Ultrafast Decarboxylation of Carbonyloxy Radicals: Influence of Molecular Structure

Article abstract of DOI:10.1021/jp0350823

Experimental and theoretical investigations on the ultrafast photoinduced decomposition of three tert-butyl peroxides of general structure R-C(O)O-O-tert-butyl with R = phenyloxy, benzyl, or naphthyloxy in solution are presented. Photoinduced O-O bond scission occurs within the time resolution (200 fs) of the pump-probe experiment. The subsequent dissociation of photochemically excited carbonyloxy radicals, R-CO₂, has been monitored on a picosecond time scale by transient absorption at wavelengths between 290 and 1000 nm. The measured decay of R-CO₂ is simulated via statistical unimolecular rate theory using molecular energies, geometries, and vibrational frequencies obtained from density functional theory (DFT) calculations. The results are compared with recent data for rerf-butyl peroxybenzoate (R = phenyl). While benzoyloxy radicals exhibit nanosecond to microsecond lifetimes at ambient temperature, insertion of an oxygen atom or a methylene group between the phenyl or naphthyl chromophore and the CO ₂ moiety significantly decreases the stability and thus lowers the lifetime of the carbonyloxy radicals in solution to picoseconds. The reasons behind this structural effect on decomposition rate are discussed in terms of barrier heights for decarboxylation on the ground-state potential energy surface and of a fast reaction channel via electronically excited states of carbonyloxy radicals. Arrhenius parameters are reported for thermal rate constants, k(T), of R-CO₂ decarboxylation as deduced from modeling of the time-resolved experimental data in conjunction with the DFT calculations.

Full text of DOI:10.1021/jp0350823

J. Phys. Chem. A 2003, 107, 9499-9510
9499
ARTICLES
Ultrafast Decarboxylation of Carbonyloxy Radicals: Influence of Molecular Structure
Bernd Abel, Jens Assmann, Michael Buback, Christian Grimm, Matthias Kling,*
Stefan Schmatz, Jo1rg Schroeder, and Thomas Witte
Institut fu¨r Physikalische Chemie, UniVersita¨t Go¨ttingen, Tammannstrasse 6, 37077 Go¨ttingen, Germany
ReceiVed: April 22, 2003; In Final Form: August 27, 2003
Experimental and theoretical investigations on the ultrafast photoinduced decomposition of three tert-butyl
peroxides of general structure R-C(O)O-O-tert-butyl with R ) phenyloxy, benzyl, or naphthyloxy in solution
are presented. Photoinduced O-O bond scission occurs within the time resolution (200 fs) of the pump-
probe experiment. The subsequent dissociation of photochemically excited carbonyloxy radicals, R-CO2,
has been monitored on a picosecond time scale by transient absorption at wavelengths between 290 and 1000
nm. The measured decay of R-CO2 is simulated via statistical unimolecular rate theory using molecular
energies, geometries, and vibrational frequencies obtained from density functional theory (DFT) calculations.
The results are compared with recent data for tert-butyl peroxybenzoate (R ) phenyl). While benzoyloxy
radicals exhibit nanosecond to microsecond lifetimes at ambient temperature, insertion of an oxygen atom or
a methylene group between the phenyl or naphthyl chromophore and the CO2 moiety significantly decreases
the stability and thus lowers the lifetime of the carbonyloxy radicals in solution to picoseconds. The reasons
behind this structural effect on decomposition rate are discussed in terms of barrier heights for decarboxylation
on the ground-state potential energy surface and of a fast reaction channel via electronically excited states of
carbonyloxy radicals. Arrhenius parameters are reported for thermal rate constants, k(T), of R-CO2
decarboxylation as deduced from modeling of the time-resolved experimental data in conjunction with the
DFT calculations.
I. Introduction
Studies on photochemical decomposition of organic peroxides
by means of microsecond and nanosecond flash photolysis in
Organic peroxides belong to a group of compounds of
fundamental as well as of application-oriented interest due to
their extensive use as initiators in free-radical polymerizations.1-4
The time scale of peroxide decomposition in different solvents
and the dynamics of active intermediates affect both initiation
rate and efficiency.4,5 To optimize the initiation efficiency, a
thorough understanding of the influence of peroxide structure
on fragmentation dynamics and kinetics is required. A large
body of literature has accumulated on the thermal decomposition
of organic peroxides,2,4 among which tert-butyl peroxides and
peroxyesters have been investigated in considerable detail.
conjunction with visible absorption or electron paramagnetic
resonance (EPR) spectroscopy15-22 have revealed that interme-
diate aroyloxy radicals are involved in the decomposition of
diaroyl peroxides. tert-Butyl peroxides such as tert-butyl per-
oxybenzoate and tert-butyl peroxy-4-methoxy-benzoate were
found to release the same reactive intermediates as the corre-
sponding diaroyl peroxides, that is, benzoyloxy and 4-methoxy-
benzoyloxy radicals, respectively.15,22 Investigations on substi-
tuted aroyloxy radicals revealed nanosecond to microsecond
lifetimes,17 in keeping with the observation of aroyloxy end
groups in polymers produced in reactions with dibenzoyl
peroxide being applied as initiator.23 However, the reason behind
the significantly different photodissociation quantum yields of
aroyloxy radicals from aromatic tert-butyl peroxyesters and
diaroyl peroxides15,22 remained unclear.
The mechanism of peroxide decarboxylation, that is, whether
bond breaking in peroxyesters, diacyl peroxides, or peroxycar-
bonates occurs in a stepwise (sequential) or a concerted fashion,
is of particular interest. This aspect has been addressed by
several groups.6-12 The question about the mechanism is
obviously related to the time scale of observation. On the basis
of thermal experiments, concerted bond breaking and stepwise
dissociation via intermediate radicals on a picosecond or
nanosecond time scale cannot be safely distinguished. Therefore,
highly time-resolved (subpicosecond) studies are required to
understand the decarboxylation mechanism in more detail.13,14
Intermediate radicals from light-induced decomposition of
tert-butyl peroxyesters were also identified in experiments with
nanosecond to picosecond time resolution.15,22,24-26 Kochi and
co-workers studied the decarboxylation of carbonyloxy radicals
after photogeneration via electron donor-acceptor complexes
by femtosecond visible spectroscopy.27,28 However, insufficient
time resolution or spectral overlap prevented a detailed insight
into the dissociation mechanism in prior studies. In particular,
the dependence of the dynamics on molecular structure, internal
* Corresponding author. Present address: Department of Chemistry,
University of California, Berkeley, D90 Hildebrand Hall, Berkeley,
California 94720. E-mail: mkling@uclink.berkeley.edu.
10.1021/jp0350823 CCC: $25.00 © 2003 American Chemical Society
Published on Web 10/22/2003

