Atmospheric degradation of glycidaldehyde: Photolysis and reaction with OH radicals

Article abstract of DOI:10.1021/es9804159

Epoxide aldehydes have recently been detected among the oxidation products of aromatic hydrocarbons. Many epoxides are toxic and very little is known about their atmospheric fate. The products and kinetics of the atmospheric oxidation, OH radical reaction, and photolysis of glycidaldehyde have been investigated in a large volume reactor at 298 K using in situ long- path FT-IR spectroscopy for the analysis. A rate coefficient of k = (1.69 ± 0.04) x 10^-11 cm³ molecule^-1 s^-1 has been determined for the reaction of glycidaldehyde with the OH radical using the relative kinetic technique. The UV absorption spectrum of glycidaldehyde was measured in the range 220380 nm from which upper limits of its photolysis frequencies in the troposphere have been deduced, e.g., J (hv) ~ 1.0 x 10^-4 s^-1 (for July 1, noon, and 50°N). The OH radical initiated photooxidation of glycidaldehyde yields CO, CO₂, formic acid, formic acid anhydride, formaldehyde, and hydroperoxymethyl formate as major products. A reaction mechanism is postulated to account for the product formation. Epoxide aldehydes have recently been detected among the oxidation products of aromatic hydrocarbons. Many epoxides are toxic and very little is known about their atmospheric fate. The products and kinetics of the atmospheric oxidation, OH radical reaction, and photolysis of glycidaldehyde have been investigated in a large volume reactor at 298 K using in situ long-path FT-IR spectroscopy for the analysis. A rate coefficient of k = (1.69 ± 0.04) × 10^-11 cm³ molecule^-1 s^-1 has been determined for the reaction of glycidaldehyde with the OH radical using the relative kinetic technique. The UV absorption spectrum of glycidaldehyde was measured in the range 220-380 nm from which upper limits of its photolysis frequencies in the troposphere have been deduced, e.g., J(hv) approx. 1.0 × 10^-4 s^-1 (for July 1, noon, and 50° N). The OH radical initiated photooxidation of glycidaldehyde yields CO, CO₂, formic acid, formic acid anhydride, formaldehyde, and hydroperoxymethyl formate as major products. A reaction mechanism is postulated to account for the product formation.

Full text of DOI:10.1021/es9804159

Environ. Sci. Technol. 1998, 32, 3515-3521
Atmospheric Degradation of
Glycidaldehyde: Photolysis and
Reaction with OH Radicals
S . M A , I . B A R N E S , * A N D K . H . B E C K E R
Physikalische Chemie, Fachbereich 9, Bergische Universita¨t,
Gaussstrasse 20, D-42097 Wuppertal, Germany
The method has been simplified on two points. First, the
reaction mixture was lyophilized overnight to reduce the
amount of water instead of using a circulating evaporator.
Second, the product was extracted in dichloromethane, and
the organic phase was distilled under reduced pressure. This
variation replaced the extraction with warm cyclohexanone
and the time-consuming rectification on an Oldershaw
column. The NMR and infrared spectra showed that the
product was essentially pure. A GC-MS determination of
the purity was unsuccessful. The density of glycidaldehyde
was determined at room temperatue by weighing different
volumes of the substance in a calibrated injection syringe
and making a linear regression. The measurements gave a
Epoxide aldehydes have recently been detected among
the oxidation products of aromatic hydrocarbons. Many
epoxides are toxic and very little is known about their
atmospheric fate. The products and kinetics of the
atmospheric oxidation, OH radical reaction, and photolysis
of glycidaldehyde have been investigated in a large
volume reactor at 298 K using in situ long-path FT-IR
spectroscopy for the analysis. A rate coefficient of k )
(1.69 ( 0.04) × 10-11 cm3 molecule-1 s-1 has been determined
for the reaction of glycidaldehyde with the OH radical
using the relative kinetic technique. The UV absorption
spectrum of glycidaldehyde was measured in the range 220-
380 nm from which upper limits of its photolysis frequencies
in the troposphere have been deduced, e.g., J (hν) ∼
1.0 × 10-4 s-1 (for July 1, noon, and 50° N). The OH radical
value of d20°C ) (1.07 ( 0.05) g cm-3
.
