Does the DABCO-catalysed reaction of 2-hydroxybenzaldehydes with methyl acrylate follow a Baylis-Hillman pathway?

Article abstract of DOI:10.1039/b300360d

Evidence is presented which supports the intermediacy of dipolar Baylis-Hillman-type adducts in the synthesis of coumarin and chromene derivatives from the reaction of 2-hydroxybenzaldehydes with methyl acrylate in the presence of 1,4-diazabicyclo[2.2.2]octane (DABCO).

Full text of DOI:10.1039/b300360d

View Article Online / Journal Homepage / Table of Contents for this issue
Does the DABCO-catalysed reaction of 2-hydroxybenzaldehydes
with methyl acrylate follow a Baylis–Hillman pathway?
Perry T. Kaye,* Musiliyu A. Musa, Xolani W. Nocanda and Ross S. Robinson
Department of Chemistry, Rhodes University, Grahamstown, 6140, South Africa.
E-mail: P.Kaye@ru.ac.za
Received 15th January 2003, Accepted 13th February 2003
First published as an Advance Article on the web 10th March 2003
Evidence is presented which supports the intermediacy of dipolar Baylis–Hillman-type adducts in the synthesis
of coumarin and chromene derivatives from the reaction of 2-hydroxybenzaldehydes with methyl acrylate in the
presence of 1,4-diazabicyclo[2.2.2]octane (DABCO).
The Baylis–Hillman reaction1,2 continues to receive attention,3
and has recently enjoyed the distinction of becoming a “text-
book” reaction!4 Our research on this versatile transformation
has been focused, largely, on its use in the construction of
benzannulated heterocycles – an interest initially sparked by
the isolation of a 2-substituted indolizine during distillation
of a pyridine-2-carbaldehyde-derived Baylis–Hillman product.5
Early attempts to extend the methodology to the synthesis of
oxygenated analogues from 2-hydroxybenzaldehydes 1, how-
ever, afforded complex mixtures of various chromene and
coumarin derivatives (Scheme 1).6–8 These observations
prompted several questions. Could the regioselectivity of cyclis-
ation be controlled to afford chromene or coumarin derivatives
chemoselectively? Could the methodology be extended to the
preparation of nitrogen- and sulfur-containing analogues? Are
the Baylis–Hillman adducts 3 common intermediates (as
suggested in Scheme 1) in the formation of the chromene and
coumarin derivatives?
We have subsequently demonstrated chemoselective syntheses
of chromene9,10 and coumarin derivatives11–13 under Baylis–
Hillman conditions, and successfully extended the method-
ology to the preparation of quinoline14 and thiochromene
(benzothiopyran) derivatives.15 In this communication, we
address the remaining issue, viz., the possible implication of
Baylis–Hillman adducts 3 as common, pivotal intermediates in
the formation of the chromene and coumarin derivatives.
substitution and Baylis–Hillman reactions (Path C) would lead
to the coumarin derivative 5.17 In our attempts to elucidate
the mechanistic sequence in these transformations, several
strategies have been explored, viz.,
i, temporary protection of the phenolic hydroxy group of the
2-hydroxybenzaldehyde substrate 1 to permit isolation of a pro-
tected Baylis–Hillman adduct, subsequent deprotection and
cyclisation being expected to afford chromene and coumarin
derivatives;
ii, use of 4-hydroxybenzaldehyde as a model compound (for
which intramolecular cyclisation is not possible) to explore the
preferred initial reaction pathway (A, B or C); and
iii, the preparation and subsequent cyclisation of unprotected
Baylis–Hillman products.
The reaction of O-acetylated 2-hydroxybenzaldehydes with
methyl acrylate 2 in the presence of DABCO was shown18 to
afford chromene and coumarin derivatives of the type
produced by unprotected analogues – an observation which
indicated in situ de-acetylation but provided no information
on the sequence of events. Attention was consequently turned
to the use of O-tert-butyldimethylsilyl(TBDMS)-protected
2-hydroxybenzaldehydes, choice of the protecting group having
been influenced by the expected stability of the resulting silyl
ethers and their potential for selective removal using fluoride
ion.19,20 Treatment of salicylaldehyde 1 (R1 = H) with tert-
butyldimethylsilyl chloride in the presence of imidazole20
failed to afford the required silyl ether 8 (Scheme 3); silylation
was, however, effected in good yield by reacting the salicyl-
aldehyde phenoxide ion with tert-butyldimethylsilyl chloride.
