Radical-chain functionalisation at C-H centres using an O-oxiranylcarbinyl O-silyl ketene acetal

Article abstract of DOI:10.1016/j.tetlet.2004.03.165

A readily prepared O-oxiranylcarbinyl O-silyl ketene acetal has been used to bring about functionalisation at various types of saturated C-H group, through the intermediacy of an oxiranylcarbinyl radical that undergoes rapid ring opening to give a highly reactive allyloxyl radical.

Full text of DOI:10.1016/j.tetlet.2004.03.165

Tetrahedron
Letters
Tetrahedron Letters 45 (2004) 4405–4409
Radical-chain functionalisation at C–H centres using an
O-oxiranylcarbinyl O-silyl ketene acetal
Yudong Cai, Hai-Shan Dang and Brian P. Roberts*
Department of Chemistry, University College London, 20 Gordon Street, London WC1H 0AJ, UK
Received 30 January 2004; accepted 26 March 2004
Abstract—A readily prepared O-oxiranylcarbinyl O-silyl ketene acetal has been used to bring about functionalisation at various
types of saturated C–H group, through the intermediacy of an oxiranylcarbinyl radical that undergoes rapid ring opening to give a
highly reactive allyloxyl radical.
ꢀ 2004 Elsevier Ltd. All rights reserved.
The ring-opening rearrangement of an oxiranylcarbinyl
radical to give an allyloxyl radical, the simplest example
of which is shown in Eq. 1, provides a useful means of
converting a carbon-centred radical into an oxygen-
centred species that is a more potent abstractor of
hydrogen atoms from C–H groups.1 This type of rear-
rangement, which has often been exploited in organic
synthesis,1 takes place much more rapidly than ring
opening of the equivalent cyclopropylcarbinyl radical,
where the oxygen atom is replaced by the isoelectronic
methylene group.2 We have reported recently that the O-
cyclopropylcarbinyl O-tert-butyldimethylsilyl ketene
acetal 1 can be used to bring about radical [3+2]
annulation of an electron-poor alkene, as generalised in
Eq. 2(EWG ¼ electron-withdrawing group).3 A key
step in the chain mechanism of this reaction involves
ring opening of the cyclopropylcarbinyl radical 2 to give
the but-3-enyl radical 3 (Eq. 3), which then adds to the
electron-poor alkene.3;4
O
ð1Þ
O
TBSO
O
2
EWG
O EWG
OTBS
EWG
1
ð2Þ
ð3Þ
2
3
dimethylsilyl ketene acetal 4, which was prepared as
shown in Scheme 1.6 Our main reasons for choosing 4
were, (i) the ready availability of isophorone oxide 5,7
(ii) the stereoelectronic predisposition of the derived
oxiranylcarbinyl radical 8 to undergo rapid regioselec-
tive ring opening to give the allyloxyl radical 9 and (iii)
the inability of this allyloxyl radical to rearrange by
intramolecular 1,5-hydrogen-atom transfer from the
sterically-inaccessible allylic C–H groups. The first type
of reaction we envisaged applies to functionalisation at a
C–H centre with an attached electron-withdrawing
substituent and the proposed chain propagation cycle is
generalised in Scheme 2. For a similar thermodynamic
driving force and similar steric effects, a relatively elec-
trophilic radical of the type 11 that is produced by
hydrogen-atom abstraction from this class of substrate,
should add more rapidly to the electron-rich double
bond in 4 than will a nucleophilic radical.8 However, the
It occurred to us that an oxiranylcarbinyl analogue of 1
might provide a useful source of allyloxyl radicals that
could react subsequently in a number of ways, in par-
ticular by direct or catalysed5 abstraction of hydrogen
from a C–H group in a suitable substrate molecule,
leading ultimately to functionalisation at this site. In
order to explore this possibility, we chose to investigate
radical reactions of the O-oxiranylcarbinyl O-tert-butyl-
Keywords: Epoxides; Radicals and radical reactions; Addition
reactions; Fragmentation reactions; Polarity-reversal catalysis.