OH-Kinetic and Product Studies. The experiments were
performed in a 1080-L quartz chamber surrounded with 32
low-pressure mercury vapor lamps (Philips TUV/ 40W; Λmax
) 254 nm) and 32 fluorescent lamps (Philips TL05/ 40W; Λmax
) 360 nm). The reactor was equipped with a multiple
reflection mirror system (White system) interfaced to a
BRUKER IFS88 FTIR-spectrometer (total optical path length
579 m). A relative kinetic technique was applied to determine
the OH reaction rate constant at 1000 mbar total pressure
of synthetic air at 298 ( 2 K. Ethene with a rate coefficent
of kOH ) 8.5 × 10-12 cm3 molecule-1 s-1(4) was used as the
reference substance. Details of this method can be found
elsewhere (5). The method depends on reaction with OH
radicals being the only loss process for the reactants:
initiated photooxidation of glycidaldehyde yields CO, CO ,
2
formic acid, formic acid anhydride, formaldehyde, and
hydroperoxymethyl formate as major products. A reaction
mechanism is postulated to account for the product
formation.
Introduction
Glycidaldehyde is the simplest epoxide aldehyde and has
been classified as an animal carcinogen substance based on
long-term in vitro animal carcinogenicity studies and in-
halation studies in humans (1). Recently glycidaldehyde
along with other epoxides has been identified by means of
GC-MS as products in the OH radical initiated degradation
of benzene and toluene (2). Although the yields were not
determined and may not be particularly high, their impor-
tance in atmospheric chemistry arises principally from their
high toxicity. The atmospheric chemistry of epoxides has
received very little attention and to our knowledge nothing
is presently known about the atmospheric fate of glycidal-
dehyde.
k1
OH + glycidaldehyde
98 products
k2
OH + ethene
98 products
The initial concentrations of glycidaldehyde and ethene were
in the range (3-6) × 1013 molecules cm-3 and the photolysis
of ∼1014 molecules cm-3 of H2O2 with the UV lamps was
employed as the OH radical source:
H2O2 + hν (λ ) 254 nm) f 2OH
Mixtures of glycidaldehyde/ ethene/ H2O2/ air were photolyzed
for approximately 30 min, and the degradation of the
compounds was monitored in situ using FTIR spectrometry
In this study glycidaldehyde has been synthesized and
the kinetics and products of its OH radical initiated degrada-
tion under atmospheric conditions investigated. The UV
spectrum of glycidaldehyde has also been recorded in an
attempt to access the relative importance of its loss via
reaction with OH radicals compared to photolysis.
with a resolution of 1 cm-1
.
Experiments were carried out to test for possible con-
tributions to the measured glycidaldehyde decay rate under
the experimental conditions employed through wall loss and
photolysis, such losses for ethene are known to be negligible.
The wall loss of glycidaldehyde was very low (<10-5 s-1). In
the tests for contributions due to photolysis a large excess
of propene was added to a mixture of glycidaldehyde and air
the purpose of which was to scavenge any OH radicals which
might be produced in the reactor, e.g. from reactions of
impurities at the reactor wall, and thus ensure that the
measured decay was actually due to photolysis. The levels
Experimental Section
Synthesis of Glycidaldehyde. Glycidaldehyde was prepared
on the basis of a method described in the literature (3) through
aqueous oxidation of acrolein by H2O2 under strictly con-
trolled pH conditions (8.0-8.5) at a temperature between 25
and 30 °C:
9
10.1021/es9804159 CCC: $15.00
Published on Web 09/18/1998
1998 Am erican Chem ical Society
VOL. 32, NO. 22, 1998 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 3 5 1 5

FIGURE 2. UV absorption cross section of glycidaldehyde (base e)
in 1000 mbar of N2 at 298 ( 2 K.
FIGURE 1. Plot of the experimental kinetic data according to eq
I for measurements on the reaction of OH radicals with glycidal-
dehyde relative to ethene at 298 ( 2 K in 1000 mbar of synthetic
air.