The O-TBDMS-protected salicylaldehyde 8 was then treated
with methyl acrylate 2 in the presence of DABCO in antici-
pation of obtaining the O-silylated Baylis–Hillman product
9. In the event, starting material 8 was isolated together with
salicylaldehyde 1 (R1 = H), the chromene-3-carboxylic ester 4,
the coumarin derivatives 10 and 11,6–8 and the bis-silylated
Baylis–Hillman product 12!
Results and discussion
The reaction of 2-hydroxybenzaldehydes 1 with methyl acrylate
2 in the presence of 1,4-diazabicyclo[2.2.2]octane (DABCO)
could, in principle, proceed via a number of possible pathways,
as illustrated for salicylaldehyde 1 (R1 = H) in Scheme 2.
An initial Baylis–Hillman reaction (Path A) would afford
the adduct 3, which could cyclise via conjugate addition–
elimination or acyl substitution to give the chromene 4 or cou-
marin derivative 5, respectively. Alternatively, initial conjugate
addition (Path B) and subsequent aldol cyclisation of the result-
ing enolate 6 would give the chromene 4,16 while tandem acyl
A consolidated mechanistic sequence, which could account
for the formation of all five of these compounds, is outlined
in Scheme 4. Thus, the Baylis–Hillman adduct 14, once formed
Scheme 1
T h i s j o u r n a l i s © T h e R o y a l S o c i e t y o f C h e m i s t r y 2 0 0 3
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 1 1 3 3 – 1 1 3 8
1133

View Article Online
Scheme 2
Scheme 3
by addition of the “Baylis–Hillman zwitterion” 13 to the O-
In order to demonstrate the ability of unprotected Baylis–
silylated salicylaldehyde 8, may attack another substrate mole-
cule 8. Nucleophilic displacement of salicylaldehyde 1 (R1 = H)
and elimination of DABCO would then afford the bis-silylated
derivative 12 (Path I; Scheme 4). Concomitant intramolecular
attack (Path II) would lead, following loss of DABCO, to the
phenoxide ion 15; subsequent cyclisation, via conjugate addi-
tion (Path III) or acyl substitution (Path IV), would then afford
the chromene 4 or the intermediate coumarin derivative 16,
respectively. Allylic displacement of TBDMS-OH from the
intermediate 16 (via a 6-centred transition-state complex)
would then explain the formation of the coumarin derivative
11. Subsequent addition of the Baylis–Hillman zwitterion 13a,
elimination of salicylaldehyde and DABCO from the inter-
mediate adduct 17 and, finally, tautomerism would then
account for the formation of the 4-substituted chromene 10.
Hillman products to undergo intramolecular cyclisation, the
disilylated derivative 12 was treated with tetrabutylammonium
fluoride in THF at 0 ЊC (Scheme 5).20 Preparative layer chrom-
atography of the resulting mixture afforded the chromene-3-
carboxylic ester 4 and, as the major product, the hitherto
somewhat elusive Baylis–Hillman product 3.21,22 The latter
compound, which was fully characterised, was then dissolved in
CDCl3 and the resulting solution subjected to periodic 1H
NMR analysis. After one week, there was no evidence of cyclis-
ation, and DABCO was added to the solution on the assump-
1
tion that the cyclisation might well be DABCO-catalysed. H
NMR analysis of the resulting mixture, however, indicated sub-
stantial formation of salicylaldehyde 1 (R1 = H) and methyl
acrylate 2. This result supports the proposed23 reversibility of
the Baylis–Hillman reaction, but clearly fails to demonstrate
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 1 1 3 3 – 1 1 3 8
1134

View Article Online
Scheme 4
Scheme 5
implication of a Baylis–Hillman pathway in the formation of
MVK dimer.10 When methyl acrylate 2, which exhibits little
chromene and coumarin derivatives!
tendency to dimerise under Baylis–Hillman conditions, was
used as the activated alkene, the only product to be isolated
[together with starting material (60%)] was, in fact, the Baylis–
Hillman adduct 23 (10%; Scheme 6). None of the conjugate-
addition product 24 could be detected. Further support for an
initial Baylis–Hillman step in the cyclisation reactions under
The competition between the Baylis–Hillman and conjugate-
addition pathways (Scheme 2; Paths A and B, respectively) was
investigated by treating 4-hydroxybenzaldehyde 22 with methyl
vinyl ketone (MVK) in the presence of DABCO. Under these
conditions, however, the only product to be isolated was the
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 1 1 3 3 – 1 1 3 8
1135

View Article Online
Scheme 6
Scheme 7
investigation is provided by the isolation of Baylis–Hillman
would initiate the observed retro-Baylis–Hillman process.
