A rationale of the BaeyerVilliger oxidation of cyclohexanone to ε-caprolactone with hydrogen peroxide: Unprecedented evidence for a radical mechanism controlling reactivity

Article abstract of DOI:10.1002/chem.201001777

We demonstrate, for the first time, in the BaeyerVilliger oxidation of cyclohexanone with aqueous hydrogen peroxide under conditions aimed at obtaining ε-caprolactone, that a thermally activated radical reaction leads to the concurrent formation of adipic acid, even when a stoichiometric amount of the oxidant is used. In fact, ε-caprolactone is the primary reaction product, but it is more reactive than cyclohexanone, and quickly undergoes consecutive transformations. When titanium silicalite-1 (TS-1) is used as a catalyst, the high concentration of hydroxy radicals within its pores accelerates the reaction rates, and the consecutive formation of adipic acid (and of lighter diacids as well) becomes largely kinetically preferred. The proper choice of the solvent, which also may act as a radical scavenger, both without catalyst and with TS-1, is a powerful tool for controlling the rates of the various reactions involved.

Full text of DOI:10.1002/chem.201001777

DOI: 10.1002/chem.201001777
A Rationale of the Baeyer–Villiger Oxidation of Cyclohexanone to
e-Caprolactone with Hydrogen Peroxide: Unprecedented Evidence
for a Radical Mechanism Controlling Reactivity
Fabrizio Cavani,*[a] Katerina Raabova,[a] Franca Bigi,[b] and Carla Quarantelli[b]
Abstract: We demonstrate, for the first
time, in the Baeyer–Villiger oxidation
of cyclohexanone with aqueous hydro-
gen peroxide under conditions aimed
at obtaining e-caprolactone, that a ther-
mally activated radical reaction leads
to the concurrent formation of adipic
product, but it is more reactive than cy-
clohexanone, and quickly undergoes
consecutive transformations. When ti-
tanium silicalite-1 (TS-1) is used as a
catalyst, the high concentration of hy-
droxy radicals within its pores acceler-
ates the reaction rates, and the consec-
utive formation of adipic acid (and of
lighter diacids as well) becomes largely
kinetically preferred. The proper
choice of the solvent, which also may
act as a radical scavenger, both without
catalyst and with TS-1, is a powerful
tool for controlling the rates of the var-
ious reactions involved.
Keywords: heterogeneous catalysis ·
acid, even when
a
stoichiometric
lactones
·
oxidation
·
radical
amount of the oxidant is used. In fact,
e-caprolactone is the primary reaction
reactions · titanium silicalite
Introduction
Active heterogeneous catalysts for the BV oxidation of
cycloalkanones with HP may be divided into four different
classes: a) Brønsted-type acid catalysts, such as zeolites (H-b
and USY),[4,5] alumina,[6] and polyoxometalates;[7] b) Lewis-
type acid catalysts, based on SnIV (Sn-b, Sn-hydrotalcite, Sn-
clay),[8–12,20] SbV,[13,14] or supported Pt complexes;[2,15] c) basic
oxides;[16,17] and d) catalysts based on elements that will
The Baeyer–Villiger (BV) reaction has aroused great inter-
est as a tool for the preparation of pharmaceuticals (e.g., an-
tibiotics, steroids) and compounds for the fine chemicals and
intermediates industry.[1–3] Various substrates have been con-
sidered as reagents for this reaction, the main interest being
focused on cyclic ketones and aromatic aldehydes, with or-
ganic peracids and peroxides, and hydrogen peroxide (HP)
as oxidants; HP appears to be the ideal oxidant for develop-
ing cleaner and more sustainable chemistry. Within this con-
text, in recent years many studies have been devoted to the
investigation of new heterogeneous systems that can selec-
tively transform cyclohexanone to e-caprolactone with HP
as the oxidant,[4–20] because this lactone is a monomer of
great industrial interest for the synthesis of polycaprolac-
tone, a biodegradable polyester with a low melting point.
allow the formation of Me OOH species, such as TiIV incor-
ꢀ
porated into silicalite or b zeolite.[18,19] In general, catalysts
belonging to classes a) and c) show limited preference for
production of the lactone; this could be due either to subse-
quent lactone hydrolysis, possibly catalyzed by acid sites, or
to the formation of unknown heavy compounds. An acidic
functionality is regarded as most important for BV oxida-
tion, according to the mechanism proposed by Criegee,[21]
because an acid activation of the ketone enables nucleophil-
ic oxidation by nonactivated HP.[1]
Outstanding results have been reported with class b) cata-
lysts; for example, with the highly regioselective Sn-b cata-
lyst developed by Corma and co-workers.[20,22,23] In this case,
the key reason for the high selectivity is the interaction be-
tween the carbonyl group of the reactant and the Lewis acid
sites (i.e., the SnIV atom incorporated in the zeolite lattice);
this interaction activates the reactant to the HP attack.
