Acid-Mediated Formation of Radicals or Baeyer-Villiger Oxidation from Criegee Adducts

Article abstract of DOI:10.1002/anie.201505648

The acid-mediated reaction of ketones with hydroperoxides generates radicals, a process with reaction conditions similar to those of the Baeyer-Villiger oxidation but with an outcome resembling the formation of hydroxyl radicals via ozonolysis in the atmosphere. The Baeyer-Villiger oxidation forms esters from ketones, with the preferred use of peracids. In contrast, alkyl hydroperoxides and hydrogen peroxide react with ketones by condensation to form alkenyl peroxides, which rapidly undergo homolytic O-O bond cleavage to form radicals. Both reactions are believed to proceed via Criegee adducts, but the electronic nature of the peroxide residue determines the subsequent reaction pathways. DFT calculations and experimental results support the idea that, unlike previously assumed, the Baeyer-Villiger reaction is not intrinsically difficult with alkyl hydroperoxides and hydrogen peroxide but rather that the alternative radical formation is increasingly favored. At the crossroads: Criegee adducts of hydroperoxides and ketones are key intermediates in the Baeyer-Villiger oxidation, but they can also generate radicals via formation of alkenyl peroxides. The fate of the Criegee adduct is determined by the electronic nature of the peroxide: peracids favor rearrangement, alkylhydroperoxides favor radical formation. Hydrogen peroxide is equally suitable for both pathways, explaining its poorer performances in the two reactions.

Full text of DOI:10.1002/anie.201505648

.
Angewandte
Communications
International Edition: DOI: 10.1002/anie.201505648
German Edition: DOI: 10.1002/ange.201505648
Reaction Mechanisms
Acid-Mediated Formation of Radicals or Baeyer–Villiger Oxidation
from Criegee Adducts
Bertrand Schweitzer-Chaput, Theo KurtØn,* and Martin Klussmann*
Abstract: The acid-mediated reaction of ketones with hydro-
peroxides generates radicals, a process with reaction conditions
similar to those of the Baeyer–Villiger oxidation but with an
outcome resembling the formation of hydroxyl radicals via
ozonolysis in the atmosphere. The Baeyer–Villiger oxidation
forms esters from ketones, with the preferred use of peracids. In
contrast, alkyl hydroperoxides and hydrogen peroxide react
with ketones by condensation to form alkenyl peroxides, which
À
rapidly undergo homolytic O O bond cleavage to form
radicals. Both reactions are believed to proceed via Criegee
adducts, but the electronic nature of the peroxide residue
determines the subsequent reaction pathways. DFT calcula-
tions and experimental results support the idea that, unlike
previously assumed, the Baeyer–Villiger reaction is not
intrinsically difficult with alkyl hydroperoxides and hydrogen
peroxide but rather that the alternative radical formation is
increasingly favored.
Scheme 1. Distinct pathways for the Baeyer–Villiger oxidation and
radical formation via Criegee adducts.
T
he Baeyer–Villiger oxidation is an important chemical
reaction that converts ketones into esters and lactones.[1]
Peracids and hydrogen peroxide under acidic conditions are
the oxidants of choice but the latter is usually less efficient
although economically and environmentally preferable. We
have recently found that under strikingly similar conditions,
the action of an alkylhydroperoxide or hydrogen peroxide
and a strong Brønsted acid catalyst on a ketone opens
a previously overlooked reaction pathway to generate
radicals. Here we provide further mechanistic details and
show how the electronic nature of the peroxide influences the
outcome of the reaction.
