Selective radical-chain epimerisation at electron-rich chiral tertiary C-H centres using thiols as protic polarity-reversal catalysts

Article abstract of DOI:10.1039/b103558b

Radical-chain epimerisation at chiral tertiary CH centres adjacent to ethereal oxygen atoms can be brought about in the presence of thiols, the function of which is to act as protic polarity-reversal catalysts for hydrogen-atom transfer between pairs of nucleophilic α-alkoxyalkyl radicals. The viability of the method is demonstrated by epimerisation of a series of simple molecules that contain two chiral centres and then the procedure is applied to more complex carbohydrate-based systems, where it is possible to convert a readily available diastereoisomer into a rarer one in a straightforward manner. Of necessity, epimerisation always proceeds in the direction of thermodynamic equilibrium and, in general, the results obtained are in accord with the predictions of molecular mechanics calculations using the MMX force-field. When the required isomer is less stable than the starting diastereoisomer, thiol-catalysed epimerisation of a suitable derivative of the parent can provide a means to obtain the desired compound in satisfactory yield, after deprotection of the epimerised derivative. This strategy is demonstrated for the conversion of trans-cyclohexane-1,2-diol into the less stable cis-isomer and for related contra-thermodynamic isomerisation of some carbohydrates, as well as for the conversion of meso-1,2-diphenylethane-1,2-diol into the dl-form. Thiol-catalysed epimerisation at a CH centre adjacent to an ether-oxygen atom is much faster than at a similar centre adjacent to an amido-nitrogen atom, a result that can be understood in terms of the importance of polar effects on the rate of abstraction of hydrogen by electrophilic thiyl radicals.

Full text of DOI:10.1039/b103558b

View Article Online / Journal Homepage / Table of Contents for this issue
Selective radical-chain epimerisation at electron-rich chiral tertiary
C–H centres using thiols as protic polarity-reversal catalysts
Hai-Shan Dang, Brian P. Roberts* and Derek A. Tocher†
1
Christopher Ingold Laboratories, Department of Chemistry, University College London,
20 Gordon Street, London, UK WC1H 0AJ
Received (in Cambridge, UK) 24th April 2001, Accepted 7th August 2001
First published as an Advance Article on the web 10th September 2001
Radical-chain epimerisation at chiral tertiary CH centres adjacent to ethereal oxygen atoms can be brought about in
the presence of thiols, the function of which is to act as protic polarity-reversal catalysts for hydrogen-atom transfer
between pairs of nucleophilic α-alkoxyalkyl radicals. The viability of the method is demonstrated by epimerisation of
a series of simple molecules that contain two chiral centres and then the procedure is applied to more complex
carbohydrate-based systems, where it is possible to convert a readily available diastereoisomer into a rarer one in a
straightforward manner. Of necessity, epimerisation always proceeds in the direction of thermodynamic equilibrium
and, in general, the results obtained are in accord with the predictions of molecular mechanics calculations using the
MMX force-field. When the required isomer is less stable than the starting diastereoisomer, thiol-catalysed epi-
merisation of a suitable derivative of the parent can provide a means to obtain the desired compound in satisfactory
yield, after deprotection of the epimerised derivative. This strategy is demonstrated for the conversion of trans-
cyclohexane-1,2-diol into the less stable cis-isomer and for related contra-thermodynamic isomerisation of some
carbohydrates, as well as for the conversion of meso-1,2-diphenylethane-1,2-diol into the dl-form. Thiol-catalysed
epimerisation at a CH centre adjacent to an ether-oxygen atom is much faster than at a similar centre adjacent to
an amido-nitrogen atom, a result that can be understood in terms of the importance of polar effects on the rate of
abstraction of hydrogen by electrophilic thiyl radicals.
Selective epimerisation of organic molecules that contain two
or more chiral centres has significant potential as a simple
method for the conversion of a readily accessible diastereo-
isomer into a more desirable one. In a preliminary communi-
cation,1 we have reported that thiol-catalysed radical-chain
epimerisation takes place selectively at a chiral tertiary CH
centre that is activated by an attached ether-oxygen atom. The
role of the thiol is to act as a protic polarity-reversal catalyst
that mediates reversible hydrogen-atom abstraction at the
CH centre.2 In the present paper we report full details of
these experiments and describe further applications of the
Scheme 1
methodology.
A thermoneutral hydrogen-atom exchange reaction of the
type shown in eqn. (1) usually has a relatively high activation
the nucleophilic α-alkoxyalkyl radical and the electrophilic thiyl
radical are significantly different.3
The thiol-catalysed epimerisation process is necessarily under
thermodynamic control. However, it should prove possible to
obtain an adequately high yield of the required epimer from its
more stable or similarly stable parent by first converting the
latter into a suitable derivative, such that the corresponding
derivative of the desired epimer is now appreciably more stable.
This strategy is outlined in Scheme 2 and a preliminary report
of its realisation has been published.4
energy and is slow at moderate temperatures. This can be
explained in terms of the importance of ‘polar effects’ in
determining the rates of reaction of electrically neutral radicals,
because of the absence of charge-transfer stabilisation of the
symmetrical transition state 1, in which the incoming and out-
going α-alkoxyalkyl radicals have the same electronegativity.3 In
the presence of a thiol catalyst, the direct hydrogen-abstraction
process is replaced by the sequence of hydrogen-atom transfer
reactions shown in Scheme 1. Each of these elementary reac-
tions proceeds via the transition state 2, which benefits from
charge-transfer stabilisation because the electronegativities of
† Correspondence concerning the X-ray crystallography should be
directed to this Author.
Scheme 2
2452
J. Chem. Soc., Perkin Trans. 1, 2001, 2452–2461
This journal is © The Royal Society of Chemistry 2001
DOI: 10.1039/b103558b

View Article Online
Table 1 Thiol-catalysed epimerisation of isosorbide dimethyl ether 7
Results and discussion
in octane at 125 ЊC in the presence of DBPB initiatora
1
Epimerisation of simple molecules
Isomeric composition (%)
Epimerisation at the oxygen-activated tertiary CH centres in
the cyclic ketals 3 and 4 and in 1,2-dimethoxycyclopentane 5
was examined initially in order to explore the viability of the
approach. When a nonane solution containing 3-cis (ca. 1 M)
and 5 mol% of 2,2-bis(tert-butylperoxy)butane 6 (DBPB,
present as an initiator) was heated under argon or nitrogen at
125 ЊC for 3 h, no conversion to the trans-isomer was observed
by GLC analysis. The half-life for decomposition of DBPB to
give alkoxyl radicals, which would go on to abstract hydrogen
from 3, is ca. 1 h at 125 ЊC; nonane was used as solvent in order
to avoid interference with the GLC analysis. However, when the
experiment was repeated in the presence of 5 mol% tert-
dodecanethiol‡ (TDT), slow epimerisation of 3-cis to the more
stable 3-trans was observed and a final cis : trans ratio of 63 : 37
was achieved after 1 h. In the additional presence of 2,4,6-
trimethylpyridine (collidine, 10 mol%), the function of which
is probably to remove traces of acid formed from thiols under
the reaction conditions, the cis : trans ratio was 46 : 54 after 1 h
and reached a final value of 43 : 57 after 2 h. We have found
previously5,6 that silanethiols are often more effective protic
polarity-reversal catalysts than are simple alkanethiols and
when the TDT was replaced by triphenylsilanethiol (Ph3SiSH;
TPST), isomerisation of 3-cis proceeded further and more
rapidly (cis : trans = 24 : 76 after 1 h). However, when collidine
was also present very little isomerisation took place, evidently
because TPST reacts with collidine under the conditions used
for epimerisation. Tri-tert-butoxysilanethiol [(ButO)3SiSH;
TBST], which is much less sensitive to nucleophilic attack at
silicon, proved to be a very efficient catalyst for the epimeris-
ation of 3-cis, especially in the presence of 10 mol% of collidine.
After 2.5 h at 125 ЊC a cis : trans ratio of 16 : 84 was reached and
this corresponds to thermodynamic equilibrium, because the
same isomer ratio was obtained when starting from pure
3-trans. Molecular mechanics calculations using the MMX
force-field7 indicate that the trans-isomer is 6.6 kJ molϪ1 more
stable than the cis and, assuming that this value corresponds
approximately to the free-energy difference between the two
isomers at 125 ЊC, the predicted cis : trans equilibrium ratio is
12 : 88.
Entry
Thiol catalyst
7
8
9
1
2
3
4
None
MeO2CCH2SH
TPST
(4-FC6H4)3SiSH
TBST
TBST
100
89.1
62.4
58.4
40.0
38.5
not detected
not detected
10.0
36.4
40.4
52.5
1.1
1.2
1.2
1.5
1.5
5
6b
60.0
a The initiator and the thiol (5 mol% of each) were added initially and
again after 1 h. The total reaction time was 4 h although, usually, no
further change occurred after 3 h. b Collidine (10 mol%) was present
initially.
The thiol-catalysed epimerisation of 1,2-dimethoxycyclo-
pentane 5 required rather more forcing conditions to drive it
to thermal equilibrium at 125 ЊC. In total, three additions of
TBST and DBPB (3 mol% of each) were made, initially, after
30 min and after 1 h to 5-cis in octane and the progress of the
epimerisation was monitored by GLC. After heating for 3 h in
all, the ratio 5-cis : 5-trans was 5 : 95 and this isomer ratio was
also obtained when the pure trans isomer was subjected to the
same treatment, showing that thermal equilibrium had been
established. Molecular mechanics calculations predict that the
trans isomer is more stable than the cis by 8.9 kJ molϪ1, which
translates to a trans : cis ratio of ca. 6 : 94 at 125 ЊC if entropy
differences are neglected.
2
Epimerisation of more complex molecules
Di-O-methyl-1,4 : 3,6-dianhydro--glucitol (isosorbide dime-
thyl ether) 7 is readily available commercially. Assuming that
the cis ring junction is preserved, epimerisation can take place
either at C-2 to give the corresponding dianhydro--mannitol
derivative 8 or at C-5 to give the dianhydro--iditol derivative 9.
