Direct Observation of Ultrafast Decarboxylation of Acyloxy Radicals via Photoinduced Electron Transfer in Carboxylate Ion Pairs

Article abstract of DOI:10.1021/jo9617833

Charge-transfer (CT) photoactivation of the electron donor-acceptor salts of methylviologen (MV²⁺) with carboxylate donors (RCO₂^-) including benzilates [Ar₂C(OH)CO₂^-] and arylacetates (ArCH₂CO₂^-) leads to transient [MV^.+, RCO₂^.] radical pairs. Femtosecond time-resolved spectroscopy reveals that the photogenerated acyloxy radicals (RCO₂^.) rapidly lose carbon dioxide by C-CO₂ bond cleavage, in competition with back-electron transfer to restore the original ion pair, [MV²⁺, RCO₂^-]. The decarboxylation rate constants for ArCH₂CO₂^. lie in the range (1-2) × 10⁹ s^-1, in agreement with previous reports. In striking contrast, the C-CO₂ bond scission in Ar₂C(OH)CO₂^. occurs within a few picoseconds (k_CC = (2-8) × 10¹¹ s^-1). The rate constants for decarboxylation of these donors approach those of barrier-free unimolecular reactions. Thus, real-time monitoring of the decarboxylation of benziloxy radicals represents the means for the direct observation of the transition state for C-C bond scission.

Full text of DOI:10.1021/jo9617833

2210
J . Org. Chem. 1997, 62, 2210-2221
Dir ect Obser va tion of Ultr a fa st Deca r boxyla tion
of Acyloxy Ra d ica ls via P h otoin d u ced Electr on Tr a n sfer
in Ca r boxyla te Ion P a ir s
T. Michael Bockman, Stephan M. Hubig, and J ay K. Kochi*
Department of Chemistry, University of Houston, Houston, Texas 77204-5641
Received September 17, 1996X
Charge-transfer (CT) photoactivation of the electron donor-acceptor salts of methylviologen (MV2+
)
)
-
with carboxylate donors (RCO2-) including benzilates [Ar2C(OH)CO2-] and arylacetates (ArCH2CO2
leads to transient [MV•+, RCO2 ] radical pairs. Femtosecond time-resolved spectroscopy reveals
•
•
that the photogenerated acyloxy radicals (RCO2 ) rapidly lose carbon dioxide by C-CO2 bond
cleavage, in competition with back-electron transfer to restore the original ion pair, [MV2+, RCO2-].
The decarboxylation rate constants for ArCH2CO2 lie in the range (1-2) × 109 s-1, in agreement
•
with previous reports. In striking contrast, the C-CO2 bond scission in Ar2C(OH)CO2• occurs within
a few picoseconds (kCC ) (2-8) × 1011 s-1). The rate constants for decarboxylation of these donors
approach those of barrier-free unimolecular reactions. Thus, real-time monitoring of the decar-
boxylation of benziloxy radicals represents the means for the direct observation of the transition
state for C-C bond scission.
In tr od u ction
interest.8-11 Previous indirect measurements have shown
that this decarboxylation process takes place on time
scales of less than 1 ns.8-11 In order to monitor directly
the rapid fragmentation in eq 1, a means of generating
Recent advances in laser spectroscopy on the pico-
second and femtosecond time scales1,2 have made it
possible to monitor in real time the course of a single
reaction step, from reactants to products through the
transition state.3 Electron transfer,4 bond scission,5
proton transfer,6 and structural isomerization7 are among
the elementary reactions that have been successfully
analyzed by the use of the new time-resolved techniques.
A process of fundamental importance that can be ob-
served by this experimental approach is carbon-carbon
bond cleavage, and the loss of carbon dioxide from acyloxy
radicals, i.e.,
•
the reactive radical, RCO2 , on very short time scaless
picoseconds to femtosecondssis thus required. This goal
is accomplished by one-electron oxidation of the corre-
sponding carboxylate anion (RCO2-), which is instanta-
neously12 effected by the charge-transfer photoactivation
(hνCT) of its preformed complex with an electron acceptor
(A).13-16 Excitation of such electron donor-acceptor
(EDA) complexes can thus be employed to convert a
carboxylate anion to the corresponding acyloxy radical,
i.e.,
kCC
•
RCO2
8 R• + CO2
(1)
(8) (a) Braun, W.; Rajbenbach, L.; Eirich, F. R. J . Phys. Chem. 1962,
66, 1591. (b) Kaptein, R.; Brokken-Zijp, J .; de Kanter, F. J . J . J . Am.
Chem. Soc. 1972, 94, 6280. See also: (c) Sheldon, R. A.; Kochi, J . K. J .
Am. Chem. Soc. 1970, 92, 5175.
is an example that has recently drawn considerable
(9) Falvey, D. E.; Schuster, G. B. J . Am. Chem. Soc. 1986, 108, 7419.
(10) Hilborn, J . W.; Pincock, J . A. J . Am. Chem. Soc. 1991, 113, 2683.
(11) (a) Chateauneuf, J .; Lusztyk, J .; Ingold, K. U. J . Am. Chem.
Soc. 1987, 109, 897. (b) Chateauneuf, J .; Lusztyk, J .; Ingold, K. U. J .
Am. Chem. Soc. 1988, 110, 2886. (c) Tateno, T.; Sakuragi, H.;
Tokumaru, K. Chem. Lett. 1992, 1883. (d) Rauk, A.; Yu, D.; Armstrong,
D. A. J . Am. Chem. Soc. 1994, 116, 8222. (e) Davidson, R. S.; Steiner,
P. R. J . Chem. Soc., Perkin Trans. 2 1971, 1357. (f) Prasad, D. R.;
Hoffman, M. Z.; Mulazzani, Q. G.; Rodgers, M. A. J . J . Am. Chem.
Soc. 1986, 108, 5135. For a comprehensive review of photodecarboxy-
lations, see: (g) Budac, D.; Wan, P. J . Photochem. Photobiol. A 1992,
67, 135.
X Abstract published in Advance ACS Abstracts, March 15, 1997.
(1) Manz, J .; Wo¨ste, L. Femtosecond Chemistry; VCH: Weinheim,
1995.
(2) Simon, J . D. Rev. Sci. Instrum. 1989, 60, 3597.
(3) (a) Zewail, A. H. J . Phys. Chem. 1993, 97, 12427. (b) Kaiser, W.
Ultrashort Laser Pulses and Applications; Springer: New York, 1988.
(c) Martin, J .-L.; Migus, A.; Mourou, G. A.; Zewail, A. H. Ultrafast
Phenomena VIII; Springer: New York, 1993.
(4) (a) Kliner, D. A. V.; Alfano, J . C.; Barbara, P. F. J . Chem. Phys.
1993, 98, 5375. (b) Schoenlein, R. W.; Peteanu, L. A.; Wang, Q.;
Mathies, R. A.; Shank, C. V. J . Phys. Chem. 1993, 97, 12087. (c) J oly,
A. G.; Nelson, K. A. Chem. Phys. 1991, 152, 69. (d) Zhang, H.; J onkman,
A. M.; van der Meulen, P.; Glasbeek, M. Chem. Phys. Lett. 1994, 224,
551. (e) Reid, P. J .; Barbara, P. F. J . Phys. Chem. 1995, 99, 3554. (f)
Walker, G. C.; Barbara, P. F.; Doorn, S. K.; Dong, Y.; Hupp, J . T. J .
Phys. Chem. 1991, 95, 5712. (g) Wynne, K.; Galli, C.; Hochstrasser, R.
M. J . Chem. Phys. 1994, 100, 4797. (h) Sakata, Y.; Tsue, H.; Oneil, M.
P.; Wiederrecht, G. P.; Wasielewski, M. R. J . Am. Chem. Soc. 1994,
116, 6904. (i) Laermer, F.; Elsaesser, T.; Kaiser, W. Chem. Phys. Lett.
1988, 148, 119
(5) (a) Pederson, S.; Baumert, T.; Zewail, A. H. J . Chem. Phys. 1993,
97, 12460. (b) Dantus, M.; Rosker, M. J .; Zewail, A. H. J . Chem. Phys.
1988, 89, 6128. (c) Banin, U.; Ruhman, S. J . Chem. Phys. 1993, 98,
4391. (d) King, J . C.; Zhang, J . Z.; Schwartz, B. J .; Harris, C. B. J .
Chem. Phys. 1993, 99, 7595. (e) Dougherty, T. P.; Heilweil, E. J . Chem.
Phys. Lett. 1994, 227, 19. (f) Lenderink, E.; Duppen, K.; Wiersma, D.
A. Chem. Phys. Lett. 1993, 211, 503
(12) Time-resolved spectroscopy4g,15 has established that CT excita-
tion of the EDA complex effects the transfer of an electron from the
donor to the acceptor in less than 500 fs.
(13) Mulliken theory describes the linear relationship between the
charge-transfer transition energies (hνCT) of these new absorption
bands and the electron-richness of the donors, as measured by their
ionization potentials (IP) in the gas phase and their thermodynamic
oxidation potentials (E°ox) in solution. See: (a) Mulliken, R. S. J . Am.
Chem. Soc. 1952, 74, 811. (b) Mulliken, R. S.; Person, W. B. Molecular
Complexes: A Lecture and Reprint Volume; Wiley: New York, 1969.
(14) (a) Hilinski, E. F.; Masnovi, J . M.; Amatore, C.; Kochi, J . K.;
Rentzepis, P. M. J . Am. Chem. Soc. 1983, 105, 6167. (b) Hilinski, E.
F.; Masnovi, J . M.; Kochi, J . K.; Rentzepis, P. M. J . Am. Chem. Soc.
1984, 106, 8071. (c) Hubig, S. M.; Bockman, T. M.; Kochi, J . K. J . Am.
Chem. Soc. 1996, 118, 3842.
(15) (a) Ojima, S.; Miyasaka, H.; Mataga, N. J . Phys. Chem. 1990,
94, 4147. (b) Ojima, S.; Miyasaka, H.; Mataga, N. J . Phys. Chem. 1990,
94, 5834. (c) Ojima, S.; Miyasaka, H.; Mataga, N. J . Phys. Chem. 1990,
94, 7534.
(6) Schwartz, B. J .; Peteanu, L. A.; Harris, C. B. J . Phys. Chem.
1992, 96, 3591.
(7) Sension, R. J .; Repinec, S. T.; Szarka, A. Z.; Hochstrasser, R. M.
J . Chem. Phys. 1993, 98, 6291.
(16) Ebbesen, T. W.; Manring, L. E.; Peters, K. S. J . Am. Chem. Soc.
1984, 106, 7400.
S0022-3263(96)01783-5 CCC: $14.00 © 1997 American Chemical Society