9500 J. Phys. Chem. A, Vol. 107, No. 45, 2003
Abel et al.
SCHEME 1: Scheme of Sequential Dissociation for UV
Photoinduced Decomposition of tert-Butyl Peroxyesters
with R ) Phenyl or Naphthyl and X ) O or CH2
SCHEME 2: Chemical Structures and Abbreviations of
tert-Butyl Peroxide Compounds Studied within the
Present Work, tert-Butyl Peroxyphenyl Carbonate
(TBPPC), tert-Butyl Peroxyphenylacetate (TBPPA), tert-
Butyl 2-Peroxynaphthyl Carbonate (TBNC), and
tert-Butyl Peroxybenzoate (TBPB)a
energy distribution, and solvent environment is not adequately
understood so far.4,5
In recent studies with picosecond to femtosecond time
resolution, we directly observed the formation and decarboxy-
lation of intermediate aroyloxy radicals13,29 and the formation
of aryl radicals and CO2 upon photoinduced decomposition of
tert-butyl peroxyesters and diaroyl peroxides.5,30,31 The results
of these experiments clearly show that peroxide photofragmen-
tation occurs as a fast sequential process, in which the primary
step, that is, the breaking of the O-O single bond, takes place
on a subpicosecond time scale.5,13,30,31 Thus the mechanism
anticipated on the basis of intuitive chemical arguments, in
particular, on the different nature of chemical bonds and bond
energies, is confirmed.32 These studies further demonstrate that
the average lifetime of aroyloxy radicals is significantly reduced
in the case that these radicals are produced in an electronically
excited state or in the electronic ground state with considerable
vibrational excess energy. Taking this finding into account, we
recently proposed a model for the two-step sequential dissocia-
tion of organic peroxides based on molecular parameters taken
from both experiments and quantum-chemical calculations.
The decarboxylation of aroyloxy radicals is considered to
consist of two processes: (i) the fast direct reaction from an
electronically excited state and (ii) the fragmentation in the
electronic ground state. The rate of the latter process is strongly
affected by the rate of vibrational cooling via coupling to the
bath modes. The vibrational energy transfer to the solvent
environment has been included in our model in a heuristic
manner. We were able to adequately represent and understand
benzoyloxy radical decarboxylation over the femtosecond to
microsecond time range, as well as differences in quantum yields
of benzoyloxy radicals from decomposition of tert-butyl per-
oxybenzoate and of dibenzoyl peroxide.29
a While TBPPC, TBPPA, and TBNC are explicitly studied, data for
TBPB from ref 13 are included to assist discussion of the kinetics
results.
onyloxy radicals will be analyzed with a model that has already
been applied for the quantitative description of decarboxylation
of benzoyloxy and naphthoyloxy radicals from photolysis of
TBPB13 and of di(1-naphthoyl) peroxide,29 respectively. The
application of experimental and theoretical methods in conjunc-
tion with the kinetic modeling of highly time-resolved concen-
tration versus time profiles is the key toward the understanding
of reaction mechanisms and time scales.
The paper is organized as follows: In section II, a brief
account of our experimental technique and theoretical approach
is given. In section III, we present the UV-vis spectra of the
reacting species, the primary experimental data, that is, the time-
resolved results of our pump-probe experiments, and the results
from quantum-chemical calculations to be used in section IV.
In section IV, we introduce our model for the decarboxylation
of carbonyloxy radicals and present the modeling of the
measured absorbance versus time profiles, which reflect the time
dependence of species concentrations. Subsequently, the influ-
ence of molecular structure on both ultrafast decarboxylation
via an electronically excited state and thermal decarboxylation
via the electronic ground state of carbonyloxy radicals are
discussed. Our conclusions are summarized in section V.
II. Experimental Technique and Theoretical Approach
In extension of the preceding work, the present paper
addresses photoinduced dissociation dynamics of peroxides of
general structure R-X-C(O)O-O-tert-butyl, with R ) phenyl
or naphthyl and X ) O or CH2. In Scheme 1, the general
decomposition mechanism is given. The primary photodisso-
ciation of the parent peroxide molecule yields the R-X-CO2/
tert-butoxy radical pair. Subsequent decarboxylation of R-X-
CO2 produces the radical R-X and CO2.
The peroxides under investigation are tert-butyl peroxyphenyl
carbonate (TBPPC), tert-butyl peroxyphenylacetate (TBPPA),
and tert-butyl 2-peroxynaphthyl carbonate (TBNC). The kinetic
results will be compared with recent data on tert-butyl peroxy-
benzoate (TBPB).13 The four peroxides are displayed in
Scheme 2.
The primary interest of the present study focuses on the
structural effect of the isolobal bridging groups between the
ring chromophore and the CO2 moiety. Electronic delocalization
decreases in going from TBPB to the systems TBPPC and
TBPPA. Comparison of the decomposition rates of TBPPC and
TBNC will afford studying the dependence of the kinetics on
the size of the aromatic chromophore. The formation of
intermediate radicals and their decarboxylation is monitored with
femtosecond time resolution. The decarboxylation of carb-
A Ti:sapphire-based regeneratively laser amplified system
(VERDI, COHERENT, ORC1000, CLARK), described in more
detail elsewhere,33 delivers 40 fs pulses (800 nm, 0.7 mJ/pulse)
at a 1 kHz repetition rate. The laser pulses were split into two
parts. The first part, about 30% of total pulse energy, was used
for harmonic generation (SHG, THG) of 266 nm pulses for
peroxide excitation. The second part serves as a pump pulse
for an optical parametric amplifier (TOPAS, Light Conversion).
The TOPAS output was frequency-mixed to provide probe
pulses in the wavelength range from 290 to 1000 nm. The delay
time between pump and probe pulses was set with a computer-
controlled translation stage (Newport). Both laser pulses were
weakly focused (f ) 200 mm) and overlapped spatially and
temporally at nearly collinear pump-probe geometry (5°) in a
fast-flow cell (0.1-0.2 mm thickness). Pulses with relative
polarization at the magic angle (54.7°) were chosen for the
experiments. Probe pulse intensities were recorded at a 1 kHz
repetition rate in front of and behind the cell, which contained
the peroxide solution at an optical density of about 2 (at the
pump wavelength). Some of the experiments were carried out
with a commercial Ti:sapphire regenerative amplifier system
(CPA-2001, Clark-MXR) and two optical parametric amplifiers
(TOPAS, Light Conversion and NOPA, Clark-MXR). Whereas

Ultrafast Decarboxylation of Carbonyloxy Radicals
J. Phys. Chem. A, Vol. 107, No. 45, 2003 9501
probe pulses for the same wavelength range as above were
provided by the TOPAS, tunable pump pulses (230-350 nm)
were obtained by second harmonic generation of the NOPA
signal. A time resolution of typically 100-200 fs was achieved
in solution. The peroxides were synthesized in high purity and
kindly provided by AKZO Nobel Polymer Chemicals. Solvents
of highest available quality were purchased from Merck.
Peroxides and solvents were used without further purification.
As in our previous work on the benzoyloxy radical,13 density
functional theory (DFT) calculations of geometries and energies
were performed for isolated phenyloxycarbonyloxy (Ph-O-
CO2), benzylcarbonyloxy (Ph-CH2-CO2), and naphthyloxy-
carbonyloxy (Nph-O-CO2) radicals. We expect the results for
isolated radicals to apply also to radicals in nonpolar and perhaps
even weakly polar aprotic solvents, such as the ones employed
in our experiments. Details of the quantum-chemical calculations
on the three radicals have been published elsewhere.34 We will
here restrict ourselves to a brief account of the procedure and
outline only those results for the carbonyloxy radicals that are
relevant for the present study. Geometry optimizations were
carried out with the DFT variant UB3LYP.35,36 We used the
6-31G(d), 6-311+G(d,p), 6-311+G(2d,p), and 6-311+G(2df,
2pd) basis sets in the calculations that were carried out
employing the Gaussian 98 program package.37 Three stationary
points (reactants, products, and first-order saddle points pertinent
to decarboxylation) on the potential energy surface (PES) of
each of the radicals were fully optimized using the 6-31G(d)
basis set, and the optimized geometries were then refined
employing the 6-311+G(d,p) basis. At the obtained structures,
additional single-point calculations with the two larger basis
sets were carried out. The stationary points were characterized
by calculation of the Hessian matrices, and transition-state
geometries were further confirmed using the intrinsic reaction
coordinate (IRC) method.38,39 On the basis of comparisons with
large-scale coupled-cluster calculations carried out in our
previous study,13 we expect the UB3LYP geometries to be
highly reliable. Because the saddle-point structures are very
close to the corresponding reactant geometries,34 the breaking
bonds are only slightly elongated and nondynamical electron
correlation effects are not expected to play a major role. Thus,
the B3LYP functional appears to be an adequate choice for the
description of the energetics of the decarboxylations.
by Misawa et al. in a flash photolysis experiment.21 On the basis
of quantum-chemical calculations, we recently assigned the
broad vis-near-IR absorption band of the benzoyloxy radical
to the X˜ 2B2 f B˜2A1 transition.13 As detailed in section IVA,
the energy of excited states plays a crucial role in determining
the contribution of the direct process from an electronically
excited state to overall dissociation of the carbonyloxy radicals.
Unfortunately, unperturbed transient spectra of the carbonyloxy
radicals formed in decomposition of TBPPA, TBPPC, and
TBNC have not been reported so far. We expect them to be
similar to the spectra of alkyloxycarbonyloxy radicals41-43 and
of other types of carbonyloxy radicals17,22 because optical
transitions in the UV-near-IR spectral region mainly originate
from the CO2 moiety of these radicals.
tert-Butoxy radicals (formed upon primary bond scission of
the peroxides) do not show any significant absorption in the
near-UV and visible spectral ranges.13 Radicals formed upon
decarboxylation of carbonyloxy radicals (the second step in the
sequential decomposition of the organic peroxides discussed
here) show significant absorption at wavelengths above 290 nm.
The absorption spectrum of the benzyl radical (produced upon
decarboxylation of TBPPA) was found to have a maximum at
306 nm at ambient temperature.44 The matrix spectrum of the
phenyloxy radical, as recently reported by Radziszewski et
al.,45,46 has a maximum at 295 nm in argon at 7 K. In addition,
phenyloxy radicals show a weak structureless band at around
625 nm. The absorption spectrum of the naphthyloxy radical
has not been reported so far. We assume its characteristic
features to be similar to those of phenyloxy, apart from a minor
red shift caused by the more extended aromatic system.
B. Decarboxylation Measured with Femtosecond Pump-
Probe Spectroscopy. The formation of CO2 after photolysis
of TBPPC and TBNC was probed in the IR spectral range with
a time resolution of a few picoseconds,5,30 which is insufficient
for obtaining detailed insight into decarboxylation mechanisms.
In most cases, intermediate carbonyloxy radicals may be
monitored with subpicosecond time resolution in the vis-near-
IR spectral range, and product radicals from decarboxylation
can be detected in the UV. Absorption of product radicals and
carbonyloxy radicals may overlap. We thus measured the
transient absorption over a wide UV to near-IR range to identify
suitable spectral windows with exclusive or at least predominant
absorption of either the intermediate or the product free radical
species. It should be noted that time-dependent spectral evolu-
tions due to solvation and vibrational energy transfer (which is
always present) are less important here and in previous cases.5,30
The differences in the features of the time-resolved traces for
the different molecules under investigation can be attributed
almost exclusively to spectral overlap of intermediate and
product radicals (see section IIIA).
III. Results
A. UV/Vis Spectra of Parent Molecules, Intermediates,
and Products. Stationary and transient spectra of parent
molecules, intermediates, and products will first be discussed.
The UV absorption spectra of TBPB in propylene carbonate
(PC) solution are given in ref 13, and those of TBPPA, TBPPC,
and TBNC are contained in the Supporting Information.
Significant UV absorption is found below 300 nm for TBPB,
TBPPA, and TBPPC and below 320 nm for TBNC. Excitation
at 266 nm into one of the repulsive electronically excited states
of these peroxides results in ultrafast direct O-O bond fission.40
According to Scheme 1, primary dissociation of the peroxides
under investigation yields a carbonyloxy/tert-butoxy radical pair.
The intermediate carbonyloxy radicals subsequently decompose
to product radicals (phenyl with TBPB, benzyl with TBPPA,
phenyloxy with TBPPC, and naphthyloxy with TBNC) and to
CO2.
TBPPC Decomposition. After photolysis of TBPPC in pro-
pylene carbonate (PC) solution at 266 nm, the transient
absorbance was recorded at various probe wavelengths between
290 and 830 nm. Typical experimental traces are shown in
Figure 1. The absorbance versus time traces measured in the
UV region from 290 to 320 nm (left column in Figure 1) differ
from each other: at 290 nm, after an initial peak, an off-set
and no significant temporal change in absorbance is seen,
whereas at 297 nm, a slight increase in absorbance occurs. At
and above 320 nm, apart from amplitude modulations, an initial
rise in absorbance occurs at t ) 0 ps, which is followed by a
decay on a picosecond to nanosecond time scale. The initial
peaks at around t ) 0 ps are in part due to nonlinear artifacts
generated in the solvent medium by the pump and probe pulses.
In solution, carbonyloxy radicals exhibit a characteristic broad
absorption band in the visible to near-infrared region and a
sharper absorption component with peak maximum below 450
nm. The spectrum of the benzoyloxy radical has been measured

