Photochemistry of 2-(3-benzoylphenyl)propionic acid (ketoprofen): Part 1. A picosecond and nanosecond time resolved study in aqueous solution

Article abstract of DOI:10.1039/a700367f

The photochemistry of ketoprofen (KPF), 2-(3-benzoylphenyl)propionic acid, has been studied in aqueous solutions by time resolved (picosecond and nanosecond) spectroscopic techniques. Excited-state calculations were performed. The overall reactivity of th

Full text of DOI:10.1039/a700367f

View Article Online / Journal Homepage / Table of Contents for this issue
Photochemistry of 2-(3-benzoylphenyl)propionic acid (ketoprofen)
Part 1 A picosecond and nanosecond time resolved study in aqueous solution
Sandra Monti,a* Salvatore Sortino,a,b Guido De Guidib and Giancarlo Marconia
a Istituto di Fotochimica e Radiazioni dÏAlta Energia del CNR, Area della Ricerca,
V ia Piero Gobetti 101, I-40129 Bologna, Italy
b Dipartimento di Scienze Chimiche, Universita di Catania, V iale A. Doria 6, I-95125, Catania,
Italy
The photochemistry of ketoprofen (KPF), 2-(3-benzoylphenyl)propionic acid, has been studied in aqueous solutions by time
resolved (picosecond and nanosecond) spectroscopic techniques. Excited-state calculations were performed. The overall reactivity
of the molecule was rationalized within a scheme in which the lowest triplet state can undergo intramolecular electron transfer
and photoionization. The main pathway for the formation of the decarboxylated photoproducts was shown to involve intramole-
cular electron transfer and to pass through triplet biradicals.
D O was purchased from Carlo Erba. Water was puriÐed by
passage through a Millipore MilliQ system or puriÐed by
1
Introduction
2
Ketoprofen (KPF), 2-(3-benzoylphenyl)propionic acid, is a
non-steroidal anti-inÑammatory drug, frequently used for
therapeutic topical applications and known to act as a photo-
sensitizer for biological substrates both in vivo and in vitro.1,2
The process at the basis of the phototoxic reactions has
been recognized to be a very efficient photodecarboxylation
(quantum yield of 0.75), occurring in aqueous solutions and
leading to radicals and photoproducts with lytic activity on
cell membranes.1h3 The main products under aerobic condi-
tions are (3-benzoylphenyl)ethane, (3-benzoylphenyl)ethyl
hydroperoxide, (3-benzoylphenyl)ethanol and (3-benzoyl-
phenyl)ethanone and under anaerobic conditions only (3-ben-
zoylphenyl)ethane is formed.1,4 Minor traces of additional
products were isolated.4 The decarboxylation reaction was
shown to occur exclusively from the dissociated acid form
triple distillation. Phosphate bu†ers, 10~2 or 10~1 mol l~1,
were used in the pH range 6.3È8.7; higher pHs were obtained
by addition of NaOH. Acetonitrile (ACN), Carlo Erba RS
Ñuorimetric grade, was used as received.
Methods
Ultraviolet absorption spectra were measured by a Perkin-
Elmer Lambda 9 spectrophotometer.
Picosecond laser Ñash photolysis. Transient absorption
spectra in the subnanosecond time region were obtained by a
pumpÈprobe technique, described in detail elsewhere.6 Pump
pulses of 35 ps (FWHM) at 266 nm were generated by an
activelyÈpassively mode-locked Nd-YAG laser source
(Continuum PY62-10), by quadrupling the fundamental 1064
nm wavelength. The probe was a picosecond pulse with a con-
tinuum spectral distribution from 400 nm to the near infrared,
split in two beams to interrogate both the excited and the
unexcited portions of the sample. The pump and probe beams
were positioned to pass collinearly through the sample and to
overlap over a 1 cm pathlength. Sample and reference beams
emerging from the cell were focused into two optical Ðbres
coupled to a spectrograph (Spectrapro 275, Acton Research
Corp., 0.275 cm focal length) and to an optical multichannel
analyser, with a double diode array detector placed on the
focal plane (Princeton Instruments). The spectral resolution
was 0.58 nm. Controllers and an integrated software package
(P-OSMA) were provided by Spectroscopy Instruments. Time
resolution was obtained by varying the pathlength of the exci-
tation beam by means of a mechanical, computer-controlled,
delay line on which dichroic mirrors were mounted. In this
way, the time interval between excitation and probing was
adjusted. Consecutive spectra, taken at several time delays,
allowed the reconstruction of the time proÐles. The maximum
delay achieved was ca. 3 ns. Time calibration at 266 nm was
performed using Rhodamine 6G in methanol.7 The build up
of the S ] S absorption corresponded to the proÐle of the
(pK \ 4.7).4
a
Extrusion of CO ~ and a pathway involving the benzylic
2
radical has been hypothesized.1 The primary photochemical
event has been proposed as photoionization,4 with subsequent
release of carbon dioxide, producing the benzylic radical, and
efficient scavenging of the ejected electron by oxygen having
been suggested to justify the formation of the oxydation pro-
ducts. In the absence of oxygen, recombination of the benzylic
radical/electron pair to form a carbanion has been hypothe-
sized to account for (3-benzoylphenyl)ethane as the Ðnal
product. A path initiated by an intramolecular electron trans-
fer and involving the benzylic anion has also been proposed.5
Up to now, however, no time resolved investigations have
been performed to determine the photochemically active
excited states and the reaction intermediates.