* Corresponding author. Fax: +20-7679-7463; e-mail: b.p.roberts@
ucl.ac.uk
0040-4039/$ - see front matter ꢀ 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetlet.2004.03.165

4406
Y. Cai et al. / Tetrahedron Letters 45 (2004) 4405–4409
dimethyl malonate although, presumably because of the
O
OH
weaker B–H bond in this complex compared with that in
the amine–borane,14 it was slightly less effective (see
Table 1, entry 1). A number of other esters containing
electron-poor a-C–H groups reacted with 4 in the
presence of a ligated borane according to Scheme 2and
the results are summarised in Table 1. For the corre-
sponding functionalisation of ethyl acetate, in which the
a-C–H bonds are relatively strong,14 the phosphine–
borane was ineffective as a polarity-reversal catalyst
(entry 4).
i
O
O
5
6
7
ii
OTBS
O
O
O
iii
1
O
O
5
3
4
O
Scheme 1. Reagents and conditions: (i) MeLi in Et2O–THF, 1 h at
)78 ꢁC, 20% aq NH4Cl quench (cf. Ref. 10); (ii) Ac2O/Et3N + 20 mol %
DMAP, 20 h at rt; (iii) LDA in THF–hexanes at )78 ꢁC, then TBSCl
followed by 1 equiv HMPA at )78 ꢁC, then rt for 12h.
4
MeO2C
CO2Me
CO2Me
1
NMe
3'
3
MeO2C
O
12
13
direct abstraction of hydrogen from the substrate by the
allyloxyl radical 9 to give 11 will be disfavoured by polar
effects, because the allyloxyl radical is also electro-
philic.5;9 To overcome this problem, a hydridic polarity-
reversal catalyst5 (an amine– or phosphine–borane
complex) is included to mediate the desired hydrogen-
atom transfer, as shown in Scheme 2.
On the other hand, if electron-donating groups (EDGs,
which include simple alkyl groups) are attached to the
C–H centre in the substrate, polar effects5;9 will now
favour the direct abstraction of hydrogen by the elec-
trophilic allyloxyl radical 9 to give a relatively nucleo-
philic carbon radical 14. In contrast, polar effects now
disfavour the addition of 14 to the electron-rich ketene
acetal 4 and so the hydridic polarity-reversal catalyst
was replaced by a stoichiometric amount of an electron-
poor alkene, in order to convert 14 to a more electro-
philic addendum radical 15, as shown in the propaga-
tion cycle generalised in Scheme 3.
When a benzene solution containing the ketene acetal
4 (1.0 mmol) and dimethyl malonate (3.0 mmol) was
heated at 70 ꢁC for a total of 2.5 h, in the presence of
di-tert-butyl hyponitrite11 (TBHN; three additions of
0.05 mmol made initially and after 20 and 40 min) as a
1
thermal source of tert-butoxyl radicals, H NMR spec-
troscopic analysis showed that ca. 95% of 4 remained
unchanged with only ca. 5% converted to the allylic
alcohol 10. However, repetition of this experiment in
the additional presence of quinuclidine–borane12
(0.20 mmol) resulted in the complete consumption of 4
and the trimethyl ester 12 was isolated in 87% yield,
after treatment of the initially-formed silyl ester with
tetrabutylammonium fluoride (TBAF) and methyl
iodide.13 In the presence of the amine–borane, but
without the TBHN initiator, no reaction took place.
Tributylphosphine–borane also functioned as a hydridic
polarity-reversal catalyst for the reaction of 4 with
Functionalisation at C–H centres with attached elec-
tron-donating groups proceeded as anticipated accord-
ing to Scheme 3, although the isolated yields of methyl
esters were only moderate; the results are included in
Table 1. For example, reaction of 4 with excess cyclo-
hexane in the presence of TBHN as initiator and N-
methylmaleimide (NMM) as the electron-poor alkene,
afforded the pyrrolidine-2,5-dione 13 in 57% yield (entry
6).17 Analogous functionalisation of cyclooctane, ada-
mantane and diisopropyl ether was also successful
using NMM as the electron-poor alkene. Adamantane
O
O
8
9
OTBS
L
BH3
EWG
R2
O
C
R1
OH
10
BH2
OTBS
O
EWG
R2
R1
R2
EWG
L
R1
R1
O
4
L
BH3
11
EWG
C
H
R2
Scheme 2.