Herein, we demonstrate, for the first time, that the ther-
mally activated BV oxidation of cyclohexanone with HP
leads fatally to the formation of adipic acid (AA), because
of the concomitant occurrence of reactions involving radi-
[a] Prof. F. Cavani, Dr. K. Raabova
Dipartimento di Chimica Industriale e dei Materiali
ALMA MATER STUDIORUM Universitꢀ di Bologna
Viale Risorgimento 4, 40136 Bologna (Italy)
Fax : (+39)0-512-093-680
E-mail: fabrizio.cavani@unibo.it
[b] Prof. F. Bigi, Dr. C. Quarantelli
Dipartimento di Chimica Organica e Industriale dellꢁUniversitꢀ
Parco Area delle Scienze 17A, 43100 Parma (Italy)
12962
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 12962 – 12969

FULL PAPER
cals formed during HP decomposition. However, strict con-
trol of reaction conditions may enable a shift in favor of the
intermediate product, e-caprolactone. Even more remarka-
bly, strong evidence was obtained that shows that the forma-
tion of lactone first, and adipic acid later, occurs through a
radical mechanism, even when titanium silicalite (TS-1) is
used as catalyst.
Results and Discussion
The thermal, uncatalyzed oxidation of cyclohexanone with
HP: The results from the uncatalyzed, thermal oxidation
(T=908C) of cyclohexanone with HP are shown in Figure 1.
This behavior, usually overlooked in studies of the catalyzed
BV oxidation to lactones, indeed proved to be the key to un-
derstanding the reaction mechanism and, subsequently, the
role of the catalyst and reaction medium as well.
In a few hours, cyclohexanone conversion reached 40%.
e-Caprolactone was the only primary product, and was
largely predominant at low cyclohexanone conversion; how-
ever, the e-caprolactone was quite reactive. Secondary prod-
ucts observed were 6-hydroxyhexanoic acid and AA. Note
that the secondary products were not kinetically related:
this means that both compounds were formed by direct
transformation of e-caprolactone; the former possibly by hy-
drolysis, and the latter by direct oxidation. Moreover, we
observed that at these conditions the hydroxy acid was
stable, and was not extensively transformed into AA. Gluta-
ric and succinic acids were apparently formed as the final
products in the reaction (the initial level of their formation
was nil or very low at up to 1 h reaction time), which means
that these compounds were formed by subsequent reactions
on a secondary product, either 6-hydroxyhexanoic or adipic
acid. A small amount of 6-oxohexanoic acid was also
formed.
~
Figure 1. Top: Cyclohexanone conversion ( ); proportional yields of cap-
~
&
*
rolactone ( ), 6-hydroxyhexanoic acid ( ), 6-oxohexanoic acid ( ), AA
), and to glutaric+succinic acids (ꢀ), all calculated with respect to
amount of cyclohexanone converted, versus reaction time, in the absence
of catalyst. Reaction conditions: T=908C, 1 mmol cyclohexanone
(0.103 mL), 3 mmol HP (30% aqueous solution; overall volume
^
(
a
0.304 mL); solvent=water (0.100 mL). Bottom: Residual HP versus reac-
tion time for the tests shown in the top part ( ), and for tests carried out
~
without cyclohexanone (&).
Figure 1 also shows the corresponding variation in HP
concentration with time for the thermally activated oxida-
tion of cyclohexanone, and for the decomposition of HP
without cyclohexanone. The decomposition of HP in the ab-
sence of cyclohexanone was only slightly less rapid than it
was in the presence of cyclohexanone. This means that the
presence of cyclohexanone in the solution did not have addi-
tional effects on HP conversion that occurred regardless of
the presence of cyclohexanone. This supports the hypothesis
that, in the absence of catalyst (i.e., without any specific cat-
alytic activation of the ketone) the formation of lactone, and
then of AA, occurs by reaction
Reactivity tests made in the presence of radical scaveng-
ers confirmed our hypothesis (Table 1). Both cyclohexanone
conversion and the product distribution were greatly affect-
ed by the type and amount of radical scavenger added. First,
an experiment was carried out under the same conditions as
for the tests in Figure 1, but in the presence of tert-butanol.
The amount of tert-butanol added was in a molar ratio of
1:1 with cyclohexanone. We chose this compound because
1) it is soluble in water, 2) it generates tert-butyl hydroper-
oxide by reaction with HP, a hydroperoxide that is stable
and does not react with cyclohexanone in the absence of a
with the radical species formed
Table 1. Comparison of reactivity tests made without catalyst, and either with or without radical scavenger.[a]
by HP decomposition. Consid-
ering the amount of HP re-
quired to produce the oxidation
products and the HP conver-
sion, we can deduce that HP
was partially (ca. 50%) decom-
posed into water and oxygen.
Radical
scavenger
Cyclohexanone
conversion [%]
Proportional yield [mol%]
AA
e-CL[b]
6-HHA[c]
6-OHA[d]
GA+SA[e]
–
34
18
6
51
51
18
11
23
71
14
9
5
–
–
16
17
10
tert-butanol
dioxane
–
[a] Reaction time=3 h. Cyclohexanone/tert-butanol and cyclohexanone/dioxane 1:1 (molar ratios). No catalyst.
Other conditions as in Figure 1. [b] e-CL = e-caprolactone. [c] 6-HHA: 6-hydroxyhexanoic acid. [d] 6-OHA:
6-oxohexanoic acid. [e] GA+SA: glutaric acid+succinic acid.
Chem. Eur. J. 2010, 16, 12962 – 12969
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
12963

F. Cavani et al.
proper catalyst, and 3) tert-butanol may diffuse into the
TS-1 pores (see experiments with TS-1 catalyst, below).