While 8 is readily obtainable by methylation of commercial
isomannide, the ether 9 is much less accessible. However,
molecular mechanics calculations indicate that the order of
stability is 9 (0) > 7 (ϩ5.8 kJ molϪ1) > 8 (ϩ15.0 kJ molϪ1). The
epimerisation of 7 was investigated in octane solution at 125 ЊC,
using DBPB as initiator and various thiols as catalysts. The
progress of the reaction was monitored by GLC and the results
are summarised in Table 1. The silanethiols TPST and TBST
were much more effective catalysts than methyl thioglycolate
(MeO2CCH2SH) and no epimerisation was detectable in the
absence of a thiol. Tris(4-fluorophenyl)silanethiol, which was
investigated on the basis that it should yield a more electrophilic
thiyl radical than that from TPST, was no more effective than
the latter. The presence of collidine improved the efficiency of
the epimerisation somewhat (entries 5 and 6) and pure 9 was
easily isolated in 56% yield from the latter run by flash chroma-
tography on silica gel. When pure isomannide dimethyl ether 8
was subjected to the conditions of entry 6, the final epimeric
proportions 7 : 8 : 9 were 41.5 : 45.5 : 13.0. Although 7 can be
formed directly from 8, compound 9 is presumably formed
indirectly by epimerisation of 7.
Epimerisation of the 1,3-dioxanes 4, obtained from pentane-
2,4-diol, was carried out under similar conditions in octane
solvent at 125 ЊC (bath temp.) in the presence of collidine
and TBST as polarity-reversal catalyst. The pure trans-isomer
was prepared from the (R,R)-diol and a 53 : 47 cis : trans
mixture was prepared from a commercially available mixture
of meso- and dl-diols. Whatever the isomeric composition of
the starting dioxane, an equilibrium mixture consisting of
93% 4-cis and 7% 4-trans was obtained within 1 h. Molecular
mechanics calculations predict the cis-isomer to be the more
stable by 13.7 kJ molϪ1, which corresponds to an equilibrium
ratio of 98 : 2 in favour of 4-cis at 125 ЊC.
The carbocyclic analogue 118a of isomannide dimethyl ether
was readily prepared by methylation of endo,endo,cis-bicyclo-
[3.3.0]octane-2,6-diol8b,c and underwent epimerisation in a
similar manner to 8 under the conditions of entry 6. Thus,
after 4 h at 125 ЊC the epimeric proportions 10 : 11 : 12 deter-
mined by GLC were 42.0 : 51.5 : 6.5. According to molecular
mechanics calculations, the relative energies (kJ molϪ1) of these
molecules increase in the order 12 (0) < 10 (ϩ1.2) < 11 (ϩ7.0),
following a trend qualitatively similar to that shown by 7–9 and
as expected on steric grounds. Although the exo,exo-epimer 12
is the most stable, it cannot be formed directly from 11 and
must arise from isomerisation of 10.
‡ Commercial mixture of isomers.
J. Chem. Soc., Perkin Trans. 1, 2001, 2452–2461
2453

View Article Online
The 3-O-methyl derivative 139 of the readily available
diacetone--glucose proved more resistant to epimerisation,
which could, in principle, take place at C-3, C-4 or C-5. How-
ever, when 13 was heated under reflux in nonane in the presence
of di-tert-butyl peroxide initiator and TBST as catalyst, partial
conversion (32%) to another isomer was achieved; no other
products were detected by NMR spectroscopy. The product was
isolated in 25% yield by flash chromatography and shown to be
the β--idofuranose 14, on the basis that the only significant
Fig. 1 Structure of methyl 2,3 : 5,6-di-O-isopropylidene-β--gulo-
furanoside 18 determined by X-ray crystallography. The crystal-
lographic numbering system differs from the conventional system used
in the text. Selected geometrical parameters (bond lengths in Å, bond
angles in degrees): O(1)–C(2) 1.395(5), O(2)–C(2) 1.422(4), O(2)–C(8)
1.434(5), C(2)–C(3) 1.521(5), C(3)–C(7) 1.539(5), C(7)–C(8) 1.546(4),
C(8)–C(9) 1.510(5), C(9)–C(10) 1.496(6); C(2)–O(1)–C(1) 113.3(4),
C(2)–O(2)–C(8) 106.5, C(3)–C(7)–C(8) 103.7(3), O(2)–C(8)–C(7)
104.9(3), O(2)–C(8)–C(9) 110.0(3), C(8)–C(9)–C(10) 114.7(3).
1
change in the H NMR spectrum that accompanies epimeris-
ation of 13 takes place for the lines assigned to the two protons
attached to C-6. In 13 these protons give rise to the AB part of
an ABX pattern (δ 4.01 and 4.08; JAB = 8.6, JAX = 5.4 and JBX
=
4.5 Hz), while in the product 14 the chemical-shift difference
between the geminal protons increases substantially (δ 3.62 and
4.10; JAB = 8.1, JAX = 7.6 and JBX = 6.6 Hz) such that the
appearance becomes close to first-order. According to molecu-
lar mechanics calculations, the energies of the C-3, C-4 and C-5
epimers relative to 13 are ϩ3.7, ϩ3.2 and ϩ2.3 kJ molϪ1,
respectively, consistent with the observed partial epimerisation
at C-5. Indeed, when a pure sample of 14 was subjected to the
same conditions used to produce it from 13, a 67 : 33 mixture of
13 and 14 was obtained, showing that this ratio corresponds
essentially to thermodynamic equilibrium between these two
epimers.
be carried out in octane at 125 ЊC with DBPB as initiator and
TBST as catalyst, in the presence of collidine, to give 32%
conversion to a single isomeric product which was identified as
the C-5 epimer 16; none of the C-4 epimer was detected in the
reaction mixture. The -lyxo-hexofuranose 16, the -antipode
of which has been reported previously,11 was isolated in 25%
yield. When a pure sample of 16 was subjected to conditions
used to epimerise 15, GLC analysis showed that a mixture of 15
and 16 was obtained in the ratio 63 : 37, indicating that thermo-
dynamic equilibrium is almost reached under the reaction con-
ditions and implying that 15 is in fact slightly more stable than
16, in agreement with the molecular mechanics calculations.
The α--mannofuranoside 179,12 also underwent epimeris-
1
ation at C-5 and the accompanying changes in the H NMR
spectrum were very similar to those described above for the
epimerisation of 13 to 14. Conversion to the β--gulo-
furanoside 18 was 30% (by GLC) in refluxing octane and the
identity of the product was confirmed by single-crystal X-ray
diffraction (see Fig. 1) after its isolation in 24% yield. According
to molecular mechanics calculations, the C-5 (C-9 in Fig. 1)
epimer 18 is marginally (by 1.1 kJ molϪ1) less stable than its
parent 17. In accord with this prediction, when pure 18 was
subjected to the isomerisation conditions, the final ratio 17 : 18
was 65 : 35 in favour of the mannoside, showing that this is
indeed the more stable epimer and that thermodynamic equi-
librium is again almost achieved between these two epimers.
3
Contra-thermodynamic epimerisation
Clearly, it would be advantageous to circumvent the limitation
on the yield of a desired epimer that is imposed by its thermo-
dynamic stability relative to its parent. The viability of
our strategy for achieving this, via indirect epimerisation as
generalised in Scheme 2, was demonstrated for the cyclic
acetonide derivatives of a number of 1,2-diols. For these diols
the method relies on the different structural preferences of the
acetonides and the ease with which acetonides may be prepared
from, and converted back to, their parent diols.
Deoxygenation of diacetone--glucose at C-3, via reaction of
the corresponding xanthate with triphenylsilane in 1,4-dioxane
at 60 ЊC,10 afforded 15 in excellent yield. Compound 15
could potentially undergo epimerisation at C-4 or C-5 to give
products with relative energies of ϩ13.2 and ϩ0.6 kJ molϪ1,
respectively, according to molecular mechanics. The epimeris-
ation of 15 proceeded more readily than that of 13 and could
According to molecular mechanics calculations, the most
stable conformation of trans-cyclohexane-1,2-diol 19 is more
stable than that of the cis-isomer 20 by 2.2 kJ molϪ1. Both
2454
J. Chem. Soc., Perkin Trans. 1, 2001, 2452–2461

View Article Online
isomers of the diol are readily converted to the corresponding
acetonides 21 and 22 by treatment with an excess of 2,2-dimeth-
oxypropane in the presence of an acidic ion-exchange resin
(Amberlyst-15). The trans-diastereoisomer is destabilised with
respect to the cis-form by the extra strain present when a five-
membered dioxolane ring is trans-fused to the cyclohexane ring
and molecular mechanics calculations indicate that the trans-
acetonide is less stable than the cis-form by 3.8 kJ molϪ1. Heat-
ing either the trans- or the cis-acetonide in refluxing octane
(bp 126 ЊC, bath temp. 140–145 ЊC, internal temp. ca. 130 ЊC)
for 2.5 h in the presence of TBST (3 ×3 mol%), DBPB (3 ×
3 mol%) and collidine (1 × 10 mol%) led to the same equi-
librium mixture of isomers in which the cis-acetonide pre-
dominated to the extent of 95%. No detectable epimerisation
takes place in the absence of the thiol. The acetonide 22 could
be readily deprotected by treatment with an excess of methanol
in the presence of Amberlyst-15 at room temperature to give
the cis-diol in a pure state after one recrystallisation from
ethyl acetate. The sequence of derivatisation, epimerisation
and deprotection thus provides an efficient procedure for the
conversion of trans-cyclohexane-1,2-diol into the less stable
cis-isomer, based on the principle generalised in Scheme 2.
temperature) gave the free cis-diol, methyl 4-deoxy-β--erythro-
pentopyranoside 29. The structure of 28 was confirmed by its
independent synthesis from -ribose, which was first converted
to a 1 : 3 mixture of the 2,3- and 3,4-O-isopropylidene-β-
-ribopyranosides 30 and 31.14 These isomers proved very dif-
ficult to separate completely and so a mixture enriched in 30
was converted to a mixture of the xanthates 32 and 33, from
which 32 could be readily obtained in a pure state and then
reduced with triphenylsilane10 to give 28.
The large difference in reactivity of 25 and 27 towards thiol-
catalysed epimerisation at C-3 is probably the result of steric
shielding of H-3 by the 4-methoxy group in the former, coupled
with the deactivating polar effect of this β-methoxy substituent
on abstraction of H-3 by the electrophilic thiyl radical.15
According to molecular mechanics, the energies of the most
stable conformations of the dl- and meso-forms of 1,2-di-
phenylethane-1,2-diol (hydrobenzoin) are almost the same.
