Photochemistry of methylglyoxal in the vapor phase

Article abstract of DOI:10.1021/jp981915r

The photolysis of methylglyoxal (CH₃COCHO) in the presence of synthetic air was studied by laboratory experiments in a static reactor in order to determine its atmospheric lifetime. Quantum yields of the molecular photolysis products CO and HCHO were determined at 298 K as a function of wavelength (260 ≤ λ ≤ 440 nm) and pressure (30 ≤ P ≤ 900 Torr) using an optical resolution of 8.5 nm. The results can be distinguished with respect to both UV/VIS-absorption bands of methylglyoxal. For the short-wavelength band (260 ≤ λ ≤ 320 nm) photolysis quantum yields were found to be unity, independent of wavelength and pressure, consistent with a dissociation mechanism yielding peroxy radicals and CO according to CH₃COCHO →^hv →^O₂,^M CH₃COO₂ + HO₂ + CO (P1). For the long-wavelength band (380 ≤ λ ≤ 440 nm) two different processes were distinguished. The major process is photodissociation (P1) with quantum yields (φ_D) decreasing with increasing wavelength and pressure following the Stern - Volmer relationship: 1/φ_D(λ) = 1/φ₀(λ) + P/k′_D(λ) with φ₀(λ) = (8.15 ± 0.5)(10^-9) [exp(7131 ± 267)] nm/λ and k′_D(λ)(A) = (7.34 ± 0.1)(10^-9 Torr) [exp(8793 ± 300)] nm/λ. The minor process could be described by an H-atom transfer between electronically excited MG and ground-state MG, yielding the experimentally observed products [CH₃COCHO]* + CH₃COCHO →^O₂,^M CH₃COO₂ + HCHO + CO + CH₃COO₂ (R10). The atmospheric lifetime due to photolysis (τ_phot) was calculated using an atmospheric radiation model and the above expression, where φ₀(λ) = 1 for λ < 380 nm, resulting in τ_poht = (4.1 ± 0.7) h for a solar zenith angle of 50° at ground level. Therefore, photolysis can be identified as the most important degradation process of atmospheric methylglyoxal.

Full text of DOI:10.1021/jp981915r

9142
J. Phys. Chem. A 1998, 102, 9142-9153
Photochemistry of Methylglyoxal in the Vapor Phase
Stephan Koch* and Geert K. Moortgat
Max-Planck-Institut fu¨r Chemie, Atmospheric Chemistry DiVision, Postfach 3060, 55020 Mainz, Germany
ReceiVed: April 17, 1998; In Final Form: July 1, 1998
The photolysis of methylglyoxal (CH3COCHO) in the presence of synthetic air was studied by laboratory
experiments in a static reactor in order to determine its atmospheric lifetime. Quantum yields of the molecular
photolysis products CO and HCHO were determined at 298 K as a function of wavelength (260 e λ e 440
nm) and pressure (30 e P e 900 Torr) using an optical resolution of 8.5 nm. The results can be distinguished
with respect to both UV/VIS-absorption bands of methylglyoxal. For the short-wavelength band (260 e λ
e 320 nm) photolysis quantum yields were found to be unity, independent of wavelength and pressure,
O2,M
hν
consistent with a dissociation mechanism yielding peroxy radicals and CO according to CH3COCHO f f
CH3COO2 + HO2 + CO (P1). For the long-wavelength band (380 e λ e 440 nm) two different processes
were distinguished. The major process is photodissociation (P1) with quantum yields (φD) decreasing with
increasing wavelength and pressure following the Stern-Volmer relationship: 1/φD(λ) ) 1/φ0(λ) + P/k′D(λ)
with φ0(λ) ) (8.15 ( 0.5)(10-9) [exp(7131 ( 267)] nm/λ and k′D(λ) ) (7.34 ( 0.1)(10-9 Torr) [exp(8793 (
300)] nm/λ. The minor process could be described by an H-atom transfer between electronically excited
MG and ground-state MG, yielding the experimentally observed products [CH3COCHO]* + CH3COCHO
O2,M
f CH3COO2 + HCHO + CO + CH3COO2 (R10). The atmospheric lifetime due to photolysis (τphot) was
calculated using an atmospheric radiation model and the above expression, where φ0(λ) ) 1 for λ < 380 nm,
resulting in τphot ) (4.1 ( 0.7) h for a solar zenith angle of 50° at ground level. Therefore, photolysis can
be identified as the most important degradation process of atmospheric methylglyoxal.
Introduction
CH3COCHO + OH f CH3COCO + H2O
CH3COCO f CH3CO + CO
(R1)
(R2)
(R3)
Methylglyoxal (CH3COCHO) is an atmospherically important
bicarbonyl compound since it is a major product of natural
isoprene oxidation. The oxidation of isoprene by OH radicals1-3
as well as O34-6 leads, besides other products, to the formation
of methyl vinyl ketone and methacrolein, which are further
oxidized by OH7,8 or O34,9 to methylglyoxal and other products.
Methylglyoxal (MG) is also a product of the OH initiated
oxidation of hydroxy acetone.10,11 In anthropogenically polluted
air, MG may be formed by the oxidation of toluene, undergoing
ring cleavage after OH attack.12-14
HCO + O2 f HO2 + CO
CH3CO + O2
9
M8 CH3COO2
(R4)
The reactions of the intermediately formed peroxy radicals give
rise to the experimentally observed stable products CO and
HCHO as discussed below.
The lifetime due to OH reaction can be estimated to be 16 h
for an OH concentration of 1 × 106 molecules cm-3 and a room
temperature rate coefficient of k3 ) 1.3 × 10-11 cm3 molecule-1
Because MG is potentially an atmospheric important precursor
of HO2, CH3COO2 radicals, and peroxyacetylnitrate (PAN),15
it has recently been observed in field experiments, where
atmospheric mixing ratios of 50 pptv were detected.16,17
s-1 21
.
The photolysis of MG was already the subject of several
laboratory investigations, resulting in estimates of the atmo-
spheric lifetime to 2,23 0.6,19 and 2.7 h ( 0.7 h,20 respectively.
The lifetime given by Plum et al.,23 is based on an
environmental chamber study by use of an unfiltered Xe arc
lamp to simulate the solar spectrum. They also determined
adsorption cross sections that were half as large as the values
found by Meller et al.10 and Staffelbach et al.20
The low concentrations of MG in the troposphere are due to
efficient degradation pathways by photolysis,19,20 OH reaction,21
and dry and wet deposition. NO3 as well as O3 reactions are
atmospherically not relevant.22,23 No quantitative data are
available concerning the efficiency of the deposition processes.
Photolysis (P1a) and OH reaction (R1) lead, following O2
addition, to the formation of CO (R2, R3), acetyl peroxy radicals
(R4), and HO2 radicals (R3).24
In the laboratory studies published by Raber and Moortgat,18,19
the photolysis of MG in the presence of 50-760 Torr synthetic
air was investigated using two types of broad-band emitting
light sources (275-380 and 390-470 nm) and FTIR detection.
From the analysis of the photolysis products (CO, CO2, HCHO,
CH3OH, CH3O2H, HCOOH, CH3COOH, CH3COO2H,
CH3CHO (275-380 nm), O3, H2O, CH3COCOOH), the major
photodissociation mechanism P1a could be proposed and the
CH3COCHO
9
hν8 CH3CO + HCO
(P1a)
* Author to whom correspondence should be addressed. E-mail: koch@
mpch-mainz.mpg.de.
10.1021/jp981915r CCC: $15.00 © 1998 American Chemical Society
Published on Web 10/21/1998