TABLE 1. Absorption Cross Section of Glycidaldehyde and
Actinic Flux Data for Noon, July 1 at a Latitude of 50° N
of “rogue” OH radical production were found to be very low
and no difference was found between experiments performed
with and without addition of propene. Tests showed,
however, the photolysis of glycidaldehyde to be nonnegligible
and consequently this loss process had to be taken into
account in the kinetic analysis. The following equation has
σ(λ) × 1020
J(λ) × 10-14
σ(λ)J(λ) × 106
λ (nm)
(cm2 molecule-1
)
(photons cm-2 s-1
)
(s-1
)
290-295
295-300
300-305
305-310
310-315
315-320
320-325
325-330
330-335
335-340
340-345
345-350
350-355
355-360
360-365
6.3268
6.6389
6.6238
6.2510
5.6155
4.5325
3.7464
2.2647
1.6399
0.9288
0.2632
0.08229
0.04157
0.01311
0.01901
0.0
0.0
0.022
0.269
1.074
2.591
3.720
4.754
6.921
7.388
7.341
7.949
7.963
9.049
8.356
9.273
0.146
1.782
6.714
14.550
16.861
17.810
15.674
12.116
6.818
2.092
0.655
0.376
0.110
0.176
kp
glycidaldehyde + hν (λ ) 254 nm)
98 products
been used to evaluate the kinetic data:
[glycidaldehyde]t
[ethene]t
0
k
0
ln
- kp
)
1 ln
(I)
t
(
)
(
)
k2
[glycidaldehyde]t
[ethene]t
where [glycidaldehyde]t and [ethene]t represent the con-
0
0
centrations of glycidaldehyde and ethene at the initial time
t0; [glycidaldehyde]t and [ethene]t represent the concentra-
tions of glycidaldehyde and ethene at the reaction time t, k1
and k2 are the rate coefficients for the reactions of OH with
glycidaldehyde and ethene, respectively. kp is the photolysis
frequency of glycidaldehyde and also includes a small
sum
95.880
diode array detector. The UV and IR light sources were a
deuterium lamp and a globar, respectively. The optical
pathlengths in the UV and IR were preadjusted to 39.1 and
51.6 m, respectively. The wavelength scale was calibrated
using atomic emission lines from a low-pressure mercury
lamp.
The infrared absorption-concentration behavior of gly-
cidaldehyde was calibrated in the reactor by means of precise
vapor pressure measurements using a Baratron pressure
gauge and the absorption peak at 842.5 cm-1. After gaseous
glycidaldehyde had been added, the reactor was filled with
N2 to 1000 mbar, and the absorbance of the peak at 842.5
cm-1 was measured. The infrared spectra were recorded
with a Nicolet 520 FT-IR spectrometer using a resolution of
1 cm-1. A plot of absorbance against concentration (mol-
ecules per cubic centimeter) resulted in straight line in
contribution due to wall loss. Plots of ln([glycidaldehyde]t /
0
t
[glycidaldehyde]t) - kp against ln([ethene]t / [ethene]t) re-
0
sulted in straight lines with a slope of k1/ k2 and zero intercept.
The spectra for the product analyses were recorded under
conditions similar to those given above but without the
inclusion of ethene in the reactant mixture.
UV Absorption Spectrum . The UV absorption spectrum
of glycidaldehyde was measured in a 480-L volume reactor
equipped with multiple reflection mirror optics for simul-
taneous in situ measurement in the UV and IR. The UV
spectrometer consists of a modified 22-cm monochromator
(SPEX) and a diode array detector (EG & G PAR 1412). A
spectral resolution of 0.6 nm was achieved with a grating of
1200 lines/ mm covering a spectral range of 70 nm on the
9
3 5 1 6 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 32, NO. 22, 1998

TABLE 2. Product Yields of the OH Initiated Oxidation of
Glycidaldehyde at 298 ( 2 K in 1000 mbar of Synthetic Air in
the Absence of NO as Function of Time
x
reaction
time
(min)
yield, %
FAN HCOOH HCHO CO a
CO
HPMF ketene
2
3.12 43 ( 6 23 ( 3 22 ( 4 20 11 ( 2
6
7
4
3
1
1
<1
<1
10.9
29 ( 4 39 ( 3 15 ( 2 46 45 ( 4
20.35 15 ( 4 47 ( 3 11 ( 2 56 68 ( 4
33.20
6 ( 3 55 ( 3 9 ( 2 69 86 ( 9
a
Yields of CO2 are subject to a high degree of uncertainty due to
fluctuations in the background levels in the gas used to purge the
external optical in the experim ental setup.
