A new type of photodimerization reaction for coumarin derivatives

Article abstract of DOI:10.1016/S0040-4039(02)00969-3

Irradiation (λ >300 nm) of coumarin-3-carboxylic acid gives the decarboxylated 4,4′-dimer of chroman-2-one, the structure of which is quite different from that of [π2s+π2s] cycloaddition dimer as is observed in the past in the photoreaction of coumarin or some of its derivatives.

Full text of DOI:10.1016/S0040-4039(02)00969-3

TETRAHEDRON
LETTERS
Pergamon
Tetrahedron Letters 43 (2002) 5161–5163
A new type of photodimerization reaction for coumarin
derivatives
Hiroki Kawata,a,* Satoko Ichikawa,a Tsutomu Kumagaib and Shigeya Niizumaa
aFaculty of Humanities and Social Sciences, Iwate University, Ueda, Morioka 020-8550, Japan
bGraduate School of Science, Tohoku University, Sendai 980-0845, Japan
Received 3 April 2002; accepted 17 May 2002
Abstract—Irradiation (u >300 nm) of coumarin-3-carboxylic acid gives the decarboxylated 4,4%-dimer of chroman-2-one, the
structure of which is quite different from that of [p2s+p2s] cycloaddition dimer as is observed in the past in the photoreaction of
coumarin or some of its derivatives. © 2002 Elsevier Science Ltd. All rights reserved.
The photoreaction of coumarin is well-known to give
cis-head-to-head (I) or trans-head-to-head (II) dimers
via the [p2s+p2s] cycloaddition reaction.1–6 This reac-
tion depends on the multiplicity of the excited state of
coumarin, that is, I is formed from the excited singlet
state in ethanol solution and II from the excited triplet
state which is performed by a photosensitizer such as
benzophenone in benzene solution, respectively. In
1995, Shim7 and Chen8 reported the photocleavage of
the cycloaddition dimer of coumarin derivatives with
UV light. We have found a new type photodimerization
reaction for coumarin-3-carboxylic acid (1). Irradiation
of 1 in ethanol solution gives three kinds of products,
4,4%-dimer of chroman-2-one (2), 3-(1%-hydroxyethyl)-
coumarin (3) and coumarin itself (4). The main photo-
product is 2. Further, these are decarboxylated
compounds. In this paper, we report the photodimer-
ization and decarboxylation reaction of 1 by the detec-
tion of the transients using the flash apparatus.
as follows: after evaporation of the solvent of the
photolysis mixture under reduced pressure, the precipi-
tate appears. It is filtered and recrystallized from acetic
acid. With the TLC procedure (silica gel 60 with ben-
zene/methanol, 8:2) of the filtrate, 3 and 4 were
obtained, respectively. The quantum yield for the disap-
pearance of 1 (1.0×10−4 mol dm−3) is 0.05 with a 150 W
Xe lamp equipped with a 313 nm interference filter. The
light intensity of the lamp was measured with a chemi-
cal actinometer. The value of 0.05 seems to be very high
compared with that for the cycloaddition-type dimer
because a few days irradiation was needed to obtain the
cycloaddition-type dimer even for the diluted solu-
tion.6,7 In the aerated solution of 1, no changes in the
UV absorption spectrum or HPLC on the irradiation
are observed. Schenk3 and Hammond4 reported the
photoreaction of coumarin proceeded under very high
concentration, ꢀ0.5 mol dm−3, and the excited singlet
and the triplet state coumarin reacted with the ground
state coumarin to yield I and II, respectively. In con-
trast, 1 does not dissolve so much in ethanol, the
maximum concentration being ꢀ1.5×10−3 mol dm−3 at
room temperature. So, we had to experiment under
lower concentration of 1 than in the case of the litera-
ture. However, it should be noted that even the 1×10−2
mol dm−3 solution of coumarin4,5 or the 1×10−3 mol
dm−3 solution of coumarin derivative7 also gave the
cycloaddition-type dimer with the irradiation. The pho-
toreaction for the cycloaddition-type dimer formation
seems not to depend on the concentration of coumarin
or its derivatives. Further, irradiation of argon satu-
rated ethanol solution of 4 (1×10−3 mol dm−3) for 10 h
gives no products. This means that under our experi-
mental light intensities of the Xe lamp and the irradia-
Using a 500 W Xe lamp with a Pyrex filter, irradiation
of 1 (Aldrich Chem. Co., recrystallized twice from
ethanol) in argon saturated ethanol solution (40 mg in
200 cm3 cylindrical quartz cell, 1×10−3 mol dm−3) for 8
h afforded the three kinds of products. The above
procedure was repeated four times and finally, com-
pounds 29 (56 mg, yield 45.2%), 310 (8 mg, 5.0%) and
411 (12 mg, 9.8%) were obtained for each run. The
reaction was monitored by HPLC with acetonitrile/
water (6:4 in volume) solvent. Compound 2 is obtained
Keywords: coumarin-3-carboxylic acid; photodimerization; decar-
boxylation; hydrogen abstraction.
* Corresponding author.
0040-4039/02/$ - see front matter © 2002 Elsevier Science Ltd. All rights reserved.
PII: S0040-4039(02)00969-3