Photochemistry of Methylglyoxal in the Vapor Phase
J. Phys. Chem. A, Vol. 102, No. 46, 1998 9143
atmospheric lifetime was estimated from the average photolysis
quantum yield over the lower-wavelength range. Methodologi-
cally similar experiments were presented by Staffelbach et al.20
They investigated the photolysis products of MG in 20 Torr O2
and 740 Torr N2 using the light of a Xe arc lamp in conjunction
with several broad-band filters (410-418, 355-480, 280-420,
280-420, and 240-420 nm) and FTIR detection. The pho-
tolysis products found by Raber and Moortgat (except CH3-
CHO, CH3O2H, and CH3COCOOH) and the major dissociation
mechanism were confirmed by Staffelbach et al., but a five times
larger atmospheric lifetime was determined from their quantum
yields.
With respect to the discrepancies and the insufficiencies of
the broad-band quantum yields determined in earlier experi-
ments, the quantitative determination of the quantum yields of
the photolysis of MG as a function of wavelength and pressure
were necessary in order to obtain reliable information on the
atmospheric lifetime of MG. Therefore, the quantum yields of
the molecular photolysis products CO and HCHO were inves-
tigated in order to derive quantitative information about the
primary photodissociation processes.
Figure 1. Chromatograms of photolysis mixtures ([MG] ) 5.3 × 1015
molecules cm-3, 300 Torr synthetic air, 280 nm) recorded after different
photolysis times, using the RGD.
Samples were analyzed using a gas chromatograph (Hewlett
Packard 5890) equipped with a flame ionization detector (FID),
a quadrupole mass selective detector (HP 5970 MSD), and a
reduction gas detector (Trace Analytical RGD 2). Samples to
be analyzed with the reduction gas detector (RGD) or the FID
were isothermally separated within 80 s (Figure 1) on a 60 m,
0.5 mm i.d. quartz capillary column coated with a 5 µm poly-
methylsiloxane phase (Quadrex 007) with a He carrier gas flow
rate of about 60 mL/min. The MSD was connected directly to
a 30 m, 0.25 mm i.d. column (1 µm poly-methylsiloxane, J&B
5) using a flow rate of 0.8 mL/min.
Quantitative detection of the photolysis products CO, HCHO,
and CH3CHO was performed with the reduction gas detector
(RGD) which is particularly selective and sensitive for CO and
carbonyl compounds. Yellow HgO in a stainless steel tube of
35 mm length and 4 mm i.d. is reduced by CO and carbonyls
at 550 K to Hg vapor which is detected by absorption of the
254 nm line of a Hg lamp.
Experimental Section
Experiments were performed in a static reactor made of a
piece of Pyrex tubing of 100 mm length and 35 mm inner
diameter, fitted with double-sided evacuated quartz windows
on both ends. The inner surface of the reactor had to be
deactivated by silanization with hexamethyldisilazan (HMDS)
at 650 K25 in order to minimize adsorption of MG on the reactor
walls, since in earlier experiments a significant amount of the
products observed were due to photolysis of MG adsorbed on
the reactor walls.
∼Si-O-H + H-N(-Si(-CH3)3)2 650 K8
1
2
1
2
CO + HgO 550 K8 CO2 + Hgv
∼Si-O-Si(CH3)3 + NH3v
HCHO + 2HgO 550 K8 CO2 + H2O + Hgv
As light source a 150 W Xe arc lamp (Cermax LX150 UV)
equipped with a parabolic reflector was used, emitting a
continuous spectrum (240-800 nm) without sharp maxima. The
nearly parallel light beam was focused with a quartz lens (f )
120 mm) onto the entrance slit (5 mm) of a f/2 monochromator
(Jobin-Yvon H.L.) equipped with a 1484 grooves/mm grating.
The divergent light cone was made parallel by a lens (f ) 120
mm) and passes through the reactor with little illumination of
the side walls. The photon flux was registered with a
manufacturer-calibrated photodiode (Newport 818 UV with
attenuator), and the digitized signals were stored in a computer.
The spectral resolution of the optical set-up was determined
using a second monochromator (AMKO Metrospec, 1200
grooves/mm, slit width 0.5 mm) to be 8.5 ( 0.1 nm FWHM of
an almost Gaussian distribution.
The gaseous mixtures (0.03-1 Torr MG, 30-900 Torr
synthetic air) to be photolyzed were prepared in a mixing system
equipped with capacitance manometers (MKS Baratron, 10 and
1000 Torr range) and expanded into the evacuated reactor.
Samples were taken in intervals of 5 min, by expansion of
the gas mixture into an evacuated sample loop consisting of a
1 m, 0.5 mm i.d. deactivated fused silica capillary connected
to a six-port valve (Valco) which served as a splitless injector.
Each sampling step reduced the pressure in the reactor by 0.85%
( 0.05% which had to be taken into account when calculating
photochemical yields.
Due to adsorption of analytes on the HgO surface, peak tailing
(Figure 1) and saturation effects were observed. The signals
were digitized and integrated on a computer.
The specified purities of the gases used in the photolysis
experiments and for chemical actinometry were, if not stated
otherwise, 99.999%: He, synthetic air, N2, O2, H2, CO 99.997%,
Cl2 99.96%, CH4 99.95%, C2H6 99.95%. The carrier gas was
further purified by use of filter cartridges with respect to traces
of CO, O2, and hydrocarbons.
Acetaldehyde was obtained with a purity of 99% and degassed
by repeated trap-to-trap distillation. Formaldehyde was prepared
by gentle heating of paraformaldehyde (95%), and traces of
water and oligomers were removed using a cold trap at 170 K.
Methylglyoxal, obtained as a 40% solution in water, was
concentrated by pumping off the water for several hours. The
resulting high-viscous polymer was distilled at 330 K and 10-3
Torr over a column of 300 mm length and 30 mm i.d. packed
with phosphorus pentoxide (98%).10 The freshly prepared MG
still contained impurities of CO (<2%), HCHO (<0.5%),
CH3CHO (<0.5%), and traces of biacetyl and pyruvic acid
which were identified by their mass spectra. Although CO,
HCHO, and CH3CHO were also found as photolysis products
(Figure 1), their concentration also increased during storage in
the dark. A 5% decomposition of the sample as determined by

9144 J. Phys. Chem. A, Vol. 102, No. 46, 1998
Koch and Moortgat
khProd(λ , P) ) φh(λ , P) σj(λ ) I (λ) dλ (2)
∫
0
0
0
0
The photon flux (∫I0(λ) dλ) was determined as a function of
wavelength using Cl2 as gaseous chemical actinometer, as it
can be used in the same apparatus as used in the MG
experiments. Molecular chlorine photodissociates with quantum
yields of unity and shows sufficiently large cross sections in
the wavelength range of interest (260 < λ < 440 nm).26
Absorption Cross Sections. The absorption cross sections
were determined at 298 K in the wavelength range between
260 and 440 nm in intervals of 20 nm with the same
experimental set-up and the same optical resolution as used in
the photolysis experiments. The cross sections of Cl2 were
obtained with a relative precision of 2% but were somewhat
lower (≈2%) than the values reported in the literature,26 which
is expected due to the lower resolution used in this work.
The UV/VIS absorption spectrum of MG, consisting of two
absorption bands (Figure 2), determined by Meller et al.10 and
Staffelbach et al.20 were in very good agreement. The short-
wavelength band (λ < 340 nm) is assigned by an electronically
n- f π4 transition, the long-wavelength band (λ > 340 nm)
by a n+ f π3 transition.27,28
Figure 2. Absorption spectrum of MG. Symbols: cross sections
determined in this work, at a resolution of 8.5 nm. Line: high-resolution
spectrum.10
the CO production, occurred within a period of several hours
to several weeks. Therefore, the MG used in experiments was
freshly distilled from the polymer, as the accuracy of the product
yield determination was limited by the amount of impurities,
not by the sensitivity of detection.
The absorption cross sections found for MG, displayed in
Figure 2, are systematically about 10% lower than given in the
literature.10,20 This discrepancy can be explained by heteroge-
neous losses of MG, as reported by Meller et al.10 The relative
statistical errors in the determination of the cross sections were
found to be (6% and are caused by the uncertainty of the MG
concentration.
Actinometry
The photodissociation of a molecule is generally described
by a first-order kinetic process with the photodissociation
coefficient (kphot) given as an integral over the observed
wavelength range,
Photon Flux. Mixtures of C2H6 (5 × 1013 to 5 × 1014
molecules cm-3) and Cl2 in at least 10-fold excess (5 × 1014 to
9 × 1015 molecules cm-3) in 760 Torr synthetic air were
photolyzed at 298 K to an amount of ca. 10% removal of C2H6.
The decay of ethane was monitored by use of the FID, and the
results were numerically fitted to the kinetic model shown in
Table 1 using Facsimile.29 The photon flux (∫ I0(λ) dλ) obtained
at the selected wavelengths was in the range between (0.81 (
0.07) × 1014 photons cm-2 s-1 at 260 nm and (3.84 ( 0.22) ×
1014 photons cm-2 s-1 at 440 nm. The relative statistical error
kphot(P) ) φ(λ, P) I (λ) σ(λ) dλ
(1)
∫
0
where φ represents the quantum yield of the process, I0 the
photon flux per unit wavelength, and σ the absorption cross
section. Under the assumption that these parameters do not
change significantly, the integral can be substituted by mean
values (eq 2) as it was done in this work.
TABLE 1: Reactions of the Ethyl Peroxy Radical
k(298) [cm3 s-1
]
ref
24
O , M8
2
Cl + C2H6
C2H5O2 + C2H5O2
C2H5O2 + HCl
C2H5OH + CH3CHO + O2
(5.9 ( 0.3) 10-11
(4.4 ( 0.8) 10-14
41
8
O , M8
2
2CH3CHO + 2HO2
C2H5O2H + O2
(2.3 ( 0.4) 10-14
(5.4 ( 1.0) 10-12
41
42, 43
HO2 + C2H5O2
Cl + C2H5OH
Cl + CH3CHO
Cl + C2H5O2H
8
O , M8
2
CH3CHO + HCl + HO2
CH3COO2 + HCl
products + HCl
(9.4 ( 1.0) 10-11
(7.2 ( 1.0) 10-11
(1.1 ( 0.2) 10-10
24
24
44
O , M8
2
O , M8
2
TABLE 2: Reactions of the Acetyl Peroxy Radical
k(298) [cm3 s-1
]
ref
45
O , M8
2
CH3COO2 + CH3COO2
CH3O2 + CH3COO2
2CH3O2 + 2CO2
(1.4 ( 0.2) 10-11
O , M8
2
CH3O2 + CO2 + HCHO + HO2
HCHO + CH3COOH + O2
CH3COO2H + O2
CH3COOH + O3
CH3O2H + O2
(8.8 ( 1.5) 10-12
(1.0 ( 0.5) 10-12
(8.5 ( 4.5) 10-12
(2.9 ( 0.7) 10-12
(5.2 ( 0.3) 10-12
(1.6 ( 0.2) 10-12
(1.6 ( 0.2) 10-13 a
45
45
46
46
24
24
24
8
8
8
8
8
8
HO2 + CH3COO2
HO2 + CH3O2
HO2 + HO2
H2O2 + O2
H2O2 + O2
M
a 100 Torr synthetic air.

