Evidence for cyclodextrin dioxiranes

Article abstract of DOI:10.1016/S0008-6215(98)00111-6

α-Cyclodextrin, β-cyclodextrin, 2,6-di-O-methyl-β-cyclodextrin, methyl-β-cyclodextrin and sucrose have been oxidised by aqueous bromine solution at neutral pH. Both ketone and carboxylic acid containing materials are among the products of the oxidations. For α-cyclodextrin there is clear ¹³C NMR evidence for the presence of a ketone group and its hydrate form. This together with the continued ability of the product to complex p-nitrophenol indicates that the ketone is present at the secondary rim of an intact cyclodextrin ring. A pH dependence for the reaction of bromine with cyclodextrin shows that the maximum rate of bromine loss roughly coincides with the maximum concentration of hypobromous acid, HOBr, indicating that this is the reactive species in these oxidations. The results are consistent with a mechanism involving attack by one of the secondary hydroxyls of cyclodextrin on HOBr, with Br^- leaving to yield an intermediate dehydroxy hydroperoxy cyclodextrin that subsequently decomposes to a keto-cyclodextrin via a Kornblum-De La Mare-type reaction. An alternative pathway prevails when the reaction is carried out under alkaline conditions, where carboxylic acids are the principle products. The keto derivatives produced by bromine oxidation at neutral pH are capable of catalysing the oxidation of p-nitrophenol and aryl-alkyl sulfoxides by peroxomono-sulfate in an analogous way to cyclohexanone, which is known to form a dioxirane upon reaction with peroxomonosulfate. It is likely that dioxirane formation is responsible for the observed catalysis in the present case also.

Full text of DOI:10.1016/S0008-6215(98)00111-6

Carbohydrate Research 309 (1998) 17±29
Evidence for cyclodextrin dioxiranes
Michael E. Deary, D. Martin Davies*
Department of Chemical and Life Sciences, University of Northumbria at Newcastle, Newcastle Upon
Tyne, UK, NE1 8ST
Received 18 December 1997; accepted 17 April 1998
Abstract
ꢀ-Cyclodextrin, ꢁ-cyclodextrin, 2,6-di-O-methyl-ꢁ-cyclodextrin, methyl-ꢁ-cyclodextrin and
sucrose have been oxidised by aqueous bromine solution at neutral pH. Both ketone and car-
boxylic acid containing materials are among the products of the oxidations. For ꢀ-cyclodextrin
1
3
there is clear C NMR evidence for the presence of a ketone group and its hydrate form. This
together with the continued ability of the product to complex p-nitrophenol indicates that the
ketone is present at the secondary rim of an intact cyclodextrin ring. A pH dependence for the
reaction of bromine with cyclodextrin shows that the maximum rate of bromine loss roughly
coincides with the maximum concentration of hypobromous acid, HOBr, indicating that this is the
reactive species in these oxidations. The results are consistent with a mechanism involving attack
-
by one of the secondary hydroxyls of cyclodextrin on HOBr, with Br leaving to yield an inter-
mediate dehydroxy hydroperoxy cyclodextrin that subsequently decomposes to a keto-cyclodex-
trin via a Kornblum±De La Mare-type reaction. An alternative pathway prevails when the
reaction is carried out under alkaline conditions, where carboxylic acids are the principle products.
The keto derivatives produced by bromine oxidation at neutral pH are capable of catalysing the
oxidation of p-nitrophenol and aryl±alkyl sulfoxides by peroxomono-sulfate in an analogous way
to cyclohexanone, which is known to form a dioxirane upon reaction with peroxomonosulfate. It
is likely that dioxirane formation is responsible for the observed catalysis in the present case also.
#
1998 Elsevier Science Ltd. All rights reserved
Keywords: Cyclodextrin; Bromine oxidation of cyclodextrin; Ketone; Dioxirane; Phenol oxidation
1
. Introduction
(50 mM) and ꢀ-cyclodextrin (2.4 mM) bu€ered at
pH 7.8, the generated bromine rapidly dis-
appeared, presumably consumed in a reaction with
the cyclodextrin. We subsequently learned of sev-
eral studies on the bromine oxidation of a range of
carbohydrates at neutral pH [2], including sucrose
[3], methyl ꢀ- and ꢁ-pyranosides of d-Gal, d-Glc
and d-Man [4], ꢀ,ꢀ- and ꢁ,ꢁ-trehalose [5], amylo-
pectin [6], and starches [7,8]. In the bromine oxi-
dation of methyl ꢀ- and ꢁ-pyranosides of d-Gal, d-
Glc and d-Man, for example, the secondary
We recently decided to extend our work on the
a€ect of cyclodextrin on the peracid±iodide reac-
tion [1] to the analogous bromide oxidation reac-
tion. During preliminary work on this system we
were surprised to see that on addition of perox-
omonosulfate to a solution containing bromide
* Corresponding author. Tel.: 0191-227-4721; fax: 0191-
27-3519.
2
0
008-6215/98/$19.00 # 1998 Elsevier Science Ltd. All rights reserved
PII S0008-6215(98)00111-6