9502 J. Phys. Chem. A, Vol. 107, No. 45, 2003
Abel et al.
absence of a temporal evolution of absorbance at 290 and 297
nm is probably due to overlapping absorption of the (decaying)
intermediate phenyloxycarbonyloxy radical and the (rising)
phenyloxy product radical. At wavelengths where both species
exhibit identical specific absorption, the dynamics may not be
detected from the absorbance versus time trace. Because the
absorption of the phenyloxy product radical is much stronger
in the UV, we expect this compensation in ∆OD to be more
pronounced at wavelengths around 295 nm, which is close to
the reported maximum of phenyloxy absorbance (see section
IIIA). The slight increase in absorbance with time seen at 297
nm (Figure 1) is thus best rationalized as being due to the
increase in phenyloxy radical absorbance, dominating over the
absorbance of the carbonyloxy radical. We conclude that the
unperturbed decarboxylation dynamics in this particular case
is best seen in the vis-near-IR region by the observation of
the absorbance decay associated with the decrease of intermedi-
ate carbonyloxy radical concentration.
TBPPA Decomposition. Examples of optical density versus
time traces measured at several wavelengths between 297 and
766 nm after photodissociation (at 266 nm) of TBPPA in PC
solution are shown in Figure 3. Again, the temporal behavior
of the transient absorption after photolysis of TBPPA strongly
depends on the probe wavelength, which indicates absorption
contributions of both intermediate and product radicals.
A weak initial increase in absorption is seen at 310 and 320
nm, which wavelengths are close to the reported maximum of
the benzyl radical spectrum at 306 nm (see section IIIA). We
assume the increase, for example, at 310 nm, to primarily reflect
the formation of benzyl radicals in the decomposition of TBPPA.
Figure 1. Transient absorbance after photolysis of TBPPC (tert-butyl-
O-O(O)C-O-Ph) in propylene carbonate solution at 266 nm. Probe
wavelengths are indicated.
The experimental trace at 297 nm does not show much
temporal change after a few picoseconds. At this wavelength,
which is shifted by only about 10 nm to the blue with respect
to a reported maximum of the benzyl radical absorbance, we
just observe an off-set that is due to absorption of both types
of radicals. Toward larger wavelengths (see Figure 3), we
observe a picosecond decay in absorption in addition to an off-
set. This decay is due to dominant absorption of the carbonyloxy
radical, which decarboxylates on a picosecond time scale
yielding a benzyl radical and CO2.
In contrast to TBPPC decomposition, in which all experi-
mental curves above 320 nm exhibit similar absorbance versus
time behavior, the situation is more complex with TBPPA
decomposition. Between 594 and 700 nm, after an initial peak
at t ) 0 ps, no pronounced temporal change in absorption is
seen with the exception of a slight increase or decrease on the
picosecond to nanosecond time scales. This behavior is similar
to the one observed at a probe wavelength of 297 nm, which
we attributed to the simultaneous absorption of the intermediate
and the product radical formed upon decomposition of TBPPA.
On the other hand, the absorbance at 766 nm clearly shows a
picosecond decay similar to the one seen at 330 and 340 nm.
Here the carbonyloxy radical is assumed to be dominant in its
absorption. In TBPPA decomposition, no region is found where
the absorbance may be unambiguously assigned to the carbo-
nyloxy radical. We decided to use the curve measured at 766
nm for analysis of decarboxylation dynamics of the carbonyloxy
radical. The reason for this choice is that the absorbance
contribution of product radicals is lowest at this wavelength,
as can be seen from the small off-set measured at larger reaction
times (Figure 3).
Figure 2. Experimental optical density versus time curves recorded
after photolysis of TBPPC in propylene carbonate at 266 nm and probe
wavelengths at and above 320 nm. The curves are normalized for better
comparison. For further details, see text.
In Figure 2, we compare the traces measured at wavelengths
at and above 320 nm. For better comparison, the curves have
been normalized by subtraction of an off-set and scaling of the
amplitude. The kinetics is very similar throughout the probe
region above 320 nm, supporting the assignment of the decay
dynamics mostly or even exclusively to the decay dynamics of
the phenyloxycarbonyloxy radical. After formation within the
experimental time resolution of 100-200 fs, the carbonyloxy
radical from TBPPC photolysis decarboxylates on a picosecond
to nanosecond time scale. The off-set in absorbance and the
TBNC Decomposition. The third molecule investigated is tert-
butyl-peroxy-2-naphthyl carbonate (TBNC). In the wavelength
range above 400 nm, an immediate absorbance increase within

Ultrafast Decarboxylation of Carbonyloxy Radicals
J. Phys. Chem. A, Vol. 107, No. 45, 2003 9503
Figure 3. Transient absorbance between 297 and 766 nm after photolysis of TBPPA (tert-butyl-O-O(O)C-CH2-Ph) in propylene carbonate
solution at 266 nm. Probe wavelengths are indicated.
less than a picosecond occurs. The subsequent decay on the
picosecond time scale is attributed to the absorption of the
intermediate 2-naphthyloxycarbonyloxy radical. Characteristic
experimental traces for 594 and 766 nm are displayed in Figure
4. The absorption at 297 nm is primarily due to the naphthyloxy
product radical.
measured at large experimental times corresponding to several
half-lives of the carbonyloxy radical.
In Figure 6, the naphthyloxy product formation measured at
297 nm and the naphthyloxycarbonyloxy radical decomposition
determined at 594 nm are compared. Note that the carbonyloxy
radical curve is inverted. The agreement between both curves
is very satisfying. Also shown in Figure 6 is the formation of
CO2 as measured by a completely independent IR experiment.30
As can be seen, the decay of the intermediate 2-naphthyloxy-
carbonyloxy radicals agrees well with the formation of both
products, the 2-naphthyloxy radical and CO2. The analysis of
the CO2 product concentration further indicates that CO2
formation has come to completion after 300 ps.47 The decar-
boxylation of TBNC after UV excitation thus clearly is a
picosecond process even at ambient temperatures. Experiments
on the decomposition of TBNC were also carried out in other
solvents,14 and the decarboxylation rate was found to be fairly
insensitive toward solvent polarity and viscosity.
In Figure 5, we compare normalized experimental absorbance
versus time traces recorded after decomposition of TBNC with
pump pulses at 266 nm in PC solution. When the different off-
sets at t < 0 ps and the peak at t ) 0 ps are ignored,
experimental traces throughout the vis-near-IR region show
the same time evolution. We assign this decay to the decar-
boxylation of the intermediate 2-naphthyloxycarbonyloxy radi-
cal. Identical decay dynamics of the intermediate carbonyloxy
radical are deduced from measurements at the six vis-near-IR
wavelengths.
According to the previous arguments for TBPPA, an off-set
on the experimental absorbance versus time curves at a particular
probe wavelength and for large t is due to absorbance of the
product radical. The temporal evolution of the carbonyloxy
radical concentration is obtained by its absorbance in the visible
region after subtraction of the product radical contribution as
C. Quantum-Chemical Calculations. Because of the flatness
of the potential energy surfaces in the vicinity of the saddle
points pertinent to decarboxylation of the phenyloxy-, benzyl-,
and naphthyloxycarbonyloxy radicals, the search for these