Photochemistry of Methylglyoxal in the Vapor Phase
J. Phys. Chem. A, Vol. 102, No. 46, 1998 9145
in the determination of the photon flux is (7% and is caused
by the error of d[C2H6]/dt. Possible systematic errors, intro-
duced by uncertainties of the reaction rates (Table 1), were
estimated to affect the results not more than 3%. The output
of the lamp was monitored during each experiment and corrected
for long-term fluctuations.
detector; therefore, the initial concentrations of methylglyoxal
could only be quantified volumetrically.
The product quantum yields (φProd) were derived from the
slope of the concentration-time profiles (d[Prod]/dt), which
were linear in all cases.
d[Prod]/dt(λ, P)
Results and Discussion
φProd(λ, P) )
(3)
Depending on the energy of the photon absorbed, the
following thermodynamically allowed pathways of dissociation
of electronically excited MG are possible.
[MG] σ(λ) I (λ) dλ
∫
0
0
The quantum yields of CO, CH3CHO, and HCHO are displayed
in Figures 3, 4, and 5 as a function of the pressure of air (P)
for all wavelengths investigated. The error bars represent the
standard deviation of the slope of the concentration-time profiles
(∆(d[Prod]/dt)). In separate experiments using the FID, the dark
loss of MG to the reactor walls was determined to be kdark ) (2
( 1.5) × 10-5 s-1, of the same order of magnitude as the
photolysis rates of MG. Therefore, dark losses resulted in
underestimations of the quantum yields, comparable to of the
amount of reaction, less than 5%. The quantum yields can,
like the UV/VIS spectrum, be distinguished with respect to both
absorption bands, consequently they will be discussed separately.
Short-Wavelength Band (260 e λ e 320 nm). The product
quantum yields obtained in this region are independent of
wavelength as well as pressure (Figures 3, 4, 5).
CH3COCHO
9
hν8 CH3CO + HCO (λ < 422 nm) (P1a)
hν8 CH3COCO + H (λ < 385 nm) (P1b)
hν8 CH3 + COCHO (λ < 370 nm) (P2a)
9
9
9
hν8 CH3 + CO + HCO (λ < 343 nm)
(P2b)
9
hν8 CH3CHO + CO (all λ)
hν8 CH3CHO + CO (all λ)
(P3)
9
(P3)
φCO (260 e λ e 320 nm) ) 1.20 ( 0.09
φHCHO (260 e λ e 320 nm) ) 0.19 ( 0.02
The threshold wavelengths were calculated from enthalpy data
(∆H°(298)).24,30 However, considering that the mean enthalpy
f
of vibration and rotation is 2.5 kJ/mol (200 cm-1) per degree
of freedom, it follows that P1a is thermodynamically allowed
at all wavelengths investigated in this study.
φCH CHO (260 e λ e 320 nm) ) 0.18 ( 0.04
3
The primary steps P1a and P1b cannot be distinguished in
the presence of O2, since their primary products yield by fast
reactions the same peroxy radicals (R3, R4 and R2, R5,
respectively). P2a and P2b are also indistinguishable and
yield two molecules of CO per molecule dissociated (R7, R3).
CO and CH3CHO Quantum Yields. In order to explain
the experimentally observed CO quantum yields higher than
unity, gas phase processes with yields higher than 1.0 (P2, P4),
or heterogeneous photolysis of MG adsorbed onto the reactor
walls have to be assumed, since the radical intermediate of P1,
CH3COO2, does not represent a source of CO (Table 2).
The photodissociation channel P4 can be excluded, since no
CH4 could be detected as photolysis product in experiments (260
e λ e 440 nm) using the FID. Note that, with respect to the
HCO + O2
9
M8 HO2 + CO
M8 CH3COO2
CH3COCO f CH3CO + CO
(R3)
CH3CO + O2
9
(R4)
(R2)
detection limit of CH4, one obtains φCH < 0.002, in accordance
4
with earlier work.19,20
Assuming that CO is formed via P1′, P2′, and P3 (eq 4) and
that CH3CHO is produced via P3 (eq 5),
H + O2
9
M8 HO2
M8 CH3O2
(R5)
(R6)
(R7)
hν8 O ,M8 CH3COO2 + HO2 + CO (P1′)
2
CH3COCHO
9
CH3 + O2
9
9
hν8 O ,M8 CH3O2 + HO2 + 2CO (P2′)
2
COCHO 9
M8 CO + HCO
The formation of HCHO and other molecular and intermediate
radical products are expected due to the known peroxy radical
reactions given in Table 2.
Therefore, in principle, the primary dissociation pathways
(P1-P4) can be quantitatively distinguished by examining the
quantum yields of the molecular photolysis products CO,
HCHO, CH3CHO, and CH4.
Product Quantum Yields in Synthetic Air. Mixtures of
MG (1 × 1015 to 1 × 1016 molecules cm-3) in 30-900 Torr of
synthetic air were photolyzed for up to 60 min, consuming not
more than 5% of the initial MG, at 298 K and wavelengths
between 260 and 440 nm. CO, HCHO, and CH3CHO could
be detected as products with the RGD. Relatively large amounts
of MG as used in these experiments led to saturation of the
9
hν8 CH3CHO + CO
(P3)
one obtains under the condition that the sum of primary quantum
yields is unity (eq 6) the following equations:
φ1 + 2 φ2 + φ3 ) 1.2 ) φCO
(4)
(5)
φ3 ) 0.2 ) φCH CHO
3
φ1 + φ2 + φ3 ) 1
φ1 + 2φ2 ) 1
(6)
with (4) and (5),
(7)