To gain insight into the mechanism of the KPF decarboxyl-
ation process and to assess the role of the intermediates
involved in the photosensitization reactions in physiological
conditions, we carried out a laser Ñash photolysis study with
picosecond and nanosecond time resolution in neutral
aqueous medium. Excited-state calculations were also per-
formed. The production of singlet oxygen, one of the most
phototoxic agents, was investigated.
1
n
excitation pulse. “ZeroÏ time was assumed to be the time when
one half of the maximum absorbance was formed.
Experimental
Aerated solutions (typically, 1000 ml by volume, 1.0È1.5
absorbance over 1 cm at 266 nm) were circulated in a quartz
Ñow cell by a peristaltic pump to minimize the photo-
degradation. The laser, run at 10 Hz, delivered typically 2 mJ
Materials
Ketoprofen from Sigma was used as received. Benzophenone
(BP) from Baker was recrystallized three times from ethanol.
J. Chem. Soc., Faraday T rans., 1997, 93(13), 2269È2275
2269

View Article Online
pulse~1. Because of solubility limitations, picosecond experi-
ments were not performed with 355 nm excitation.
tion being very close to that of S . For BP the polar solvent
appears able to reverse the position of these states. This e†ect,
0
even if not experimentally observed, stands for the di†erent
character of the lowest triplet in the two molecules: in BP it is
a localized carbonyl n,n* triplet, and the solvent tends to
increase its energy di†erence with respect to the ground state,
whereas in KPF it exhibits more of a n,n* character. The elec-
tron density is in the latter case, more delocalized and is
mainly displaced on the phenyl carbon atoms adjacent to the
CxO group.
Nanosecond laser Ñash photolysis. The set-up for the nano-
second absorption measurements has been described pre-
viously.8 The laser beam (Nd-YAG, JK Lasers, pulse 20 ns
FWHM, 266 or 355 nm, or the picosecond 266 nm pulse) was
focused onto a 3 mm high ] 10 mm wide rectangular area of
the cell and the Ðrst 2 mm were analysed at right-angle
geometry. The energy used was O4 mJ pulse~1 (nanosecond
excitation) and ca. 4 mJ pulse~1 (picosecond excitation). Spec-
tral resolution was 2 nm. The sample absorbance was ca. 0.5
at 266 nm and 0.1 at 355 nm over 1 cm. Oxygen was removed
by vigorously bubbling the solutions with a constant Ñux of
argon, previously passed through a water trap to prevent
evaporation of the sample. The same procedure was used to
Picosecond laser Ñash photolysis
Aerated solutions of KPF, 7.0 ] 10~5 mol l~1, in phosphate
bu†er, 10~2 mol l~1, at pH 7.4, were excited at 266 nm by
laser pulses of 35 ps. Fig. 1 shows the absorbance changes 33
ps after time zero (see Experimental section), and at several
delays up to ca. 700 ps. The spectrum at 33 ps is characterized
by a broad absorption band extending from 400 to 700 nm
prepare either oxygen or N O saturated solutions. Care was
2
taken to renew the solution at each laser shot. The tem-
perature was 295 ^ 2 K.