9504 J. Phys. Chem. A, Vol. 107, No. 45, 2003
Abel et al.
Figure 6. Absorbance change due to the decay of 2-naphthyloxycar-
bonyloxy radical concentration superimposed onto the formation of
2-naphthyloxy and of CO2. The data are from different experiments:
(O) inverted profile of 2-naphthyloxycarbonyloxy radicals obtained at
594 nm; (4) formation of 2-naphthyloxy radicals at 297 nm; (b) CO2
absorbance deduced from integrated transient absorption at 4.3 µm
during photolysis (308 nm) of TBNC in CCl4.30 The bold solid line
denotes the yield of direct decarboxylation as described in the text.
TABLE 1: Reaction Energies (in kcal mol-1) for
Decarboxylationa
Figure 4. TBNC photolysis at 266 nm in propylene carbonate. Shown
is the transient absorption of the naphthyloxycarbonyloxy radical as a
function of time at different probe wavelengths of 766 and 594 nm.
The trace for 297 nm primarily reflects the absorption of the
naphthyloxy radical.
carbonyloxy radical
∆RE
∆RH°(298 K)
Ph-O-CO2
Ph-CH2-CO2
Nph-O-CO2
-29.7
-28.3
-30.6
-30.8
-29.9
-31.2
a UB3LYP/6-311+G(2df,2pd) results calculated at the UB3LYP/6-
311+G(d,p) geometries are given. ∆RE denotes the electronic reaction
energy; ∆RH°(298 K) is the reaction enthalpy. ZPE and thermal
corrections for 298 K were calculated employing the 6-311+G(d,p)
basis set.
Figure 5. Comparison between experimental traces recorded after
decomposition of TBNC at 266 nm in propylene carbonate solution at
probe wavelengths in the vis-near-IR region (wavelengths indicated).
The curves have been shifted and normalized in intensity for better
comparison. The decay dynamics is assigned to the intermediate
2-naphthyloxycarbonyloxy radical.
Figure 7. Calculated reactive normal modes at the saddle point for
decarboxylation of carbonyloxy radicals: (1) phenyloxycarbonyloxy;34
(2) benzylcarbonyloxy;34 (3) naphthyloxycarbonyloxy;34 (4) benzoyl-
oxy.13
structures is quite complicated. Because the detailed results of
our DFT calculations are reported elsewhere,34 we here give
only a brief account. Reaction energies, ∆RE, and enthalpies,
∆RH°(298 K), are listed in Table 1. Harmonic vibrational
frequencies, ωi, of the reactant and the saddle-point species are
reported in ref 34. They are crucial parameters for the statistical
rate constant calculations described in section IV.
Phenyloxycarbonyloxy Radical (Ph-O-CO2). The results of
our UB3LYP/6-311+G(d,p) calculations indicate that the equi-
librium reactant and saddle-point geometries of this radical in
its X˜ 2A electronic ground state are very close to each other,
which is referred to as an “early” transition state. The transition
mode (normal coordinate with imaginary vibrational frequency)
is characterized by a shortening of the Ph-O bond, while the
O-CO2 bond is stretched and eventually broken. Simulta-
neously, the CO2 group becomes coplanar with the phenyl ring
and attains a linear configuration. The motion is governed by
the upper oxygen atom of the CO2 group (see Figure 7), whereas
the displacement vectors of the other two oxygen atoms and of
the carboxylic carbon atom are small. Despite the fact that the
transition mode seems to end up in a σ-phenoxy radical with
the unpaired electron in the ring plane, IRC calculations34 clearly
show that the phenoxy radical is produced in its X˜ 2B1 electronic
ground state with the unpaired electron being part of the

Ultrafast Decarboxylation of Carbonyloxy Radicals
J. Phys. Chem. A, Vol. 107, No. 45, 2003 9505
delocalized π-system. In a convergence study with respect to
basis set size, we calculated the barrier height at the geometry
optimized at the UB3LYP/6-311+G(d,p) level of theory em-
ploying the large 6-311+G(2d,p) and 6-311+G(2df,2pd) basis
sets.34 After correction of zero-point energy (ZPE), we obtain
radical are markedly longer than those with the other radicals.
The O-C-O bond angle, on the other hand, is similar for all
species with the exception of Nph-O-CO2 in which this angle
is enlarged by about 10°. While the X-CO2 bond distances are
shortest for the two aryloxycarbonyloxy radicals (X ) O) and
longest for the benzylcarbonyloxy species with its pure σ-type
bonding situation, benzoyloxy exhibits an intermediate situation.
Furthermore, whereas the three saddle-point structures of this
work can be classified as “early” transition states, the saddle
point in benzoyloxy is shifted much more to the product side.
a barrier height of E0 ) 2.2 kcal mol-1
.
Benzylcarbonyloxy Radical (Ph-CH2-CO2). The UB3LYP/
6-311+G(d,p) results for Ph-CH2-CO2, which are also given
in detail in ref 34, indicate that the transition state is very close
to the reactant structure (“early” TS). The transition mode is
described by a normal coordinate that considerably differs from
the case of the Ph-O-CO2 species. Here, the reactive normal
coordinate is characterized by a stretching of the CH2-CO2 bond
and a simultaneous widening of the CO2 angle, while the Ph-
CH2 skeleton is less affected by the reactive normal mode
(Figure 7). The CH2 group rotates into the ring plane toward
the C2V-symmetric structure of the product benzyl radical. The
displacement vector of the carbon atom in the CO2 group clearly
dominates the reaction path at the transition state. Visual
inspection of the transition mode shows that the benzyl radical
is produced in its X˜ 2B1 electronic ground state48 with the
unpaired electron being part of the π-system. The barrier height
for decarboxylation, for which we obtain a value of 1.07 kcal
mol-1, is only about one-half of that for Ph-O-CO2. It is
concluded that substitution of the bridging oxygen atom by a
methylene group results in a rather different transition mode
for decarboxylation. This is quite surprising because the oxygen
atom with its two lone electron pairs should behave similarly
to the carbon atom with the attached two hydrogen atoms.
Obviously, the bonding situation in phenyloxycarbonyloxy is
more complex, and lone pair electrons of the oxygen atom
interact with the π-system of the ring.
At 8.4 kcal mol-1 13
the barrier for decarboxylation of the
,
benzoyloxy radical is much higher than those for the other three
radicals. The reason for this behavior may be found in the strong
π-electron delocalization in the planar C2V-symmetric benzoyl-
oxy system.
To shed some light on the structure dependence of the
decarboxylation of carbonyloxy radicals from a theoretical
perspective, the particular motions of atoms in the normal modes
corresponding to the transition modes may be inspected (see
Figure 7). In the case of the benzoyloxy radical decarboxylation,
the transition mode is governed by a strong motion of the carbon
atom of the CO2 moiety. The motions of atoms at the saddle
points pertinent to decarboxylation are similar for the benzyl-
carbonyloxy and naphthyloxycarbonyloxy radicals. In both
cases, the decarboxylation is dominated by the motion of the
carbon atom of the CO2 group. In contrast to the benzoyloxy
radical, the oxygen atoms move in opposite directions so that
the formation of linear CO2 does not involve large changes in
the X-CO2 (X ) O, CH2) bond distances. The decarboxylation
dynamics of benzylcarbonyloxy and naphthyloxycarbonyloxy
radicals are similar, which is in agreement with our experimental
findings.
2-Naphthyloxycarbonyloxy Radical (Nph-O-CO2). The
UB3LYP/6-31G(d) calculations for the 2-naphthyloxycarbonyl-
oxy radical yield a very small barrier to decarboxylation, which
forced us to use the larger 6-311+G(d,p) basis set to obtain a
meaningful optimized geometry. If the harmonic zero-point
vibrational energy is added to the classical potential, the effective
(or adiabatic) barrier is calculated to be close to zero. Unfor-
tunately, the small barrier height in this case is below the
accuracy of the energy estimates by the applied method. To
obtain a reliable value for the barrier height to decarboxylation
in Nph-O-CO2, high-level quantum dynamical calculations
have to be performed, which are out of reach at present. As
will be shown below, we obtain a barrier height of E0 ) (2.9 (
0.1) kcal mol-1 from the analysis of our experimental data.
IV. Discussion
A. Modeling Sequential Decomposition of Peroxides.
Concerted breaking of two different bonds of a peroxyester (to
yield CO2) after excitation at wavelengths between 250 and 300
nm into the repulsive S1 state49 is rather unlikely as has been
pointed out in our recent paper.13 Due to the different nature of
the chemical bonds involved, fast sequential dissociation is
supposed to be the common decomposition pathway of aryl-
peroxyesters and diaroyl peroxides.32 Our model, which allows
for a quantitative prediction of the decarboxylation dynamics
of benzoyloxy radicals from decomposition of TBPB,13 assumes
that after absorption of a photon the peroxide decomposes within
one vibrational period and generates a vibrationally excited
carbonyloxy radical either in the X˜ ground state or in an
electronically excited state. The second fragment, which may
also be vibrationally hot, is not detected in our experiments. A
ground- (X˜ ) to excited-state transition in the near-IR range has
been used to probe the ground-state population of carbonyloxy
radicals.13
Electronically excited carbonyloxy radicals are supposed to
react in a fast barrierless “downhill reaction” to products, as is
depicted in Figure 8. We identified the electronically excited
state leading to direct decarboxylation of benzoyloxy radicals
as produced in TBPB decomposition to be the B˜2A1 state.13 For
carbonyloxy radicals, which exhibit no symmetry elements (thus
belonging to point group C1), such as the ones of the present
study, the specific type of electronically excited states involved
may not be unambiguously assigned, and no symmetry cor-
relations between carbonyloxy and product radical states can
be established. However, coupling between the states is assumed
to be sufficiently strong to ensure that reaction takes place
Our examination reveals that the Nph-O-CO2 geometry
differs markedly from the structure of Ph-O-CO2. Most of
the geometrical parameters in the saddle-point structure are only
very slightly changed with respect to the reactant species,
indicating, as expected, again an “early” transition state. The
transition mode differs considerably from the Ph-O-CO2
counterpart. While in the latter case the outer carboxylic oxygen
atom dominates the normal mode, the motion of the carboxylic
carbon atom is most pronounced with Nph-O-CO2 (see Figure
7). The O-CO2 bond stretches, along with a simultaneous
widening of the O-C-O angle, and the Nph-O bond becomes
shorter. In summary, substitution of the phenyl chromophore
by a naphthyl unit should result in significantly different
decarboxylation dynamics.
Comparison with the Benzoyloxy Radical (Ph-CO2). As
compared with the results for the benzoyloxy radical,13 the three
systems studied in the present paper show significant differ-
ences: the carboxylic C-O bond distances of the benzoyloxy