We had postulated[3] that the mechanism of the acid-
mediated radical formation involves Criegee adducts as well,
but instead of the Baeyer–Villiger rearrangement taking
place, the acid catalyzes elimination of water via peroxycar-
benium ion[4] 3 to form an alkenyl peroxide 4 (Scheme 1). This
highly reactive species rapidly fragments into radicals 5 and 6,
which can undergo addition reactions to olefins[3] and mediate
C–H functionalization reactions.[5]
Interestingly, the formation and reactivity of alkenyl
peroxides has been studied extensively by theoretical chem-
ists in the context of atmospheric chemistry, in contrast to the
chemistry of these compounds in solution. Gas-phase ozo-
nolysis of alkenes leads to Criegee carbonyl oxide intermedi-
ates that can rearrange to alkenyl hydroperoxides, which
rapidly decompose into resonance-stabilized carbonyl and
hydroxyl radicals.[6] Ozonolysis of biogenic and other alkenes
is thus believed to be the major source of atmospheric
hydroxyl radicals at night.[7]
In order to improve or expand both the Baeyer–Villiger
oxidation and the radical formation, one has to better
understand their commonalities and differences. At first we
looked for further indications that the radical formation
indeed takes place as suggested. As a model reaction, we
chose the previously reported oxidative functionalization of
styrene with cyclohexanone and tert-butyl hydroperoxide
(tBuOOH), catalyzed by p-toluenesulfonic acid (pTsOH)
(Scheme 2a).[3] On closer inspection of the resulting reaction
mixture, we detected the a-oxygenated ketones 8a and 8b as
byproducts next to the main product 7. The oxygenated 8a
was formed in around 10% yield, irrespective of whether an
aqueous or an organic solution of tBuOOH was used, while
The mechanism of the Baeyer–Villiger oxidation is
generally accepted to involve two distinct steps, both of
which can be catalyzed by acids (Scheme 1).[1b] The first step is
the addition of the peroxide to the carbonyl compound to
afford a tetrahedral intermediate 1, known as a Criegee
adduct (not to be confused with the Criegee carbonyl oxide
intermediate involved in the ozonolysis of olefins[2]). In the
second step, rearrangement of the Criegee adduct affords the
desired ester 2.
[*] Dr. B. Schweitzer-Chaput, Dr. M. Klussmann
Max-Planck-Institut für Kohlenforschung
Kaiser-Wilhelm-Platz 1, 45740 Mülheim an der Ruhr (Germany)
E-mail: klusi@mpi-muelheim.mpg.de
Dr. T. KurtØn
Laboratory of Physical Chemistry, University of Helsinki
P.O. BOX 55, 00014 Helsinki (Finland)
E-mail: theo.kurten@helsinki.fi
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201505648.
11848
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2015, 54, 11848 –11851

Angewandte
Chemie
These experimental findings clearly support the mecha-
nism proposed in Scheme 1, but the question regarding the
difference between Baeyer–Villiger oxidation and radical
formation remains. We therefore evaluated the postulated
mechanism using density functional theory (DFT) calcula-
tions of a simple model system derived from acetone. At first
À
we compared the O O bond dissociation enthalpies (BDE) of
several postulated peroxide intermediates. For comparison,
we looked at compounds derived from tBuOOH, peracetic
acid (as a simple model for peracids in general), and hydrogen
peroxide. As can be seen in Figure 1, tBuOOH (12), bisper-
Scheme 2. a) Product distribution in the addition reaction to styrene,
dependent on the oxidant used; yields of 8 are relative to cyclo-
hexanone. b) Product distribution in the acid-catalyzed reaction of
cyclohexanone with tBuOOH, effect of radical inhibitor. BHT=
bis(2,6-tert-butyl)-4-methylphenol.
the peroxygenated 8b was formed in around half the amount.
Di-tert-butylperoxide did not mediate the reaction, support-
ing the premise that a condensation reaction to an alkenyl
peroxide is necessary.
In the absence of styrene, the reaction leads to the
formation of 8a and 8b in slightly larger amounts, with the
latter now as the major compound (Scheme 2b). When one
equivalent of the radical inhibitor BHT was added, 8a was
formed in essentially unchanged amount while formation of
8b was completely suppressed. The formation of 8a can be
rationalized by geminate recombination of radical pair 10,
formed by decomposition of alkenylperoxide 9. Escape from
the solvent cage, fast hydrogen atom transfer (HAT) between
the oxyl radical and excess tBuOOH,[8] and subsequent
recombination of the ketone with the peroxyl radical 11 will
lead to 8b (Scheme 3). This coupling of two radicals, present
Figure 1. O–O bond dissociation enthalpies (BDE) of peroxide species
potentially formed under the reaction conditions; calculated in vacuum
using the wB97XD functional with the aug-cc-pVTZ basis set.