Y. Cai et al. / Tetrahedron Letters 45 (2004) 4405–4409
4407
Table 1. Functionalisation at C–H centres using the O-oxiranylcarbinyl O-tert-butyldimethylsilyl ketene acetal 4
Class of C–H group
Entry Conditionsa
C–H donor (equiv)
Catalystb or co-reactant
(equiv)
Productc (isolated yield)
8
CO2Me
QuinfiBH3 (0.2)
Bu3PfiBH3 (0.2)
(87%)
(83%)
>
>
>
>
1
A
MeO2C CO2Me (3.0)
MeO2C
CO2Me
>
>
>
>
>
>
>
>
CO2Et
EtO2C CO2Et
>
>
>
>
2
3
A
A
(3.0)
Bu3PfiBH3 (0.2)
(65%)
EtO2C CO2Et
>
>
>
R1
EWG C H
R2
>
>
>
>
>
<
CO2Et
(3.0)
EtO2C CO2Et
EtO2C
CO Et
2
EtO2C CO2Et
QuinfiBH3 (0.2)
Bu3PfiBH3 (0.2)
(85%)
(64%)
>
>
(Scheme 2)
>
>
>
>
>
QuinfiBH3 (0.2)
Bu3PfiBH3 (0.2)
>
>
>
4
5
B
B
CH3CO2Et (4.0)
(55%)
(14%)d
EtO2C
CO2Et
>
>
>
>
>
>
>
>
>
>
CO2Me
>
>
(3.0) QuinfiBH3 (0.2)
CO2Me
(60%)
MeO2C
>
>
:
MeO2C
MeO2C
CO2Me
O
O
8
>
NMe
(57%)
6
7
C
C
(ca. 14)
(1.5)
(1.5)
NMe
>
>
>
>
>
O
O
O
>
>
>
>
>
O
>
>
>
>
>
MeO2C
NMe
(ca. 11)
(54%)
NMe
>
>
>
>
>
O
O
>
>
>
R1
EDG C H
R2
>
>
O
>
>
<
MeO2C
NMe
O
O
1-Ad
>
>
>
>
(Scheme 3)
8
D
(5.0)
(1.5)
79:21 (50%)e
NMe
>
>
O
>
>
>
>
>
O
MeO2C
2-Ad
NMe
O
>
>
>
>
>
>
>
>
>
>
>
O
O
>
>
>
>
>
>
:
O
MeO2C
O
f
NMe
9
C
(ca. 14)
(1.5)
(42%)
NMe
O
O
a Generally 1 mmol of acetal 4; the total reaction time was usually 2.5 h. A ¼ benzene solvent, bath temp 70 ꢁC; TBHN (0.05 equiv) added initially
and again after 20 and 40 min; B ¼ refluxing benzene solvent, TBHN initiator (0.20 equiv) in benzene added by syringe pump during 100 min;
C ¼ acetal 4 dissolved in CH donor, bath temp 90 ꢁC, separate solutions of TBHN (0.20 equiv) and of co-reactant in benzene added by syringe
pump during 1 h; D ¼ acetal 4 and CH donor dissolved in benzene, otherwise as C.
b Quin ¼ Quinuclidine.
c The initially-formed silyl ester was converted to the alkyl ester shown.13
d Estimated by 1H NMR spectroscopy.
e Combined yield.
f The low yield may reflect the relatively low boiling point (69 ꢁC) of diisopropyl ether.
afforded a 79:21 mixture of two regioisomeric products
containing either a 1-adamantyl group or a 2-adamantyl
group, respectively, implying that the allyloxyl radical 9
has a greater tendency to abstract hydrogen from a
tertiary C–H group than does the tert-butoxyl radical.18
This difference in regioselectivity presumably reflects the
greater steric demands of 9 compared with those of ButOÅ.
Methodology for C–H functionalisation that is based
on heavy-metal catalysed reactions can suffer from

4408
Y. Cai et al. / Tetrahedron Letters 45 (2004) 4405–4409
O
O
OTBS
EWG
R1
R1
C
O
8
9
R2
EDG
H
C
R2
EDG
OH
10
R1
R2
OTBS
EWG
R1
R2
EDG
R1
O
R2
EDG
EDG
O
EWG
4
15
EWG
14
Scheme 3.
environmental and/or economic drawbacks.19 The pres-
ent preliminary work suggests that suitable precursors of
oxiranylcarbinyl radicals, that rapidly evolve into highly
reactive allyloxyl radicals, offer promise as reagents for
metal-free functionalisation at a range of types of C–H
group, including those in simple hydrocarbons.
8. Fischer, H.; Radom, L. Angew. Chem., Int. Ed. 2001, 40,
1340–1371.