adducts as intermediates in our reported syntheses of quin-
Addition of HCl (Nu: = ClϪ) in strongly acidic medium, how-
ever, would afford the protonated species 35, which could
undergo rapid and irreversible cyclisation to the corresponding
3-(chloromethyl)coumarin 29a (Scheme 7) via acid-catalysed
acyl substitution (path II).
oline14 and thiochromene15 systems.
We have subsequently found that reaction of 2-hydroxy-
benzaldehydes 1a–e (Scheme 7) with tert-butyl acrylate 25 in the
presence of DABCO affords the isolable Baylis–Hillman
products 26a–e.13 On treatment with DABCO in CDCl3, the
salicylaldehyde-derived adduct 26a was shown to undergo a
retro-Baylis–Hillman reaction. In refluxing acetic acid, how-
ever, the adduct 26a was converted to the chromene and cou-
marin derivatives 27 and 28,13 respectively, demonstrating the
capacity of Baylis–Hillman adducts to undergo intramolecular
cyclisation, albeit under acidic conditions! In fact, treatment
of the tert-butyl ester adducts 26a–e with hydrochloric acid
has been shown to provide convenient, efficient and chemo-
selective access to the corresponding 3-(chloromethyl)-
coumarins 29a–e.13 The isolability of the adducts 26a–e may be
attributed to the steric and electronic effects of the tert-butyl
group, which appear to inhibit spontaneous, intramolecular
nucleophilic attack at both the vinylic and carbonyl centres.
While it is apparent that the 2-hydroxybenzaldehyde-derived
Baylis–Hillman adducts (3 and 26a–e) cyclise in acidic media,
their failure to undergo such cyclisation in the presence of
DABCO was, initially, somewhat puzzling. A possible explan-
ation lies in the interplay between the reversible and non-revers-
ible transformations outlined in Scheme 8. Thus, reversible
addition of the Baylis–Hillman zwitterions (13 or 30) to the
2-hydroxybenzaldehyde 1 (R1 = H) produces the corresponding
dipolar adducts 31 and 32. In the case of the methyl ester 31,
proton transfer affords the more stable isomeric species 33,
which contains a nucleophilic phenoxide moiety and two highly
electrophilic centres, C-1 (activated by intramolecular hydro-
gen-bonding) and C-3Ј (activated by the adjacent quaternary
nitrogen). Dipolar adducts of type 33 are, we suggest, the
pivotal intermediates, which undergo slow but irreversible
intramolecular (SN) cyclisation and dehydration (via path I) or
acyl substitution (via path II) to afford chromene or coumarin
products, respectively. In the case of the tert-butyl ester 32,
cyclisation is inhibited and elimination of DABCO affords the
relatively stable Baylis–Hillman product 26a. Formation of the
adduct 26a is reversible, and conjugate addition by DABCO
In the light of the cumulative evidence, we conclude that in
the formation of chromene and coumarin derivatives from
DABCO-catalysed reactions of 2-hydroxybenzaldehydes 1 with
methyl acrylate 2: i) the Baylis–Hillman reaction precedes
conjugate addition or acyl substitution; and ii) the pivotal
intermediates are, in fact, highly activated dipolar adducts of
type 33, rather than the Baylis–Hillman products per se.