Ultrafast Decarboxylation of Acyloxy Radicals
J . Org. Chem., Vol. 62, No. 7, 1997 2211
hνCT
Ch a r t 1
•
[RCO2-, A]
8 [RCO2 , A-•]
(2)
This novel method for the photogeneration of the acyloxy
radical (eq 2) contrasts with other photochemical meth-
ods, such as the cleavage of peroxides,9 esters,10 etc., since
it yields the reactive radicals in their electronic ground
states.17
Earlier reports noted the formation of EDA com-
plexes18-20 between carboxylate donors and the dicationic
acceptor, 1,1′-dimethyl-4,4′-bipyridinium, commonly re-
ferred to as methylviologen (MV2+).21-23 [Such EDA
complexes between negatively-charged donors and posi-
tively-charged acceptors constitute charge-transfer (CT)
ion pairs.] The properties of the CT ion pair formed
between MV2+ and the benzilate anion (Ph2C(OH)CO2
)
-
were particularly intriguing, since its photolysis led to
the efficient (φ ) 0.8-1.2) formation of decarboxylation
products.18 Moreover, the characteristic absorption spec-
trum of reduced methylviologen (A•- ) MV•+ in eq 2)22
provides a means of monitoring the reaction by time-
resolved pump-probe absorption spectroscopy. To access
the critical femtosecond/picosecond time scales relevant
to electron transfer and bond cleavage, we designed and
constructed a time-resolved femtosecond spectrometer
based on a Ti:sapphire laser with a short (230 fs) pulse
and a broad tuning range (360-450 nm, frequency-
doubled) as the excitation source. We chose a laser
system with a high-energy pulse, which allowed us to
record a continuous visible spectrum (360-900 nm) in a
single laser shot. This approach not only provided kinetic
traces at all relevant wavelengths, but more importantly,
it enabled us to recognize subtle spectral changes in the
intermediates during the decarboxylation process.23
I. Ch a r ge-Tr a n sfer Ba n d s of Meth ylviologen /
Ca r b oxyla t e Ion P a ir s. When the colorless methyl-
viologen ditriflate was added to an aqueous solution of
sodium benzilate, a light yellow color developed im-
mediately. 1H-NMR analysis of the colored solution (in
D2O) showed no changes in the proton resonances of
MV2+ or of the benzilate anion [compared with those of
pure MV(OTf)2 or Ph2C(OH)CO2Na]. The intensity of the
yellow color faded upon extensive dilution or upon the
addition of inert salt. By analogy with previous observa-
tions,25 the color change was attributed to the formation
of an ion pair26,27 between the positively charged meth-
ylviologen and the negatively charged benzilate, i.e.
KIP
-
MV2+ + Ar2C(OH)CO2- y z [MV2+, Ar2C(OH)CO2
]
(3)
UV-vis spectroscopy revealed that the yellow color was
caused by a new electronic absorption band in the
spectral region between 340 and 420 nm, where neither
MV(OTf)2 nor Ph2C(OH)CO2Na absorbed. Similar color
changes were observed upon addition of methylviologen
to other substituted benzilate anions in aqueous solution.
Figure 1 shows the visible absorption bands of the ion
pairs in eq 3, with their progressive red shift as the anion
is changed from benzilate (Ar ) phenyl) to 4,4′-dimethox-
ybenzilate (Ar ) p-anisyl) and to 2,2′,5,5′-tetramethoxy-
benzilate (Ar ) 2,5-dimethoxyphenyl). Such a red shift
as the aromatic nucleus becomes progressively more
electron-rich accords with the Mulliken theory13 of charge-
transfer interactions. This charge-transfer theory further
implies that irradiation of the new absorption bands in
Figure 1 effects the transfer of a single electron from the
benzilate donor to the viologen acceptor, according to the
Mulliken formulation in eq 4.
Resu lts
To prepare the methylviologen benzilate ion pairs, a
series of benzilic acids, Ar2C(OH)CO2H, with electron-
donating substituents on the aromatic rings were syn-
thesized. The carboxylate anions were generated in situ
by dissolution of the free carboxylic acids in aqueous
sodium bicarbonate. Methylviologen was prepared and
used as the bis(trifluoromethanesulfonate) or ditriflate
salt, MV(OTf)2.24 The benzilate substrates are shown in
Chart 1.
(17) Bond homolysis by the direct photoexcitation of these precursors
involves excited states, the temporal relaxation of which leads to the
acyloxy radical on the same time scale as the cleavage process itself.
The overall kinetics of the photodecarboxylation is thus a convolution
of the formation as well as the decomposition of the acyloxy radical,
and rate constants (kCC) on the picosecond time scale cannot be readily
extracted. For a discussion of this problem in the context of peroxide
decomposition, see ref 9.
(18) Barnett, J . R.; Hopkins, A. S.; Ledwith, A. J . Chem. Soc., Perkin
Trans. 2 1973, 80.
(19) (a) Deronzier, A.; Esposito, F. Nouv. J . Chem. 1983, 7, 15. (b)
J ones, G., II; Zisk, M. B. J . Org. Chem. 1986, 51, 947. (c) Mandler, D.;
Willner, I. J . Chem. Soc., Perkin Trans. 2 1988, 997. (d) Ford, W. E.;
Rodgers, M. A. J . J . Phys. Chem. 1991, 95, 5827.
hνCT
-
•
[MV2+, Ar2C(OH)CO2
]
8 [MV+•, Ar2C(OH)CO2 ]
(4)
Therefore, the methylviologen/benzilate ion pairs were
identified as charge-transfer (CT) ion pairs.27
(20) Usui, Y.; Sasaki, Y.; Ishii, Y. Tokumaru, K. Bull. Chem. Soc.
J pn. 1988, 61, 3335.
(21) (a) Michaelis, L. Biochem. Z. 1932, 250, 564. (b) Michaelis, L.;
Hill, E. S. J . Gen. Physiol. 1933, 16, 859.
(25) (a) Bockman, T. M.; Kochi, J . K. J . Am. Chem. Soc. 1989, 111,
4669. (b) Kochi, J . K.; Bockman, T. M. Adv. Organomet. Chem. 1991,
33, 51.
(26) Kosower, E. M. J . Am. Chem. Soc. 1956, 80, 3253.
(27) The possibility of further aggregation (to form triple ions, ion
quadrupoles, etc.) is unlikely in view of the high polarity of the aqueous
medium and the relatively low concentrations (<0.1 M) of methylvi-
ologen and benzilate salts needed to observe the CT absorption bands.
[Triple-ion formation constants for MV2+ and halide anions are on the
order of 1.0 M-1. See ref 24b.]
(22) (a) Watanabe, T.; Honda, K. J . Phys. Chem. 1982, 86, 2617. (b)
Bockman, T. M.; Kochi, J . K. J . Org. Chem. 1990, 55, 4127.
(23) See Bockman, T. M.; Hubig, S. M.; Kochi, J . K. J . Am.
Chem.Soc. 1996, 118, 4502 for a preliminary report of these results.
(24) In contrast to the more commonly-utilized chloride or bromide
salts,24b methylviologen triflate showed no interfering CT bands
between anion and cation. (b) Ebbesen, T. W.; Ferraudi G. J . Phys.
Chem. 1983, 87, 3717.

2212 J . Org. Chem., Vol. 62, No. 7, 1997
Bockman et al.
F igu r e 1. Charge-transfer absorption spectra of methyl-
viologen (0.05 M) in aqueous solution ion-paired with (A)
2,2′,5,5′-tetramethoxybenzilate, (B) 4,4′-dimethoxybenzilate,
and (C) benzilate. The dashed lines are the spectral cutoffs of
methylviologen ditriflate (‚‚‚) and sodium tetramethoxy-
benzilate (-‚-).
F igu r e 3. Deconvolution of the 600-nm band in Figure 2 as
the sum of the absorbance of methylviologen (dotted spectrum)
and benzophenone ketyl radical (dashed spectrum). The inset
shows the authentic spectrum of the ketyl radical observed
55 ps following charge-transfer irradiation (355 nm) of the
[N-methyl-4-cyanopyridinium, benzilate] ion pair.
centered at λ ≈ 600 nm. The first band was readily
identified as the near-UV absorption band of the reduced
form of methylviologen (MV•+), the absorption spectrum
of which has two maxima (390 and 605 nm)22 in aqueous
solution. Subtraction of the visible absorbance of MV•+
from the composite 600 nm band (see Figure 3) yielded
an additional absorption band (λmax ) 530 nm), which
was identified as the visible band of the ketyl radical
Ph2C(OH)• (vide infra). Since the formation of the ketyl
radical and MV•+ were both observed on the early
picosecond time scale, we conclude that the transfer of
an electron from the benzilate anion to methylviologen
was accompanied by decarboxylation, i.e.,
hν
[MV2+, Ph2C(OH)CO2
]
CT8
-
[MV+•, Ph2C(OH)•, CO2] (5)
F igu r e 2. Transient absorption spectrum recorded 1.0 ps
following the 230-fs laser excitation (375 nm) of MV2+/benzilate
CT ion pairs in aqueous solution. The inset shows the rise
(P -behavior) of the absorption of MV•+, monitored at 605 nm
(open circles), and the rise of benzophenone ketyl radical,
monitored at 530 nm (closed circles).
The derived extinction coefficient of ꢀ530 ) 6200 M-1 cm-1
for the ketyl radical in Figure 3 was in reasonable
agreement with the reported value of 5500 M-1 cm-1 28
,
and, thus, confirmed the equimolar formation of MV•+
and Ph2C(OH)• in eq 5. The formation of MV•+ and Ph2C-
(OH)• could be separately monitored by following the
absorbance changes at 605 and 530 nm, respectively. The
inset to Figure 2 shows that the rise of the MV•+
absorbance occurred on the same time scale (700 fs) as
the instrumental response of the laser system, but the
growth of the ketyl radical absorbance was significantly
retarded. The absorbance of both species reached a
maximum after 3 ps and remained unchanged through-
out the observation time window of 50 ps.29
Charge-transfer ion pairs were also formed upon
addition of MV(OTf)2 to aqueous solutions of sodium
arylacetates (ArCH2CO2Na). An increasing red shift in
the CT absorption band was observed as the para
substituent was changed in the order Cl < H < Ph <
OMe. The CT absorption bands of the [MV2+, benzilate]
ion pairs could not be distinguished from those of the
arylacetates with the same ring substituents. Thus, the
CT bands of methylviologen benzilate and phenylacetate
were both observed as broad tailing bands, extending
from 340 to 420 nm, while the bands of the ion pairs
formed from 4,4′-dimethoxybenzilate and 4-methoxy-
phenylacetate both had distinct maxima at λmax ) 370
nm.
II. F em tosecon d Tr a n sien t Sp ectr oscop y a n d
Kin etics of [MV2+, Ar 2C(OH)CO2-] Ion P a ir s. Ben -
zila te a n d 4,4′-Dim eth ylben zila te. Immediately upon
375-nm laser photolysis of the CT ion pair formed from
methylviologen and benzilate ions, the transient spec-
trum shown in Figure 2 was observed. The spectrum
Nearly identical transient spectra and kinetics were
obtained upon CT-photolysis of the ion pair formed from
MV2+ and 4,4′-dimethylbenzilate. In this case, the
absorption band of the ketyl radical (λmax ) 550 nm) and
that of MV•+ (λmax ) 605 nm) grew in with the same
risetime as the response time of the laser spectrometer
(28) Hayon, E.; Ibata, T.; Lichtin, N. N.; Simic, M. J . Phys. Chem.
1972, 76, 2072.
(29) (a) Since the experimental spectrum in Figure 2 was unchanged
over a period of more than 50 ps, a correction for group velocity
dispersion29b was unnecessary. (b) Sharma, D. K.; Yip, R. W.; Williams,
D. F.; Sugamori, S. E.; Bradley, L. L. T. Chem. Phys. Lett. 1976, 41,
460
exhibited two bands: a strong, sharp band with λmax
390 nm and a broad absorbance in the visible region,
)

