Efficient photodecarboxylation of 3- and 4-acetylphenylacetic acids in aqueous solution

Article abstract of DOI:10.1139/v04-122

The photochemistry of 3- and 4-acetylphenylacetic acids (6 and 7) has been studied in aqueous solution. This work is a continuation of research efforts aimed at understanding the structural effects on the efficacy for benzyl carbanion photogeneration via photodecarboxylation. The nitro group (at the 3- and 4-positions) is known to be an exceptionally good activating group on the benzene ring - because of its enhanced electron-withdrawing effect in the excited triplet state - for photodecarboxylation and the related photo-retro-aldol type process. It is shown in this work that the acetyl group is an equally good activating group for the photodecarboxylation. Thus, the photochemistry of 6 and 7 parallels much of what was observed for the corresponding nitrophenylacetic acids 1 and 2. Both 6 and 7 photodecarboxylate efficiently (Φ_d = 0.60 and 0.22, respectively, at pH 7) via the carboxylate form, to give observable (by laser flash photolysis) benzyl carbanion or related intermediates. The meta isomer 6 displays an acid-catalyzed pathway for photodecarboxylation at pH < 3 and along with its enhanced overall reactivity, is consistent with a meta effect of the acetyl group. Triplet state reactivity is inferred from sensitization and laser flash photolysis experiments. Based on the results of this work, the acetyl group may now be viewed as an "enhanced" electron-withdrawing group (in the excited state) when attached to a benzene ring (i.e., in acetophenone derivatives) that is capable of inducing ionic reactions, which is not the traditional photochemistry that is expected from such compounds.

Full text of DOI:10.1139/v04-122

1760
Efficient photodecarboxylation of 3- and 4-
acetylphenylacetic acids in aqueous solution
Lawrence A. Huck, Musheng Xu, Kaya Forest, and Peter Wan
Abstract: The photochemistry of 3- and 4-acetylphenylacetic acids (6 and 7) has been studied in aqueous solution.
This work is a continuation of research efforts aimed at understanding the structural effects on the efficacy for benzyl
carbanion photogeneration via photodecarboxylation. The nitro group (at the 3- and 4-positions) is known to be an ex-
ceptionally good activating group on the benzene ring — because of its enhanced electron-withdrawing effect in the
excited triplet state — for photodecarboxylation and the related photo-retro-aldol type process. It is shown in this work
that the acetyl group is an equally good activating group for the photodecarboxylation. Thus, the photochemistry of 6
and 7 parallels much of what was observed for the corresponding nitrophenylacetic acids 1 and 2. Both 6 and 7
photodecarboxylate efficiently (Φd = 0.60 and 0.22, respectively, at pH 7) via the carboxylate form, to give observable
(by laser flash photolysis) benzyl carbanion or related intermediates. The meta isomer 6 displays an acid-catalyzed
pathway for photodecarboxylation at pH < 3 and along with its enhanced overall reactivity, is consistent with a meta
effect of the acetyl group. Triplet state reactivity is inferred from sensitization and laser flash photolysis experiments.
Based on the results of this work, the acetyl group may now be viewed as an “enhanced” electron-withdrawing group
(in the excited state) when attached to a benzene ring (i.e., in acetophenone derivatives) that is capable of inducing
ionic reactions, which is not the traditional photochemistry that is expected from such compounds.
Key words: photodecarboxylation, acid catalysis, acetophenone, photogenerated carbanion, enol, enolate.
Résumé : On a étudié la photochimie des acides 3- et 4-acétylphénylacétiques (6 et 7) en solution aqueuse. Ce travail
est une continuation des efforts de recherche déployés pour comprendre les effets structuraux sur l’efficacité de la pho-
togénération du carbanion benzyle par le biais d’une photodécarboxylation. Il est connu que, en raison de
l’augmentation de leur effet électroaffinitaire dans l’état triplet excité, les groupes nitro (en positions 3 et 4) activent le
noyau benzénique d’une façon exceptionnelle lors de réactions de photodécarboxylation et autres processus apparentés
du type photo-rétro-aldolique. Dans ce travail, on a démontré que le groupe acétyle est aussi un bon groupe activant
pour la réaction de photodécarboxylation. La photochimie des composés 6 et 7 est donc parallèle à ce qui a été ob-
servé avec les acides nitrophénylacétiques correspondants, 1 et 2. Les composés 6 et 7 se décarboxylent d’une façon
efficace (Φd = 0,60 et 0,22 respectivement, à un pH de 7), par le biais de la forme carboxylate, pour conduire à la for-
mation du carbanion benzyle et d’autres intermédiaires apparentés qui peuvent être observés par photolyse éclair au la-
ser. À des pH < 3, l’isomère méta, 6, donne lieu à une photodécarboxylation impliquant une catalyse acide et, combiné
avec sa plus grande réactivité globale, cette observation est en accord avec un effet méta du groupe acétyle. On déduit
la réactivité de l’état triplet d’expériences de sensibilisation et de photolyse éclair au laser. Sur la base des résultats de
ce travail, on peut considérer que, dans l’état excité, l’effet électroaffinitaire d’un groupe acétyle attaché à un noyau
benzénique (comme dans les dérivés de l’acétophénone) est rehaussé, qu’il est capable d’induire des réactions ioniques
ce qui ne correspond pas à la photochimie traditionnelle à laquelle on peut s’attendre avec de tels composés.
