Photosensitized oxygenation of some benzyl sulfides. The role of persulfoxide

Article abstract of DOI:10.1002/(SICI)1099-1395(199909)12:9<703::AID-POC178>3.0.CO;2-L

Benzyl ethyl sulfide (6a) is photo-oxidized to benzaldehyde in benzene, whereas diethyl sulfide is known to give inefficiently the sulfoxide under these conditions. Oxidative C-S cleavage is the main process also with benzhydryl ethyl sulfide (6c), but not with α-methylbenzyl ethyl sulfide (6b), which mainly gives the sulfoxide. The carbonyl derivatives reasonably arise from S-hydroperoxy ylides (3). Consistently with this finding, calculations at the PM3 level suggest that the first intermediate, the persulfoxide (1), undergoes intramolecular hydrogen transfer when an activated α-hydrogen is available and gives 3. This is the case for the above benzyl sulfides (ΔH* for the process decreases with decreasing C-H BDE). However, only some of the persulfoxide conformations are correctly oriented for this rearrangement, and this may slow this process and make other reactions compete, as happens with 6b. Copyright

Full text of DOI:10.1002/(SICI)1099-1395(199909)12:9<703::AID-POC178>3.0.CO;2-L

JOURNAL OF PHYSICAL ORGANIC CHEMISTRY
J. Phys. Org. Chem. 12, 703–707 (1999)
Photosensitized oxygenation of some benzyl sul®des. The
role of persulfoxide
Sergio M. Bonesi, Mauro Freccero and Angelo Albini*
Department of Organic Chemistry, University of Pavia, via Taramelli 10, 27100 Pavia, Italy
Received 3 November 1998; revised 26 April 1999; accepted 26 April 1999
ABSTRACT: Benzyl ethyl sulfide (6a) is photo-oxidized to benzaldehyde in benzene, whereas diethyl sulfide is
known to give inefficiently the sulfoxide under these conditions. Oxidative C—S cleavage is the main process also
with benzhydryl ethyl sulfide (6c), but not with a-methylbenzyl ethyl sulfide (6b), which mainly gives the sulfoxide.
The carbonyl derivatives reasonably arise from S-hydroperoxy ylides (3). Consistently with this finding, calculations
at the PM3 level suggest that the first intermediate, the persulfoxide (1), undergoes intramolecular hydrogen transfer
when an activated a-hydrogen is available and gives 3. This is the case for the above benzyl sulfides (DH≠ for the
process decreases with decreasing C—H BDE). However, only some of the persulfoxide conformations are correctly
oriented for this rearrangement, and this may slow this process and make other reactions compete, as happens with 6b.
Copyright 1999 John Wiley & Sons, Ltd.
KEYWORDS: single oxygen; sulfides; photochemistry; oxidation
INTRODUCTION
can be trapped with sulfides (to give sulfoxides) and
possibly it rearranges to a sulfone. Recent computations
by Jensen et al.2 and McKee3 do not support the inter-
mediacy of 2, while suggesting that another rearrange-
ment, intramolecular hydrogen transfer to give the
hydroperoxy ylide 3, is possible (path c). Calculations
show that further rearrangement of 3 to the sulfone is
viable,2,3 it has been considered that 3 may be involved in
oxygen transfer to sulfides (viz. that it is the second
intermediate as proposed by Foote), but this has found no
clear support.2
Further processes attributed to the persulfoxide are
intermolecular reactions, viz addition of alcohols to give
a hydroperoxyalkoxy sulfide (4, path d), an important
reaction because this is what makes the photo-oxidation
efficient in alcohols, and oxygen transfer to sulfoxides
(path e), a mechanistically interesting process which
allows recognition of the nucleophilic character of this
species (as opposed to the electrophilic second inter-
mediate).1
The photosensitized oxygenation of sulfides is currently
rationalized as involving an adduct, the persulfoxide 1, as
the first-formed intermediate.1 This is a weakly bound
adduct, and the main (ꢀ95%) fate of this species in
aprotic media is chemically unproductive decay to the
ground-state components (Scheme 1, path a), so that
under these conditions sulfides are essentially physical
1
quenchers of O2. On the basis of a thorough kinetic
analysis, Foote1d suggested that a small fraction of 1
undergoes isomerization to a further intermediate, for
which a possible structure was that of thiadioxirane 2
(path b). The second intermediate is an electrophile; it
Although there is a general consensus on the role of 1,
there is no direct identification of such an intermediate or
of a rearranged isomer. A way to tackle this problem is to
study a chemical process directly related to these
intermediates. This is difficult with dialkyl sulfides, since
in aprotic media the photooxygenation is very inefficient
and mainly gives the sulfoxide. This is not mechan-
istically indicative since, as mentioned above, it is not
clear whether this may arise directly from 1 or 3, while it
is formed much more efficiently in protic solvents, where
a secondary intermediate, presumably adduct 4, is
involved. Products different from sulfoxides have been
Scheme 1
*Correspondence to: Angelo Albini, Department of Organic
Chemistry, University of Pavia, via Taramelli 10, 27100 Pavia, Italy.