oxides 13a and 13b, as well as Criegee adducts 14a and 14b,
all have BDEs in the 33 to 42 kcalmolÀ1 range, in accordance
with known values for peroxide bonds.[11] By contrast, the
alkenyl peroxides 15a–c were considerably less stable, with
BDE values between 6 and 17 kcalmolÀ1. This is in accord-
ance with previous studies of related compounds, which also
suggested that alkenyl, alkynyl, and aryl peroxides generally
do not exist as stable species.[6a,b,d,12] Knowing that tBuOOH
requires temperatures above 1008C to decompose at syntheti-
cally useful rates, we can safely assume that products 13 and
14 require similarly elevated temperatures for relevant rates
[13]
À
of homolytic O O bond cleavage.
Interestingly, the acyl-substituted 15c was found to be the
least stable species within the series of alkenyl peroxides
evaluated, with an extremely low BDE value of 6.2 kcalmolÀ1
(see Figure 1). As peracids are the oxidants of choice in
Baeyer–Villiger reactions, the formation of 15c cannot be
significant under reaction conditions. We therefore compared
the two reaction pathways for the three different oxidants
(tBuOOH, H2O2 and AcOOH, Figure 2).
We used Criegee adducts 14 (plus methane sulfonic acid)
as references and starting points. As can be seen, the Baeyer–
Villiger products 16 are thermodynamically more favored
than the alkenyl peroxides by about 70 kcalmolÀ1, for all cases
evaluated. The decisive factor must therefore be the kinetic
barrier for each product. We first calculated the transition
states leading to Baeyer–Villiger products (TS1) and found
them to be 14.2 kcalmolÀ1 above the reactants for 14a,
15.5 kcalmolÀ1 for 14b, and 13 kcalmolÀ1 for 14c. As
Scheme 3. Suggested formation of 8a and 8b.
in only low concentrations, might appear unlikely. However,
tertiary peroxyl radicals are known to be stabilized and
relatively unreactive, and dimerize relatively slowly, while
their reaction with alkyl radicals is very fast.[9]
This model can explain the results shown in Scheme 2. In
the presence of styrene, 8a is formed in larger amounts than
8b, possibly because the peroxyl radical formed after cage
escape and HAT is largely consumed by addition to styrene.[10]
In its absence, 8b is the major product of the two. The
addition of BHT quenches all radicals that escape their
solvent cage, but the geminate recombination of 10 within the
cage is not affected.
Angew. Chem. Int. Ed. 2015, 54, 11848 –11851
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 11849

.
Angewandte
Communications
Scheme 4. Comparison of oxidants for the Baeyer–Villiger oxidation.
1
Yields were determined by H NMR analysis of the reaction mixture
using an internal standard.
trace of caprolactone in the oxidation of cyclohexanone,
consistent with our previous experimental results;[3,5] instead
8a was detected in usual amounts (10%). However, bicyclic
adamantanone, which cannot form an internal double bond,
cleanly underwent rearrangement to lactone 18 with both
peroxides. The efficiency of tBuOOH was only slightly lower
than with mCPBA, in line with the small energy differences
presented in Figure 2. This strongly supports the fact that the
Baeyer–Villiger reaction is not intrinsically problematic with
hydrogen peroxide or tBuOOH, but rather that a competing
pathway—formation of alkenyl peroxides—is increasingly
favored with these oxidants.
In summary, we have found experimental evidence for the
intermediacy of alkenyl peroxides in the acid-mediated
formation of radicals from ketones and hydroperoxides. The
mechanism helps to rationalize previous reports of radical
polymerization or hydrogen peroxide decomposition that
lacked a satisfying explanation.[16] Alkenyl peroxides could
also be utilized in new synthetic applications and might have
implications for industrial autoxidation processes, where the
necessary starting materials are formed from hydrocarbons.[17]
Finally, this chemistry could play a role in organic aerosol
particles containing carbonyl compounds, hydroperoxides,
and acid.[18]
Figure 2. Comparison of Baeyer–Villiger rearrangement and formation
of alkenyl peroxides from Criegee adducts 14, calculated using the
wB97XD functional with the 6-31++G(d,p) basis set for geometries
and frequencies and the aug-cc-pV(T+d)Z basis set for energies, and
the IEFPCM solvent model (with acetone as a solvent). Where not
shown, the total enthalpies include the enthalpy of one molecule of
methanesulfonic acid.