Mots clés : photodécarboxylation, catalyse acide, acétophénone, carbanion photogénéré, énol, énolate.
[Traduit par la Rédaction] Huck et al. 1768
Introduction
sively for a variety of organic substrates, where a number of
mechanistic pathways are known to operate for the loss of
CO2 initiated by light (3). With respect to the photodecar-
boxylation of aromatic compounds, phenylacetic acids have
been the most studied as they provide the obvious basis for
Hammett-type studies, by varying substituents on the phenyl
ring (3). Although we are not aware of any extended system-
atic studies in this approach — perhaps thwarted by the fact
that the parent and many simple alkyl-substituted phenyl-
acetic acids react with only low quantum efficiency (Φ <
0.02) (3) — it is known that 3- and 4-nitrophenylacetic acids
(1 and 2, respectively, and related compounds) both react
with very high quantum efficiency in the carboxylate form
(Φ = 0.6) (4) via nitrobenzyl carbanion intermediates (e.g.,
Decarboxylation is one of the simplest extrusion reactions
in organic chemistry with wide-relevance in both organic
synthesis (1) and in biological chemistry (1, 2). Photodecar-
boxylation is also well-known and has been studied exten-
Received 19 April 2004. Published on the NRC Research
Press Web site at http://canjchem.nrc.ca on 3 February 2005.
L.A. Huck, M. Xu, K. Forest, and P. Wan.1 Department of
Chemistry, Box 3065, University of Victoria, Victoria, BC
V8W 3V6, Canada.
1Corresponding author (e-mail: pwan@uvic.ca).
Can. J. Chem. 82: 1760–1768 (2004)
doi: 10.1139/V04-122
© 2004 NRC Canada

Huck et al.
1761
3-nitrobenzyl carbanion (3) in eq. [1]). More recently, we
have shown (5) that both the benzoyl (PhCO) and acetyl
(CH3CO) moieties also act as “enhanced” electron-
withdrawing groups (when attached to a benzene ring) in the
excited state, capable of mediating highly efficient photo-
decarboxylations of the corresponding substituted phenyl-
acetic acid, providing a rational explanation for the
previously reported efficient photodecarboxylation of keto-
profen (4) (via benzyl carbanion 5, eq. [2]), a nonsteroidal
anti-inflammatory drug that is known to be phototoxic (6).
In this work, we detail our complete results on a study of the
photodecarboxylation of 3- and 4-acetylphenylacetic acids
(6 and 7, respectively) and show that their photochemical
behaviour essentially parallels that observed for the corre-
sponding nitro-substituted compounds 1 and 2. Moreover,
the results show that the acetyl group may be thought of as
an enhanced “electron-withdrawing” group in the excited
state, capable of inducing photochemistry that would not
have been normally be anticipated when these compounds
are viewed as substituted acetophenones.
initial NBS bromination of the methyl group, followed by
substitution of the bromine by cyanide ion, and subsequent
acid hydrolysis. Photolysis of 6 in H2O–CH3CN (1:1, pH 7,
water portion, 10–3 mol L–1, 254 or 300 nm, Rayonet RPR
100 photochemical reactor, 1–10 min, deoxygenated by ar-
gon purge) gave exclusively the corresponding photodecar-
boxylation product 8 in high yield (>80%, eq. [3]).
Quantitative conversion was observed at sufficiently long
photolysis times. No reaction was observed in the absence of
light or when photolyzed in neat CH3CN, the latter consis-
tent with photoreactivity via the carboxylate ion (vide infra).
Photolysis in D2O–CH3CN (1:1, pD 1 and 7) gave exclu-
sively the corresponding α-deuterated product 8-αD, consistent
with the formation of a benzyl-carbanion-type intermediate
in the photodecarboxylation. The overall reaction yield was
only about 10%–20% lower in D2O, indicative of a second-
ary solvent isotope effect for reaction, although the exact
source of this effect is not known.
Similar photolysis of the para-isomer 7 gave two products
at all conversions, the expected photodecarboxylation prod-
uct 9 and a “dimer” 10 (formally derived from the coupling
of two 4-acetylphenylmethyl radicals) (eq. [4]). Independent
photolysis of 9 did not produce 10. Photolysis in D2O–
CH3CN gave the α-deuterated product 9-αD, consistent with
the formation of a benzyl-carbanion-type intermediate; no
deuterium was incorporated into dimer 10. The ratio of
Results and discussion
Product studies
The substrates for study (6 and 7) were readily prepared
from the corresponding methylacetophenones (8 and 9) by
© 2004 NRC Canada

1762
Can. J. Chem. Vol. 82, 2004
12 (τ ≈ 60 s, aqueous solution, pH 7) (7) would indicate
slow reaction via simple protonation of the anion (carbon
protonation is generally “slow” even in thermodynamically
favourable cases), and hence, the possibility of dimerization
to give radical dianion 14. Such a mechanism does not oper-
ate for the 3-isomer 1 since the corresponding benzyl car-
banion intermediate 3 is expected to be shorter lived.
