Photochemistry of methyl ethyl ketone: Quantum yields and S 1/S0-diradical mechanism of photodissociation

Article abstract of DOI:10.1002/cphc.201000522

Pulsed laser photolysis (PLP) at λ=248 and 308 nm coupled with gas-chromatographic analysis is applied to determine the photodissociation quantum yield (QY) of methyl ethyl ketone (MEK). Temperature dependent UV absorption cross-sections [σ_MEK(λ,T)] are also determined. At 308 nm, the QY decreases with decreasing temperature (T=323-233 K) and with increasing pressure (P=67-998 mbar synthetic air). Stern-Volmer (SV) analysis of the T and P dependent QYs provides the experimental estimate of E_S1=398±9 kJ mol^-1 (=300±6 nm) for the barrier of the first excited singlet state (S₁). The QY at 248 nm is close to unity and independent of pressure (T=298 K). Theoretical reaction pathways are examined systematically on the basis of CASPT2/6-31+G* calculations. Among three possible pathways, a S₁/S ₀-diradical mechanism, which involves H atom transfer on the S ₁ surface, followed by a nonadiabatic transition at a diradical isomer of MEK, explains the experimental data very well. Therefore, this unusual mechanism, which is not seen in any smaller carbonyl compounds, is proposed as an important pathway for the MEK dissociation. Our study supports the view that both the absorption cross-sections and the QYs of carbonyls have significant temperature dependences that should be taken into account for accurate modelling of atmospheric chemistry.

Full text of DOI:10.1002/cphc.201000522

DOI: 10.1002/cphc.201000522
Photochemistry of Methyl Ethyl Ketone: Quantum Yields
and S1/S0-Diradical Mechanism of Photodissociation
Rebeka Nꢀdasdi,[a] Gꢀbor L. Zꢁgner,[a] Mꢀria Farkas,[a] Sꢀndor Dꢂbꢃ,*[a] Satoshi Maeda,[b, c] and
Keiji Morokuma*[b, d]
Pulsed laser photolysis (PLP) at l=248 and 308 nm coupled
with gas-chromatographic analysis is applied to determine the
photodissociation quantum yield (QY) of methyl ethyl ketone
(MEK). Temperature dependent UV absorption cross-sections
[sMEK(l,T)] are also determined. At 308 nm, the QY decreases
with decreasing temperature (T=323–233 K) and with increas-
ing pressure (P=67-998 mbar synthetic air). Stern–Volmer (SV)
analysis of the T and P dependent QYs provides the experi-
mental estimate of ES1 =398ꢁ9 kJmolꢀ1 (=300ꢁ6 nm) for the
barrier of the first excited singlet state (S1). The QY at 248 nm
is close to unity and independent of pressure (T=298 K). Theo-
retical reaction pathways are examined systematically on the
basis of CASPT2/6-31+G* calculations. Among three possible
pathways, a S1/S0-diradical mechanism, which involves H atom
transfer on the S1 surface, followed by a nonadiabatic transi-
tion at a diradical isomer of MEK, explains the experimental
data very well. Therefore, this unusual mechanism, which is
not seen in any smaller carbonyl compounds, is proposed as
an important pathway for the MEK dissociation. Our study sup-
ports the view that both the absorption cross-sections and the
QYs of carbonyls have significant temperature dependences
that should be taken into account for accurate modelling of at-
mospheric chemistry.
1. Introduction
Methyl ethyl ketone (MEK) or butanone, CH3C(O)CH2CH3,
occurs in significant concentration in the free troposphere.[1] It
has been found in relatively high concentrations in air masses
as varied as urban plumes,[2] coastal marine atmosphere,[3] and
rural mountain air,[4] occasionally amounting up to 13–17% of
the total organic carbon detected.[4] MEK has both direct an-
thropogenic and biogenic sources, but it is mostly produced
from the photo-oxidation of n-butane and >C5 iso-alkanes in
the atmosphere.[1,5] Atmospheric loss of MEK occur via photoly-
sis and reactions with OH radicals.[6,8] While the OH reaction
rate coefficient is well established, little is known about the
photolysis quantum yield (QY) of MEK as a function of wave-
length, l, temperature, T, and pressure, P, which is required for
quantitative atmospheric modelling studies. Similar to other
abundant carbonyl molecules in the atmosphere, most notably
acetone, CH3C(O)CH3,[9,10] the atmospheric degradation of MEK
may provide a significant source for the HOx (OH and HO2) rad-
icals and peroxy acetyl nitrate (PAN), especially in the upper
troposphere.
ods have become available to study such basic questions as
the femtosecond dynamics of the cleavage of the a-CꢀC
bond,[12] the classical Norrish Type I reaction, and to address
such mechanistic features as the conical intersections (CI) and
seams of crossings (SX) of potential energy surfaces (PESs),
which are central for understanding the nonadiabatic mecha-
nism of photoreactions (see reference [13]and references there-
in). The photochemistry of acetone has been by far the most
studied, the literature of which has recently been reviewed.[14]
In contrast, just a few experimental studies have been reported
on MEK,[8,15–17] and we are only aware of one single theoretical
[a] Dr. R. Nꢀdasdi, G. L. Zꢁgner, M. Farkas, Prof. S. Dꢂbꢃ
Institute of Materials and Environmental Chemistry
Chemical Research Center of the Hungarian Academy of Sciences
Pusztaszeri fflt 59-67, 1025 Budapest (Hungary)
Fax: +36-1-438-1147
E-mail: dobe@chemres.hu
[b] Prof. S. Maeda, Prof. K. Morokuma
Fukui Institute for Fundamental Chemistry, Kyoto University
Kyoto 606-8501 (Japan)
The photochemistry and photophysics of aliphatic carbonyls
are probably the most thoroughly studied of any class of poly-
atomic molecules. Noyes, Porter, and Jolley[11] cited more than
100 references in their review paper on the photochemistry of
simple ketones, published as early as 1956. The interest in the
photochemistry of aliphatic aldehydes and ketones has even
increased over the past few decades by the recognition of
their atmospheric importance and since these molecules have
been, and continue to be, excellent models for fundamental
photochemical studies. In recent years, highly sophisticated ex-
perimental techniques and powerful quantum chemical meth-
Fax: +81-75-781-4757
E-mail: morokuma@fukui.kyoto-u.ac.jp
[c] Prof. S. Maeda
The Hakubi Center, Kyoto University
Kyoto 606-8501 (Japan)
[d] Prof. K. Morokuma
Department of Chemistry and
Cherry L. Emerson Centre for Scientific Computation
Emory University
Atlanta, GA 30322 (USA)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cphc.201000522.
ChemPhysChem 2010, 11, 3883 – 3895
ꢄ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3883