After 3 h reaction time, the conversion of cyclohexanone
was 18% (versus 34% without tert-butanol), of which 51%
(versus 51%) was converted to AA, 23% to caprolactone
(versus 11%), and 9% to 6-hydroxyhexanoic acid (versus
14%). Thus, the distribution of products was slightly differ-
ent from that achieved in the absence of tert-butanol; how-
ever, the main effect was on the reaction rate at which cy-
clohexanone was transformed into e-caprolactone (the only
primary reaction, see Figure 1), which was definitely lower
than without tert-butanol. These results were confirmed by
using dioxane (molar ratio dioxane/cyclohexanone=1:1), an
efficient radical scavenger that is also soluble in water.[24] In
this case, the effect was even more pronounced, because the
conversion of cyclohexanone was only 6%, but with a high
proportional yield (71%) of lactone, and a low yield (18%)
of AA. Furthermore, when a large excess of dioxane was
used (molar ratio dioxane/cyclohexanone 34:1, cyclohexa-
none/HP 1.5:1), the conversion was nil. In conclusion, these
results confirm the hypothesis that the radical species
formed by HP decomposition is involved in the noncata-
lyzed transformation of cyclohexanone into lactone and of
the latter into AA.
ment in a nitrogen atmosphere. Cyclohexanone conversion
after 6 h reaction time and under continuous feeding of ni-
trogen was 42% (40% in air), with 58% proportional yield
of AA (55% in air).
We could also rule out the possible role of metals present
in traces in cyclohexanone for two reasons. First, the overall
content of transition metals in the cyclohexanone used was
less than 0.2 ppm (see Experimental Section for details).
Second, we carried out reactivity tests by measuring the HP
decomposition after 1 h without a catalyst at 908C, and by
using cyclohexanone of two different purities, either ꢁ99.5
or ꢁ99.9%. The same extent of HP decomposition was ob-
served in each case.
Experiments were also carried out by directly reacting e-
caprolactone or 6-hydroxyhexanoic acid, which are postulat-
ed to be reaction intermediates in the formation of AA; the
results of these experiments are reported in Table 3. The lac-
tone was much more reactive than cyclohexanone under the
Table 3. Reactivity of e-caprolactone (e-CL) and of 6-hydroxyhexanoic
acid (6-HHA) without catalyst.[a]
Reactant
Conversion
[%]
Proportional yield [mol%]
AA
6-HHA[b]
6-OHA[c]
GA+SA[d]
e-CL
78
99
13
28
56
62
44
9
–
trace
trace
4
28
35
34
e-CL[e]
6-HHA
Tests made with a cyclohexanone/HP 1:1 molar ratio (sto-
ichiometric for lactone formation) showed that even in
these conditions it is possible to achieve both a remarkable
cyclohexanone conversion (30% after 6 h reaction time)
and a high proportional yield of AA (see Table 2). The pro-
[a] Reaction time=3 h, except [e]. No catalyst. Other conditions as in
Figure 1 (e-CL or 6-HHA replacing cyclohexanone). [b] 6-HHA: 6-hy-
droxyhexanoic acid. [c] 6-OHA: 6-oxohexanoic acid. [d] GA+SA: gluta-
ric acid+succinic acid. [e] 6 h.
Table 2. Comparison of reactivity tests made without catalyst, at differ-
ent cyclohexanone/HP feed ratio.[a]
conditions examined, conversion being 78% after 3 h, and
almost complete after 6 h reaction time; this supports the
view that in the uncatalyzed, thermally activated oxidation
of the ketone, stopping the reaction at the formation of the
lactone intermediate is not an easy task. The main product
of lactone transformation was 6-hydroxyhexanoic acid; how-
ever, the latter was converted into AA, and to a minor
extent to lighter diacids, when the conversion of the lactone
increased. In fact, when the reaction time was increased
from 3 to 6 h, conversion to AA increased from 28 to 56%,
whereas the proportional yield of 6-hydroxyhexanoic acid
correspondingly decreased from 44 to 9%. This was con-
firmed in an experiment carried out by direct reaction of 6-
hydroxyhexanoic acid (also reported in Table 3). This com-
pound was less reactive than the lactone (conversion 13%
after 3 h reaction time); it was converted to AA (propor-
tional yield 62%) and to lighter diacids, with minor amounts
of 6-oxohexanoic acid.
Cyclohexanone/HP Cyclohexanone
molar feed ratio
Proportional yield [mol%]
conversion [%] AA e-CL 6-HHA[b] GA+SA[c]
1:3
1:1
40
30
55
43
2
7
15
37
28
13
[a] Reaction time=6 h. No catalyst. Other conditions as in Figure 1.
[b] 6-HHA: 6-hydroxyhexanoic acid. [c] GA+SA: glutaric acid+succin-
ic acid.
portion of 6-hydroxyhexanoic acid produced (37%)was
higher than that achieved with excess HP (15%), whereas
the proportion of AA produced was lower (43% versus
55%), as was that of the lighter diacids (13% versus 28%).
This means that under conditions of excess HP (cyclohexa-
none/HP 1:3, molar ratio, as seen in Figure 1), the consecu-
tive degradation into glutaric and succinic acids is kinetically
more probable. With a cyclohexanone/HP 1:1 molar ratio,
the yield with respect to the converted HP was around 75%
for an HP conversion of 87% after 6 h reaction time (yield
80% after 1 h, for an HP conversion of 38%). Therefore,
when the cyclohexanone/HP molar ratio was increased, the
majority of the radicals generated by HP transformation re-
acted with the ketone.
All of these experiments support the proposed reaction
network for the noncatalyzed, thermally activated BV reac-
tion depicted as shown in Scheme 1. It is worth noting that,
while it has been proposed in the past,[25] this is the first re-
ported experimental evidence of the reaction network con-
necting the various products formed in the oxidation of cy-
clohexanone with HP.