5162
H. Kawata et al. / Tetrahedron Letters 43 (2002) 5161–5163
O
O
OH
OH
tion time, we do not have to consider the possibility of
the photoisomerization of the primary product 4. Thus,
the contribution of I and II type photoproducts of 1 is
ruled out.
EtOH
.
31*
.
COOH
COOH
H
Dimeization
O
Enol - keto
tautomerization
With a conventional microsecond flash apparatus, the
photolysis of 1 in argon saturated ethanol solution
gives a relatively long-lived transient which decays in
the second order with the rate constant of 1.4×m×104
dm3 mol−1 s−1, where m is the molar extinction coeffi-
cient of the transient at 420 nm. This transient is
observed below the wavelength of 600 nm (umax 420 and
540 nm). No transient was observed in the aerated
ethanol or the argon saturated benzene solution of 1.
Nanosecond laser flash photolysis (266 nm light pulse
excitation of Nd3+–YAG laser with a full width at
half-maximum intensity of 5 ns) of 1 in argon saturated
ethanol solution also gives a transient which decays in
the first order with the rate constant of 2.4×105 s−1,
while no transient is observed in the aerated system.
Fig. 1 shows the time-dependent transient absorption
spectra at various delay times, 300 ns, 1.5 ms and 2.5 ms
after flashing, respectively. This should be a T–T
absorption spectrum of 1 because it is quenched by
ferrocene or oxygen molecule, a good triplet quencher.
The quenching rate constant by ferrocene is 3.4×109
dm3 mol−1 s−1. The amount of the long-lived transient
by a dose observed in a microsecond flash photolysis
decreases with increasing the concentration of fer-
rocene, though the additive does not affect its decay
rate. Thus the photoreactive state of 1 is inferred to be
the excited triplet state (31*) and the long-lived tran-
O
OH
O
.
COOH
COOH
H
H
H
H
COOH
OH
O
Dimeization
Enol - keto
tautomerization
..
O
OH
O
O
H
H
- CO2
H
O
H
C
H
COOH
OH
H
H
O
H
COOH
O
O
O
O
Enol - keto
tautomerization
3-Keto acid type compound
O
O
O
H
H
H
H
H
H
- CO2
H
H
H
H
H
O
COOH
Enol - keto
O
O
O
tautomerization
2
3
sient is a subsequent product after the decay of 1*. We
Scheme 1.
interpret this transient as a precursor of 4, a ketyl
radical formed by the hydrogen abstraction from the
solvent, ethanol. This is also detected in a benzene
solution containing N,N-dimethylaniline, a hydrogen
atom donor, but not detected in a benzene only solu-
tion as mentioned above. Unfortunately, we could not
obtain the CIDEP signals for the ketyl radical as were
observed in the photolysis of chromone-2-carboxylic
acid.12
We propose the Scheme 1 for the formation of 2. The
3
photoreactive state is 1* from the flash photolysis. The
fate of the ketyl radical is as follows. The equilibrium
between the 2-position and 4-position radical should
shift to the latter judging from the structure of the end
product. Two pathways for the formation of 3-keto
acid-type compounds can be depicted as in Scheme 1;
the dimerization of radicals followed by the enol-keto
tautomerization or quite the reverse. However, it is
uncertain which way is plausible to give the 3-keto
acid-type compound. 3-Keto acids, especially free car-
boxylic acids, are known to decompose very easily to
give carbon dioxide and ketone by a concerted reaction
involving a cyclic transition state.13 Similarly, 3-keto
acid-type dimer in Scheme 1 can be thought to undergo
the two-step reaction through the decarboxylation and
the enol-keto tautomerization in the dark, and finally to
yield 2. Further, photoillumination of 4-hydroxycou-
marin or 4-methyl-3-phenylcoumarin in argon satu-
rated ethanol solution gives no transients and no
photoproducts. The hydrogen abstraction and the fol-
lowing decarboxylation reaction in this paper seem to
be characteristic of the carboxylic group of coumarin.
Figure 1. T–T absorption spectra of 2×10−4 mol dm−3 ethanol
solution of 1: (a) 300 ns, (b) 1.5 ms and (c) 2.5 ms after 266 nm
laser flashing.