Ultrafast Decarboxylation of Acyloxy Radicals
J . Org. Chem., Vol. 62, No. 7, 1997 2213
(700 fs). Both transient bands persisted unchanged in
intensity and appearance for times exceeding 50 ps.
The spectrum of the ketyl radical derived from benzi-
late could be more clearly distinguished in the absence
of strongly-absorbing MV•+. Thus, 355-nm excitation30
of the charge-transfer band of benzilate paired with the
(monocationic) acceptor N-methyl-4-cyanopyridinium
(NCP+) generated a transient band (λmax ) 530 nm; see
inset to Figure 3) identical to the (deconvoluted) band of
Ph2C(OH)• generated from the corresponding [MV2+
,
benzilate] ion pair in Figure 3. The ketyl radical was
generated by CT excitation, in the same manner as
described for the MV2+ ion pairs, i.e.,
hν
[NCP+, Ph2C(OH)CO2
]
CT8
-
[NCP•, Ph2C(OH)•, CO2] (6)
F igu r e 4. Transient absorption spectra recorded following the
230-fs laser excitation (398 nm) of MV2+/2,2′,5,5′-tetrameth-
oxybenzilate CT ion pairs in aqueous solution. [The narrow
UV absorption band of MV•+ at 395 nm is obscured by
scattered light from the laser excitation pulse.] The inset shows
the rise and decay (D-behavior) of the absorbance of MV•+
monitored at 605 nm.
In this experiment, however, only the absorption band
of the ketyl radical was observed, since the NCP• radical
does not absorb visible light.31 To confirm the assign-
ment of the 530- and 550-nm transients to the ketyl
radicals, diphenylhydroxymethyl and bis(4-methyl-
phenyl)hydroxymethyl were generated independently by
photolysis of benzophenone and 4,4′-dimethylbenzo-
phenone,32,33 respectively, in aqueous isopropyl alcohol.
(See the Experimental Section.) The absorption spectra
of these transients were identical to the ketyl spectra
generated from the MV2+ and NCP+ ion pairs.
In summary, the spectroscopic and kinetic behavior of
MV2+ ion pairs with both benzilate and 4,4′-dimethyl-
benzilate followed a pattern characterized by (1) simul-
taneous or near-simultaneous formation of MV•+ and
Ar2C(OH)• on a time scale comparable to the risetime of
the laser pulse followed by (2) persistence of both species
for long times (t > 50 ps). This behavior is labeled P to
denote the persistence of the transients.
2,2′,5,5′-Tetr a m eth oxyben zila te a n d 9-Hyd r oxy-9-
flu or en eca r boxyla te. Quite different behavior was
observed upon excitation of the CT ion pair formed from
MV2+ and 2,2′,5,5′-tetramethoxybenzilate. Photostimu-
lation at λ ) 398 nm led to the formation of MV•+, but
the ketyl radical was not observed. (See Figure 4.) The
undistorted transient spectrum of MV•+ decayed to the
spectral base line within 5 ps after laser excitation. This
decay was fitted to a first-order rate law, and the lifetime
of MV•+ was calculated as τMV ) 1.9 ps. Analogous
results were obtained upon photolysis of the charge-
transfer band of MV2+ with 9-hydroxy-9-fluorenecarboxy-
late. The undistorted spectrum of MV•+ decayed to the
base line by simple first-order kinetics (τMV ) 20 ps), and
the fluorenone ketyl radical was not present.34 This
pattern of behavior, characterized by the formation of
MV•+ unaccompanied by ketyl radical, and the subse-
F igu r e 5. Transient absorption spectra recorded following the
230-fs laser excitation (398 nm) of MV2+/4,4′-dimethoxy-
benzilate CT ion pairs in aqueous solution. (The narrow UV
absorption band of MV•+ at 395 nm is obscured by scattered
light from the laser excitation pulse.) The inset shows the rise
and partial decay (DP -behavior) of the absorbance of MV•+
monitored at 605 nm.
quent decay of its absorbance, is labeled D to emphasize
the complete disappearance of the spectral transient.
4,4′-Dim eth oxyben zila te a n d 4-Meth oxyben zila te.
A third type of behavior characterized these CT ion pairs.
At early times (t < 5 ps), MV•+ was the only transient
observed, and its spectrum decayed to a residual absor-
bance within 5 ps. During this decay process, the
spectral band of MV•+ centered at 600 nm underwent a
distinct broadening, as shown in Figure 5. No further
evolution of the transient absorption was observed, and
the broad band persisted unchanged for times greater
than 50 ps after the laser pulse. The broadening of the
band at longer time intervals was attributed to the
absorption of the ketyl radicals, Ar2C(OH)• (Ar ) anisyl
(30) The CT band of [NCP+, benzilate] was blue-shifted relative to
the corresponding band of the [MV2+, benzilate] ion pair and could
not be excited with the femtosecond laser (λexc > 370 nm). The
experiment thus was carried out with the frequency-tripled output of
a 25-ps mode-locked Nd:YAG laser.
(31) (a) The NCP. radical31b has an absorption maximum at λmax
)
390 nm. A band at λmax ) 650 nm, due to intermolecular association,
would be too weak to be observed with our apparatus. (b) Itoh, M.;
Nagakura, S. Bull. Chem. Soc. J pn. 1966, 39, 369.
(32) (a) J ohnston, L. J .; Lougnot, D. L.; Wintgens, V. Scaiano, J . C.
J . Am. Chem. Soc. 1988, 110, 518.
(33) Nagarajan, V.; Fessenden, R. W. Chem. Phys. Lett. 1984, 112,
207.
(34) (a) The ketyl radical of fluorenone has λmax ) 520 nm and ꢀmax
) 1100 M-1 cm-1 in aqueous solution.28 See also: (b) Davidson, R. S.;
Santhanam, M. J . Chem. Soc., Perkin Trans. 2 1972, 2355 for its
spectrum in various nonaqueous solvents. (c) For the spectrum of the
triethylsilyl-substituted analogue, see: Chatgilialoglu, C.; Ingold, K.
U.; Scaiano, J . C. J . Am. Chem. Soc. 1982, 104, 5119.

2214 J . Org. Chem., Vol. 62, No. 7, 1997
Bockman et al.
Ta ble 1. Tr a n sien t Deca ys a n d Qu a n tu m Yield s of MV+•
u p on La ser Ir r a d ia tion of [MV2+, Ben zila te] Ion P a ir sa
a Q-switched Nd:YAG laser,36 we determined the con-
centration of MV•+ at the end of the 10-ns laser pulse
and measured the light absorbed (Iabs) by transient
actinometry.37 Thus, the CT bands of the various ion
pairs in Table 1 were excited with the third harmonic
(355 nm) of the laser, and the quantum yields of MV•+
(ΦMV) were determined by comparison of its absorbance
at 605 nm (AMV) with that of the benzophenone triplet
(ABP) monitored at 530 nm.37 The quantum yields of
MV•+ in Table 1 decreased smoothly from 0.79 to 0.05 as
the donor anion was changed from benzilate to dimethyl-,
monomethoxy-, dimethoxy-, and tetramethoxybenzilate.
In the cases of DP behavior, for which we were able to
apply both methods for quantum-yield determination, the
values of ΦMV were in good agreement. (See Table 1.)
For this reason, we concluded that no significant loss or
generation of MV•+ occurred between the end of the
femtosecond/ picosecond time window (50 ps) and the
beginning of the time window of the nanosecond experi-
ment (10 ns).
benzilate donorb
τMVc (ps)
ΦMV
d
benzilate
4,4′-dimethylbenzilate
4-methoxybenzilate
4,4′-dimethoxybenzilate
2,2′,5,5′-tetramethoxybenzilate
9-hydroxy-9-fluorenecarboxylate
no decay
no decay
2.7
1.8
1.9
0.79
0.46
0.28 (0.28)e
0.24 (0.25)e
0.05
20
0.05
a
b
In aqueous solution at 25 °C. Generated as the sodium salt
by dissolution of the free acids in sodium bicarbonate solution.
c Lifetime of reduced methylviologen (MV+•), measured by moni-
toring the decay of its absorbance at 605 nm following femtosecond
d
laser irradiation. Quantum yield of residual methylviologen at
10 ns following (nanosecond) laser excitation. e Quantum yield of
MV+• determined 50 ps after excitation in parentheses.
and/or phenyl), the absorption bands of which overlapped
with that of MV•+. This behavior is labeled DP to
emphasize the (partial) decay and persistence of the
transient species following laser photolysis.
The absorbance of MV•+ on the nanosecond time scale
did not remain constant but increased over the interval
from 10 to 200 ns. This increase was observed for all
the benzilate ions in Table 1, and it was temporally
distinct from the initial formation of MV•+ on the fem-
tosecond/picosecond time scale. This secondary phase of
MV•+ production was complete after 200 ns, and the
absorbance remained unchanged for time periods exceed-
ing the temporal range (2 ms) of the laser spectrometer.
Since the transient quantum yields in Table 1 (ΦMV) were
obtained by extrapolating the absorbance traces to zero
time, they apply to the primary (femtosecond/picosecond)
phase of MV•+ production.
In contrast to the diverse kinetic behaviors of the
benzilates, the ion pairs formed from MV2+ and aryl-
acetates were all characterized by decay of MV•+ to the
spectral base line, i.e., D-type behavior. A clear trend
was apparent, however, in the lifetime of MV•+, which
decreased uniformly as the substituents on the aryl group
became more electron-releasing, in the order Cl > H >
Ph > OMe. The low quantum yields for persistent MV•+,
as determined by transient actinometry on the nano-
second time scale (vide supra), showed a parallel trend,
with ΦMV decreasing along the same sequence as τMV in
Table 2.
IV. Stea d y-Sta te P h otolysis of MV2+/Ben zila te
Ion P a ir s. In general, the degassed aqueous solutions
of MV(OTf)2 and the sodium salts of the various benzilate
anions, Ar2C(OH)CO2-, turned deep blue to indicate the
formation of reduced methylviologen, MV•+.22 The blue
reaction mixture was extracted with dichloromethane,
and GC-MS analysis of the organic extract revealed the
formation of the diaryl ketone, Ar2CO, in accord with
previous photodecarboxylation studies.18 The two pho-
toproducts, MV•+ and diaryl ketone, were the only organic
products detected in the reaction mixture. In order to
clarify the stoichiometry of the reaction these two prod-
ucts were quantified in terms of their photochemical
quantum yields.
III. Tr a n sien t Qu a n tu m Yield s F ollow in g Na n o-
secon d P h ot olysis of [MV2+, Ar 2C(OH )CO2-] Ion
P a ir s. The presence of persistent phototransientssMV•+
and ketyl radicalsswas established by the observation
of residual absorptions on the femtosecond/picosecond
time scale, as described in section II. To quantify and
compare the concentrations of persistent MV•+ obtained
with the various benzilates in Table 1, we converted the
residual absorbances at 605 nm to quantum yields, ΦMV
,
of persistent MV•+ by normalizing them with respect to
the light absorbed, i.e., ΦMV ) [MV•+]res/Iabs ) Ares/ꢀMVIabs
,
where [MV•+]res was the concentration of residual reduced
methylviologen, Ares the residual absorbance (at 605 nm),
ꢀMV the extinction coefficient of reduced methylviologen,
and Iabs the number of moles of photons absorbed per
volume unit.
Two methods were used to determine these quantum
yields by laser photolysis experiments. In the first
method the residual concentration of MV•+ at 50 ps, A50
ps/ꢀMV, was compared with the concentration of MV•+
extrapolated to time zero, A0/ꢀMV
. This extrapolated
concentration of MV•+ was set equal to Iabs by assuming
that the radical pair formation in eq 4 occurred with unit
efficiency.35 This gives ΦMV ) (A50ps/ꢀMV)/(A0/ꢀMV) ) A50ps
/
A0. This method, however, has problems when applied
to the ion pairs with D or P behavior. Thus, in the case
of the kinetic traces with D behavior, the measurement
of the weak residual absorbance was not sufficiently
accurate due to the experimental error ((0.005 absor-
bance units) in the quantification of the absorbance
signals. In the case of P behavior, a strong absorbance
signal was obtained that did not decay. Extrapolation
of this residual absorbance to time zero would result in
a unit quantum yield, ΦMV ) 1.0. However, such a
procedure would overlook any ultrafast decay processes
occurring in the time domain of the laser pulse (vide
infra). In the intermediate (DP ) cases, the method could
be successfully applied, since the absorption band of MV•+
decayed to a residual absorbance that was accurately
measurable. (See column 3 in Table 1.)
Owing to the restricted applicability of the method
described above, a second approach was chosen to deter-
mine ΦMV for all the systems (P , D, and DP ). Utilizing
a slower but more sensitive laser spectrometer based on
(36) Bockman, T. M.; Karpinski, Z. J .; Sankararaman, S.; Kochi, J .
K. J . Am. Chem. Soc. 1992, 114, 1970.
(37) (a) Sandros, K. Acta Chem. Scand. 1969, 23, 2815. (b) Chatto-
padhyay, S. K.; Kumar, C. V.; Das, P. K. J . Photochem. 1985, 30, 81.
(c) Hurley, J . K.; Sinai, N.; Linshitz, H. Photochem. Photobiol. 1983,
38, 1.
(38) The ion pair formed from MV2+ and tetramethoxybenzilate was
an exception, and the orange solution remained unchanged in color
even after prolonged irradiation.
(35) In accord with the Mulliken formulation of the charge-transfer
transition in ref 13.

