Radical-chain redox rearrangement of cyclic benzylidene acetals to benzoate esters in the presence of thiols

Article abstract of DOI:10.1016/S0040-4039(00)01904-3

Cyclic benzylidene acetals derived from 1,2- and 1,3-diols undergo an efficient ring-opening redox rearrangement to give benzoate esters in the presence of a peroxide initiator and a thiol, which acts as a polarity-reversal catalyst to promote the radical-chain reaction.

Full text of DOI:10.1016/S0040-4039(00)01904-3

TETRAHEDRON
LETTERS
Pergamon
Tetrahedron Letters 42 (2001) 137–140
Radical-chain redox rearrangement of cyclic benzylidene acetals
to benzoate esters in the presence of thiols
Brian P. Roberts* and Teika M. Smits
Department of Chemistry, University College London, 20 Gordon Street, London WC1H 0AJ, UK
Received 25 September 2000; accepted 26 October 2000
Abstract—Cyclic benzylidene acetals derived from 1,2- and 1,3-diols undergo an efficient ring-opening redox rearrangement to
give benzoate esters in the presence of a peroxide initiator and a thiol, which acts as a polarity-reversal catalyst to promote the
radical-chain reaction. © 2000 Elsevier Science Ltd. All rights reserved.
1,2- and 1,3-Diols are commonly protected as cyclic
benzylidene acetals 1 (n=0 or 1) during organic syn-
thesis.1 In 1962, Huyser and Garcia2 reported that some
simple benzylidene acetals underwent an intramolecular
redox rearrangement to give alkyl benzoates 2 when
heated with di-tert-butyl peroxide (DTBP) at ca. 135°C
for 18 h. For example, 2-phenyl-4-methyl-1,3-dioxane 3
gave a mixture of the benzoates 4 and 5, together with
by-products including benzaldehyde. When the starting
ratio 3:DTBP was 5:1, 47% of 3 was converted to
products and the ester ratio 4:5 was 5.4:1, but when
3:DTBP was increased to 36.8:1 conversion was only
9.8% and the ratio 4:5 was 4.5:1.
The reaction was thought to follow the radical-chain
mechanism shown in Scheme 1, although the kinetic
chain length was very short (1.3–1.9), probably because
step B of the propagation cycle is relatively slow even
Scheme 1.
Keywords: radicals and radical reactions; catalysis; thiols; diols; rearrangements; carbohydrates.
* Corresponding author.
0040-4039/01/$ - see front matter © 2000 Elsevier Science Ltd. All rights reserved.
PII: S0040-4039(00)01904-3

