclohexane/
conversionc (%)
adipic acid
selectivityc (%)
catalyst
0 none
0
0.36
0.46
1.36
0.72
0.66
6.6
10.9
12.5
10.8
10.0
16
17
32
26
21
4 CoAPO
5 CoAPO
6 CoAPO
7 CoAPO
0.080
0.145
0.22
0.26
a Same conditions as Table 1. b Rate of oxygen consumption determined after 2 h of reaction. c After 3 h of reaction.
1762 J. Chem. Soc., Faraday T rans., 1998, V ol. 94

View Article Online
Table 3 Catalytic activity of cobalt acetate in the oxidation of cyclohexanea
cobalt concentration/
reaction rateb/
mmol min~1
cyclohexane
conversionc (%)
adipic acid
selectivityc (%)
mmol dm~3
0
0.36
0.72
1.50
0.60
0.52
0.46
6.6
10.0
19.7
9.5
7.7
6.5
16
24
40
25
22
23
0.10
0.17
0.27
0.53
1.06
a Same conditions as Table 1. b Rate of oxygen consumption determined after 2 h of reaction. c After 3 h of reaction.
of cobalt in the reaction medium. This observation demon-
strates that the activity is only dependent on the cobalt con-
centration in the reaction medium. Thus, no superactive
species are produced by dissolution of cobalt-containing
molecular sieves, rather the solids act as distributors of suit-
able amounts of cobalt for observable catalytic activity. So,
the unique behaviour of such cobalt solutions is related to the
unusual activity of cobalt(II) acetate at low concentration.
high concentration. Taking into account the Ðrst point
(obtained without cobalt) and the last three, a monotonic
curve (dashed line) can be drawn that overshadows the cataly-
sis and the phenomenon of catalystÈinhibitor conversion
observed at low concentration. This might explain why this
phenomenon has never been described before in the oxidation
of cyclohexane in acetic acid.
The variation in cyclohexanone and adipic acid selectivities
as a function of cobalt concentration, at a constant conversion
of cyclohexane (ca. 15%), is presented in Fig. 3. Changes in
these selectivities were observed in parallel with changes in
activity: selectivity in adipic acid followed the same variation
as activity, while cyclohexanone followed the opposite evolu-
tion. Such a variation in selectivity is remarkable because, in
the case of successive reactions and at constant conversion,
the selectivity towards intermediate and Ðnal products must
be constant. Therefore, we are led to assume a change in the
reaction mechanism has occurred, as developed below.
Global activity of cobalt(II) acetate
The variation in the activity of cobalt(II) acetate over a large
range of concentrations, covering the concentration domain
where homogeneous catalysis usually operates,2 is given in
Fig. 2. Such a representation shows the narrowness of the
domain in which catalytic activity is observable at low con-
centration, and the emergence of another type of catalysis at
Mechanistic interpretation
At low concentration (\0.17 mmol dm~3), the increase in the
autooxidation rate by traces of cobalt could be due to the
catalysis of the cyclohexylhydroperoxide decomposition into
alkoxy and alkyloxy radicals, according to a modiÐed HaberÈ
Weiss mechanism.17
ROOH ] CoII ] RO~ ] CoIII ] OH~
ROOH ] CoIII ] ROO~ ] CoII ] H`
2 ROOH ] ROO~ ] RO~ ] H O
2
Such reactions would increase the rate of free radical forma-
tion in the chain reaction, explaining the global catalysis of
the autooxidation.16b
At higher concentration (between 0.17 and 2.5 mmol dm~3),
cobalt(II) would compete with cyclohexane for the alkylperoxy
radical.
Fig. 1 Activity of cobalt(II) acetate, CoAPO-5, Ðltrates of CoAPO-5
and Ðltrates of Co-zeolites, as a function of cobalt concentration in
the reaction medium (the conditions are as given in Table 1)
Fig. 3 Adipic acid and cyclohexanone selectivities as a function of
the cobalt concentration at a cyclohexane conversion of 15% (details
at low concentration are shown in the inset)
Fig. 2 Activity of cobalt(II) acetate as a function of cobalt concentra-
tion (the conditions are as given in Table 1)
J. Chem. Soc., Faraday T rans., 1998, V ol. 94
1763

View Article Online
catalytic activity observed was due, as in the case of cobalt-
containing zeolites, to the cobalt leaching into the reaction
medium. In spite of the very low concentration of dissolved
cobalt, an efficient catalysis was observed. This was not the
result of the formation, by dissolution of the solid, of a pecu-
liar species more active than cobalt(II) acetate. In fact, cobalt
acetate itself exhibited such an exceptional catalytic behaviour
at low concentration. A maximum in the catalytic activity of
cobalt(II) acetate was observed which was explained by the
catalysis of the autooxidation, followed by a phenomenon
similar to the so-called catalystÈinhibitor conversion already
known in non-polar media. At high concentration, the cataly-
sis certainly occurred by another mechanism like the direct
oxidation of cyclohexane by cobalt(III). Taking into account
the amount of dissolved cobalt, the increase in the oxidation
activity observed in this work in the presence of the CoAPO-5
catalyst and previously with Ðltrates of cobalt containing zeo-
lites, were in perfect agreement with the catalytic activity of
cobalt(II) acetate at the same low concentration levels. Thus,
cobalt molecular sieves did not act as heterogeneous catalysts,
but as distributors of adequate amounts of cobalt in the reac-
tion medium.