FIGURE 3. Gas-phase infrared absorption spectrum of glycidal-
dehyde in 1000 mbar of synthetic air at 298 ( 2 K.
FIGURE 5. Concentration-time profiles of the products observed
in the reaction of OH radicals with glycidaldehyde at 298 ( 2 Kin
1000 mbar of synthetic air.
experiments. Reasonable linearity is observed. Combined
photolysis and wall losses contributed e 3% to the decay of
glycidaldehyde under the conditions employed. Using a
value of k2 ) 8.52 × 10-12 cm3 molecule-1 s-1 for the reference
hydrocarbon results in a rate coefficient of k1 ) (1.69 ( 0.04)
× 10-11 cm3 molecule-1 s-1 for the reaction of OH radicals
with glycidaldehyde. The error is the 2σ standard deviation
of the linear regression fit and only reflects the precision of
the measurements. This work is the first determination of
a rate coefficient for the reaction of OH with glycidaldehyde.
Taking, e.g. a global mean value of (9.7 ( 0.6) × 105 cm-3 for
the OH radical concentration (6, 7) would imply a global
atmospheric lifetime τ of ∼17 h for glycidaldehyde through
reaction with OH radicals.
In common with other unsaturated expoxides the reaction
of OH radicals with glycidaldehyde is expected to proceed
via a H atom abstraction (8, 9). Glycidaldehyde has 4 H
atoms from which H atom abstraction by OH radicals can
occur, one attached to the aldehyde group and the other
three situated on the epoxide ring. The rate coefficients for
the reactions of OH with ethene oxide, formaldehyde,
acetaldehyde, and propanal are 1.13 × 10-13 (8-11), 0.94 ×
10-11 (4), 1.58 × 10-11 (4), and 1.96 × 10-11 (4), respectively.
Comparing these rate coefficients, the rate coefficient for H
atom abstraction from the epoxide ring is approximately 100
times slower than H atom abstraction from an aldehyde
FIGURE 4. The upper trace shows a residual product spectrum
obtained from a glycidaldehyde/H O /air mixture after ∼5 min of
2
2
irradiation with 32 UV lamps and subtraction of absorptions from
glycidaldehyde, H O , and water. The middle trace shows an infrared
2
2
spectrum of formic acid anhydride, H(O)COC(O)H, in air and the
lower trace an infrared spectrum of hydroperoxymethyl formate,
HOOCH OC(O)H, in air.
2
accordance with Lambert-Beer’s law, ln(Io/ I) ) ꢀcl. A value
of the absorption coefficient ꢀ (842.5 cm-1) ) (2.45 ( 0.12)
× 10-19 cm2 molecule-1, base e (equivalent to (6.03 ( 0.30)
× 10-4 ppm-1 m-1, base e) was obtained.
Calibrated UV spectra of glycidaldehyde were obtained
by simultaneously measuring the UV and IR spectra of the
compound in the chamber at a total pressure of 1000 mbar
with a spectral resolution of 0.6 nm. The concentration of
glycidaldehyde was determined from the corresponding IR
absorption and the cross sections of glycidaldehyde were
subsequently obtained from a regression of optical depth
against concentration.
