Kinetic and Product Study of the Atmospheric Photooxidation of 1,4-Dioxane and Its Main Reaction Product Ethylene Glycol Diformate

Article abstract of DOI:10.1021/jp990273k

A FTIR kinetic and product study of the OH-radical initiated oxidation of 1,4-dioxane (DOX) has been performed in a quartz-glass photoreactor in the laboratory under different conditions and also in the outdoor EUPHORE simulation chamber in Valencia, Spain. Using the relative kinetic technique, a rate coefficient of k = (1.24 ± 0.04) × 10^-11 cm³ molecule^-1 s^-1 was determined for the reaction at 298 K in 1000 mbar of synthetic air, which is in good agreement with other published values. The major reaction product both in the presence and absence of NO was ethylene glycol diformate (EDF). This compound has been synthesized, and authentic samples have been used for calibration. Integrated band intensities have been calculated for the three strongest bands of EDF: (4.99 ± 0.06) × 10^-17 cm molecule^-1 for 1100-1225 cm^-1, (3.90 ± 0.05) × 10^-17 cm molecule^-1 for 1670-1820 cm^-1, (9.34 ± 0.11) × 10^-18 cm molecule^-1 for 2775-3075 cm^-1. In the laboratory reactor, yields for EDF of 87 ± 9 and 95 ± 10 mol % were obtained using the photolysis of MeONO/NO/air and H₂O₂/NO/air as the OH radical sources, respectively. Using only the photolysis of H₂O₂/ air as the OH source resulted in a molar yield of 55 ± 6 mol % for EDF. In the outdoor EUPHORE simulation chamber a yield of 95 ± 10 mol % was obtained from irradiation of a DOX/NO_x/air mixture. The OH-radical- and Cl-atom-initiated oxidation of EDF has also been investigated. Rate coefficients of k_OH = (4.72 ± 0.31) × 10^-13 cm³ molecule^-1 s^-1 and k_cl = (3.52 ± 0.09) × 10¹² cm³ molecule^-1 s^-1 have been determined for the reaction of EDF with OH radicals and Cl atoms, respectively, at 298 K and 1000 mbar total pressure. The products determined in the Cl-initiated oxidation in the presence of NO_x were formic acid anhydride (FAA), formic acid (FA), and carbon monoxide (CO) with yields of 173 ± 34 mol %, 45 ± 9 mol %, and 41 ± 8 mol %, respectively. Formation of a peroxy formyl nitrate was also observed. In the absence of NO_x the yields of FAA, FA and CO were 144 ± 29 mol %, 39 ± 8 mol %, and 22 ± 4 mol %, respectively.

Full text of DOI:10.1021/jp990273k

5032
J. Phys. Chem. A 1999, 103, 5032-5039
Kinetic and Product Study of the Atmospheric Photooxidation of 1,4-Dioxane and Its Main
Reaction Product Ethylene Glycol Diformate
Tobias Maurer and Heinz Hass
FORD-Forschungszentrum Aachen, D-52068 Aachen, Germany
Ian Barnes* and Karl H. Becker
Bergische UniVersita¨t GH Wuppertal/FB 9 - Physikalische Chemie, Gauss Strasse 20,
D-42097 Wuppertal, Germany
ReceiVed: January 22, 1999; In Final Form: April 12, 1999
A FTIR kinetic and product study of the OH-radical initiated oxidation of 1,4-dioxane (DOX) has been
performed in a quartz-glass photoreactor in the laboratory under different conditions and also in the outdoor
EUPHORE simulation chamber in Valencia, Spain. Using the relative kinetic technique, a rate coefficient of
k ) (1.24 ( 0.04) × 10-11 cm3 molecule-1 s-1 was determined for the reaction at 298 K in 1000 mbar of
synthetic air, which is in good agreement with other published values. The major reaction product both in the
presence and absence of NO was ethylene glycol diformate (EDF). This compound has been synthesized,
and authentic samples have been used for calibration. Integrated band intensities have been calculated for the
three strongest bands of EDF: (4.99 ( 0.06) × 10-17 cm molecule-1 for 1100-1225 cm-1, (3.90 ( 0.05) ×
10-17 cm molecule-1 for 1670-1820 cm-1, (9.34 ( 0.11) × 10-18 cm molecule-1 for 2775-3075 cm-1. In
the laboratory reactor, yields for EDF of 87 ( 9 and 95 ( 10 mol % were obtained using the photolysis of
MeONO/NO/air and H2O2/NO/air as the OH radical sources, respectively. Using only the photolysis of H2O2/
air as the OH source resulted in a molar yield of 55 ( 6 mol % for EDF. In the outdoor EUPHORE simulation
chamber a yield of 95 ( 10 mol % was obtained from irradiation of a DOX/NOx/air mixture. The OH-
radical- and Cl-atom-initiated oxidation of EDF has also been investigated. Rate coefficients of kOH ) (4.72
( 0.31) × 10-13 cm3 molecule-1 s-1 and kCl ) (3.52 ( 0.09) × 10-12 cm3 molecule-1 s-1 have been determined
for the reaction of EDF with OH radicals and Cl atoms, respectively, at 298 K and 1000 mbar total pressure.