9506 J. Phys. Chem. A, Vol. 107, No. 45, 2003
Abel et al.
The function f(E,t) represents the time-dependent “decay” of
the initial energy distribution f(E,t ) 0) being given by
E
kBT
F(E) exp -
(
)
f(E,t)0) )
(3)
Qvib
Qvib is the molecular partition function, and T is the initial
temperature of the carbonyloxy radical after photolysis of the
corresponding peroxide precursor.
The factor frot in eq 2 is introduced to account for rotational
effects on the reaction:
3
frot
)
B /Bq
(4)
Figure 8. Schematic potential energy diagram by which the two
decarboxylation pathways of excited carbonyloxy radicals R-X-CO2
are illustrated. R is phenyl or naphthyl, and X is O or CH2. f(E) denotes
a particular initial energy distribution of the carbonyloxy radical, k(E)
is the specific rate constant for decarboxylation of the intermediate
radical, ∆E denotes the energy separation between the electronic ground
and the relevant electronically excited state of the carbonyloxy radical,
and ∆RE is the reaction energy for dissociation into CO2 and a product
radical R-X. For further details, see the text.
∏
i
i
x
i)1
with Bi and Bq being the rotational constants of the reactant
i
and transition-state species on the electronic ground-state PESs
of carbonyloxy radicals, respectively. These quantities are
calculated from the optimized DFT geometries. While the
change in rotational constants is negligible for naphthyloxy-
carbonyloxy radicals with frot ) 1.00 and very small for
benzylcarbonyloxy radicals with frot ) 1.02, it becomes some-
what more pronounced for phenyloxycarbonyloxy and benzoy-
loxy radicals, frot ) 1.04 and frot ) 1.06, respectively.
Intermolecular vibrational energy transfer was globally taken
into account by assuming first-order relaxation of f(E,t) with
an overall phenomenological relaxation time, τVET ) (7.5 (
1.5) ps, as observed for molecules of similar complexity and
of a similar number of vibrational degrees of freedom.51 τVET
was kept constant for all molecules dissolved in propylene
carbonate. For less-efficient solvents, such as CCl4, τVET was
selected to be 10 ps. Decay curves, S(t)norm, have been calculated
via eq 5
whenever sufficient excess energy for a particular reaction
channel is available. Depending on the particular reaction
channel, that is, whether the decarboxylation takes place
exclusively on the ground-state PES or also involves electroni-
cally excited states, CO2 is generated at different initial excess
energies, which may be monitored in transient picosecond IR
experiments.5,30,31
The contribution of immediate carbonyloxy fragmentation
from the electronically excited state to overall fragmentation
crucially depends on the location of the excited-state energy
level with respect to the initial energy of the carbonyloxy radical
as represented by the function f(E). For some diaroyl peroxides,
such as di(1-naphthoyl) peroxide and di(4-methoxy-benzoyl-
oxy) peroxide, the electronically excited-state level of the
intermediate radical is too high to be populated to any relevant
extent after peroxide photolysis at 266 nm.14 With these species,
carbonyloxy fragmentation occurs exclusively on the ground-
state PES. The process may be adequately treated by statistical
unimolecular rate theory.50
S(t)norm ) A(t) e-k(t)t
(5)
We used a stepladder model for the calculation of S(t)norm
,
analogous to a master-equation formulation as given in ref 13.
The decay curve was simulated using a stepwise procedure for
time intervals ∆t at constant internal energy E. Energy relaxation
was taken into account after each time interval ∆t. The optimum
size of an individual simulation time interval was found to be
∆t ) 10 fs. A scaling factor A(t) was calculated in each iteration
step, which ensures a continuous profile for connection of the
individual time slices. Prior to comparison with the experimental
data, the simulated decay curves, S(t)norm, have been convoluted
with the experimental cross-correlation function.
Our model assumes, first, that the initial excess energy of
the peroxide molecule is statistically distributed between the
two fragments and, second, that the internal energy of the
intermediate radical is statistically distributed over all intra-
molecular vibrational modes. Microcanonical reaction rate
constants for the decarboxylation of carbonyloxy radicals at
energy E are calculated via eq 1:50
B. Simulation of Time-Resolved Carbonyloxy Radical
Concentrations. Phenyloxycarbonyloxy Radical from TBPPC
Decomposition. The time evolution of the concentration of
phenyloxycarbonyloxy radicals is unambiguously deduced from
the visible spectral region. Off-sets due to contributions of other
species in this region are small and will be neglected in our
analysis of the experimental curves. In Figure 9, experimental
traces and simulations (solid line) are given for the time
evolution of the phenyloxycarbonyloxy radical from TBPPC
photolysis between 257 and 266 nm in solvents of different
viscosity and polarity. As has been mentioned above, the “spike”
on the signal at around t ) 0 ps is mainly due to a coherence
artifact (here only present in PC as a solvent), which occurs on
the time scale of the cross-correlation of our experiment, shown
by the inset in Figure 9.
Wq(E - E0)
k(E) )
(1)
hF(E)
with Wq(E - E0) and F(E) being the number of energetically
accessible transition-state energy levels and the density of
reactant states at energy E, respectively. The latter two quantities
may be calculated in the rigid-rotor and harmonic-oscillator
approximation using barrier heights, E0, and reactant and
transition-state frequencies available from DFT calculations (see
above). The thermal time-dependent rate constant is given by
averaging over the time-dependent energy distribution, f(E,t):
k(t) ) frot ∞f(E,t)k(E) dE
(2)
∫
0