9146 J. Phys. Chem. A, Vol. 102, No. 46, 1998
Koch and Moortgat
Figure 3. Quantum yields of CO as a function of pressure obtained
in synthetic air. Symbols: experimental data at the indicated wave-
lengths; horizontal line: mean values of experimental data obtained at
wavelengths between 260 and 320 nm; curves: fit according to eq 12.
Figure 4. Quantum yields of CH3CHO as a function of pressure
obtained in synthetic air. Symbols: experimental data at the indicated
wavelengths; horizontal line: mean values of experimental data obtained
at wavelengths between 260 and 320 nm.
with (5) and (6),
observed dark loss of MG. Consequently, the CO quantum
yields (φCO′) mentioned further in the text reflect the experi-
mentally observed CO quantum yields reduced by the CH3CHO
quantum yields (eq 9).
φ1 + φ2 ) 0.8
φ1 ) 0.6
(8)
with (7) and (8),
and
HCHO Quantum Yields. The quantum yields of formal-
dehyde, initiated by P1, can very well be explained by the
reaction sequence of the intermediately formed peroxy radicals,
as given in Table 2. Numerical simulation of this system shows
a very good agreement between the experimentally observed
quantum yield φHCHO ) 0.19 ( 0.02 and the result of the model
calculation (0.21 ( 0.05). The uncertainty of the kinetic model
was investigated by repeated random variation of the rate
coefficients within their error limits, given in Table 2. The
assigned error is the FWHM of the Gaussian-like distribution
of the quantum yield obtained after 1000 computations.
φ2 ) 0.2
From model calculations of this case, according to the reaction
scheme in Table 2, a HCHO quantum yield of 0.15 is expected.
However, this explanation is unrealistic for several reasons.
Staffelbach et al. did not find CH3CHO as photolysis product
of MG20 using a stainless steel reactor. On the other hand, Raber
and Moortgat, using a reactor made of quartz, observed
CH3CHO with quantum yields of 0.1 when photolyzing MG in
the wavelength range between 275 and 380 nm.19 However,
constant and pressure independent quantum yields over the
wavelength range 260-320 nm are not expected if they do not
correspond to a primary process with a quantum yield of unity.
Also, formation of CO and CH3CHO via chain reactions of
the acetyl radical with MG (R8, R2) can be excluded with
respect to the fast reaction with O2 (R4).
Photodissociation of MG yielding 2 equivalents of CO
according to P2′
hν8 O ,M8 CH3O2 + HO2 + 2 CO (P2′)
2
CH3COCHO
9
can be ruled out, since quantum yields of φCO ) 2 and φHCHO
) 0.12 are expected from numerical analysis of the kinetic
model in contradiction to the experimental results.
Long-Wavelength Band (380 e λ e 440 nm). The quantum
yields of CO and HCHO from the long-wavelength transition
increase with the energy of the photon absorbed. The acetal-
dehyde yields do not exceed 5% of the CO yields, hence they
are practically negligible.
CO Quantum Yields. The CO yields decrease with increas-
ing pressure and show in the Stern-Volmer representation a
nonlinear relationship between the reciprocal quantum yields
and total pressure (Figure 6). These findings can be empirically
described, if the experimentally observed quantum yields are
CH3CO + CH3COCHO f CH3CHO + CH3COCO (R8)
CH3COCO f CH3CO + CO
M8 CH3COO2
(R2)
(R4)
CH3CO + O2
9
No oxygen partial pressure dependence of the CH3CHO
quantum yield was observed under the experimental conditions
employed in this work.
Assuming that the formation of acetaldehyde is caused by
photolysis of MG adsorbed onto the reactor walls
assumed as a sum of two independent yields, φD and φCO
.
∞
φCO(λ, P) ) φD(λ, P) + φCO (λ)
∞
(CH3COCHO)wall 9
hν8 CH3CHOv + COv
(φD)-1(λ, P)
(φCO)-1(λ, P) )
(10)
the CO quantum yield of the gas phase (φCO′) is then
1 + φCO (λ) × (φD)-1(λ, P)
∞
φCO′ ) φCO - φCH CHO ) 1.0 ( 0.1
(9)
3
The first term (φD) is supposed to represent the photodissociation
as a function of pressure, obeying the Stern-Volmer relationship
This supposition is reasonable with respect to the experimentally

Photochemistry of Methylglyoxal in the Vapor Phase
J. Phys. Chem. A, Vol. 102, No. 46, 1998 9147
TABLE 3: Coefficients of CO Quantum Yield According to
Equation 12
λ[nm]
φ0
φCO
kD′ [Torr]
∞
380
400
420
440
1.10 ( 0.12
0.37 ( 0.03
0.18 ( 0.03
0.05 ( 0.01
0.043 ( 0.040
0.030 ( 0.013
0.030 ( 0.004
0.018 ( 0.002
76 ( 21
36 ( 9
9 ( 2
4 ( 1
displayed in Figure 6 with coefficients given in Table 3.
(φ0)-1 + P/kD′
(φCO)-1(λ) )
(12)
1 + φCO ((φ0)-1 + P/kD′ )
∞
HCHO Quantum Yields. Quantum yields of formaldehyde
increase with photon energy and show at 380 nm little and in
the range of 400-440 nm no dependence upon pressure (Figure
5). To explain these unusual observations, the quantum yields
are, as in the case of CO, assumed to be described by a sum of
two functions. If the first term is given by a Stern-Volmer
function analogous to eq 11, then the second term must show
a positive pressure dependence (in order to compensate the first)
so to reach constant values at high pressure, when the contribu-
tion of the Stern-Volmer function becomes zero. Therefore,
a process has to be postulated which reflects a rate of HCHO
formation increasing with pressure and reaching a limit at high
pressures.
Figure 5. Quantum yields of HCHO as a function of pressure obtained
in synthetic air. Symbols: experimental data; horizontal line: mean
values of experimental data obtained at wavelengths between 260 and
320 nm.
The rate of the reaction R9
CH3O + O2 f HCHO + HO2
(R9)
is proportional to the O2 partial pressure and might therefore
appear as a pressure-dependent source of HCHO in experiments
performed in synthetic air, since the [O2]/P ratio is constant.
However, quantum yields obtained at wavelengths shorter than
320 nm show no O2 partial pressure dependence. Therefore, it
can be concluded that eventually competing reactions are slow
compared to R9 under the experimental conditions employed
in this work.
Figure 6. Stern-Volmer representation of reciprocal quantum yields
of CO obtained in synthetic air. Symbols: experimental data at the
indicated wavelengths; curves: fit according to eq 12.
For the intramolecular rearrangement of an H atom
(eq 11),
CH3COCHO 9
hν8 H2CdCO + HCHO
(φD)(λ, P) )
leading to the formation of ketene and formaldehyde, pressure-
dependent quantum yields would be expected since this process
represents a photodissociation. No hint of this pressure
dependence has been observed in this study or earlier investiga-
tions. This is reasonable, since the thermodynamically favored
pathway leading to methane (P4) also does not occur.
A reaction path yielding HCHO is feasible via intermolecular
H atom transfer between electronically excited MG and ground-
state MG, followed by fragmentation of the C-C bond leading
to CO, HCHO, and acetyl radicals.
kD(λ)
kD′ (λ)
)
kD(λ) + kX(λ) + kP(λ) × P k′ (λ) + kX′ (λ) + P
D
1
1
P
)
+
(11)
φD(λ, P) φ0(λ) k′ (λ)
D
with
kX′ (λ)
1
[CH3COCHO]* + HCOCOCH3 f
) 1 +
φ0(λ)
k′ (λ)
D
CH3CO + HCHO + CO + CH3CO (R10)
A comparable gas phase reaction was observed by the photo-
fragmentation of 2-pentanone. This Norrish Type II reaction
proceeds via formation of a biradical intermediate undergoing
an intramolecular H shift, followed by fragmentation of the C-C
bond yielding enol-acetone (CH3C(OH)dCH2) and ethene as
products.31
where kD, kX, und kP represent the rate coefficients of dissocia-
tion, the pressure-independent, and the pressure-dependent
deactivation of vibronically excited molecules. The coefficients
kD′ und kX′ are normalized to kP, and φ0 indicates the quantum
yield extrapolated to zero pressure. The minor yield (φCO ) is
∞
assumed to be pressure independent, giving the limit of the CO
quantum yield at infinite pressure. Substituting eq 11 in eq 10
CH3COCH2CH2CH3 9
hν8 [CH3C(OH)CH2CH2CH2]*
-1
yields the expression of (φCO
)
as a function of pressure as