The products determined in the Cl-initiated oxidation in the presence of NOx were formic acid anhydride
(FAA), formic acid (FA), and carbon monoxide (CO) with yields of 173 ( 34 mol %, 45 ( 9 mol %, and
41 ( 8 mol %, respectively. Formation of a peroxy formyl nitrate was also observed. In the absence of NOx
the yields of FAA, FA and CO were 144 ( 29 mol %, 39 ( 8 mol %, and 22 ( 4 mol %, respectively.
Introduction
two previous determinations of the rate coefficient for the
reaction of OH radicals with DOX (Dagout et al.4 and Porter et
al.5). Platz et al.6 have examined the atmospheric chemistry of
DOX and identified ethylene glycol diformate (EDF) as the only
product with unit yield. However, the evidence for the identi-
fication and quantification was based on a comparison of the
observed infrared product features with an infrared spectrum
available in the literature.
We present here a kinetic and product study of the OH-radical
initiated oxidation of DOX in which the major reaction product
EDF has been identified and quantified using a synthesized
sample of the pure compound. In addition, the variation of the
yield of EDF with and without the presence of NO in the
reaction system has been investigated. Further, first determina-
tions have been made of the rate coefficients for the reaction
of OH and Cl with EDF. Finally the products resulting from
the Cl atom initiated oxidation with and without NO present
have been identified.
Branched and cyclic ethers are either currently in use or are
under discussion as additives to fuels. The use of ethers in fuels1
serves as a replacement for aromatic hydrocarbons, apart from
enhancing the octane value they also help to reduce the carbon
monoxide content of the automobile exhaust.2 Ethers are also
widely employed in industry as solvents. The cyclic ether 1,4-
dioxane (DOX), for example, is mainly employed as a solvent
for cellulose acetate, resins, oils, and waxes.3
The increased employment of DOX, and indeed any com-
pound, as solvent or fuel additive inevitably results in increased
emissions of these compounds to the atmosphere and raises the
question of the atmospheric fate of the substances. The first
major degradation step of most chemicals released into the
troposphere is reaction with OH radicals. A detailed understand-
ing of the kinetics and mechanisms of the reactions radicals
with OH is, therefore, required in order to assess their potential
atmospheric impact such as their contribution to the formation
of ozone and other photooxidants or possible toxicity of the
reaction products. DOX is a cyclic diether with symmetric
structure and four equivalent -CH2- groups. There have been
Experimental Section
The different experiment setups used for the investigations
are described briefly below. Details can be found in the
literature.7-9 Most experiments were carried out in a 1080 l
* To whom correspondence should be addressed.
10.1021/jp990273k CCC: $18.00 © 1999 American Chemical Society
Published on Web 06/15/1999

Photooxidation of 1,4-Dioxane
J. Phys. Chem. A, Vol. 103, No. 26, 1999 5033
quartz-glass room-temperature photoreactor equipped with a
built-in White mirror system. Either the UV photolysis of
hydrogen peroxide (H2O2) with 32 low-pressure mercury lamps
(Philips TUV 40, λmax ) 254 nm) or the irradiation of methyl
nitrite (CH3ONO)/NO mixtures with 32 fluorescent lamps
(Philips Tl 40W/05, 320 < λ < 450 nm) was used for the
production of OH radicals. Chlorine atoms were produced by
the photolysis of molecular chlorine (Cl2) with the fluorescent
lamps. Concentration-time profiles of reactants and products
were monitored by long path in situ FTIR spectroscopy. The
FTIR spectrometer (Bruker IFS-88) was operated with a
resolution of 1 cm-1 using a path length of 484.7 m. All
experiments were carried out at 1000 mbar total pressure in
O2 + N2 at (298 ( 2) K. Measurements were performed with
and without NOx. During irradiation between 5 and 20 min, 10
spectra with 32/64/128 scans (0.5/1/2 min) were collected.
Typical start concentrations employed for the product study were
1.2 ppm DOX, 0.5 ppm EDF, 1-2 ppm CH3ONO, 10-20 ppm
H2O2, 5-6 ppm Cl2, and 0.5-10 ppm NO (1 ppm ) 2.46 ×
1013 molecule cm-3 at 1000 mbar and 298 K).
The rate coefficients for the reaction of OH with DOX and
EDF were determined in the 1080 l photoreactor using the
relative kinetic method.10 In the case of EDF the Cl atom relative
kinetic study was carried out in a 480 l Duran glass reactor in
1000 mbar of synthetic air at (298 ( 2) K. Detection of reactants
was by in situ FTIR long path spectroscopy using a Nicolet
520 spectrometer and a path length of 51.6 m. The duration of
one experiment was 40 min; over this period 20 spectra were
collected with a resolution of 1 cm-1 by coadding 83 scans per
2 min. Details can be found in the literature.9
mbar at (298 ( 2) K were established in the tube by injection
of quantities of the compound. Spectra were recorded at a
resolution of 1 cm-1 over a path length of 46 cm with a Bruker
IFS 88 FTIR spectrometer by coadding 128 interferograms over
2 min.