Ultrafast Decarboxylation of Carbonyloxy Radicals
J. Phys. Chem. A, Vol. 107, No. 45, 2003 9507
Figure 9. Simulation of the transient absorption of phenyloxycarbo-
nyloxy radicals during photoinduced decomposition of TBPPC in
acetonitrile (4, 257 nm pump/620 nm probe), n-heptane (O, 257 nm
pump/600 nm probe), and propylene carbonate (0, 266 nm pump/594
nm probe). The solid line (s) represents the simulation as described
Figure 10. Normalized absorption of benzylcarbonyloxy radicals (4)
at 766 nm during photoinduced (266 nm) decomposition of TBPPA in
propylene carbonate (contributions due to the product benzyl radical
have been subtracted). The solid line (s) represents the simulation of
benzylcarbonyloxy radical decarboxylation with the parameters Evib
) 3000 cm-1, E0 ) 3.3 kcal mol-1, and τVET ) 7.5 ps. For further
details, see the text.
in the text with the parameters Evib ) 7100 cm-1, E0 ) 4.3 kcal mol-1
and τVET ) 7.5 ps.
,
Because accurate estimates of barrier height, E0, are com-
plicated because of the flatness of the PES in the saddle-point
region (see section IIIC), we treated E0 as a fit parameter. Energy
transfer to the solvent is taken into account by a time constant,
decarboxylation of the phenyloxycarbonyloxy radical after
TBPPC excitation at 266 nm in propylene carbonate is estimated
to proceed with (50 ( 30)% via an electronically excited state.
Benzylcarbonyloxy Radical Decarboxylation from TBPPA
Decomposition. According to the observed decay of the
intermediate benzylcarbonyloxy radical (dominant absorption
at, for example, 766 nm) and in agreement with our quantum-
chemical calculations, the decarboxylation dynamics has been
found to be a picosecond process. Figure 10 shows experimental
and simulated traces for the decarboxylation dynamics of the
benzylcarbonyloxy radical. The experimental signal is obtained
by subtraction of the benzyl radical contribution at 766 nm
measured at large delay times.
τVET.
13 The barrier height may be accurately obtained by fitting
the ∆OD vs t data at larger delay times, which part of the
experimental traces reflects the thermal decarboxylation of
carbonyloxy radicals at ambient temperature. The best fit was
obtained with E0 ) (4.3 ( 0.1) kcal mol-1, which is reasonably
close to the value deduced from our DFT calculations (2.2 kcal
mol-1). We simulated the decarboxylation of the carbonyloxy
radical on its ground-state PES with different initial internal
vibrational energies and found that a value of Evib ) 7100
cm-1, corresponding to a vibrational temperature of 720 K,
yields very satisfying agreement with the experimental data at
early delay times. The uncertainty associated with the relaxation
time constant of τVET ) (7.5 ( 1.5) ps results in an uncertainty
of up to 100 K for the initial temperature of the carbonyloxy
radicals.
The total internal energy of phenyloxycarbonyloxy radicals
that is instantaneously available at t ) 0 with TBPPC photolysis
at 257 and 266 nm may be estimated from the reaction enthalpy
of peroxide dissociation into the two primary radicals, which
was determined to be 23.0 kcal mol-1 at 298 K (UB3LYP/6-
311+G(2d,p)). The obtained initial energies of 15 300 cm-1
(257 nm excitation) and 14 600 cm-1 (266 nm excitation)
correspond to temperatures of 1130 and 1100 K, respectively,
which are almost twice as high as the value found from
simulation of the statistical ground-state reaction. This discrep-
ancy may be explained by a significant contribution of ultrafast
phenyloxycarbonyloxy decarboxylation via an electronically
excited state. The latter reaction depletes the high-energy part
of the phenyloxycarbonyloxy radical population on a subpico-
second time scale. The remaining radicals react statistically on
the ground-state PES. This interpretation is consistent with the
observation from previous picosecond IR experiments that the
produced CO2 has high initial internal vibrational energy at time
t being close to zero.5 The branching ratio of decarboxylation
via an electronically excited state to decarboxylation on the
ground-state potential energy surface may be estimated from
both CO2 excess energies and initial internal energies available
for ground-state decarboxylation of carbonyloxy radicals. The
A barrier height of E0 ) (3.3 ( 0.1) kcal mol-1 was used in
the simulation, which number is in reasonable agreement with
the value from our quantum-chemical calculations (1.07 kcal
mol-1). We determined the rate constant for the thermal
decarboxylation of benzylcarbonyloxy radicals at 300 K to be
k ) 1.3 × 1010 s-1. With an initial internal vibrational energy
of the ground-state benzylcarbonyloxy radical of Evib ) 3000
cm-1, corresponding to an internal temperature of 450 K,
satisfactory agreement between experiment and simulation is
obtained (see Figure 10). The total initial internal energy of the
benzylcarbonyloxy radical available after 266 nm photolysis of
TBPPA is estimated using ∆RH°(298 K) ) 24.7 kcal mol-1
(calculated at UB3LYP/6-311+G(2d,p)). One obtains an initial
energy of 15 000 cm-1, corresponding to an initial temperature
of 1050 K. As compared to TBPPC decomposition, the
discrepancy between (estimated) initial total internal energy and
initial vibrational energy of the benzylcarbonyloxy radical, as
obtained from kinetic modeling, is even larger with TBPPA.
This suggests that direct decarboxylation from the excited-state
plays an even more important role. The contribution is estimated
to be (75 ( 20)%. This value is in agreement with the large
instantaneous absorbance change as compared to the subsequent
minor time-dependent change observed in the spectral trace
taken at 310 nm (see Figure 3).
Our findings are in agreement with what has been reported
on the thermal decomposition of TBPPA,7-12 for which a close-
to-concerted decarboxylation mechanism was postulated. Bock-
man et al. studied the decarboxylation of benzylcarbonyloxy