9148 J. Phys. Chem. A, Vol. 102, No. 46, 1998
Koch and Moortgat
Figure 9. Coefficients of quantum yield of dissociation as a function
of reciprocal wavelength. Symbols: coefficients as given in Table 4;
lines: fit according to eqs 17 and 18, respectively.
Figure 7. Quantum yields of HCHO as a function of MG concentration
(380 nm: P ) 400 Torr, [O2] ) 90 Torr; 440 nm: P ) 760 Torr, [O2]
) 150 Torr). Symbols: experimental data.
product yields of each process. If these coefficients are known,
the quantum yields φD and φR can be extracted from the
experimentally observed quantum yields.
φD ) (δφCO′ - γφHCHO) ꢀ
φR ) (RφHCHO - âφCO′) ꢀ
(15)
(16)
with
ꢀ ) (Rδ - âγ)-1
The relative product yields can be predicted by numerical
analysis of the reaction scheme given in Table 2. For the
dissociation yield via P1a, one obtains R ) 1.00 ( 0.00 and â
) 0.21 ( 0.06, which means that per molecule dissociated, one
molecule of CO as well as 0.2 molecules of HCHO are formed.
These results agree with the quantum yields obtained at short
wavelengths. The coefficients due to R10 were calculated using
the reactions given in Table 2, γ ) 1.00 ( 0.00 and δ ) 2.03
( 0.03 with ꢀ ) 0.55 ( 0.02. The assigned errors were
estimated regarding the uncertainties of the rate coefficients of
the peroxy radical reactions. By variation of the boundary
conditions it was found that the values of the coefficients R, â,
γ, and δ do not change significantly, if both processes take place
simultaneously with variable contributions.
Quantum Yields of Photodissociation. The quantum yields
of photodissociation (P1a) were extracted from experimental
data according to eq 15. The reciprocal yields are shown as a
function of total pressure (P) in Figure 8 and are found to be
consistent with linear Stern-Volmer functions (eq 11). The
coefficients of these functions are given in Table 4 and are in
very good agreement with those in Table 3. The assigned errors
represent the standard deviations of the parameters obtained
Figure 8. Stern-Volmer representation of reciprocal quantum yields
of dissociation obtained in synthetic air. Symbols: calculated from
experimental data according to eq 15; curves: fit according to eq 11.
[CH3C(OH)CH2CH2CH2]* f
CH3C(OH)dCH2 + H2CdCH2
If reaction R10 occurs, then the yield of HCHO should increase
with MG concentration. Testing this hypothesis, further experi-
ments were performed at 380 and 440 nm, under constant
pressure and variable MG concentrations. Figure 7 shows the
increase of φHCHO with MG concentration reaching a plateau at
partial pressures higher than about 1 Torr. This experiment
clearly demonstrates the dependence of the HCHO quantum
yields on MG concentration; however, no information concern-
ing the total pressure dependence is obtained. Therefore, further
experiments were performed in order to verify this approach.
Elementary Processes
from weighted least-squares fits. The coefficients k′ and φ0
D
can also be described by an exponential function of the
reciprocal wavelength as shown in Figure 9.
Supposing that in the long-wavelength band the experimen-
tally observed yields are caused by two different processes,
photodissociation (P1a) and chemical reaction of electronically
excited MG (R10), then the CO and HCHO quantum yields
represent a linear combination of the quantum yields of both
elementary processes, φD and φR,
k′ (λ) ) (7.34 ( 0.1)(10-9 Torr) [exp(8793 ( 300)] nm/λ
D
(17)
φ0(λ) ) (8.15 ( 0.5)(10-9) [exp(7131 ( 267)] nm/λ
for λ > 380 nm (18)
φCO ) RφD + γφR
(13)
(14)
φ0(λ) ) 1 for λ < 380 nm
φHCHO ) âφD + δφR
where the coefficients R, â, γ, and δ represent the relative
Using eq 11 together with the expressions given in eqs 17 and

Photochemistry of Methylglyoxal in the Vapor Phase
J. Phys. Chem. A, Vol. 102, No. 46, 1998 9149
Figure 10. Quantum yields of dissociation as a function of wavelength
for various pressures. Symbols: calculated according to eq 11 using
the coefficients given in Table 4; lines: calculated according to eqs
11, 17, 18.
Figure 11. Quantum yields of R10 as a function of total pressure
obtained in synthetic air. Symbols: calculated from experimental data
according to eq 16; curves: fit according to eq 20.
TABLE 4: Coefficients of Quantum Yield of Dissociation
According to Equation 11
λ[nm]
φ0
k′ [Torr]
k′ [Torr]
D
X
380
400
420
440
1.10 ( 0.14
0.47 ( 0.09
0.21 ( 0.06
0.08 ( 0.02
78 ( 9
31 ( 4
10 ( 1
4 ( 0.3
-7 ( 8
41 ( 12
32 ( 6
46 ( 7
TABLE 5: Relative Errors of the Quantum Yield of
Dissociation
error source
relative error
photolysis rate (d[Prod]/dt])
actinometry (I0)
adsorption cross section (σ)
concentration of MG ([MG])
(7%
(10%
(6%
(6%
5
dark loss of MG (kdark
)
(
%
0
Figure 12. Quantum yields of R10 as a function of MG concentration
(380 nm: P ) 400 Torr, [O2] ) 90 Torr; 440 nm: P ) 760 Torr, [O2]
) 150 Torr). Symbols: calculated from experimental data according
to eq 16; curves: fit according to eq 23.
coefficients (δ, ꢀ)
optical resolution (∆λ)
quantum yield (φD)
(5%
(06%
(17%
type
18, the quantum yield of photodissociation can be specified as
a function of wavelength and pressure (Figure 10), as required
for the calculation of the lifetime of MG. In the range between
335 and 345 nm, these functions were interpolated to form a
smooth transition to φD ) 1 of the short-wavelength band.
k
MG[MG]
KMG
φR ∼
)
(19)
k∑MG[MG] + ki 1 + k′i/[MG]
where (kMG)[MG] represents the effective rate of R10, k∑MG
the sum of all quenching processes dependent on [MG], and ki
all deactivation processes obeying first- as well as pseudo-first-
order kinetics. The ratio kMG/k∑MG ) KMG gives the limit of
the yield at high concentrations.
The effect of O2 partial pressure on φR was investigated under
conditions of constant MG partial and constant total pressure
at 380 and 440 nm. The results shown in Figure 13 can also
be represented by a saturation function,
The relative standard deviations of the quantum yields as a
function of wavelength were determined to be 7%, possible
systematic errors introduced by uncertainties of the coefficients
δ and ꢀ are thought to be less than 5%. The finite optical
resolution (8.5 nm) leads to an overestimation of the quantum
yields of not more than 6%. Summarizing all statistical as well
as systematical errors, shown in Table 5, results in a relative
uncertainty of the quantum yields determined of ∆φD/φD
(17%.
)
k [O2]
K
O2
Quantum Yields of Reaction R10. The quantum yields φR
of the reaction of electronically excited MG with ground-state
MG (R10) were calculated from experimental data using eq 16.
O2
φR ∼
)
(20)
k∑O [O2] + kT 1 + kT′ /[O2]
2
The results of the experiments performed in synthetic air are
displayed as a function of total pressure in Figure 11 and can
be empirically described by a function analogous to eq 20.
with KO giving the limit of φR at high O2 partial pressure. By
2
comparison of the plots in Figures 12 and 13, it can be observed
that the limits of φR reached at large [MG] and [O2], respec-
tively, show nearly the same values. The interpretation of these
results requires product-forming as well as deactivation path-
ways of the reactive state under participation of MG as well as
oxygen.
The dependence of the quantum yields φR of MG concentra-
tion, determined at 380 and 440 nm, all other experimental
conditions kept constant, is shown in Figure 12. The experi-
mental data can be represented by saturation functions of the