Three experiments on the OH-radical initiated oxidation of
DOX were carried out in the outdoor EUPHORE8 simulation
chamber in Valencia, Spain. The chamber consists of a half-
spherical 200 m3 Teflon bag with a transmission of more than
80% of the sunlight between 280 and 640 nm and a cooling
system under the chamber floor panels to compensate the heating
of chamber by solar radiation. The photooxidation was initiated
in the chamber by irradiation of a mixture of DOX and nitrogen
oxide (NO) or nitrogen dioxide (NO2) in 1000 mbar of dry air
with natural sunlight. The concentration-time profiles of
reactants and products were monitored by in situ FTIR-
spectroscopy with a Nicolet Magna 550 spectrometer using a
resolution of 1 cm-1 and a path length of 326 m. Each spectrum
consisted of 515 interferograms co-added over a time period of
15 min. Typical start concentrations were 400-770 ppb DOX
and 50 ppb NO or 45 ppb NO2, and the mixtures were irradiated
for approximately 6 h.
Chemicals and Gases. 1,4-Dioxane (99+% GC) was pur-
chased from Aldrich and hydrogen peroxide (85.0%) from
Peroxid Chemie. Ethane 3.5, propene 3.5, Cl2 2.8, NO (e
99.95%), NO2 3.5, ethyl chloride 2.0, and synthetic air
(99.995%) were supplied by Messer Griesheim with the stated
purities.
Methyl nitrite was synthesized as described in the literature13
by adding sulfuric acid dropwise to an ice cooled mixture of
methanol and sodium nitrite and collection in a cold trap at
-76 °C. The reference spectra of formic acid anhydride, formic
acid and carbon monoxide were taken from an available
calibrated infrared spectrum data bank collected by this group.
EDF was synthesized in a manner similar to a method
described in the literature for t-amyl formate.14 Briefly, mixed
anhydride (formic and acetic acid) was slowly added to ethylene
glycol and stirred at room temperature for 24 h. The mixture
was than neutralized with a saturated solution of sodium
bicarbonate, washed with water, and extracted with dimethyl
ether. Separation of the product was performed by distillation
under reduced pressure. The purity of the compound was
determined by GC/ NMR with >98%.
In both the OH radical and Cl atom relative kinetic experi-
ments, if there is no significant loss of the reactant (RA) or the
reference (RF) compound by photolysis or wall loss, the
observed decays can be described by
RA + OH/Cl f Product; k1
RF + OH/Cl f product; k2
and the kinetic data can be treated using eq 1
[RA]t)0
[RA]t
k
[RF]t)0
[RF]t
ln
)
RA ln
kRF
(1)
Spectroscopic Data. (IR (vapor) CdO 1754 cm-1, C-H
where [RA]t)0 and [RF]t)0 are the concentrations of the reactant
and reference compound at time t ) 0, respectively, and [RA]t
and [RF]t are the concentrations at time t, and kRA and kRF are
the corresponding rate coefficients for the reaction of reactant
and reference compound with OH or Cl, respectively. Plots of
ln([RA]t)0/[RA]t) against ln([RF]t)0/[RF]t) should yield a
straight line with slope kRA/kRF and zero intercept. The experi-
ments were performed at 298 ( 2 K with propene (kOH+propene
1
2942 cm-1, C-O 1162 cm-1; H NMR (CDCl3, 400 MHz)
δ ) 7.97 ppm (s, 2H, -OC(O)-H), δ ) 4.29 ppm (s, 4H,
-CH2-); 13C NMR (CDCl3, 400 MHz) δ ) 160.7 ppm
(-OC(O)H), δ ) 61.6 ppm (-CH2-).
Results
Rate Coefficients for the Reaction of 1,4-Dioxane (DOX)
with OH Radicals and Ethylene Glycol Diformate (EDF)
with OH Radicals and Cl Atoms. Over the time period of all
the experiments both adsorption on the wall and photolysis of
the compounds were negligible. Figure 1, parts A and B, shows
examples of the data from the kinetic experiments plotted
according to eq 1 for the reaction of OH with DOX and Cl
with EDF, respectively. From slopes of these plots the following
rate coefficients have been derived were the quoted errors, which
do not include possible errors in the reference rate coefficient,
represent the uncertainty at the 95% confidence level:
) (2.63 ( 0.53) × 10-11 cm3 molecule-1 s-1 11
)
as reference
compound for the DOX + OH kinetic investigations and ethane
(kOH+ethane ) (2.50 ( 0.10) × 10-13 cm3 molecule-1 s-1 11
) and
ethyl chloride (kCl+ethyl
) (8.7 ( 1.0) × 10-12 cm3
chloride
molecule-1 s-1 12
) as reference compounds for the EDF + OH
and EDF + Cl kinetic investigations, respectively. Typical start
concentrations for the OH kinetic experiments were 1 ppm
DOX, 0.5 ppm EDF, 4 ppm propene, 5 ppm ethane, and 25
ppm H2O2. For the kinetic experiments with Cl radicals 2-3
ppm EDF, 40 ppm ethyl chloride, and 40 ppm Cl2 were used.
The infrared absorption cross sections of EDF were measured
in a glass tube with a volume of 900 cm3. The tube was
evacuated, and vapor pressures of EDF between 0.7 and 0.25
kOH-DOX ) (1.24 ( 0.04) × 10-11 cm3 molecule-1 s-1

5034 J. Phys. Chem. A, Vol. 103, No. 26, 1999
Maurer et al.