Ultrafast Decarboxylation of Acyloxy Radicals
J . Org. Chem., Vol. 62, No. 7, 1997 2215
Ta ble 2. Tr a n sien t Deca ys a n d Qu a n tu m Yield s of MV+•
u p on La ser Ir r a d ia tion of [MV2+, Ar yla ceta te] Ion P a ir sa
-•
MV•+ + O2 f MV2+ + O2
(7)
d
arylacetate donor
σ+ b
τMVc (ps)
ΦMV
the buildup of MV•+ was prevented. The second approach
relied on the catalyzed reaction of MV•+ with H+ in the
aqueous medium.43,44 Platinum dioxide (Adams catalyst)
was added to the reaction mixture prior to photolysis, so
that photogenerated MV•+ was consumed by the redox
reaction
p-chlorophenylacetate
phenylacetate
4-biphenylacetate
p-methoxyphenylacetate
1-naphthylacetate
diphenylacetate
+0.11
0.0
-0.21
-0.78
-0.3
59
39
15
4.8
8.3
5.5
0.096
0.070
0.023
0.007
<0.005
0.035
a
b
In aqueous solution at 25 °C. Hammett coefficient for the
aromatic substituent from ref 59b. c Lifetime of reduced methyl-
viologen (MV+•) measured by monitoring the decay of its absor-
2MV•+ + 2H+ [PtO2]8 2MV2+ + H2
(8)
d
bance at 605 nm following femtosecond laser irradiation. Quan-
Indeed, the presence of either O2 or PtO2 in the reaction
mixture completely eliminated the accumulation of MV•+,
and the solutions did not develop the characteristic blue
coloration of the reduced viologen throughout the irradia-
tion period. Irradiation times were limited to 60-80 min,
and conversions were kept low in order to minimize
tum yield of residual methylviologen 10 ns after laser irradiation.
Ta ble 3. Stea d y-Sta te P h otoch em ica l Qu a n tu m Yield s
for th e Rea ction of MV2+ w ith su bstitu ted Ben zila tesa
b
c
d
benzilate anion
benzilate
4,4′-dimethylbenzilate
4-methoxybenzilate
4,4′-dimethoxybenzilate
2,2′,5,5′-tetramethoxybenzilate
ΦSS
Φaer
Φcat
reactions of MV2+ or Ar2C(OH)CO2 with O2 or H2
formed in eqs 7 and 8. The photochemical yields of Ar2-
CO were determined by means of gas chromatography,
with 4-chlorobenzophenone as the internal standard. The
quantum efficiencies for the formation of ketones in the
aerated and catalyzed reactions are listed in Table 3 as
Φaer and Φcat, respectively.
Three observations should be emphasized with regard
to the quantum yields in Table 3: (1) The photochemical
quantum yields of diaryl ketone are the same in both the
catalyzed (Φcat) and the aerated (Φaer) reactions. (2) The
formation of diaryl ketone is approximately one half as
efficient as the steady-state formation of MV•+ (Φss). (3)
The photochemical quantum yields for ketone formation
(Φaer or Φcat) are identical to the transient yields of MV•+
(ΦMV). (Compare Tables 1 and 3.)
-
•-
1.14
0.96
0.47
e
0.78
0.55
0.23
0.26
<0.03
0.71
0.52
0.29
0.20
e
<0.01
a
In aqueous solution at 25 °C using the 366-nm light of a
b
mercury lamp as irradiation source. Photochemical quantum
yield of MV+• in deaerated solution, as determined by UV-vis
spectroscopy. c Photochemical quantum yield of diaryl ketone in
d
aerated solution, as determined by GC analysis. Photochemical
quantum yield of diaryl ketone in the presence of platinum dioxide
(see text) as determined by GC analysis. e Not measured.
The light source for the steady-state experiments was
the 366-nm emission of a medium-pressure mercury
lamp, and the light flux was determined by actinometry
using Aberchrome-540 according to the procedure of
Heller and Langan.39 The formation of MV•+ was quanti-
fied by UV-vis spectrophotometry. The steady-state
quantum yields (ΦSS) for the formation of MV•+ are listed
in Table 3. A comparison of the transient yields, ΦMV in
Table 1, with the steady-state yields, ΦSS in Table 3,
shows that ΦSS was consistently greater than ΦMV in all
cases by a factor of about 2.
Attempts to measure the photochemical quantum
yields for ketone formation were not successful in de-
gassed solutions. Photochemical conversion of the ben-
zilate anion to the diaryl ketone remained low even after
prolonged periods (several hours) of irradiation. We
concluded that the photogenerated MV•+ absorbed the
actinic light, and this “inner filter” effect40,41 prevented
efficient photolysis. To overcome this problem, two
methods were devised to remove the interfering MV•+ by
chemical means. In the first method, the photolysate was
continuously purged with air to provide a constant
concentration of dissolved dioxygen. Since MV•+ reacts
quantitatively with O2,42 i.e.,
Discu ssion
Charge-transfer irradiation of [MV2+, Ar2C(OH)CO2
]
-
ion pairs results in efficient oxidative decarboxylation of
the benzilate anion, and the scission of the C-CO2 bond
leads to diaryl ketones as the ultimate products. To
clarify the course of this reaction, and, particularly, to
focus on the intermediate cleavage step, we first deter-
mine the overall stoichiometry of this transformation.
The femtosecond/picosecond reaction dynamics subse-
quent to charge-transfer excitation is then analyzed in
terms of the observed P , D, and DP kinetic behaviors.
Finally, the relationship between the rate and efficiency
of the cleavage processes and the structures of the
carboxylate donors will be explored.
I. Stoich iom etr y of th e P h otor ed ox Rea ction s of
[MV2+, Ben zila te] Ion P a ir s. Charge-transfer photo-
lysis of the ion pairs formed from methylviologen dication
(MV2+) and benzilate anion (Ph2C(OH)CO2-) yields the
reduced form of the viologen acceptor (MV•+) and
benzophenone as products. Formally, the decarboxyla-
tion of benzilate to produce benzophenone is a two-
electron oxidation,
(39) Heller, H. G.; Langan, J . R. J . Chem. Soc., Perkin Trans. 2 1981,
341. See also: Kuhn, H. J .; Braslavsky, S. E.; Schmidt, R. Pure Appl.
Chem. 1989, 61, 187.
(40) Eaton, D. R., in CRC Handbook of Photochemistry, Vol 1;
Scaiano, J . C., ed; CRC Press: Boca Raton, FL, 1989, p. 1233.
(41) The intense absorption of reduced methylviologen22 allows
quantities of MV•+ as small as 0.1 µmol to be assayed after short (1
min) irradiation times. Under these conditions, the formation of the
byproduct diaryl ketone (0.05 µmol) would be too small to be reliably
quantified. As a result, the ratio of the quantum yields for formation
of reduced methylviologen and of diaryl ketone could not be determined
under the same conditions of photolysis. [Continued irradiation was
unproductive, since the photogenerated MV•+ absorbed more than 90%
of the incident actinic light and the reaction effectively ceased.]
(42) (a) Rodgers, M. A. J . Radiat. Phys. Chem. 1984, 23, 245. (b)
Tsukuhara, K. Wilkins, R. G. J . Am. Chem. Soc. 1985, 107, 2632.
Ph2C(OH)CO2 -2e8 Ph2CO + CO2 + H+
(9)
-
whereas the reduction of the methylviologen dication is
a one-electron-transfer step, i.e.
(43) Kalyanasundaram, K.; Kiwi, J .; Gra¨tzel, M. Helv. Chim. Acta
1978, 61, 2720.
(44) Harriman, A.; West, M. A. Photogeneration of Hydrogen;
Academic: New York, 1982.