with j \ 526 nm, closely similar in shape and intensity to
max
that of BP in water (taken under the same experimental
Time resolved IR emission. Singlet oxygen was generated in
air-saturated acetonitrile or deuteriated bu†er solutions from
triplet sensitizers excited at 532, 355 or 266 nm. The near IR
conditions) and is attributed to the lowest triplet state of the
ketone (transient 1 in Scheme 1). The di†erence spectrum
undergoes
a fast evolution during the subsequent few
luminescence resulting from the forbidden transition 3& ~ ^
g
1* of molecular oxygen was observed by means of a time
g
hundreds of picoseconds: the absorbance decreases at 526 nm
while it increases at longer wavelengths. An absorption band
with a maximum at 570È580 nm appears, which corresponds
to the formation of a new transient 2 (in Scheme 1). The time
proÐles of the absorbance changes at 526 nm and 570 nm are
shown in Fig. 2. The kinetics, in both the increasing and the
decreasing part, is described well by an exponential function
resolved detection system, as already described.9 Single expo-
nential analysis of the emission decay proÐle was performed,
with the exclusion of the initial fast signal resulting from scat-
tered excitation light or from sensitizer Ñuorescence. The life-
time was 55 ls in acetonitrile with Rose Bengal as sensitizer
(j \ 532 nm), and 60 ls in D O with benzophenone as sen-
exc
2
with lifetime q \ 250 ^ 50 ps. A change of pH up to 13 does
sitizer (j \ 266 nm), in agreement with previous literature.7
1
exc
not vary either the spectral features of transients 1 and 2 or
The value of the intercept at t \ 0 was taken as proportional
to the singlet-oxygen concentration. Quantum yields (U ) of
KPF in acetonitrile were calculated from the slope of a linear
the kinetics of their decay and formation, respectively.
*
Nanosecond laser Ñash photolysis
plot obtained by varying the laser energy and by using opti-
cally matched acetonitrile solutions (A B 0.2) of BP (U \
Detection of transient absorption with nanosecond resolution
from KPF, 3.5 ] 10~5 mol l~1, in argon-saturated phosphate
bu†er, upon excitation with the 35 ps laser pulse at 266 nm,
provides the absorbance changes reported in Fig. 3. The spec-
trum, measured 4 ns after the pulse, is characterized by
maxima at 330 and at 570È580 nm and in the visible region is
qualitatively similar to that measured by means of the pico-
second detection system. Its time evolution, if described by an
exponential function, is characterized by lifetimes (ca. 180 ns)
that somewhat depend on the observation wavelength. The
di†erence spectrum 800 ns after the pulse is characterized by
maxima at 330 and 520 nm.
355
*
0.37) or Rose Bengal (U \ 0.54) as standard.10
*
Acquisition and processing of the transient absorption and
emission signals were performed by a home-made program
using Asyst 3.1 (Software Technologies).
Results
Electronic states of KPF
The absorption spectrum of KPF in phosphate bu†er 10~2
mol l~1 at pH 7.4 is closely similar to that of the parent BP
molecule. It is, in fact, characterized by a band with j
at
max
260 nm (e \ 15 700 l mol~1 cm~1) attributed to the
260
S ] S transition of n,n* nature, and by a lower energy
0
2
shoulder in the 300È350 nm region corresponding to the
S ] S forbidden transition of n,n* parentage.
0
1
The properties of the excited states of KPF and BP were
investigated by theoretical methods. The geometries of the iso-
lated molecules were optimized using the MM3-92 AllingerÏs
molecular mechanics program.11 The MM3 optimized struc-
tures were taken as the starting points for a CNDO/S
(complete neglect of di†erential overlap/spectroscopic) calcu-
lation of the excited states [repulsion integrals in the OhnoÈ
Klopmans approximation, conÐguration interaction (CI) \ 60
singly excited conÐgurations], with inclusion of the solvent
e†ect through the method of virtual charges.12 The lowest
singlets are calculated at 3.44 (360 nm, f \ 0.0108) and 3.29 eV
(378 nm, f \ 0.0009) for KPF and BP, respectively, the former
having more of a n,n* nature, whereas the latter has a local-
ized carbonylic n,n* character. As far as the triplets are con-
cerned, the results of the calculation show the presence of two
very close triplets at 3.08 and 3.15 eV for KPF and at 3.09 and
3.145 eV for BP in vacuo. The two triplets of KPF are almost
una†ected in energy by the solvent, their electronic distribu-
Fig. 1 Absorbance changes evolution (cell pathlength 1 cm) observed
in a KPF 7.0 ] 10~5 mol l~1 solution in aerated phosphate bu†er
10~2 mol l~1 at pH 7.4, after a 35 ps laser pulse of 2.8 mJ at 266 nm
2270
J. Chem. Soc., Faraday T rans., 1997, V ol. 93

View Article Online
Fig. 2 Time proÐle of the absorbance changes at 526 (K) and 570
Fig. 5 Dependence of the ratio between the absorbance change (at
10 ns delay) and the pulse energy (A/E) on the pulse energy (E) for
KPF in the same experimental conditions as those in Fig. 4. K, 590
nm (…). Experimental conditions as in Fig. 1. Laser pulse energy 1.7
mJ.
nm in N O-saturated solution and …, 720 nm in Ar-saturated solu-
2
tion.