Figure 2. Infrared spectra recorded for a H2O2/NOx/DOX photolysis
system in the wavenumber region 2000-700 cm-1. Spectrum A was
recorded before irradiation, B is a difference spectrum derived from a
spectrum recorded after 10 min irradiation minus spectrum A, C is a
reference spectrum of EDF, and D is the residual spectrum obtained
after 10 min irradiation and subtraction of DOX and all identified
products. The N2O is introduced into the chamber as impurity with
NO.
TABLE 1: Molar Yields of Ethylene Glycol Diformate
(EDF) Obtained from Experiments on the
OH-Radical-Initiated Oxidation of 1,4-Dioxane (DOX)
EDF
OH-
source
light
yielda
part
reactor
condition
source (mol%)
A
B
C
D
1080 L reactor
1080 L reactor
1080 L reactor
without NOx H2O2
UV
UV
55 ( 6
95 ( 10
87 ( 9
with NOx
with NOx
H2O2
CH3ONO vis
Figure 1. Kinetic data plotted according to eq 1 for reaction of (A)
OH radicals with 1,4-dioxane (DOX) and (B) Cl atoms with ethylene
glycol diformate (EDF).
200 m3 smog chamber with NOx
NOx/RH sunlight 95 ( 10
a The errors are estimated for a relative uncertainty of 10%.
kOH-EDF ) (4.72 ( 0.31) × 10-13 cm3 molecule-1 s-1
kCl-EDF ) (3.52 ( 0.09) × 10-12 cm3 molecule-1 s-1
obtained after subtraction of all identified compounds. The N2O
in spectrum D is introduced into the chamber as impurity with
NO. In contrast to the residual spectrum of the NOx-containing
system, significant absorptions in the area of the CdO stretching
vibration between 1750 and 1650 cm-1 and in the area of the
C-O stretching vibration between 1200 and 1000 cm-1 remain
in the NOx-free system (not shown).
From the analysis of the infrared data with the calibrated EDF
spectrum the main product of the OH-initiated oxidation of DOX
has been positively identified as EDF in all of the systems
investigated. The EDF molar yield for the different experimental
systems has been obtained from the slopes of plots of the
concentration of EDF formed against reacted DOX. Examples
of such plots for a NOx-containing (part A)) and a NOx-free
system (part B)) are shown in Figure 3. The molar yields of
EDF obtained for the different experimental systems are
summarized in Table 1. The EDF yields have not been corrected
for secondary reaction with OH radicals since this reaction is
slow and over the time scale of the experiments the correction
is negligibly small.
Infrared integrated band intensity (IBI) of ethylene glycol
diformate (EDF). Figure 2 shows the infrared spectrum of EDF
(spectrum C) recorded with a resolution of 1 cm-1 in the region
of 700 to 2000 cm-1. The spectrum contains three main
absorption bands, one in the region of the C-H stretching
vibration between 2875 and 3075 cm-1 (not shown in Figure
2), one in the region of the CdO stretching vibration between
1670 and 1820 cm-1, and one in the region of the C-O
stretching vibration between 1100 and 1225 cm-1. For these
three bands the following integrated band intensities (IBI) have
been calculated where the estimated errors represent the
uncertainty at the 95% confidence level:
IBI (2775-3075 cm-1) )
(9.34 ( 0.11) × 10-18 cm molecule-1
IBI (1670-1820 cm-1) )
Products from the Cl-atom-initiated oxidation of ethylene
glycol diformate (EDF). Because of the very slow reaction of
EDF with OH radicals the Cl-atom initiated oxidation was used
to study the products with and without NOx in the reaction
system. The oxidation processes with OH radicals and Cl atoms
are very similar and only minor differences in site selectivity
is expected.
(3.90 ( 0.05) × 10-17 cm molecule-1
IBI (1100-1225 cm-1) )
(4.99 ( 0.06) × 10-17 cm molecule-1
Products from the OH-Radical Initiated Oxidation of 1,4-
Dioxane(DOX). Figure 2 shows infrared spectra obtained from
the experimental system with NOx. Part A shows a spectrum
recorded before radiation, B is a difference of a spectrum
recorded after 10 min irradiation minus spectrum A, C is the
EDF reference spectrum, and D is the residual product spectrum
Figure 4 shows infrared spectra obtained from the experi-
mental system with NOx. Shown in part A is a spectrum
recorded before irradiation, B is a difference spectrum derived
from a spectrum recorded after 10 min irradiation minus
spectrum A, C is a calibrated reference spectrum of formic acid
anhydride (FAA),15 and D is the residual spectrum obtained after

Photooxidation of 1,4-Dioxane
J. Phys. Chem. A, Vol. 103, No. 26, 1999 5035
Figure 5. Plots of the measured concentrations of formic acid
anhydride (FAA), formic acid (FA), and CO as a function of the amount
of ethylene glycol diformate (EDF) consumed. The data are from a
Cl2/NO/EDF photolysis experiment. The plot of the data for the peroxy
formyl nitrate (PAN-EDF) shows curvature, the increase in slope with
the amount of consumed EDF is due to changes in the NO/NO2 ratio
as described in the text.