H. Kawata et al. / Tetrahedron Letters 43 (2002) 5161–5163
5163
.
6. Hoffman, R.; Wells, P.; Morrison, H. J. Org. Chem.
1971, 36, 102.
7. Shim, S. C.; Jeong, B. M.; Paik, Y. H. Bull. Korean
COOH or
EtOH
O
O
.
31*
+ CO2 or
.
4
CH3CHOH
O
O
Chem. Soc. 1992, 13, 684.
8. Chen, Y.; Chou, C. F. J. Polym. Sci., Part A: Polym.
Chem. 1995, 33, 2705.
.
4'
CH3CHOH
OH
CH3
9. Compound 2: Mp 278–280°C. Anal. found: C, 73.29; H,
4.44. Calcd for C18H14O4: C, 73.46; H, 4.79%. HRMS:
294.0892 (calcd for C18H14O4; 294.3014). IR (KBr): 1771,
3
Scheme 2.
1760 cm−1 1H NMR (CDCl3, 500 MHz): l 2.73 (1H,
.
ddd, J=2.0, 5.2, 16.0 Hz), 2.85 (1H, dd, J=2.0, 16.0 Hz),
2.91 (1H, dd, J=2.0, 2.0 Hz), 7.14 (dd, J=1.2, 8.0 Hz),
7.22 (ddd, J=1.2, 7.6, 7.6 Hz), 7.29 (dd, J=1.6, 7.6 Hz),
7.37 (ddd, J=1.6, 7.6, 8.0 Hz). 13C NMR (CDCl3, 150
MHz): l 32.981 (CH2), 38.681 (CH), 117.64 (C4),
124.844 (C6), 129.542 (C5), 129.709 (C7), 167.733 (C1).
10. Compound 3: Colorless oil. IR (KBr): 3424, 1718, 1699,
The reaction mechanism for the minor photoproducts,
3 and 4, is represented in Scheme 2. First, the direct
homolytic cleavage of the carboxyl group should occur
3
from 1* as is also observed in the photo-decarboxyla-
tion reaction of pyridine carboxylic acid.14 A hydrogen
atom in the 3-position of 4 most likely comes from the
carboxyl group, though the hydrogen abstraction from
ethanol cannot be ruled out. As a result, the 1-hydroxy-
ethyl radical, which is formed in Scheme 1 or in both
Schemes 1 and 2, attacks 4% to yield 3.
1
1610 cm−1. H NMR (CDCl3, 500 MHz): l 1.55 (3H, d,
J=6.5 Hz), 3.23 (OH, brs), 4.93 (1H, dq, J=0.8, 6.5 Hz),
7.29 (ddd, J=1.0, 7.6, 8.0 Hz), 7.32 (dd, 1.0, 8.7 Hz),
7.506 (dd, J=1.7, 7.6 Hz), 7.508 (ddd, 1.7, 8.0, 8.7 Hz),
7.78 (1H, s). 13C NMR (CDCl3, 150 MHz): l 21.860
(CH3), 65.919 (CH), 116.397 (C4), 124.539 (C6), 127.904
(C7), 131.249 (C5), 137.237 (C3).
References
11. Mp, mass and NMR spectra of 4 are identical with those
of the authentic sample, coumarin.
12. Kawata, H.; Kumagai, T.; Morita, T.; Niizuma, S. J.
Photochem. Photobiol., A: Chem. 2001, 138, 281.
13. Vollhardt, K. P. C., et al. Organic Chemistry; Freeman
and Company: New York, 1994; Chapter 23.
1. Ciamician, G.; Silber, P. Chem. Ber. 1902, 35, 4128.
2. Anet, F. A. L. Can. J. Chem. 1962, 40, 1249.
3. Schenck, G. O.; von Wilucki, I.; Krauch, C. H. Ber. 1962,
95, 1409.
4. Hammond, G. S.; Stout, C. A.; Lamola, A. A. J. Am.
Chem. Soc. 1964, 86, 3103.
14. Azuma, C.; Sugimori, A. Kougyou Kagaku Zasshi 1969,
72, 239.
5. Morrison, H.; Curtis, H.; MacDowell, T. J. Am. Chem.
Soc. 1966, 88, 5415.