The role of molecular oxygen as a possible reactant for
radical autoxidation was ruled out by carrying out an experi-
12964
www.chemeurj.org
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 12962 – 12969

Baeyer–Villiger Oxidation of Cyclohexanone to e-Caprolactone
FULL PAPER
Lactone is then converted
into various products by means
of
parallel
reactions
(Scheme 1): 1) hydrolysis to 6-
hydroxyhexanoic acid, which
appears to be not very reactive
under the chosen conditions; or
2) radical reactions to generate
either AA (Scheme 3), or light-
er diacids, or, to a minor extent,
6-oxohexanoic acid.
Scheme 1. Reaction network of cyclohexanone oxidation with HP.
Our results demonstrate that, even in the absence of a
catalyst, under conditions typically used for the catalytic BV
oxidation of ketones, cyclohexanone is converted to e-capro-
lactone. Therefore, the radicals generated by HP decomposi-
tion, as shown in Equations (1) and (2) are also involved in
the cyclohexanone transformation:
Reactivity tests with TS-1 catalyst: Results of experiments
carried out with TS-1 catalyst are shown in Figure 2; condi-
tions were similar to those used for the thermally activated
reaction. In contrast to the experiments carried out without
catalyst, there was no formation of either the 6-hydroxyhex-
anoic or the 6-oxohexanoic acids, and only a very low
amount of e-caprolactone was formed. The main product
over the entire range of cyclohexanone conversion was AA,
but glutaric and succinic acids were also produced in signifi-
cant quantities. The distribution of products was not affected
by the conversion, thus supporting the hypothesis that AA
is mainly formed by a direct, parallel reaction of cyclohexa-
none, or that the intermediate e-caprolactone is very effi-
ciently transformed into AA inside the pores of the silicalite
before it diffuses out into the bulk liquid phase.
C
H2O2 ! 2 HO
ð1Þ
ð2Þ
C
C
HO þ H2O2 ! H2O þ HO2
In other words, even though an acid-type activation of the
ketone may accelerate the reaction, the carbonyl moiety in
cyclohexanone does not need this chemical activation to
react. The Criegee intermediate may indeed form either by
the classical addition of HOOH, or by a two-step reaction
C
Figure 2 also depicts the effect of reaction time on HP
conversion; the amount of unconverted HP is compared
with that achieved under the same conditions but without a
catalyst. The HP conversion was greatly enhanced in the
presence of TS-1, but the initial proportional yield of oxy-
genated compounds still remained lower than 50%. This im-
plies that Ti sites are able to activate HP and develop the
TiOOH species, but that a large part of the latter then de-
composes into radical species, and then to water and
oxygen. However, it is also shown that, even though the
conversion of cyclohexanone increased continuously over
the entire range of reaction time (0–6 h), the amount of re-
sidual HP was already very low after 1 h reaction time. At
the same time, the proportional yield of AA, calculated with
respect to the converted HP, increased over time.
with OH (Scheme 2). However, cyclohexanone activation
by hydrogen bonding with water could also occur (also
shown in Scheme 2).
Scheme 2. Left: Conventional (upper route) and nonconventional radical
(lower route) routes to e-caprolactone, via the common Criegee inter-
mediate. Right: Possible carbonyl activation by water through hydrogen
bonding.
The only possible explanation for these observations is
that the rate-determining step of the process is not the gen-
Scheme 3. Transformation of e-caprolactone into AA through a radical mechanism.
Chem. Eur. J. 2010, 16, 12962 – 12969
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
12965

F. Cavani et al.
When the reaction was carried out under the same experi-
mental conditions as for Figure 2, but in the presence of tert-
butanol (cyclohexanone/tert-butanol molar ratio 1:1), the
conversion of cyclohexanone was only slightly lower than
that observed during the same reaction time without tert-bu-
tanol (Figure 3). However, the main difference concerned
~
Figure 3. Cyclohexanone conversion ( ); proportional yield of caprolac-
&
^
tone ( ), AA ( ), and glutaric+succinic acids (ꢀ), all calculated with re-
spect to amount of cyclohexanone converted, versus reaction time. Reac-
tion conditions as in Figure 2, but with tert-butanol as cosolvent (cyclo-
hexanone/tert-butanol 1:1 molar ratio). Catalyst TS-1 (0.077 g;
0.032 mmol Ti).
~
Figure 2. Top: Cyclohexanone conversion ( ); proportional yield of cap-
&
^
rolactone ( ), AA ( ), and glutaric+succinic acids (ꢀ), all calculated
with respect to amount of cyclohexanone converted, versus reaction time.
Reaction conditions: T=908C, 1 mmol cyclohexanone (0.103 mL),
3 mmol HP (30% aqueous solution; overall volume 0.304 mL); solvent=
water (0.100 mL). Catalyst TS-1 (0.077 g; 0.032 mmol Ti). Bottom: Resid-
the distribution of products. In the presence of tert-butanol,
e-caprolactone formed in higher proportions; in other
words, its consecutive transformation into AA was much
slower than it was in the absence of tert-butanol. It is also
important to note that the HP conversion over time was ex-
actly the same for tests carried out either with or without
tert-butanol; a result that agrees with the assumption that
the radical scavenger does not have any effect on the HP
decomposition rate. On the other hand, when a large excess
of tert-butanol was used (cyclohexanone/tert-butanol 1:6),
while keeping the other reaction parameters constant, the
cyclohexanone conversion was largely reduced (8% after
6 h reaction time), and once again the proportional yield of
e-caprolactone was relatively high (34 versus 42% for the
proportional yield of AA). This means that tert-butanol has
a major effect on the rate of oxidation of the lactone into
AA, due to the higher reactivity of this compound than cy-
clohexanone, but when higher amounts of tert-butanol are
used, the inhibitory effect also extends to the less reactive
cyclohexanone. Note that the same detrimental effect was
not observed when methanol was used as solvent.