Contract/grant sponsor: CNR.
Contract/grant sponsor: MURST.
Copyright 1999 John Wiley & Sons, Ltd.
CCC 0894–3230/99/090703–05 $17.50

704
A. ALBINI ET AL.
Table 1. Products from the tetraphenylporphine photo-
sensitized oxidation of benzyl sul®des in benzene
Product yield (%)
Scheme 2
Substrate
R
Relative reactivity
7
8
9
6a
6b
6c
H
Me
Ph
3.5
1.8
2.8
1
80
10
60
3
70
9
8
5
22
reported for some thiazolidines, which give the corre-
sponding a-hydroperoxides (see formula 5).4 Calcula-
tions support that this arises by shift of the OOH group
from ylide 3.2,3 A related reaction has been reported for
some benzyl sulfides, where the products actually
obtained are aldehydes, presumably resulting from the
fragmentation of 5 (Scheme 1).5,6
The mechanism of benzyl sulfide photo-oxidation has
received little experimental study and has not as yet been
considered computationally. We thought that a detailed
investigation was a unique opportunity to understand the
reactivity of the persulfoxide, since the cleavage to
aldehydes appears to depend on the rearrangement of this
intermediate and should provide information on its
structure. Therefore, we began the study of this reaction
through a kinetic analysis of the oxidation of benzyl ethyl
sulfide.7 A further way of exploring the mechanism could
be to test the sensitivity of the reaction to substitution at
the key a-position. Here we report the photo-oxygenation
of three benzyl sulfides and compare the experimental
results with a semiempirical computational study energy
surface of the corresponding persulfoxides.
Et2S
Et2SO (90%)
1 the yields obtained at ca 25% conversion are reported
(see Experimental). These were measured by gas
chromatography (GC) when the products were stable to
this technique (this is not the case for sulfoxide 8b; see
1
below and Experimental) and H NMR spectroscopy.
The irradiation of the sulfides was carried out under the
same conditions for a different time until a ca 25%
conversion was reached. In this way, a reactivity order
was established. As an example, Table 1 shows that the
oxidation of 6a was much more efficient than that of
diethyl sulfide under the same condition.
The TPP photo-oxidation of a-methylbenzyl ethyl
sulfide (6b) in benzene gave only a small amount of
acetophenone (7b), the main product being by far
sulfoxide 8b, accompanied by some sulfone 9b. With
benzhydryl ethyl sulfide (6c) the oxidative cleavage was
again predominant, with benzophenone (7c) as the main
product, a considerable amount of sulfone (9c) and a
small amount of sulfoxide (8c).
The efficiency of the photo-oxidation for the sulfides
tested decreased in the order 6a > 6c > 6b > Et2S. Thus,
the introduction of a phenyl group a to the sulfur atom
makes photo-oxidation of alkyl sulfides more efficient,
while directing it towards oxidative C—S bond frag-
mentation rather than to the sulfoxidation consistently
observed with dialkyl sulfides. However, introducing a
further a-substituent again affects both efficiency and
competition between the above two processes in a non-
intuitively obvious way: a second a-phenyl simply slows
the reaction, but with an a-methyl slowing is more
marked and the main product is the sulfoxide.