TABLE 2: Molar Yields of Products Observed in the
Cl-Atom-Initiated Oxidation of Ethylene Glycol Diformate
(EDF) in Presence and Absence of NOx
molar yields %a
molar yields %a
product of EDF oxidation
[Cl, synthetic air] [Cl/NO, synthetic air]
formic acid anhydride (FAA)
formic acid (FA)
carbon monoxide (CO)
144 ( 29
39 ( 8
22 ( 4
173 ( 34
45 ( 9
41 ( 8
Figure 3. Example of a plot of the measured EDF concentration versus
the amount of DOX consumed for the OH-radical-initiated oxidation
using (A) the photolysis of CH3ONO/NO/air as the OH-radical source
and (B) the photolysis of H2O2 as the radical source.
a Errors are the total overall estimated uncertainty of 20%.
identified products in the region at 1150 cm-1; however, in the
residual product spectrum from the system with NOx (see Figure
4 spectrum D) distinctive spectral features remain with bands
at 1835, 1746, 1306, 1223, and 793 cm-1. These absorptions
are very similar to the characteristic absorptions of peroxy acetyl
nitrate (PAN) and are consequently assigned to a peroxy formyl
nitrate (PAN-EDF). The formation of this compound is de-
pendent on the NO/NO2 ratio in the experimental system and
consequently the yield at the beginning of the experiment with
high NO/NO2 ratio is low and increases gradually during the
course of the experiment as the NO/NO2 ratio decreases. A
rough estimate has been made of the concentration of the
peroxynitrate using the value of 1.09 × 10-3 ppm-1 m-1 for
the absorption cross section of peroxy acetyl nitrate at 1835
cm-1 available in the literature16; its yield is found to vary
between approximately 8 and 30 mol % during the progression
of an experiment.
Figure 4. Infrared spectra obtained from a Cl2/NO/ethylene glycol
diformate (EDF) photolysis system in the wavenumber region 2350-
700 cm-1. Spectrum A was recorded before irradiation, B is a difference
spectrum derived from a spectrum recorded after 10 min irradiation
minus spectrum A, C is a calibrated reference spectrum of formic acid
anhydride (FAA), and D is the residual spectrum obtained after 10 min
irradiation and subtraction of EDF and all identified products multiplied
by a factor of 10.
Discussion
Kinetic Experiments. The measured rate coefficient for the
reaction of DOX with OH is in reasonable agreement with the
values reported in the literature. The value reported by Porter
et al.5 (k ) 1.26 . 10-11 cm3 molecule-1 s-1) using an absolute
technique agrees best with that measured in the present work.
It is somewhat higher than the values measured by Dagaut et
al.4 (k ) 1.09 . 10-11 cm3 molecule-1 s-1) and Porter et al.5 (k
10 min irradiation and subtraction of EDF. All of the absorption
values of the identified products are multiplied by a factor of
10 in order to better highlight the residual absorptions.
In reaction systems with or without NOx, three products were
identified, formic acid anhydride (FAA), formic acid (FA), and
carbon monoxide (CO). Molar yields for the products have been
obtained from plots of the measured concentrations as a function
of EDF reacted as illustrated by the exemplary plots in Figure
5 for a Cl2/NO/EDF photolysis experiment. The measured molar
yields are listed in Table 2.
.
) 9.7 10-12 cm3 molecule-1 s-1) using absolute and relative
techniques, respectively.
The measured rate coefficient for the reaction of OH with
DOX is lower than the value of k ) 3.9 . 10-11 cm3 molecule-1
s-1 calculated using the structure activity relationship (SAR)
technique. The rate can be compared with that of the open chain
diether dimethoxy ethane (DMEt), CH3OCH2CH2OCH3, which
contains one -OCH2CH2O- unit. Rate coefficients of (3.3 (
In the residual product spectrum of the NOx-free system only
weak residual absorptions remain after subtraction of the

5036 J. Phys. Chem. A, Vol. 103, No. 26, 1999
Maurer et al.
0.2) × 10-11 cm3 molecule-1 s-1 5 and (2.7 ( 0.5) × 10-11 cm3
molecule-1 s-1 17 have been reported in the literature for the
reaction of DMEt with OH-radicals. The value is slightly more
than a factor of 2 higher than that measured for DOX which
contains 2 -CH2CH2- units joined in a cyclic fashion by 2
ether linkages. Porter et al.5 have postulated that this apparent
discrepancy can be explained by the formation of six-membered
cyclic transition state in the case of the reaction of OH with the
linear ether dimethoxy ethane which is not possible in the case
of the cyclic ether DOX.
by cleavage of the C-C bond to form an open chain alkoxy
radical with one formate group (III). In the presence of NO
reaction of the cyclic alkoxy radical (II) with O2 to form a
lactone (VI) ether and HO2 appears to be negligible. The open
chain alkoxy radical (III) reacts with O2 to form a peroxy radical
(IV), which further reacts with NO to form the alkoxy radical
(V). EDF is then formed via reaction of this alkoxy radical with
O2. Isomerization reactions of radical (V) appear to be unim-
portant. This is supported by the 100% carbon balance and the
lack of observation of FAA, FA, and HCHO which are expected
isomerization products.