S. Dꢂbꢃ, K. Morokuma et al.
study.[12] Herein, we present joint experimental and theoretical
investigations of the photodissociation of MEK.
pulsed laser photolysis (PLP). Details of the experimental tech-
niques and methodology can be found in ref. [20].
Like other aliphatic ketones, MEK has a weak and broad ab-
sorption band between ~220 nm (~544 kJmolꢀ1 excitation
energy) and ~340 nm (~352 kJmolꢀ1), which corresponds to
an n!p* transition from the ground state (S0) to the first elec-
tronically excited singlet state (S1) upon photoexcitation. By
analogy with acetone,[14] transformations of S1 are expected to
be dominated by nonradiative processes. S1 can undergo inter-
system crossing (ISC) to the lowest triplet state (T1) and inter-
nally convert (IC) to the vibrationally excited ground state (S0*).
The role of the S1, T1, and S0* states in product formation is the
focus of our current theoretical study. There are a number of
photolysis channels that are thermochemically accessible,[18]
but acetyl, CH3CO, and ethyl, C2H5, radicals are the dominant
products in the photolysis of MEK (Reaction (R2) below) at and
above ~248 nm.[11,15–17,19] We have performed photolysis ex-
UV Absorption Cross-Section Measurements: Temperature-depen-
dent absorption cross-sections, sMEK(l,T), were determined in a
home-constructed single-path UV/Vis spectrometer. The collimated
beam of a 30 W D2 lamp (Hamamatsu L2196HBSQ) was passed
through a 50.2 cm long jacketed absorption cell before being fo-
cused onto the entrance slit of
a 0.25 m monochromator
(Oriel 77200). The wavelength of the monochromator was calibrat-
ed using a low-pressure mercury lamp (Ultraviolet Products, Pen
Ray) with an accuracy of 0.4 nm. The spectral resolution was
~0.4 nm (0.1 mm entrance slit). The monochromator was thermo-
stated (to 295ꢁ1 K), which was found of crucial importance for
obtaining accurate absorption cross-sections. A photomultiplier
(Thorn EMI 9781B) was attached to the exit slit of the monochro-
mator. The signal from the photomultiplier was fed to a digital
control unit (Oriel, Merlin System), which served as a wide dynamic
range lock-in amplifier and provided control and data acquisition
functions for automatic recording of spectra. The absorption cell
was temperature regulated by flowing heating or cooling fluid
through its jacket (Julabo F81-MV thermostat). The jackets extend-
ed up to the end windows eliminating temperature gradients
along the length of the cell. The windows were purged with dry ni-
trogen to prevent deposition of water vapour and ice crystals in
the low temperature measurements. The temperature over the op-
tical path was constant within ꢁ1 K (errors throughout refer to 2s
of the measurements precision, if not otherwise stated). The ab-
sorption measurements were performed under slow flow condi-
tions. This minimised the concentration depletion of MEK via pho-
tolysis by the analytical light beam and its deposition on the opti-
cal windows of the cell. For measurements at wavelength longer
than 300 nm, a cut-off filter (Schott, WG-295) was placed between
the light source and the absorption cell to prevent the photolytic
formation of diketones (e.g. biacetyl), which absorb strongly at
longer wavelengths.
periments at l=248 nm (~482 kJmolꢀ1
(~388 kJmolꢀ1) [Reactions (R1)–(R4)]:
)
and 308 nm
CH3CðOÞC2H5 þ hn
! CH3 þ C2H5CO D1H298 ¼ 353 kJ molꢀ1, lthreshold ¼ 339 nm
ðR1Þ
! C2H5 þ CH3CO D2H298 ¼ 349 kJ molꢀ1, lthreshold ¼ 342 nm
ðR2Þ
! CH3 þ C2H5 þ CO D3H298 ¼ 395 kJ molꢀ1
! CH3CHO þ C2H4 D4H298 ¼ 125 kJ molꢀ1
ðR3Þ
ðR4Þ
Herein we report experimental QYs and the photodissocia-
tion mechanism of MEK from theory for photolysis wave-
lengths relevant to the free and upper troposphere. In the
course of the experiments, the concentration depletion of
CH3C(O)C2H5 was measured in the photolysed samples, but no
product channels were quantified. That is, we have determined
the total primary QY of the photodissociation, FMEK [Reac-
tions (R1)–(R4)]:
Absorption cross-sections were determined at each wavelength by
applying the Beer–Lambert law [Eq. (1)]:
lnfðI0Þ=ðIÞg ¼ A ¼ sMEKðl,TÞ‘½MEKꢂ
ð1Þ
where I0 and I are the transmitted light intensities in the absence
and presence of MEK, respectively, A is the absorbance, sMEK(l,T) is
the absorption cross-section [cm2 moleculeꢀ1] at wavelength l, and
temperature T, ‘ is the path length [cm] and [MEK] is the concen-
tration [moleculecmꢀ3]. Cross-section determinations were carried
out over a wide range using at least ten different concentrations at
each wavelength, and different total pressures, using premixed
MEK/N2 gas mixtures (5–10%). Gas mixtures were prepared mano-
metrically by using a Bourdon-type pressure meter (Texas Instru-
ments M-145).
CH3CðOÞC2H5 þ hn ! Products
ðR1--R4Þ
FMEK was measured in synthetic air over a range of tempera-
tures and pressures. In the theoretical part of this study, local
minimum (MIN), transition state (TS), and minimum on seam of
crossing (MSX) structures were explored systematically as criti-
cal points of adiabatic and nonadiabatic dissociation pathways
on the basis of the CASPT2 theory. Three possible pathways
were located theoretically, and the dissociation mechanism of
photolysis at 248 nm and 308 nm is discussed by comparing
their reaction profiles and experimental data.
Photolysis Quantum Yield Measurements: Determination of the
photodissociation QY, FMEK(l,P,T), involved PLP and the off-line
measurement of the concentration depletion of MEK using gas-
chromatographic (GC) analysis. An excimer laser (Lambda Physik,
ComPex 201) provided the pulsed laser light at 248 nm and
308 nm and the repetition rate was 10 Hz. The experimental
method applied is similar to that described by Gierczak et al.[21]
and in our previous study[22] for acetone photolysis. Most of the in-
vestigations were carried out in a 19.8 cm (optical path)ꢅ2.1 cm
(internal diameter) cylindrical quartz cell, which was surrounded by
a thermostatting jacket through which temperature-regulated fluid
could be circulated by means of a thermostat (Julabo F81-MV). The
reaction temperature inside the cell was measured with a retracta-
Experimental Section
The experimental part of our work consisted of two separate, but
related studies: 1) UV absorption cross-section measurements of
MEK and 2) measurements of the photodissociation QY using
3884
www.chemphyschem.org
ꢄ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemPhysChem 2010, 11, 3883 – 3895

Photochemistry of Methyl Ethyl Ketone
ble thermocouple and was found constant at 1 K. The jacketed cell bination of the free radicals that are produced in the primary pho-
was, at each end, mounted with evacuated compartments with Su- tochemical processes. The experiments were conducted in synthet-
prasil windows at the Brewster’s angle. The evacuated joints pre- ic air where O2 converted the primary radical species to peroxy
vented deposition of moisture and formation of ice on the optical radicals that practically did not react with MEK. To investigate the
windows in measurements below ambient conditions. The photol- reliability of the experimental approach and to assess potential sys-
ysis cell was equipped with a GC sampling port, which included a tematic errors, we performed test experiments (at l=308 nm, T=
septum joint and could be evacuated separately. During analysis, 298 K, P=133 mbar, E=30 mJpulseꢀ1, or varied, [MEK]0 =2.1ꢅ
the sampling line was flushed through with the reaction mixture 1016 moleculecmꢀ3, or varied). The tests included the following:
and samples for GC analyses were withdrawn by a gastight syringe. 1) The trapping efficiency of O2 was investigated in N2 buffer gas
The GC analysis was performed on a 30 m HP-5 quartz capillary by varying the oxygen concentration (the total pressure was kept
column at 303 K using flame-ionization detection and N2 as the at 133 mbar). The measured QYs decreased when the partial pres-
carrier gas (HP 5840 A GC).
sure of O2 was less than ~8 mbar indicating the increased impor-
The photolysis was carried out in synthetic air buffer gas (20% tance of the radical recombination reaction CH3CO+CH3CH2!
O2 +80% N2). The reaction mixture usually contained an inert GC CH3C(O)CH2CH3 which led to underestimations of the QY. Based on
internal standard as well, perfluoromethylcyclohexane (c-C7F14), en- this observation, the lowest range in the regular measurements
abling accurate measurement of the concentrations. Gas mixtures was set to 67 mbar air pressure. 2) The measured QYs were the
were prepared using calibrated pressure meters (MKS Baratron, 10 same within the experimental uncertainty when the photolysis
and 1000 mbar pressure heads). The cell was filled with MEK-GC- energy was varied in the range of 2–80 mJpulseꢀ1. 3) The QYs de-
standard-synthetic-air
gas
mixture
containing
~2ꢅ creased slightly when the conversion was greater than ~35% due
1016 MEKmoleculescmꢀ3 and was filled-up to the desired pressure to the build-up of products that interfered with the GC analysis.
with synthetic air. The photolysis laser beam entered the cell along Thus, the conversion was kept below ~25% in the regular meas-
the optical axis and an iris blend of 1.2–1.8 cm i.d. was placed in urements. 4) FMEK has been found invariant to the initial MEK con-
front of the cell. The initial concentration of MEK, [MEK]0, was mea- centration varied in the range of (0.5–5.0)ꢅ1016 moleculecmꢀ3
.
sured by the GC and the sample was photolysed with a measured 5) The accuracy of the energy measurements was tested by using
number, n, of laser pulses. After irradiation, the concentration of NO2 actinometry. The cell was filled with a diluted NO2/He gas mix-
MEK remaining in the mixture, [MEK], was measured by the GC. ture instead of MEK. The NO2 was photolysed and its depletion
This procedure was repeated six to eight times by varying the pho- was monitored by measuring the absorption at 435 nm. Data were
tolysis time, but always starting with a fresh photolysis mixture. analysed employing Equation (2) with NO2 substituting MEK. The
The consumption of MEK was usually in the range of 5–25%. The consumption QY of NO2 is known to be FNO2 =2 in the 300–
laser energy, E(l), required for the calculation of the QY, was mea- 398 nm range[23] and sNO2(308 nm)=1.65ꢅ10ꢀ19 cm2 moleculeꢀ1 [24]
.
sured after the iris blend by using a calibrated laser energy meter The value of E(308 nm) obtained with the NO2 actinometry agreed
(Gentec, OE235SP). E(l) was typically ~30 mJpulseꢀ1 within 4% with that measured by using the laser energy meter.
(~17 mJcmꢀ2 pulseꢀ1 fluence).
This good agreement confirms the correctness of Equation (2).
Considering the low light absorption, the small fractional photoly-
sis, and that the laser energy was measured in front of the en-
trance window of the cell and the laser beam did not fill the cell
completely, Equation (2) can be derived for the photodepletion of
MEK:[20]
Results and Discussion
2.1. UV Absorption Cross-Sections
The absorption spectrum for MEK was determined in the wave-
length and temperature ranges of l=220–350 nm and T=
253–363 K, respectively. The spectrum is presented in Figure 1
and the corresponding absorption cross-sections are listed in
Table SI1 (Supporting Information) in 1 nm intervals.
lnð½MEKꢂn=½MEKꢂ0Þ ¼ꢀn ꢃ EðlÞ ꢃ fw ꢃ EPhðlÞꢀ1
ð2Þ
ꢃFMEKðl,P,TÞ ꢃ sMEKðl,TÞ ꢃ ‘ ꢃ Vꢀ1
where fw is the transmittance of the entrance window (a typical
value was fw =0.920 for one window), EPh(l) is the energy of one
photon (mJphotonꢀ1), ‘ is the optical path length (cm) and V is
the total volume of the cell (cm3). Plots of ln([MEK]n/[MEK]0) vs nꢅE
yielded straight lines in all experiments. The FMEK(l,P,T) results
were obtained from linear least-squares slopes by making use of
the temperature-dependent cross-sections determined herein. Rep-
resentative decays plotted according to Equation (2) are presented
in Figures 2,3.
Materials: MEK (Sigma–Aldrich,>99.7%) and c-C7F14 (Fluka, 97%)
were degassed by several freeze–pump–thaw cycles prior to use.
NO2 (Messer–Griesheim, 98%) was purified by repeated trap-to-
trap distillations in vacuum until the colour of the sample frozen in
N2(l) became bright white. Nitrogen (>99.9%), oxygen (>99.9%),
and helium (>99.9%), all from Messer–Griesheim, were used as ob-
tained.
Optimisation and Consistency Tests: The photodissociation QY was
determined by monitoring the loss rate of MEK. That is, the meth-
odology implies the assumption that the photolyte is not con-
sumed in any secondary reactions and is not reformed via recom-
Figure 1. UV absorption spectrum of MEK at different temperatures.
ChemPhysChem 2010, 11, 3883 – 3895
ꢄ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemphyschem.org 3885