2216 J . Org. Chem., Vol. 62, No. 7, 1997
Bockman et al.
+e
Sch em e 1
MV2+ 8 MV•+
(10)
A comparison of the quantum yields for the formation of
MV•+ (Φss ≈ 1.2) and of benzophenone (Φaer ≈ Φcat ≈ 0.8)
in Table 3 indicates that MV•+ and benzophenone are
formed in an approximate 2:1 ratio. Similar ratios
prevail in the reactions of other (substituted) benzilate
anions, Ar2C(OH)CO2-, with MV2•+ The experimental
ratios of the quantum yields in Table 3 point to an overall
2:1 stoichiometry, i.e.,
hν
2MV2+ + Ar2C(OH)CO2
CT8
-
2MV•+ + Ar2CO + CO2 + H+ (11)
(which are complete in 10 ps) does not allow diffusive
separation to compete.46
to accord with the formal stoichiometry that results from
the combination of eqs 9 and 10. The time-resolved
measurements show that the 2 equiv of reduced meth-
ylviologen are formed in distinct steps on the femtosecond
and on the nanosecond time scales, respectively. The
first of these processes is the photoinduced electron
The various kinetic behaviors of the [MV•+, benziloxy]
radical pairs, i.e., P , D, and DP , can be described in terms
of the competing processes in Scheme 1. P -Beh a vior :
For the parent [MV2+, Ph2C(OH)CO2-] ion pair, the ketyl
radical intermediate is observed within 1.0 ps of the laser
pulse (Figure 2). Decarboxylation (kCC) thus occurs very
rapidly, and it successfully competes with back-electron
transfer (kbet). As a result, high quantum yields (both
steady-state and transient) for the formation of reduced
methylviologen and benzophenone are obtained. (See
Tables 1 and 3.) D-Beh a vior : By contrast, the radical
pair derived from 2,2′,5,5′-tetramethoxybenzilate and
MV2+ undergoes back-electron transfer (kbet) predomi-
nantly, as indicated by the absence of the ketyl radical
as a spectral transient in Figure 4 and the smooth decay
of the absorbance of MV•+ to the spectral base line. The
yields of photoproducts in this case are negligible. DP -
Beh a vior : In the radical pairs derived from MV2+ and
monomethoxy- and dimethoxybenzilate, the competing
processes of bond scission (kCC) and back-electron transfer
(kbet) occur with comparable probabilities. Accordingly,
the femtosecond spectral transients decay partially to a
non-zero base line, and intermediate values of the
quantum yields are obtained. In summary, the various
femtosecond/picosecond kinetic behaviors (P , D, and DP )
are explained in qualitative terms by the simple competi-
tion between decarboxylation (kCC) and back-electron
transfer (kbet) as depicted in Scheme 1. However, both
of these processes occur in the same time domain as the
photoactivation of the CT ion pair. Thus, a complete
quantitative simulation of the radical pair dynamics
requires the explicit inclusion of the temporal profile of
the laser excitation process.
transfer in eq 4 to generate the [MV•+, Ar2C(OH)CO2 ]
•
radical pair. The second process is the oxidation of the
ketyl radical by MV2+, i.e.,
MV2+ + Ar2C(OH)• f MV•+ + Ar2CO + H+ (12)
as originally suggested by Barnett et al.18 (Ketyl radicals
are oxidized by MV2+ with rate constants on the order of
108 M-1 s-1 45
. ) The crucial step linking the photoinduced
electron transfer in eq 4 to the formation of the diaryl
ketone in eq 12 is the decarboxylation of the acyloxy
•
radical, Ar2C(OH)CO2 . Consequently, we concentrate on
the steps immediately following the electron transfer in
eq 4, particularly the C-C bond cleavage implied by the
formation of diaryl ketone (eq 11).
II. F em tosecon d /P icosecon d Dyn a m ics of [MV•+,
Acyloxy] Ra d ica l P a ir s. The spectroscopic observation
of benzophenone ketyl radical (Figure 2) within a pico-
second of laser excitation of the ion pair in eq 5 clearly
demonstrates the rapid cleavage of the C-CO2 bond in
the benziloxy radical. However, the picosecond decays
in Figures 4 and 5, combined with the quantum yields
of less than unity in Tables 1 and 3, point to the existence
of a competing back-electron transfer process to restore
the original CT-ion pair. Back-electron transfer is ubiq-
uitous in photoinduced redox processes, and it typically
dominates the reactivity of the radical pairs generated
by charge-transfer excitation.14-16 The various dynamic
behaviors of the MV•+/acyloxy radical pairs illustrated
in Figures 2-5 can be simply analyzed in terms of this
We first consider the rise and decay of reduced meth-
ylviologen for the simplest case, considering only the
back-electron-transfer pathway (i.e., kbet . kCC in Scheme
1). The mathematical description of the kinetics in this
case is given by the convolution of the experimental decay
(kbet) with the laser pulse (τ), as described in the
Experimental Section.47,48 Figure 6 shows the simulated
kinetic traces for the absorbance of MV•+ (AMV) as a
competition between carbon-carbon bond cleavage (kCC
)
and back-electron transfer (kbet) as depicted in Scheme
1.
The scission of the R-CO2 bond yields a triad of MV•+,
CO2, and the radical, R•, while the competing back-
electron transfer restores the original CT ion pair.
Diffusive separation of the radical pair is not considered
explicitly in Scheme 1 since the rapidity of the chemical
transformations (kCC and kbet) of the benziloxy radicals
function of kbet
. As the value of kbet increases, the
temporal behavior of the absorbance changes from an
(46) The escape of the benziloxy radicals out of the solvent cage is
not included in the femtosecond kinetics, since it occurs on much slower
time scales. See: Scott, T. W.; Liu, S. N. J . Phys. Chem. 1989, 93, 1393.
(45) (a) Small, R. D.; Scaiano, J . C. J . Photochem. 1977, 6, 453. (b)
Ketyl radicals are electron rich by virtue of their very low oxidation
potentials and ease of oxidation in solution. See: (c) Baumann, H.;
Merckel, C.; Timpe, H.-J .; Graness, A.; Kleinschmidt, J .; Gould, I. R.;
Turro, N. J . Chem. Phys. Lett. 1984, 103, 497. (d) Kemp, T. J .; Martins,
L. J . A. J . Chem. Soc., Perkin Trans. 2 1980, 1708. (e) Engel, P. S.;
Wu, W. X. J . Am. Chem. Soc. 1989, 111, 1830.
(b) Cage escape may compete with the slower reactions of the [MV•+
arylacetoxy] radical pairs.
,
(47) Compare: Peters, K. S.; Li, B. J . Phys. Chem. 1994, 98, 401.
(48) Asaki, M. T.; Huang, C.-P.; Garvey, D.; Zhou, J .; Kapteyn, H.
C.; Murnane, M. M. Opt. Lett. 1993, 18, 977.

Ultrafast Decarboxylation of Acyloxy Radicals
J . Org. Chem., Vol. 62, No. 7, 1997 2217
7A with curve c in Figure 7B.) In summary, the full
range of kinetic behaviors (P , D, and P D) can be
simulated using the simple kinetics of Scheme 1, with
only two adjustable parameters, k and R. The simulation
parameters, k and R, can now be related to experimen-
tally observed quantities, i.e., the picosecond decays (τMV
)
and the transient quantum yields (ΦMV). The partition
ratio, R, as defined in eq 14, reflects the extent to which
reduced methylviologen escapes back-electron transfer to
appear in the triad. Thus, R is equal to ΦMV, the
transient quantum yield of reduced methylviologen.
Since all the kinetic behaviors can be simulated by
varying R and k (see above), and since R is experimen-
tally determined as ΦMV, one can now fit the simulated
absorbance profiles to the experimental data by adjusting
one single parameter, k. Figure 8 shows fits of the
absorbance of MV•+ to the simulated curves. The k and
R values may now be used to calculate the values of the
rate constants kbet and kCC for ion pairs that show D or
DP behavior according to eqs 13 and 14 (see Table 4).
F igu r e 6. Simulation of D-behavior for various values of kbet
(in reciprocal laser pulsewidths, τ-1): (a) 0.1, (b) 0.2, (c) 0.5,
and (d) 1.0.
exponential decay (curve a in Figure 6) to a symmetrical
rise and fall that closely follows the temporal profile of
the laser pulse (curve d). In addition to the change in
shape, we call attention to the diminution of the maxi-
mum value of AMV as kbet increases. Despite the varia-
tions in the size and shape of the temporal responses,
all of the simulated traces reach the spectral base line.
These simulated kinetic behaviors thus correspond to the
decay-dominated kinetics (D-type) of the radical pairs in
Table 1.
If the cleavage reaction in Scheme 1 is to be included
in the kinetics (i.e., if kCC ≈ kbet), it is convenient to define
an overall decay rate constant, k, and a partition ratio,
R, in terms of the rate constants for decarboxylation and
back electron transfer.
For the benzilate ion pair, which shows P -behavior,
kCC can be obtained from the growth kinetics of the
absorbance of the Ph2C(OH)• radical, since the rise of its
absorbance depends on kCC in Scheme 1, while the rise
of MV•+ depends only on the pulse width. We observed
a significant time lag of 2-3 ps between the formation
of MV•+ (monitored at 605 nm) and the slower formation
of Ph2C(OH)• (monitored at 530 nm). (See the inset to
Figure 2.) The best fit for the rise of the 530-nm
absorbance was obtained for kCC ) 8 × 1011 s-1 in Figure
8C. (See the Experimental Section for the details of this
analysis.) Consequently, kbet must have a value of 2 ×
1011 s-1 to accommodate the observed quantum yield, ΦMV
) R ) 0.79 in eq 14. For the ion pair derived from 4,4′-
dimethylbenzilate, the rate constants, kCC and kbet, could
not be determined in this way, since there was no
measurable time lag between the rise of the ketyl radical
and the rise of MV•+. A lower limit for k can, however,
be estimated from the transient quantum yield ΦMV. To
successfully simulate the observed P behavior of this ion
pair with an observed R value (ΦMV) of 0.46, it is
necessary to choose k in eq 13 to be greater than 1012
s-1. (Note that this represents a case in which there is
no apparent decay even though R is considerably less
than unity.) On the basis of these values of R and k we
k ) kbet + kCC
(13)
(14)
R ) kCC/(kbet + kCC
)
These parameters, k and R, are varied to generate the
family of simulated traces in Figure 7. In Figure 7A the
time-dependent absorptions of MV•+ are shown for a fixed
value of k (1012 s-1) and for varying partition ratios, R.
For R ) 0.05 (curve a), the simulated absorbance decays
nearly to the base line and thus approximates D-
behavior. On the other hand, a R value of 0.8 results in
a rise with no apparent decay component, to simulate
P -behavior (curve d). (Note that R need not be unity to
yield simulations of a simple rise to a nondecaying
plateau.) The intermediate values of R lead to a rise and
decay to a raised base line, in accordance with DP
behavior (curves b and c).
In the case of partial decay (DP behavior), the shape
of the temporal profile of MV•+ absorbance is sensitive
not only to R, but also to the overall rate constant, k.
Figure 7B shows the changes in the simulated absor-
bance profiles at a fixed value of R ) 0.5 and variable k.
As k approaches the time constant of the laser pulse (τ-1),
the decay component in the traces becomes less pro-
nounced. Ultimately, this leads to a kinetic profile
consisting of a simple rise to a persisting plateau (P
behavior) at k ) 2τ-1. In other words, P -behavior can
be explained either by a fast decarboxylation rate com-
pared with back-electron transfer (kCC . kbet) or by
comparable rates (kCC ≈ kbet) combined with overall fast
kinetics (kCC + kbet ≈ τ-1). (Compare curve d in Figure
estimate that kCC ≈ kbet g 5 × 1011 s-1
.
Scheme 1 can also be applied to the radical pairs
formed by photolysis of arylacetate CT-ion pairs. Since
these decay rates are all substantially slower than the
laser pulse, the kinetics of the excitation need not be
taken into account. The observed first-order decay
constant is thus equal to k, and the values of ΦMV in Table
2 can be equated to R. The rate constants for back-
electron transfer and C-C bond cleavage in Table 5 were
thus extracted from the picosecond decays and transient
quantum yields of reduced methylviologen according to
eqs 13 and 14.
III. Tr en d s in th e Ra tes of Deca r boxyla tion a n d
Ba ck -Electr on Tr a n sfer . The most remarkable obser-
vation in this study is the extremely rapid fragmentation
of benziloxy radicals. The rate constants for carbon-
carbon bond cleavage in Table 4, which are on the order
of 1012 s-1, approach those of barrier-free unimolecular
reactions, for which rate constants of 1013 s-1 are calcu-