Results and Discussion
OH Kinetic Studies. Figure 1 shows a plot of the kinetic
data evaluated according to eq I using data from four
9
VOL. 32, NO. 22, 1998 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 3 5 1 7

From the absorption data an upper limit of the photolysis
frequency, J, for glycidaldehyde in the troposphere can be
estimated by using the following:
λi
J(s-1) )
σ(λ)φ(λ)I(λ)
(IV)
∑
λ)290nm
where σ(λ) is absorption cross section of glycidaldehyde
averaged over a wavelength interval ∆λ, centered at λ (base
e, in cm2 molecule-1); φ(λ) is the primary quantum yield for
the loss of glycidaldehyde averaged over the wavelength
interval ∆λ, centered at λ; I(λ) represents actinic flux (in
photons cm-2 s-1) summed over the wavelength interval ∆λ
centered at λ, at a solar zenith angle θ for July 1, noon at a
latitude of 50° N (13). Table 1 summarizes the data of this
calculation. Assuming a primary quantum yield φ(λ) ) 1
gives an upper limit of J e 0.96 × 10-4 s-1 for the photolysis
frequency of glycidaldehyde in the troposphere under the
given conditions (July 1, noon and 50° N). A similar
calculation for 40° N results in a value of J e 1.03 × 10-4 s-1
.
For comparison the importance of photolysis of glycid-
aldehyde in the atmosphere has been assessed by another
procedure. The photolysis frequency of glycidaldehyde has
been determined in 1000 mbar of synthetic air in the 1080-L
reactor by monitoring the concentration decrease of gly-
cidaldehyde on irradiation with 32 fluorescent lamps as
described in the Experimental Section. The evaluation of
the recorded spectra resulted in a first-order decay and a
photolytic rate constant of 1.9 × 10-5 s-1
. To adjust the
measured NO2 photolysis rate constant measured in the
reactor to atmosphere conditions eq V was used:
k
NO (atmosphere)
2
FIGURE 6. Proposed reaction pathways for the primary steps in the
OH-initiated degradation of glycidaldehyde in air leading to the
formation of the oxiranoxy radical.
J(atmosphere) )
kp(reactor) (V)
kNO (reactor)
2
where kNO is the photolysis rate constant for NO2 in the
2
group. The measured rate coefficient for the reaction of OH
with glycidaldehyde of k ) (1.69 ( 0.04) × 10-11 cm3
molecule-1 s-1 lies between those measured for acetaldehyde
and propanal which strongly supports that the reaction
proceeds predominantly via H atom abstraction from the
aldehyde group.
UV Absorption Cross Section and Photolysis Frequency.
The UV absorption spectrum was recorded in 4 sections:
215-280, 260-325, 310-375, and 355-420 nm. At the
present experimental sensitivity, absorptions from glycid-
aldehyde above 350 nm were below the detection level. The
various portions of the spectrum have been merged using a
procedure which compensates for any differences, for
example in concentration, in the regions of overlap. Below
250 nm and particularly below 230 nm the spectrum was
noisy due to a combination of low lamp intensity and low
sensitivity of the diode array. Therefore the result plotted
in Figure 2 shows only the region 230-375 nm. Values of
the UV absorption cross section have been calculated
according to Lambert-Beer’s law, eq III:
atmosphere at noon and in the reactor as indicated and J
represents the photolysis frequency for glycidaldehyde in
the atmosphere at noon and in the reactor as indicated. Using
a value of kNO2 (reactor) ) 2.7 × 10-3 s-1 determined under
the experimental conditions in this laboratory and a literature
value of kNO2 (atmosphere) ) (8.5 ( 0.5) × 10-3 s-1 (14) results
in an approximative value of 6.0 × 10-5 s-1 for the photolysis
frequency of glycidaldehyde at July 1, noon and 40° N. The
ratio of the values of J determined from the comparison with
NO2 photolysis rates and the UV absorption spectrum of
∼0.58 suggests that the photolysis quantum efficiency is
probably only of the order of 60%.
These first estimations of the photolysis frequency suggest
that photolysis will probably be somewhat more important
than removal with OH radicals.
Product Studies. Figure 3 shows the gas-phase infrared
spectrum of glycidaldehyde in the range of 4000 to 800 cm -1
.
In the liquid-phase spectrum of glycidaldehyde reported in
the literature (3) an absorption at 5.75 µ (1739 cm-1) has
been assigned to a carbonyl absorption and bands at 8.15 µ
(1227 cm-1) and 11.76 µ (850 cm-1) to epoxide vibrations. In
the gas-phase spectrum shown in Figure 3 the corresponding
carbonyl band is centered at 1754 cm-1 and the epoxide
I
ln 0 ) ꢀ c l
(III)
I
absorption bands are located at 1162 and 835 cm-1
.