S. Dꢂbꢃ, K. Morokuma et al.
The absorption spectrum of MEK displays the characteristic
spectral feature of aliphatic aldehydes and ketones with the
wide np* transition in the near UV region.[11] The absorption
maximum is at lmax(MEK)=278 nm (T=298 K), which displays a
few nanometer red-shift compared to the spectrum of ace-
tone, lmax(acetone)=274 nm.[21] Our results show a definite
temperature dependence of the absorption cross-sections. As
seen in Figure 1, the absorption curve before the maximum at
shorter wavelengths has only a smaller temperature depend-
ence. After the maximum, the temperature dependence is
more pronounced; the absorption cross-sections show a sub-
stantial increase with increasing temperature. The maximum of
the spectrum shifts towards longer wavelengths with increas-
ing temperature, but the effect is small, ~2 nm, on going from
253 to 363 K. The results below ~230 nm, above ~320 nm, and
in particular at the lowest temperature (253 K) are considered
less reliable. Under these conditions, the light absorption was
small and the measured absorbance values became increasing-
ly scattered in the Beer–Lambert plots.
Equation (3) describes the measurements within an average
absolute error of 7% in the 253–363 K temperature and
230–320 nm wavelength ranges. However, the recommended
parameterised absorption cross-section is valid only in the
given T and l ranges and extrapolation outside this region
could lead to erroneous results.
The room-temperature UV absorption spectrum of MEK has
been reported in several publications, which have recently
been evaluated by IUPAC.[6] The proposed 298 K cross-sec-
tions[6] are based on the data reported by Martinez et al.[25]
who applied two-beam spectrophotometry. The cross-sections
proposed by IUPAC[6] are in excellent agreement (within 3%)
with our present determination between 230 and 320 nm. We
are only aware of one temperature-dependent study published
by Hynes and coworkers.[26] These authors also observed an in-
crease of the absorption cross-section with increasing tempera-
ture, but because they only presented sMEK(l,T) in a graphical
form, no quantitative comparison to our data can be made.
The temperature dependence of the absorption cross-sec-
tions is presented at 248, 278, and 308 nm in Table 1. We mea-
sured the temperature dependence of the QY of MEK at
308 nm, where the temperature dependence of the absorption
cross-section is significant; increasing the temperature from
253 to 263 K, increases sMEK by ~60%. Clearly, the temperature
dependence of the cross-section should be taken into account
in deriving accurate QYs for the photodissociation of
CH3C(O)CH2CH3.
2.2. Experimental Quantum Yield Results
The photodissociation QY for CH3C(O)CH2CH3 was determined
at two photolysis wavelengths, 308 nm and 248 nm, in syn-
thetic air buffer gas. At 308 nm, temperature and pressure de-
pendency studies were performed over the T and P ranges of
233–323 K and 67–998 mbar, respectively. At 248 nm, the QY
was measured at room temperature and two pressures, 133
and 998 mbar. As described in detail in the Experimental Sec-
tion, gas mixtures containing MEK were irradiated with an exci-
mer laser and the concentration depletion was measured by
GC analysis. QYs were obtained by plots of the experimental
observables according to Equation (2).
Table 1. Temperature dependence of the absorption cross-section of
MEK at selected wavelengths in 10ꢀ20 cm2 moleculeꢀ1
.
l [nm]
253 K
ꢁ2s sMEK
273 K
ꢁ2s sMEK
283 K
ꢁ2s sMEK
298 K
ꢁ2s
sMEK
2.2.1. FMEK (308 nm)
248
278
308
2.050 0.334 2.392 0.072 2.399 0.034 2.246 0.034
5.754 0.266 5.781 0.046 5.826 0.040 5.807 0.024
1.459 0.534 1.688 0.030 1.795 0.032 1.885 0.056
Representative semi-logarithmic decay plots used to determine
the photolysis QY of MEK at 308 nm are presented in Figure 2.
(T=298 K). The data points form straight lines, the slopes of
which decrease with increasing pressure indicating a substan-
tial pressure dependence of the QY. A similar picture is seen in
experiments performed at other temperatures as well. At the
lowest temperature, T=233 K, the decay plots show straight
lines with larger scattering of the data, likely due to the in-
creased adsorption of MEK on the cell windows (all QY deter-
minations were carried out under static conditions). The upper
end of the temperature range (323 K) was limited by the onset
of interfering hydrogen abstraction reactions, which might
have caused an overestimation of the measured consumption
QYs.
l [nm]
323 K
ꢁ2s sMEK
2.250 0.002 2.261 0.074 2.289 0.002 2.206 0.002
5.910 0.001 6.029 0.082 6.074 0.002 6.110 0.002
2.031 0.002 2.125 0.008 2.131 0.002 2.334 0.002
333 K
ꢁ2s sMEK
343 K
ꢁ2s sMEK
363 K
ꢁ2s
sMEK
248
278
308
We fitted an asymmetric double sigmoidal function to the
data points in the wavelength range 230–320 nm. In this way,
Equation (3) was returned from the parameter estimations for
the temperature and wavelength dependence of the absorp-
tion cross-section of MEK (the units of T, l, and sMEK are K, nm,
and cm2 moleculeꢀ1, respectively):
The 308 nm photodissocia-
tion QYs for CH3C(O)CH2CH3 at
sMEKðl; TÞ¼ ðꢀ1:488 ꢃ 10ꢀ20 þ 3:6069 ꢃ 10ꢀ23TÞ
different temperatures and
0
1
pressures are summarised in
8:6674 ꢃ 10ꢀ20
1
Table 2. The FMEK values de-
@
A
h
i
h
i
þ
1 ꢀ
lꢀð270:5976þ0:0215TÞþð12:7414þ0:0223TÞ
12:2062
lꢀð270:5976þ0:0215TÞþð12:7414þ0:0223TÞ
crease significantly with de-
creasing temperature and also
with increasing pressure. As for
1 þ exp
1 þ exp ꢀ
10:5243
ð3Þ
3886
www.chemphyschem.org
ꢄ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemPhysChem 2010, 11, 3883 – 3895

Photochemistry of Methyl Ethyl Ketone
Figure 2. Consumption of MEK in 308 nm photolysis experiments at room
temperature. n is the number of laser shots and E is the energy per pulse.
The slopes are proportional to the photodissociation quantum yields
[Eq. (2)].
Figure 3. Consumption of MEK in 248 nm photolysis experiments at room
temperature. n is the number of laser shots and E is the energy per pulse.
The slopes are proportional to the photodissociation quantum yields
[Eq. (2)].
instance, at 133 mbar air pressure, the QY decreases to half
when decreasing the temperature from 323 to 233 K. Further-
more, at 253 K the atmospheric pressure QY is just ~22% of
that measured at 67 mbar. To our knowledge, this is the first
report on the T and P dependence of the photodissociation QY
of MEK in air buffer gas under conditions relevant to the at-
mosphere.
stood by the primary formation of acetyl and ethyl radicals in
the photodissociation and the subsequent secondary reactions
of peroxy radicals.
2.2.3. Error Analysis
The overall uncertainty in FMEK was estimated by the observed
statistical errors and estimated systematic uncertainty for the
308 nm results. The average of the statistical errors of the QYs
listed in Table 2 is ꢁ12% (2s). Systematic errors were assessed
for the parameters used in Equation (2), for example, it is ꢁ3%
for [MEK]0 judged by calibrations of the pressure meters, and
by considering the results of test experiments outlined in the
Experimental Section. The potential systematic error of the
measurements is estimated to be ꢁ6% (2s). Combination of
the root mean squares of the statistical and systematic errors
gives ꢁ15%, which is the proposed overall uncertainty in
FMEK(308 nm) at the 2s (95% confidence) level.
2.2.2 FMEK(248 nm)
Experimental results measured for CH3C(O)CH2CH3 depletion at
248 nm are presented in Figure 3. The following QYs were de-
termined at 298 K in synthetic air at two pressures:
FMEK(248 nm, 133 mbar)=0.937ꢁ0.200 and FMEK(248 nm,
998 mbar)=0.953ꢁ0.094. That is, the FMEK values are close to
unity and are independent of bath gas pressure within experi-
mental uncertainty at 248 nm. By analogy to our results on
acetone,[22] unity QYs were anticipated and the slightly smaller-
than-unity yields by ~6 and ~5% may indicate minor system-
atic error of unknown origin occurring at this wavelength.
The stable reaction products CH3OH, CH3CHO, and C2H5OH
were identified in the photolysed samples (at both 248 and
308 nm), which is in accordance with the results of the photo-
oxidation study of MEK by Raber and Moortgat,[15] and under-
2.3. Stern–Volmer Analysis of the Experimental Quantum
Yields
The 308 nm T and P dependent photodissociation QYs are pre-
sented in Stern–Volmer (SV) plots in Figure 4, where the recip-
Table 2. Temperature and pressure dependence of the photodissociation quantum yield of MEK at 308 nm in synthetic air. (The errors represent 2s preci-
sion. In parenthesis, the number of experimental measurements are given.)
P(298 K) [mbar]
T [K]
323
298
273
253
233
67
–
0.962ꢁ0.124 (8)
0.817ꢁ0.002 (63)
0.667ꢁ0.038 (6)
0.586ꢁ0.002 (19)
0.408ꢁ0.020 (7)
0.410ꢁ0.002 (12)
0.337ꢁ0.046 (6)
0.294ꢁ0.034 (7)
0.738ꢁ0.036 (7)
0.518ꢁ0.002 (12)
0.458ꢁ0.054 (7)
0.374ꢁ0.074 (5)
0.280ꢁ0.052 (5)
0.243ꢁ0.002 (8)
0.205ꢁ0.024 (5)
0.174ꢁ0.002 (14)
0.693ꢁ0.092 (6)
0.463ꢁ0.084 (6)
0.366ꢁ0.104 (5)
0.270ꢁ0.002 (12)
0.201ꢁ0.002 (11)
0.232ꢁ0.014 (5)
0.191ꢁ0.002 (14)
0.155ꢁ0.002 (15)
0.578ꢁ0.058 (21)
133
266
400
532
665
798
998
0.831ꢁ0.002 (31)
0.721ꢁ0.010 (13)
0.580ꢁ0.284 (7)
0.509ꢁ0.070 (7)
0.468ꢁ0.170 (8)
0.384ꢁ0.090 (6)
0.336ꢁ0.060 (9)
0.416ꢁ0.006 (15)
0.320ꢁ0.078 (6)
0.222ꢁ0.022 (21)
0.122ꢁ0.044 (6)
–
–
–
ChemPhysChem 2010, 11, 3883 – 3895
ꢄ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemphyschem.org
3887