Experimental
NMR spectra were recorded on Bruker AMX400 or AVANCE
400 MHz spectrometers at 303 K in CDCl3 and calibrated using
solvent signals. Infrared spectra were recorded on a Perkin
Elmer FT-IR Spectrum 2000 spectrometer. Low-resolution (EI)
mass spectra were obtained on a Finnigan-Mat GCQ mass
spectrometer and high-resolution (EI) mass spectra on a VG70-
SEQ Micromass double-focusing magnetic sector spectrometer
(Cape Technikon Mass Spectrometry unit). The preparation
and characterisation of compounds 4,10,11,26a–e,27,28 and
29a–e have been reported previously.6–8,13,18,21
2-(tert-Butyldimethylsilyloxy)benzaldehyde 8
Salicylaldehyde (1.9 ml, 18 mmol) was added dropwise to a
stirred suspension of washed NaH (50% dispersion in oil; 0.94
g, 20 mmol) in dry THF (50 ml) under N2. The mixture was
stirred at room temperature for 30 min, tert-butyldimethylsilyl
chloride (3.0 g, 20 mmol) in dry THF was then added and the
resulting mixture was stirred at room temperature for 12 h. The
reaction was quenched by the addition of aq. NaHCO3 (20 ml)
and the resulting mixture extracted into diethyl ether (3 ×
20 ml). The ethereal extracts were washed with saturated brine
(50 ml) and dried over anhydrous MgSO4. The solvent was
removed in vacuo to give the crude product (6.4 g), which was
purified by flash chromatography [elution with hexane–EtOAc
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 1 1 3 3 – 1 1 3 8
1136

View Article Online
Scheme 8
(9 : 1)] to afford 2-(tert-butyldimethylsilyloxy)benzaldehyde 8
25.8 [2 × C(CH3)3], 51.5 (OCH3), 66.7 (OCH), 117.8, 120.7,
(2.7 g, 65%) (Found: MH؉, 237.1310. C13H20SiO2 requires
Mϩ1, 237.1311); νmax (hexachlorobutadiene mull/cmϪ1) 1742
(CO); δH (400 MHz; CDCl3) 0.26 [6H, s, Si(CH3)2], 1.02 [9H, s,
C(CH3)3], 6.86 (1H, d, ArH), 7.01 (1H, t, ArH), 7.43 (1H, t,
ArH), 7.80 (1H, d, ArH) and 10.46 (1H, s, CHO); δC (100 MHz;
CDCl3) Ϫ4.4 [Si(CH3)2], 18.3 [SiC(CH3)3], 25.6 [SiC(CH3)3],
120.2, 121.4, 127.3, 128.3, 135.6 and 158.8 (ArC) and 189.9
(CHO); m/z 236 (Mϩ, 0.9) and 179 (100%).
124.8, 124.9, 128.1, 132.4, 143.9 and 152.3 (ArC and C᎐CH )
᎐
2
and 166.9 (CO); m/z 436 (Mϩ, 0.05) and 179 (100%).
Deprotection of methyl 3-(tert-butyldimethylsilyloxy)-3-[2-(tert-
butyldimethylsilyloxy)phenyl]-2-methylenepropanoate 12
Tetrabutylammonium fluoride (1.0 g, 3.2 mmol) was added to
a stirred solution of methyl 3-(tert-butyldimethylsilyloxy)-3-[2-
(tert-butyldimethylsilyloxy)phenyl]-2-methylenepropanoate 12
(0.12 g, 0.27 mmol) in dry THF (1.0 ml) at 0 ЊC. After
5 minutes, water was added (0.2 ml), followed by dilute HCl
until the solution was just acidic, and the mixture was then
extracted with diethyl ether. The ethereal extracts were
combined and dried (anhyd. MgSO4), the solvent was removed
in vacuo and the residue purified by preparative layer
chromatography [elution with hexane–EtOAc (3 : 2)] to afford
the following two products. i) Methyl 3-hydroxy-3-(2-hydroxy-
phenyl)-2-methylenepropanoate 3, as a colourless oil (0.04 g,
71%), (Found: M؉, 208.0733. C11H12O4 requires: M, 208.0736);
δH (400 MHz; CDCl3) 3.82 (3H, s, OCH3), 4.12 (1H, br s,
Attempted preparation of methyl 3-hydroxy-3-[2-(tert-butyl-
dimethylsilyloxy)phenyl]-2-methylenepropanoate 9
A solution of 2-(tert-butyldimethylsilyloxy)benzaldehyde 8
(1.5 g, 6.4 mmol), methyl acrylate 2 (0.72 ml, 7.9 mmol) and
DABCO (0.07 g, 0.6 mmol) in CHCl3 (1.0 ml) was stirred in a
stoppered flask under N2 for 3 d. The crude mixture was puri-
fied by flash chromatography followed by preparative layer
chromatography (elution with chloroform) to afford the follow-
ing six compounds. i) Salicylaldehyde 1 (R1 = H) (0.30 g, 39%).
ii) Methyl 2H-1-chromene-3-carboxylate 4 (0.08 g, 6.6%).
iii) 2-(tert-Butyldimethylsilyloxy)benzaldehyde 8 (0.22 g, 15%).