However, the meso-form 34 is appreciably easier to prepare than
the dl-isomer 35 and the latter is much more costly to obtain
commercially. On the other hand, the trans-acetonide 36
derived from the dl-diol is calculated to be more stable by 14.6
kJ molϪ1 than the cis-acetonide 37 which can be obtained from
the meso-diol, implying that radical-chain-epimerisation of
37 should provide an efficient means to convert the meso-diol to
the dl-diol under mild neutral conditions.
Methyl β--xylopyranoside 23 was converted into the aceto-
nide 24 using the published method13 and subsequent methyl-
ation afforded 25, which is predicted by molecular mechanics
calculations to be less stable than both its C-3 and C-2 epimers,
by 6.9 and 8.2 kJ molϪ1, respectively. However, attempted iso-
merisation of 25 failed under the conditions used to epimerise
13 to 14, and 25 remained unchanged. Under more forcing
conditions, in refluxing nonane (bp 151 ЊC) with di-tert-butyl
peroxide (20 mol%) as initiator and three additions of TBST
(3 × 5 mol%) during 3 h, partial isomerisation of 25 took place
to give 26 as the major product (30%), but the remainder of
the starting material was unchanged. The structure of 26 was
assigned on the basis of the nuclear Overhauser effects evident
in its 2D NOESY spectrum. Thus, H-2 shows a strong correl-
ation with both H-1 and H-3, indicating that all three protons
are located on the same side of the pyranose ring. Assuming
epimerisation at one site only, the configuration at C-1 must be
the same as that in 25 and thus epimerisation must have taken
place at C-2 to give 26.
In accord with this, treatment of 37 with 2,4,6-tris-
(trifluoromethyl)thiophenol16 as catalyst and DBPB as initiator
in refluxing octane resulted in complete (≥98%) conversion to
the trans-acetonide 36, which could be deprotected (Amberlyst-
15, excess of MeOH, reflux) to give the free dl-diol 35. The
choice of thiol catalyst for epimerisation of 37 proved to be
critical and no significant conversion to 36 was observed when
the arenethiol was replaced by TBST or tert-dodecanethiol,
under otherwise identical conditions. The key point here is that
epimerisation is taking place at a benzylic centre and involves
an oxygen-stabilised secondary benzylic radical. For efficient
radical-chain epimerisation, both abstraction of hydrogen from
37 by the thiyl radical and abstraction of hydrogen from
the thiol by the resulting benzylic radical must be rapid, and
it is likely that the latter reaction is significantly endothermic
and thus relatively slow for alkane- or silane-thiols. The S–H
bond in an arenethiol is appreciably weaker than that in an
alkanethiol (366 kJ molϪ1 in MeSH and 349 kJ molϪ1 in
PhSH)17 and, in general, the choice of thiol catalyst needs to be
made such that the strengths of the C–H and S–H bonds
involved are fairly similar. A number of other arenethiols were
investigated as catalysts for the epimerisation of 37 to 36 and
the results are summarised in Table 2.
In contrast to the sluggish isomerisation of 25, its 4-deoxy
analogue 27 underwent epimerisation readily in refluxing
octane, essentially under the conditions used to isomerise 15,
and was converted almost quantitatively into the C-3 epimer 28.
The acetonide 27 is calculated to be less stable than 28 by 18.9
kJ molϪ1 and less stable than its C-2 epimer by 12.0 kJ molϪ1.
Deprotection of 28 (Amberlyst-15, excess of MeOH, room
J. Chem. Soc., Perkin Trans. 1, 2001, 2452–2461
2455

View Article Online
Table 2 Epimerisation of cis-2,2-dimethyl-4,5-diphenyl-1,3-dioxolane
37 in refluxing octane using DBPB as initiator and different arenethiols
as catalystsa
diphenylethanol by treatment with acetone in the presence of
an excess of aluminium powder, followed by acylation at nitro-
gen with valeryl chloride in the presence of triethylamine, as
illustrated for 38 in Scheme 3. The parent oxazolidine 42 is very
Arenethiol
Final ratio 36 : 37
2,4,6-Tris(trifluoromethyl)thiophenol
4-(Trifluoromethyl)thiophenol
Pentafluorothiophenol
Pentachlorothiophenol
2,4-Dichlorothiophenol
4-Fluorothiophenol
4-Chlorothiophenol
Thiophenol
4-Methylthiophenol
4-Methoxythiophenol
≥98 : ≤2
≥98 : ≤2
≥98 : ≤2
95 : 5
81 : 19
68 : 32
66 : 34
60 : 40
36 : 64
14 : 86
Scheme 3 Reagents and conditions: (i) Me2CO, Al powder, cat.
MeC(O)Cl in CH2Cl2, rt, 20 h; (ii) BuC(O)Cl, Et3N in Et2O, 0 ЊC to rt,
12 h.
a The thiol and DBPB initiator (3 mol% of each) were added initially,
then again after 20 min and 40 min (9 mol% of each in all). The total
reaction time was 2.5 h and the final ratio 36 : 37 was determined by 1H
NMR spectroscopy.
sensitive to acid-catalysed hydrolysis and was converted directly
to the stable crystalline derivative 38. An authentic sample of
the (4R,5R)-oxazolidine 40 was prepared in the same way from
(1R,2R)-2-amino-1,2-diphenylethanol.20
Epimerisation of the cis-(4R,5S)-oxazolidine 38 was carried
out in refluxing nonane using pentafluorothiophenol as catalyst
(3 × 5 mol%) added initially, again after 20 min and again after
40 min; the total reaction time was 2.5 h. The initiator was
di-tert-butyl peroxide (20 mol% present initially, followed by
It is evident that electron-withdrawing substituents on the
benzene ring increase the effectiveness of the arenethiol as a
catalyst, relative to thiophenol itself, whilst electron-donating
groups reduce the effectiveness of the thiol as a protic catalyst.
It seems likely that there are two related reasons for this
behaviour. First, the presence of electron-withdrawing sub-
stituents, particularly at ortho- and para-sites, leads to an
increase in the strength of the S–H bond18 and probably to
better matching of the S–H and benzylic C–H bond strengths,
while electron-donating groups weaken the S–H bond relative
to that in thiophenol, leading to poorer matching. Second,
electron-withdrawing substituents render the thiyl radical more
electrophilic, resulting in increased charge-transfer stabilisation
of the transition state (see structure 2) for abstraction of the
benzylic hydrogen atom to give the nucleophilic benzylic
radical.3 Electron-donating groups render the thiyl radical less
electrophilic than that from thiophenol, resulting in a lower rate
of hydrogen abstraction because of less favourable polar effects.
Pentafluorothiophenol is particularly acidic with a pKa of 2.6,19
compared with 6.2 for thiophenol, and it seems likely that
1
further additions of 5 mol% after 20 and 40 min). H NMR
analysis showed that the final cis : trans ratio of oxazolidines
was 27 : 73 and the enantiomeric ratio 40 : 41 for the trans-
isomer was found to be 95 : 5 by chiral stationary-phase HPLC
analysis, using authentic 40 as a reference. Epimerisation thus
takes place with high selectivity at the carbon centre adjacent to
oxygen. The trans-oxazolidine was isolated in 72% yield and
readily upgraded to >99% ee by a single recrystallisation from
hexane. Similar reactions of the cis-(4S,5R)-oxazolidine 39 gave
a final cis : trans ratio of 24 : 76 and the enantiomeric ratio 40 :
41 was 4 : 96; the trans-isomer was isolated in 71%.
The final cis : trans ratio of ca. 25 : 75 obtained in refluxing
nonane is close to the value corresponding to thermodynamic
equilibrium at ca. 155 ЊC (estimated internal temp.) for epimeri-
sation at C-5 adjacent to oxygen, but far from equilibrium for
epimerisation at C-4 adjacent to nitrogen. Thus, when a pure
sample of the trans-(4R,5R)-oxazolidine 40 was subjected to
the conditions used to epimerise 38 and 39, the final cis : trans
ؒ
C6F5S will behave as an exceptionally electrophilic thiyl
radical, in accord with the effectiveness of C6F5SH as a protic
catalyst for the epimerisation of 37.
1
ratio was 20 : 80 as determined by H NMR spectroscopy.
However, the enantiomeric ratio 38 : 39 for the cis-isomer was
ca. 97 : 3 (by HPLC), whereas thermodynamic equilibrium
between these two enantiomers (established through epimeris-
ation adjacent to nitrogen) would correspond to a racemic
mixture of 38 and 39. Consistent with this interpretation, no
racemisation of recovered 40 (which would result from a double
epimerisation via 39) was detectable by HPLC analysis.
The relative ease of epimerisation at C-4 and C-5 in the
oxazolidines 38 and 39 can be understood in terms of polar
effects on the rates of benzylic hydrogen-atom abstraction by
the electrophilic thiyl radical;21 the strengths of the C5-H and
C6-H bonds are likely to be very similar. The abilities of alkoxy
and amido groups to stabilise positive charge on an attached
carbon atom (cf. structure 2) should be reflected in their
Hammett σ or σϩ substituent constants and for p-Me2N,
p-MeO and p-MeC(O)NH groups the values are Ϫ0.32 (Ϫ1.7),
Ϫ0.27 (Ϫ0.78) and 0.0 (Ϫ0.6), respectively (σϩ-values in
parentheses).22 Thus, the rate of hydrogen-atom abstraction
by electrophilic radicals from otherwise similar C–H bonds
adjacent to amino,23 alkoxy and amido substituents would be
expected to decrease in this order.
Experiments were carried out to investigate the formal
possibility that epimerisation of 37 might be brought about by
a heterolytic ring-opening mechanism under the influence of
simple acid catalysts. When 37 was heated in refluxing octane
in the presence of toluene-p-sulfonic acid (5 mol%), with or
without added DBPB, decomposition of the dioxolane was
almost complete within 40 min to give unidentified products
which did not include any of the trans-isomer 36. When these
runs were repeated, replacing the free sulfonic acid with its
pyridine salt (5 mol%), no 36 was produced during 3 h and the
cis-isomer 37 remained unchanged. In the absence of initiator,
but under conditions otherwise identical to those used for
radical-chain epimerisation, no conversion of 37 to 36 was
observed in the presence of pentafluorothiophenol.