Indeed, the 3-nitrobenzyl carbanion (3) has never been de-
tected as a transient, although the reason for this is still not
clear.
The high yields and clean photochemistry observed for
both 6 and 7 is further evidence for the enhanced activating
effect of the acetyl group for photodecarboxylation. More-
over, no products were observed that are derived from the
photochemistry of the carbonyl end of the molecule (e.g.,
via photoreduction or hydrogen abstraction), which is con-
sistent with the notion that this group participates only as an
enhanced electron-withdrawing group. Without the acetyl
group, phenylacetic acid itself photodecarboxylates with
only marginal quantum yields (<0.02) (3). That there is sub-
stantial charge transfer character (with negative charge po-
larized towards the carbonyl oxygen) in the reactive excited
state of these compounds is further evidenced by observation
of acid catalysis for the photodecarboxylation of 6 (vide in-
fra).
Table 1. Ratio of products 9 vs. 10 observed in the
photodecarboxylation of 7.a
pHb
9 (%)
10 (%)
2.5
3.2
5.2
7.0
8.0
10.2
12.0
90
83
80
78
79
77
72
10
17
20
22
21
23
28
1
aRelative yields were calculated by H NMR. Estimated
errors: 10% of quoted value. Conversions were 30%–70%.
The relative yields were independent of conversion.
bpH of the aqueous portion (H2O–CH3CN, 1:1).
products 9 vs. 10 was found to be pH dependent (Table 1),
with increasing yield of the dimer 10 (maximum relative
yield ≈ 30%) with increasing pH. This trend is consistent
with a mechanism in which the initially formed benzyl car-
banion intermediate (19a) can react via two pathways, sim-
ple protonation to give 9 and a more complex “bimolecular”
pathway (disproportionation-type route, vide infra) to give
10. The protonation pathway would be favoured at lower
pH, and this was indeed observed. Since the dimer 10 was
observed as a photoproduct under relatively low lamp inten-
sities (Rayonet photochemical reactor), it is unlikely that it
arises via secondary photolysis of the anticipated intermedi-
ate carbanion (19a) of the photodecarboxylation (vide infra).
The above results observed for 4-acetylphenylacetic acid
(7) is very similar to what is known in the photodecar-
boxylation of 4-nitrophenylacetic acid (2) (4) (Scheme 1).
Photodecarboxylation of 2 gave a mixture of the expected 4-
nitrotoluene (11) via simple protonation of the initially
formed 4-nitrobenzyl carbanion (12) and a “dimeric” prod-
uct (4,4′-dinitrobibenzyl, 13). The relative yield of dimer 13
is enhanced at higher pH, becoming the major (>80%) prod-
uct at pH > 10. Thermally generated 12 (via the depro-
tonation of 11 using strong bases and where there is no
possibility of photoexcitation of the carbanion 12) also gave
13 (7). The mechanism for the formation of 13 from 12 ap-
pears to be complex. Radical ions are believed to be in-
volved (7) as strong EPR signals are observed. For the
photodecarboxylation pathway, Craig and Pace (8) suggested
the possibility of dimerization of the initially formed 12
based on kinetic studies for its decay, to give the radical
dianion 14 (expected to be EPR active and consistent with
the EPR signals observed) (8), which on work-up, is oxi-
dized to give 13. The very long lifetime of photogenerated
Quantum yields
Quantum yields for photodecarboxylation were measured
as a function of pH, using the photodecarboxylation of keto-
profen (4, Φd = 0.75 (6a)) as a secondary standard (Fig. 1).
H2O–CH3CN (1:1, v/v) was employed as the solvent in these
studies because of experimental difficulties in performing
semipreparative photolysis in 100% water. All the quoted pH
are of the aqueous portion prior to mixing except for pH 4–
8, where the pH was measured in the mixed solvent as ad-
justment to the pH had to be made with NaOH upon dissolu-
tion of the substrate. Since the quantum yield plot is for
qualitative purposes only, the exact pH values are of less
consequence. The quantum yields for photodecarboxylation
(Φd) for both 6 and 7 are substantial at pH 7 (0.60 and 0.22,
respectively), indicative of the highly activating nature of the
acetyl group. For both compounds, the highest quantum
yields were observed for the carboxylate form of the mole-
cule (in the region where pH > pKa; estimated pKa for these
compounds ≈ 3 to 4). Indeed, this is suggestive that only the
carboxylate form is reactive in photodecarboxylation, as is
the case for most arylacetic acids reported to date (3, 4).
This is indeed the case for the para isomer 7 since the quan-
tum yield vs. pH plot for this compound approaches zero at
© 2004 NRC Canada

Huck et al.
1763
Scheme 1.
tum yield vs. pH plot observed for 6 in Fig. 1 over the entire
pH region is due to photodecarboxylation and not some
other contributing photochemical reaction that may have
been overlooked in the product studies.