~
~
ual amount of HP, in the absence ( ) and in the presence of TS-1 ( ),
and proportional yield of diacids with respect to converted HP ( ),
^
versus reaction time.
eration of the TiOOH species, but its subsequent reaction
with cyclohexanone; in other words, the few Ti sites (the
molar ratio TiIV/cyclohexanone was equal to 0.032) act as
reservoirs for the oxidizing species, the hydroxy radicals, as
shown in Equations (3) to (7):
H2O2 þ ꢀOꢀTiIV ! ðHOÞꢀTiꢀOOH Ð ðHOÞꢀTiꢀO þ OH
ð3Þ
C
C
C
C
ðHOÞꢀTiꢀO þ H2O2 ! ðHOÞꢀTiꢀOH þ HO2
ð4Þ
ð5Þ
ð6Þ
ð7Þ
C
C
ðHOÞꢀTiꢀO þ C6H10O ! ðHOÞꢀTiꢀOH þ C6H9O
C
C
C6H10O þ OH ! C6H9O þ H2O
Furthermore, similarly to what is observed in tests carried
out without catalysts, with TS-1 an experiment conducted in
a nitrogen atmosphere gave the same result as that obtained
in the presence of air: cyclohexanone conversion was 51%
after 3 h reaction time (versus 51% in air), with 82% pro-
portional yield of AA (versus 78% in air).
ðHOÞꢀTiꢀOH þ H2O2 ! ðHOÞꢀTiꢀOOH þ H2O
Due to both the higher reactivity of e-caprolactone than
cyclohexanone, and the very high local concentration of the
OH (and O2H) radicals generated and trapped inside the
C
C
TS-1 pores, lactone is very efficiently transformed into AA
and lighter diacids.
Lastly, experiments were conducted by direct reaction of
either e-caprolactone or 6-hydroxyhexanoic acid (Table 4);
12966
www.chemeurj.org
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 12962 – 12969

Baeyer–Villiger Oxidation of Cyclohexanone to e-Caprolactone
FULL PAPER
Table 4. Reactivity of e-caprolactone (e-CL) and of 6-hydroxyhexanoic acid (6-HHA) with TS-1 catalyst.[a]
used in the tests shown in Fig-
ures 2 and 3; under these condi-
tions, we could not load all re-
actants from the very begin-
ning, because the very rapid de-
composition of HP led to the
evolution of large amounts of
oxygen, causing a sudden and
very dangerous pressure in-
crease in the reactor. Therefore,
we slowly added HP over one
Radical
scavenger
Reactant
Conversion [%],
t [h]
Proportional yield [mol%]
AA
6-HHA[b]
6-OHA[c]
GA+SA[d]
–
–
e-CL
e-CL
e-CL
6-HHA
40, 0.5
84, 1
22, 1
17, 6
93
87
59
83
3
7
27
–
2
2
–
5
2
4
14
12
tert-butanol[e]
–
[a] Reaction conditions as in Figure 2 (e-CL or 6-HHA replacing cyclohexanone). [b] 6-HHA: 6-hydroxyhexa-
noic acid. [c] 6-OHA: 6-oxohexanoic acid. [d] GA+SA: glutaric acid+succinic acid. [e] e-caprolactone/tert-bu-
tanol 1:1 (molar ratio).
these tests were aimed at confirming the high reactivity of
the lactone. The reaction conducted for 1 h, under the same
conditions as for cyclohexanone (Figure 2, with the same
molar ratio of reactant/TiIV/HP), led to 84% lactone conver-
sion, with 87% proportional yield of AA. However, when
the reaction was carried out in the presence of tert-butanol,
e-caprolactone conversion was only 22% after 1 h reaction
time, with 59% proportional yield of AA and 27% 6-hy-
droxyhexanoic acid. This means that the presence of this al-
cohol inhibited the transformation of e-caprolactone to AA,
whereas the parallel route of lactone hydrolysis to 6-hydroxy-
hexanoic acid was left substantially untouched. 6-Hydroxy-
hexanoic acid was much less reactive than e-caprolactone,
which is similar to what is observed in experiments conduct-
ed without catalysts; after 6 h reaction time, conversion was
only 17%, the main product of oxidation being AA.
Reactivity tests were carried out by using both greater
amounts of TS-1 and reactants and solvent volumes, to eval-
uate the isolated yield of AA. Figure 4 shows a plot of the
results of these experiments; the yield of the products and
the conversion of HP are plotted as a function of reaction
time for tests carried out either in water only, or in water+
tert-butanol. During these experiments, all the reactants and
catalyst weights loaded were 50 times higher than those
hour, and then continued the reaction over the desired time.
Because of this, the rates of transformation and extent of
conversion of both HP and cyclohexanone were lower than
those obtained for the same reaction times in the experi-
ments shown in Figures 2 and 3.