In order to compare the ease of epimerisation at CH groups
adjacent to oxygen and adjacent to amido nitrogen, the cis–
trans isomerisation of the enantiomeric oxazolidines 38 and 39
was examined. The oxazolidines were prepared from the com-
mercially available (1S,2R) and (1R,2S) forms of 2-amino-1,2-
Conclusions
In summary, it appears that thiol-catalysed selective epi-
merisation at tertiary CH centres adjacent to ether-type oxygen
atoms can provide a useful method for the conversion of one
2456
J. Chem. Soc., Perkin Trans. 1, 2001, 2452–2461

View Article Online
diastereoisomer into another, more valuable, diastereoisomer.
The procedures used here are very straightforward, although
sometimes careful choice of the thiol catalyst is essential for
efficient epimerisation. When the required isomer is less stable
than the starting diastereoisomer, thiol-catalysed epimerisation
of a suitable derivative of the parent can provide a means to
obtain the desired compound in satisfactory yield, after depro-
tection of the epimerised derivative.
was added and the mixture was stirred and heated under
reflux (stirring-heating mantle) under nitrogen for 72 h. The
cooled reaction mixture was filtered through Celite to remove
unchanged silicon disulfide and the filter cake was washed with
diethyl ether. Excess of alcohol and diethyl ether were removed
from the filtrate by rotary evaporation and the residual oil was
distilled under reduced pressure to give the silanethiol (18.1 g,
31%) as a colourless liquid, bp 95 ЊC/15 mmHg (lit.,26 113–115
ЊC/35 mmHg); δH 0.18 (1 H, s, SH), 1.34 (27 H, s, But); δC 31.4,
74.4.
2,4,6-Tris(trifluoromethyl)thiophenol16 was prepared as
described previously and all other thiols were obtained
commercially (Aldrich) and were used as received.
Experimental
NMR spectra were recorded using a Bruker ADVANCE
1
500 instrument (500 MHz for H, 125.7 MHz for 13C). Unless
stated otherwise, the solvent was CDCl3 and chemical shifts are
reported relative to Me4Si; J-values are quoted in Hz and the
use of [multiplet] indicates an apparent multiplet associated
with an averaged coupling constant. Column chromatography
and TLC were carried out using Merck Kieselgel 60 (230–400
mesh) and Kieselgel 60 F254 aluminium-backed pre-coated
plates, respectively. Determination of enantiomeric excess by
HPLC was carried out using Chiralcel-OD or Chiralpak-AD
columns (4.6 mm × 250 mm; Daicel Chemical Industries Ltd.)
in conjunction with hexane–isopropyl alcohol eluent (flow rate
1 cm3 minϪ1). The proportion of alcohol in the eluent is given
in the text and UV detection was at 254 nm. Optical rotations
were measured on an AA Series Polaar 2000 polarimeter
(Optical Activity Ltd.) using a 1 dm cell and specific rotations
are given in units of 10Ϫ1 deg cm2 gϪ1.
Preparation of starting isomers and standards
cis-2,2,4,5-Tetramethyl-1,3-dioxolane 3-cis. meso-Butane-2,3-
diol (5.00 g, 56 mmol) in acetone (25 cm3) was heated under
gentle reflux for 4 h in the presence of 50% aq. sulfuric acid
(ca. 30 mm3). The acetone was removed by slow distillation at
atmospheric pressure and the residue was dissolved in diethyl
ether (50 cm3), shaken with anhydrous K2CO3 (2 g), and dried.
Distillation gave 3-cis as a clear oil, bp 117–118 ЊC/755 mmHg
(lit.,27 111–114 ЊC); δH 1.13 (6 H, d, J 5.9, Me), 1.32 (3 H, s,
Me), 1.44 (3 H, s, Me), 4.22 (2 H, m, OCH); δC 15.5, 25.7, 28.5,
73.9, 107.2.
trans-2,2,4,5-Tetramethyl-1,3-dioxolane 3-trans. This was
prepared by the procedure used for the cis-isomer, starting from
dl-butane-2,3-diol, itself prepared by hydrolytic ring opening
of cis-2,3-epoxybutane. Distillation gave 3-trans as a clear oil,
109–111 ЊC/758 mmHg (lit.27 109–112 ЊC); δH 1.23 (6 H, d, J 5.8,
Me), 1.39 (6 H, s, Me), 3.63 (2 H, m, OCH); δC 16.8, 27.3, 78.2,
107.5.
GLC analysis was carried out using an HP 6890 Series
instrument in conjunction with capillary columns (25 m ×
0.32 mm i.d.) coated with BPX5 (non-polar, 0.32 µm film) or
BP20 (polar, 0.10 µm film) stationary phases, both obtained
from SGE International Ltd. The carrier gas was helium and
the flow rate was 1.2 cm3 minϪ1.
All manipulations and reactions of air-sensitive compounds
were carried out under an atmosphere of dry argon or nitrogen
and all extracts were dried over anhydrous MgSO4, unless
stated otherwise. Petroleum spirit refers to the fraction of
distillation range 40–60 ЊC.
trans-2,2,4,6-Tetramethyl-1,3-dioxolane 4-trans. This was
prepared from (R,R)-pentane-2,4-diol according to the pub-
lished method,28 bp 135 ЊC/755 mmHg (lit.,28 135 ЊC); δH 1.19
(6 H, d, J 6.3, Me), 1.36 (6 H, s, Me), 1.59 (2 H, t, J 7.5, CH2),
3.95 (2 H, m, OCH); δC 21.7, 25.2, 41.5, 62.7, 100.0. A 53 : 47
mixture of cis- and trans-4 was prepared similarly from a
commercially available mixture of dl- and meso-pentane-2,4-
diol. The NMR spectrum of the cis-isomer showed δH 1.15
(6 H, d, J 6.3, Me), 1.40 (3 H, s, Me), 1.44 (3 H, s, Me), 1.46
(1 H, m, HA-5), 1.50 (1 H, m, HB-5), 3.94 (2 H, m, OCH);
δC 19.8, 22.2, 30.3, 40.4, 65.0, 98.3.
Materials
Anhydrous octane, nonane and 1,4-dioxane (Aldrich) were
used as received. Di-tert-butyl peroxide (Aldrich) was distilled
under reduced pressure and stored under argon in a refriger-
ator; 2,2-bis(tert-butylperoxy)butane 6 (50% w/w in mineral oil)
was obtained commercially (Laporte Organics or Aldrich) and
was used as received.
trans-1,2-Dimethoxycyclopentane 5-trans.29 The published
method29b was modified by the use of iodomethane as the
methylating agent, instead of dimethyl sulfate, in order to
facilitate isolation of the product. trans-Cyclopentane-1,2-diol
(5.00 g, 49 mmol) in dry dimethylformamide (DMF, 20 cm3)
was added dropwise to a vigorously stirred suspension of
sodium hydride (60% in mineral oil, 1.80 g, 125 mmol) in DMF
(80 cm3) cooled in an ice–water-bath. After the addition,
the mixture was allowed to warm to room temperature and
stirred vigorously until a homogeneous solution was obtained.
The solution was cooled to 0 ЊC and iodomethane (21.3 g,
150 mmol) was added dropwise with stirring. The reaction
mixture was then stirred at room temperature overnight before
being quenched with water (100 cm3) and extracted with diethyl
ether (4 × 20 cm3). The combined extracts were washed with
saturated brine (2 × 20 cm3) and dried; the diethyl ether
and unchanged iodomethane were removed by careful rotary
evaporation and the residue was distilled under reduced
pressure to give the product as a clear oil (3.5 g, 55%). Further
purification by chromatography on silica gel (petroleum spirit–
diethyl ether, 5 : 1 v/v eluent) was necessary and re-distillation
then afforded pure 5-trans, bp 63–64 ЊC/35 mmHg (lit.,29b 45 ЊC/
15 mmHg); δH 1.57 (2 H, m, CH2), 1.65 (2 H, m, CH2), 1.89
Tris(4-fluorophenyl)silanethiol. This thiol was prepared
according to the method described in the literature for similar
triarylsilanethiols.24 A mixture of tris(4-fluorophenyl)silane25
(5.0 g, 16.0 mmol) and powdered elemental sulfur (0.51 g, 16.0
mmol) in decalin (20 cm3) was stirred and heated under reflux
for 36 h. Decalin was removed by distillation under reduced
pressure (12 mmHg) and the residue was distilled to give the
thiol as an oil (3.8 g, 58%), bp 146–148 ЊC/0.02 mmHg, which
slowly solidified on storage and was then recrystallised from
hexane, mp 72–74 ЊC; δH 0.46 (1 H, s, SH), 7.11 (6 H, m, ArH),
7.57 (6 H, m, ArH); δC 115.5 (d, JCF 20.2), 129.5 (d, JCF 3.7),
137.4 (d, JCF 7.9), 164.5 (d, JCF 251.1) (Found: C, 62.4; H, 3.6; S,
9.2. C18H13F3SSi requires C, 62.4; H, 3.8; S, 9.3%).
Tri-tert-butoxysilanethiol (TBST) was prepared according to
a modification of the method described in the literature.26
Silicon disulfide (Alfa Aesar) was crushed to a fine powder
using a dry stainless steel pestle and mortar enclosed in a
polythene bag filled with nitrogen. Powdered silicon disulfide
(95% pure; 20.2 g, 0.21 mol) was charged into a 100 cm3 round-
bottomed flask containing a robust stirrer bar and equipped
with a reflux condenser. tert-Butyl alcohol (60.0 g, 0.81 mol)
J. Chem. Soc., Perkin Trans. 1, 2001, 2452–2461
2457

View Article Online
(2 H, m, CH2), 3.34 (6 H, s, OMe), 3.64 (2 H, m, OCH); δC 21.3,
29.7, 56.8, 86.3.
2,3-O-isopropylidene-β--xylopyranoside 2413 via the corre-
1
sponding xanthate, mp 95–97 ЊC (lit.,35 93–97 ЊC). The H
NMR spectrum of 27 was identical to that described in the
literature;35 δC 26.5, 26.7, 56.5, 62.9, 64.2, 76.2, 79.2, 103.4,
110.3.
cis-1,2-Dimethoxycyclopentane 5-cis. This was prepared in
60% yield from the cis-diol by the method used for the trans-
isomer; bp 70–72 ЊC/42 mmHg; δH 1.75 (6 H, m, CH2), 3.38
(6 H, s, OMe), 3.66 (2 H, m, OCH); δC 19.0, 27.2, 57.1, 82.0
(Found: C, 64.4; H, 10.9. C7H14O2 requires C, 64.6; H, 10.8%).