S. Dꢂbꢃ, K. Morokuma et al.
ꢀ1
ꢀ1
ð4Þ
F
¼ 1 þ ðkQkIC Þ ꢃ ½Mꢂ
MEK
Implicit in the above mechanism is the assumption that the
rate of the collisional desactivation of S*0 is negligible com-
pared with its conversion to products. If the pressure quench-
ing of S*0 were significant, a nonlinear dependence of FꢀM1EK on
[M] would be expected in the SV plots. There is no obvious de-
viation from linearity, but minor quenching of S*0 cannot be
excluded by the data. The mechanism does not include ISC,
S1!T1, since there was no indication for the occurrence of this
process from the experiments (note that the experiments were
performed in synthetic air, and that O2 is known to be very effi-
cient quencher of triplet carbonyls). It is also clear, that the
direct S1 mechanism, the one that does not involve the inter-
mediate S0*, provides the same SV equation as Equation (4).
The indirect S1/S0-diradical mechanism has been preferred by
arguments given in Sections 3.3 and 4.
Figure 4. Photodissociation quantum yield of MEK in Stern–Volmer represen-
tation at 308 nm.
ꢀ1
The kQkIC data determined by the SV plots of Figure 4 are
listed in Table SI2. The rate coefficient ratios decrease by a
factor of ~6 when increasing the temperature between 233
and 323 K. It is reasonable to assume that the temperature de-
pendence is due to the increase of kIC with temperature, since
quenching is expected to have only a slight temperature de-
pendence, if any at all, in the relatively small temperature
range investigated. That is, there is experimental evidence that
the photodissociation occurs through an energy barrier. The
height of the energy barrier, ES1, has been estimated by an em-
pirical Arrhenius expression [Eq. (5)] proposed by Blitz et al.:[27]
rocal of the QY of MEK is plotted vs. the number density of the
buffer gas air, [M]. Weighted least-squares fitting was applied
to the data, which obey reasonably well straight lines over the
whole temperature and pressure ranges. Unity intercepts were
returned from the fitting procedure at 323 and 298 K (within
10%). The intercepts became higher at lower temperatures, in-
creasing up to 1.6 (T=253 K), but showed no systematic varia-
tion with decreasing temperature. No mechanistic interpreta-
tion of this observation is warranted because of the relatively
large experimental uncertainties of the low temperature ex-
periments.
ꢂꢀ
ꢁ
ꢃ
kQ
kIC
hc hc
1
kT
¼ AS1 exp
ꢀ
ð5Þ
The observed linear SV plots and their temperature depend-
ences are consistent with (but, of course, do not prove) the in-
direct S1/S0-diradical mechanism we propose by our quantum
chemical computations as the most probable mechanism for
the photodissociation of MEK at 308 nm (Sections 3.3 and 4).
This mechanism is given in the simplified form below to inter-
pret the experimental findings [Reactions (R5)–(R8)]:
ES1
l
where AS1 is a constant; h=6.621ꢅ10ꢀ34 Js is Planck’s con-
stant; c=2.998ꢅ108 msꢀ1 is the velocity of light; k=1.381ꢅ
10ꢀ23 JKꢀ1, Boltzmann’s constant, and l=3.08ꢅ10ꢀ7 m is the
photolysis wavelength in the present investigation.
Equation (5) was fitted to the experimentally determined
ꢀ1
kQkIC values using a (1/si2)-weighted nonlinear fitting proce-
S0 þ hn ! S1
*
S1 ! S
0
ðR5Þ
ðR6Þ
ðR7Þ
ðR8Þ
dure. The results are presented in Figure 5. In the Arrhenius
plot, the ln(kQkICꢀ1) values plotted against the reciprocal of the
kIC
*
S
! Products
kP
kQ
0
S1 þ M ! S0 þ M
In the sketched mechanism, S0 and S1 are the ground and
first excited singlet states, designated in the detailed molecular
mechanism as S0-MIN1 and S1-FC, respectively. kIC, kP and kQ are
rate coefficients of the respective elementary steps. S*0 is a vi-
brationally–rotationally excited CH3C(O)CH2CH3 molecule,
which undergoes fast decomposition to products (preferably
CH3CO+C2H5) on the ground PES. The rate-limiting step is the
IC that occurs through a reaction barrier (through S1-TS3 in the
detailed mechanism). Pressure dependence is a consequence
of the competition between unimolecular decomposition and
collisional quenching (Q) of S1. The mechanism predicts the
simple SV expression [Eq. (4)]:
ꢀ1
Figure 5. Arrhenius plot for the rate coefficient ratio kQkIC
.
3888
www.chemphyschem.org
ꢄ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemPhysChem 2010, 11, 3883 – 3895

Photochemistry of Methyl Ethyl Ketone
absolute temperature define a straight line. The parameter es-
timation has provided the following barrier for the photodisso-
ciation of MEK by our experiments: ES1 =300ꢁ6 nm=398ꢁ
9 kJmolꢀ1 where the quoted uncertainty is given at the 95%
confidence level and includes estimated systematic errors. As
discussed,[27] the use of Equation (5) is only a first approxima-
tion to derive the barrier for photodissociation by QY data
mainly because the S1-FC state is not a Boltzman distribution,
but the ground state population is distorted to an unknown
extent by Franck–Condon factors. Therefore, the barrier esti-
mated by the experiments may be significantly more inaccu-
rate than reflected by the given error margins.
from S1 to T1 is very slow; which should be the reason for the
large pressure dependence of the T1 mechanism. The ultrafast
dissociation on the S1 surface explains the small pressure de-
pendence of the S1 mechanism, while it is minor at low pres-
sures because of a high barrier on the S1 surface (further com-
parison of the mechanism of the photodissociation of acetone
and MEK is presented in Sections 3.3 and 4).
2.4. Comparison with Previous Quantum Yield
Measurements
Relatively few QY determinations for MEK have been published
to which our results can be compared. The 2005 IUPAC[6] evalu-
ation only reports the photolysis study by Raber and Moort-
gat.[15] Raber and Moortgat determined the QY in synthetic air
for wavelengths >275 nm using broad-band photolysis lamps.
The consumption of MEK and formation of stable products
were measured by infrared absorption. The photolysis QY was
found to depend on pressure, decreasing from 0.89 at 68 mbar
to 0.34 at 1000 mbar. Although Raber and Moortgat[15] used
polychromatic light, the agreement with our current determi-
nation done with 308 nm laser light, namely FMEK(67 mbar)=
0.962 and FMEK(998 mbar)=0.294 (T=298 K) is surprisingly
good.
We have measured QYs close to unity and independent of
pressure for the 248 nm photolysis of MEK indicating the very
fast dissociation of the photoexcited molecule. At this wave-
length, the excitation energy is significantly higher
(~482 kJmolꢀ1) than the S1 barriers estimated by the 308 nm
experiments and ab initio computations, and so, besides the
S1/S0-diradical mechanism, the higher energy S1-direct mecha-
nism (Section 4) may become operative as well.
In a previous study, we applied the same experimental ap-
proach as herein to study the photodissociation of acetone.[22]
The 308 nm photodissociation QYs of both acetone and MEK
have positive T and negative P dependences, but there are im-
portant differences as well. In Figure 6 the 298 K photodissoci-
More recently, QYs for acetyl radical production, FCH3CO, from
MEK photolysis as a function of temperature (T=198–295 K)
and pressure (Pꢄ10–600 mbar) were determined by Baeza-
Romero et al.[8] at the University of Leeds using 308 and
320 nm PLP. Most of the QY measurements were carried out at
308 nm. He containing O2 at a low concentration was the
buffer gas. The authors measured CH3CO yields via pulsed
laser induced (LIF) fluorescence monitoring of OH, which is a
minor product of the CH3CO+O2 reaction, providing a sensi-
tive and selective tracer for the acetyl radical.[30]
Similarly to our findings, the Leeds group has established
the QY of MEK photodissociation to have a positive tempera-
ture dependence and a negative dependence on the pressure
of the buffer gas (note that at the relatively long wavelength
of 308 nm, FCH3CO =FMEK). While the QYs reported by Baeza-
Romero and coworkers[8] are on average only ~20% higher
than proposed herein, which may appear to be a reasonable
agreement. More quantitative comparison of the results is not
feasible because of a range of reasons: 1) In the study of
Baeza-Romero and coworkers, the yield of CH3CO at 248 nm
photolysis was used as a reference and in their data analysis it
was assumed to be independent of pressure. In contrast to
this assumption, the most recent experimental studies have re-
vealed FCH3CO to be pressure dependent at 248 nm[16,17] (see
below). 2) It is likely that the Leeds group used room-tempera-
ture absorption cross-sections of MEK for deriving the QYs at
all temperatures (295–198 K) in lack of reported temperature-
dependent sMEK data. Note that we found sMEK to decrease sig-
nificantly as the temperature was decreased (Section 2.1).
3) Baeza-Romero et al.[8] carried out the measurements in He
buffer gas, while we performed our experiments in synthetic
air, and there is no straightforward way to correct the QYs for
the different quenching efficiencies of the buffer gases.
Figure 6. Comparison of the photodissociation quantum yields for ace-
tone[22] and MEK in Stern–Volmer plots (T=298 K, l=308 nm).
ation QYs for acetone and MEK are compared in SV plots. The
acetone data were taken from our previous study.[22] The QYs
of acetone are significantly smaller than those of MEK, for ex-
ample, at 133 mbar air pressure, FAcetone =0.319, while at the
same pressure FMEK =0.817. A characteristic difference is also
that the SV plot of acetone is curved, while that of MEK dis-
plays linearity. The nonlinear behaviour of acetone was inter-
preted by proposing the product formation to occur from
more than one electronically excited state, namely both from
the S1 and T1 states with the latter being more efficiently
quenched by pressure.[22,27] This interpretation may, however,
need refinement in view of recent results from experiment[28]
and theory.[29] The theoretical study[29] showed that the ISC
ChemPhysChem 2010, 11, 3883 – 3895
ꢄ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemphyschem.org
3889