2218 J . Org. Chem., Vol. 62, No. 7, 1997
Bockman et al.
F igu r e 7. Simulation of DP -behavior for various values of k and R. A: k ) 1.0 reciprocal laser pulsewidths and R ) (a) 0.05, (b)
0.2, (c) 0.5, and (d) 0.8. B: R ) 0.5 and k ) (a) 0.2, (b) 1.0, and (c) 2.0 reciprocal laser pulsewidths.
F igu r e 8. Experimental data for the normalized absorbance of MV•+ (large open circles) superimposed on simulated kinetic
traces. A: MV2+/4,4′-dimethoxybenzilate, simulated as DP -behavior, with k ) 1.0 × 1012 s-1 and R ) 0.24. B: MV2+/2,2′,5,5′-
tetramethoxybenzilate, simulated as D-behavior, with k ) 8.3 × 1011 s-1 and R ) 0.05. C: MV2+/benzilate, simulated as P -behavior,
with kCC ) 8 × 1011 s-1; the dotted line shows the simulated response of the laser system. Error bars represent experimental
errors of 10% of the maximum absorbance.
Ta ble 4. Ra te Con sta n ts for Deca r boxyla tion a n d Ba ck
Electr on tr a n sfer in Meth ylviologen /Ben zila te Ion P a ir sa
Ta ble 5. Ra te Con sta n ts for Deca r boxyla tion a n d
Ba ck -Electr on Tr a n sfer in Meth ylviologen /Ar yla ceta te
Ion P a ir sa
b
c
kbet
kCC
(1011 s-1
)
(1011 s-1
)
arylacetate
kbet (1011 s-1
)
kCCc (109 s-1
)
b
benzilate donor
benzilate
2
8
p-chlorophenylacetate
phenylacetate
4-biphenylylacetate
p-methoxyphenylacetate
1-naphthylacetate
diphenylacetate
0.15
0.26
0.64
2.1
1.2
1.7
1.6
1.8
1.5
1.5
<0.2
6.1
4,4′-dimethylbenzilate
5d
5d
4-methoxybenzilate
3
1
4,4′-dimethoxybenzilate
2,2′,5,5′-tetramethoxybenzilate
9-hydroxy-9-fluorenecarboxylate
8.0
8.3
0.48
2.0
0.4
0.02
a
b
a
b
Calculated using eqs 13 and 14. Rate constant for back-
electron transfer within the radical pair in Scheme 1. c Rate
constant for decarboxylation of the acyloxy radical derived from
Calculated using eqs 13 and 14. Rate constant for back-
electron transfer from reduced methylviologen to the acyloxy
radicals derived from the arylacetate anions in column 1. c Rate
constant for decarboxylation of the acyloxy radical.
d
the benzilate anion in column 1. Lower limit; see text.
lated.49 These rapid rates imply that the benziloxy
) 1-2 × 109 s-1) that pertains to the arylacetoxy radicals
in Table 5. The latter are within the range of previously
reported values8-11 of kCC determined for various aliphatic
acyloxy radicals. The ultrafast rates of decarboxylation
of the benziloxy radicals are dependent on the substitu-
ents in the aromatic ring, with electron-donating sub-
stituents inducing a diminution in kCC. The ultrafast
rates may be rationalized on the basis of the reaction
radicals are separated from the transition states for C-C
bond cleavage by activation barriers of only 1-2 kcal
mol-1 50
The ultrafast decarboxylation of Ar2C(OH)CO2
.
•
should be contrasted with the relatively slower rate (kCC
(49) Moore, J . W.; Pearson, R. G. Kinetics and Mechanism, 3rd ed.;
Wiley: New York, 1981, p 173.
(50) Calculated by means of the Eyring equation in ref 49.

Ultrafast Decarboxylation of Acyloxy Radicals
J . Org. Chem., Vol. 62, No. 7, 1997 2219
energetics, while the substituent effects point to the
influence of electronic factors.
the aromatic π-system (as illustrated in structure I) also
explains the anomalously low rates of decomposition of
the acyloxy radicals derived from hydroxyfluorenecar-
boxylate and naphthylacetate. (See entries 6 in Table 1
and 5 in Table 2.) In both of these cases, the cation
radical structure analogous to I (for which decarboxyla-
tion is not favorable) is strongly preferred, since the
positive charge is delocalized over an extended aromatic
system.
The C-C bonds of the acyloxy radicals derived from
substituted phenylacetates are cleaved with a uniform
rate constant of kCC ) 1.6 ( 0.3 × 109 s-1, regardless of
the substituents on the phenyl ring. (See entries 1-4 in
Table 5.) The nonvariant decarboxylation rates of these
ArCH2CO2• radicals are in accord with indirect measure-
ments of the decarboxylation of a series of substituted
Rea ction En er getics. Although decarboxylation of
aliphatic acyloxy radicals (R ) alkyl in eq 1) is exer-
gonic,51 theoretical and experimental studies agree that
the loss of carbon dioxide requires a significant activation
energy.52 Ab initio calculations indicate that decarboxy-
lation corresponds to an intended crossing of the π ground
state with low-lying σ states of the acyloxy radical.52
Accordingly, the increasing stability of R• in eq 1 lowers
the barrier to C-C bond scission by moving the crossing
point closer to the reactant surface. Thus, the increased
stability of ketyl radicals, Ar2C(OH)•, relative to benzyl
•
radicals, ArCH2 , implies a lowering of the transition-
state free energy (∆Gq) for the decarboxylation of the
corresponding benziloxy radicals.53 The transition state
for this decarboxylation is lower in energy (relative to
those of arylacetoxy radicals) owing to more extensive
delocalization (over two aromatic rings rather than one)
of the radical center that is forming on the benzilic carbon
during the decarboxylation.
acyloxy radicals, which have cleavage rate constants, kCC
,
between 1 and 11 × 109 s-1 57
.
These relatively slow rates
imply higher activation barriers for decarboxylation,
which in turn points to a strengthening of the C-CO2
bonds of the arylacetoxy radicals relative to their benzil-
oxy counterparts. Thus, we conclude that the remote
substituent effect is only pronounced for the cleavage of
Su bstitu en t Effects. As shown in Table 4, decar-
boxylation rates decrease systematically as electron-
donating groups are added to the aromatic system. Thus,
the unsubstituted benziloxy radical decarboxylates at
ultrafast rates, while its dimethoxy-substituted analogue
fragments four times more slowly. Additional methoxy
substitution (to form tetramethoxybenzilate) results in
a further 5-fold decrease in kCC. To explain this effect,
let us consider the critical intermediate, the benziloxy
•
the very weak C-C bonds of Ar2C(OH)CO2 and not for
•
the relatively stronger ones of ArCH2CO2 .
Ba ck -Electr on Tr a n sfer . The rate constants for
back-electron transfer (kbet) in Tables 4 and 5 increase
monotonically as the aromatic nuclei in the carboxylate
donors are substituted with electron-releasing groups, in
the order phenyl < methyl < methoxy. For the benzi-
lates, this trend extends over a factor of 4 in kbet
proceeding from benzilate to tetramethoxybenzilate. (See
Table 4.) An even clearer tendency is apparent for the
arylacetate donors, as reflected in an increase of more
than 1 order of magnitude as the para substituent is
changed from chloro to hydrogen, phenyl, and methoxy.
(See Table 5.) The trend is correlated with the increasing
donor strength of the substituent, as assessed by the
decreasing values of the Hammett parameter, σ+,58,59 and
also follows the increasing red-shift of the charge-transfer
absorption bands. This spectral shift relates directly to
the decreasing oxidation potentials (E°ox) of the carboxy-
late donors13 and thus to the decreasing driving force,
-∆Gbet ) F (E°ox + E°red) for the back-electron transfer
in Scheme 1. (E°red represents the reduction potential
of the methylviologen acceptor and F the Faraday
constant of 23.1 kcal mol-1 V-1.) Thus, the decreased
•
radical Ar2C(OH)CO2 , which can be thought of as either
a zwitterionic species, consisting of an arene cation
radical with a remote carboxylate substituent (I), or as
a simple acyloxy radical with the unpaired electron
centered on the carboxyl group (II). The relative contri-
butions of structures I and II are controlled by the
distribution of electron density over the benziloxy radical,
which in turn depends on the substituents on the
aromatic ring.54 Electron-releasing groups (G) will sta-
bilize I more than II. Since structure I, with an anionic
carboxyl group, is less prone to loss of CO2 than structure
II, which contains an acyloxy group, the presence of
electron-donating substituents will disfavor decarboxy-
lation.55 This “trapping” of the electron deficiency within
exergonicity for the electron-transfer process is ac-
60
companied by an increase in kbet
.
Such a free energy
correlation is termed “inverted” to contrast with the
“normal” behavior of increasing reaction rate with in-
creasing thermodynamic driving force. The inverted
(51) (a) Edge, D. J .; Kochi, J . K. J . Am. Chem. Soc. 1973, 95, 2635.
(b) Benassi, R.; Taddei, F. Tetrahedron 1994, 50, 4795.
(56) (a) Schlag, E. W.; Schneider, S.; Fischer, S. Ann. Rev. Phys.
Chem. 1971, 22, 465. (b) Kompa, K. L.; Levine, R. D. Acc. Chem. Res.
1994, 27, 91.
(52) Pacansky, J .; Brown, D. W. J . Phys. Chem. 1983, 87, 1553.
(53) In this context, the enhanced rate for the loss of CO2 from the
diphenylacetoxy radical (kCC ) 6.1 × 109 s-1) relative to that from
(57) The minor discrepancy between the values of kCC for decar-
boxylation of PhCH2CO2 obtained by Hilborn and Pincock (5 × 109
•
phenylacetoxy (1.6 × 109 s-1) results in
a
similar (though less
s-1) in methanol10 and by us (1.6 × 109 s-1) in water may be due to a
solvent effect, since decarboxylations are slower in polar than in
nonpolar solvents.11c Alternatively, the rate of the clock reaction used
by these authors (decarboxylation of the 9-methylfluorenecarboxyl
radical) may be somewhat overestimated. (See footnote 20 in ref 9.)
(58) The linear free-energy relation yields F+ ) -1.73 for the back-
electron transfer.
pronounced) effect (see Table 5).
(54) The radical is conceived of as a resonance hybrid of structures
I and II. The possibility that I and II are distinct chemical species
and that there is a chemical transformation prior to decarboxylation
seems unlikely at this juncture (since I and II differ only in their
electronic configuration) but cannot be ruled out a priori. Note that
the distinction between nuclear and electronic motion becomes difficult
to draw as the time scale of fast (fs) reactions approaches the time
domain for electronic redistribution.56
(59) (a) Brown, H. C.; Okamoto, Y.; Inukai, T. J . Am. Chem. Soc.
1958, 80, 4964. (b) Taylor, R. Electrophilic Aromatic Substitution;
Wiley: New York, 1990; p 455f.
(55) The effect of electron-withdrawing substituents such as chloro
could not be examined, since the charge-transfer absorption band of
(60) (a) Marcus, R. A. Annu. Rev. Phys. Chem. 1964, 15, 155. (b)
Marcus, R. A. J . Chem. Phys. 1965, 43, 679. (c) Marcus, R. A. Faraday
Discuss. Chem. Soc. 1982, 74, 7. (d) Marcus, R. A. J . Chem. Phys. 1956,
24, 966.
the [MV2+, 4,4′-dichlorobenzilate] ion pair was blue-shifted to λCT
<
360 nm and thus could not be excited with the Ti:sapphire laser.