Methyl
4-deoxy-2,3-O-isopropylidene-ꢁ-D-erythro-pento-
pyranoside 28. A 1 : 3 mixture of methyl 2,3-O-isopropylidene-
β--ribofuranoside 30 and methyl 3,4-O-isopropylidene-β--
ribofuranoside 31 was prepared according to the published
procedure.14 All attempts to isolate pure 30 by flash chroma-
tography were unsuccessful and a mixture of the two com-
pounds was converted to a mixture of the corresponding
xanthates. Sodium hydride (60% in mineral oil; 0.55 g, 13.8
mmol) was added with stirring to a chromatographically
enriched 45 : 55 mixture of 30 and 31 (2.57 g, 12.6 mmol) and
imidazole (10 mg) in dry tetrahydrofuran (50 cm3) under argon
at 0 ЊC. The mixture was allowed to warm to room temperature
and was stirred until liberation of hydrogen ceased (ca. 25 min).
Carbon disulfide (2.51 g, 33 mmol) was added in one portion
and the mixture was stirred for a further 30 min. Iodomethane
(2.84 g, 20 mmol) was added dropwise and the mixture was then
stirred at room temperature for 1 h. The reaction mixture was
quenched with water (60 cm3), the organic layer was separated,
and the solvent was removed by rotary evaporation. The
aqueous phase was extracted with diethyl ether (3 × 15 cm3)
and the combined extracts were added to the residue from
the organic layer. The ethereal solution was washed with
saturated brine (20 cm3) and dried. The solvent was removed
by rotary evaporation and the residue was purified by flash
chromatography using petroleum spirit followed by petroleum
spirit–diethyl ether (10 : 1 then 5 : 1 v/v) as eluent. The xanthate
32 (1.48 g, 89% based on 30) was isolated from the earlier frac-
tions as a pale yellow oil; δH 1.37 (3 H, s, Me), 1.57 (3 H, s, Me),
2.59 (3 H, s, SMe), 3.45 (3 H, s, OMe), 3.85 (1 H, dd, J 11.0
and 8.1, HA-5), 3.95 (1 H, dd, J 11.0 and 5.0, HB-5), 4.11 (1 H,
dd, J 6.3 and 3.2, H-2), 4.60 (1 H, d, J 3.2, H-1), 4.63 (1 H, dd,
J 6.3 and 4.0, H-3), 6.06 (1 H, ddd, J 8.1, 5.0 and 4.0, H-4);
δC 25.5, 26.8, 56.2, 58.9, 67.9, 71.2, 74.0, 75.5, 100.6, 110.8,
215.9.
Isosorbide dimethyl ether 7. This was used as received from
Aldrich. Isomannide dimethyl ether 8 was prepared by methyl-
ation of isomannide with dimethyl sulfate in the presence of
potassium hydroxide, as described in the literature.30
endo,endo-cis-Bicyclo[3.3.0]octane-2,6-diol. This was pre-
pared according to the published method,8b bp 110–113 ЊC/0.02
mmHg (lit.,8b 130–144 ЊC/2.7 mmHg). The corresponding di-
methyl ether 118a was prepared from the diol by methylation
with dimethyl sulfate as described for isomannide dimethyl
ether,30 bp 45 ЊC/0.05 mmHg or 100–102 ЊC/15 mmHg; δH 1.40
(4 H, m, CH2), 1.80 (4 H, m, CH2), 2.55 (2 H, m, CH2), 3.29
(6 H, s, OMe), 3.72 (2 H, m, OCH); δC 20.4, 31.0, 43.0, 57.2,
84.4.
1,2:5,6-Di-O-isopropylidene-3-O-methyl-ꢀ-D-glucofuranoside
13. This was prepared according to the literature procedure;9
δH 1.32 (3 H, s, Me), 1.36 (3 H, s, Me), 1.43 (3 H, s, Me), 1.50
(3 H, s, Me), 3.45 (3 H, s, OMe), 3.77 (1 H, d, J 3.0, H-3), 4.01
(1 H, dd, J 8.6 and 5.4, HA-6), 4.08 (1 H, dd, J 8.6 and 4.5,
HB-6), 4.11 (1 H, dd, J 7.9 and 3.0, H-4), 4.30 (1 H, d[t], J 7.9
and 5.0, H-5), 4.56 (1 H, d, J 3.8, H-2), 5.86 (1 H, d, J 3.8, H-1);
δC 25.4, 26.2, 26.8, 26.9, 58.2, 67.2, 72.3, 81.0, 81.8, 83.6, 105.1,
109.0, 111.7.
3-Deoxy-1,2:5,6-di-O-isopropylidene-ꢀ-D-ribo-hexofuranose
15.31 This was prepared by reduction with triphenylsilane of the
xanthate derived from diacetone -glucose,10,32 bp 80–82 ЊC/0.5
mmHg (lit.,31 72–73 ЊC/0.2 mmHg); δH 1.32 (3 H, s, Me), 1.35
(3 H, s, Me), 1.42 (3 H, s, Me), 1.51 (3 H, s, Me), 1.77 (1 H, m,
HA-3), 2.18 (1 H, dd, J 13.7 and 4.1, HB-3), 4.13 (4 H, m, H-4,
-5 and -6), 4.75 (1 H, t, J 4.1), 5.82 (1 H, d, J 3.7, H-1); δC 25.1,
26.1, 26.4, 26.8, 35.2, 67.2, 76.7, 78.6, 80.4, 105.6, 109.6, 111.3.
The xanthate 32 (294 mg, 1.0 mmol) was treated with tri-
phenylsilane (310 mg, 1.2 mmol) and di-tert-butyl hyponitrite36
(8.4 mg, 5 mol%) in 1,4-dioxane (2 cm3) at 60 ЊC as described
previously10,32 to give 28 (155 mg, 82%) [α]2D0 Ϫ106.1 (c 3.8,
CHCl3); δH 1.36 (3 H, s, Me), 1.52 (3 H, s, Me), 1.89 (1 H, d[q],
J 14.6 and 4.0, HA-4), 2.01 (1 H, dd[t], J 14.9, 9.8 and 4.9,
HB-4), 3.46 (3 H, s, OMe), 3.66 (1 H, ddd, J 11.5, 9.8 and 3.7,
HA-5), 3.77 (1 H, d[t], J 11.5 and 4.7, HB-5), 3.86 (1 H, [t], J 5.1,
H-2), 4.38 (1 H, [q], J 4.8 H-3), 4.46 (1 H, d, J 4.7, H-1); δC 25.9,
27.3, 27.9, 56.2, 58.9, 71.5, 74.7, 101.8, 109.0 (Found: C, 57.2;
H, 8.7. C9H16O4 requires C, 57.4; H, 8.6%).
Methyl
2,3:5,6-di-O-isopropylidene-ꢀ-D-mannofuranoside
17.9,12 This was prepared according to the literature procedure;
δH 1.32 (3 H, s, Me), 1.38 (3 H, s, Me), 1.45 (3 H, s, Me), 1.46
(3 H, s, Me), 3.31 (3 H, s, OMe), 3.90 (1 H, dd, J 7.7 and 3.6,
H-4), 4.04 (1 H, dd, J 8.7 and 4.4, HA-6), 4.11 (1 H, dd, J 8.7
and 6.3, HB-6), 4.40 (1 H, ddd, J 7.7, 6.3 and 4.4, H-5), 4.56
(1 H, d, J, 5.9, H-2), 4.76 (1 H, dd, J 5.9 and 3.6, H-3), 4.87
(1 H, s, H-1); δC 24.6, 25.2, 25.9, 26.9, 54.6, 67.0, 73.2, 79.5,
80.3, 85.0, 107.4, 109.2, 112.6.
Compound 28 was also prepared by reduction of the xan-
thate 32 (2.94 g, 10.0 mmol) with tributyltin hydride to give
28 (1.77 g, 94%), bp 48–50 ЊC/0.05 mmHg.
trans- and cis-Acetonides 21 and 22. These were prepared
from trans- and cis-cyclohexane-1,2-diol, respectively, following
the published procedure.33
trans-2,2-Dimethyl-4,5-diphenyl-1,3-dioxolane 3637b,c and the
cis-isomer 37.37a These were prepared from the corresponding
diols as described in the literature.
Methyl 2,3-O-isopropylidene-4-O-methyl-ꢁ-D-xylopyranoside
25. This was prepared from 2413 by methylation with iodo-
methane, as described above for the preparation of 5,34 bp 70–
72 ЊC/0.05 mmHg; [α]2D0 Ϫ32.9 (c 0.85, CHCl3); δH 1.44 (3 H, s,
Me), 1.45 (3 H, s, Me), 3.28 (1 H, dd, J 12.0 and 7.4, HA-5), 3.33
(1 H, dd J 9.3 and 7.4, H-2), 3.47 (3 H, s, OMe), 3.52 (3 H, s,
OMe), 3.56 (1 H, dd, J 8.7 and 6.3, H-3), 3.50 (1 H, m, H-4),
4.10 (1 H, dd, J 12.0 and 4.7, HB-5), 4.53 (1 H, d, J 7.4, H-1);
δC 26.6, 26.8, 56.4, 57.8, 65.2, 76.7, 77.9, 80.2, 102.6, 111.7
(Found: C, 55.1; H, 8.2. C10H18O5 requires C, 55.0; H, 8.3%).
(4R,5S)-2,2-Dimethyl-4,5-diphenyl-N-valeryloxazolidine 38.
A mixture of (1S,2R)-2-amino-1,2-diphenylethanol (2.13 g,
10.0 mmol), aluminium powder (1.20 g, 44.4 mmol) and a
catalytic amount of acetyl chloride (20 mm3) in a mixture of
dry acetone (10 cm3) and dry dichloromethane (20 cm3) was
stirred under argon for 20 h at room temperature.38 Excess of
aluminium powder was removed by filtration and washed with
dichloromethane. The solvent was removed from the filtrate
by evaporation to give the essentially pure oxazolidine 42 as a
viscous oil (2.40 g, 95%), which was partially hydrolysed during
chromatography on silica gel; δH 1.58 (3 H, s, Me), 1.81 (3 H, s,
Methyl
4-deoxy-2,3-O-isopropylidene-ꢀ-L-threo-pento-
pyranoside 27.35 This was prepared by deoxygenation of methyl
2458
J. Chem. Soc., Perkin Trans. 1, 2001, 2452–2461

View Article Online
Me), 2.85 (1 H, br s, NH), 4.94 (1 H, d, J 7.7, H-5), 5.38 (1 H, d,
J 7.7, H-4), 6.86 (1 H, m, Ph), 6.95 (1 H, m, Ph), 7.05 (8 H, m,
Ph); δC 26.2, 27.9, 66.7, 81.8, 94.9, 126.4, 126.7, 127.0, 127.4(1),
127.4(2), 127.8, 137.4, 140.0.