138
B. P. Roberts, T. M. Smits / Tetrahedron Letters 42 (2001) 137–140
though the acetal was present as the solvent. The
benzaldehyde was believed to arise from ring-opening
rearrangement and subsequent fragmentation of the
radical produced by abstraction of hydrogen from C-4
of 3 by the reactive and relatively unselective tert-bu-
toxyl radical.2 The same methodology was applied by
Jeppesen et al.3 to bring about intramolecular redox
rearrangement of cyclic benzylidene acetals derived
from carbohydrates; DTBP was usually used as solvent,
but no improvements in yields were found when
chlorobenzene was present as co-solvent, despite the
low solubility of some of the acetals in neat DTBP. For
example, the 4-benzoate 7 was isolated in 41% yield
after heating the 4,6-O-benzylidene acetal 6 in DTBP at
140°C for 7 h; none of the primary benzoate 8 was
detected in this reaction.3
was made after 40 min. Removal of the solvent and
examination of the residual oil by NMR spectroscopy
showed complete (]99%) conversion of 9 to 3-methyl-
butyl benzoate 10; tert-pentyl benzoate (which would
arise from cleavage of the primary CꢀO bond) was not
detected. Similar results were obtained when tri-tert-bu-
toxysilanethiol [(ButO)3SiSH, TBST] was used as cata-
lyst. However, in the absence of any thiol catalyst,
under otherwise identical conditions, only ca. 2% con-
version of 9 to 10 was observed.
Analogous thiol-catalysed redox rearrangements take
place with the benzylidene acetals 3 and 11. Essentially
complete conversion of 11 to the secondary alkyl ben-
zoate 12 was observed in the presence of TDT or TBST
and only a trace of the isomeric tertiary benzoate 13
Although step B of Scheme 1 should be appreciably
exothermic, because of the benzylic stabilisation of the
product radical, the reaction still suffers from adverse
polar effects in the transition state and should, there-
fore, be subject to polarity-reversal catalysis4 by a thiol.
Indeed, we have recently reported5 that radical-chain
redox reactions of the type shown in Eq. (1) are pro-
moted by thiols, which act as protic polarity-reversal
catalysts for hydrogen-atom transfer from the methyl-
was detected by GLC analysis (12:13=99:1). In the
absence of thiol only ca. 2% of 11 rearranged. The
TDT-catalysed rearrangement of 3 also took place
quantitatively to give a 6.7:1 mixture of the benzoates 4
and 5, similar to that obtained by Huyser and Garcia.2
The regioselectivity of these redox rearrangements is
determined by the relative rates of cleavage of the two
RꢀOC bonds in the intermediate 2-phenyl-1,3-dioxanyl
radicals (Scheme 1, step A) and follows the expected
‘enthalpic’ trend viz. R=3°>2°>1°.
’
ene group of the acyclic acetal to the tertiary alkyl
t
’
radical R .
Similar results were obtained with the 1,3-dioxolane 14,
except that the effect of thiol catalysis was less dra-
matic, presumably because direct abstraction of the
benzylic hydrogen (Scheme 1, step B) is faster when a
five-membered ring is involved.9 Quantitative rear-
rangement of 14 to 15 was observed in the presence of
TDT under the standard conditions, while 23% conver-
RtOCH2OMe“RtH+MeOCHO
(1)
We reasoned that thiols should also function as polar-
ity-reversal catalysts for step B of Scheme 1 and should
thus promote the redox rearrangement of benzylidene
acetals to give benzoate esters. In the present paper we
report the experimental verification of this proposal.
A solution in dry octane (2 ml) containing the benzyli-
dene acetal 9 (1 mmol), 2,2-di-tert-butylperoxybutane6
(DBPB, 5 mol%) as a thermal source of initiating
sion of 14 was seen in the absence of thiol; no tert-butyl
benzoate was detected. Although b-acyl- and b-aroyl-
’
oxyalkyl radicals (including BzOCMe2CH2 ) are well
tert-butoxyl radicals, tert-dodecanethiol7 (TDT,
5
known to rearrange by a 1,2-migration of the ester
group,10 the regioselective formation of isobutyl ben-
zoate 15 almost certainly10,11 results from highly selec-
tive opening of the intermediate 2-phenyl-4,4-dimethyl-
mol%) and collidine8 (10 mol%) was heated under
reflux (bath temp. 140–145°C) for a total of 3 h. A
second addition of DBPB and TDT (5 mol% of each)

B. P. Roberts, T. M. Smits / Tetrahedron Letters 42 (2001) 137–140
139
1,3-dioxolanyl radical to give directly the more stable
tertiary radical BzOCH2CMe2 .
step in such a conversion (see Scheme 2). Thus, when
24 was subjected to the usual conditions, using either
TDT or TBST as catalyst and refluxing octane–
chlorobenzene (3:1 v/v) as the solvent, conversion to
diethyl (R)-O-benzoylmalate 25 was 95% complete;
without thiol catalyst only 16% of the tartrate rear-
ranged. Chiral stationary-phase HPLC analysis con-
firmed that no racemisation of 25 takes place under the
reaction conditions. Diethyl tartrate derived from natu-
ral tartaric acid is readily converted to 24 and 25 can be
easily debenzoylated by treatment with Ti(OEt)4 in
ethanol,15 providing a simple route to unnatural diethyl
malate. This procedure represents a significant and
environmentally-friendly improvement over the previ-
ous methodology15 which involves classical Hanessian-
Hullar16 ring-opening of 24 by N-bromosuccinimide,
followed by tributyltin hydride reduction of the derived
bromomalate to give 25.
’
Thiol catalysis was also applied successfully to some of
the carbohydrate benzylidene acetals investigated by
Jeppesen et al.3 Under the standard conditions (DBPB
initiator+collidine) in refluxing octane–chlorobenzene
(1:1 v/v) solvent, 85% of 6 rearranged to the benzoate 7
when TDT was used as catalyst and conversion could
be increased to ]98% by the use of TBST. From the
latter reaction mixture 712 was isolated in 89% yield;
only a trace of the benzoate 8 was detectable by NMR
spectroscopic analysis of the crude product (7:8=ca.
97:3). Similar treatment of the 5,6-O-benzylidene
derivative 16, using TBST as catalyst, resulted in ]
98% conversion to a mixture of the benzoates 17 and 18
1
(4.9:1 by H NMR), which were isolated in yields of 72
and 14%, respectively.
It has been reported13 that treatment of the 4,6-thiono-
carbonate 19 with tributyltin hydride under radical
conditions gives the 4-deoxy compound 20, after hy-
drolysis of the initially-formed thiolester 21, pre-
sumably itself derived from regioselective ring-opening
b-scission of the intermediate radical 22 to give the
secondary C-4 alkyl radical. This result contrasts
markedly with the highly selective ring opening of the
benzylic radical 23, to give almost exclusively the pri-
mary C-6 alkyl radical, that occurs in the redox rear-
rangement of 6 to 7 described above. The latter result
also contrasts with the enthalpic control of regioselec-
tivity observed for ring opening of the 2-phenyl-1,3-
dioxanyl radicals derived from the monocyclic
benzylidene acetals 3, 9 and 11. However, Ziegler and
Zheng have reported clear-cut examples of contra-ther-
modynamic ring opening of bicyclic thionocarbonates
mediated by tributyltin hydride and rationalised their
results in terms of angle strain effects in the transition
states for b-scission of radicals related to 22.14 There is
clearly a need for further studies of the regiochemistry
of both types of redox ring-opening reactions.
In conclusion, we emphasise that in the thiol-catalysed
redox rearrangement of benzylidene acetals the chain is
propagated by thiyl radicals, which are much less reac-
tive and more selective in their hydrogen-atom abstrac-
tion reactions than are alkoxyl radicals and seek out
specifically the benzylic CꢀH group. Thus, our proce-
dure is not only higher yielding than the uncatalysed
processes of Huyser2 and of Jeppesen,3 but is also
cleaner and easier to control.
Acknowledgements
We thank the EPSRC for financial support.
References
1. Green, T. W.; Wuts, P. G. M. Protective Groups in
Organic Synthesis; Wiley-Interscience: New York, 1999;
3rd ed.
2. Huyser, E. S.; Garcia, Z. J. Org. Chem. 1962, 27, 2716.
3. Jeppesen, L. M.; Lundt, I.; Pedersen, C. Acta Chem.
Scand. 1973, 27, 3579.
4. Roberts, B. P. Chem. Soc. Rev. 1999, 28, 25.
5. Dang, H.-S.; Franchi, P.; Roberts, B. P. Chem. Commun.
2000, 499.
Simple methods for the conversion of natural (R,R)-
tartaric acid derivatives to those of unnatural (R)-malic
acid would be of appreciable interest15 and we have
now shown that thiol-catalysed redox rearrangement of
the (R,R)-benzylidene tartrate 24 can be used as the key
Scheme 2.