S. Dꢂbꢃ, K. Morokuma et al.
In 2007, Khamaganov et al.[16] reported the QY of CH3 forma-
tion, FCH3, in 248 nm PLP of MEK at 298 K and at pressures of
~7–2000 mbar N2. CH3 was detected by transient UV absorp-
tion spectroscopy. Pressure-dependent CH3 QYs were observed
concurrently with the increase in the CH3CO QYs indicating
that the change in CH3 yield was a result of the competition
between decomposition and collisional relaxation of the vibra-
tionally excited CH3CO radical formed in the primary photolysis
step. Khamaganov and coworkers found the pressure-depen-
dent FCH3 data to be well described by a modified exponential
function, the extrapolation to infinitely large pressure provided
F1CH3 =0.19ꢁ0.03=a, proposed as the branching ratio for
Reaction (R1) [a=F1/(F1+F2)].
Recently, Rajakumar et al.[17] reported measurements of the
CH3CO QY following 248 nm PLP of CH3C(O)CH2CH3 at 296 K in
He and N2 buffer gases at total pressures of 80–890 mbar.
CH3CO was detected using cavity ring-down absorption spec-
troscopy. The CH3CO QYs were found to increase with pressure
and to depend on the identity of the bath gas, increasing with
collision efficiency [FCH3CO (N2)>FCH3CO (He)]. The CH3CO QY
extrapolated to zero pressure was F8CH3CO =0.41ꢁ0.08. At the
highest N2 pressure, FCH3CO(890 mbar N2, 248 nm)ꢄ0.9 was
measured,[17] which is consistent with our determination of the
primary photodissociation QY of FMEK(998 mbar air, 248 nm)=
0.953ꢁ0.094 and indicates that Reaction (R2) is the dominant
primary photodissociation process.
plored by Liu et al. at the CASSCF level.[52] A minimum on CI
structure between the S1 and S0 surfaces was discovered by
Diau et al. at the CASSCF level .[44] The S0, S1, and S2 PESs and
their intersections were studied by Antol et al. at the CASPT2
and MRCI levels.[53] Dissociation mechanisms involving the T1
state were revisited by Maeda et al. at the CASPT2 level.[29] Re-
cently, a dissociation channel called “roaming channel”, known
in formaldehyde and acetaldehyde photodissociation,[54–56] has
been proposed to take place on the S0 PES of acetone.[57] Four
mechanisms generating the S0 species before the roaming
channel by either S1!T1!S0 or direct S1!S0 transitions were
systematically explored in a recent theoretical study.[29] In spite
of these extensive studies on the smallest aldehydes and ke-
tones, only one theoretical study on the photodissociation of
MEK has been reported. Only the barriers of the CꢀC bond
cleavage on the S1 PES were examined by Diau et al. on the
basis of the CASSCF and TD-DFT calculations.[12] Although PESs
and their intersections of MEK might be very similar to those
of acetone, this point has never been examined systematically
hitherto.
3.2. Theoretical Method
Critical points, that is, MIN, TS, and MSX, on the S0, S1, and T1
PESs of MEK were optimized at the CASPT2/6-31+G* level
with an eight electrons and seven orbitals active space (8e,7o).
This (8e,7o) active space includes two s and two s* orbitals in
addition to p, n, and p* orbitals. The multi-state CASPT2[58] and
the single-state CASPT2[59] were employed in calculations of
the singlet states and the triplet state, respectively. The shift
parameter 0.3 was applied in the CASPT2 calculations to avoid
the intruder state problem.[60] The term “seam of crossing” in-
cludes 3N-8 dimensional CI between states with the same spin
and space symmetry as well as 3Nꢀ7 dimensional seam of
crossing between states of different spin or space symmetry.
Energy values and gradient vectors of the CASPT2 method
were computed by the MOLPRO2006 program.[61] By using
these quantities, critical points were optimized by the GRRM
program.[62–64] TS structures were located by the d-ADDF
method[65,66] in the GRRM program. MSX structures were
searched by a combination of the model-function[41] method
and the d-ADDF method. MSX structures were optimized using
a combination of the gradient projection[67,68] method and the
rational function optimization[69] method, in which the branch-
ing-plane updating method[70] was employed to avoid compu-
tations of nonadiabatic coupling derivative vectors (CDVs). It
should be noted that our MSX optimizer without use of CDVs
gives the same optimized structures as those optimized with a
method using CDVs within a given optimization threshold as
demonstrated in reference [70].
3. Theoretical Study
3.1. Recent Theoretical Studies on the Photochemistry of
Carbonyls
The photodissociation mechanism of the smallest aldehyde,
formaldehyde, has been studied extensively by theoretical cal-
culations. Since its decomposition had been postulated to take
place on the ground state PES after an S1 to S0 nonadiabatic
transition,[31] exploration for dynamics[32–34] as well as stationary
points[35–38] on the S0 PES have been the subjects of most theo-
retical studies. In the last two years, the nonadiabatic decay
mechanisms were revealed in detail by Morokuma et al.[13,39–41]
and Robb et al.[42,43] A new T1/S0 crossing seam has recently
been discovered inside the potential well of formaldehyde,[13,41]
which has provided an answer to a long-standing mystery in
the low energy photolysis how the S0 species can be generat-
ed inside the potential well before the S0 dynamics. Separately,
S1/S0 and T1/S0 intersections with partially dissociated geome-
tries (H···HCO) have been discovered,[42,43] and demonstrated to
be important in the photolysis where the photon energy is
higher than the barriers of the CꢀH bond cleavage on the S1
and T1 PESs, respectively. Another higher energy channel
through a potential well of hydroxycarbene has also been lo-
cated.[39,40] In contrast to formaldehyde, the CꢀC bond cleav-
age in acetone takes place mainly on excited-state PESs; occur-
ring on the S1 PES if the photon energy is much higher than
the barrier on the S1 PES,[12,44–48] while it happens on the T1 PES
after the S1 to T1 nonadiabatic transition at low photon ener-
gies.[49–51] Relevant excited-state stationary structures were ex-
3.3. Features of the S0, S1 and T1 Potential Energy Surfaces
Optimized (8e,7o)-CASPT2/6-31+G* geometries are shown in
Figure 7, in which one MIN on the S0 PES (MIN1), one MIN on
the S1 PES (MIN2), and two MINs on the T1 PES (MIN3 and
MIN4) are presented. Eight TS structures were located on S1
3890
www.chemphyschem.org
ꢄ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemPhysChem 2010, 11, 3883 – 3895

Photochemistry of Methyl Ethyl Ketone
Figure 7. Optimized MIN, TS, and MSX structures for MEK at the (8e,7o)-CASPT2/6-31+G* level.
and T1 PESs; dissociation channels (DCs) of the methyl group
(TS1 and TS6), DCs of the ethyl group (TS2 and TS7), and alkyl
to oxygen H transfer channels (TS3-5 and TS8). Among them,
only TS1 and TS2 have been reported in the literature on the
basis of CASSCF and TD-DFT calculations.[12] Figure 7 shows
three MSXs between S1 and S0 (MSX1-3), two MSXs between T1
and S0 (MSX4 and 5), and one MSX between S1 and T1 (MSX6).
acetone. A key step is the S1!T1 transition. For acetone, al-
though it was previously proposed on the basis of CASSCF cal-
culations that a low energy S1/T1-MSX exists and that the S1!
T1 takes place quickly through the MSX,[52] a different conclu-
sion has been proposed very recently, based on the more relia-
ble CASPT2 calculations, that the S1/T1-MSX is higher in energy
than both the S1 dissociation barrier and the T1 barrier,[29]
where the latter calculations better explain (especially recent)
experimental observations. Similarly to acetone, in the present
case of MEK, there is no low-energy crossing seam between S1
and T1 PESs. The lowest crossing point (MSX6) is located at a
high-energy region in Figure 8. Therefore, the S1-to-T1 transi-
tion at low photon energy is possible only by trickling down
from S1 to T1 while the molecule in the S1 state spends a long
MSX1 and MSX2 are similar to
a
S1/S0-MSX of
acetone.[29,44,53]Figure 8 summarizes the energy profiles on S0,
S1, and T1 PESs. As shown in the supporting information,
energy changes due to conformation rearrangements are very
small (only a few kJmolꢀ1) on excited state PESs.
At first, we discuss DCs on the T1 PES (TS6 and 7) after an
S1!T1 transition. A similar DC on the T1 PES is also known for
Figure 8. Potential energy profiles of the S0, S1, and T1 potential energy surfaces of MEK at the (8e,7o)-CASPT2/6-31+G* level.
ChemPhysChem 2010, 11, 3883 – 3895
ꢄ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemphyschem.org
3891