Method B. A solution of the substrate (1.0 mmol) in dry
octane (1.5 cm3) containing collidine (7 mm3, 0.10 mmol),
DBPB (50% w/w in mineral oil; 20 mm3, 0.03 mmol) and TBST
(9 mm3, 0.03 mmol) was stirred under argon and heated under
reflux in an oil-bath that had been pre-heated to 145–150 ЊC.
The same amounts of DBPB and TBST were added after
30 min and again after 1 h; the total reaction time was 3 h.
Before work-up, the reaction mixture was examined by NMR
spectroscopy or GLC to determine the extent of isomerisation.
Crude 42 (1.26 g, 5.0 mmol) was dissolved in dry diethyl ether
(10 cm3) containing triethylamine (0.70 g, 6.9 mmol) and the
solution was stirred and cooled in an ice–water-bath while
valeryl chloride (0.70 g, 5.8 mmol) was added dropwise from a
syringe. The mixture was stirred at room temperature for 12 h,
and the precipitated amine hydrochloride was removed by fil-
tration and washed with diethyl ether. The solvent was removed
from the filtrate by evaporation and the residue was purified by
chromatography, eluting with petroleum spirit–diethyl ether
(from 10 : 1 then 5 : 1 then 5 : 2 v/v), to give 38 as an oil (1.24 g,
74%) which was crystallised from hexane at Ϫ10 ЊC, mp 88–89
ЊC; [α]2D0 ϩ114.3 (c 1.05, CHCl3). The ee was determined by
HPLC to be >99.5%, using the OD column with hexane–
isopropyl alcohol (99 : 1) as eluent (tR 11.9 min); δH 0.74 (3 H, t,
J 7.3 CH2Me), 1.14 (2 H, m, CH2), 1.45 (2 H, m, CH2), 1.80 (3
H, s, Me), 1.91 [1 H, ddd, J 15.5, 8.6 and 6.3, HACHC(O)], 2.06
(3 H, s, Me), 2.17 [1 H, ddd, J 15.5, 9.7 and 6.3, HBCHC(O)],
5.05 (1 H, d, J 6.3, H-4 or -5), 5.54 (1 H, d, J 6.3, H-5 or -4),
6.90 (1 H, m, Ph), 6.99 (1 H, m, Ph), 7.11 (8 H, m, Ph); δC 13.7,
22.2, 23.4, 25.9, 26.7, 35.4, 66.4, 80.1, 95.1, 126.8, 127.3, 127.4,
127.7; (2-C), 127.9, 135.4, 137.6, 171.1 (Found: C, 78.1; H, 8.2;
N, 4.1. C22H27NO2 requires C, 78.3; H, 8.1; N, 4.2%).
Method C. A solution of the substrate (1.0 mmol) in dry
nonane (1.5 cm3) containing collidine (7 mm3, 0.10 mmol),
di-tert-butyl peroxide (58 mm3, 0.2 mmol) and TBST (9 mm3,
0.03 mmol) was stirred under argon and heated under reflux in
an oil-bath that had been pre-heated to 165–170 ЊC. The same
amounts of DBPB and TBST were added after 30 min and
again after 1 h; the total reaction time was 3 h. Before work-up,
the reaction mixture was examined by NMR spectroscopy or
GLC to determine the extent of isomerisation.
Results of epimerisation experiments
Epimerisation of the dioxolanes 3 was carried out under the
conditions of method A, except that nonane was used as the
solvent to avoid interference by octane in the GLC analysis.
After 1 h, the cis : trans ratio reached 16 : 84 and this remained
unchanged when further amounts of DBPB and TBST (5 mol%
of each) were added and heating was continued. The epimeris-
ation of 4 and of 5 was carried out using methods A and B,
respectively.
(4S,5R)-2,2-Dimethyl-4,5-diphenyl-N-valeryloxazolidine 39.
This was obtained in the same way from (1R,2S)-2-amino-1,2-
diphenylethanol, mp 88–89 ЊC; [α]2D0 Ϫ113.7 (c 1.10, CHCl3), ee
> 99.5%, tR 16.4 min (Found: C, 78.4; H, 8.2; N, 4.1. C22H27NO2
requires C, 78.3; H, 8.1; N, 4.2%).
Di-O-methyl-1,4 : 3,6-dianhydro-L-iditol 9 (isoidide dimethyl
ether). An octane solution of di-O-methyl-1,4 : 3,6-dianhydro-
-glucitol 7 (174 mg, 1.0 mmol) was subjected to the conditions
of method A. GLC analysis of the reaction mixture, using the
BP20 column at 175 ЊC (tR for 7, 8 and 9 was 9.7, 10.9 and 8.7
min, respectively), showed the proportions of the ethers 7 : 8 : 9
to be 38.5 : 1.5 : 60.0. The epimer 9 was isolated as an oil
by flash chromatography using petroleum spirit–diethyl ether
(10 : 1) as eluent, [α]D17 Ϫ8.0 (c 8.3, CHCl3); δH 3.39 (6 H, s,
2 Me), 3.80–3.88 (6 H, m, Hendo-1,-6, Hexo-1,-6, Hendo-2,-5), 4.58
(2 H, s, H-3,-4); δC 57.2, 71.8, 84.8, 85.0 (Found: C, 54.9;
H, 8.0. C8H14O4 requires C, 55.2; H, 8.1%).
(4R,5R)-2,2-Dimethyl-4,5-diphenyl-N-valeryloxazolidine 40.
This was prepared from (1R,2R)-2-amino-1,2-diphenylethanol,
itself prepared in an overall yield of 70% from (1S,2R)-
2-amino-1,2-diphenylethanol according to the published
method.20 The acetonide was obtained as described above for
the cis-isomer; δH 1.64 (3 H, s, Me), 1.65 (3 H, s, Me), 2.80 (1 H,
br s, NH), 4.24 (1 H, d, J 8.6, H-4 or -5), 4.80 (1 H, d, J 8.6, H-5
or -4), 7.28 (10 H, m, Ph); δC 28.4(5), 28.4(7), 70.0, 85.9, 95.4,
126.2, 127.2, 127.6, 127.9, 128.3, 128.7, 137.7, 139.7.
This crude oxazolidine was acylated to give 40 as colourless
needles, mp 75–76 ЊC; [α]D20 ϩ22.7 (c 1.78, CHCl3), ee > 99.5%,
determined by HPLC using the AD column and eluting with
hexane–isopropyl alcohol (99 : 1, tR 8.9 min); δH 0.70 (3 H, t,
J 7.3, CH2Me), 1.02 (2 H, m, CH2), 1.26 (1 H, m, CHAH), 1.45
(1 H, m, CHBH), 1.61 [1 H, ddd, J 14.9, 8.7 and 5.8,
HACHC(O)], 1.85 [1 H, ddd, J 14.9, 8.7 and 6.7, HBCHC(O)],
1.89 (6 H, s, CMe2), 4.59 (1 H, d, J 8.3, H-4), 4.85 (1 H, d, J 8.3,
H-4), 7.10 (1 H, m, Ph), 7.20 (1 H, m, Ph), 7.33 (8 H, m, Ph);
δC 13.7, 22.2, 24.7, 26.7 (2 C), 36.3, 70.2, 85.4, 96.7, 126.5,
126.7, 128.1, 128.5, 128.6, 129.1, 136.8, 138.9, 171.6 (Found:
C, 78.0; H, 8.1; N, 4.1. C22H27NO2 requires C, 78.3; H, 8.1; N,
4.2%).
1,2 : 5,6-Di-O-isopropylidene-3-O-methyl-ꢁ-L-idofuranose 14.
This was obtained in 32% conversion (NMR) by epimerisation
of 13 (274 mg, 1.0 mol) under the conditions of method A.
Isolation by flash chromatography using petroleum spirit–
diethyl ether (10 : 1 then 5 : 1) as eluent gave 14 as a clear oil (70
mg, 25%), [α]2D0 Ϫ67.5 (c 0.85, CHCl3); δH 1.32 (3 H, s, Me), 1.37
(3 H, s, Me), 1.45 (3 H, s, Me), 1.50 (3 H, s, Me), 3.36 (3 H,
s, OMe), 3.62 (1 H, dd, J 8.1 and 7.6, HA-6), 3.64 (1 H, d, J 3.9,
H-3), 4.10 (1 H, dd, J 8.1 and 6.6, HB-6), 4.17 (1 H, dd, J 8.1
and 3.9, H-4), 4.35 (1 H, ddd, J 8.1, 7.6 and 6.6, H-5), 4.57 (1 H,
d, J 3.9, H-2), 5.97 (1H, d, J 3.9, H-1); δC 25.3, 26.3, 26.8, 26.9,
57.5, 65.8, 75.1, 81.3, 82.3, 84.9, 105.5, 109.9, 111.9 (Found: C,
57.1; H, 8.2. C13H22O6 requires C, 56.9; H, 8.1%).
Conditions for radical-chain epimerisation
3-Deoxy-1,2 : 5,6-di-O-isopropylidene-ꢁ-L-lyxo-hexofuranose
16. Epimerisation of 15 (246 mg, 1.0 mmol) was carried out
using method A, with the modification that further portions of
TBST and DBPB (5 mol% of each) were added after 1 and 2 h.
GLC analysis (BP20 column at 200 ЊC, tR 10.6 min for 15 and
11.5 min for 16) showed the final ratio 15 : 16 to be 68 : 32 and
no evidence was found for the C-4 epimer. The epimer 16
(65 mg, 28%) was isolated by flash chromatography using
petroleum spirit–diethyl ether (10 : 1 then 5 : 1) as eluent and
then was recrystallised from hexane; mp 57–58 ЊC; [α]2D0 ϩ22.5
(c 1.47, CHCl3); δH 1.32 (3 H, s, Me), 1.37 (3 H, s, Me), 1.43
(3 H, s, Me), 1.52 (3 H, s, Me), 1.72 (1 H, ddd, J 13.3, 10.7 and
Most reactions were carried out using one of the following
three procedures.