To our knowledge this represents the first determination of
the rate coefficients for the reaction of OH radicals and Cl atoms
with EDF. Because of the lack of data for the formate group, it
is difficult to reliably estimate a rate coefficient for the reaction
of OH with EDF using the SAR method. A survey of the kinetic
data for other formates in the literature,18,19 shows that the
presence of a formate group greatly reduces the reactivity of a
compound. For example, if DMEt5 is compared with methoxy
ethyl formate,17 CH3OCH2CH2OCHO, replacing one of the two
CH3O-groups by a formate group decreases the OH rate
coefficient by a factor of approximately 6.5 from (3.3 ( 0.2)
× 10-11 cm3 molecule-1 s-1 to (5.10 ( 1.03) × 10-12 cm3
molecule-1 s-1. If such effects are additive one would predict
a value of approximately 7.9 × 10-13 cm3 molecule-1 s-1 for
the reaction of OH with EDF which considering the crude
approximation is in fair agreement with the actual measured
value of (4.72 ( 0.31) × 10-13 cm3 molecule-1 s-1. The rate
coefficient of (3.52 ( 0.09) × 10-12 cm3 molecule-1 s-1 is
intermediate between the values of (1.83 ( 0.21) × 10-12 and
(13.4 ( 0.15). 10-12 cm3 molecule-1 s-1 reported for the reaction
of Cl with the simple formates methyl and ethyl formate,
respectively.20
The oxidation mechanism of the DOX in a NOx-free system
is more complicated because of the self-and cross peroxy radical
reactions, which can occur as shown in Figure 6. Reaction steps
which can affect the EDF yield are the self-reaction of the cyclic
peroxy radicals (I) giving the lactone (VI) and alcohol (VII) or
reaction of the peroxy radicals (I) with HO2 to form a
hydroperoxide (VIII). The self-reaction of the open chain peroxy
radical (IV) to form an alcohol (IX) or with HO2 to form a
hydroperoxide (X) can also affect the EDF yield.
As can be seen from Table 1, in the absence of NOx the yield
of EDF in the oxidation of DOX is only 55 mol %. Varying
the O2 partial pressure between 0 and 600 mbar at a total
pressure of 1000 mbar (N2 + O2) did not affect the EDF yield.
This observation is consistent with the results of Platz et al.6
who also found no dependence of the EDF formation yield on
the O2 concentration. The missing 45 mol % products are likely
to consist of ether lactone (VI), an alcohol (VII and IX), and
a hydroperoxide (VIII and X). In the residual product spectrum
of a NOx free system the absorption observed at 1760 cm-1 is
typical for six-membered lactones and that at 3460 cm-1
probably belongs to a hydroperoxide. Absorptions in the region
of 3500 to 3700 cm-1, which are typical for alcohols, were not
observed.
For both the OH-radical and Cl-atom, H-atom abstraction
from the formate group is known to be very slow. Therefore,
in the OH- and Cl-initiated oxidation of DOX, H-atom abstrac-
tion from the -OCH2CH2O- moiety is expected to dominate
over abstraction from the formate group. For example, the rate
coefficient for the reaction of Cl-radicals with formic acid is
(2.00 ( 0.25) × 10-13 cm3 molecule-1 s-1 21 and is thought to
proceed mainly by H abstraction from the H-CO- group.
Using this value suggests that for the reaction of Cl with EDF
approximately 11% should proceed via H atom abstraction from
the formate group and the remainder from the -OCH2CH2O-
moiety.
The reason for the considerably lower yield of EDF in this
study measured in the absence of NOx compared to that of Platz
et al.6 is not clear. Platz et al. used high concentrations of Cl2
in their study and probably had high steady-state concentrations
of Cl-atoms. The reactions of Cl-atoms with peroxy radicals
are fast; it may be that in the system of Platz et al. peroxy-
peroxy reactions were suppressed by efficient conversion of the
peroxy radicals to alkoxy radicals via reaction with Cl atoms.
Products Studies on Ethylene Glycol Diformate. The major
product observed in the Cl-atom-initiated oxidation of EDF is
FAA with yields of 86 ( 17%C in the presence of NOx and 72
( 15%C without NOx in the system. The other products
observed are FA and CO. The carbon balances were 108 (
21% and 87 ( 18% with and without NOx, respectively,
showing that all the major products have been identified. The
yields of CO and FA were approximately the same in the system
with NOx; however, in the NOx-free system the CO yield is
approximately half the FA yield. The observation of a PAN-
type product in the presence of NOx shows that H-abstraction
also occurs at the terminal formate H-atoms. Although no
evidence could be found for the presence of other products in
the residual spectra after subtraction of the identified products
the carbon balance in the absence of NOx suggests that small
amounts of other products may be formed.
Products Studies on 1,4-Dioxane (DOX). The product
studies show quite conclusively that, in the presence of NOx,
the OH-radical oxidation of DOX leads to the formation of EDF
in near 100% yield. The result confirms the earlier assignment
of Platz et al.6. However, the results obtained in the absence of
NOx differ significantly from those of Platz et al. In the present
study a yield of 55 mol % is found for EDF without NOx;
whereas Platz et al. found only 10 mol % less EDF in their
NOx-free system compared to their NOx-containing system.