In Figure 2, two absorption maxima are discernible. The
strongest absorption with a maximum about 300 nm is very
similar in form to the absorption features observed for simple
aldehydes (12) and also for unsaturated carbonyl compounds,
e.g., methacrolein or methyl vinyl ketone (9). The maximum
is close to those reported for the simple carbonyls such as
formaldehyde, acetaldehyde, and acetone. A weaker ab-
sorption which peaks at 230 nm is thought to be from the
epoxide ring.
Figure 4, upper trace, shows the residual product spectrum
which is obtained after ∼5 min irradiation of a glycidalde-
hyde/ H2O2/ air mixture with 32 UV lamps and subtraction of
absorptions due to glycidaldehyde, H2O2, and water. The
middle and lower traces show the infrared spectra of formic
acid anhydride (H(O)COC(O)H, hereafter abbreviated to FAN)
and hydroperoxymethyl formate (HOOCH2OC(O)H, hereafter
abbreviated to HPMF), respectively. An inspection of Figure
4, upper trace, shows the products consist of carbon
9
3 5 1 8 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 32, NO. 22, 1998

FIGURE 7. Proposed reactions of the oxiranoxy radical which account for the products in the OH-initiated oxidation of glycidaldehyde.
monoxide, carbon dioxide, formic acid, formaldehyde, FAN,
and HPMF. Careful inspection of the residual spectrum after
subtraction of carbon monoxide showed the formation of
ketene in low yields.
setup. The CO2 concentraion values given in Table 2 and
shown in Figure 5 are only intended to indicate the
concentration trend. Figure 5 shows the concentration-
time profiles for all the positively identified products in the
reaction of OH with glycidaldehyde during the entire
observation period. Error bars have been omitted in Figure
5 to preserve clarity. From Figure 5 and Table 2 it is evident
that FAN is the major product at short reaction times and
that it undergoes further oxidation as the reaction progresses
producing CO, CO2, and HCOOH.
Due to the slowness of the glycidaldehyde photolysis rate
under the present experimental conditions the photolysis
products could not be unambiguously determined. However,
the photolysis products are anticipated to be identical to
those produced in the OH-initiated oxidation since the major
pathway is expected to be loss of the aldehyde functional
group, -CHO, which is equivalent to the OH radical reaction
pathway as discussed below.
The product formation yields and estimated errors (where
appropriate) are tabulated in Table 2 for selected reaction
times. The quoted errors are estimated total errors and
include uncertainties for the subtraction procedure and the
calibration spectra. Under the conditions of this work the
error was of the order of 15% for middle concentrations and
usually somewhat higher for low concentrations. Normally
in studies of this type it is a common practice to plot the
yields of the products versus the consumption of the reactant
to obtain information on the overall mass balance. However,
such an analysis in this particular case is of very limited use
because of the instability of the major primary product and
uncertainies in the yields of the ensuing products as outlined
below.
The absolute yield of CO2 is subjected to a high degree
of uncertainty due to fluctuations in its concentration in the
air used to purge the external optics in the experimental
Mechanism of the Reaction of OH Radicals with Gly-
cidaldehyde. On the basis of the kinetic and product
information given above a reaction mechanism for the OH
9
VOL. 32, NO. 22, 1998 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 3 5 1 9

FIGURE 8. Residual product spectrum of a glycidaldehyde/H O /air mixture obtained after ∼12 min of irradiation with 32 UV lamps and
2
2
subtraction of all identified product spectra. The insert shows the spectral range 3200-3700 cm-1.
radical initiated oxidation of glycidaldehyde has been
constructed. The mechanism is outlined in Figures 6 and 7.
It is suggested that in the primary step (Figure 6), the OH
radical mainly reacts via H atom abstraction from the
aldehyde group, -CHO, to give a carbonyl radical (I) which
combines with an oxygen molecule to yield a peroxy radical
(II). Under the NOx-free conditions of the present study the
peroxy radical undergoes self-reaction or cross reactions with
other peroxy radicals to form the corresponding oxy radical,
in this case an oxiranyl carboxy radical (III). Reaction of the
peroxy radical (II) with HO2 to form an instable peroxy acid
is also a further possibility.
experimental conditions in this work glyoxal was always below
the detection limit.