Method A. A solution of the substrate (1.0 mmol) in dry
octane (1.5 cm3) containing collidine (7 mm3, 0.10 mmol),
DBPB (50% w/w in mineral oil; 34 mm3, 0.05 mmol) and TBST
(14 mm3, 0.05 mmol) was stirred under argon and heated in
an oil-bath, the temperature of which had been adjusted
previously to 125 ЊC. The same amounts of DBPB and TBST
were added after 1 h and the progress of the reaction was moni-
tored by GLC for 2–3 h, until no further change occurred.
J. Chem. Soc., Perkin Trans. 1, 2001, 2452–2461
2459

View Article Online
4.9, Hexo-3), 2.02 (1 H, dd, J 13.3 and 4.6, Hendo-3), 3.80 (1 H, dd,
J 8.3 and 6.6, H-6), 4.02 (1 H, dd, J 8.3 and 6.9, HЈ-6), 4.13
(1 H, ddd, J 7.0, 6.6 and 5.3, H-5), 4.26 (1 H, ddd, J 10.7, 5.3
and 4.6, H-4), 4.73 (1 H, dd, J 4.9 and 3.7, H-2), 5.83 (1H, d,
J 3.7, H-1); δC 25.5, 26.2, 26.3, 26.8, 34.6, 65.6, 76.8, 78.0, 80.3,
105.8, 109.8, 111.4 (Found: C, 59.0; H, 8.2. C12H20O5 requires
C, 59.0; H, 8.3%). For the -enantiomer,11b mp 60–61 ЊC;
[α]1D8 Ϫ24.7 (c 1.90, CHCl3) have been reported.
4.1 and 3.4, H-3), 4.22 (1 H, d, J 3.3, H-1); δC (in CD3OD) 30.5,
55.6, 59.9, 67.1, 71.1, 103.4.
trans-2,2-Dimethyl-4,5-diphenyl-1,3-dioxolane 36. This was
obtained by epimerisation of the cis-isomer 37 (254 mg,
1.0 mmol) using method B, except that 2,4,6-tris(trifluoro-
methyl)thiophenol or pentafluorothiophenol was used as the
catalyst and collidine was not used, because salt formation
is essentially complete with these strongly acidic thiols. The
solvent was removed by evaporation, and NMR analysis
showed that conversion was complete (≥98%). The trans-isomer
36 (228 mg, 90%) was readily isolated by recrystallisation from
petroleum spirit, mp 47–48 ЊC, (lit.,37a 46–48 ЊC); δH 1.69 (6 H, s,
2 Me), 4.76 (2 H, s, OCH), 7.20–7.33 (10 H, m, Ph); δC 27.2,
85.4, 109.4, 126.7, 128.2, 128.4, 136.8.
The trans-isomer 36 obtained from the above reaction was
dissolved in methanol (5 cm3) and Amberlyst-15 (10 mg)
was added. The mixture was stirred under reflux for 6 h, the
resin was removed by filtration, the solvent was removed from
the filtrate under reduced pressure and the residue was
recrystallised from dichloromethane–petroleum spirit to give
dl-hydrobenzoin 35 as colourless crystals (180 mg, 84%). The
mp (118–120 ЊC) and NMR spectra were identical with those of
the commercially available (Aldrich) authentic compound.
Methyl 2,3 : 5,6-di-O-isopropylidene-ꢁ-L-gulofuranoside 18.39
This was obtained from 17 using method B in 30% conversion
(GLC, BPX5 column) at 170 ЊC, tR11.4 min for 17 and 14.0 min
for 18. The epimer 18 was isolated by flash chromatography,
using petroleum spirit–diethyl ether (10 : 1 then 5 : 1) as eluent,
as a clear oil (66 mg, 24%) which was crystallised from hexane,
mp 77–78.5 ЊC (lit.,39 76–77 ЊC); δH 1.28 (3 H, s, Me), 1.40 (3 H,
s, Me), 1.45 (3 H, s, Me), 1.46 (3 H, s, Me), 3.37 (3 H, s, OMe),
3.71 (1 H, dd, J 8.4 and 7.1, HA-6), 3.91 (1 H, dd, J 8.4 and 3.9,
H-4), 4.20 (1 H, dd, J 8.4 and 6.5, HB-6), 4.38 (1 H, ddd, J 8.4,
7.1 and 6.5, H-5), 4.57 (1 H, d, J 5.9, H-2), 4.64 (1 H, dd, J 5.9
and 3.9, H-3), 4.97 (1 H, s, H-1); δC 24.8, 25.3, 25.9, 26.8, 54.8,
66.0, 75.5, 79.8, 81.9, 85.1, 107.3, 109.7, 112.8. The structure
was confirmed by single-crystal X-ray diffraction and is shown
in Fig. 1.
Epimerisation of the trans-acetonide 21 to give the cis-
isomer 22 (156 mg, 1.0 mmol) was carried out using method B;
the cis : trans ratio at equilibrium was 95 : 5. Octane was
removed by careful rotary evaporation, the residual oil
was taken up in methanol (2 cm3) and Amberlyst-15
acidic ion-exchange resin (15 mg) was added. The mixture
was stirred overnight at room temperature, then filtered to
remove the catalyst. Methanol was removed from the filtrate
by evaporation, and recrystallisation of the residue from
ethyl acetate gave cis-cyclohexane-1,2-diol 20 (98 mg, 85%), mp
(4R,5R)-2,2-Dimethyl-4,5-diphenyl-N-valeryloxazolidine 40.
(4R,5S)-2,2-Dimethyl-4,5-diphenyl-N-valeryloxazolidine
38
(337 g, 1.0 mmol) was subjected to the conditions of method C,
except that pentafluorothiophenol (6 mm3, 0.05 mmol) was
used as catalyst and collidine was absent; repeat additions of
thiol were made after 20 and 40 min. After 2.5 h the ratio 38 : 40
was 27 : 73 (by NMR). Nonane was removed at room temper-
ature under reduced pressure (0.02 mmHg) and 40 (235 mg,
70%) was isolated by flash chromatography, using petroleum
spirit–diethyl ether (10 : 1 then 5 : 1 then 5 : 2) as eluent; the ee
of the product was 91% [OD column, tR 8.9 min using hexane–
isopropyl alcohol (99 : 1) as eluent]. Recrystallisation from
hexane gave enantiomerically pure 40 as colourless needles,
mp 75–76 ЊC, [α]2D0 ϩ22.5 (c 1.10, CHCl3); δH 0.70 (3 H, t, J 7.3,
Me), 1.02 (2 H, m, CH2), 1.26 (1 H, m, CHAH), 1.45 (1 H, m,
CHBH), 1.61 [1 H, ddd, J 14.9, 8.7 and 5.8, HACHC(O)], 1.85
[1 H, ddd, J 14.9, 8.7 and 6.7, HBCHC(O)], 1.89 (6 H, s, CMe2),
4.59 (1 H, d, J 8.3, H-4 ), 4.85 (1 H, d, J 8.3, H-4), 7.10 (1 H, m,
Ph), 7.20 (1 H, m, Ph), 7.33 (8 H, m, Ph); δC 13.7, 22.2, 24.7,
26.7 (2-C), 36.3, 70.2, 85.4, 96.7, 126.5, 126.7, 128.1, 128.5,
128.6, 129.1, 136.8, 138.9, 171.6 (Found: C, 78.3; H, 8.1; N, 4.1.
C22H27NO2 requires C, 78.3; H, 8.1; N, 4.2%).
1
100–101 ЊC, which showed H and 13C NMR spectra identical
with those of the authentic compound.
Methyl 2,3-O-isopropylidene-4-O-methyl-ꢁ-D-lyxopyranoside
26. This was obtained in 30% conversion (NMR) by epimeris-
ation of methyl 2,3-O-isopropylidene-4-O-methyl-β--xylo-
pyranoside 25 (218 mg, 1.0 mmol) using method C. Isolation
by flash chromatography using petroleum spirit–diethyl ether
(10 : 1 then 5 : 1) as eluent gave 26 as a clear oil (55 mg, 25%)
that still contained a small amount (2–3%) of 25, δH 1.39 (3 H,
s, Me), 1.56 (3 H, s, Me), 3.47 (3 H, s, OMe), 3.48 (3 H, s, OMe),
3.64 (1 H, [t], J 10.2, Hax-5), 3.75 (1 H, ddd, J 10.0, 5.4 and 3.5,
H-4), 3.90 (1 H, ddd, J 10.4, 5.4 and 0.9, Hax-5), 3.99 (1 H, dd,
J 6.0 and 4.6, H-2), 4.40 (1 H, d, J 4.6, H-1), 4.58 (1 H, ddd,
J 6.0, 3.5 and 0.9, H-3); δC 25.7, 27.4, 56.6, 57.6, 60.7, 72.2, 73.3,
76.1, 102.0, 110.5 (Found: C, 55.2; H, 8.2. C10H18O5 requires C,
55.0; H, 8.3%).
(4S,5S)-2,2-Dimethyl-4,5-diphenyl-N-valeryloxazolidine 41.
This was produced by epimerisation of 39 following the pro-
cedure used to obtain 40. Before recrystallisation, the product
(240 mg, 71%) showed an ee of 92% [OD column, tR 8.1 min
using hexane–isopropyl alcohol (99 : 1) eluent]. The enantio-
merically pure product showed mp 75–76 ЊC, [α]2D0 Ϫ22.7 (c 0.95,
Methyl
4-deoxy-2,3-O-isopropylidene-ꢁ-D-erythro-pento-
pyranoside 28. The epimerisation of 27 (188 mg, 1.0 mmol) was
carried out using method B to give >98% conversion to 28 (by
NMR). The latter was isolated as an oil (175 mg, 93%) by flash
chromatography using petroleum spirit–diethyl ether (5 : 1 v/v)
as eluent. It showed NMR spectra identical with those of the
authentic compound, prepared as described above.
1
CHCl3). The H and 13C NMR spectra were indistinguishable
from those of 40 (Found: C, 78.2; H, 8.1; N, 4.1. C22H27NO2
requires C, 78.3; H, 8.1; N, 4.2%).