The yield of the final solid product (which consisted of
75% AA, and 25% glutaric and succinic acids, with only
traces of e-caprolactone; see the Experimental Section for
details of yield measurement) was 21% after 3 h reaction
time, and 32% after 6 h reaction time. The experiment was
repeated after the recovery of the spent TS-1, which was
either reloaded as such for another experiment, or regener-
ated either by simply washing with water and acetone, or by
a thermal treatment at 4508C in air for 3 h. In all cases, the
isolated yield obtained after 6 h reaction time was very simi-
lar to that obtained with the fresh catalyst (results also
shown in Figure 4).
Again, Figure 4 shows the results of experiments carried
out in the presence of tert-butanol. The yield after 3 h reac-
tion time was much lower than that obtained without tert-
butanol, but the conversion of HP did not differ in the two
cases; in fact, during tests carried out with tert-butanol, most
of the HP decomposed into water and oxygen. This confirms
the results obtained in tests carried out without a catalyst;
the radical scavenger does not so much affect the conversion
of HP, but it does affect the reaction pathway by which radi-
cals generated by the HP conversion react further.
Our results highlight that the BV oxidation of cyclohexa-
none, and presumably of cyclic ketones in general, indeed
comprises a complex network of reactions that have often
been overlooked in the literature, such as the consecutive
transformation of the very reactive e-caprolactone into 6-hy-
droxyhexanoic acid, AA, and lighter diacids. This would
suggest that it might not be possible to achieve both high
conversion of cyclohexanone and high proportional yield of
e-caprolactone at the same time; a result which is in contrast
with most reports on the catalyzed BV oxidation of cyclo-
hexanone with HP. Remarkably, this reaction network is the
same in both thermal and TS-1-catalyzed reactions, which
strongly suggests that the same active oxidizing species are
involved in the two cases. This supports the unprecedented
view that the Ti-OOH hydroperoxo species does not act as
an electrophile toward the activated carbonyl in the sub-
strate, but acts as a sort of reservoir for the oxidant species
through the controlled decomposition of the hydroperoxo
into hydroxy radicals.
~ ~
Figure 4. Isolated yield of (adipic+glutaric+succinic) acids (
, ) and
&
HP conversion (&, ) as a function of reaction time. Reaction conditions:
Open symbols: 5 mL cyclohexanone, 5 mL H2O, 15.2 mL H2O2 (35%)
(molar ratio HP/cyclohexanone 3:1). Full symbols: 5 mL cyclohexanone,
5 mL H2O, and 29 mL tert-butanol (molar ratio tert-butanol/cyclohexa-
none 6.3), 15.2 mL H2O2 (35%) (molar ratio HP/cyclohexanone 3:1), TS-
^
1 3.83 g. T=908C. Isolated yield with filtered and reused TS-1 ( ); iso-
lated yield with regenerated TS-1 catalyst: after regeneration of TS-1 by
washing with water and acetone ( ), and after regeneration of TS-1 by
*
*
calcination at 4508C in air for 3 h ( ).
Chem. Eur. J. 2010, 16, 12962 – 12969
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
12967

F. Cavani et al.
framework incorporation of TiIV were checked by means of XRD, FTIR,
and UV/Vis diffuse reflectance spectroscopy. The amount of Ti incorpo-
rated was 2 wt%. The crystal size of this sample was <0.2 mm. The slurry
containing TS-1 crystallites was spray dried after the addition of silica
Ludox, by using a Lab-Spray drier (T of nebulization chamber 2008C,
nozzle diameter 2 mm). The final average size of agglomerated particles
was about 25 mm. The agglomerated samples were then calcined in air at
5508C.
One possible explanation for discrepancies between our
data and previously reported results is that the distribution
of products, and therefore, the relative importance of each
reaction rate in the mentioned network is dramatically af-
fected by both the solvent used (because the solvent may
act as a radical scavenger and thus slow down the consecu-
tive transformation of the very reactive lactone) and by the
characteristics of the catalyst used. For instance, we found
that the Lewis-type Sn-b catalyst reported by Corma et
al.[22,23] offers unique reactivity, and that its behavior is enor-
mously affected by reaction conditions. When we used this
catalyst under conditions similar to those reported in the
present work (i.e., temperature 908C, reaction time 3 h,
0.06 g catalyst, H2O (0.2 mL), cyclohexanone (2 mmol), HP
(6 mmol)), the Sn-b afforded very low cyclohexanone con-
version (0.9%), with 20% proportional yield of e-caprolac-
tone and 70% AA. However, when the same catalyst was
used under conditions closer to those described in papers by
Corma et al.[22,23] (i.e., with dioxane as solvent, temperature
908C, reaction time 3 h, 0.06 g catalyst, 2 mmol cyclohexa-
none, 1.3 mmol HP, to achieve a cyclohexanone/HP molar
ratio of 1.5 and a dioxane/cyclohexanone wt ratio of 30), cy-
clohexanone conversion was 19.4%, with 98% proportional
yield of e-caprolactone and only 2% AA.