9. (a) Roberts, B. P.; Steel, A. J. J. Chem. Soc., Perkin Trans.
2 1994, 2155–2162; (b) Roberts, B. P. J. Chem. Soc. Perkin
Trans. 2 1996, 2719–2725.
10. Doris, E.; Dechoux, L.; Mioskowski, C. J. Am. Chem. Soc.
1995, 117, 12700–12704.
11. (a) Kiefer, H.; Traylor, T. G. Tetrahedron Lett. 1966, 7,
6163–6168; (b) Mendenhall, G. D. Tetrahedron Lett. 1983,
24, 451–452.
References and notes
12. (a) Dang, H.-S.; Roberts, B. P. J. Chem. Soc., Perkin
Trans. 1 1993, 891–898; (b) VanPaasschen, J. M.; Gean-
angel, R. A. J. Inorg. Nucl. Chem. 1976, 38, 2321–2322,
This amine–borane complex is relatively stable towards
thermal dissociation.
1. (a) Motherwell, W. B.; Crich, D. Free Radical Chain
Reactions in Organic Synthesis; Academic: London, 1992,
€
pp 166–175; (b) Gansauer, A.; Pierobon, M. Rearrange-
ments of Cyclopropanes and Epoxides. In Radicals in
Organic Synthesis; Renaud, P., Sibi, M. P., Eds.; Wiley-
VCH: Weinheim, 2001; Vol. 2, pp 207–220, Chapter 3.3.
2. (a) Krosley, K. W.; Gleicher, G. J. J. Phys. Org. Chem.
1993, 6, 228–232; (b) Krishnamurthy, V.; Rawal, V. H. J.
Org. Chem. 1997, 62, 1572–1573.
3. Cai, Y.; Roberts, B. P. Tetrahedron Lett. 2004, 45, 1485–
1488.
4. (a) Cai, Y.; Roberts, B. P. Tetrahedron Lett. 2003, 44,
4645–4648; (b) Cai, Y.; Roberts, B. P.; Tocher, D. A.;
Barnett, S. A. Org. Biomol. Chem., in press.
13. Sakaitani, M.; Ohfune, Y. J. Org. Chem. 1990, 55, 870–
876.
14. The B–H bond in H3PfiBH3 is about 40 kJ molꢀ1 weaker
than that in H3NfiBH3; see Baban, J. A.; Roberts, B. P. J.
Chem. Soc., Perkin Trans. 2 1987, 497–505. Our unpub-
lished ab initio calculations at the G3 level15 indicate that
the values of DH298 for the B–H bonds in Me3PfiBH3 and
Me3NfiBH3 are 387.9 and 427.7 kJ molꢀ1, respectively,
compared with the corresponding calculated values for the
C–H bonds in Me4C and C2H6 of 429.3 and 423.2 kJ
molꢀ1, respectively; the experimental value for ethane is
423.0 1.6 kJ molꢀ1. For an a-C–H bond in ethyl acetate
5. Roberts, B. P. Chem. Soc. Rev. 1999, 28, 25–35.
6. Each of the compounds 4, 6 and 7 contains ca. 10% of the
isomer in which Me-1 and Me-3 are trans. The trans isomer
of the ketene acetal was consumed more rapidly than the
cis compound 4 during C–H functionalisation reactions;
the same product results from either isomer. Data for 4
containing 10% trans isomer (66% yield from 5); purified by
Kugelrohr distillation at 0.05 mmHg (oven temp. 125 ꢁC)
from a trace amount of 4,40-methylenebis(2,6-di-tert-bu-
tylphenol) as a radical scavenger. Found: C, 65.9; H, 10.3.
C18H34O3Si requires C, 66.2; H, 10.5%. NMR (500 MHz
the value of DH298 is ca. 402kJ mol ꢀ1 16
.
15. Curtiss, L. A.; Raghavachari, K.; Redfern, P. C.; Rasso-
lov, V.; Pople, J. A. J. Chem. Phys. 1998, 109, 7764–7776.
16. Bordwell, F. G.; Zhang, S.; Zhang, X.-M.; Liu, W.-Z. J.
Am. Chem. Soc. 1995, 117, 7092–7096.