Fig. 1. Plots of quantum yields for photodecarboxylation (Φd) vs.
pH for acetylphenylacetic acids 6 (squares) and 7 (circles) in
H2O–CH3CN (1:1). The short-dashed line shows the expected
trend for 6 if acid catalysis does not operate; the long-dashed
line shows the resolved acid-catalyzed photodecarboxylation
pathway.
The above “unusual” photoreactivity displayed by the
meta acid 6 may be interpreted as another example of the
well-known “meta effect” for the photochemistry of substi-
tuted benzenes that has been well-documented and studied
by Zimmerman (9). That is, the meta-acetyl group in 6 can
activate (in the excited state, by electron withdrawal) the
corresponding attached acetic acid moiety more than a cor-
responding para-acetyl group. This results in a more reactive
compound (as manifested in the higher quantum yields ob-
served for 6 at all pH) as well as the operation of an observ-
able acid-catalyzed pathway (vide infra).
Nanosecond laser flash photolysis (LFP) and reaction
mechanisms
Although the results presented so far, based on product
studies and quantum yield measurements vs. pH, show a
close analogy in photochemical behaviour between the
nitrophenylacetic acids (1 and 2) and the acetylphenylacetic
acids (6 and 7), LFP results show mechanistic details for 6
and 7 that appear to be more complex than those reported
for 1 and 2. LFP studies of 6 and 7 were first performed in
dry CH3CN, where there is no photoreaction. Only one
strong and broad (∆OD ≈ 0.12) transient formed immedi-
ately after the laser pulse was observed for both compounds,
with λmax 330 nm (Fig. 2, shown for 7), which was effi-
ciently quenched by oxygen. This transient exhibited the
same decay kinetics over the 330–430 nm region of the
spectrum and hence assignable to one transient. For 6, the
lifetime (all single exponential fits) was 800 ns under N2 and
50 ns under O2; for 7, 1200 ns under N2 and 170 ns under
O2. This transient is assigned to the triplet state, in agree-
ment with the known triplet state absorption for
acetophenone itself (λmax 330 nm (10)), although there is a
pH < 1. For 3-acetylphenylacetic acid (6), the quantum yield
is substantial (≈0.13) in acidic solution, which is indicative
of an acid-catalyzed mechanism that operates only for the
meta isomer. As shown in Fig. 1, the short-dashed line rep-
resents the expected trend in the absence of an acid-
catalyzed pathway. The long-dashed line is the “resolved”
contributing acid-catalyzed pathway. The “plateau” region
observed in the pH 2 to Ho –2 region is therefore the sum of
the acid- and “base”-catalyzed mechanisms. It should be em-
phasized at this point that product studies of 6 in the acid re-
gion discussed here also gave only one photoproduct (viz.,
8). Additionally, photolysis of 6 at pD 1 (where the mecha-
nism is believed to be almost exclusively in the acid-
catalyzed regime) gave 8-αD exclusively. That is, the quan-
© 2004 NRC Canada

1764
Can. J. Chem. Vol. 82, 2004
Fig. 2. Transient observed on LFP of 7 in 100% CH3CN
(266 nm excitation, N2 purged). The signal at 340 nm is as-
signed to the triplet state. (130 ns (᭜), 220 ns (ٗ), 410 ns (᭹),
3000 ns after the laser pulse (×)). Inset: first-order decay at
340 nm; τ ≈ 1000 ns.
Table 2. Lifetimes of observed transients for 6 and 7 by LFP.
Observed lifetimes (ns)a
λb
6 (In H2O)c
6 (In D2O)c
7 (In H2O)c
7 (In D2O)c
310
320
340
400
4600 (600)
3400 (400)
1800 (260)
1700 (420)
4400 (630)
3200 (460)
1700 (300)
3900 (710)
390 (220)
460 (140)
910 (190)
920 (190)
aLifetimes obtained from first-order decays at the respective wavelength
under N2, except for 310, 320, and 340 nm of 7, which did not obey first-
order kinetics and the lifetimes are estimated based on force fitting to a
single exponential decay. Values in brackets obtained under O2. Estimated
error: 10% of quoted value. All transients were observed immediately af-
ter the laser pulse.
bWavelength at which decay was monitored. Corresponds to λmax for 6
at 340 nm and for 7 at 310 and 400 nm (shoulder).
cAt pH(D) = 7. Cosolvent is 50% CH3CN.
be the triplet state. Indeed, the absence of the triplet state in
aq. CH3CN would suggest that it may be the reactive state in
the photodecarboxylation, given that aromatic ketones are
known to intersystem cross very efficiently. Triplet sensiti-
zation experiments were carried out in which D2O–CD3CN
(1:1) was replaced with D2O–acetone-d6 (1:1, with λex
=
Fig. 3. Transient observed on LFP of 6 in H2O–CH3CN (1:1,
pH 7) (266 nm excitation, N2 purged). The transient is assigned
to carbanion 15 or 16. The decay is first order (130 ns (᭜), 280
ns (ٗ), 480 ns (᭹), 3000 ns after the laser pulse (×)). Inset:
first-order decay at 340 nm; τ ≈ 400 ns.