Excitation of the argon-saturated solution with a 20 ns laser
pulse at 266 nm, produces the absorbance changes reported in
Fig. 4. The spectrum taken 10 ns after the pulse is character-
ized by maxima at 330 and ca. 600 nm, with the visible
absorption extending toward 800 nm in a broad and intense
band. Upon saturation of the solution with N O (2.2 ] 10~2
2
mol l~1), a large part of the visible contribution disappears
and a well deÐned band with a maximum at 570È580 nm is
evidenced. Correspondingly, small variations are observed in
the UV region. It is concluded that the hydrated electron (e ),
aq
characterized by a broad absorption extending from 300 to
1200 nm with a maximum at 720 nm (e \ 1.9 ] 104 l
720
mol~1 cm~1) in aqueous medium,13 contributes signiÐcantly
to the early spectral variations and is completely scavenged by
N O in a few ns (k \ 9.1 ] 109 l mol~1 s~1).14
2
q
The dependence of the transient absorbance on the energy
of the nanosecond laser pulse was examined in the range 0È15
mJ. At 720 nm in argon-saturated solution it has a parabolic
shape at low energies, indicating the dominance of a two-
photon process. By admitting a mixed one-photon and two-
photon mechanism, in the limit of low pulse energies, the
relation A \ aE ] bE2 is valid,15 E being the pulse energy,
720
proportional to the photon Ñuence, a a coefficient that
depends on the quantum yield of the one-photon process, and
b a coefficient that depends on the absorption coefficients and
yields of the intermediate states of the consecutive two-photon
Fig. 3 Absorbance changes (cell pathlength 1 cm) observed in a
KPF 3.5 ] 10~5 mol l~1 solution in Ar-saturated phosphate bu†er,
10~2 mol l~1, at pH 7.4, upon excitation with a 35 ps, 4.2 mJ laser
pulse at 266 nm. …, 4 ns and =, 800 ns after the pulse.
process. Fig. 5 shows a plot of A /E vs. E. Linearity is good
720
up to 1 mJ pulse~1. From the intercept, by using K Fe(CN)
4
6
(Ue \ 0.52)16 as standard, a one-photon quantum yield of
0.017 for e production from KPF can be determined. The
positive slope shows that the biphotonic contribution is not
aq
aq
negligible even at moderate pulse energies. With the 35 ps
pulse the hydrated electron absorption is negligible, indicating
the dominance of a path that requires the population of the
triplet state for the absorption of the second photon.17h19 ¤ At
590 nm, in N O-saturated solutions, the dependence of the
2
absorbance change on the pulse energy is linear at low ener-
gies and has a saturating behaviour at higher energies in
agreement with the occurrence of a monophotonic process.
Fig. 5 shows that A /E vs. E is constant up to 0.4 mJ
590
pulse~1.
On the basis of the above results it can be safely concluded
that production of hydrated electrons and the production of
transient 2 are parallel, independent processes. This can be
conÐrmed by changing the excitation wavelength. Ar-
saturated KPF 1.2 ] 10~3 mol l~1 solutions in phosphate
Fig. 4 Absorbance changes (cell pathlength 1 cm) observed in a
KPF 3.5 ] 10~5 mol l~1 solution in phosphate bu†er, 10~2 mol l~1,
at pH 7.4, upon excitation with a 20 ns, 2.5 mJ laser pulse at 266 nm.
¤ The possible occurrence of other biphotonic processes from
highly excited triplets, as already reported for benzophenone deriv-
atives (see ref. 8), cannot be excluded.
…, Ar-saturated solution and =, N O-saturated solution, 10 ns after
2
the pulse.
J. Chem. Soc., Faraday T rans., 1997, V ol. 93
2271

View Article Online
Fig. 6 Absorbance changes (cell pathlength 1 cm) observed in a
KPF 1.2 ] 10~3 mol l~1 solution in Ar-saturated phosphate bu†er,
10~1 mol l~1, at pH 7.4, upon excitation with a 20 ns, 3.5 mJ laser
pulse at 355 nm. …, End of pulse and =, 400 ns after pulse.