The reaction of OH with DOX can only proceed via an
H-atom abstraction mechanism to form EDF and because of its
symmetric cyclic structure all the -CH2- groups are equivalent.
Figure 6 shows the reactions in the oxidation of DOX, with
and without NOx in the system, which lead to the formation of
EDF and other possible products. The reactions dominating
under most atmospheric conditions are highlighted by the arrows
in bold type. Irrespective of whether NOx is present or not the
first step is abstraction of a ring H atom followed by addition
of O2 to form a cyclic alkylperoxy radical (I). With NO this is
transformed to a cyclic alkoxy radical (II), which reacts further
With the exception of the PAN-type compound plots of the
concentration of all the observed products against the amount
of EDF reacted are linear as it is shown by the exemplary plots
in Figure 5 for the NOx-containing system. The linearity of the
plots supports that the products are all formed directly and not
in secondary reactions. The reason for the nonlinear behavior

Photooxidation of 1,4-Dioxane
J. Phys. Chem. A, Vol. 103, No. 26, 1999 5037
Figure 6. Mechanism for the OH-radical-initiated oxidation of 1,4-dioxane (DOX) depicting possible reaction channels both with and without the
presence of NOx. The reactions dominating under most atmospheric conditions showing arrows with bold type.
of the PAN-type compound is due to the variation of the NO/
NO2 ratio with time as discussed earlier.
Tuazon et al.23 have recently reported that alkoxy radicals of
structure RC(O)OCH(O•)R can undergo a rapid rearrangement
and decomposition to RC(O)OH and RC(•)O via a five-
membered ring transition state. The authors observed the so-
called "R-ester rearrangement" for several acetates. This type
of rearrangement is possible with the HC(O)OCH(O•)CH2-
OC(O)H (XI) radical formed from the oxidation of EDF. The
R-ester rearrangement for this radical is shown in Figure 7. The
rearrangement would result in formation of FA (XIV) and is
obviously a possible source of the acid in this study. The R-ester
rearrangement of the HC(O)OCH(O•)CH2OC(O)H radical apart
from forming FA will give the C(O•)CH2OC(O)H radical (XV).
The fate of this radical is expected to be addition of O2 and
eventual cleavage of CO2 with formation of the (•)CH2OC(O)H
radical (XVI). This radical is also formed in the C-C bond
scission of radical (XI). The further reactions of (•)CH2OC-
(O)H will give (•O)CH2OC(O)H (XVII). The (•O)CH2OC(O)H
radical can either react with O2 to produce FAA (XIIb) or
undergo an R-ester rearrangement to form FA (XIVb) and HCO.
Further reaction of HCO with O2 will give CO.
As for DOX, both the OH-radical- and Cl-atom-initiated
oxidation of EDF will proceed by H-atom abstraction. However,
in contrast to DOX, there are two different positions for radical
attack i.e., H-atom abstraction from either the H-C(O)-O-
or the O-CH2CH2O- entities leading to the corresponding alkyl
radicals.
From the discussion in the kinetic section it is expected that
in the Cl2 system used for the product analysis approximately
89% of the Cl-atom attack will be at the methylene groups in
the center of the molecule (see Figure 7). Attack at the
methylene group will lead to formation of the alkoxy radical
HC(O)OCH(O•)CH2OC(O)H (XI) which can decompose giving
two molecules of FAA (XII/XIIb) as shown in Figure 7. No
indication was observed for the formation of a mixed anhydride
(XIII) by reaction of the alkoxy radical (XI) with O2. This
supports that the alkoxy radical undergoes C-C bond scission
rather than reaction with O2 as has been observed for similar
peroxy radicals formed from other ethers, e.g., DOX. Decom-
position of FAA in the liquid phase is known to produce FA;22
however, the short time scale and a stability test performed in
the reactor rules out this route as the source of this product.
The approximately equal yields of FA and CO and also the
somewhat higher yield of FAA in the reaction system with NO
suggests that in the NO-system the major reaction of the HC-

5038 J. Phys. Chem. A, Vol. 103, No. 26, 1999
Maurer et al.
Figure 7. Mechanism for the Cl-atom initiated oxidation of ethylene glycol diformate (EDF) depicting reaction channels with Cl attack at the
methylene group. To maintain clarity the reaction channels involving RO2 are only shown in the initial steps. The reactions dominating under most
atmospheric conditions showing arrows with bold type.
(O)OCH(O•)CH2OC(O)H (XI) radical will be the C-C bond
cleavage and that the R-ester rearrangement is negligible. If the
R-ester rearrangement channel was operative to any degree, a
disproportionation between the yields of FA and CO would
occur, i.e., higher yields of FA compared to those of CO. This
is exactly what is observed in the NOx-free experiments where
the yield of CO is approximately a factor of 2 lower than the
FA yield. A possible explanation of this effect may be the
formation of a “hot alkoxy radical” in the NO-containing system.
The “hot alkoxy radical” effect has recently been postulated to
explain differences between the behavior of several alkoxy
radicals in the presence and absence of NO.24 This effect causes
a significant fraction of the available energy to be deposited
into the internal modes of the oxy radical product, which results
in an enhanced tendency for the radical to undergo unimolecular
decomposition. Thus, in the present case, in the presence of
NO, formation of a “hot alkoxy radical” favors C-C bond
fission over an R-ester rearrangement. While, in the absence of
NO the peroxy-peroxy reactions forming the oxy radical are
not so exothermic and consequently the oxy radical will be less
excited and R-ester rearrangement can occur to some extent.