18
M.E. Deary, D.M. Davies/Carbohydrate Research 309 (1998) 17±29
hydroxyls are oxidised to keto-groups (2-, 3- or 4-
uloses). Characterisation was accomplished by
using H NMR, elemental analysis and GC-MS
after ®rst stabilising these sensitive compounds
case of ꢀ-cyclodextrin, whereas for the other com-
1
pounds it was between 2= and 5 h; this is in general
2
1
agreement with the time taken for complete bro-
mine loss when the reaction was followed spectro-
photometrically. Additionally, it was generally
observed that the pH would start to increase at this
point, reaching approximately 7.5 to 7.8 before
returning to 7.0 (there was no acid dispensing
facility on the pH-stat to counter this rise). For
samples that were subject to bromide removal,
generally only one fraction was eluted from the ion
[
4,9] by conversion to the corresponding O-methyl
oxime [4]. A small degree of oxidation of the pri-
mary alcohols was also reported, although these
are rapidly converted to uronic acids [4].
The bromine oxidation of cyclodextrins should
give rise to similar products to that of the methyl
pyranosides. We report here on studies of this
reaction and the characterisation of the products,
which show that a keto-cyclodextrin is indeed pro-
duced. In addition to the native cyclodextrins
similar studies were carried out on a randomly O-
methylated ꢁ-cyclodextrin and on 2,6-di-O-methyl-
ꢁ
-cyclodextrin and sucrose, which is known to
form a ketone derivative when subjected to mild
oxidation with bromine [3]. 2,6-Di-O-Methyl-ꢁ-
cyclodextrin was used because it has only one
hydroxyl group available for reaction with bromine
(C-3), allowing us some certainty in the position of
any keto-group and also, hopefully, improving the
stability of the keto derivative.
The keto-cyclodextrins, like other ketones, are
capable of catalysing a number of oxidation reac-
tions of peroxomonosulfate, most probably due to
the formation of the more reactive dioxirane inter-
mediate [10]. Preliminary results are presented on
the oxidation of a chiral aryl alkyl sulfoxide and an
oxidatively inert substituted phenol. The latter
oxidation is compared with literature reports of
conventional chemical methods for the treatment
of high concentration chemical process waste water
streams.
Fig.
1. Sequential
3
spectra
for
the
reaction
of
�
� 3
� 3
� 3
1
.8Â10 mol dm
bromine with 2.2Â10 mol dm
ꢀ-
cyclodextrin in pH 7.8 phosphate bu€er, ionic strength
� 3
ꢀ
0
.2 mol dm
at 25 C. The dashed and solid lines show,
respectively, the decrease in absorbance across the spectrum
due to the reaction of bromine, and the subsequent emergence
of absorbance bands.
2
. Results and discussion
General comments on the reaction of bromine
with a-cyclodextrin.ÐFig. 1 shows the e€ect of
adding ꢀ-cyclodextrin to bromine in pH 7.8 phos-
ꢀ
phate bu€er at 25 C. Over the course of 20 min the
absorbance due to the bromine disappeared and a
colourless intermediate was formed. Then over the
next few hours absorbance bands emerged between
2
40 and 340 nm. The product of this reaction
complexes p-nitrophenol (Fig. 2) con®rming that
the ring integrity of the reacted cyclodextrin is lar-
gely maintained. In preparations carried out using
the pH-stat, it was observed that the majority of
the NaOH was dispensed in the ®rst 30 min in the
�
4
� 3
Fig. 2. Spectrum of 1.0Â10 mol dm p-nitrophenol in pH
�
3
ꢀ
7
line; and in the presence of: 0.01 mol dm
.8 phosphate bu€er, ionic strength 0.2 mol dm , 25 C, solid
ꢀ-cyclodextrin,
�
3
� 4
� 3
dotted line; and 3.8Â10 mol dm of the ketone containing
product, dashed line.

M.E. Deary, D.M. Davies/Carbohydrate Research 309 (1998) 17±29
19
exchange column, however when ꢀ-cyclodextrin
2.4 mM) was reacted with a 10-fold excess of bro-
ꢁ-cyclodextrin and 2,6-di-O-methyl-ꢁ-cyclodextrin
was accomplished using 13C NMR spectroscopy.
This con®rmed that signi®cant amounts of ketone
and carboxylic acid were present in the oxidation
products of ꢀ-cyclodextrin (Fig. 3). No ketone
resonance could be observed for either ꢁ-cyclodex-
trin or 2,6-di-O-methyl-ꢁ-cyclodextrin, though
there were strong carboxylic acid resonances. The
ketone carbonyl carbon signals were identi®ed by
comparison with published values for bromine
oxidation products of methyl ꢀ-d-glucopyranoside
in which the secondary hydroxyls at either the C-2
or C-4 are oxidised to give the keto-derivatives 1
and 2, respectively [4]. Comparisons were also made
with data published for the corresponding 3-ulose
3 [4], and with the bromine oxidation products of
potato starch [11] and dextran 4, which is an ꢀ-
(1!6) linked d-glucan [13]. Table 1 summarises
the literature ꢂ values and those obtained for bro-
mine oxidation products of cyclodextrins. In addi-
tion to the ketone and carboxylic acid shifts Table 1
lists the shifts for ketone hydrates, >C(OH)2,
which exist in equilibrium with the keto forms in
aqueous solution [14,15]. Compounds 1 and 2 for
(
mine at least three fractions were eluted from the
column. This, together with the observation that
none of the three fractions were able to
bind p-nitrophenol suggests that the cyclodextrin
ring is broken up under these conditions producing
charged ring fragments (presumably carboxylic
acid containing [11]). When 2,6-di-O-methyl-ꢁ-
cyclodextrin was oxidised under similar conditions
only one fraction was obtained, although sig-
ni®cantly reduced p-nitrophenol complexation was
observed compared to the parent compound. A
general characterisation of the main products of
these reactions was accomplished using 13C NMR
spectroscopy, as detailed in the next section. We
did not, however, attempt to characterise the col-
ourless intermediate of this reaction although there
is literature evidence to suggest that it may be a
dehydroxy hydroperoxy cyclodextrin and this is
discussed later [12].
13C NMR characterisation of bromine oxidation
products of cyclodextrin.ÐA general characterisa-
tion of the bromine oxidation products of ꢀ- and
Fig. 3. Fully decoupled 13C NMR spectrum of keto-ꢀ-cyclodextrin in H
O at 67.8 MHz.
2