S. Dꢂbꢃ, K. Morokuma et al.
time oscillating on the S1 surface. This may be possible be-
cause the two PESs, although without crossing, are very close
and nearly parallel to each other in energy at any place in the
basin of S1-MIN2. The transition probability at each geometry
should be small because the spin–orbital coupling between
the two states belonging to the same electronic configuration
is expected to be small. Therefore, this trickling transition
mechanism should be very slow. This is a common S1!T1
mechanism among formaldehyde,[13,41] acetone,[29] and MEK.
Such slow pathways are overtaken by more efficient S1 path-
ways for higher photon energy.
switch from the T1 or S0 mechanisms to the S1 mechanism de-
creases the branching ratio of the methyl radical.
4. Mechanism of Photodissociation
As discussed above, we have three possible channels: 1) the S1
! T1 trickling transition followed by the CꢀC bond cleavages
(TS7 and 6) on the T1 PES (denoted as T1 pathway) to produce
C2H5 +CH3CO and CH3 +C2H5CO, respectively, 2) the excited-
state H atom transfer (TS3) on S1 followed by the S1!S0 transi-
tion (MSX3), regeneration of the ground state MEK, and the
CꢀC bond cleavages on the S0 PES (denoted as S1/S0-diradical
pathway) also to yield C2H5 +CH3CO and CH3 +C2H5CO, and
3) the direct CꢀC bond cleavages (TS1 and 2) on the S1 PES
(denoted as S1 pathway), which also produces C2H5 +CH3CO
and CH3 +C2H5CO.
At the 248 nm (~482 kJmolꢀ1) photolysis, all these pathways
are energetically open even at 0 K. As discussed above and in
ref. [29] for acetone, the T1 pathway can be neglected when
the photon energy is higher than the barriers on the S1 surface
because the trickling transition without explicit crossing is very
slow. Of the other two, the S1/S0-diradical pathway is energeti-
cally slightly the more favourable, which is enhanced when the
vibrational zero-point energy (ZPE) and tunnelling corrections
are taken into account. Table 3 shows harmonic ZPE correc-
At higher photon energy, two very different channels, that
is, CꢀC bond cleavages and an H atom transfer on the S1 PES,
open. Here, our computations show that the H atom transfer
via TS3 is preferred by 5.7 kJmolꢀ1 (in potential energy) over
the dissociations via TS2 (or TS1). Although a similar H atom
transfer on the S1 PES of acetone, that is, an excited state
keto–enol isomerisation, exists as a minor channel, its barrier is
30.6 kJmolꢀ1 higher than the TS of methyl dissociation.[29] Simi-
larly, in MEK, two TSs of excited state keto–enol isomerisations,
that is, for the H atom transfer channel from the methyl group
of MEK via TS4 and from the CH2 (in ethyl group) of MEK
through TS5, were located at 487.9 kJmolꢀ1 and at
476.5 kJmolꢀ1, respectively. TS3 for H atom transfer from CH3
in the ethyl group is most stable among these three TSs for
possible H atom transfer reactions. After going over TS3, the
system can come down to the S0 PES via S1/S0-MSX3. Starting
from the S1/S0-MSX3 geometry, a geometry optimization direct-
ly converges to S0-MIN1. Although we found a shallow local
minimum (an open-shell singlet species at 335.9 kJmolꢀ1) close
to the geometry of S1/S0-MSX3 on the PES of the UB3LYP/6-
31+G* method, the barrier from the local minimum to MEK is
only 4.3 kJmolꢀ1. Therefore, regeneration of the S0 state MEK
mainly takes place after the S1!S0 transition because of the
downhill potential from S1/S0-MSX3 to S0-MIN1. From S1/S0-
MSX3, a DC generating C2H4 +CH3COH might exist as a minor
channel since the corresponding dissociation energy
(362.7 kJmolꢀ1 relative to MIN1) is only slightly above S1/S0-
MSX3. In the dissociation reactions on S0 the ethyl radical dis-
sociation is preferred by 2.9 kJmolꢀ1 compared to the methyl
group dissociation. Since the ethyl dissociation is preferred by
12.4 kJmolꢀ1 on the T1 pathway, if a switch from the T1 mecha-
nism to the S0 mechanism takes place, the branching ratio of
methyl radical increases.
Table 3. Barrier height estimations of TS2 and TS3 with the harmonic
zero-point-energy corrections and the Wigner tunneling corrections.
MIN1
MIN2
TS2
TS3
Obsd.
E [kJmolꢀ1
ZPE [kJmolꢀ1
E+ZPE [kJmolꢀ1
w°[a] [cmꢀ1
]
0.0
301.7
0.0
–
–
–
–
–
358.1
297.1
353.5
–
–
–
–
–
443.9
286.6
428.8
467.2i
75.3
79.3
78.4
427.9
438.2
284.0
420.5
2091.6i
67.0
67.4
63.6
416.7
–
–
–
–
–
–
–
]
]
]
Ea1[b] (kJmolꢀ1
)
]
]
Ea2[c] [kJmolꢀ1
Ea3[d] [kJmolꢀ1
[e]
ES1 [kJmolꢀ1
]
398ꢁ9[f]
[a] Imaginary frequencies for TS2 and TS3. [b] Differences in E+ZPE
values between TSs and MIN2. [c] Activation energies from MIN2 to TS2
or TS3 obtained by the Arrhenius plot of computed canonical rates with-
out the tunnelling correction. [d] Activation energies from MIN2 to TS2 or
TS3 obtained by the Arrhenius plot of computed canonical rates with the
tunnelling correction. [e] Estimated ES1 as (E+ZPE+Ea3ꢀEa2). [f] Experi-
mental ES1 (see Section 2.3).
When TS1 or TS2 are passed instead of TS3, dissociation
occurs directly on the S1 PES. There are S1/S0-MSX1 and S1/S0-
MSX2 outside the basin of MEK. The CH3CO or C2H5CO moiety
in these MSX structures are similar to the D1/D0-MSX structures
of isolated radicals with a (almost) linear O=CꢀC backbone,
and hence, they correlate to D1/D0-MSX structures of isolated
radicals. These radicals can come down to their ground state
through either S1/S0-MSXs or D1/D0-MSXs before or after com-
pletion of the dissociations. This mechanism is very similar to
the one found for acetone.[29,44,53] The DC of ethyl radical is
preferred by 24.0 kJmolꢀ1 compared to the DC of methyl
group in this S1 dissociation mechanism, and therefore a
tions and crude estimations of the tunnelling effects to the
barriers of S1-TS2 and S1-TS3. The tunnelling effects were esti-
mated by the Arrhenius plot of canonical reaction rate con-
stants in Equation (6) with one-dimensional Wigner correction
in Equation (7):
kBT QTS
eꢀE =RT
a
ð6Þ
ð7Þ
k ¼ G
h QMIN
!
2
z
1
hn
G ¼ 1 þ
24 kBT
3892
www.chemphyschem.org
ꢄ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemPhysChem 2010, 11, 3883 – 3895