300 nm), which resulted in 30% higher yield (when acetone-
d6 was used). Since most of the exciting light is absorbed by
acetone in these experiments, the inference is that the triplet
state of 6 is reactive towards photodecarboxylation, consis-
tent with the LFP data. However, we cannot rule out reac-
tion from the singlet state, although such a pathway is
probably insignificant.
Scaiano and Monti and their respective co-workers (6b–
6d) have carried out extensive LFP studies of the
photodecarboxylation of ketoprofen (4), a reaction that is
closely related to the ones being studied in the present work.
They assigned a transient with λmax at 330 and 580 nm as
owing to carbanion 5, although its spectral characteristics
suggested that this anion might have substantial contribution
from the radical anion (non-Kekulé) resonance form 5a.
weak residual longer lived signal at ca. 310 that was not as-
signed.
Lifetimes of this species has been reported to be about 200
ns in aqueous solution with much longer lifetimes when
photogenerated in some organic solvents (6b–6d). In addi-
tion, a simple benzyl anion is known to absorb at ca. 350 nm
(11). In view of these literature precedents and the observed
primary solvent isotope effect on decay of the observed
340 nm LFP transient for 6, it seems reasonable that it be as-
signed to (singlet ground state) benzyl carbanion 15a/15b
(Scheme 2). It is assumed that if the benzyl carbanion was
initially photogenerated in its triplet state that intersystem
crossing has occurred during the laser pulse. We cannot rule
out formation of the corresponding (ground state) non-
Kekulé enol 16 at this pH since the pKa of some aromatic
ketyls are known to be in the 7 to 8 range (6c). Indeed,
LFP studies of 6 and 7 were then carried out in H2O–
CH3CN (1:1, at several different pH), a solvent where these
compounds photodecarboxylate efficiently (except for 7 at
pH < 2). The transient observed immediately upon the laser
pulse for 6 in pH 7 is shown in Fig. 3. The spectral band is
narrower and weaker than that observed in 100% CH3CN,
with λmax at 340 nm. The same decay kinetics (single expo-
nential decay) was observed across the whole spectral range
with no significant residual signals. In addition, the transient
is significantly longer lived in D2O (about twofold) and is
somewhat sensitive to oxygen but not to the same extent as
the transient observed in 100% CH3CN (Table 2). Therefore,
although this transient has absorption in about the same
spectral region as the triplet state assigned above, it cannot
© 2004 NRC Canada

Huck et al.
1765
Scheme 2.
protonation at oxygen would be diffusion controlled in this
case (at pH 7 or lower), further arguing that the transient
may actually be the non-Kekulé biradical 16. This species
would still be quenchable by oxygen and sensitive to D2O.
LFP studies at pH 3 gave essentially the same transient but
at pH 0, only very weak signals were observed, consistent
with a change in reaction mechanism as required from data
presented in Fig. 1. The acid-catalyzed photodecarboxyla-
tion pathway is given in eq. [5]. Support for this mechanism
comes from the recent work of Wirz and co-workers (12)
who showed that aromatic ketones such as benzophenone
and acetophenone are readily photoprotonated via the triplet
excited state, giving rise to reactive electronically excited
carbocations that can undergo overall hydration (which for
the parent systems studied are entirely reversible). In the
case of acetophenone – phenylacetic acid 6, such a
photoprotonation would give rise to 17a/17b, since Wirz and
co-workers (12) have also shown that most of the positive
charge of such cations reside at the meta positions (a form
of meta activation (9)). An activated species such as 17b
would be expected to decarboxylate (with concerted loss of
the carboxylic acid proton to water), to generate a species
such as 18 (which may or may not be identical to 16). In any
event, in an acid medium, we anticipate that 18 will quickly
form 8. This mechanistic scheme also gives an explanation
as to why acid catalysis for photodecarboxylation was not
observed for the para-acid 7 since the positive charge of the
protonated species would not be expected to be localized
significantly at the para position. The question of why the
quantum yields measured in acid reach a plateau below pH 1
(Fig. 2) can be explained simply by noting that 17a/17b can
deactivate back to ground state 6 without continuing on to
18. In addition, triplet excited state 6 could also suffer sim-
ple nonproductive quenching by protons, which would com-
pete with the acid-catalyzed photodecarboxylation process
itself. Thus, one would not expect the quantum yields to
continue to rise until quantum yield of unity is reached.
LFP results for the para-isomer 7 are more complex. Al-
though strong signals (∆OD up to 0.1, Fig. 4) were observed
immediately upon the laser pulse, three major transients
could be discerned at 310, 355, and 400 nm (Table 2). At
least two transients are contained within the broad absorp-
© 2004 NRC Canada

1766
Can. J. Chem. Vol. 82, 2004
Scheme 3.
equipment. When H2O was replaced with D2O, an increase
in lifetime (by about twofold) of the 400 nm transient was
observed, while the lifetimes of the other transients were
essentially unaffected. The products of the photodecar-
boxylation of 7 are formally derived from a benzyl carban-
ion 19 (to give 9) and a benzyl radical 20 (to give 10).