20
M.E. Deary, D.M. Davies/Carbohydrate Research 309 (1998) 17±29
Table 1
Comparison of literature 13C NMR data with the main 13C NMR resonances obtained for the bromine oxidation, at neutral pH, of
native and substituted cyclodextrins, dextran and potato starch
Chemical shifts (ppm)
Compound
Reference
>
C=O
-COOH
>C(OH)
2
Bromine oxidation products of:
-cyclodextrin
ꢀ
208.8
171.6
176.0
93.0
95.6
92.6
93.7
This work
This work
2
06.8
Ð
ꢁ
-cyclodextrin
1
71.1
9
9
9
9
5.3
6.4
7.4
8.1
2,6-O-methyl-ꢁ-cyclodextrin
Potato Starch
Ð
206.4
162.4
94.4
Ð
This work
[11]
9
5.4
93.4
93.8
Dextran 4
204.2
174.0
176.3
[13]
2
07.2
Ð
1
1
76.7
76.8
Other compounds:
1
2
n/aa
n/a
n/a
n/a
n/a
93.4
94.1
Ð
Ð
98.7
[13]
[13]
[13]
[16]
[18]
204.9
207.2
211.1
Ð
3
d-xylo-hexos-5-ulose(5-keto-glucose)
Keto-ꢁ-cyclodextrin
aNot applicable.
example exist almost exclusively as the hydrate and
keto form, respectively, whereas for 3 there are
signi®cant proportions of both forms [13].
The 13C NMR data are qualitative and therefore
ketone resonances are likely to be signi®cantly
under-represented, which may also explain the dif-
that observed by Czarnik et al. for a keto-ꢁ-cyclo-
dextrin [18]. For 2,6-di-O-methyl-ꢁ-cyclodextrin, a
very weak 13C spectrum was obtained in which no
ketone carbonyl carbons or the corresponding
ketone hydrates could be detected, though there
was a signi®cant carboxylic acid resonance.
Nevertheless, the accelerative e€ect of this com-
pound on sulfoxide oxidation, as reported later,
indicates the presence of some ketone groups.
®culties in observing these signals for ꢁ-cyclodex-
trin, which is expected to undergo a similar
transformation to that seen for ꢀ-cyclodextrin.
Nevertheless, a series of weak signals were
observed between ꢂ 92.6 and ꢂ 98.1 for the bromine
oxidation product of ꢁ-cyclodextrin, which is the
region in which resonances for the ketone hydrate
would be expected [17]. Two of these resonances
coincide with the much stronger ketone hydrate
resonances observed for ꢀ-cyclodextrin, and the
resonance observed at ꢂ 98.1 is in agreement with
E€ect of pH.ÐBromine speciation is dependent
on pH, as shown by the distribution diagram (Fig.
4) and, consequently, pH should be expected to
have a signi®cant e€ect on its reactivity. Fig. 4 has
been calculated by obtaining a solution to the sys-
tem of eqs (1)±(3), for which it was necessary to
assume the presence of an excess of bromide.
Values of 2Â10� mol dm , 16.85 dm mol and
9
� 3
3
� 1

M.E. Deary, D.M. Davies/Carbohydrate Research 309 (1998) 17±29
21
Fig. 4. Composition of an aqueous bromine solution as a
function of pH, where the total bromine concentration is
Fig. 5. E€ect of pH on absorbance changes at 350 nm during
the initial and subsequent stages of the reaction between
�
4
� 3
and there is an excess of bromide
0
.5Â10 mol dm
� 3
� 3
� 3
� 3
1
cyclodextrin. Bu€ers were phosphate or carbonate, ionic
0.32Â10 mol dm
bromine and 3.33Â10 mol dm
ꢀ-
�
3
(
0.1 mol dm ).
�
3
ꢀ
strength 0.2 mol dm , 25 C. The inset shows the e€ect of pH
on the initial rate of bromine loss, conditions as for the main
®gure.
�
9
2
� 6
9
respectively [19,20].
Â10 mol dm , were used for K , K and Kh,
a
Br3
trins [22] and starch [23]. In these reactions a
hypochlorite/bromide system was used whereby
�
‡
‰
BrO ŠÁ‰H Š
K ˆ
ꢀ1†
ꢀ2†
ꢀ3†
a
�
�
‰
HOBrŠ
there was rapid Br /OCl interconversion to the
�
proposed reactive species, HOBr and BrO .
�
‰
Br Š
A further signi®cant ®nding is the observation
that if the bromine oxidation reaction was carried
out at pH 7.2 and the solution subsequently raised
to pH 10.8 then a rapid and substantial increase in
absorbance was observed (Fig. 6). This is very dif-
ferent from the behaviour observed when the reac-
tion was carried out entirely at high pH (Fig. 5).
3
KBr
ˆ
�
‰
Br ŠÁ‰Br Š
2
‡
�
‰
HOBrŠÁ‰H ŠÁ‰Br Š
K ˆ
h
‰
Br Š
2
The initial rate of bromine loss upon its reaction
with ꢀ-cyclodextrin, and the extent to which
absorbance bands emerge was signi®cantly a€ected
by pH as shown in Fig. 5. When the reaction was
carried out above pH 9 there was minimal emer-
gence of the absorbance bands and this is most
likely due to a pathway whereby dicarboxylic acids
are formed rather than ketone groups. The 13
C
NMR studies detailed in Table 1 show that sig-
ni®cant amounts of carboxylic acid are formed
during the bromine oxidation of cyclodextrin and,
moreover, that the proportion increases if the
reaction is carried out at pH 8.5 compared to pH
7
.0 (results not shown). Literature studies on the
bromine oxidation of a variety of carbohydrate
substrates also report signi®cant carboxylic acid
formation at high pH and this reaction has been
used in the production of calcium-chelating poly-
carboxylate derivatives of inulin [21], linear dex-
Fig. 6. E€ect of raising the pH to 10.85, using NaOH, in a
�
3
� 3
solution containing 3.33Â10 mol dm ꢀ-cyclodextrin that
� 3
� 3
had previously reacted with 10.32Â10 mol dm bromine in
�
3
ꢀ
pH 7.20 phosphate bu€er, ionic strength 0.2 mol dm , 25 C.