Neither FAN nor HPMF is thermally stable. In the gas
phase, HPMF is known to decompose to FAN and water (16),
and FAN further decomposes to formic acid and carbon
monoxide (17). Examination of Figure 5 and Table 2 shows
a clear increase in the concentrations of formic acid and CO
as the concentration of FAN decreases. Experiments in this
laboratory on samples of pure synthesized FAN (17) have
shown that its decay can be qualitatively described by:
HC(O)OC(O)H f HCOOH + CO
The oxiranyl carboxy radical (III) is expected to lose carbon
dioxide and form an oxiranyl radical (IV). Niki et al. (15)
have investigated the Cl-initiated reaction of ethene oxide
in O2/ N2 diluent systems in which the initial step is also
formation of an oxiranyl radical. With the exception of
chlorine-containing compounds the products observed by
Niki et al. are very similar to the products observed in this
study. As in the work of Niki et al. the major fate of the oxiranyl
radical (IV) is addition of O2 to form a peroxy radical (V)
followed by further self-reactions or cross reactions to form
an oxiranoxy radical (VI). Carbon-carbon bond fission of
the oxiranyl radical (IV) can lead to the formation of ketene
as indicated in Figure 6; however, this appears to be only a
minor pathway.
however, the stoicheiometry of HCOOH and CO to another
as well as to the decay of FAN is not given. The missing
products as well as the magnitude of the observed decay are
in line with a mainly heterogeneous decomposition of FAN
at the reactor surface.
Niki et al. (15) assigned the absorption bands in their
study of Cl-initiated reaction of ethylene oxide to hydroxy-
methyl formate, CH2(OH)OC(O)H, based on a comparison
with a spectrum obtained from an ozone-ethene system.
This identification has recently been shown to be incorrect
and that the absorptions are actually due to hydroperoxy-
methyl formate (HPMF) (16-19). Thamm et al. (16) have
synthesized HPMF by the ozonolysis of ethyl vinyl ether in
solution and characterized it by NMR and mass spectrometry.
The recorded IR spectrum of the pure compound in the gas
phase has confirmed that the intermediate observed in the
gas-phase ozonolysis of ethene is HPMF and not hydroxy-
methyl formate as previously thought. It is interesting to
note that although the ozonolysis studies (16-19) have shown
that FAN and HPMF are two important products of the gas-
phase ozonolysis of ethene they are formed via a completely
different mechanisms to the OH-initiated oxidation of
glycidaldehyde, i.e., from reactions of the stablized Criegee
biradical, CH2O2.
The further reactions of the oxiranyl radical (VI), as
outlined in Figure 7, determine the final product distribution.
Two important intermediate products observed in the system,
FAN and HPMF, are formed from the alkyl-type radical, •CH2-
OCOH, which results from C-C bond scission of the
•
oxiranoxy radical (VI). The CH2OCOH radical adds O2 to
•
form a peroxy radical, OOCH2OCOH, for which there are
three possible reaction routes: (i) reaction with HO2 to form
hydroperoxymethyl formate (HPMF), (ii) self-reaction or
possibly cross reaction to form formic acid anhydride (FAN)
and hydroxymethyl formate, and (iii) self-reaction or cross
In Figure 7 the formation of hydroxymethyl formate, CH2-
(OH)OC(O)H, has been proposed as one of the possible
products of the self-reaction of the peroxy radicals, •OOCH2-
OC(O)H. Figure 8 shows the residual product spectrum of
the glycidaldehyde/ H2O2/ air system after ∼15 min of ir-
radiation with 32 UV lamps and subtraction of all the
identified product spectra including HPMF. The residual
•
reaction to form an oxy radical, OCH2OC(O)H. Further
reactions of the oxy radical will yield FAN, HCHO, and CO2
as indicated in Figure 7.
The oxiranoxy radical (VI) may also undergo C-O scission
and form glyoxal as the main product. However, this is
obviously not an important pathway since under the
9
3 5 2 0 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 32, NO. 22, 1998

(5) Bierbach, A.; Barnes, I.; Becker, K. H.; Wiesen, E. Environ. Sci.