Fig. 7 Absorbance changes (cell pathlength 1 cm) observed in KPF
1.2 ] 10~3 mol l~1 solutions in Ar-saturated phosphate bu†er, 10~1
mol l~1, at di†erent pH values, upon excitation with a 20 ns, 2.5 mJ
laser pulse at 355 nm, taken 2 ls after the pulse. …, 330 nm and =,
520 nm.
At pH [ 8.7 no transients with microsecond lifetimes are
detected. The rate of disappearance of the absorption changes
occurs in the submicrosecond time domain over the whole
wavelength range: the absorbance changes at the end of the
laser pulse show a maximum at 570 nm and a decay with
apparent single exponential kinetics with time constant ca.
250 ns, maintaining the maximum at 570 nm for the whole
decay. In Fig. 8(c) the decay proÐle observed at 520 nm is
reported.
bu†er, 10~1 mol l~1,” at pH 7.4, excited by a 20 ns laser pulse
of 355 nm, gives the 570 nm transient, without any contribu-
tion of the solvated electron (Fig. 6). The end of pulse absorb-
ance decays, both in the UV and in the visible region, with
very good single exponential kinetics with lifetime q \ 120
2
^ 15 ns, conÐrming the presence of only one species. The
kinetics do not depend on the ground-state concentration, on
the laser pulse energy or on the presence of oxygen. It is
attributed to an intramolecular reaction.
The di†erence spectrum is characterized by maxima at ca.
330 nm and 520 nm 400 ns after the pulse (Fig. 6). The decay
kinetics appear biexponential with time constants 4 ^ 1 ls
and 12 ^ 3 ls. The short-lived component is largely domi-
nant. Saturation with air and oxygen makes the decay mono-
exponential with lifetimes of 1.6 ls (air) and 0.5 ls (oxygen). It
can be concluded that the intermediate with a 4 ls lifetime
Singlet-oxygen production
Excitation of KPF 4.4 ] 10~3 mol l~1 in acetonitrile at 355
nm produces singlet oxygen (1O ) by energy transfer from
2
triplet KPF to molecular oxygen. The 1270 nm 1* emission
g
of this species was detected by means of a time resolved
apparatus. The lifetime was found to be 25 ls, shorter than
that observed with BP under the same experimental condi-
tions (45 ls). This could be in agreement with a reactivity of
interacts with O with a bimolecular rate parameter of
2
2 ] 109 l mol~1 s~1, thereby no longer producing the species
with a 12 ls lifetime. The pH dependence was examined in the
range 6.3È12.5. The intensity at both 330 and 520 nm depends
on pH following a sigmoidal behaviour (Fig. 7) consistent
1O with the starting compound and/or with the photo-
2
products. However, this does not a†ect the estimation of the
yield U , which is obtained by a procedure of extrapolation of
*
with an equilibrium of pK B 7.6. The decay kinetics are biex-
the signal at t \ 0 (see Experimental section). By using either
a
ponential from pH ca. 7.2 up to pH 8.7. The 4 ls lifetime
BP (U) \ 0.37) or Rose Bengal (U \ 0.54) in acetonitrile as
*
*
remains constant. The longer lifetime varies with pH and the
wavelength of analysis. At pH \ 7.2 the time dependence of
the microsecond absorption cannot be described by a biex-
ponential treatment. In Fig. 8 the decay proÐles observed at
520 nm are reported [(a) and (b)].
standards,10 U (KPF) is 0.3 ^ 0.05. This value is in agree-
*
ment with an n,p* character for the lowest triplet state.20 The
same U (KPF) is obtained approximately by using 266 nm
*
excitation, provided U (BP), taken as reference, is assumed not
*
to vary with the excitation wavelength.
KPF, 3.8 ] 10~5 mol l~1, in deuteriated phosphate bu†er
solution, does not give any detectable signal of 1O emission.
” In these experiments the ionic strength was 10~1 M; however, no
2
inÑuence of this parameter was observed.
Only after several laser pulses have been delivered to the
Fig. 8 Time proÐles of the absorbance changes at 520 nm observed in KPF 1.2 ] 10~3 mol l~1 solutions in Ar-saturated phosphate bu†er, 10~1
mol l~1, upon excitation with a 20 ns, 2.5 mJ laser pulse at 355 nm. (a) pH 6.5; (b) pH 7.4 and (c) pH 12.5.