Typical conditions for the thermal, uncatalyzed oxidation of cyclohexa-
none with HP: In a 10 mL pyrex reactor with screw stopper, a mixture of
H2O (0.1 mL), H2O2 (30%, 3 mmol, 0.340 g, 0.304 mL; source Carlo
Erba), and cyclohexanone (1 mmol, 0.098 mg, 0.103 mL; sources: 1) Al-
drich, code 29140, ꢁ99.5%, containing ꢂ0.1 ppm Fe; ꢂ0.02 ppm Co;
ꢂ0.02 ppm Cr; ꢂ0.02 ppm Cu; 2) Aldrich code 02482, ꢁ99.9%) were
stirred at 908C for the required time. The reaction mixture was then
cooled to room temperature and diluted with a 1:1 mixture of H2O/
CH3CN to 50 mL. After addition of a known amount of benzoic acid as
internal standard, subsequent dilutions were required to prepare samples
for analysis, in particular 5mm of internal standard for positive ion and
0.5mm for negative ion mode. In TS-1-catalysed reactions, a mixture con-
taining H2O, 30 % H2O2, and TS-1 (0.077 g) was stirred at room tempera-
ture for 15 min. Cyclohexanone was then added, and the mixture was
heated under stirring for the required time. After cooling the reaction
mixture to room temperature, the catalyst was removed by Buchner fil-
tration, and washed with H2O (5 mL), then with CH3CN (5 mL). The fil-
trate was diluted with a 1:1 mixture of H2O/CH3CN (10 mL). Hydrogen
peroxide conversion was determined by iodometric titration.
Product concentration was determined by ultraperformance liquid chro-
matography (UPLC)–mass spectrometry (gas chromatography (GC) is
not suitable for the identification of dicarboxylic acids.[26] Indeed, in most
reports on the oxidation of cyclic ketones with hydrogen peroxide (with a
few exceptions[4,10]), the analysis of the reaction products is carried out
Conclusion
by means of GC, which does not allow the detection of diacids.[5–9,11–20]
)
UPLC analysis was performed by using an Acquity Ultra Performance
Waters UPLC System equipped with autosampler. UPLC separation was
conducted on a Acquity UPLC BEH C18 column (1.7 mm, 2.1ꢃ100 mm)
at a flow rate of 0.3 mLminꢀ1 with an elution gradient: eluent A H2O
(0.2% HCOOH), eluent B (0.2% HCOOH); elution: 0–2 min isocratic
100% A, 2–10 min linear gradient to 65% A-35% B, 10–10.5 min linear
gradient to 100% B, 10.5–13.5 min isocratic 100% B, 13.5–14 min linear
gradient to 100% A, 14–19 min isocratic 100% A; flow rate
0.3 mLminꢀ1. Both the column and the autosampler were maintained at
258C. Mass spectrometry conditions in positive ion mode: capillary volt-
age 3.0 kV, cone voltage 28.0 V, source temperature 1008C, desolvation
temperature 2008C, cone gas flow (N2) 100 Lh,ꢀ1 desolvation gas flow
(N2) 600 Lhꢀ1. Mass spectrometry conditions in negative ion mode: capil-
lary voltage 2.8 kV, cone voltage 23.0 V, source temperature 1008C, de-
solvation temperature 2008C, cone gas flow (N2) 100 Lhꢀ1, desolvation
In the Baeyer–Villiger oxidation of cyclohexanone with
aqueous hydrogen peroxide under conditions aimed at ob-
taining e-caprolactone, an uncatalyzed, thermally activated
radical reaction leads to the concurrent formation of adipic,
succinic, and glutaric acids, even when a stoichiometric
amount of the oxidant is used. e-Caprolactone is the primary
reaction product, but it is more reactive than cyclohexanone,
and it quickly undergoes consecutive transformations by one
of two different reaction pathways: 1) a hydrolythic pathway
to 6-hydroxyhexanoic acid, which also is oxidized to AA, al-
though more slowly than the concurrent ones, or 2) a direct
oxidative scission to AA.
The relevant reaction rates are modified when TS-1 is
used as catalyst. In this case, the high concentration of hy-
droxy radicals within pores accelerates the reaction rates, es-
pecially the consecutive formation of AA and of lighter di-
acids. The proper choice of solvent, which may also act as a
radical scavenger, both without catalyst and with TS-1, is a
powerful tool for controlling the rates of the various reac-
tions involved. With either tert-butanol or dioxane, which
are both efficient radical scavengers, all reaction rates are
slowed down, especially the consecutive oxidation of the
very reactive e-caprolactone into diacids.
gas flow (N2) 600 Lhꢀ1
.
Experiments focused on the determination of the isolated yield were car-
ried out by using cyclohexanone (5 mL), H2O (5 mL), H2O2 (35%,
15.2 mL, to give molar ratio HP/cyclohexanone 3:1), and TS-1 (3.83 g)
with no tert-butanol or with 29 mL tert-butanol. In the latter case, the HP
solution was slowly added over 1 h, and then the reaction was left to pro-
ceed for the desired time. Note that it was not possible to add all of the
reactants at once, because the very rapid decomposition of HP leads to a
sudden and very dangerous release of a large volume of oxygen. This
causes a sudden pressure increase in the vessel, even when it is unsealed
and exposed to air. The weight of the isolated yield was measured as fol-
lows: the hot slurry downloaded from the reactor was filtered to recover
the TS-1 catalyst; then the filtrate was concentrated under vacuum at
808C to remove unconverted cyclohexanone and solvents (water and/or
tert-butanol). The solid obtained was washed four times with cold water,
dried overnight at 808C, and then weighed. Finally, a part of the solid
was dissolved in hot water and analyzed to determine the relative
amount of each product. These experiments were also aimed at deter-
mining the activity of the spent TS-1. The recovery of the spent catalyst
always led to a loss of approximately 10% of the original catalyst; the
Experimental Section
General: Titanium-silicalite-1 (TS-1) was synthesized according to a pre-
viously described procedure.[27] The structural integrity of the sample and
12968
www.chemeurj.org
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 12962 – 12969

Baeyer–Villiger Oxidation of Cyclohexanone to e-Caprolactone
FULL PAPER
amount lost was replaced with an analogous amount of catalyst recov-
ered after another experiment. Then the catalyst was either reused as
such, or regenerated by simply washing with water and acetone, or by
washing and then by thermal treatment in air at 4508C for 3 h.