RESULTS
Photoreactions
Tetraphenylporphine (TPP) photosensitized oxygenation
of benzyl ethyl sulfide (6a) in benzene mainly gave
benzaldehyde (7a) with virtually no sulfoxide (8a) and
only a little sulfone (9a) at the beginning of the reaction.
The proportion of the sulfoxide grew somewhat during
the course of the reaction, but did not alter the
predominance of benzaldehyde (see Scheme 2). In Table
Table 2. Conformations of the persulfoxide 1 and activation energy for the transition state leading
to 3 (see Scheme 3)
1
1
dO—H (A) TS, DH≠ (kcal mol
)
dO—H (A)
˚
˚
Conformer R'
R''
R''' 1, DHf (kcal mol
)
1da
1a a
b
H
H
H
Me
Et
4.68
31.40
30.07
28.88
28.12
27.98
27.71
64.37
63.20
3.15
2.90
3.21
4.57
3.05
3.10
3.74
3.01
4.72
19.1
15.9
16.0
—
13.9
13.8
—
1.32
1.33
1.33
—
1.34
1.33
—
Ph
1b a
b
Ph Me
Et
g
ꢀ
1c a
Ph
Ph
Et
12.2
—
1.34
—
b
a
The energy data for 1d and the related TS reproduce well those previously reported by Ishiguro et al.8
Copyright 1999 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 12, 703–707 (1999)

PHOTOSENSITIZED OXYGENATION OF BENZYL SULFIDES
705
Owing to the lower symmetry, more stable conforma-
tions were expected for the currently considered benzyl
persulfoxides than for the dimethyl derivative. Indeed,
two minima were located for persulfoxide 1a (a and b in
Table 2). In both of them, the geometry (angle of the
O—O bond with respect to the sulfide moiety, dO—H) was
close to that calculated for the single minimum of 1a (see
Scheme 3, Table 2). For both conformers, conversion to
ylide 3a occurred smoothly via a transition state (TS)
where the hydrogen was equidistant between the oxygen
and the carbon atom, again similarly to the Me2S case,
but DH≠ was considerably smaller, 16 rather than
19 kcal mol 1 (see Table 2). As for ylide 3a, this showed
an S—CH distance of 1.60 A (compared with 1.84 A for
the S—CH2 distance in persulfoxide 1a) and the dihedral
angle CH2—S—C(Ph)—H was 160°. This evidence
supported the double bond nature of the S—CH bond.
The geometry of the persulfoxide (conformation a), the
corresponding transition state and the resulting ylide are
shown in Fig. 1 as an example.
Scheme 3
Computational analysis
The above-mentioned calculations by Jensen et al.2
[MP2/6–31G(d) and advanced levels] and McKee3
(B3LYP/6–31 ‡ G and advanced levels) on the
Me2S–1O2 system showed that ylide 3 is slightly
stablized with respect to persulfoxide 1 and is separated
by a barrier of only a few kcal mol 1. Benzyl substitution
was difficult to manage at this level of theory. However,
in 1996 Ishiguro et al.8 reported a study of the Me2S–1O2
system at the PM3 level and also found the rearrangement
to ylide 3 to be a viable path. Given the difficulty of using
ab initio methods, we thus felt it worthwhile to perform a
PM3 study on benzyl sulfides 6a–c, since this would give
at least qualitative information on the difference of such
substrates with respect to aliphatic sulfides.
˚
˚
We first carried out the calculation on dimethyl
persulfoxide (1d). The conversion to ylide 3d involved,
as reported by Ishiguro et al.8 a sizeable barrier (DH≠ =
19 kcal mol 1, see Table 2). This should be compared
1
with the 5.8–7.8 kcal mol obtained by CCSD(T) 6–31
G(d) or 6–3111 ‡ G(2df) methods2 and with the
1
9–12.6 kcal mol obtained with the B3LYP/a(‡2PC)
or the QCISD(T)/a methods.3 Apparently there is a
scaling factor of ca 2. However, we observed that the
minimum located by the PM3 computation corresponded
favourably with that found for the persulfoxide using
more advanced methods2,9 and had the O—O bond
A further decrease in symmetry led to four minima for
persulfoxide 1b and these are arranged in order of
increasing stability in Table 2. It can be seen that
conformers b and g were similar to those discussed above
˚
roughly bisecting the Me—S—Me angle and 3 A
˚
distance between the a hydrogen and the persulfoxide
oxygen (dO—H, see Table 2 and Scheme 3). Therefore, we
felt that any difference with substitution observed by the
PM3 method should be indicative of the reactivity order.