22
M.E. Deary, D.M. Davies/Carbohydrate Research 309 (1998) 17±29
hydrogen bonding interactions between hydroxyls
at the secondary rim. The relatively low solubilities
of native cyclodextrins have been attributed to
hydrogen bonding between the hydroxyls, resulting
in reduced interaction with the bulk solvent [27].
ꢁ
ꢀ
-Cyclodextrin is considerably less soluble than
-cyclodextrin, possibly implying greater hydrogen
bonded interactions. It follows that this may well
a€ect the extent of reaction of hypobromous
acid with the hydroxyl groups. The fact that the
ꢁ
-cyclodextrin derivatives that we have looked at
also react slowly with bromine does not necessarily
detract from the latter argument since although
solubility is greatly increased as a result of the
derivatisations disrupting the hydrogen bonded
structure, there are less reactive sites.
Fig. 7. Comparison of the relative reactivities of ꢀ-and ꢁ-
�
3
� 3
)
cyclodextrin
�
(3.33Â10 mol dm
towards
bromine
3
� 3
(
strength 0.2 mol dm , 25 C.
10.32Â10 mol dm ) in a pH 6.66 phosphate bu€er, ionic
�
3
ꢀ
Mechanisms.ÐIt is clear from the pH depen-
dence that there are two separate pathways for the
initial oxidation step of cyclodextrins by bromine;
one occurring at neutral pH, where ketones are
signi®cant products, and the other taking place
under alkaline conditions where carboxylic acids
are largely produced. Additionally, the inter-
mediate formed when the oxidation takes place
under neutral conditions subsequently decomposes
in either neutral or alkaline conditions as demon-
strated by Fig. 6.
Whilst there have been a number of studies car-
ried out on the bromine oxidation of secondary
alcohols over the past three decades, many invol-
ving carbohydrates, there is still no general con-
sensus on the role of the main bromine species in
these reactions over the full pH range. Palou, has
recently reviewed the mechanism of bromine oxi-
dations of a range of organic molecules, including
secondary alcohols, however the author mainly
The inset to Fig. 5 shows a bell-shaped curve for
the e€ect of pH on the initial rates of bromine
consumption. The maximum rate appears to occur
at about pH 8.7, however it is dicult to de®nitely
state whether the maximum actually coincides with
the pKa of BrOH (8.69 [24]) or the maximum
BrOH concentration (ꢁ pH 7.9 [25]) because of the
possibility of bu€er e€ects [26]; the change over to
carbonate from phosphate bu€er occurred at
approximately the point where the maximum
initial rate was observed.
E€ect of ring size of the cyclodextrin on the rate
of reaction.ÐAbsorbance traces for the reaction of
ꢀ
- and ꢁ-cyclodextrin with bromine at pH 6.7 are
shown in Fig. 7. The emergence of absorbance
bands due to the keto-cyclodextrin clearly begins
signi®cantly earlier for ꢀ-cyclodextrin compared to
ꢁ-cyclodextrin. It is possible that this is due to a
size factor whereby the reactive bromine species
concentrates on reactions involving Br [25]. The
2
binds strongly with ꢀ-cyclodextrin but less so with
most widely accepted mechanism for this species,
which predominates at pH's below 6, involves rate
limiting hydride transfer from the carbon to Br2
followed by a rapid proton removal step to yield
the ketone (Scheme 1) [25,28,29]. The e€ect of
electron withdrawing or donating groups on the
rate of bromine oxidation of a range of secondary
alcohols supports this view [30]. Alternative
ꢁ
-cyclodextrin, which because of proximity e€ects
leads to a much faster rate with ꢀ-cyclodextrin. As
far as we are aware there are no published stability
�
constants for BrOH or BrO with cyclodextrins.
Whilst di€erent binding anities of hypobromous
acid with the cyclodextrins may be an obvious
explanation, another plausible reason concerns the
Scheme 1.

M.E. Deary, D.M. Davies/Carbohydrate Research 309 (1998) 17±29
23
mechanisms at low pH involve the formation of a
hypobromite ester by direct attack of the bromine
on the hydroxyl together with proton abstraction
from the carbon [31]. In studies conducted at pHs
above 6, where hypobromous acid and hypo-
bromite predominate, mechanisms involving
hydroxide. It was suggested by Czarnik et al. that
the dehydroxy hydroperoxy ꢁ-cyclodextrin decom-
poses to a dialdehyde [12], however the evidence is
not conclusive and it is plausible that a ketone is
formed, as is observed for the decomposition of the
intermediate in the present work. The Kornblum±
De La Mare reaction gives a precedent for the
conversion of ꢁ-hydroxy hydroperoxides to
ketones, and whilst our case is not strictly analo-
gous because of the absence of a neighbouring
aromatic group, it is still a reasonable mechanism
(Scheme 3) [32]. This is particularly so in the light
of the enhanced decomposition of the intermediate
upon raising the pH (Fig. 6).
hydride transfer to Br are still often referred to,
2
despite observed rate dependencies on HOBr con-
centration [4]. Doane and Whistler [6], and more
recently, Besumer and Van Bekkum [21±23] have
suggested the involvement of both HOBr and
�
BrO in the oxidation of a range of poly-
saccharides, in which carbonyl groups and car-
boxylic acids were the main products, with the
latter increasing in proportion at high pHs. How-
ever, no clear mechanisms were postulated for the
involvement of these species.
It is still necessary in this mechanism for the
carbon hydrogen to be removed (as a proton) and
so there is still consistency with the arguments of
1
It seems reasonable that in the case of cyclodex-
trins, and most probably for other carbohydrates
also, the mechanism at neutral pH could involve
direct attack of the secondary hydroxyl on HOBr,
Theander et al. who have produced H NMR evi-
dence which shows that the position at which the
keto groups are formed in the bromine oxidation
of carbohydrates is signi®cantly a€ected by steric
factors [3,4]. Thus, for methyl-ꢀ-d-Glc, virtually no
oxidation takes place at the C-3 position for which
the axial hydrogen in a syn diaxial relationship
with the bulky O-methyl group at the C1 position
and, consequently, only 2-uloses and 4-uloses are
formed [4]. It is uncertain whether the same steric
factors would be important in the case of cyclo-
dextrins.
�
with Br then leaving to yield a hydroperoxy
cyclodextrin; this would correspond to the colour-
less intermediate that we observe (Scheme 2). We
have no experimental evidence for the nature of
this intermediate but it has recently been reported
that ꢁ-cyclodextrin can react with hydrogen per-
oxide to yield a dehydroxy hydroperoxy cyclodex-
ꢀ
trin (8.8 mM cyclodextrin, 30% H O , 60 C for 3
2
2
days) [12]. This species decomposes with a half-life
of about 3 h at pH 8.5, which is similar to the
decomposition of the intermediate that we observe.
The authors suggest that hydrogen bonding
between the secondary hydroxyls of the cyclodex-
trin activates one of these with respect to nucleo-
philic attack on the hydrogen peroxide. A similar
attack on BrOH would be facilitated by the better
leaving group ability of bromide compared with
The mechanism involved when the bromine oxi-
dation takes place entirely under alkaline condi-
tions, and for which carboxylic acids are signi®cant
products, is unclear although the participation of a
dehydroxy hydroperoxy intermediate is unlikely.
Substrate oxidation.ÐThere is clear con®rma-
tion from 13C NMR spectroscopy that keto-cyclo-
dextrins are among the products obtained from the
bromine oxidation of cyclodextrins. At neutral pH
Scheme 2.
Scheme 3.