17. Representative procedure: A solution of the ketene acetal 4
(327 mg, 1.0 mmol) in cyclohexane (1.5 mL) was stirred and
heated under reflux (bath temp. 90 ꢁC) while NMM
(167 mg, 1.5 mmol) in benzene (2.0 mL) and TBHN
(35 mg, 0.20 mmol) in benzene (0.5 mL) were added sepa-
rately by syringe pump during 1 h. Care was taken to keep
the TBHN solution at room temperature prior to its
addition, by passing the PTFE transfer tube down the
centre of the water-cooled condenser and ensuring that the
solution dripped directly into the hot reaction mixture.
After the additions, the solution was heated for a further
1.5 h before the solvent was removed by evaporation. The
residue was dissolved in THF (1.0 mL), cooled in an ice
bath and iodomethane (250 lL, 4.0 mmol) was added,
followed by TBAF in THF (1.0 M, 1.2mL). The solution
was stirred at 0 ꢁC for 2h before saturated aqueous NH 4Cl
solution (10 mL) was added. The mixture was extracted
1
for H, 125.7 MHz for 13C; CDCl3 solvent, J in Hz): dH
0.20 (6H, s, SiMe2), 0.88 (3H, s, Me-5), 0.94 (9H, s, But),
1.01 (3H, s, Me-5), 1.20 (1H, d, J 14.5, ring-CH2), 1.34 (3H,
s, Me-3), 1.44 (3H, s, Me-1), 1.55 (1H, dd, J 14.8 and 1.6,
ring-CH2), 1.62(1H, d, J 14.8, ring-CH2), 1.80 (1H, dd, J
14.5 and 1.6, ring-CH2), 3.03 (1H, s, H-2), 3.43 (1H, d, J
1.7, @CHAHB), 3.45 (1H, d, J 1.7, @CHAHB); dC–4.8, 18.0,
24.0, 24.4, 25.7, 29.3, 29.4, 31.3, 42.5, 42.8, 59.9, 63.8, 72.0,
78.4, 157.7. Identifying peaks for the trans isomer of 4: dH
t
0.92(9H, s, Bu ), 1.32(3H, s, Me-3), 1.39 (3H, s, Me-1),
2.98 (1H, s, H-2), 3.48 (1H, d, J 1.5, @CHAHB), 3.57 (1H,
d, J 1.5, @CHAHB).
7. Wasson, R. L.; House, H. O. Org. Synth. 1957, 37, 58–59,
This compound is also available from Aldrich.

Y. Cai et al. / Tetrahedron Letters 45 (2004) 4405–4409
4409
with CH2Cl2 (3 · 10 mL), the combined extracts were dried
(MgSO4) and the solvent was removed under reduced
pressure. The residue was purified by flash chromatogra-
phy on silica gel, using light petroleum (bp 40–60 ꢁC)-
diethyl ether–CH2Cl2 (16:4:5) as eluent, to afford trans-
1-methyl-3-cyclohexyl-4-(methoxycarbonylmethyl)pyrroli-
dine-2,5-dione3 13 (153 mg, 57%). Mp 76–78 ꢁC (from
hexane–CH2Cl2); dH 1.00–1.35 (5H, complex, ring-CH2),
1.48 (1H, m, ring-CH2), 1.63–1.80 (4H, complex, ring-
CH2), 1.94 (1H, m, H-30), 2.53 (1 H, apparent t, J 4.6, H-3),
2.66 (1 H, dd, J 17.0 and 4.6, MeO2CCHAHB), 2.82 (1H,
apparent q, J 5.1, H-4), 2.93 (1H, dd, J 17.0 and 5.6,
MeO2CCHAHB), 2.99 (3H, s, NMe), 3.67 (3H, s, OMe); dC
24.8, 25.8(9), 25.9(3), 26.2, 28.4, 30.1, 34.6, 38.6(6), 38.7(0),
50.9, 52.0, 171.2, 178.2, 178.5. Found: C, 62.8; H, 8.0; N,
5.2. C14H21NO4 requires C, 62.9; H, 7.9; N, 5.2%.
18. Minisci, F.; Fontana, F.; Zhao, L.; Banfi, S.; Quici, S.
Tetrahedron Lett. 1994, 35, 8033–8036, The comparison
between the selectivities of 9 and ButOÅ assumes that a
competing chlorine-atom chain mechanism can be
neglected in the reaction of tert-butyl hypochlorite with
adamantane reported by Minisci et al.
19. Fokin, A. A.; Schreiner, P. R. Adv. Synth. Catal. 2003,
345, 1035–1052.