iv) Methyl 2-(3-methyl-1-benzopyran-2-on-4-yl)propenoate 10,
as white crystals (0.03 g, 2%), mp 104–106 ЊC (Found: M؉,
244.0745. C14H12O4 requires: M, 244.0736); νmax (Nujol mull /
cmϪ1) 1707 and 1667 (2 × CO); δH (400 MHz; CDCl3) 2.09 (3H,
CHOH ), 5.59 and 6.35 (2H, 2 × s, C᎐CH ), 5.74 [1H, s,
᎐
2
CH(OH)], 6.85 (1H, t, ArH), 6.93 (1H, d, ArH), 6.96 (1H, d,
ArH), 7.21 (1H, t, ArH), 7.97 (1H, s, ArOH); δC (100 MHz;
CDCl3) 52.4 (OCH3), 73.5 [CH(OH)], 117.6, 120.0, 123.8,
127.8, 128.0, 129.7, 139.4 and 155.9 (ArC and C᎐CH ) and
᎐
2
s, CH ), 3.76 (3H, s, OCH ), 5.82 and 6.87 (2H, 2 × s, C᎐CH ),
᎐
3
3
2
167.7 (CO). ii) Methyl 2H-1-chromene-3-carboxylate 4 (0.01 g,
19%).
7.21–7.46 (4H, m, ArH); δC (100 MHz; CDCl3) 14.6 (CH3), 52.7
(OCH3), 116.8, 119.7, 123.9, 124.2, 125.6, 130.7, 131.4, 135.3,
145.8 and 152.4 (ArC, C᎐C and C᎐CH ), 161.8 (CO CH ) and
᎐
᎐
2
2
3
Methyl 3-hydroxy-3-(4-hydroxyphenyl)-2-methylenepropanoate
23
165.0 (CO). v) 3-[(2-Formylphenoxy)methyl]coumarin 11, as
white crystals, (0.05 g, 3%), mp 178–180 ЊC (lit.6180–182 ЊC). vi)
Methyl 3-(tert-butyldimethylsilyloxy)-3-[2-(tert-butyldimethyl-
silyloxy)phenyl]-2-methylenepropanoate 12 as a light yellow oil
(0.33 g, 12%), (Found: MH؉, 437.2544. C23H40Si2O4 requires
Mϩ1, 437.2543); νmax (hexachlorobutadiene mull /cmϪ1) 1727
(CO); δH (400 MHz; CDCl3) Ϫ0.020, 0.067, 0.23 and 0.26 [12H,
4 × s, 2 × Si(CH3)2], 0.86 and 0.99 [18H, 2 × s, 2 × C(CH3)3],
A mixture of 4-hydroxybenzaldehyde 22 (0.20 g, 1.6 mmol),
methyl acrylate 2 (0.22 ml, 2.5 mmol) and DABCO (0.07 g,
0.7 mmol) in CHCl3 (0.5 ml) was stirred in a stoppered flask
under N2 at room temperature for 7 d. The resulting mixture
was purified by flash chromatography [elution with hexane–
EtOAc (3 : 2)], to afford starting material (0.12 g, 60%) and, as
colourless oil, methyl 3-hydroxy-3-(4-hydroxyphenyl)-2-methyl-
enepropanoate 23 (0.03 g, 10%) (Found: M؉, 208.0743.
C11H12O4 requires: M, 208.0736; νmax (hexachlorobutadiene
mull /cmϪ1) 3407 (br, OH) and 1710 (CO); δH (400 MHz;
3.71 (3H, s, OCH ), 5.44 and 6.14 (2H, 2 × s, C᎐CH ), 6.05 (1H,
᎐
3
2
s, OCH), 6.75 (1H, d, ArH), 6.93 (1H, t, ArH), 7.12 (1H, t,
ArH) and 7.42 (1H, d, ArH); δC (100 MHz; CDCl3) Ϫ4.76,
Ϫ4.17, Ϫ4.11 and Ϫ4.12 [2 × Si(CH3)2], 18.2 [2 × C(CH3)3],
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 1 1 3 3 – 1 1 3 8
1137

View Article Online
8 R. S. Robinson and P. T. Kaye, Synth. Commun., 1996, 26,
CDCl3) 3.13 [1H, br s, CH(OH )], 3.70 (3H, s, OCH3), 5.49 [1H,
2085.
s, CH(OH)], 5.85 and 6.31 (2H, 2 × s, C᎐CH ), 6.10 (1H, br s,
᎐
2
9 P. T. Kaye and X. W. Nocanda, J. Chem. Soc., Perkin Trans. 1, 2000,
1331.