Therefore, we assign the 400 nm transient to (ground state)
carbanion 19a/19b (Scheme 3), although we cannot rule out
that it is in fact the corresponding enol 21. The mechanism
of photodecarboxylation of 7 (Scheme 3) proceeds in a simi-
lar fashion as proposed for 6. After extrusion of CO2 (via the
carboxylate form), a benzyl carbanion (19) is generated,
which can follow two possible routes to the product. It is ei-
ther carbon protonated immediately to give 9 or is pro-
tonated at the oxygen to give enol 21. Either pathway
eventually leads to product 9. The LFP transient at 400 nm
could be reasonably assigned to either 19 or 21. Compared
to the analogous transient 15/16 for the meta acid 6, the
present transient is longer lived since the structure would
have better resonance stablization. Improved resonance sta-
bilization also explains the observed red shift (≈60 nm) of
this transient.
Fig. 4. Transients observed on LFP of 7 in H2O–CH3CN (1:1,
pH 7, 266 nm excitation, N2 purged); 500 ns (᭜), 800 ns (ٗ),
1600 ns (᭹), 13 000 ns after the laser pulse (×)). The transient
at 400 nm (shoulder) is assigned to 19 or 21 and decays with
first-order kinetics (inset) with τ ≈ 1700 ns.
The benzyl radical 20 is likely contained within the broad
strong 300–380 nm absorption. These transients are all sen-
sitive to oxygen, as would be required for such a radical.
The parent benzyl radical itself is known to absorb at
≈315 nm (11). Although our results cannot discern among
the various mechanisms possible for formation of 10, we
propose that one possible pathway is formation of 20 via ox-
idation of 19 (either by residual oxygen or by absorption of
a photon). Another possibility is the coupling of two mole-
tion in the 300–380 nm region, where the decays do not
obey simple first-order kinetics, likely caused by the over-
lapping absorptions; only the 400 nm transient exhibited
first-order decay. This increased complexity is not surprising
since this compound gave two photoproducts (9 and 10) at
all pH. Analogous experiments to 6 (vide supra) were per-
formed to identify as many of the observed transients as pos-
sible, within the scope of the time resolution of the laser
© 2004 NRC Canada

Huck et al.
1767
cules of 19, giving a diradical dianion, which is then oxi-
dized, similar to that proposed for 2 (Scheme 1). However,
the much shorter lived 19 (1700 ns) compared to the para ni-
tro analogue 12 (60 s) makes this less likely. However, it
would appear that the longer lifetime of 19 compared to the
corresponding meta carbanion 15 does allows for the minor
“radical” pathway giving rise to the corresponding dimer 10.
4-Acetylphenylacetic acid (7)
The desired product was synthesized in a similar proce-
dure and with comparable yield to 6 previously. The com-
pound is a white powder, mp (uncorrected) 106–108 °C (lit.
1
value (13) 119 °C). H NMR (300 MHz, CDCl3) δ: 2.58 (s,
3H, CH3), 3.70 (s, 2H, ArCH2), 7.36 (d, J = 8 Hz, 2H, ArH),
7.92 (d, J = 8 Hz, 2H, ArH), 12.4 (in acetone-d6, br,
COOH). 13C NMR (250 MHz) δ: 26.6 (CH3), 40.8 (CH2),
128.7 (CH), 129.7 (CH), 136.2 (ipso), 138.6 (ipso), 176.1
(CO2H), 197.8 (R2CO). HR-MS calcd. for C10H10O3:
178.0630; found: 178.0629.
Summary
We have shown that acetophenone derivatives 6 and 7 un-
dergo efficient photochemistry that is best rationalized as be-
ing due to the enhanced electron-withdrawing character of
the acetyl group in the triplet excited state. The photo-
chemistry mirrors some of the observed behaviour for the
much-studied nitrophenylacetic acids 1 and 2. The triplet
state is reactive, although we cannot rule out residual singlet
state reaction. Our results suggest that new types of photo-
chemistry via ionic intermediates may be available for aro-
matic ketones, which will require a reevaluation of the
present theoretical understanding of ketone photochemistry,
at least in polar solvents such as water.
Product studies
Preparative photolyses for 6 and 7 were carried out with
samples (10–100 mg) dissolved in the appropriate solvent
(40–100 mL) and transferred to a 100 mL quartz tube. The
solution was irradiated in a RPR-100 photochemical reactor
fitted with 254 nm or 300 nm lamps with continuous cooling
(by a cold finger) and purging by a stream of argon for ap-
proximately 10 min before and continuously during irradia-
tion (via a long stainless steel needle). Photolysis times
ranged from 1 to 30 min, depending on the conversion de-
sired, the efficiency of the reaction, and the size of the sam-
ple. After photolysis, aqueous samples were acidified (if
necessary) and extracted with CH2Cl2. Direct evaporation of
the organic solvent was used when samples were irradiated
in wholly organic solvents. As the photoproducts are some-
what volatile, the temperature was kept under 30 °C during
evaporation and the flask was returned to atmospheric pres-
sure as soon as the solvent was removed. Typical experi-
ments are described in the following.