24
M.E. Deary, D.M. Davies/Carbohydrate Research 309 (1998) 17±29
ketones such as acetone and cyclohexanone can
catalyse the decomposition of peroxomonosulfate
and enhance the rate of many oxidation reactions
of peroxomonosulfate and it was these original
observations which led Montgomery to propose
the formation of dioxiranes, generated in situ from
the ketone and peroxomonosulfate [33]. Edwards
continued this work producing strong evidence
from 18O labelling experiments for the presence of
dioxiranes and also showing that dioxiranes are
powerful oxidising agents [34]. In view of this it is
reasonable to expect the keto-cyclodextrins to per-
form similar catalysis via the formation of a tran-
sient cyclodextrin dioxirane species. The possibility
of a dioxirane moiety attached to a chiral cyclo-
dextrin cavity raises interesting implications for
both catalysis and chiral oxidations.
comparison of absorbance changes for the oxidation
of (S)-(-)-methyl-p-tolyl sulfoxide in the presence of
peroxomonosulfate alone, peroxomonosulfate/
cyclohexanone
and
peroxomonosulfate/keto-
(ꢀ-cyclodextrin. There is a modest two-fold
increase in the initial rate of oxygen transfer for the
latter system, which is comparable to the cyclo-
hexanone system once relative concentrations are
accounted for. Table 2 summarises the e€ect of a
range of ketones on sulfoxide oxidation by perox-
omonosulfate. The observation of a reduced cata-
lytic e€ect of keto-ꢀ-cyclodextrin on the oxidation
of 4-bromophenyl methyl sulfoxide compared to
(S)-(-)-methyl-p-tolyl sulfoxide, where the former
would be expected to form the stronger inclusion
complex, and both would be expected to bind in an
orientation where the sulfoxide group is at the sec-
ondary hydroxyl rim [35], suggests that the cyclo-
dextrin dioxirane reacts principally with unbound
sulfoxides. An increased catalytic e€ect might have
been expected for the oxidation of the stronger
binding bromophenyl methyl sulfoxide if the con-
verse were true. It is possible, however, that the
orientation of sulfoxides within the cyclodextrin
cavity is unsuitable for interaction with a proximal
dioxirane moiety. Other classes of substrates may
show di€erent e€ects. For the O-methylated cyclo-
dextrins there are contrasting e€ects, with keto-
methyl-ꢁ-cyclodextrin actually showing rate inhi-
bition whereas the keto derivative of 2,6-di-O-
methyl-ꢁ-cyclodextrin is reasonably e€ective. It is
noteworthy that the keto-sucrose derivative proves
to be even more e€ective than keto-ꢀ-cyclodextrin.
In the context of interpretation of the results of
observed catalysis it is important to note that per-
oxomonosulfate has negligible anity for ꢀ-cyclo-
dextrin [36] and that this is almost certainly the
case for the keto-ꢀ-cyclodextrin derivatives also.
The observed rate enhancements are not, therefore,
due to the peroxide activation e€ects of cyclodex-
trin that were observed for perbenzoic acids and
long chain alkyl peracids [37].
We have looked at the e€ects of perox-
omonosulfate/ketone systems on the decomposi-
tion of peroxomonosulfate, the oxidation of aryl
alkyl sulfoxides to the corresponding sulfones and
the oxidation of p-nitrophenol. Fig. 8 shows a
Fig. 8. Catalytic e€ect of keto-ꢀ-cyclodextrin on the oxidation
of (S)-(-)-methyl-p-tolyl sulfoxide and on the decomposition of
peroxomonosulfate. The main ®gure shows the reaction of
�
�
3
� 3
� 3
4
mol dm (S)-(-)-methyl p-tolyl sulfoxide in the presence of
3
.25Â10 mol dm
peroxomonosulfate with 0.109Â10
3
�
3
� 3
� 3
Â10 mol dm
keto-ꢀ-cyclodextrin, solid line; 8Â10
The keto-ꢀ-cyclodextrin was also e€ective in the
decomposition of peroxomonosulfate, as shown in
the inset to Fig. 8, although this decomposition
coincides with a signi®cant and permanent loss in
absorbance over the wavelength range associated
with the keto-cyclodextrin. Thus, in contrast to
dioxiranes generated from acetone and cyclohex-
anone [10], minimal re-generation of the keto-
cyclodextrin was achieved during reaction with
peroxomonosulfate.
�
3
mol dm cyclohexanone, dashed line; and bu€er alone, dot-
ted line. The reactions were carried out in pH 7.6 phosphate
�
3
ꢀ
bu€er, ionic strength 0.2 mol dm , 25 C. Absorbance changes
were followed at 244 nm and were corrected for the back-
ground change in absorbance due to the keto-ꢀ-cyclodextrin
or cyclohexanone. The inset shows the e€ect of keto-ꢀ-cyclo-
dextrin on the decomposition of 1.92Â10 mol dm perox-
omonosulfate in pH 7.6 phosphate bu€er, where the symbols
represent: peroxomonosulfate/phosphate bu€er alone, open
�
3
� 3
�
3
� 3
squares; 12Â10 mol dm keto-ꢀ-cyclodextrin, ®lled circles;
�
3
� 3
0
.75Â10 mol dm
keto-ꢀ-cyclodextrin, open circles. All
�
5
� 3
solutions contained 2Â10 mol dm EDTA.