X-Ray crystallography§
Data were collected on a Nicolet R3mV diffractometer at 20 ЊC
using graphite-monochromated Mo-Kα radiation. Three stand-
ard reflections were monitored throughout the data collection
and these showed no variation with time. The data were cor-
rected for Lorentz and polarisation effects. The structures were
solved by direct methods (SHELXS-86)41 and developed using
Methyl 4-deoxy-ꢁ-D-erythro-pentopyranoside 29. This was
prepared by deprotection of 28. Compound 28 (200 mg,
1.06 mmol) was stirred in methanol (3 cm3) containing
Amberlyst-15 (10 mg) for 20 h at room temperature. The resin
was removed by filtration, and evaporation of the filtrate under
reduced pressure gave essentially pure 29 (138 mg, 93%) as a
viscous oil; [α]D20 Ϫ95.5 (c 1.24, H2O) {lit.,40 [α]1D5 Ϫ95.9 (c 1.0,
H2O)}; δH (in CD3OD) 1.25 (1 H, dq, J 13.0 and 4.7, HA-4), 1.54
(1 H, ddd, J 13.0, 9.9 and 4.1, HB-4), 3.02 (3 H, s, OMe), 3.20
(1 H, [t], J 3.2, H-2), 3.35 (2 H, m, H2-5), 3.55 (1 H, ddd, J 9.9,
§ CCDC reference number(s) 163544. See http://www.rsc.org/suppdata/
p1/b1/b103558b/ for crystallographic files in .cif or other electronic
format.
2460
J. Chem. Soc., Perkin Trans. 1, 2001, 2452–2461

View Article Online
1079; (c) B. P. Roberts and A. J. Steel, J. Chem. Soc., Perkin Trans. 2,
1994, 2411.
alternating cycles of least-squares refinement and difference-
fourier synthesis (SHELXL-93).42 Non-hydrogen atoms were
refined anisotropically, while hydrogen atoms were placed in
idealised positions and assigned a common isotropic thermal
parameter.
16 (a) H.-S. Dang and B. P. Roberts, J. Chem. Soc., Perkin Trans. 1,
1998, 67; (b) G. E. Carr, R. D. Chambers and T. F. Holmes,
J. Organomet. Chem., 1987, 325, 13.
17 CRC Handbook of Chemistry and Physics, CRC Press, Boca Raton,
78th edn., 1997.
18 (a) F. G. Bordwell, X.-M. Zhang, A. V. Satish and J.-P. Cheng, J. Am.
Chem. Soc., 1994, 116, 6605; (b) W. M. Nau, J. Org. Chem., 1996, 61,
8312; (c) W. M. Nau, J. Phys. Org. Chem., 1997, 10, 445.
19 K. N. Dalby and W. P. Jencks, J. Chem. Soc., Perkin Trans. 2, 1997,
1555.
20 (a) J. Weijlard, K. Pfister, E. F. Swanezy, C. A. Robinson and
M. Tishler, J. Am. Chem. Soc., 1951, 73, 1216; (b) F. Effenberger,
B. Gutterer and T. Ziegler, Liebigs Ann. Chem., 1991, 269;
(c) R. Lou, A. Mi, Y. Jiang, Y. Qin, Z. Li, F. Fu and A. S. C. Chan,
Tetrahedron, 2000, 56, 5857.
Crystal data for methyl 2,3 : 5,6-di-O-isopropylidene-ꢁ-L-
gulofuranoside 18. C13H22O6, M = 274.3, orthorhombic, space
group P212121, a = 5.371(1), b = 10.704(2), c = 25.652(5) Å, V =
1474 Å3 (by least-squares refinement of diffractometer angles
for 28 reflections in the range 17 < 2θ < 26Њ, λ = 0.71073 Å), Z =
4, F(000) = 592, Dc = 1.24 g cmϪ3, µ(Mo-Kα) = 0.9 cmϪ1, colour-
less needle 0.76 × 0.22 × 0.16 mm. Full matrix least-squares
refinement on 173 parameters gave R = 0.0483 (Rw = 0.1230) for
1154 independent reflections [I > 2σ(I )] and R = 0.0691 (Rw =
0.1468) for all 1534 independent reflections in the range 5 ≤ 2θ ≤
50Њ. The final electron density map was featureless with the
largest peak 0.21 e ÅϪ3.
21 G. A. Russell, in Free Radicals ed. J. K. Kochi, Wiley-Interscience,
New York, 1973, vol. 1, pp. 293–298.
22 N. S. Isaacs, Physical Organic Chemistry, Longman, Harlow, 2nd
edn., 1995, p. 152.
23 P. E. Elford and B. P. Roberts, J. Chem. Soc., Perkin Trans. 2, 1998,
1413.
24 B. Becker and W. Wojnowski, Synth. React. Inorg. Met.-Org. Chem.,
1982, 12, 565.
Acknowledgements
25 V. E. Fisher, G. Schott and A. D. Petrow, J. Prakt. Chem., 1963, 21,
149.
We acknowledge support from EPSRC for this work.
26 V. R. Piekos and W. Wojnowski, Z. Anorg. Allg. Chem., 1962, 318,
212.
References
27 B. N. Blackett, J. M. Coxon, M. P. Hartshorn, A. J. Lewis,
G. R. Little and G. J. Wright, Tetrahedron, 1970, 26, 1311.
28 G. M. Kellie and F. G. Riddell, J. Chem. Soc. B, 1971, 1030;
G. M. Kellie and F. G. Riddell, J. Chem. Soc., Perkin Trans. 2, 1975,
740.
29 (a) S. Uemura and S. Fukuzawa, J. Chem. Soc., Perkin Trans. 1,
1985, 471; (b) D. Amurrio, K. Khan and E. P. Kundig, J. Org. Chem.,
1996, 61, 2258.
1 H.-S. Dang and B. P. Roberts, Tetrahedron Lett., 1999, 40, 4271.
2 B. P. Roberts, Chem. Soc. Rev., 1999, 28, 25.
3 B. P. Roberts and A. J. Steel, J. Chem. Soc., Perkin Trans. 2, 1994,
2155; B. P. Roberts, J. Chem. Soc., Perkin Trans. 2, 1996, 2719.
4 H.-S. Dang and B. P. Roberts, Tetrahedron Lett., 2000, 41, 8595.
5 Y. Cai and B. P. Roberts, Chem. Commun., 1998, 1145 (corrigendum
p. 1607).
6 Y. Cai and B. P. Roberts, J. Chem. Soc., Perkin Trans. 1, 1998, 467
(corrigendum p. 3653).
30 D. Achet, D. Rocrelle, I. Murengezi, M. Delmas and A. Gaset,
Synthesis, 1986, 642.
31 S. Iacono and J. R. Rasmussen, Org. Synth., 1986, 64, 57.
32 S. J. Cole, J. N. Kirwan, B. P. Roberts and C. R. Willis, J. Chem. Soc.,
Perkin Trans. 1, 1991, 103.
33 (a) R. U. Lemieux and J. W. Lown, Can. J. Chem., 1964, 42, 893;
(b) J. E. McMurry and J. G. Rico, Tetrahedron Lett., 1989, 30, 1169.
34 U. E. Diner, F. Sweet and R. K. Brown, Can. J. Chem., 1966, 44,
1591.
35 L. E. Rabow, J. Stubbe and J. W. Kozarich, J. Am. Chem. Soc., 1990,
112, 3196.
36 (a) H. Kiefer and T. G. Traylor, Tetrahedron Lett., 1966, 6163;
(b) G. D. Mendenhall, Tetrahedron Lett., 1983, 24, 451.
37 (a) D. H. R. Barton, P. D. Magnus, G. Smith, G. Strecket and
D. Zurr, J. Chem. Soc., Perkin Trans. 1, 1972, 542; (b) E. J. Enholm
and J. P. Schulte, J. Org. Chem., 1999, 64, 2610; (c) C. Rosini,
S. Scammzzi, M. P. Focati and P. Salvadori, J. Org. Chem., 1995,
60, 8289.
38 R. Bureau, J. Motier and M. Joucla, Bull. Soc. Chim. Fr., 1993, 130,
584.
39 M. E. Evans and F. W. Parrish, Carbohydr. Res., 1973, 28, 359.
40 T. Kinoshita, N. Nakamura and T. Miwa, Carbohydr. Res., 1982,
102, 298.
7 K. E. Gilbert, PCMODEL version 7.5, Serena Software,
Bloomington, IN 47402–3076, USA. The GMMX conformational
search routine was used to find the most stable conformation of each
molecule. Some of the numerical energy differences reported herein
differ slightly from those reported in refs. 1 and 4, which were
obtained using version 7.0 of this program.
8 (a) P. K. Wong and J. K. Stille, J. Organomet. Chem., 1974, 70, 121;
(b) J. Perard-Viret and A. Rassat, Tetrahedron: Asymmetry, 1994, 5,
1; (c) M. A. Djadchenko, K. K. Pivnitsky, F. Thiel and H. Schink,
J. Chem. Soc. Perkin Trans. 1, 1989, 2001.
9 M. Bessodes, J. Shamsazer and K. Antonakis, Synthesis, 1988, 560.
10 Y. Cai and B. P. Roberts, Tetrahedron Lett., 2001, 42, 763.
11 (a) C. R. Haylock, L. D. Melton, K. N. Slessor and A. S. Tracey,
Carbohydr. Res., 1971, 16, 375; (b) R. J. Conway, J. P. Nagel,
R. V. Stick and D. M. G. Tibrook, Aust. J. Chem., 1985, 38, 939.
12 (a) M. H. Randall, Carbohydr. Res., 1969, 11, 173; (b) J. Leroux and
A. S. Perlin, Carbohydr. Res., 1978, 67, 163.
13 R. F. Helm, J. Ralph and L. Anderson, J. Org. Chem., 1991, 56,
7015.
14 N. A. Hughes and C. D. Maycock, Carbohydr. Res., 1974, 35, 247.
15 (a) W. K. Busfield, I. D. Grice, I. D. Jenkins and M. J. Monteiro,
J. Chem. Soc., Perkin Trans. 2, 1994, 1071; (b) W. K. Busfield,
I. D. Grice and I. D. Jenkins, J. Chem. Soc., Perkin Trans. 2, 1994,
41 G. M. Sheldrick, SHELXS-86, University of Göttingen, 1990.
42 G. M. Sheldrick, SHELXL-93, University of Göttingen, 1993.
J. Chem. Soc., Perkin Trans. 1, 2001, 2452–2461
2461