(dO—H ꢁ 3 A) and the TS for the rearrangement to the
ylide was lowered by a further 2 kcal mol 1. However,
conformations a and ꢀ, although of similar energy and
again with the O—O bond bisecting the R—S—R angle,
Figure 1. Geometries of 1a (a conformation), the TS and 3a
Copyright 1999 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 12, 703–707 (1999)

706
A. ALBINI ET AL.
had the outer oxygen closer to the a-methyl than to the a-
hydrogen (see Fig. 2). The search for a TS leading to 3b
gave no result in this case.
determinant. Thus, to the extent that such a distinction is
meaningful, the rearrangement of persulfoxide 1 to ylide
3 should be envisaged as radical hydrogen transfer rather
than proton transfer (peroxy radicals are known as poor
hydrogen abstractors; however, an intramolecular pro-
cess may be sufficiently fast) On the other hand,
persulfoxides, like in general mesomeric 1,3-dipoles,10
can be considered as diradical as well as charge-localized
zwitterions. Hence it is suggested that rearrangement to
ylide 3 is significant when a radical-stabilizing (phenyl)
group is present in the a position.
The introduction of a second a substituent further
decreases DH≠ for rearrangement, but on the other hand
increases the number of persulfoxide configurations, and
for some of these there is no accessible path to the ylide.
Since ab initio calculations (to which our PM3 data
should be scaled) suggest that the barrier for hydrogen
transfer is below 10 kcal mol 1, i.e. of the same order of
magnitude as rotation barriers, and collapse of the
persulfoxide to the components encounters in every case
a small barrier, it is expected that the presence of
conformations unable to rearrange significantly di-
minishes the overall efficiency of rearrangement. Hence
the two effects above operate in opposite directions.
This may explain the result with sulfide 6b, with which
the overall reaction is slower than with 6a and gives
much more sulfoxide, despite the fact that two out of four
conformations of persulfoxide 1b rearrange to the ylide
through a TS lying lower than those from 1a. We attribute
this fact to the overwhelming effect of adding two
roughly equally populated, but unproductive, conforma-
tions, which decreases the overall rate for hydrogen
transfer (path c) and increases the proportion of both
decay to the ground-state components (path a) and other
chemical processes leading to the sulfoxide.
Finally, two minima were located for persulfoxide 1c
(Table 2). Conformer a had a short O—H distance and
the DH≠ for conversion to the ylide was diminished by a
further 2 kcal mol 1 with respect to 1b, while the slightly
˚
less energetic conformer b had dO—H = 4.7 A and no
accessible TS for rearrangement to the ylide.
Previous calculations, both ab initio2,3 and at the PM3
level,8 showed the viability of the rearrangement of the
ylide 3d to the a-hydroperoxy sulfide (and hence to a
carbonyl derivative after C—S bond cleavage) or,
presumably through a higher barrier, to the sulfone (see
Scheme 1). In view of the multiplicity of paths, it
appeared that semiempirical calculations on the benzyl
derivatives could add little to this point and were
therefore not investigated further.
DISCUSSION
The present data show that the oxidation of benzyl sul-
fides in an aprotic solvent is considerably more efficient
than that of diethyl sulfide and finally leads to oxidative
C—S bond cleavage. The literature proposal that this
reaction is initiated by intramolecular hydrogen transfer
competing with persulfoxide dissociation to the compo-
nents is supported by calculations showing that the
barrier to this process is substantially decreased when a
a-phenyl substituent is present. This is support for the
role of ylide 3 in the photo-oxidation of sulfides that
would be difficult to obtain in the reaction of dialkyl
sulfides, since in that case the yield of sulfone, possibly
formed via 3, is very low in an aprotic solvent, where the
main products is the sulfoxide, formed much more
efficiently in protic media.