Table 2
E€ect of cyclohexanone and ketone containing products of the bromine oxidation of ꢀ-cyclodextrin (ꢀ-CD), 2,6-di-O-methyl-ꢁ-cyclodextrin (2,6-DM-ꢁ-CD), methyl-ꢁ-
cyclodextrin (M-ꢁ-CD) and sucrose on initial rate for the oxidation of aryl methyl sulfoxides by peroxomonosulfate (PMS)
�
�
3
Bu€era
� 3
[PMS]/10
mol dm
Ketone or ketone
containing products
of bromine oxidation
Substrate (S)
[S]
o
/10
mol dm
[Ketone]
� 3
Initial rate
/10 mol dm s
Acceleration of
initial rate
3
� 3
� 3
� 7
� 3 � 1
/10 mol dm
Cyclohexanone
(S)-(-)-Methyl-p-tolyl sulfoxide
0.109
pH 7.8, I=0.52 M
pH 7.6, I=0.2 M
pH 7.8, I=0.52 M
pH 7.6, I=0.2 M
5.13
4.38
5.13
4.25
0.00
2.89‹0.08
4.30‹0.13
2.20‹0.06
7.00‹0.19
3.76‹0.10
4.67‹0.13
2.04‹0.04
3.78‹0.04
2.86‹0.05
2.23‹0.03
2.73‹0.09
2.91‹0.04
1.76‹0.04
3.67‹0.15
5.07‹0.26
1.78‹0.02
3.04‹0.08
2.20‹0.35
1.95‹0.06
2.72‹0.06
2.19‹0.04
1.98‹0.04
1.54‹0.14
2.24‹0.09
4.49‹0.13
5.72‹0.15
2.20‹0.04
3.18‹0.18
3.90‹0.19
Ð
1.49
Ð
3.12
Ð
2
.00
0.00
.00
0.00
.00
0.00
0
.109
8
4
-Bromophenyl methyl sulfoxide
0.100
0.109
2
1.24
ꢀ
-CD
(S)-(-)-Methyl-p-tolyl sulfoxide
3
6
.30
.50
1.85
1.40
Ð
1.22
1.30
Ð
2.09
2.88
Ð
1.71
Ð
0.88
1.23
Ð
0.90
0.70
Ð
2.00
2.55
Ð
1.44
1.77
4
-Bromophenyl methyl sulfoxide
0.100
0.109
pH 7.6, I=0.2 M
pH 7.6, I=0.2 M
4.25
4.44
0.00
1
3
.63
.26
2
2
,6-DM-ꢁ-CD
,6-DM-ꢁ-CDb
(S)-(-)-Methyl-p-tolyl sulfoxide
0.00
.20
5
1
0.4
0.00
1.06
0.00
(S)-(-)-Methyl-p-tolyl sulfoxide
(S)-(-)-Methyl-p-tolyl sulfoxide
0.109
0.109
pH 7.6, I=0.2 M
pH 7.6, I=0.2 M
4.57
4.38
M-ꢁ-CD
3
7
.58
.16
4
-Bromophenyl methyl sulfoxide
0.100
0.109
0.100
pH 7.6, I=0.2 M
pH 7.6, I=0.2 M
pH 7.6, I=0.2 M
4.25
4.38
4.25
0.00
3
7
.58
.16
Sucrose
(S)-(-)-Methyl-p-tolyl sulfoxide
0.00
3
6
.14
.28
4
-Bromophenyl methyl sulfoxide
0.00
3
6
.14
.28
a
Phosphate.
Second batch, made with a 10-fold excess of bromine.
b

26
M.E. Deary, D.M. Davies/Carbohydrate Research 309 (1998) 17±29
When the peroxomonosulfate/keto-ꢀ-cyclodex-
oxidation of mono-substituted phenols due to the
complexity of the product mixture and, instead,
oxidation was carried out on hindered phenols
such as 2,6-di-tert-butyl phenol where the product
mix was more manageable [42]. The generally
accepted mechanism for the initial step in the oxi-
dation of phenols, including that by dioxiranes, is a
single electron transfer from the phenol to the oxi-
dant yielding a phenoxy radical which can then
react further to form a complex mixture of pro-
ducts including catechols, hydroxyquinones, qui-
nones, diphenyls and other polymeric compounds
[43]. Additionally, the hydroxylated compounds
and benzoquinones can undergo further oxidation
in which there is ring cleavage, yielding muconic
and 2,5-dioxo-3-hexenedioc acids and subsequent
products [41,44]. It is reasonable to assume that
similar product mixtures result from the oxidation
of p-nitrophenol by the keto-sugar peroxomono-
sulfate systems.
trin system was applied to p-nitrophenol oxidation,
as shown in Fig. 9, the observed catalytic e€ect was
greater than for sulfoxide oxidation. Again the
involvement of dioxirane intermediates, generated
from the keto-derivatives, is an explanation for the
observed accelerative e€ects. The oxidation of
phenols by chemical methods has been widely
investigated for the treatment of waste water
streams originating from chemical processes where
the concentration of organics is often high enough
to be toxic to conventional biological treatment
processes. Methods employed for phenol degrada-
tion include wet air oxidation [38], photocatalysis
using TiO [39], and supercritical water systems
2
[
40]. Peroxyacids such as peroxomonosulfate react
slowly with p-nitrophenol as shown in Fig. 9,
however in the presence of keto-ꢀ-cyclodextrin and
the keto-sucrose, the acceleration of p-nitrophenol
decomposition was comparable to that seen for
cyclohexanone yet with signi®cantly lower con-
centrations. Non-derivatised ꢀ-cyclodextrin inhi-
bits p-nitrophenol oxidation. No attempt to
identify the products of p-nitrophenol oxidation
was made. This would have been a signi®cant
study in its own right and the oxidation of phenols
by a variety of oxidants, including dioxiranes
3. Summary
We have described a convenient method
whereby ketone groups can be introduced onto the
secondary hydroxyl rim of cyclodextrins using
aqueous bromine solution at neutral pH. The
ketone-derivatised cyclodextrins and an analogous
sucrose derivative are capable of catalysing the
peroxomonosulfate oxidation of sulfoxides and
phenols, with the formation of a dioxirane inter-
mediate being the most likely explanation for the
observed catalysis. Preliminary results indicate,
however, that it is the unbound form of the sub-
strate that is oxidised. Additionally, these com-
pounds are unstable and are subject to oxidation
by peroxomonosulfate. Nevertheless, there remains
a wide range of possible avenues for future
research on these compounds. For example, we
have only looked at a limited number of substrates
and it may well be the case that di€erent substrates
form inclusion compounds where the reactive sub-
strate group is in a more favourable orientation
with respect to the dioxirane moiety, allowing oxi-
dation of bound substrate. Enantioselective oxida-
tions could be carried out in this way.
[
[
41,42], has been widely reported in the literature
43]. In the case of phenol oxidation by dioxiranes
the authors did not analyse the products of the
Fig. 9. Absorbance changes at 400 nm for the reaction of
�
4
� 3
� 3
� 3
1
.0Â10 mol dm
p-nitrophenol and 1.9Â10 mol dm
We have extended this work to look at the for-
mation of keto-derivatised hydroxypropyl cyclo-
dextrins where the hydroxyls of the isopropyl
derivative are available to react with bromine in
addition to secondary hydroxyls of the cyclodextrin.
peroxomonosulfate in pH 7.8 phosphate bu€er, ionic strength
ꢀ
�
3
0
0
.2 mol dm , 25 C, solid line; and in the presence of
� 3
�
3
� 3
.01 mol dm ꢀ-cyclodextrin, dotted line; 3.8Â10 mol dm
�
3
� 3
keto-ꢀ-cyclodextrin, dashed line; 7.8Â10 mol dm
keto-
sucrose, dot-dot-dashed line; and 5Â10 mol dm cyclohex-
�
3
� 3
anone, dot-dashed line.