9150 J. Phys. Chem. A, Vol. 102, No. 46, 1998
Koch and Moortgat
TABLE 6: Quantum Yields of Intersystem Crossing
kIX
k+
k′
X
φX
)
0
λ[nm]
λ[nm]
k′ + k′
(this work)X
D
(ref 32)
0
380
400
420
440
401
420
442
0.66 ( 0.09
0.77 ( 0.11
0.99 ( 0.14
0.55 ( 0.05
0.76 ( 0.02
0.92 ( 0.01
energy as well as pressure, were obtained. In the second
absorption band (λ > 340 nm) dissociation quantum yields
decrease with increasing wavelength and increasing total
pressure, obeying the Stern-Volmer relationship (eq 11).
Further, a chemical reaction of electronically excited MG with
MG in the ground state (R10) was proposed.
Figure 13. Quantum yields of R10 as a function of O2 partial pressure
(380 nm: P ) 400 Torr, [MG] ) 0.25 Torr; 440 nm: P ) 760 Torr,
[MG] ) 0.25 Torr). Symbols: calculated from experimental data
according to eq 16; curves: fit according to eq 23.
[CH3COCHO]* + HCOCOCH3 f
CH3CO + HCHO + CO + CH3CO (R10)
This mechanism may be discussed with respect to the previously
published studies dealing with the photophysics of MG of van
der Werf et al.32 and Coveleskie and Yardley.28,33,34 They
investigated the fluorescence and phosphorescence of MG at
excitation wavelengths between 398 and 455 nm. Their results
relevant to this work are summarized as follows:
(i) Methylglyoxal in its first electronically excited state (S1)
undergoes fast intersystem crossing to a triplet state with rates
in the order of magnitude of 107 s-1 32,34
.
(ii) The collisionless quantum yields of intersystem crossing
(kST/k+, with kST representing the rate coefficient of intersystem
crossing and k+ the sum of all monomolecular deactivation
processes) obtained by van der Werf et al.32 are in excellent
agreement with the quantum yields of monomolecular deactiva-
tion extrapolated to zero pressure (φX ) kX/(kX + kD) ) 1 -
0
φ0, Table 6). Hence kD and kX can be identified as rate
coefficients of dissociation and intersystem crossing to a
nondissociative triplet state, respectively. Moreover, these
processes can be regarded as the only significant monomolecular
dissipation pathways of vibrationally excited MG in the S1 state.
(iii) The results at 380 nm must be regarded as somewhat
different. The quantum yield of dissociation, extrapolated to
zero pressure (φ0), is unity. Since it cannot be expected that
intersystem crossing becomes negligible at this wavelength,35
it is concluded that the corresponding triplet state is also
dissociative.
Figure 14. Quantum yields of R10 as a function of total pressure
obtained under constant O2 and MG partial pressure. (380 nm: [O2] )
150 Torr, [MG] ) 0.20 Torr; 440 nm: [O2] ) 150 Torr, [MG] ) 0.25
Torr). Symbols: calculated from experimental data according to eq
16; curves: fit according to eq 23.
Quantum yields as a function of total pressure, at constant
MG and O2 partial pressure, are displayed in Figure 14. Within
experimental error no pressure effect could be observed.
Therefore the pressure effect observed in the experiments
performed in synthetic air (Figure 11) has to be interpreted as
an O2 partial pressure dependence.
The errors indicated by the bars in Figures 11-14 represent
the weighted sum of ∆(d[CO]/dt) and ∆(d[HCHO]/dt). Sys-
tematic errors caused by the uncertainties of δ and ꢀ do not
change the calculated values of φR by more than 10%, while
the shape of the curves remains unaffected.
(iv) Collisional quenching of vibrationally excited triplets (cq
≈ 107 s-1 Torr-1) is orders of magnitudes faster than the
competing first-order processes (kT < 106 s-1).32
(v) No experimental data concerning the triplet quenching
of MG by oxygen is available. Nevertheless, the thermalized
phosphorescence of the homologous biacetyl is strongly quenched
by O2 with a rate of 2.8 × 104 s-1 Torr-1 36
.
Hence, triplet MG
In the following section these experimental findings are
discussed with respect to the photophysics of dicarbonyl
compounds.
must also be considered to be efficiently electronically deacti-
vated by oxygen.
(vi) Intersystem crossing from the thermalized S1 state
therm
) 4.9 × 107 s-1) is orders of magnitudes faster than
Photochemical Model
(k
IX
fluorescence (kFluorescence ≈ 105 s-1).33
The photochemistry of MG in the presence of O2 is described
by a photodissociation mechanism according to P1′.
Following these arguments, the conclusions can be drawn
that under the experimental conditions employed in this work,
all molecules in triplet states are thermalized and all nondis-
sociating molecules undergo intersystem crossing, reaching the
thermalized triplet states. Therefore, the states responsible for
the reaction of electronically excited MG should be thermalized
triplets and the experimentally determined quantum yield of this
hν8 O ,M8 CH3COO2 + HO2 + CO (P1′)
2
CH3COCHO
9
For absorptions in the short-wavelength band (λ < 340 nm)
dissociation quantum yields of unity, independent of photon

Photochemistry of Methylglyoxal in the Vapor Phase
J. Phys. Chem. A, Vol. 102, No. 46, 1998 9151
TABLE 7: Coefficients of the Quantum Yield of R10a
reaction should be proportional to the triplet quantum yield (φT),
which is equivalent to the quantum yield of nondissociation (1
- φD).
λ[nm]
KR
ki′ [Torr]
k′ [Torr]
T
380
440
0.072 ( 0.01
0.040 ( 0.005
1.45 ( 0.7
40 ( 18
0.10 ( 0.05
0.05 ( 0.03
φR(λ, P) ∼ φT(λ, P)
(21)
(22)
a The assigned errors represent one standard deviation obtained by
least-squares fitting.
φT(λ, P) ) 1 - φD(λ, P)
13, and 14 can be given,
This relationship is shown by the lines in Figure 14, with the
values of 1 - φD calculated from the coefficients given in Table
4.
kO [O2]
kMG[MG]
2
φR ) (1 - φD)
(
)(
)
k∑O [O2] + kT k∑MG[MG] + ki
In order to explain the experimentally observed photon
energy, [O2] as well as [MG] dependence of the quantum yield
φR, detailed information concerning the chemical and physical
properties of the involved triplet states of MG is necessary. The
properties of triplet MG and the homologous biacetyl (CH3-
COCOCH3), respectively, were examined by several experi-
mental as well as theoretical studies. The results needed for
the discussion of this work are summarized as follows:
(i) The CO quantum yield of biacetyl at 436 nm increases
with O2 partial pressure, the yields were described by a
saturation function (φCO ) ((0.07 ( 0.03) s-1)/(1 + (0.012 (
0.002) s-1 Torr-1)/[O2]).37 A similar behavior was observed
for the CO quantum yield of acetaldehyde (313 nm).38 These
results were interpreted in terms of a bimolecular interaction
of oxygen with triplet biacetyl and acetaldehyde, respectively,
following product formation.
2
KR
) (1 - φD)
(23)
(
)
(1 + kT′ /[O2])(1 + k′/[MG])
i
where KR ) (kO /k∑O )(kMG/k∑MG) ) KO KMG, and all coefficients
2
2
2
have to be considered as functions of the wavelength as given
in Table 7.
For explanation of the observed wavelength dependence of
φR, the participation of two distinct triplet states, populated with
different energy-dependent probabilities is assumed. Therefore
the coefficients entered in Table 7 have to be considered as
some mean values of the different states involved.
The thermodynamics of R10a require triplet energies corre-
sponding to a threshold wavelength of 434 nm, whereas the
formation of acetylperoxy instead of acetyl radicals diminishes
this threshold to 935 nm.
(ii) Molecular orbital considerations on the basis of biacetyl
with C2h-symmetry show the existence of two triplet states that
can be reached by intersystem crossing from the first excited
singlet. The first triplet, lying about 3000 cm-1 below the
[CH3COCHO]* + HCOCOCH3 f
CH3CO + HCHO + CO + CH3CO (R10a)
I
II
3
3
singlet,33 is assigned to be of Au, the second of Bg sym-
O2
98
metry.27
[CH3COCHO]* + HCOCOCH3
(iii) However, there exists some disagreement concerning the
energy differences of these states. Drent et al. assume the
energy difference of both states to be as small as 2500 cm-1
CH3COO2 + HCHO + CO + CH3CO (R10b)
Therefore, the participation of oxygen might thermodynamically
open the product formation channel (R10b). However, a
mechanistic interpretation might be too speculative. A further
confirmation of the proposed reaction mechanism can be
obtained by experiments using partially deuterated methylgly-
oxal (CH3COCDO). Reaction R10 yields exclusively deuterated
formaldehyde (DCDO) if the exchange of deuterium between
ground-state molecules is sufficiently slow. The reaction rate
is expected to be reduced according to the kinetic isotope effect.
II
3
(29 kJ/mol),39 while Coveleskie and Yardley estimate the Bg
state lying at 27000 cm-1, inaccessible for excitation wave-
lengths longer than 370 nm.33
(iv) An experimental indication to the population of two
different triplet states can be taken from the biexponential time
profiles of the slow fluorescence of MG, initiated by reverse
intersystem crossing which was observed at all wavelengths
investigated.32
The experimentally obtained dependence of φR on MG and
oxygen partial pressure as well as photon energy can be brought
into consistency with the known physical properties of elec-
tronically excited MG by the photochemical model described
below.
The increase of φR with O2 partial pressure (Figure 13, eq
20) requires the participation of O2 in product formation. We
propose that a minor portion of the collisions of triplet MG
Comparison to Recent Investigations
The quantum yields of CO and HCHO determined in this
work agree only qualitatively with those obtained by Raber and
Moortgat18,19 and Staffelbach et al.,20 respectively, as outlined
in Table 8.
Raber and Moortgat18,19 observed nonlinear product-
concentration time profiles. The ratio ∆[CO]/∆[MG] increased
linearly with photolysis time, while the ratio ∆[HCHO]/∆[MG]
slightly decreased with time (Table 8). These observations could
not be explained by their kinetic model, consisting of the
reactions as given in Table 2 with slightly different rate constants
and reaction of MG with HO2 (R11) included.
with O2 leads to some reactive intermediate (kO ), not identified,
2
undergoing further collisions with MG, yielding the reaction
products according to R10 (kMG). This assumes that the major
amount of triplets is electronically deactivated by oxygen (k∑O
)
2
or other first-order kinetic processes (kT) like phosphorescence,
internal conversion, or intersystem crossing.
In order to explain the [MG] dependence of φR as displayed
in Figure 12 (eq 19), besides a product promoting pathway via
R10 (kMG), competitive deactivation of the assumed intermediate
by first-order (ki) as well as MG dependent (k∑MG) kinetic
processes is suggested.
Following these considerations, an analytical expression for
φR which combines eqs 19-22 as displayed in Figures 12,
CH3COCHO + HO2 h CH3COCHO-O2H f Products
(R11)
Possibly, secondary chemical processes occurred, caused by
higher concentrations of radical and molecular intermediates,
since their photon flux was about two orders of magnitude
higher than in this work. Wall effects are not mentioned.