Experimental
General
1H NMR spectra were recorderd on a Bruker AC300
(300 MHz) instrument in CDCl3 unless otherwise specified.
MS and HR-MS were obtained on a Kratos Concept H (EI)
instrument. UV–vis spectra were recorded on a Varian Cary
5 instrument. Acetonitrile (HPLC or spectral grade) used for
LFP experiments was dried by distillation from calcium hy-
dride. Organic extracts were dried over anhydrous magne-
sium sulfate and filtered. pH(D) values were measured using
a Fisher Accumet 915 pH meter.
Photolysis of 6 (pH 7)
Compound 6 (50 mg) was dissolved in 100 mL of H2O–
CH3CN (1:1) and the pH was adjusted to 7 using aq. NaOH.
The solution was irradiated using 4 × 254 nm lamps for
15 min. After photolysis, the solution was made acidic with
concd. HCl (to pH < 2) and extracted with CH2Cl2 (3 ×
40 mL). The solvent was dried over MgSO4 and concen-
Materials
1
3-Acetylphenylacetic acid (6)
trated on a rotary evaporator. The H NMR (300 MHz) was
3-Methylacetophenone (Aldrich, 5.17 g, 38 mmol) was
added to a solution consisting of 100 mL CCl4, 1.2 equiv. N-
bromosuccinimide (8.23 g, 46 mmol), and a catalytic
amount of benzoyl peroxide, and refluxed for 20 h under N2.
The bromination product was isolated by first evaporation of
the solvent and subsequent dissolution in CH2Cl2 (with
back-wash with water). This crude material was then dis-
solved in 10% H2O–CH3CN and refluxed for 4 h with
1.1 equimolar of NaCN. The resulting crude cyano deriva-
tive was hydrolyzed directly with concd. HCl (refluxed in
50 mL for 5 h). The crude acid product was isolated by
evaporation of the solvent to give a brown solid, which was
then dissolved in a minimum of hot water, filtered, and
cooled. The crude product was recrystallized again in water
to yield a white powder (0.38 g, 12% overall yield); mp (un-
corrected) 96–99 °C (lit. value (13) 106 °C). 1H NMR
(300 MHz) δ: 2.59 (s, 3H, CH3), 3.71 (s, 2H, ArCH2), 7.43
(m, 2H, ArH), 7.86 (m, 2H, ArH), 10.8 (in acetone-d6, br,
COOH). 13C NMR (250 MHz) δ: 26.6 (CH3), 40.7 (CH2),
127.5 (ArCH), 128.9(ArCH), 129.3 (ArCH), 133.8 (ipso),
134.1(ArCH), 137.5 (ipso), 176.6 (CO2H), 197.9 (R2CO).
HR-MS calcd. for C10H10O3: 178.0630; found: 178.0631.
obtained in CDCl3 and gave a spectrum identical to that of
the commercial 3-methylacetophenone δ: 2.40 (s, 3H,
ArCH3), 2.58 (s, 3H, COCH3), 7.37 (m, 2H, ArH), 7.74 (m,
2H, ArH). Conversion to 8 was 100%. MS (m/z): 134 (M+).
Photolysis of 6 (pH ≈ 0)
Compound 6 (50 mg) was dissolved in 10% H2SO4–
CH3CN (1:1). Procedure as for pH 7 previously. The NMR
spectrum of the product was the same as pH 7, with
1
unreacted 6. Conversion to 8 was 15% by H NMR.
Photolysis of 6 (pD 7)
Compound 6 (100 mg) was dissolved in 100 mL of D2O–
CH3CN (1:1) and the pD was adjusted to 7 with a solution
1
of NaOD–D2O. The H NMR (300 MHz) spectrum of the
photoproduct (76% yield) was identical to that of commer-
cial 3-methylacetophenone except that the aromatic methyl
group was monodeuterated (product 8-αD). It was separated
from the product mixture using column chromatography
(CH2Cl2, Rf = 0.73): δ: 2.38 (“t”, 2H, J = 2 Hz, ArCH2D),
2.58 (s, 3H, COCH3), 7.34 (m, 2H, ArH), 7.74 (m, 2H,
ArH).
© 2004 NRC Canada

1768
Can. J. Chem. Vol. 82, 2004
irradiated in 7 mm quartz cells. Flow cells were used to ob-
tain spectra and kinetic data. Runs using static cells in aque-
ous solution gave irreproducible results because of efficient
depletion of substrate because of efficient photochemical
decarboxylation. Solutions were purged (with N2 or O2)
prior to and continuously during LFP studies.
Photolysis of 7
Compound 7 (100 mg) was dissolved in 80 mL of H2O–
CH3CN (1:1) and the pH was adjusted to 7.5 using aq.