Technol. 1994, 28, 8, 715.
(6) Prinn, R. G.; Weiss, R. F.; Miller, B. R.; Huang, J.; Alyea, F. N.;
Cunnold, D. M.; Fraser, P. J.; Hartley, D. E.; Simmond, P. G.
Science 1995, 269, 187.
(7) Crutzen, P. J.; Zimmermann, P. H. Tellus 1991, 43AB, 136.
(8) Schindler, R. N.; Benter, Th.; Dethlefs, J.; Jungkamp, T.; Liesner,
M.; Wille, U. In EUROTRAC Annual Report, Part 8 LACTOZ;
Commission of the European Communities: Brussels, Belgium,
1992; p189.
spectrum still shows absorption bands which are very similar
to those of HPMF. The main difference lies in the region of
the OH stretching mode. The insert in Figure 8 displays this
region. Two relatively strong absorption bands at 3320 and
3583 cm-1 are tentatively attributed to the cis and trans
isomers of hydroxymethyl formate, i.e.:
(9) Le Bras, G., Ed. Transport and Chemical Transformation of
Pollutants in the Troposphere. In Chemical Processes in
Atmospheric Oxidation, Springer-Verlag: Berlin, 1997; Vol. 3.
(10) Lorenz K.; Zellner, R. Ber. Bunsen-Ges. Phys. Chem. 1984, 88,
1228.
(11) Wallington, T. J.; Liu, R.; Dagant, P.; Kurylo, M. J. Int. J. Chem.
Kinet. 1988, 20, 41.
(12) Atkinson, R.; Baulch, D. L.; Cox, R. A.; Hampson, R. F., Jr.; Kerr,
J. A.; Rossi, M. J.; Troe, J. J. Phys. Chem. Ref. Data 1997, 27, 521.
(13) Finlayson-Pitts, B. J.; Pitts, J. N., Jr. Atmospheric Chemistry; John
Wiley & Sons: New York, 1986; p 131.
(14) Tuazon, E. C.; Atkinson, R.; Carter, W. P. L. Environ. Sci. Technol.
1985, 19, 9, 256.
The lower frequency at 3320 cm-1 is attributed to the cis
isomer whose O-H bonding is weakened by the intramo-
lecular hydrogen bond, whereas the O-H vibration in the
trans form is expected to behave normally.
(15) Chen, J.; Young, V.; Hooshiyar, P. A.; Niki, H. J. Phys. Chem.
1995, 99, 9, 4071.
(16) Thamm, J.; Wolff, S.; Turner, W. V.; Ga¨b, S.; Thomas, W.; Zabel,
F.; Fink, E. H.; Becker, K. H. Chem. Phys. Lett. 1996,258, 155.
(17) Thomas, W. Ph.D. Thesis, University of Wuppertal, Germany,
1997.
(18) Neeb, P.; Horie, O.; Moortgat, G. K. Chem. Phys. Lett. 1995, 246,
150.
Acknowledgments
Financial support of this work by the BMBF (Bundesminister
fu¨ r Bildung, Wissenschaft, Forschung und Technologie)
within the framework of a TFS/ LT3 project is gratefully
acknowledged.
(19) Neeb, P.; Horie, O.; Moortgat, G. K. Int. J. Chem. Kinet. 1996,
28, 721.
Literature Cited
(1) Golding, B. T.; Slaich, G.; Kennedy, G.; Bleasdale, C.; Watson,
W. P. Chem. Res. Toxicol. 1996, 6, 147.
(2) Yu, Y.; Jeffries, H.; Sexton, K. Atmos. Environ. 1997, 31, 2261.
(3) Payne, G. B. J. Am. Chem. Soc. 1959, 81, 4901.
Received for review April 21, 1998. Revised manuscript re-
ceived July 7, 1998. Accepted July 16, 1998.
(4) Atkinson, R. J. Phys. Chem. Ref. Data 1994, Monograph No. 2.
ES9804159
9
VOL. 32, NO. 22, 1998 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 3 5 2 1