Photochemistry of Methyl Ethyl Ketone
In Equation (6), QTS and QMIN are the total partition functions close to unity in MEK in Figure 6, which is also inconsistent
for TSs and MIN2, respectively, based on harmonic and rigid- with the S1 mechanism that requires significant thermal activa-
rotor approximations. Table 3 lists the activation energies, Ea, tion by a probability much smaller than unity. Among the
obtained as the slopes of the Arrhenius plot with and without pathways considered in this study, only the S1/S0-diradical
the tunnelling correction (Ea3 and Ea2, respectively). The final pathway can explain both the lack of contribution of the T1
effective barriers (ES1 in Table 3) for S1-TS2 and S1-TS3 were esti- mechanism and the high QY at low pressures. The most impor-
mated by adding the ZPE correction and (Ea3-Ea2) to the po- tant point is that the pathway for H atom transfer has a low
tential barriers. As expected, the tunnelling effect on S1-TS2 is barrier and can take place without thermal activation through
very small and is much larger in S1-TS3. Moreover, the final esti- H atom tunnelling. In the case of acetone without such a tun-
mated barrier ES1 of S1-TS3 is lower than S1-TS2, by nelling channel, dissociation does not happen below the barri-
11.2 kJmolꢀ1. Therefore, we suggest that the S1/S0-diradical er on the T1 surface (403.0 kJmolꢀ1),[29] which is slightly higher
pathway may be a major mechanism in the 248 nm photolysis. than at 308 nm. Consequently, the QY in acetone does not
It should be noted that such a small difference in barrier become unity even at very low pressures. On the other hand,
height may be negligible when the available kinetic energy is if the H atom tunnelling is possible without thermal activation,
very high in high energy photolysis starting from the S2 it is convincing that the QY in MEK can be near unity at low
PES.[12,44,45]
pressures. The lack of contribution of the T1 mechanism can
At the 308 nm (~388 kJmolꢀ1) photolysis, all the three path- also be interpreted by an assumption that the H atom tunnel-
ways are close in energy at 0 K. At a finite temperature, the ling is faster than the ISC. To obtain further evidence for the
system has a certain population of vibrationally excited mole- importance of this mechanism, more experimental and theo-
cules depending on the temperature, and they may undergo retical studies are required. An experiment using deuterated
one of the three pathways with the photon energy plus the in- MEK would provide additional information. Theoretically, a sim-
itial vibrational energy. As discussed, the straight SV plot indi- ulation is required to see how vibrational wavefunctions leak
cates that only one single mechanism is dominant in the out of the S1 basin through the tunnelling or the ISC, although
308 nm photolysis at 298 K. We propose that it is the S1/S0-dir- direct-dynamics simulations using reliable multi-reference ab
adical mechanism in the following discussions. Firstly, the T1 initio methods seem to be very difficult because both the tun-
mechanism can be excluded as discussed above. From the the- nelling and the ISC are expected to be very slow. In such
oretical calculations, it can be pointed out that the ISC from S1 cases, an accurate multi-dimensional global potential function
to T1 is slower in MEK than in acetone, because the energy is needed,[71] although development of such potential func-
gap between S1-MIN2 and T1-MIN3 (29.0 kJmolꢀ1) in MEK is tions for systems greater than ten atoms is still very challeng-
larger than that in acetone (21.2 kJmolꢀ1), at the same compu- ing.
tation level.[29] Moreover, the barrier of the S1 pathway in MEK
(443.9 kJmolꢀ1
is slightly lower than that in acetone the 248 nm photoexcitation, an H atom transfer occurs on the
The proposed mechanisms are summarized in Figure 9. After
)
(454.9 kJmolꢀ1). Hence, one may conclude that these two are S1 PES through S1-TS3 to generate S0/S1-MSX3. A nonadiabatic
reasons for the lack of contribution of a T1 pathway in MEK. transition takes place in the CI, then the transferred H atom
However, at least at low pressure, the T1 mechanism should be immediately comes back to the methyl group on the S0 PES to
dominant because it can take place with the excess energy of regenerate the ground state MEK. On the S0 PES, MEK decom-
391.3 kJmolꢀ1, which is much lower than 443.9 kJmolꢀ1, re- poses into either CH3CO+CH2CH3 or CH3CH2CO+CH3. In the
quired in the S1 mechanism. Note that the barrier for ethyl dis-
sociation does not change significantly depending on compu-
tation levels. Diau et al.[12] also obtained similar barrier height
of 442 kJmolꢀ1 and 454 kJmolꢀ1 at CASSCF and TD-DFT levels,
respectively.
In the 308 nm photolysis at 298 K, the population of mole-
cules with excess energy of 443.9 kJmolꢀ1 should be much
smaller than that of 391.3 kJmolꢀ1. Therefore, it is highly un-
likely that the S1 mechanism is dominant and the T1 mecha-
nism has no contribution at low pressure in this low energy
photolysis. Actually, in acetone, the T1 mechanism is dominant
at low pressure but is over taken by the S1 mechanism at high
pressures (because of the large and small pressure dependen-
ces of QY in the T1 and S1 mechanisms, respectively), and the
resulting SV plot is curved. More strangely, the QY of MEK with-
out the T1 mechanism is higher than the QY of acetone in
Figure 6 even at low pressures. Hence, another mechanism
that can take place with excess energy of around or below
391.3 kJmolꢀ1 is necessary to explain the lack of contribution
Figure 9. A summary of the photodissociation mechanisms of MEK at the
of the T1 mechanism. Furthermore, the QY at low pressures is
248 nm and 308 nm photoexcitations.
ChemPhysChem 2010, 11, 3883 – 3895
ꢄ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemphyschem.org 3893

S. Dꢂbꢃ, K. Morokuma et al.
308 nm photolysis, the excitation energy is not sufficient to
overcome the barrier of the excited state H atom transfer. We
have proposed that the barrier is passed through by an H
atom tunnelling.
Acknowledgements
This work has been supported by the European FP6 Project
SCOUT-O3 and the Hungarian Scientific Fund OTKA (contracts
K68437 and OMFB-00992/2009). This work is also partly support-
ed by a grant from Japan Science and Technology Agency with a
Core Research for Evolutional Science and Technology (CREST) in
the Area of High Performance Computing for Multiscale and Mul-
tiphysics Phenomena as well as grants from US AFOSR (Grant
No. FA9550-07-1-0395 and FA9550-10-1-030). The computational
resources at the Research Center for Computational Science at In-
stitute for Molecular Science are gratefully acknowledged.
The estimated ES1 for S1-TS3 (416.7 kJmolꢀ1) in Table 3 is
~10 kJmolꢀ1 higher than the experimentally determined barri-
er height of 398ꢁ9 kJmolꢀ1. Several reasons can be consid-
ered for this; the computation level, the one-dimensional ap-
proximation in the tunnelling correction, and some assump-
tions in the fitting of the experimental data. Computationally,
more expensive calculations including larger active space,
higher-order electron correlations, larger basis functions, and
multidimensional tunnelling corrections can be considered for
better energetics in the future studies. We recall here that we
had to use several approximations in deriving the experimental
ES1 value as well (Section 2.3). Further QY measurements, in
particular at longer wavelengths and low temperatures, would
help to better define the barrier height on the S1 surface. In
summary, we believe the agreement between experiment and
theory is reasonable, lending credence to the proposed photo-
dissociation mechanism of MEK.
Keywords: adiabatic
and
nonadiabatic
pathways
·
·
atmospheric chemistry
·
computational chemistry
photodissociation · quantum yield
[1] H. Singh, L. Salas, R. Chatfield, E. Czech, A. Fried, J. Walega, M. Evans, B.
Field, D. Jacob, D. Blake, B. Heikes, R. Talbot, G. Sachse, J. Crawford, M.
Avery, S. Sandholm, H. Fuelberg, J. Geophys. Res. 2004, 109, D15S07.
[2] F. Fehsenfeld, G. Ancellet, T. Bates, A. Goldstein, R. Hardesty, R. Honrath,
K. Law, A. Lewis, R. Leaitch, S. McKeen, J. Meagher, D. Parrish, A. Pszen-
ny, P. Russell, H. Schlager, J. Seinfeld, R. Talbot, R. Zbinden, J. Geophys.
Res. 2006, 111, D23S01.
[3] I. Galbally, S. Lawson, I. Weeks, S. Bentley, R. Gillett, M. Meyer, A. Gold-
stein, Environ. Chem. 2007, 4, 178–182.
[4] H. Khwaja, A. Narang, Chemosphere 2008, 71, 2030–2043.
[5] R. Sommariva, J. A. de Gouw, M. Trainer, E. Atlas, P. D. Goldan, W. C.
Kuster, C. Warneke, F. C. Fehsenfeld, Atmos. Chem. Phys. Disc. 2008, 8,
12371–12408.
[6] M. Ammann, R. Atkinson, R. A. Cox, J. Crowley, R. Hynes, M. E. Jenkin, W.
Mellouki, M. J. Rossi, J. Troe, T. Wallington in IUPAC subcommittee for
gas kinetic data evaluation for atmospheric chemistry. Evaluated kinetic
data-Data Sheet P8, http://www.iupac-kinetic.ch.cam.ac.uk/, 2005.
[7] M. Ammann, R. Atkinson, R. A. Cox, J. Crowley, M. E. Jenkin, R. Hynes, W.
Mellouki, M. J. Rossi, J. Troe, T. Wallington in IUPAC subcommittee for
gas kinetic data evaluation for atmospheric chemistry. Evaluated kinetic
data - Data Sheet HOx VOC20, http://www.iupac-kinetic.ch.cam.ac.uk/,
2007.
[8] M. T. Baeza-Romero, M. A. Blitz, D. E. Heard, M. J. Pilling, B. Price, P. W.
Seakins, L. M. Wang, Faraday Discuss. 2005, 130, 73–88.
[9] H. Singh, M. Kanakidou, P. Crutzen, D. Jacob, Nature 1995, 378, 50–54.
[10] L. Jaeglꢃ, D. J. Jacob, W. H. Brune, P. O. Wennberg, Atmos. Environ. 2001,
35, 469–489.
5. Implications for Tropospheric Chemistry
The rate of atmospheric photodissociation of MEK, JMEK(sꢀ1), is
given by Equation (8):
Z
JMEK
¼
Fðl,z,cÞ FMEKðl,P,TÞ sMEKðl,TÞdl
ð8Þ
where F is the actinic photon flux (quanta cmꢀ2 nmꢀ1 sꢀ1) that
depends on the wavelenght, l, the altitude, z, and the solar
zenith angle, c. The tropospherically more relevant wavelength
from our experimental study is 308 nm. The T-and P-depen-
dent FMEK and T-dependent sMEK values we propose imply that
the photolysis rate of MEK decreases with altitude at this
wavelength up to a few kilometres. For example, the decrease
of the FMEK ꢅsMEK product is ~60% when going from ground
level (T=298 K and P=1000 mbar) to 5 km altitude (Tꢄ250 K
and Pꢄ550 mbar), while F increases only by ~30% at 308 nm.
At even higher altitudes, the actinic photon flux increases rap-
idly with z, which probably becomes the decisive factor lead-
ing to an increase of the photodissociation rate. Nevertheless,
our assessment may be taken as an indication for the photoly-
sis rate of MEK indeed being slower in the free and upper tro-
posphere than previously thought, in accordance with the esti-
mations by Hynes et al.[26] and Baeza-Romero et al.[8] Conse-
quently, the relative importance of the OH reaction increases,
which is the other important tropospheric sink of MEK beside
photolysis. While the computation of the atmospheric rate of
OH reactions is straightforward, more information on photodis-
sociation QYs as a function of wavelength (~290-330 nm), tem-
perature (~200–300 K), and air pressure (~100–1000 mbar) is
required in order to provide accurate parameterised JMEK for
the use in atmospheric modelling studies.
[11] W. A. Noyes, G. B. Porter, J. E. Jolley, Chem. Rev. 1956, 56, 49–94.
[12] E. W. G. Diau, C. Kçtting, A. H. Zewail, ChemPhysChem 2001, 2, 273–293.
[13] P. Zhang, S. Maeda, K. Morokuma, B. Braams, J. Chem. Phys. 2009, 130,
114304.
[14] Y. Haas, Photochem. Photobiol. Sci. 2004, 3, 6–16.
[15] W. H. Raber, G. K. Moortgat in Photooxidation of Selected Carbonyl Com-
pounds in Air: Methyl Ethyl Ketone, Methyl Vinyl Ketone, Methacrolein and
Methylglyoxal, (Ed. J. Barker), World Scientific Publishing, Singapore,
1987.
[16] V. Khamaganov, R. Karunanandan, A. Rodriguez, J. Crowley, Phys. Chem.
Chem. Phys. 2007, 9, 4098–4113.
[17] B. Rajakumar, T. Gierczak, J. F. Flad, A. R. Ravishankara, J. B. Burkholder, J.
Photochem. Photobiol. A 2008, 199, 336–344.
[18] S. P. Sander, B. J. Finlayson-Pitts, R. R. Friedl, D. M. Golden, R. E. Huie, H.
Keller-Rudek, C. E. Kolb, M. J. Kurylo, M. J. Molina, G. K. Moortgat, V. L.
Orkin, A. R. Ravishankara, P. H. Wine, JPL Publication 2006, Evaluation
Number 15.
[19] M. T. Baeza-Romero, M. A. Blitz, D. E. Heard, M. J. Pilling, B. Price, P. W.
Seakins, Chem. Phys. Lett. 2005, 408, 232–236.
[20] R. Nꢀdasdi, Ph.D. Thesis, Budapest University of Technology and Eco-
nomics, 2009.
3894
www.chemphyschem.org
ꢄ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemPhysChem 2010, 11, 3883 – 3895