Fig. 4 Adipic : glutaric acid ratio as a function of cobalt concentra-
tion at a cyclohexane conversion of 15%
ROO~ ] RH ] ROOH ] R~
ROO~ ] Co(II) ] ROOCo(III)
If the cobaltic complex was stable and inactive, a termination
step would take place instead of a propagation step. This
could justify the conversion of the catalyst into inhibi-
tor.11,12,14
At the highest concentration levels ([3 mmol dm~3), the
catalysis was operating again and activity was directly con-
nected to the cobalt concentration. A direct attack of cyclo-
hexane by CoIII was proposed.2 The rate determining step
should be a one-electron transfer from the alkane to the tri-
valent cation which should be followed by fast proton loss.15c
From the present study, it seems difficult to Ðnd a heter-
ogeneous catalyst for the aerial oxidation of cyclohexane, at
least, as long as acetic acid remains the best solvent to
produce adipic acid.
References
1
2
3
4
R. A. Sheldon and J. Dakka, Catal. T oday, 1994, 19, 215.
K. Tanaka, Hydrocarbon Proc., 1974, 53, 114.
J. Kollar, W. O. Pat., 1993, 94.07833.
S. S. Lin and H. S. Weng, Appl. Catal. A, 1993, 105, 289; Appl.
Catal. A, 1994, 118, 21; J. Chem. Eng. Jpn, 1994, 27, 211.
B. Kraushaar-Czarnetzki and W. G. M. Hoogervorst, Eur. Pat.,
1992, 519.569.
D. L. Vanoppen, D. E. de Vos, M. J. Genet, P. G. Rouxhet and
P. A. Jacobs, Angew. Chem., Int. Ed. Engl., 1995, 34, 560.
B. Kraushaar-Czarnetzki, W. G. M. Hoogervorst and W. H. J.
Stork, Stud. Surf. Sci. Catal., 1994, 84, 1869.
P. Conception, A. Corma, J. M. Lopez Nieto, J. Perez-Pariente,
Appl. Catal. A, 1996, 143, 17.
I. Belkhir, A. Germain, F. Fajula and E. Fache, Stud. Surf. Sci.
Catal., 1997, 110, 577.
RH ] CoIII] RH~` ] CoII
RH~` ] R~ ] H`
5
6
7
8
9
Such a reaction did not occur with a manganese catalyst.16
This is in agreement with the lower electrochemical potential
of the MnII/MnIII redox couple.
The study of the adipic : glutaric acid ratio conÐrmed
changes in the mechanism. It was determined that glutaric
acid was not formed appreciably by degradation of adipic
acid, but was obtained directly during the ring opening, in
competition with adipic acid.18 So, variation of this ratio
would be a reÑection of changes in mechanism. The evolution
of this ratio, as a function of cobalt concentration for a cyclo-
hexane conversion of 15%, is shown in Fig. 4. At low concen-
tration of cobalt until 0.17 mmol dm~3, the ratio was close to
6, which is the value already determined by Hendry et al. for
the autooxidation of cyclohexane.19 Above 10 mmol dm~3,
the value was constant and equal to 9, emphasizing a single
mechanism. Between these two limits, the ratio was variable:
it was decreasing when the inhibition competed with the
autooxidation, reaching the lowest value (ca. 3) for the
minimum of activity and was increasing when the attack by
cobalt(III) competed with the inhibition.
10 R. A. Sheldon, Stud. Surf. Sci. Catal., 1997, 110, 151.
11 A. T. Betts and N. Uri, Makromol. Chem., 1966, 95, 22; Adv.
Chem. Ser., 1968, 76, 160.
12 D. G. Hendry and D. Schuetzle, Am. Chem. Soc., Div. Pet. Chem.,
Prepr., 1969, 14, A31.
13 J. F. Black, J. Am. Chem. Soc., 1978, 100, 527.
14 N. M. Emanuel, Bull. Acad. Sci. USSR, Div. Chem. Sci., 1974,
1010.
15 I. Belkhir, A. Germain, F. Fajula and E. Fache, Supported
Reagents and Catalysts in Chemistry, ed. B. K. Hodnett, A. P.
Kybett, H. H. Clark and K. Smith, The Royal Society of Chem-
istry, Cambridge, 1998, p. 48.
16 R. A. Sheldon and J. K. Kochi, in Metal-catalysed oxidations of
organic compounds, Academic Press, New York, 1981, (a) p. 46;
(b) p. 38; (c) p. 137.
17 F. Haber and J. Weiss, Proc. R. Soc. L ondon A, 1934, 147, 332.
18 J. D. Druliner, J. Org. Chem., 1978, 43, 2069.
19 D. G. Hendry, C. W. Gould, D. Schuetzle, M. G. Syz and F. R.
Mayo, J. Org. Chem., 1976, 41, 1.
Conclusion
The study of the oxidation of cyclohexane in the presence of
CoAPO-5 molecular sieves as catalysts has shown that the
Paper 8/01245H; Received 12th February, 1998
1764
J. Chem. Soc., Faraday T rans., 1998, V ol. 94