NaOH. The solution was irradiated using 254 nm lamps for
30 min. After photolysis, the solution was made acidic with
concd. HCl (so the pH < 2) and extracted with CH2Cl2 (3 ×
40 mL). The solvent was dried with MgSO4 and concen-
trated. The products were separated by preparative TLC
(5% EtOAc–CH2Cl2) to give two photoproducts, 4-
methylacetophenone (9, 78%) and a dimer 10 (22%). The 1H
NMR (300 MHz) spectrum of 9 was identical to that of the
commercial 4-methylacetophenone δ: 2.25 (s, 3H, ArCH3),
2.42 (s, 3H, COCH3), 7.20 (d, J = 8 Hz, 2H, ArH), 7.80 (d,
Acknowledgments
The Natural Sciences and Engineering Research Council
of Canada (NSERC) and the University of Victoria are ac-
knowledged for continued support of this research. We thank
Matt Lukeman for help with some initial LFP experiments.
J = 8 Hz, 2H, ArH). MS (EI, m/z): 134 (M+). Dimer 10: H
1
NMR (300 MHz) δ: 2.58 (s, 6H, CH3CO), 3.00 (s, 4H,
ArCH2-), 7.20 (d, 4H, J = 8 Hz, ArH), 7.85 (d, 4H, J =
8 Hz, ArH). MS (EI, m/z): 266 (M+).
References
1. M.B. Smith and J. March. Advanced organic chemistry. 5th ed.
Wiley-Interscience, New York. 2001.
2. A. Williams. Concerted organic and bio-organic mechanisms.
CRC Press, Boca Raton, Florida. 2000.
3. D. Budac and P. Wan. J. Photochem. Photobiol. A, 67, 135
(1992).
Triplet sensitization of 6
Compound 6 (9.1 mg) was dissolved in 3.0 mL of D2O–
CD3CN (2:1). Two aliquots (0.5 mL) were removed and di-
luted, one was with 0.5 mL CD3CN and the other with
0.5 mL acetone-d6. The solutions (8 × 10–3 mol L–1) were
placed into NMR tubes and purged with argon. The solu-
tions were photolyzed for 90 s (8 × 300 nm lamps).
4. (a) J.D. Margerum and C.T. Petrusis. J. Am. Chem. Soc. 91,
2467 (1969); (b) P. Wan and S. Muralidharan. J. Am. Chem.
Soc. 110, 4336 (1988); (c) J. Morrison, H. Osthoff, and P.
Wan. Photochem. Photobiol. Sci. 1, 384 (2002).
5. M. Xu and P. Wan. J. Chem. Soc. Chem. Commun. 2147
(2000).
6. (a) L.L. Costanzo, D. DeGuidi, G. Conderelli, A. Cambria,
and M. Fama. Photochem. Photobiol. 50, 359 (1989); (b) G.
Cosa, L.J. Martinez, and J.C. Scaiano. Phys. Chem. Chem.
Phys. 1, 3533 (1999); (c) S. Monti, S. Sortino, G. DeGuidi,
and G. Marconi. J. Chem. Soc. Faraday Trans. 93, 2269
(1997); (d) M. Laferrière, C.N. Sanramé, and J.C. Scaiano.
Org. Lett. 6, 873 (2004).
1
Photodecarboxylation yields (by H NMR) were 33% and
43% for D2O–CD3CN vs. D2O–acetone-d6 solutions, respec-
tively.
Quantum yields
Photodecarboxylation quantum yields for 6 and 7 were
determined using relative methods that involved comparison
of the yields of decarboxylated product between ketoprofen
(4) with a known quantum yield (secondary actinometric
standard) of 0.75 (6) and those of the compounds in ques-
tion, as determined by 1H NMR integration. A known
amount of standard was dissolved in CH3CN–H2O (1:1, 2 ×
7. (a) G.A. Russell and E.G. Janzen. J. Am. Chem. Soc. 89, 300
(1967); (b) E. Buncel and B.C. Menon. J. Am. Chem. Soc.
102, 3499 (1980).
–1
10–3 mol L adjusted to pH 7) and photolyzed as above for
8. B.B. Craig and M.D. Pace. J. Chem. Soc. Chem. Commun.
1144 (1987).
1 min (4 lamps, 254 nm). An equimolar amount of the com-
pound of unknown quantum yield was irradiated under the
same conditions and decarboxylation determined by 1H
NMR integration of the benzyl methyl protons.
9. (a) H.E. Zimmerman. J. Phys. Chem. A, 102, 5616 (1998);
(b) H.E. Zimmerman. J. Am. Chem. Soc. 117, 8988 (1995).
10. S.L. Murov, I. Carmichael, and G.L. Hug. Handbook of
photochemistry. 2nd ed. M. Dekker, New York. 1993. p. 101.
11. F.L. Cozens, W. Ortiz, and N.P. Schepp. J. Am. Chem. Soc.
120, 13 543 (1998).
12. M. Ramseier, P. Senn, and J. Wirz. J. Phys. Chem. A, 107,
3305 (2003).
13. D. Papa, E. Schwenk, and A. Klingsberg. J. Am. Chem. Soc.
68, 2133 (1946).
Nanosecond laser flash photolysis (LFP)
LFP studies were carried out at the University of Victoria
laser flash photolysis facility. A Spectra-Physics Nd:YAG la-
ser (266 nm, ~10 ns, <30 mJ/pulse) was used for excitation
and signals were digitized with a Tektronix TDS 520 re-
corder. Samples of OD ~ 0.4 at 266 nm were prepared and
© 2004 NRC Canada