140
B. P. Roberts, T. M. Smits / Tetrahedron Letters 42 (2001) 137–140
6. Available from the Aldrich Chemical Co. as a 50%
5.02 (1H, [t], J 9.7, H-4), 5.60 (1H, [t], J 9.6, H-3), 7.39
(2H, [t], J 7.8, H-m), 7.53 (1H, tt, J 7.4 and 1.3, H-p),
7.96 (2H, dd, J 8.4 and 1.3, H-o); lC 17.3, 20.5, 20.7,
55.2, 65.2, 69.7, 71.3, 74.1, 96.6, 128.4, 129.1, 129.7,
133.3, 165.4, 169.9 and 170.1. The use of [multiplet]
indicates an apparent multiplet with line spacing corre-
sponding to an average coupling constant; the apparent
triplet from H-3 shows further small splittings due to
virtual coupling to H-1 (which is coupled to H-2 and
nearly isochronous with it).
solution in mineral oil and used as such; the half-life of
this peroxide is ca. 1 h at 125°C.
7. This is the isomeric mixture of tertiary thiols as supplied
by the Aldrich Chemical Co.
8. Yields were generally improved in the presence of col-
lidine (2,4,6-trimethylpyridine), the role of which is prob-
ably to act as a scavenger of acid resulting from reactions
between the initiator and the thiol.
9. See, for example: Malatesta, V.; Scaiano, J. C. J. Org.
Chem. 1982, 47, 1455.
13. Barton, D. H. R.; Subramanian, R. J. Chem. Soc., Perkin
Trans. 1 1977, 1718.
14. Ziegler, F. E.; Zheng, Z. J. Org. Chem. 1990, 55, 1416.
15. Hungerbu¨hler, E.; Seebach, D.; Wasmuth, D. Helv.
Chim. Acta 1981, 64, 1467.
16. Hanessian, H. Methods Carbohydr. Chem. 1972, 6, 183.
Hanessian, H.; Plessas, N. R. J. Org. Chem. 1969, 34,
1053.
10. Barclay, L. R. C.; Griller, D.; Ingold, K. U. J. Am.
Chem. Soc. 1982, 104, 4399.
11. Newcomb, M.; Glenn, A. G.; Manek, M. B. J. Org.
Chem. 1989, 54, 4603.
12. NMR (500 MHz for 1H, 125 MHz for 13C, CDCl3
solvent, J in Hz); lH 1.20 (3H, d, J 6.3, Me-6), 1.84 (3H,
s, Ac), 2.03 (3H, s, Ac), 3.38 (3H, s, OMe), 3.98 (1H, dq,
J 9.9 and 6.3, H-5), 4.87–4.92 (2H, m, H-1 and H-2),
.
.