9508 J. Phys. Chem. A, Vol. 107, No. 45, 2003
Abel et al.
the excitation wavelength does not significantly change the
observed temporal evolution. This can be easily understood in
terms of our model: within the first 5 to 10 ps, we observe the
ground-state reaction of 2-naphthyloxycarbonyloxy radicals with
a modest excess internal energy corresponding to about 370 K.
After equilibration to ambient temperature with a time constant
of about (7.5 ( 1.5) ps, the thermal decarboxylation at ambient
temperature is seen. Measurements in different solvents14
demonstrate that the latter process is affected neither by solvent
viscosity nor by polarity (e.g., n-heptane and PC as solvents do
not result in any significant change of decarboxylation dynamics,
see Figure 11). With an enthalpy of reaction, ∆RH°(298 K), of
19.7 kcal mol-1, as calculated by UB3LYP/6-311+G(d,p)//6-
31G+(d,p), the initial internal energy of 2-naphthyloxy-
carbonyloxy radicals after photodissociation of the peroxide at
wavelengths between 257 and 308 nm is estimated to be
15 500-19 500 cm-1, which corresponds to initial temperatures
in the range 900-1030 K. The initial internal energy of
carbonyloxy radicals monitored in their respective ground state
is much lower than predicted by statistical estimates. This
observation suggests a large contribution of direct decarboxy-
lation of the intermediate 2-naphthyloxycarbonyloxy radical via
an electronically excited state. The contribution of direct
decarboxylation is estimated from the amplitude ratio of prompt
versus delayed product formation as seen in Figure 6 (indicated
by the bold solid line representing the contribution of direct
decarboxylation) to be (50 ( 20)%. This percentage is consistent
with the value deduced from depletion of the high-energy tail
of the initial energy distribution of intermediate radicals and
with high excess energies of the CO2 product found in previous
picosecond IR experiments.5
Figure 11. Normalized transient absorption of 2-naphthyloxy-
carbonyloxy radicals during photoinduced decomposition of TBNC in
n-heptane (], 282 nm pump/600 nm probe) and propylene carbonate
(O, 266 nm pump/594 nm probe). The solid line (s) represents the
simulation as described in the text with the parameters Evib ) 2700
cm-1, E0 ) 2.9 kcal mol-1, and τVET ) 7.5 ps. For further details, see
text.
and similar carbonyloxy radicals by femtosecond visible spec-
troscopy28 and found a rate constant of k ) 1.6 × 109 s-1 for
the decarboxylation of benzylcarbonyloxy radicals in water.
Hilborn and Pincock report a unimolecular rate constant of k
) 5 × 109 s-1 for the decarboxylation of benzylcarbonyloxy
radicals in methanol.52 It is however not clear whether these
data28,52 are suitable for comparison with the ones from our study
in a weakly interacting solvent environment.
C. Dependence of the Decarboxylation Kinetics on Per-
oxide Structure. Our experiments on various peroxides indicate
that the ultrafast decarboxylation kinetics is clearly related to
the structure of the parent peroxide molecule. However, a clear
trend of the variations in time-resolved traces upon chemical
substitution is difficult to identify. According to the pioneering
work by Bartlett and Hiatt,6 the structural dependence of the
decarboxylation dynamics results, at least in part, from different
stabilities of the product radicals. Our calculations indicate an
increase in exothermicity of the decarboxylation step by about
30 kcal mol-1 in going from R-CO2 to R-X-CO2. According
to the Bell-Evans-Polanyi principle,53,54 this trend is ac-
companied by a lowering of barrier height. This effect may also
account for the variation in the empirical barrier values deduced
from the modeling of experimental decarboxylation data of
carbonyloxy radicals, R-X-CO2. Unfortunately, the accurate
prediction of the small barrier heights by DFT calculations
turned out to be more difficult than that in cases where the
barrier is high.14 The relative size of barrier heights is, however,
well reproduced by our quantum-chemical calculations.
2-Naphthyloxycarbonyloxy Radical Decarboxylation from
TBNC Decomposition. Fast decarboxylation on a picosecond
time scale is also seen for the 2-naphthyloxycarbonyloxy radical
generated in the photolysis of TBNC in solution. Figure 11
shows the decarboxylation dynamics of this radical as measured
at 594 nm in PC and at 600 nm in n-heptane after subtraction
of the contribution of the product radical (measured at 297 nm
and normalized in intensity to represent the ∆OD values at 300
ps of the curves around 600 nm). The decarboxylation of
2-naphthyloxycarbonyloxy radicals, formed upon decomposition
of TBNC at UV wavelengths, occurs in a similar way as does
the decarboxylation of benzylcarbonyloxy radicals. The decar-
boxylations are found to be highly exothermic for both
carbonyloxy radicals and involve a low reaction barrier. This
is consistent with our DFT calculations, which gave low barrier
values for the decarboxylation for benzylcarbonyloxy and for
2-naphthyloxycarbonyloxy radicals. Assuming these barriers to
be correct means that decarboxylation of both carbonyloxy
radicals is a picosecond process even at ambient temperature
(see Figure 11).
In addition to the effect of barrier height, the different
contributions of electronically excited states to overall decar-
boxylation that originate from chemical (structural) modification
probably play an important role. These contributions are by far
more significant than those resulting from changes in frequencies
and in the nature of the transition modes (147i cm-1 for Ph-
O-CO2 decarboxylation, 257i cm-1 for Nph-O-CO2, 795i
cm-1 for Ph-CH2-CO2 decarboxylation,34 and 253i cm-1 for
the benzoyloxy radical13). In fact, the differences in the rate
constants of more than 3 orders of magnitude between, for
example, Ph-CO2 and Nph-O-CO2 decarboxylation (see
Table 2) can obviously not be explained solely in terms of
differences in transition modes.
The decarboxylation of 2-naphthyloxycarbonyloxy radicals
was simulated using vibrational frequencies and rotational
constants from our UB3LYP calculations.34 Fitting of the
experimental data yields a barrier of E0 ) (2.9 ( 0.1) kcal
mol-1, which is the lowest barrier value found for the
investigated R-X-CO2 radicals, in agreement with the trend
in our calculations. Under the particular photodecomposition
conditions, the initial internal energy of the 2-naphthyloxycar-
bonyloxy radical was deduced to be Evib ) 2700 cm-1
,
corresponding to an initial temperature of 370 K. The decar-
boxylation was also measured in n-heptane solution after
excitation of TBNC at 282 nm (see Figure 11). When the artifact
at t ) 0 ps in PC solution is ignored, the results indicate that

Ultrafast Decarboxylation of Carbonyloxy Radicals
J. Phys. Chem. A, Vol. 107, No. 45, 2003 9509
TABLE 2: Arrhenius Parameters, A and Ea, as Deduced
from a Fit of Temperature-Dependent Rate Constants
between 200 and 2000 K Estimated via Transition-State
Theory
of an ultrafast excited-state decarboxylation channel of initially
highly excited carbonyloxy radicals. Low barrier heights toward
decarboxylation of carbonyloxy radicals are associated with
large reaction contributions via excited states of these radicals,
which suggests that the energy gap between ground and excited
states is small for less-stable carbonyloxy radicals. Direct
decarboxylation might occur via the first electronically excited
state in these carbonyloxy radicals, which is known to be
energetically close to the electronic ground state. In support of
this conclusion, a recent study55 on the highly exothermal
decarboxylation of 9-methylfluorenylcarbonyloxy radicals gen-
erated via UV photolysis of tert-butyl 9-methylfluorene-9-
percarboxylate, which exhibit a lifetime of only 55 ps in ambient
temperature solutions,24 provides evidence for a significant
contribution of an excited-state channel to the reaction.
The detailed analysis of the photoinduced decomposition of
organic peroxides also allows us to estimate Arrhenius param-
eters for thermal decarboxylation rate constants, k(T), of
carbonyloxy radicals. These rate coefficients are important
quantities for deducing initiator efficiencies of peroxides in free-
radical polymerizations. The correlation between peroxide
structure and decomposition kinetics is a key feature for
selecting suitable peroxide initiators.
carbonyloxy radical A (1013 s-1
)
Ea (kcal mol-1
)
k(300 K) (s-1
)
Ph-CO2
7.66
0.59
0.97
0.75
9.44
4.83
3.94
3.29
1.0 × 107
1.8 × 109
1.3 × 1010
3.0 × 1010
Ph-O-CO2
Ph-CH2-CO2
Nph-O-CO2
A general quantitatiVe understanding of the structure depen-
dence or predictive power will certainly benefit from additional
experimental and theoretical investigations, in particular on
systems with small activation barriers and large excited-state
contributions to the dissociation dynamics.
Finally, we would like to point out that data from these
photochemically induced measurements in combination with
quantum-chemical calculations and kinetic modeling may
provide thermal rate constants for the decarboxylation of
peroxide molecules. Benzoyloxy radicals in their electronic
ground state exhibit a lifetime in the nanosecond to microsecond
range and are thus relatively stable toward decarboxylation at
ambient temperature, whereas carbonyloxy radicals of general
structure R-X-CO2 (R ) phenyl or naphthyl, X ) O or CH2)
in their electronic ground states are considerably less stable.
Table 2 shows kinetic parameters for the decarboxylation of
carbonyloxy radicals. To deduce these numbers, temperature-
dependent thermal rate constants within the range 200-2000
K have been calculated employing transition-state theory on the
basis of our experimentally determined barrier heights and
quantum-chemical calculations. These rate coefficients have
been fitted to an Arrhenius-type expression, k(T) ) A
exp(-Ea/(RT)), to obtain the parameters A and Ea listed in Table
2.
Acknowledgment. The authors acknowledge financial sup-
port by the Deutsche Forschungsgemeinschaft (DFG) within
Sonderforschungsbereich 357 (“Molekulare Mechanismen Uni-
molekularer Prozesse”) and from the Fonds der Chemischen
Industrie. We are grateful to AKZO-Nobel Polymer Chemicals
(Deventer) for preparing and kindly providing the peroxide
samples. We thank Dipl.-Chem. J. Zerbs for experimental
support and Dr. R. Oswald for his help with preparing Figure
7. We also enjoyed stimulating discussions about various aspects
of this work with Professors P. Botschwina and J. Troe.
These thermal rate constants from ultrafast photoinduced
dissociation experiments should be very helpful for estimating
thermally actiVated decomposition rate of peroxides over a very
wide time scale.
Supporting Information Available: Linear absorption
spectra for tert-butyl peroxyphenyl carbonate (TBPPC), tert-
butyl peroxyphenylacetate (TBPPA), and tert-butyl peroxy-2-
naphthyl carbonate (TBNC) between 220 and 340 nm. This
material is available free of charge via the Internet at http://
pubs.acs.org.
V. Conclusions
The decarboxylation of carbonyloxy radicals in solution from
laser light-induced decomposition of organic peroxides of
general structure R-X-C(O)O-O-tert-butyl with R ) phenyl
and X ) O (TBPPC), R ) phenyl and X ) CH2 (TBPPA), and
R ) naphthyl and X ) O (TBNC) has been studied by
absorption spectroscopy on a picosecond time scale. The
decarboxylation kinetics is compared with recently published
data for tert-butyl peroxybenzoate (Ph-C(O)O-O-tert-butyl,
TBPB) decomposition.13 Thermally equilibrated benzoyloxy
radicals decarboxylate on a nanosecond to microsecond time
scale. In contrast, carbonyloxy radicals from UV photolysis of
TBPPA, TBPPC, and TBNC decarboxylate at much faster rates
in the picosecond time domain. The decarboxylation rate on
the ground-state PES of the carbonyloxy radicals increases in
the order phenyl-C(O)O-(O-tert-butyl) < phenyl-O-C(O)O-
(O-tert-butyl) < phenyl-CH2-C(O)O-(O-tert-butyl) < naph-
thyl-O-C(O)O-(O-tert-butyl).
With barrier heights and vibrational frequencies of the
carbonyloxy radicals obtained from DFT calculations for both
reactant states and saddle-point geometries, we have used our
recently proposed model to simulate the decarboxylation
dynamics. The observed order of decarboxylation rates of the
studied peroxides is consistent with model predictions based
on UB3LYP calculations. In addition, our analysis provides
evidence for a significant contribution to overall decarboxylation
References and Notes
(1) Fujimori, K. In Organic Peroxides; Ando, W., Ed.; Wiley: New
York, 1992; p 319.
(2) Sawaki, Y. In Organic Peroxides; Ando, W., Ed.; Wiley: New
York, 1992; p 425.
(3) Barson, C. A.; Bevington, J. C. J. Polym. Sci., Part A: Polym.
Chem. 1997, 35, 2955.
(4) Buback, M.; Sandmann, J. Z. Phys. Chem. 2000, 214, 583.
(5) Buback, M.; Kling, M.; Seidel, M. T.; Schott, F.-D.; Schroeder, J.;
Steegmu¨ller, U. Z. Phys. Chem. 2001, 215, 717.
(6) Bartlett, P. D.; Hiatt, R. R. J. Am. Chem. Soc. 1958, 80, 1398.
(7) Bartlett, P. D.; Simons, D. M. J. Am. Chem. Soc. 1960, 82, 1753.
(8) Bartlett, P. D.; Ru¨chardt, C. J. Am. Chem. Soc. 1960, 82, 1756.
(9) Koenig, T.; Wolf, R. J. Am. Chem. Soc. 1967, 89, 2948.
(10) Koenig, T.; Wolf, R. J. Am. Chem. Soc. 1969, 91, 2574.
(11) Pryor, W. A.; Smith, K. J. Am. Chem. Soc. 1970, 92, 5403.
(12) Neuman, R. C.; Behar, J. V. J. Am. Chem. Soc. 1969, 91, 6024.
(13) Abel, B.; Assmann, J.; Botschwina, P.; Buback, M.; Kling, M.;
Oswald, R.; Schmatz, S.; Schroeder, J.; Witte, T. J. Phys. Chem. A 2003,
107, 5157.
(14) Assmann, J.; Kling, M.; Abel, B. Angew. Chem., Int. Ed. 2003,
42, 2226.
(15) Chateauneuf, J.; Lusztyk, J.; Ingold, K. U. J. Am. Chem. Soc. 1988,
110, 2877.
(16) Hashimoto, J.; Segawa, K.; Sakuragi, H. Chem. Phys. Lett. 1999,
314, 261.
(17) Wang, J.; Tateno, T.; Sakuragi, H.; Tokumaru, K. J. Photochem.
Photobiol. 1995, 92, 53.