As for the mechanism of such an intramolecular step,
the original report suggested that this is a proton
abstraction by the negatively charged outer oxygen atom
of the persulfoxide.5 Both ab initio and the present PM3
calculations show indeed that there is substantial charge
separation in the persulfoxide. However, Table 2 shows
that the energy of the TS for hydrogen transfer dropped
monotonically with increasing substitution at the
a-carbon and, despite the large scaling factor between
the semiempirical and ab initio calculated DH≠, the effect
was sufficiently large and consistent to ensure that the
trend was real. The stabilization of the TS was strictly
proportional to the decrease in the a-C—H homolytic
bond dissociation energy (BDE), while it is doubful that
phenyl or methyl substituents would substantially
increase the a-hydrogen acidity. Furthermore, we found
previously that the efficiency of photooxidative C—S
bond cleavage in both the 4-nitro- and the 4-methoxy-
substituted derivatives was similar to that of parent
benzyl ethyl sulfide,7 suggesting that polar factors are not
With 1c the ratio of productive vs unproductive
conformations remains the same as with 1b (1:1), but
the further decrease of DH≠ favours rearrangement to 3c.
As a result, the oxidation of sulfide 6c, although not as
efficient as that of 6a, mainly gives the ketone.
Certainly, the final result depends on the following
reaction of ylide 3, in addition to its rate of formation.
The ab initio analysis by Jensen et al.2 suggests that a
possible path is backward rearragement to 1. The fact that
no deuterium loss occurs on recovered sulfide after
partial photooxidation of the a,a-d2-benzyl ethyl sulfide7
militates against a role of such a path with benzyl
derivatives. However, the fact sulfide 6c, for which the
yield of C—S cleavage is somewhat lower gives a higher
yield of sulfone 9c may indicate that the competition
between the reactions of 3 is also affected by substituents.
The rearrangement to the ylide and the following
conformation-dependent chemistry of this intermediate
may have a role also with other sulfides. In particular, this
may be the case for the Pummerer rearrangement
reported by Ando and co-workers with thiolanes11 and
thiazolidines4 (in the former case, attributed by the
Copyright 1999 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 12, 703–707 (1999)

PHOTOSENSITIZED OXYGENATION OF BENZYL SULFIDES
707
authors to the acidity of the a-hydrogen). This process
seems to be favoured with five-membered cyclic sulfides.
these conditions, so in this case the yields were
determined by 1H NMR spectroscopy.
Computational details. PM320 calculations were carried
out with the HyperChem program (version 5.0). Persulf-
oxide conformational minima were fully optimized.
Transition structures were located using the algorithm
eigenvector following. Transition structures have only
one negative eigenvalue (first-order saddle point) with
the corresponding eigenvector involving the formation of
an O—H bond and the cleavage of the benzylic C—H
bond.
CONCLUSION
Benzyl sulfides are photo-oxidized more easily than
dialkyl sulfides, and are cleaved to aldehydes or ketones.
The experimentally determined relative efficiency of this
fragmentation agrees with the computational prediction
about the ease of the rearrangement of the initially
formed oxygen adduct, persulfoxide 1, to ylide 3, in turn
the likely intermediate for the observed carbonyl
derivatives. The 1→3 conversion can be viewed as a
hydrogen atom transfer process, and its probability is
determined by the a-C—H BDE and the proportion of
correctly oriented conformations of 1. This is per se not
definitive evidence that this is the mechanism operating.
Another possibility is that 3 is formed through a separate
path, e.g. via a concerted reaction directly from the
sulfide and singlet oxygen, analogously to the alkene ene
reaction; this is e.g. the path indicated at the MP2/6—31
G(d) level, although there is no experimental support.2 At
any rate, the present evidence suggests that if 1 is the
intermediate it has an as yet overlooked diradicalic
character.
Acknowledgements
Financial support by CNR, Rome and MURST, Rome is
gratefully acknowledged. S.M.B. thanks FOMEC for a
fellowship.
REFERENCES
1. (a) E. L. Clennan, Sulfur Rep. 19, 171 (1996); (b) W. Ando and T.
Takada, in Singlet O2, edited by A. A. Frimer, Vol. 3, p. 1. CRC,
Boca Raton, FL (1985); (c) W. Ando, Sulfur Rep. 1, 143 (1981);
(d) C. S. Foote, in Singlet Oxygen, edited by H. H. Wasserman
and R. W. Murray, p. 139. Academic Press, New York (1979); (e)
F. Jensen, in Advances in Oxygenated Processes, edited by A. L.
Baumstark, Vol. 4, p. 49. JAI Press, Greenwich, CT (1995); (f) G.