[7] L. Balbinot, U. Schuchardt, C. Vera, J. Sepffllveda, Catal. Commun.
2008, 9, 1878–1881.
[8] A. Corma, M. T. Navarro, L. Nemeth, M. Renz, Chem. Commun.
2001, 2190–2191.
[9] Z. Lei, Q. Zhang, J. Luo, X. He, Tetrahedron Lett. 2005, 46, 3505–
3508.
[10] U. R. Pillai, E. Sahle-Demessie, J. Molec. Catal. A 2003, 191, 93–
100.
[11] C. Jimenez-Sanchidrian, J. M. Hidalgo, R. Llamas, J. R. Ruiz, Appl.
Catal. A 2006, 312, 86–94.
[12] C Li, J. Wang, Z. Yang, Z. Hu, Z. Lei, Catal. Commun. 2007, 8,
1202–1208.
[13] A. Lambert, D. J. Macquarrie, G. Carr, J. H. Clark, New J. Chem.
2000, 24, 485–488.
[14] C. Chen, J. Peng, B. Li, L. Wang, Catal. Lett. 2009, 131, 618–623.
[15] C. Palazzi, F. Pinna, G. Strukul, J. Mol. Catal. A 2000, 151, 245–252.
[16] R. Llamas, C. Jimenez-Sanchidrian, J. R. Ruiz, Tetrahedron 2007, 63,
1435–1439.
In tests for isolated yield determination, the products were also analyzed
by gas-chromatography (GC). Analysis of cyclohexanone and e-caprolac-
tone is possible by GC, but the analysis of diacids (adipic, succinic, and
glutaric acids) is not; the latter need derivatization (e.g., by reaction with
methanol) to be analyzable by GC.[26] The procedure for diacid derivati-
zation was as follows: The reaction mixture was heated to 658C in a
vacuum to remove both the unconverted cyclohexanone and the initial
reaction solvent. The residual solid was mixed with an excess of methanol
and BF3 (dissolved in methanol) at 808C for 1 h in a sealed vessel. The
esters were then extracted with n-hexane (this also led to the precipita-
tion of BF3 from the mixture); the extraction procedure was repeated
several times. Lastly, the solution of diesters in n-hexane was analyzed by
GC (internal standard n-decane). This method allowed the identification
of acids in the form of methyl esters.
[17] R. Llamas, C. Jimenez-Sanchidrian, J. R. Ruiz, Appl. Catal. B 2007,
72, 18–25.
[18] A. Bhaumik, P. Kumar, R. Kumar, Catal. Lett. 1996, 40, 47–50.
[19] M. Renz, T. Blasco, A. Corma, V. Fornes, R. Jensen, L. Nemeth,
Chem. Eur. J. 2002, 8, 4708–4717.
Acknowledgements
The ISA (Institute of Advances Studies), at the University of Bologna, is
acknowledged for financial support to KR. Prof. A. Corma is gratefully
acknowledged for providing the Sn-b sample.
[20] A. Corma, L. T. Nemeth, M. Renz, S. Valencia, Nature 2001, 412,
423–425.
[21] R. Criegee, Justus Liebigs Ann. Chem. 1948, 560, 127.
[22] M. Boronat, A. Corma, M. Renz, G. Sastre, P. M. Viruela, Chem.
Eur. J. 2005, 11, 6905–6915.
[23] M. Boronat, P. Concepcion, A. Corma, M. Renz, Catal. Today 2007,
121, 39–44.
[1] R. A. Michelin, P. Sgarbossa, A. Scarso, G. Strukul, Coord. Chem.
Rev. 2010, 254, 646–660.
[24] F. J. Johnston, J. Phys. Chem. 1983, 87, 1498–1502.
[25] Y. Usui, K. Sato, Green Chem. 2003, 5, 373–375.
[26] I. Mato, S. Suµrez-Luque, J. F. Huidobro, Food Res. Int. 2005, 38,
1175–1188.
[2] G. Strukul, Angew. Chem. 1998, 110, 1256–1267; Angew. Chem. Int.
Ed. 1998, 37, 1198–1209.
[3] C. Jimenez-Sanchidrian, J. R. Ruiz, Tetrahedron 2008, 64, 2011–
2026.
[27] M. Taramasso, G. Perego, B. Notari (Enichem), US 4410501, 1983.
[4] J. Fischer, W. Hoelderich, Appl. Catal. A 1999, 180, 435–443.
[5] M. Lenarda, M. Da Ros, M. Casagrande, L. Storaro, R. Ganzerla,
Inorg. Chim. Acta 2003, 349, 195–202.
[6] R. A. Steffen, S. Teixeira, J. Sepulveda, R. Rinaldi, U. Schuchardt, J.
Mol. Catal. A 2008, 287, 41–44.
Received: June 23, 2010
Published online: September 28, 2010
Chem. Eur. J. 2010, 16, 12962 – 12969
ꢂ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
12969