expected, the rearrangement is easiest for the acyl-substituted
Criegee adduct and more difficult for the adducts with the two
other substituents.[14]
When evaluating the alternative pathway of alkenyl
peroxide formation, we observed an opposite trend. We
found an acid-catalyzed concerted E2-type pathway from the
Criegee adducts 14 via transition state TS2 to the alkenyl
peroxides 15, formed as a hydrogen-bonded product complex
15* (see the Supporting Information for more details). This
pathway is easiest for 14a (11.2 kcalmolÀ1), followed by 14b
(14.4 kcalmolÀ1), and most unfavorable for 14c (18.6 kcal
molÀ1). Comparing the two different pathways for each
substituent perfectly explains the selectivities observed
experimentally. In the case of peracetic acid, TS1c is lower
in energy than TS2c by 5.6 kcalmolÀ1. Starting from tBu-
substituted 14a, TS2a is lower than TS1a by 3 kcalmolÀ1. For
these two cases, the selectivity for one pathway or the other is
therefore very clear: peracids form Baeyer–Villiger products
while hydroperoxides form alkenyl peroxides. In the case of
hydrogen peroxide, the energy difference between TS1b and
TS2b is much smaller (1.1 kcalmolÀ1 in favor of TS2b) and
suggests that both pathways contribute to a similar extent.
One interesting outcome of these calculations is that for
all three peroxides, the barriers to Baeyer–Villiger rearrange-
ment are quite close to each other (within 2.5 kcalmolÀ1).[14]
Most strikingly, the rearrangement is actually easier for
tBuOOH than for H2O2, by 1.3 kcalmolÀ1. This is in contra-
diction with what would be expected from the Baeyer–
Villiger literature, in which tBuOOH is virtually never
reported as an oxidant,[15] in contrast to hydrogen peroxide.[1]
In order to gain further experimental evidence in support of
this phenomenon, we subjected two different ketones to
reactions with two different oxidants (Scheme 4).
In addition, we have rationalized the relationship between
the acid-mediated radical formation from ketones and hydro-
peroxides and the Baeyer–Villiger oxidation. Both reactions
proceed via Criegee adducts, but the electronic effects of the
peroxides determine the ensuing pathways: peracids prefer
the Baeyer–Villiger oxidation, hydroperoxides the radical
formation, while for hydrogen peroxide, both are similarly
accessible.
Acknowledgements
We thank the DFG (KL 2221/3-1 and Heisenberg scholarship
to M.K., KL 2221/4-1), the MPI für Kohlenforschung (post-
doctoral scholarship to B.S.), and the Academy of Finland for
funding.
Keywords: Baeyer–Villiger oxidation · Brønsted acid catalysis ·
ketones · peroxides · radical reactions
As expected, meta-chloroperbenzoic acid (mCPBA) was
very efficient in delivering the corresponding lactones from
cyclohexanone and adamantanone (17). tBuOOH gave no
How to cite: Angew. Chem. Int. Ed. 2015, 54, 11848–11851
Angew. Chem. 2015, 127, 12015–12018
11850 www.angewandte.org
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2015, 54, 11848 –11851

Angewandte
Chemie
[12] Probably the only known exception is a heteroaryl peroxide
[1] a) M. Renz, B. Meunier, Eur. J. Org. Chem. 1999, 737 – 750;
b) G. J. ten Brink, I. W. C. E. Arends, R. A. Sheldon, Chem. Rev.
2004, 104, 4105 – 4124.
[2] R. Criegee, Angew. Chem. Int. Ed. Engl. 1975, 14, 745; Angew.
Chem. 1975, 87, 765.
[3] B. Schweitzer-Chaput, J. Demaerel, H. Engler, M. Klussmann,
Angew. Chem. Int. Ed. 2014, 53, 8737; Angew. Chem. 2014, 126,
8882.
[4] P. H. Dussault, I. Q. Lee, J. Am. Chem. Soc. 1993, 115, 6458.
[5] B. Schweitzer-Chaput, A. Sud, . Pinter, S. Dehn, P. Schulze, M.
Klussmann, Angew. Chem. Int. Ed. 2013, 52, 13228; Angew.