9152 J. Phys. Chem. A, Vol. 102, No. 46, 1998
Koch and Moortgat
TABLE 8: Comparison to Recent Studies of the Photochemistry of MG
Raber18
P [Torr]
∆[CO]/∆[MG]
∆[HCHO]/∆[MG]
φMG
φD
a
d
λ ) 275-380 nm
54
146
220
400
600
760
57
148
600
760
0.85-1.6
0.85-1.3
1.00-2.0
0.90-1.1
0.30-1.0
0.65-0.9
0.63-1.0
0.44-0.55
0.18-0.35
0.28-0.31
0.28-0.22
0.35-0.22
0.31-0.18
0.38-0.30
0.27-0.20
0.28-0.21
0.20-0.16
0.32-0.22
0.28-0.21
0.35-0.22
1.11 ( 0.03
1.07 ( 0.01
0.99 ( 0.05
0.95 ( 0.05
0.86 ( 0.04
0.72 ( 0.02
0.51 ( 0.04
0.31 ( 0.03
0.26 ( 0.03
0.26 ( 0.04
0.94 ( 0.04
0.90 ( 0.03
0.87 ( 0.05
0.82 ( 0.04
0.75 ( 0.03
0.64 ( 0.03
0.41 ( 0.04
0.28 ( 0.01
0.22 ( 0.02
0.23 ( 0.02
λ ) 390-470 nm
Staffelbach20
P [Torr]
∆[CO]/∆[MG]
∆[HCHO]/∆[MG]
φMG
φD
a
d
λ ) 410-418 nm
λ ) 355-480 nm
λ ) 240-420 nm
760
760
760
b
b
2(φMG - φR11
)
0.005
0.055
0.14
<0.78
<1.44
0.51 ( 0.06
0.30 ( 0.06
a
this work
P [Torr]
φCO′/φD
φHCHO/φD
φMG
φCO′ - φR
0.013 ( 0.03
all λ
λ ) 420 nm
all P
760
1.0 ( 0.1
0.19 ( 0.02
c
a φMG ) -(d[MG]/dt)/(∫I0 dλ σ [MG]). b Not specified. c Not determined. d φD ) (φMG - φR11).
results would be compatible if one would assume that reaction
R11 is negligibly slow, in contradiction to the findings of
Staffelbach et al. There is no explanation for these discrepan-
cies. However, a direct determination of the rate coefficient of
R11 would give a deeper understanding of the photochemistry
of MG.
Atmospheric Lifetime
The atmospheric photodissociation coefficient (J) is defined
as the integral of the solar flux per unit wavelength (I0) as a
function of height (z), the solar zenith angle (θ), the absorption
cross section (σ), and the quantum yield (φ) of the considered
process over the atmospherically relevant wavelength range. The
atmospheric lifetime due to photolysis (τphot) is the reciprocal
of the photodissociation coefficient.
Figure 15. Atmospheric lifetime of MG due to photolysis as a function
of solar zenith angle (z ) 0.5 km, P ) 760 Torr, albedo ) 0.2).
J(P, T, z, θ) ) I (λ, z, θ) σ(λ, P, T) φ(λ, P, T) dλ
∫
0
Staffelbach et al.20 studied the wavelength dependence of the
photolysis using broad-band cut-off filters specified by their
threshold wavelengths. The pressure dependence of the quan-
tum yield was not investigated, all experiments were performed
at 760 Torr. The CO quantum yields determined have to be
considered as upper limits. The results of their model calcula-
tions could not be reproduced with the given rate coefficients;
e.g., the model calculations resulted in a carbon balance of
100%, although the postulated product CH3O2H was not
included and HCHO yields higher than 60% per molecule of
MG removed were calculated.
Their experimental results are not in agreement with this
work. Their ratio ∆[CO]/∆[MG] is significantly lower and ∆-
[HCHO]/∆[MG] significantly higher than the quantum yields
determined in this study (Table 8).
In a further experiment Staffelbach et al.20 investigated the
photolysis of a mixture of H2, O2, N2, MG, and Cl2 in order to
determine the rate of the reaction of MG with HO2 (R11). The
examination of the stable products resulted in an effective rate
τphot ) J-1
The radiation model is related to that of Luther and Gelinas.40
The solar photon flux is incrementally calculated for a standard
atmosphere with respect to the extraterrestrial spectrum, absorp-
tions due to N2, O2, O3, and NO2, multiple scattering, and a
constant albedo. The calculations of the photodissociation
coefficient were performed for the height increment between 0
and 1 km (760 Torr) and an albedo of 0.2, using the absorption
cross sections given by Meller et al.10 and the quantum yields
of dissociation given by eqs 11, 17, and 18 as shown in Figure
10. Atmospherically, only photodissociation of MG is relevant,
the suggested reaction of triplet CH3COCHO is of no importance
due to the low concentrations of reactants. The results of the
model calculations for z ) 0.5 km as a function of the zenith
angle are shown in Figure 15. For an angle of 50° one obtains
a lifetime due to photolysis of
τphot ) 4.1 h ( 0.7 h
constant of k11 ) (1.4 ( 0.5) × 10-15 cm3 molecule-1 s-1
,
leading to the conclusion that during the photolysis of MG half
of the removal of MG is the result of the reaction with HO2
(R11). Their photolysis quantum yield obtained in the wave-
length range between 410 and 418 nm is about half as large as
the CO quantum yield found at 420 nm in this work. Both
with respect to the error of the quantum yield φD.
The lifetime due to OH reaction is calculated to be 16 h for
an OH concentration of 1 × 106 molecules cm-3 and a room-
temperature rate coefficient of k3 ) 1.3 × 10-11 cm3 molecule-1
s-1 21
Hence, OH reaction is of minor importance for the
.