Photochemistry of Methyl Ethyl Ketone
[21] T. Gierczak, J. B. Burkholder, S. Bauerle, A. R. Ravishankara, Chem. Phys.
1998, 231, 229–244.
[51] S. W. North, D. A. Blank, J. D. Gezelter, C. A. Longfellow, Y. T. Lee, J.
Chem. Phys. 1995, 102, 4447–4460.
[22] R. Nꢀdasdi, G. Kovꢀcs, I. Szilꢀgyi, A. Demeter, S. Dꢂbꢃ, T. Bꢃrces, F.
Mꢀrta, Chem. Phys. Lett. 2007, 440, 31–35.
[52] D. Liu, W. H. Fang, X. Y. Fu, Chem. Phys. Lett. 2000, 325, 86–92.
´
ˇ
ˇ
[53] I. Antol, M. Eckert-Maksic, M. Oncak, P. Slavicek, H. Lischka, Collect.
[23] J. Troe, Z. Phys. Chem. 2000, 214, 573–581.
Czech. Chem. Commun. 2008, 73, 1475–1494.
[24] M. F. Mꢃrienne, A. Jenouvrier, B. Coquart, J. Atmos. Chem. 1995, 20,
281–297.
[25] R. Martinez, A. Buitrago, N. Howell, C. Hearn, J. Joens, Atmos. Environ. A
1992, 26 A, 785–792.
[26] A. J. Hynes, E. A. Kenyon, A. J. Pounds, P. H. Wine, Spectrochim. Acta A
1992, 48, 1235–1242.
[27] M. A. Blitz, D. E. Heard, M. J. Pilling, J. Phys. Chem. A 2006, 110, 6742–
6756.
[28] I. Szilꢀgyi, G. Kovꢀcs, M. Farkas, G. L. Zꢁgner, A. Gola, S. Dꢂbꢃ, A. Deme-
ter, React. Kinet. Catal. Lett. 2009, 97, 437–446.
[29] S. Maeda, K. Ohno, K. Morokuma, J. Phys. Chem. Lett. 2010, 1, 1841–
1845.
[54] D. Townsend, S. A. Lahankar, S. K. Lee, S. D. Chambreau, A. G. Suits, X.
Zhang, J. Rheinecker, L. B. Harding, J. M. Bowman, Science 2004, 306,
1158–1161.
[55] A. G. Suits, Acc. Chem. Res. 2008, 41, 873–881.
[56] B. R. Heazlewood, M. J. T. Jordan, S. H. Kable, T. M. Selby, D. L. Osborn,
B. C. Shepler, B. J. Braams, J. M. Bowman, Proc. Natl. Acad. Sci. USA
2008, 105, 12719–12724.
[57] V. Goncharov, N. Herath, A. G. Suits, J. Phys. Chem. A 2008, 112, 9423–
9428.
[58] J. Finley, P.-ꢆ. Malmqvist, B. O. Roos, L. Serrano-Andrꢃs, Chem. Phys. Lett.
1998, 288, 299–306.
[59] H. J. Werner, Mol. Phys. 1996, 89, 645–661.
[30] M. Blitz, D. Heard, M. Pilling, Chem. Phys. Lett. 2002, 365, 374–379.
[31] C. Moore, Annu. Rev. Phys. Chem. 2007, 58, 1–33.
[32] G. Peslherbe, W. Hase, J. Chem. Phys. 1996, 104, 7882–7894.
[33] J. Bowman, X. Zhang, Phys. Chem. Chem. Phys. 2006, 8, 321–332.
[34] X. S. Li, J. M. Millam, H. B. Schlegel, J. Chem. Phys. 2000, 113, 10062–
10067.
[35] R. L. Jaffe, D. M. Hayes, K. J. Morokuma, J. Chem. Phys. 1974, 60, 5108.
[36] R. L. Jaffe, K. Morokuma, J. Chem. Phys. 1976, 60, 4881–4886.
[37] D. Feller, M. Dupuis, B. C. Garrett, J. Chem. Phys. 2000, 113, 218–226.
[38] L. B. Harding, S. J. Klippenstein, A. W. Jasper, Phys. Chem. Chem. Phys.
2007, 9, 4055–4070.
[39] P. Zhang, Ph.D. Thesis, Emory University, 2005.
[40] B. C. Shepler, E. Epifanovsky, P. Zhang, J. M. Bowman, A. I. Krylov, K. Mo-
rokuma, J. Phys. Chem. A 2008, 112, 13267–13270.
[41] S. Maeda, K. Ohno, K. Morokuma, J. Phys. Chem. A 2009, 113, 1704–
1710.
[42] M. Araujo, B. Lasorne, M. J. Bearpark, M. A. Robb, J. Phys. Chem. A 2008,
112, 7489–7491.
[43] M. Araujo, B. Lasorne, A. L. Magalha¯es, G. A. Worth, M. J. Bearpark, M. A.
Robb, J. Chem. Phys. 2009, 130, 144301.
[44] E. W.-G. Diau, C. Kçtting, T. I. Sølling, A. H. Zewail, ChemPhysChem 2002,
3, 57–78.
[45] T. I. Sølling, E. W.-G. Diau, C. Kçting, S. D. De Feyter, A. H. Zewail, Chem-
PhysChem 2002, 3, 79–97.
[60] B. O. Roos, K. Andersson, Chem. Phys. Lett. 1995, 245, 215–223.
[61] H.-J. Werner, P. J. Knowles, R. Lindh, F. R. Manby, M. Schꢁtz, P. Celani, T.
Korona, A. Mitrushenkov, G. Rauhut, T. B. Adler, R. D. Amos, A. Bern-
hardsson, A. Berning, D. L. Cooper, M. J. O. Deegan, A. J. Dobbyn, F.
Eckert, E. Goll, C. Hampel, G. Hetzer, T. Hrenar, G. Knizia, C. Kçppl, Y. Liu,
A. W. Lloyd, R. A. Mata, A. J. May, S. J. McNicholas, W. Meyer, M. E. Mura,
A. Nicklass, P. Palmieri, K. Pflꢁger, R. Pitzer, M. Reiher, U. Schumann, H.
Stoll, A. J. Stone, R. Tarroni, T. Thorsteinsson, M. Wang, A. Wolf, MOLPRO,
version 2006.1, a package of ab initio programs, see http://www.mol-
pro.net.
[62] K. Ohno, S. Maeda, Chem. Phys. Lett. 2004, 384, 277–282.
[63] S. Maeda, K. Ohno, J. Phys. Chem. A 2005, 109, 5742–5753.
[64] K. Ohno, S. Maeda, J. Phys. Chem. A 2006, 110, 8933–8941.
[65] S. Maeda, K. Ohno, Chem. Phys. Lett. 2005, 404, 95–99.
[66] S. Maeda, K. Ohno, J. Chem. Phys. 2006, 124, 174306.
[67] M. J. Bearpark, M. A. Robb, H. B. Schlegel, Chem. Phys. Lett. 1994, 223,
269–274.
[68] F. Sicilia, L. Blancafort, M. Bearpark, M. Robb, J. Chem. Theory Comput.
2008, 4, 257–266.
[69] A. Banerjee, N. Adams, J. Simons, R. Shepard, J. Phys. Chem. 1985, 89,
52–57.
[70] S. Maeda, K. Ohno, K. Morokuma, J. Chem. Theory Comput. 2010, 6,
1538–1545.
[71] J. M. Bowman, B. J. Braams, S. Carter, C. Chen, G. Czakꢂ, B. Fu, X. Huang,
E. Kamarchik, A. R. Sharma, B. C. Shepler, Y. Wang, Z. Xie, J. Phys. Chem.
Lett. 2010, 1, 1866–1874.
[46] W.-K. Chen, J.-W. Ho, P. Y. Cheng, Chem. Phys. Lett. 2003, 380, 411–418.
[47] W.-K. Chen, J.-W. Ho, P. Y. Cheng, J. Phys. Chem. A 2005, 109, 6805–
6817.
[48] W.-K. Chen, P. Y. Cheng, J. Phys. Chem. A 2005, 109, 6818–6829.
[49] R. A. Copeland, D. R. Crosley, Chem. Phys. Lett. 1985, 115, 362–368.
[50] H. Zuckermann, B. Schmitz, Y. Haas, J. Phys. Chem. 1988, 92, 4835–
4837.
Received: June 30, 2010
Published online on October 18, 2010
ChemPhysChem 2010, 11, 3883 – 3895
ꢄ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemphyschem.org
3895