2220 J . Org. Chem., Vol. 62, No. 7, 1997
Bockman et al.
trend in kbet is in accord with previous studies of back-
electron transfer within EDA complexes composed of a
variety of neutral and charged donors and acceptors.14,61,62
In particular, inverted behavior applies to back-electron
transfer within complexes of methylviologen with both
neutral and anionic donors.62 The driving force depen-
dency of kbet in MV2+/carboxylate ion pairs is thus
qualitatively the same as that observed with other
methylviologen EDA complexes. In addition the range
of kbet in Tables 4 and 5 is the same (109-1011 s-1) as has
been observed for other radical pairs containing MV•+.62
Thus, the distinction between the dissociative and non-
dissociative processes is not reflected in the rates of back-
electron transfer.63
reactions the transition state will be directly generated
and observed.
Exp er im en ta l Section
Ma ter ia ls. Phenylacetic acid, p-chlorophenylacetic acid,
p-methoxyphenylacetic acid, and 4-biphenylylacetic acid were
obtained from Aldrich and used as received. 1-Naphthylacetic
acid, 9-hydroxy-9-fluorenecarboxylic acid, and 9-fluorenecar-
boxylic acid (Aldrich) were recrystallized from water and dried
at 60 °C in vacuo. Diphenylacetic acid and benzilic acid
(Aldrich) were used as received.
Syn th esis of Ben zilic Acid s. The substituted benzilic
acids used in this study were prepared by rearrangement of
the corresponding benzils using potassium hydroxide in n-
butyl alcohol, according to the procedure of Ford-Moore,65 and
In summary, the diverse efficiencies for decarboxyla-
tion within radical pairs derived from benzilate donors
results from competing reactions, the decarboxylation of
were recrystallized from
a mixture of ethyl acetate and
hexanes. 4,4′-Dim eth ylben zilic Acid . From 4,4′-dimethyl-
benzil66 in 58% yield: mp 128-129 °C (lit.67 mp 127-129 °C);
1H-NMR (acetone-d6) δ 7.36 (d, J ) 8.1 Hz, 4H), 7.07 (d, J )
8.1 Hz, 4H), 5.3 (br, 2H), 2.30 (s, 6H); 13C-NMR (acetone-d6) δ
175.5, 141.2, 137.8, 129.1, 128.4, 81.1, 21.1. 4-Met h oxy-
ben zilic Acid . From 4-methoxybenzil66 in 35% yield: mp
143-144 °C (lit.66 mp 147 °C); 1H-NMR (acetone-d6) δ 7.50 (d,
J ) 7.5 Hz, 2H), 7.40 (d, J ) 8.7 Hz, 2H), 7.32 (d, J ) 7.5 Hz,
2H), 7.31 (m, 1H), 6.89 (d, J ) 8.7 Hz, 2H), 3.78 (s, 3H); 13C-
NMR (acetone-d6) δ 175.4, 160.0, 144.2, 136.0, 129.4, 128.5,
128.2, 113.9, 81.1, 55.4. 4,4′-Dim eth oxyben zilic a cid . From
4,4′-dimethoxybenzil (Aldrich) in 60% yield: mp 159 °C (lit.68
mp 155 °C); 1H-NMR68 (acetone-d6) δ 7.39 (d, J ) 8.7 Hz, 4H),
•
the acyloxy radicals, Ar2C(OH)CO2 , and back-electron
transfer. The former process is retarded and the latter
enhanced by electron-rich substituents. In the case of
the arylacetoxy radicals, on the other hand, the variation
in the decarboxylation efficiency is governed solely by the
rate of electron transfer, since kCC is invariant.
Con clu sion s
CT ion pairs formed from methylviologen (MV2+) and
substituted arylacetate or benzilate anions transfer an
electron from the carboxylate donor to MV2+ upon charge-
transfer irradiation. The acyloxy radical in the photo-
68
6.88 (d, J ) 9.0 Hz, 4H), 3.81 (s, 6H); 13C-NMR (acetone-d6)
δ 159.5, 133.5, 132.4, 128.6, 113.5, 80.5, 55.3.
2,2′,5,5′-Tetr a m eth oxyben zil. A solution of sodium cya-
nide (2.0 g) in 20 mL of water was added to a solution of 2,5-
dimethoxybenzaldehyde (20 g, 0.12 mol) in 35 mL of absolute
ethanol. The reaction mixture was stirred and refluxed for
0.5 h and then cooled to room temperature. An orange oily
phase separated. The oil was washed with cold (0 °C) ethanol
and dried in vacuo. This crude material (2,2′,5,5′-tetrameth-
oxybenzoin) was dissolved in 70 mL of 80 vol % aqueous acetic
acid. Ammonium nitrate (10 g, 0.125 mol) and cupric acetate
(0.2 g) were added, and the reaction mixture was refluxed for
1.5 h. Upon cooling to room temperature, a yellow-orange solid
precipitated from the solution. The crude tetramethoxybenzil
was recrystallized from ethanol to yield 12 g (62%) of 2,2′,5,5′-
tetramethoxybenzil as yellow rhombic crystals: mp 148-149
°C; 1H-NMR (acetone-d6) δ 7.48 (d, J ) 3.0 Hz, 2H), 7.22 (dd,
J ) 8.7, 2.7 Hz, 2H), 7.10 (d, J ) 8.7 Hz, 2H), 3.84 (s, 6H),
3.52 (s, 6H); 13C-NMR (acetone-d6) δ 192.4, 155.7, 155.0, 124.7,
123.1, 115.7, 112.8, 56.8, 56.0. Anal.69 Calcd for C18H18O6:
C, 65.45; H, 5.49. Found: C, 65.29; H, 5.51. 2,2′,5,5′-
Tetr a m eth oxyben zilic Acid . Prepared from 2,2′,5,5′-tet-
ramethoxybenzil by the procedure of Ford-Moore65 in 45% yield
generated radical pair, [MV•+, RCO2 ], rapidly loses
•
carbon dioxide by cleavage of the C-CO2 bond. Back-
electron transfer to restore the original ion pair competes
with this process. Time-resolved spectroscopy on the
femtosecond and nanosecond time scales allows the two
reaction pathways to be directly observed and quantified.
Decarboxylation rate constants for arylacetoxy radicals
lie in the range of 109 s-1 and, thus, are in agreement
with values reported previously.10 In contrast, the rate
constant for cleavage of the benziloxy radical is on the
order of 1012 s-1. The high stability of the product radical,
Ph2C(OH)•, effects a decrease in the activation energy,
which in turn leads to ultrafast rates of carbon-carbon
bond cleavage. The cleavage rates are modulated by the
electronic effects of substituents on the aromatic rings,
with electron-withdrawing groups enhancing the rate of
C-CO2 scission. By manipulation of these structural
factors, it should be possible to design carboxylate donors
that lose an electron and cleave the C-C bond in a single
step. In such concerted electron-transfer/bond-cleavage
1
as the hemihydrate: mp 105-108 °C dec; H-NMR (acetone-
d6) δ 7.02 (d, J ) 3.0 Hz, 2H), 6.89 (dd, J ) 8.7, 3.0 Hz, 2H),
6.82 (d, J ) 3.0 Hz, 2H), 5.72 (br s, 3H), 3.73 (s, 6H), 3.66 (s,
6H); 13C-NMR (acetone-d6) δ 173.8, 154.5, 152.0, 131.4, 116.0,
114.2, 114.0, 79.6, 56.8, 55.8. Anal.69 Calcd for C18H20O7‚
0.5H2O: C, 60.50; H, 5.88. Found: C, 60.27; H, 5.86.
(61) (a) Asahi, T.; Mataga, N.; Takahashi, Y.; Miyashi, T. Chem.
Phys. Lett. 1990, 171, 309. (b) Asahi, T.; Mataga, N. J . Phys. Chem.
1989, 93, 6575. (c) Asahi, T.; Mataga, N. J . Phys. Chem. 1991, 95, 1956.
(d) Segawa, H.; Takehara, C.; Honda, K.; Shimidazu, T.; Asahi, T.
Mataga, N. J . Phys. Chem. 1992, 96, 503. (e) Asahi, T.; Ohkohchi, M.;
Mataga, N. J . Phys. Chem. 1993, 97, 13132. (f) Gould, I. R.; Noukakis,
D.; Gomez-J ahn, L.; Goodman, J . L.; Farid, S. J . Am. Chem. Soc. 1993,
115, 4405.
Methylviologen was prepared as the bis(trifluoromethane-
sulfonate) as follows: A solution of 4,4′-dipyridyl (3.12 g, 0.02
mol) in 50 mL of dichloromethane was stirred as methyl
trifluoromethanesulfonate (6.56 g, 0.04 mol), dissolved in 50
mL of the same solvent, was added over a period of 15 min.
During the addition, the solution became warm, and a white
precipitate of methylviologen ditriflate formed. Diethyl ether
(100 mL) was added to precipitate the remainder of the
product. The white microcrystalline solid was recrystallized
from a mixture of acetonitrile and ethyl acetate to afford thick
(62) (a) Hubig, S. M. J . Lumin. 1991, 47, 137. (b) Hubig, S. M. J .
Phys. Chem. 1992, 96, 2903. (c) Hubig, S. M.; Kochi, J . K. J . Phys.
Chem. 1995, 99, 17578.
(63) (a) It is remarkable that the competing processes of back-
electron transfer and carbon-carbon bond cleavage are completely
independent and the rate of one process is not affected by the presence
of the other. (b) A quantitative comparison is not possible at this
juncture, however, since the thermodynamic oxidation potentials of
the carboxylate donors are not directly available from electrochemical
measurements. Owing to the rapid C-CO2 bond cleavage in the acyloxy
radical, accurate E0 values for carboxylates cannot be obtained directly
from electrochemical experiments.64
(65) Ford-Moore, A. H. J . Chem. Soc. 1947, 952.
(66) Shacklett, C. D.; Smith, H. A. J . Am. Chem. Soc. 1953, 75, 2654.
(67) Ohwada, T.; Shudo, K. J . Org. Chem. 1989, 54, 5227.
(68) Ohwada, T.; Shudo, K. J . Am Chem. Soc. 1988, 110, 1862.
(69) Elemental analyses were performed by Atlantic Microlabs,
Norcross, GA.
(64) Najdo, L.; Save´ant, J .-M. J . Electroanal. Chem. 1973, 48, 113.