M.E. Deary, D.M. Davies/Carbohydrate Research 309 (1998) 17±29
27
The preparation of these compounds and their
a€ect on a variety of oxidation reactions will be
described in a subsequent paper.
280 to 320 nnm) using a HP-8450 diode array
spectrophotometer with a ¯ow cell connected to
the column outlet. No further puri®cation was
carried out. Variations to the above procedure
involved changes to the concentrations of reactants
and are as reported in the text. In particular, sev-
eral experiments were conducted in which the bro-
mine oxidation was carried out in bu€er in order to
monitor the UV/vis spectral changes during the
oxidation. Carbonate and phosphate bu€ers were
used in these cases and the reaction was followed
on a Pharmacia Biotech Ultraspec 2000 Spectro-
photometer.
4
. Experimental
Materials.Ð2,6-di-O-Methyl-ꢁ-cyclodextrin was
obtained from Cyclodextrin Technology Develop-
ment Inc., FL and was 99% pure. All other mate-
rials were obtained from Aldrich and were the best
available grades. Methyl-ꢁ-cyclodextrin, which was
obtained from Aldrich, is a randomly O-methy-
lated derivative with an average of 1.8 methyl
groups per glucose residue. Bromine solutions
were prepared in distilled water and standardised
iodometrically.
13C NMR.ÐSpectra were obtained in distilled
water at pH 6.5±7.5 in 5 mm o.d. sample tubes
using a JEOL EX 270 MHz NMR spectrometer at
an operating frequency of 67.8 MHz. Sodium acet-
Preparation of keto-derivatised cyclodextrins and
sucrose.ÐThese materials were prepared by react-
ate (ꢂ 182.61 in H O [45]) was used as an internal
2
standard. The solutions containing the bromine
oxidation products were reduced down to one
tenth of their original volume using a vacuum oven
�
3
� 3
ing approximately 30Â10 mol dm
bromine
cyclodextrin or
sucrose, whilst maintaining the pH at 7.0 using a
�
3
� 3
solution with 18Â10 mol dm
ꢀ
at 40 C, giving solutions ca. 160±180 mM in oxi-
�
3
pH-stat dispensing 0.5 mol dm NaOH. The reac-
tion was carried out for approximately 24 h during
which time approximately 12 mL NaOH was
added. The reaction was started by adding the
bromine solution to cyclodextrin or sucrose in
powder form. After completion of the reaction,
bromide, a product of the oxidation, was removed
from the solution in order to prevent interference
with subsequent kinetic experiments involving the
keto-sugars and peroxomonosulfate. This was
accomplished by passing the solution through an
Amberlite IRA-743 ion exchange resin. The resin
dised cyclodextrin. Nevertheless, typical acquisi-
tions required between 30,000 to 50,000 scans in
order to see detail in the carbonyl region. No bro-
mide removal was carried out when the oxidised
cyclodextrins were prepared purely for NMR ana-
lysis. The pH of the solutions was approximately
6.5 to 7.5 and so the standard exists predominantly
in the anionic form for which complexation with
cyclodextrin is negligible [36].
Reactions involving the keto-derivatives.ÐOxida-
tions of S-(-)-methyl-p-tolyl sulfoxide, 4-bromo-
phenyl methyl sulfoxide and p-nitrophenol were
carried out in phosphate bu€er at pH 7.6 and ionic
�
3
had previously been equilibrated in 2 mol dm
NaNO in order remove chloride ions, which can
�
3
strength 0.2 mol dm . The substrate and perox-
omonosulfate concentrations are given in the
results tables. The reactions were carried out at
3
react with peroxides, that were originally present as
the exchange ion. The resin was then washed sev-
eral times in distilled water before packing in an
ꢀ
25 C and followed at 244 and 250 nm for (S)-(-)-
8
0 mL glass column. The bromide concentration at
methyl-p-tolyl sulfoxide and 4-bromophenyl
methyl sulfoxide, respectively, and at 400 nm for p-
nitrophenol on a Pharmacia Biotech Ultraspec
2000 spectrophotometer with thermostatted cell-
holder. In the case of the sulfoxides, which absorb
in a similar region to the keto-derivatives, it was
necessary to run a blank in which no substrate was
present. This was subtracted from the correspond-
ing run with substrate present in order to account
for the decrease in absorbance due to the keto-
derivatives as they react with peroxomonosulfate.
Con®rmation that the sulfoxides were oxidised to
sulfones in all cases was obtained by comparing the
the column outlet was monitored by reacting resi-
dual bromide with 10Â10� 3 mol dm� 3 perox-
omonosulfate in weak sufuric acid and measuring
the absorbance due to bromine at 398 nm for
which the molar absorbtivity is 86 dm mol cm .
Over 99.9% removal eciency was achieved using
the column with a residence time of 15 min. It
should be noted that the bromide concentration
was evaluated at low pH where the bromine oxi-
dation of cyclodextrin and sucrose is very slow.
The fraction containing the oxidised compound
was monitored at the peak absorbance (typically
3
� 1
� 1