2272
J. Chem. Soc., Faraday T rans., 1997, V ol. 93

View Article Online
sample does the signal appear, clearly indicating that 1O is
for KPF photodecarboxylation. The key intermediates were
thought to be the benzylic radical1,4 or the benzylic anion.4,5
The Ðrst could hardly account for the high quantum yield of
formation of (3-benzoylphenyl)ethane in water, where
abstractable H atoms are absent. The involvement of this
radical in the main photodecarboxylation path in the absence
of oxygen is therefore excluded under our experimental condi-
tions. An anionic mechanism was found to be operative in the
photodecarboxylation of the KPF sodium salt in methanol,
following the observation that deuterium is incorporated into
the benzylic position of the photodecarboxylated product
2
produced by the photoproducts of KPF.
Discussion
The ground-state absorption properties of KPF in aqueous
medium indicate that the lowest singlet excited states are
similar to those of BP. Moreover the shape and the intensity
of the TÈT absorption spectra in the visible region (with
j
\ 526 nm in KPF and 520 nm in BP) show that these
max
molecules have quite similar intersystem crossing quantum
yields (ca. 1) and triplet absorption coefficients. However semi-
empirical calculations indicate that the n,p* and p,p* charac-
ters of the excited states in KPF are less clear-cut than in BP.
In particular the lowest triplet is a localized carbonyl state of
“pureÏ n,p* character in BP, whereas in KPF it is more delo-
calized and assumes a partial p,p* character. The solvent
e†ect on the triplet energy is in agreement with the substan-
tially di†erent electronic distribution. The polarity of the
environment raises the energy of the BP triplet, which has a
dipole moment lower than the ground state,21 while it leaves
the KPF triplet almost unperturbed with a dipole moment
similar to that of the ground state. Consequently, the dynamic
behaviour of the two states is quite di†erent in a polar
environment. In BP, the triplet reactivity is mainly H-
abstraction oriented, being started by the n vacant orbital.
The process is strongly disfavoured in water, so that the state
is very long lived (q D 100 ls).22 In KPF (dissociated acid
form) the main reaction is photodecarboxylation and a very
when the reaction is conducted in CH OD.4 The ground state
3
benzyl anion was then proposed to be the key intermediate for
the formation of (3-benzoylphenyl)ethane in water at neutral
pH, in the absence of oxygen.4 On the other hand Budac and
Wan5 suggested the formation of the benzyl anion, which is
promoted by intramolecular electron transfer in the triplet
state.5 This view is conÐrmed by the present study.
In Scheme 1 an adiabatic mechanism (path I) involving
several ionic intermediates is proposed, consistent with the
observation that the undissociated acid does not undergo loss
of CO .4 In the triplet state of the carboxylate anion an intra-
2
molecular electron transfer from the carboxyl (donor) to the
carbonyl (acceptor) groups leads to the triplet biradical 2, with
j
\ 570 nm. This is in agreement with the high rate of the
max
process and with its monophotonic nature; moreover the
absorption features of 2 are consistent with its benzophenone-
like anion structure.22 Release of CO proceeds adiabatically
2
from 2, with a time constant of 120 ns, and the transient 3 is
fast triplet decay (q \ 250 ps) is observed. The quantitative
formed. At pH \ 7.4, transient 3 quickly achieves protonation
1
intersystem crossing (ISC) and the high quantum yields of the
equilibrium, the observed pK B 7.6 (Fig. 7), being closely
a
photoprocess (0.75) rule out any role for the KPF singlet state.
Previous literature considered radical or anionic pathways
similar to the pK of aromatic ketyl radicals (for example, the
a
ketyl radical of 4-carboxybenzophenone has pK \ 8.223). The
a
Scheme 1
J. Chem. Soc., Faraday T rans., 1997, V ol. 93
2273

View Article Online
biradical 4 is formed, which is identiÐed with the intermediate
reaction, the KPF triplet does not form singlet oxygen in
aqueous medium. This species is produced in the system only
upon accumulation of the decarboxylated photoproducts. In a
medium where KPF does not undergo photodecarboxylation,
as, for example, in acetonitrile, the quantum yield of singlet
oxygen produced (ca. 0.3) is similar to that of BP. This indi-
cates that the two molecules in a non-aqueous environment
have closely similar triplet lifetimes and oxygen quenching
rate parameters.