10 P. T. Kaye and X. N. Nocanda, J. Chem. Soc., Perkin Trans. 1, 2002,
1318–1323.
ArOH), 6.72 (2H, d, ArH) and 7.17 (2H, d, ArH); δC (100 MHz;
CDCl3) 52.0 (OCH3), 72.8 [CH(OH)], 115.8, 125.8, 128.1,
133.0, 142.0 and 155.6 (ArC and C᎐CH ) and 167.0 (CO).
᎐
2
11 P. T. Kaye and M. A. Musa, Synth. Commun., in press.
12 P. T. Kaye and M. A. Musa, Synthesis, 2002(18), 2701–2706.
13 (a) M. A. Musa, PhD thesis, Rhodes University, 2002; (b) P. T. Kaye
and M. A. Musa, Synthesis, in press.
14 O. B. Familoni, P. T. Kaye and P. J. Klaas, Chem. Commun., 1998,
2563.
15 P. T. Kaye and X. N. Nocanda, Synthesis, 2001(16), 2389–
2392.
16 For a preparation of 2H-chromenes by reacting salicylaldehyde (as
the phenoxide ion) with activated alkenes in an approach which
presumably involves a conjugate addition–aldol cyclisation sequence
see C. D. DeBoer, J. Org. Chem., 1974, 39, 2426.
Acknowledgements
The authors thank the National Research Foundation (NRF)/
Foundation for Research Development (FRD) for bursaries (to
X.W.N. and R.S.R), and Rhodes University and the NRF/FRD
for generous financial support.
References
1 A. B. Baylis and D. E. M. Hillman, Ger. Pat., 1972, 2.155.113 (Cl.
C07c) (Chem. Abstr., 1972, 77, 34174q).
17 The capacity of pre-formed salicylaldehyde esters of type 7 to
undergo DABCO-catalysed cyclisation to afford quaternary
ammonium coumarin derivatives has been reported elsewhwere. See
ref. 13a and S. E. Drewes, O. L. Njamela, N. D. Emslie, N. Ramesar
and J. S. Field, Synth. Commun., 1993, 23(20), 2807–2815.
18 R. S. Robinson, PhD thesis, Rhodes University, 1997.
19 E. J. Corey and A. Venkateswarlu, J. Am. Chem. Soc., 1972, 94,
6190.
2 E. Ciganeck, Org. React., 1977, 101, 201.
3 See, for example: D. Basavaiah, D. S. Sharada, N. Kumaraguru-
baran and R. M. Reddy, J. Org. Chem., 2002, 67, 7135;
D. Basavaiah, R. M. Reddy, N. Kumaragurubaran and D. S.
Sharada, Tetrahedron, 2002, 58, 3693; C. Yu, B. Liu and L. Hu,
J. Org. Chem., 2001, 66, 5413.
4 J. P. Clayden, N. Greeves, S. Warren and P. D. Wothers,
Organic Chemistry, Oxford University Press, Oxford, 2001, p. 1124.
5 M. L. Bode and P. T. Kaye, J. Chem. Soc., Perkin Trans. 1, 1990,
2612; M. L. Bode and P. T. Kaye, J. Chem. Soc., Perkin Trans. 1,
1993, 1809.
6 M. L. Bode, R. B. English and P. T. Kaye, S. Afr. J. Chem., 1992, 45,
25.
20 T. W. Green, Protective Groups in Organic Synthesis, Wiley, New
York, 1981, pp. 44–47.
21 X. W. Nocanda, PhD thesis, Rhodes University, 2001.
22 In a previous study (ref. 18) similar treatment of the bis-silylated
derivative 12 gave a mixture indicated by 1H NMR spectroscopy to
contain the chromene-3-carboxylic ester 4 (43%), its 4-hydroxy-
chroman precursor (19%), and a third component, tentatively
identified as 4-hydroxy-3-methylenecoumarin (38%).
23 S. E. Drewes and G. H. P. Roos, Tetrahedron, 1988, 44, 4653.
7 J. Bacsa, P. T. Kaye and R. S. Robinson, S. Afr. J. Chem., 1998, 51,
47.
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 1 1 3 3 – 1 1 3 8
1138