O. Schenk and C. H. Krauch, Angew. Chem. 74, 510 (1962).
2. F. Jensen, A. Greer and E. L. Clennan, J. Am. Chem. Soc. 120,
4439 (1998).
3. M. L. McKee, J. Am. Chem. Soc. 120, 3963 (1998).
4. T. Takata, K. Hoshino, E. Takeuchi, Y. Tamura and W. Ando,
Tetrahedron Lett. 25, 4767 (1984); T. Takata, K. Ishibashi and W.
Ando, Tetrahedron Lett. 26, 4609 (1985).
EXPERIMENTAL
Materials. Benzene for the photo-oxygenation was
distilled over calcium hydride before use. The sulfides
6a–c12–14 were prepared from the corresponding halides
through published procedures. Likewise, samples of the
corresponding sulfoxides 7a–c15–17 and sulfones
8a–c13,17,18 for comparison with the photoproducts were
prepared by 3-chlorobenzoic acid oxidation according to
the general procedure.19
5. E. J. Corey and C. Ouanne`s, Tetrahedron Lett. 4263 (1976).
6. W. Ando, T. Nagashima, K. Saito and S. Kohmoto, J. Chem. Soc.,
Chem. Commun. 154 (1979); T. Akasaka, A. Sakurai and W.
Ando, J. Am. Chem. Soc. 113, 2696 (1991).
7. S. M. Bonesi, M. Mella, N. d’Alessandro, G. G. Aloisi, M. Vanossi
and A. Albini, J. Org. Chem. 63, 9946 (1998).
8. K. Ishiguro, M. Hayashi and Y. Sawaki, J. Am. Chem. Soc. 118,
7265 (1996).
9. F. Jensen and C. S. Foote, J. Am. Chem. Soc. 109, 1478 (1987); F.
Jensen, J. Org. Chem. 57, 6478 (1992).
10. P. C. Hiberty, Isr. J. Chem. 23, 10 (1983).
11. T. Takata, K. Ishibashi and W. Ando, Tetrahedron Lett. 26, 4609
(1985).
12. G. Modena, Gazz. Chim. Ital. 89, 834 (1960).
13. W. E. Bacon and W. M. LeSuer, J. Am. Chem. Soc. 76, 670 (1954).
14. M. Dagonneau and J. Vialle, Bull. Soc. Chim. Fr. 2072 (1972).
15. A. Cerniani, G. Modena and P. E. Todesco, Gazz. Chim. Ital. 90, 3
(1960).
16. Y. Kodama, S. Zuschi, K. Nishihata, M. Nishio and J. Uzawa, J.
Chem. Soc., Perkin Trans. 2 1307 (1980).
17. S. Becher, Y. Fort and P. Caube`re, J. Org. Chem. 55, 6194 (1990).
18. J. Buechi, M. Prost, H. Eichenberger and R. Lieberherr, Helv.
Chim. Acta 35, 1527 (1952).
Photoreactions. The photo-oxidations were carried out
using 0.01 M solutions of the appropriate sulfide in
benzene in the presence of tetraphenylporphine. The
solutions were contained in rubber-stoppered Pyrex
tubes. These were exposed to four phosphor-coated
15 W lamps (Applied Photophysics) emitting from 350 to
700 nm while a stream of oxygen saturated with benzene
was passed in to the solution through a needle.
The products were determined by GC on the basis of
calibration curves. A Hewlett-Packard Model 5890
instrument equipped with an Innowax column (30 m Â
0.25 mm i.d., 0.5 mm film thickness) was used, with He as
the carrier gas. The injector pressure was constant
(12 psi). The temperature program was 50–140°C (5°C
min 1), 140–250°C (25°C min 1), 7 min final hold.
Sulfoxide 8b was converted to a-methylstyrene under
19. J. Drabowicz and M. Mikolajczyk, Org. Prep. Proced. Int. 14, 45
(1982).
20. J. J. P. Stewart, J. Comput. Chem. 10, 209 (1989).
Copyright 1999 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 12, 703–707 (1999)