with a 4 ls lifetime, very sensitive to dissolved oxygen (k B 2
q
] 109 dm3 mol~1 s~1). The spectral properties of this
biradical (j
at 330 and 520 nm) are in agreement with the
max
presence of a benzophenone-like ketyl radical centre.22 The
decay of 4 to the Ðnal product requires ISC and involves
further long-lived intermediates (q \ 12 ls): an intramolecular
H-shift is needed for the conversion to (3-benzoylphenyl)
ethane. At pH [ 8.7 (Fig. 8) the end of pulse absorption in
both the UV and the visible region decays with an overall
time constant similar to the lifetime of 2. Since pH has no
inÑuence on the quantum yield of the decarboxylation over
Conclusions
the range 6.3È12.5,24
a
fast conversion of
3
to
This study shows that the photochemistry of ketoprofen in
neutral aqueous medium is triplet initiated and involves triplet
biradicals with a radical centre on the carbonyl carbon. The
main reaction is an intramolecular electron transfer from the
carboxyl (donor) to the carbonyl (acceptor) groups leading to
adiabatic decarboxylation (pathway I). This pathway is largely
dominant. A minor channel for decarboxylation is photoion-
ization from the triplet state. The radical formed upon elec-
(benzoylphenyl)ethane via ISC and intramolecular electron
transfer from the oxygen site to the benzylic site, followed by
protonation from solvent of the formed carbanion, is pro-
posed to account for the formation of the Ðnal product.
In this mechanism, the triplet nature of the long-lived
biradical 4 is an important point. The species 4 has a non-
Kekule structure, essentially similar to that of the m-
quinodimethane biradical; this latter is known to be a triplet
in the ground state, long-lived enough to be trapped by a
diene in Ñuid solution at 25 ¡C.25 The observed efficient
oxygen quenching of 4 accounts well both for the lower yield
of (3-benzoylphenyl)ethane in air-saturated neutral solutions
and for the formation of the oxidation products.1 Since singlet
oxygen is not produced efficiently in the system while the
superoxide anion radical was found to be present in high
yields,1 we conclude that in the quenching act the latter is
preferentially formed.
tron release is proposed to lose CO thereby producing the
2
benzylic radical (pathway II). This pathway is scarcely efficient
at low light levels but might become important with high light
intensities.
Financial support from the Progetto Strategico “Tecnologie
Chimiche InnovativeÏ of the Consiglio Nazionale delle
Ricerche and from the Progetto “Proprieta Chimico-Fisiche
dei Medicamenti e loro Sicurezza dÏUsoÏ of the Instituto
Superiore di Sanita is acknowledged. Fruitful discussions with
Prof. Angelo Albini (University of Pavia, Italy), critical
reading of the manuscript by Dr Pietro Bortolus and technical
support of Mr M. Minghetti and Mr R. Cortesi are also grate-
fully acknowledged.
A triplet adiabatic path for decarboxylation was proposed
to be operative in p-nitrophenylacetate in aqueous medium by
Craig et al.26 The only two intermediates detected in that
system were assigned to the triplet state of the p-nitrobenzyl
anion (with a lifetime of ca. 90 ns) and the long-lived and
oxygen-insensitive ground state of the same species (a lifetime
of ca. 60 s in aerated solution). Such a mechanism could in
principle be operative in KPF. We could identify the 250 ps
References
process as the loss of CO directly from the KPF triplet state
1
2
3
L. L. Costanzo, G. De Guidi, G. Condorelli, A. Cambria and M.
Fama, Photochem. Photobiol., 1989, 50, 359.
2
and the 120 ns process as the T ] S ISC of the decarboxyl-
1
0
G. Condorelli, G. De Guidi, G. Giu†rida and L. L. Costanzo,
Coord. Chem. Rev., 1993, 125, 115.
ated benzyl anion. However, the evolution of the ground-state
benzyl anion should proceed on the microsecond timescale
through several hard to assign intermediates. This mechanism
is therefore discarded.
G. De Guidi, G. Condorelli, S. Giu†rida, G. Puglisi and G.
Giammona, J. Inclusion Phenom. Mol. Recognit. Chem., 1993, 15,
43.
F. Bosca, M. A. Miranda, G. Carganico and D. Mauleon, Photo-
chem. Photobiol., 1994, 60, 96.