Ultrafast Decarboxylation of Acyloxy Radicals
J . Org. Chem., Vol. 62, No. 7, 1997 2221
colorless plates of methylviologen ditriflate. Anal. Calcd for
C14H14F6N2O6S2: C, 34.72; H, 2.91; N, 5.78. Found: C, 34.83;
H, 2.87; N, 5.73.
analysis. Irradiation during a time period of 60-80 min was
sufficient to yield 1.0-5.0 µmol of ketone.
P r ep a r a tion of Sa m p les for Tim e-Resolved Sp ectr os-
cop y. Aqueous solutions 0.02-0.10 M in methylviologen
ditriflate and 0.1-0.3 M in the sodium salt of the carboxylate
acid were prepared for the experiments on the femtosecond/
picosecond time scale. The absorbance at the irradiation
wavelength was 1.0-1.6. Preliminary experiments indicated
that the laser irradiation of degassed solutions generated large
amounts of persistent MV•+, which interfered with the time-
resolved spectroscopic measurements; accordingly, the solu-
tions were prepared in the open atmosphere and not degassed.
The sample cell was a 1.0 cm quartz flow cell, and the sample
solution was circulated with a Labline 5514 peristaltic pump
to minimize problems arising from secondary photolysis of the
products. Connections were made with Teflon and Fluoran
tubing. The UV-vis spectra of the circulating solution were
acquired before and after the laser photolysis experiments to
ensure that no significant thermal or photochemical changes
had occurred.
The transient quantum yields of MV•+ upon excitation of
the CT ion pairs were measured by the method of relative
transient actinometry.37 A solution of benzophenone (4.0 ×
10-3 M) in deaerated benzene served as the actinometer, and
the transient absorbance of the triplet of benzophenone, with
λmax ) 530 nm and ꢀBP ) 7220 M-1 cm-1 37 was monitored. The
absorbance was determined at three laser excitation energies.
Solutions of MV(OTf)2 (5-10 mM) and sodium carboxylate
(0.02-0.05 M) were made up in the same cuvette as used for
the actinometric measurements, and the concentrations were
adjusted to match those of the actinometer solution. The
solutions were then degassed by purging with argon. The
transient absorbance of MV•+ was monitored at λmax ) 605 nm.
This absorbance was observed to increase between 10 and 100
ns following the flash. The trace was linearly extrapolated to
zero time to obtain AMV for each laser excitation energy.
Quantum yields were calculated using the following relation:
ΦMV ) (AMV/ABP)/(ꢀMV/ꢀBP), where ABP is the absorbance of
benzophenone triplet and ꢀMV and ꢀBP are the extinction
coefficient of MV•+ (13 700 M-1 cm-1 at 605 nm22) and ben-
zophenone triplet (7220 M-1 cm-1 at 530 nm37), respectively.
The values of ΦMV so obtained were invariant ((10%) with
laser energy over a range from 4 to 22 mJ per pulse.
In str u m en ta tion . 1H- and 13C-NMR spectra were recorded
on a General Electric QE-300 NMR spectrometer, and the
chemical shifts are reported in ppm units downfield from
tetramethylsilane. UV-vis absorption spectra were recorded
on a Hewlett-Packard 8450A diode-array spectrometer. Gas
chromatography was performed on a Hewlett-Packard 5890A
gas chromatograph equipped with a flame-ionization detector
and a HP 3392 integrator. GC-MS analyses were carried out
on a Hewlett-Packard 5890A chromatograph interfaced to a
HP 5970 mass spectrometer (EI, 70 eV). Melting points were
measured on a Mel-Temp apparatus (Laboratory Devices) and
are uncorrected. Time-resolved spectroscopic measurements
on the picosecond and nanosecond/microsecond time scales
were carried out with laser spectrometers described previ-
ously.36
Deter m in a tion of P h otoch em ica l Qu a n tu m Yield s. An
aqueous solution of the sodium salt of the benzilate was
prepared by addition of 0.5 mmol of the free acid and 0.090 g
(0.55 mmol) of NaHCO3 to 5.0 mL of distilled water. The
solution was stirred until dissolution appeared to be complete
and then filtered and transferred to a 1.0-cm Pyrex test tube.
Methylviologen ditriflate (0.24 g, 0.5 mmol) was added, and
the tube was closed by a rubber septum. The solution was
then degassed by purging with argon for 30 min. The
absorbance of these solutions at 366 nm was always greater
than 2.0, to indicate that more than 99% of the incident light
was absorbed.
The light source for the quantum yield measurements was
a
medium-pressure 500-W mercury lamp (Osram HBO)
equipped with a circulating water filter to remove infrared
radiation. The 366-nm line of the mercury emission spectrum
was isolated by a combination of a CoSO4 solution filter and
colored-glass filters (Corning CS 0-52 and 7-37) as recom-
mended by Murov et al.71 The light flux was determined using
Aberchrome-540 as the actinometer.39 An actinometric mea-
surement was carried out prior to each quantum yield deter-
mination, and the light flux remained essentially constant at
1.8 ( 0.2 × 10-7 einsteins min-1. The solutions of MV(OTf)2
and Ar2C(OH)CO2Na were irradiated for 10-20 s until the
faint blue color of MV•+ persisted unchanged upon stirring for
10 min. (This preirradiation removed residual oxygen from
the reaction mixture.) The optical density of the solution at
604 nm was measured and taken as the initial absorbance (Ai).
The solution was then irradiated for 1.0 min, and the final
absorbance (Af) was measured. The quantum yields were
calculated as ΦSS ) [(Af - Ai)/ꢀMV](V/F),70 where V was the
volume of the solution (5.0 mL), F the light flux in einsteins
min-1, and ꢀMV the molar extinction coefficient of MV•+ in water
(13 700 M-1 cm-1 at 605 nm22).
The quantum yields of the diaryl ketones were determined
by quantitative gas chromatography using 4-chlorobenzophe-
none as the internal standard. Correction factors were
calculated by coinjection of known amounts of authentic
samples and the internal standard. Solutions were made up
with the same volumes and concentrations as for the deter-
mination of the MV•+ quantum yields. For the quantum yields
determined on the aerated samples (Φaer in Table 3), a slow
stream of air (1-2 bubbles/s) was drawn through the solution
during the irradiation. For the catalyzed reactions (Φcat), 11
mg of amorphous platinum dioxide (Adams catalyst, Strem)
was added to the solution, which was then degassed. In both
cases the blue color of MV•+ did not develop during the
irradiation. After the photolysis, the aqueous solution was
extracted three times with 2.0 mL portions of dichloromethane.
The organic extracts were washed with saturated aqueous
NaCl and dried over MgSO4, and the internal standard was
added. The yields of diaryl ketones were determined by GC
The transient spectra of the ketyl radicals, diphenyl-
hydroxymethyl and di-(p-tolyl)hydroxymethyl, were generated
from the corresponding diaryl ketone using the nanosecond/
microsecond laser spectrometer. The ketone (benzophenone
or 4,4′-dimethylbenzophenone (ca. 4 × 10-3 M) was dissolved
in a 1:1 (v:v) mixture of water and isopropyl alcohol. The
concentration was adjusted so that the absorbance was 0.4 at
355 nm, and the solution was degassed by bubbling with argon
for 30 min. Laser flash photolysis initially generated the
ketones in their transient triplet state, which decayed within
10 µs to a residual absorbance of the ketyl radical. The band
maximum and spectral width of the ketyl radical absorbance
was essentially identical to that reported in aqueous media.28,33
Ack n ow led gm en t. We thank the National Science
Foundation, the Robert A. Welch Foundation, and the
Texas Advanced Research Program for financial sup-
port.
Su p p or tin g In for m a tion Ava ila ble: The femtosecond
laser spectrometer is described in detail, and the kinetics of
the formation and decay of the radical pairs [MV•+, Ar2C(OH)-
•
CO2 ] are analyzed (5 pages). This material is contained in
libraries on microfiche, immediately follows this article in
the microfilm version of the journal, and can be ordered from
the ACS; see any current masthead page for ordering
information.
(70) See: de Mayo, P. In Creation and Detection of the Excited State;
Ware, W. R., Ed.; Dekker: New York, 1976; Vol. 4, p 194.
(71) Murov, S. L.; Carmichael, I.; Hug, G. L. Handbook of Photo-
chemistry; Dekker: New York, 1933; p 322.
J O9617833