Photochemistry of Methylglyoxal in the Vapor Phase
J. Phys. Chem. A, Vol. 102, No. 46, 1998 9153
(17) Lee, Y. N.; Zhou, X.; Hallock, K. J. Geophys. Res. 1995, 100,
25933.
degradation of tropospheric MG. The photolysis lifetime of 2
h given by Plum et al.23 is based on experiments using an
unfiltered Xe arc lamp in order to simulate the solar spectrum.
Raber and Moortgat19 reported a value of 0.6 h based on a
wavelength-independent quantum yield of 0.23, obtained with
a light source showing a broad emission between 390 and 470
nm. Their results are possibly affected by systematic errors.
The results of Staffelbach et al. (2.7 h ( 0.7 h)20 are in good
agreement with those of this work. They assume a quantum
yield of dissociation linearly decreasing with wavelength (φD-
(300 nm) ) 0.45, φD(430 nm) ) 0), assuming that only about
half of the MG removed in their experiments is due to
photolysis. This function results in higher yields in the
atmospherically effective window between 420 and 370 nm.
Finally, in order to complete our understanding of the fate
of atmospheric MG, information concerning its dry and wet
deposition is necessary.
(18) Raber, W. PhD. Thesis, Johannes Gutenberg-University, Mainz,
Germany, 1992.
(19) Raber, W.; Moortgat, G. K. In Progress and Problems in
Atmospheric Chemistry; Barker, J., Ed.; Singapore, 1996; p 318.
(20) Staffelbach, T. A.; Orlando, J. J.; Tyndall, G. S.; Calvert, J. G. J.
Geophys. Res. 1995, 100, 14189.
(21) Tyndall, G. S.; Staffelbach, T. A.; Orlando, J. J.; Calvert, J. G. Int.
J. Chem. Kinet. 1995, 27, 1009.
(22) DeMore, W. B.; Sander, S. P.; Golden, D. M.; Molina, M. J.;
Hampson, R. F.; Kurylo, M. J.; Howard, C. J.; Ravishankara, A. R.; Kolb,
C. E. Chemical Kinetics and photochemical data for use in stratospheric
modelling, Evaluation number 11, JPL Publication 94-26, 1994.
(23) Plum, C. N.; Sanhueza, E.; Atkinson, R.; Carter W. P. L.; Pitts, J.
N., Jr. EnViron. Sci. Technol. 1983, 17, 479.
(24) Atkinson, R.; Baulch, D. L.; Cox, R. A.; Hampson, R. F., Jr.; Kerr,
J. A.; Rossi, J. M.; Troe, J. J. Phys. Chem. Ref. Data 1997, 26, (3), 521-
1011.
(25) Wright, B. W.; Lee, M. L.; Graham, S. W.; Phillips, L. V.; Hercules,
D. M. J. Chromatogr. 1980, 199, 355.
(26) Maric, D.; Burrows, J. P.; Meller, R.; Moortgat, G. K. J. Photochem.
Photobiol., A 1993, 70, 205.
(27) Drent, E.; Kommandeur, J. Chem. Phys. Lett. 1972, 14, 321.
(28) Coveleskie, R. A.; Yardley, J. T. Chem. Phys. 1975, 9, 277.
(29) Facsimile, Release H010, Version 101010, United Kingdom Atomic
Energy Authority, Harwell, 1987.
(30) Benson, S. W. Thermochemical Kinetics; New York, 1976.
(31) McMillan, G. R.; Calvert, J. G.; Pitts, J. N., Jr. J. Am. Chem. Soc.
1994, 86, 3602.
References and Notes
(1) Killus, J. P.; Whitten, G. Z. EnViron. Sci. Technol. 1984, 18, 142.
(2) Lloyd, A. C.; Atkinson, R.; Lurmann, F. W.; Nitta, B. N. EnViron.
Sci. Technol. 1983, 17, 1931.
(3) Paulson, S. E.; Flagan, R. C.; Seinfeld, J. H. Int. J. Chem. Kinet.
1992, 24, 79.
(4) Kamens, R. M.; Gery, M. W.; Jeffries, H. E.; Jackson, M.; Cole,
E. I. Int. J. Chem. Kinet. 1982, 14, 955.
(32) van der Werf, R.; Schutten, E.; Kommandeur, J. Chem. Phys. 1976,
16, 151.
(5) Niki, H.; Maker, P. D.; Savage, C. M.; Breitenbach, L. P. EnViron.
Sci. Technol. 1983, 17, 312 A.
(33) Coveleskie, R. A.; Yardley, J. T. J. Am. Chem. Soc. 1975, 97, 1667.
(34) Coveleskie, R. A.; Yardley, J. T. Chem. Phys. 1976, 13, 441.
(35) van der Werf, R.; Kommandeur, J. Chem. Phys. 1976, 16, 125.
(36) Kaskan, W. E.; Duncan, A. B. F. J. Chem. Phys. 1950, 18, 427.
(37) Padnos, N.; Noyes, W. A. J. Phys. Chem. 1964, 68, 464.
(38) Weaver, J.; Meagher, J.; Heicklen, J. J. Photochem. 1976, 6, 111.
(39) Drent, E.; van der Werf, R.; Kommandeur, J. J. Phys. Chem. 1973,
59, 2061.
(40) Luther, F. M.; Gelinas, R. J. J. Geophys. Res. 1976, 81, 1125.
(41) Bauer, D.; Crowley, J. N.; Moortgat, G. K. J. Photochem.
Photobiol., A 1992, 65, 329.
(42) Dagaut, P.; Wallington, T. J.; Kurylo, M. J. J. Phys. Chem. 1988,
92, 3836.
(6) Paulson, S. E.; Flagan, R. C.; Seinfeld, J. H. Int. J. Chem. Kinet.
1992, 24, 79.
(7) Tuazon, E. C.; Atkinson, R. Int. J. Chem. Kinet. 1990, 22, 591.
(8) Tuazon, E. C.; Atkinson, R. Int. J. Chem. Kinet. 1989, 21, 1141.
(9) Atkinson, R.; Aschmann, S. M.; Winer, A. M.; Pitts, J. N., Jr. Int.
J. Chem. Kinet. 1981, 13, 1133.
(10) Meller, R.; Raber, W.; Crowley, J. N.; Jenkin, M. E.; Moortgat, G.
K. J. Photochem. Photobiol., A 1991, 62, 163.
(11) Jenkin, M. E.; Cox, R. A.; Emrich, M.; Moortgat, G. K. J. Chem.
Soc., Faraday Trans. 1992, 89, 2983.
(12) Tuazon, E. C.; Atkinson, R.; MacLeod, H.; Biermann, H. W.; Winer,
A. M.; Carter, W. P. L.; Pitts, J. N., Jr. EnViron. Sci. Technol. 1984, 18,
981.
(13) Tuazon, E. C.; MacLeod, H.; Atkinson, R.; Carter, W. P. L. EnViron.
Sci. Technol. 1986, 20, 383.
(43) Wallington, T. J.; Japar, S. M. Chem. Phys. Lett. 1990, 166, 495.
(44) Kaiser, E. W.; Rimai, L.; Wallington, T. J. J. Phys. Chem. 1989,
93, 4094.
(45) Roehl, C. M.; Bauer, D.; Moortgat, G. K. J. Phys. Chem. 1996,
100, 4038.
(14) Seuwen, R.; Warneck, P. Int. J. Chem. Kinet. 1996, 28, 315.
(15) Lightfoot, P. D.; Cox, R. A.; Crowley, J. N.; Destiau, M.; Hayman,
G. D.; Jenkin, M. E.; Moortgat, G. K.; Zabel, F. Atmos. EnViron. 1992,
26A, 1805.
(16) Munger, J. W.; Jacob, D. J.; Daube, B. C.; Horowitz, L. W.; Keene,
W. C.; Heikes, B. G. J. Geophys. Res. 1995, 100, 9325.
(46) Bauer, D. PhD. Thesis, Johannes Gutenberg-University, Mainz,
Germany, 1996.