As for many arylalkyl carboxylates in aqueous medium,27
we observe production of hydrated electrons from the lowest
4
5
6
D. Budac and P. Wan, J. Photochem. Photobiol., A, 1992, 67, 135.
P. Bortolus, F. Elisei, G. Favaro, S. Monti and F. Ortica, J.
Chem. Soc., Faraday T rans., 1996, 92, 1841.
triplet of KPF with j \ 266 nm. The reaction is, partly
exc
monophotonic (Ue \ 0.017), partly biphotonic and is paral-
aq
lel to the main intramolecular deactivation process (Fig. 5). An
7
8
9
F. Wilkinson, W. P. Helman and A. B. Ross, J. Phys. Chem., Ref.
Data, 1995, 24, 663.
additional photocleavage mode in KPF could then involve
the radical partner of e , 2@, which can release CO and lead
S. Monti, N. Camaioni and P. Bortolus, Photochem. Photobiol.,
1991, 54, 577.
aq
2
to the benzylic radical 3@ (pathway II in Scheme 1). Species 3@
via second-order radicalÈradical recombination processes,
could account for the formation of the corresponding dimer,
which is found as a minor product of the KPF photo-
decarboxylation process.4 Under our experimental conditions
we were not able to clearly distinguish the intermediates of
P. Bortolus, S. Monti, A. Albini, E. Fasani and S. Pietra, J. Org.
Chem., 1989, 54, 534.
10 F. Wilkinson, W. P. Helman and A. B. Ross, J. Phys. Chem., Ref.
Data, 1993, 22, 113.
11 N. L. Allinger, Y. Hi. Yuh and J. M. Lii, J. Am. Chem. Soc., 1989,
111, 8551; N. L. Allinger, Quantum Chem. Program Exchange
Bull., 1989, 9, 2.
this pathway, except for e . However, transient phenomena
aq
12 I. Jano, Chem. Phys. L ett., 1984, 106, 60.
observed exclusively with 266 nm excitation are reasonably
13 G. L. Hug, Optical Spectra of Non-Metallic Inorganic T ransient
Species in Aqueous Solution, Natl. Stand. Ref. Data Ser. 69,
National Bureau of Standards, Washington DC, 1981.
14 G. V. Buxton, C. L. Greenstock, W. P. Helman and A. B. Ross, J.
Phys. Chem., Ref. Data, 1988, 17, 513.
attributed to the existence of pathway II.°
In agreement with the presence of a fast intramolecular
° The qualitative absorption properties of the radical partner of e
,
aq
2@, can be derived by subtracting from the absorbance changes
15 U. Lachish, A. Sha†ermann and G. Stein, J. Chem. Phys., 1976,
64, 4205.
16 M. Shirom, M. and G. Stein, J. Chem. Phys., 1971, 55, 3372.
17 F. Elisei, G. Favaro and H. Gorner, J. Photochem. Photobiol., A,
1991, 59, 243.
18 N. Ishiwata, H. Murai and K. Kuwata, J. Phys. Chem., 1993, 97,
7129.
19 J. Sauberlich and D. Beckert, J. Phys. Chem., 1995, 99, 12520.
obtained 10 ns after pulse in N O-saturated solution with 266 nm
2
excitation, the corresponding ones obtained using 355 nm excitation
(upon normalization of them at 580 nm). The di†erence spectrum has
a maximum at ca. 330 nm and is negligible in the visible. The presence
of this species could be related to the formation, on the microsecond
time scale, of a product absorbing at 380È400 nm, not detected with
355 nm excitation.
2274
J. Chem. Soc., Faraday T rans., 1997, V ol. 93

View Article Online
20 A. P. Darmanyan and C. S. Foote, J. Phys. Chem., 1993, 97, 4573.
21 H. Shimamori, H. Uegaito and K. Houdo, J. Phys. Chem., 1991,
95, 7664.
22 S. Baral-Tosh, S. K. Chattopadhyay and P. K. Das, J. Phys.
Chem., 1984, 88, 1404; D. A. Nelson and E. Hayon, J. Phys.
Chem., 1972, 76, 3200.
23 S. Inbar, H. Linschitz and S. G. Cohen, J. Am. Chem. Soc., 1981,
103, 7323.
24 S. Sortino, unpublished results.
25 J. L. Goodman and J. A. Berson, J. Am. Chem. Soc., 1985, 107,
5409.
26 B. B. Craig, R. G. Weiss and S. J. Atherton, J. Phys. Chem., 1987,
91, 5906.
27 S. Steenken, C. J. Warren and B. C. Gilbert, J. Chem. Soc., Perkin
T rans. 2, 1990, 335.
Paper 7/00367F; Received 15th January, 1997
J. Chem. Soc., Faraday T rans., 1997, V ol. 93
2275