28
M.E. Deary, D.M. Davies/Carbohydrate Research 309 (1998) 17±29
product spectra with spectra for the corresponding
sulfones [35]. We did not attempt to identify the
products of p-nitrophenol oxidation for the rea-
sons given in the results section. Initial rates were
calculated for the sulfoxide oxidation experiments
using the initial 200 s of data where the plot was
linear. Peroxomonosulfate concentrations for the
oxidation experiments were standardised iodome-
trically using a thiosulfate titration. The change in
peroxomonosulfate concentration during reaction
with the derivatised ꢀ-cyclodextrin was monitored
using a standard spectrophotometric iodometric
technique.
[13] O. Larm, K. Larsson, E. Scholander, B. Meyer,
and J. Thiem, Carbohydr. Res., 91 (1981) 13.
[
[
[
[
[
14] J. Kozlowski and P. Zuman, Bioelectrochem. and
Bioenergetics, 28 (1992) 43.
15] H-J. Buschman, H.-H. Fuldner, and W. Knoche.,
È
Ber. Bunsenges. Phys. Chem., 85 (1980) 41.
16] J.M. Riordan, Philip E. Morris, and D.E. Kiely, J.
Carbohydrate Chemistry, 12 (1993) 865.
17] N.E. Mackenzie, P.G. Malthouse, and A.I. Scott,
Science, 225 (1984) 883.
18] M.A. Mortellaro, S. Hong, D.T. Winn, and A.W.
Czarnik, Bioorg. Med. Chem. Lett., 4 (1994) 2041.
[19] O.S. Tee and J.M. Bennett, J. Am. Chem. Soc., 110
(1988) 269.
20] O.S. Tee and B.C. Javed, J. Chem. Soc., Perkin
Trans. 2, (1994) 23.
21] A.C. Besemer and H. van Bekkum, Recuil des
Travaux Chimiques des Pays-Bas, 113 (1994) 399.
22] A.C. Besemer and H. van Bekkum, Starch, 46
[
[
[
[
[
Acknowledgements
We are grateful to the Engineering and Physical
Science Research Council for ®nancial (ROPA)
support: grant reference GR/K34535. We thank Dr
Jonathan Bookham and Mr David McEvoy for
their invaluable help with the NMR analysis.
(
1994) 101.
23] A.C. Besemer and H. van Bekkum, Starch, 46
1994) 95.
(
24] D.R. Lide (Ed.), CRC Handbook of Chemistry and
Physics, 73rd ed., CRC Press, Boca Raton, FL,
1
25] J. Palou, Chemical Society Reviews, (1994) 357.
992.
[
References
[26] D.M. Davies and P. Jones, J. Society of Dyers and
Colourists, 98 (1983) 17.
[
1] D.M. Davies, G.A. Garner, and J.R. Savage, J.
Chem, Soc., Perkin Trans. 2, (1994) 1531.
2] O. Theander, in W. Pigman and D. Horton (Eds.),
The Carbohydrates, 2nd ed., Vol. 1B, Academic
Press, New York, 1989.
[27] M.L. Bender and M. Komiyama, Cyclodextrin
Chemistry, Springer Verlag, Berlin, 1978.
[28] C.G. Swain, R.A. Wiles, and R.F.W. Bader, J. Am.
Chem. Soc., 83 (1961) 1945.
[29] I.R.L. Barker, W.G. Overend, and C.W. Rees, J.
Chem. Soc., (1964) 3263.
[30] N. Venkatasubramian and V. Thiagarajan, Tetra-
hedron Lett., (1968) 1711.
[31] N.C. Deno and N.H. Potter, J. Am. Chem. Soc., 89
(1967) 3555.
[32] R. Hiatt in D. Swern (Ed.), Organic Peroxides, Vol
2., Wiley Interscience, New York, 1971.
[33] R.E. Montgomery, J. Am. Chem. Soc., 96 (1974)
7820.
[34] J.O. Edwards, R.H. Pater, R. Curci, and F. Di
Furia, Photochem. Photobiol., 30 (1979) 63.
[35] D.M. Davies and M.E. Deary, J. Chem. Soc., Per-
kin Trans. 2, (1995) 1287.
[
[
3] R. Andersson, O. Larm, E. Scholander, and O.
Theander, Carbohydr. Res., 78 (1980) 257.
4] O. Larm, E. Scholander, and O. Theander, Carbo-
hydr. Res., 49 (1976) 69.
[
[
[
5] E. Scholander, Carbohydr. Res., 73 (1979) 302.
6] W.M. Doane and R.L. Whistler, Starch, (1964)
1
77.
7] I. Ziderman and J. Bel-Ache, Carbohydr. Res., 27
1973) 341.
[
[
[
(
8] L.J. Torneport, B.A-C. Salomonsson, and O.
Theander, Starch, 42 (1990) 413.
9] O. Theander, Ad. Carb. Chem., 17 (1962) 223.
[
10] For reviews of dioxirane chemistry see: (a) W.A.
Adam, R. Curci, and J.O. Edwards, Acc. Chem.
Res., 22 (1989) 205; (b) R.W. Murray, Chem. Rev.,
[36] D.M. Davies and M.E. Deary, J. Chem. Soc., Per-
kin Trans. 2, (1996) 2415.
8
9 (1989) 1187.
[37] D.M. Davies and M.E. Deary, J. Chem. Soc., Per-
kin Trans. 2, (1996) 2423.
[38] J.E. Sawicki and B. Casas, Environ. Progress, 12
(1993) 275.
[
11] A.-C.B. Salomonsson, R.E. Anderson, L.J. Torne-
port, and O. Theander, Carbohydr. Res., 217 (1991)
2
21.
[
12] K.A. Martin, M.A. Mortellaro, R.W. Sweger, L.E.
Fikes, D.T. Winn, S. Clary, M.P. Johnson, and
A.W. Czarnik, J. Am. Chem. Soc., 117 (1995)
[39] M.S. Dieckmann and K.A. Gray, Water Research,
30 (1996) 1169.
[40] C.J. Martino and P.E. Savage, Ind. Eng. Chem.
Res., 36 (1997) 1385.
10443.

M.E. Deary, D.M. Davies/Carbohydrate Research 309 (1998) 17±29
29
[
41] A. Altamura, C. Fusco, L. D'Accolti, R. Mello, T.
Prencipe, and R. Curci, Tetrahedron Lett., 32
(Ed.), The Chemistry of Functional Groups, Hydro-
xyl Group, Wiley, New York, 1971.
[44] A. Pintar and J. Levec, Chem. Eng. Sci., 49 (1994)
4391.
[45] D.L. Rabenstein and T.L. Sayer, J. Magnetic
Resonance, 24 (1976) 27.
(
1991) 5445.
[
42] J.K. Crandall, M. Zucco, R.S. Kirsch, and D.M.
Coppert, Tetrahedron Lett., 32 (1991) 5441.
[
43] M. Lj. Mihailovic and Z. Cekovic, in S. Patai