9510 J. Phys. Chem. A, Vol. 107, No. 45, 2003
Abel et al.
(18) Yamauchi, S.; Hirota, N.; Takahara, S.; Sakuragi, H.; Tokumaru,
K. J. Am. Chem. Soc. 1985, 107, 5021.
(19) Yamauchi, S.; Hirota, N.; Takahara, S.; Misawa, H.; Sawabe, K.;
Sakuragi, H.; Tokumaru, K. J. Am. Chem. Soc. 1989, 111, 4402.
(20) Tateno, T.; Sakuragi, H.; Tokumaru, K. Chem. Lett. 1992, 20, 1883.
(21) Misawa, H.; Sawabe, K.; Takahara, S.; Sakuragi, H.; Tokumaru,
K. Chem. Lett. 1988, 23, 357.
(22) Chateauneuf, J.; Lusztyk, J.; Ingold, K. U. J. Am. Chem. Soc. 1988,
110, 2886.
(23) Bevington, J. C.; Lewis, T. D. Trans. Faraday Soc. 1958, 54, 1340.
(24) Falvey, D. E.; Schuster, G. B. J. Am. Chem. Soc. 1986, 108, 7419.
(25) Morlino, E. A.; Bohorquez, M. D.; Neckers, D. C.; Rodgers, A. A.
J. J. Am. Chem. Soc. 1991, 113, 3599.
(26) Allen, N. S.; Hardy, S. J.; Jacobine, A. F.; Glaser, D. M.;
Navaratnam, S.; Parsons, B. J. J. Photochem. Photobiol., A 1990, 50, 389.
(27) Bockman, T. M.; Hubig, S. M.; Kochi, J. K. J. Am. Chem. Soc.
1996, 118, 4502.
(28) Bockman, T. M.; Hubig, S. M.; Kochi, J. K. J. Org. Chem. 1997,
62, 2210.
(29) Abel, B.; Assmann, J.; Buback, M.; Kling, M.; Schmatz, S.;
Schroeder, J. Angew. Chem., Int. Ed. 2003, 42, 299.
(30) Aschenbru¨cker, J.; Buback, M.; Ernsting, N. P.; Schroeder, J.;
Steegmu¨ller, U. Ber. Bunsen-Ges. Phys. Chem. 1998, 102, 965.
(31) Aschenbru¨cker, J.; Buback, M.; Ernsting, N. P.; Schroeder, J.;
Steegmu¨ller, U. J. Phys. Chem. B 1998, 102, 5552.
(32) Maul, C.; Gericke, K.-H. Int. ReV. Phys. Chem. 1997, 16, 1.
(33) Assmann, J.; v. Benten, R.; Charvat, A.; Abel, B. J. Phys. Chem.
A 2003, 107, 1904.
(34) Kling, M.; Schmatz, S. Phys. Chem. Chem. Phys. 2003, 5, 3891.
(35) Becke, A. D. J. Chem. Phys. 1993, 98, 5648.
(36) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785.
(37) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb,
M. A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.;
Stratmann, R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A.
D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi,
M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.;
Ochterski, J.; Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick,
D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.;
Ortiz, J. V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi,
I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.;
Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.; Gill, P. M.
W.; Johnson, B. G.; Chen, W.; Wong, M. W.; Andres, J. L.; Head-Gordon,
M.; Replogle, E. S.; Pople, J. A. Gaussian 98, revision A.9; Gaussian,
Inc.: Pittsburgh, PA, 1998.
(38) Gonzales, C.; Schlegel, H. B. J. Chem. Phys. 1989, 90, 2154.
(39) Gonzales, C.; Schlegel, H. B. J. Phys. Chem. 1990, 94, 5523.
(40) Dorer, F. H.; Johnson, S. N. J. Phys. Chem. 1971, 75, 3651.
(41) Chateauneuf, J.; Lusztyk, J.; Maillard, B.; Ingold, K. U. J. Am.
Chem. Soc. 1988, 110, 6727.
(42) Behar, D.; Czapski, G.; Duchovny, I. J. Phys. Chem. 1970, 74,
2206.
(43) Chen, S.; Cope, V. W.; Hoffman, M. Z. J. Phys. Chem. 1973, 77,
1111.
(44) Ikeda, N.; Nakashima, N.; Yoshihara, K. J. Phys. Chem. 1984, 88,
5803.
(45) Radziszewski, J. G.; Gil, M.; Gorski, A.; Spanget-Larsen, J.; Waluk,
J.; Mroz, J. J. Chem. Phys. 2001, 115, 9733.
(46) Radziszewski, J. G.; Gil, M.; Gorski, A.; Spanget-Larsen, J.; Waluk,
J.; Mroz, J. J. Chem. Phys. 2002, 116, 5912.
(47) Steegmu¨ller, U. Dissertation (Ph.D.), University of Go¨ttingen,
Go¨ttingen, Germany, 1997.
(48) Rice, J. E.; Handy, N. C.; Knowles, P. J. J. Chem. Soc., Faraday
Trans. 2 1987, 83, 1643.
(49) Scaiano, J. C.; Wubbels, G. G. J. Am. Chem. Soc. 1981, 103,
640.
(50) Baer, T.; Hase, W. L. Unimolecular Reaction Dynamics: Theory
and Experiments; Oxford University Press: New York, 1996.
(51) Schwarzer, D.; Troe, J.; Zerezke, M. J. Chem. Phys. 1997, 107,
8380.
(52) Hilborn, J. W.; Pincock, J. A. J. Am. Chem. Soc. 1991, 113, 2683.
(53) Bell, R. P. Trans. Faraday Soc. 1937, 33, 496.
(54) Evans, M. G.; Polanyi, M. Trans. Faraday Soc. 1936, 32, 1333.
(55) Abel, B.; Buback, M.; Kling, M.; Schmatz, S.; Schroeder, J. J. Am.
Chem. Soc., in press (available in electronic form via the Internet at http://
pubs.acs.org).