Chem. 2013, 125, 13470.
[6] a) R. Gutbrod, R. N. Schindler, E. Kraka, D. Cremer, Chem.
Phys. Lett. 1996, 252, 221; b) N. Sebbar, J. W. Bozzelli, H.
Bockhorn, J. Phys. Chem. A 2004, 108, 8353; c) L. Vereecken,
J. S. Francisco, Chem. Soc. Rev. 2012, 41, 6259; d) T. KurtØn,
N. M. Donahue, J. Phys. Chem. A 2012, 116, 6823; e) F. Liu, J. M.
Beames, A. S. Petit, A. B. McCoy, M. I. Lester, Science 2014, 345,
1596.
[7] D. Stone, L. K. Whalley, D. E. Heard, Chem. Soc. Rev. 2012, 41,
6348.
[8] a) H. Paul, R. D. Small, J. C. Scaiano, J. Am. Chem. Soc. 1978,
100, 4520; b) D. V. Avila, K. U. Ingold, J. Lusztyk, W. H. Green,
D. R. Procopio, J. Am. Chem. Soc. 1995, 117, 2929.
[9] K. U. Ingold, Acc. Chem. Res. 1969, 2, 1.
which was reported to decompose over a period of one month at
À108C. See: H. Kropf, M. Ball, Liebigs Ann. Chem. 1976, 2331.
[13] K. Matsuyama, T. Sugiura, Y. Minoshima, J. Org. Chem. 1995,
60, 5520.
[14] Somewhat different values were reported for Baeyer– Villiger
rearrangements using AcOOH and MeOOH, albeit catalysed by
a carboxylic acid. See: R. D. Bach, J. Org. Chem. 2012, 77, 6801.
[15] B. Dutta, S. Jana, S. Bhunia, H. Honda, S. Koner, Appl. Catal. A
2010, 382, 90.
[16] a) E. Schmitz, O. Brede, J. Prakt. Chem. 1970, 312, 43; b) J.
Pavlinec, M. Lazµr, Collect. Czech. Chem. Commun. 1984, 49,
404; c) L. V. Petrov, V. M. Solyanikov, Russ. Chem. Bull. 1996,
45, 340; d) I. N. Grebenshchikov, A. S. Dykman, V. V. Pinson,
Russ. J. Org. Chem. 2008, 44, 617.
[17] I. Hermans, J. Peeters, P. A. Jacobs, Top. Catal. 2008, 50, 124.
[18] a) M. Ehn, J. A. Thornton, E. Kleist, M. Sipila, H. Junninen, I.
Pullinen, M. Springer, F. Rubach, R. Tillmann, B. Lee, F. Lopez-
Hilfiker, S. Andres, I.-H. Acir, M. Rissanen, T. Jokinen, S.
Schobesberger, J. Kangasluoma, J. Kontkanen, T. Nieminen, T.
Kurten, L. B. Nielsen, S. Jorgensen, H. G. Kjaergaard, M.
Canagaratna, M. D. Maso, T. Berndt, T. Petaja, A. Wahner, V.-
M. Kerminen, M. Kulmala, D. R. Worsnop, J. Wildt, T. F. Mentel,
Nature 2014, 506, 476; b) M. P. Rissanen, T. KurtØn, M. Sipilä,
J. A. Thornton, J. Kangasluoma, N. Sarnela, H. Junninen, S.
Jørgensen, S. Schallhart, M. K. Kajos, R. Taipale, M. Springer,
T. F. Mentel, T. Ruuskanen, T. Petäjä, D. R. Worsnop, H. G.
Kjaergaard, M. Ehn, J. Am. Chem. Soc. 2014, 136, 15596.
[10] J.-Q. Yu, E. J. Corey, Org. Lett. 2002, 4, 2727.
[11] Y.-R. Luo, J.-P. Cheng, Bond Dissociation Energies in CRC
Handbook of Chemistry and Physics, 95th ed. (Eds.: W. M.
Haynes, T. J. Bruno, D. R. Lide), 2015 Internet Version, http://
www.hbcpnetbase.com/.
Received: June 18, 2015
Published online: August 12, 2015
Angew. Chem. Int. Ed. 2015, 54, 11848 –11851
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 11851