On the basis of the yield of CO and FA as coproduct, the results
Figure 8. Mechanism for the formation of the PAN-EDF in the reaction
of ethylene glycol diformate (EDF) with Cl-atoms.
are in line with radical (XI) in Figure 7 undergoing ap-
proximately 83% C-C scission and 17% R-ester rearrangement
in the absence of NOx. Radical (XVII) cannot undergo C-C
bond scission. From a comparison of the FA to CO yields in
the NO-containing system it is estimated that the O2 reaction
and R-ester rearrangement account for approximately 63 and
37%, respectively, of the fate of this radical.
Abstraction of the H atom at the formate group forms an
alkoxy-radical (XVIII) which adds O2 to form the peroxyradical
(XIX) as shown in Figure 8. In the presence of NOx addition

Photooxidation of 1,4-Dioxane
J. Phys. Chem. A, Vol. 103, No. 26, 1999 5039
of NO2 leads to the peroxy formyl nitrate (PAN-EDF) (XX).
In the absence of NOx reactions of the formylperoxy radicals
(XIX) and formyloxy will occur. The products of these reaction
are not known but will probably produce HCO-O-CH2CH2-
O(•) radicals and eventually HC(O)-O-CH2CH(O). Although
a PAN-type compound is formed in the Cl-atom initiated
oxidation it is not expected that PAN formation will be important
in the OH-radical initiated oxidation, since OH should be more
site selective and attack at the methylene group is expected to
dominate.
Acknowledgment. Financial support of the work by the
European Commission and the “Bundesministierum fu¨r Bildung
und Forschung” (BMBF) is gratefully acknowledged. The
authors acknowledge the help of the CEAM employees in using
EUPHORE and financial support by the Generalidad Valenciana
and Fundacio´n BANCAIXA.
References and Notes
(1) Ha¨lsig, C.-P.; Gregory, R.; Peacock, T. Special Publication-Royal
Society of Chemistry Chemicals in the Oil Industry: DeVelopments and
Applications 1991, 97, 311.
Conclusions
(2) Dolislager, L. J.; J. Air Waste Manag. Assoc. 1997, 47, 775.
(3) ReV. of EnViron. Contam. Toxicol. 1988, 106, 113 (Section I), based
The atmospheric oxidation of 1,4-dioxane (DOX) and eth-
ylene glycol diformate (EDF) has been studied under different
conditions. It has been confirmed, using an authentic sample,
that EDF is the only product of the OH initiated oxidation of
DOX in the presence of NO with a molar yield of about 100%
independent of OH source. In the absence of NOx the yield of
EDF is only 55% indicating the formation of other nonidentified
products such as cyclic esters, alcohols or hydroperoxides. The
rate coefficient determined for the reaction of OH with DOX
of k ) (1.24 ( 0.04) × 10-11 cm3 molecule-1 s-1 is in good
agreement with other literature values.
on USEPA, 1987.
(4) Dagaut, P.; Wallington, T. J.; Liu, R.; Kurylo, M. J. J. Phys. Chem.
1990, 94, 1881.
(5) Porter, E.; Wenger, J.; Treacy, J.; Sidebottom, H.; Mellouki, A.;
Te´ton, S.; Le Bras, G. J. Phys. Chem. A 1997, 101, 5770.
(6) Platz, J.; Sehested, J.; Nielsen, O. J.; Wallington, T. J. J. Chem.
Soc., Faraday Trans. 1997, 93, 2855.
(7) Barnes, I.; Becker, K. H.; Mihalopoulos, N. J. Atmos. Chem. 1994,
18, 267.
(8) Becker, K. H., Ed. The European photoreactor EUPHORE. Final
Report EV5V-CT92-0059; 1996.
(9) Libuda, H. G.; Zabel, F.; Fink, E. H.; Becker, K. H. J. Phys. Chem.
1990, 94, 5860.
Rate coefficients of (4.72 ( 0.31) × 10-13 and (3.52 ( 0.09)
× 10-12 cm3 molecule-1 s-1 have been determined for the
reaction of OH-radicals and Cl-atoms with EDF, respectively.
Using an annual average global tropospheric OH concentra-
tion of 1 × 106 molecules cm-3 25 the maximum lifetime of
DOX and EDF will be 22.4 h and 24 days, respectively. The
lifetime of EDF with respect to reaction with OH will enable it
to be transported to remote areas. However, because of the good
solubility of EDF in water, wet precipation or rain out will be
an important loss process which will considerably shorten its
atmospheric lifetime in humid or wet climatic regions.
The products of the Cl-atom initiated EDF oxidation, in order
of importance, are formic acid anhydride (FAA), formic acid
(FA), and carbon monoxide (CO). The yields of all compounds
were highest in a system containing NO. A mechanism has been
proposed to explain the formation of the products. The formation
of FA and CO includes an R-ester rearrangement recently
invoked to explain the acid formation in the oxidation of
acetates.
(10) Bierbach, A.; Barnes, I.; Becker, K. H.; Wiesen, E.; EnViron. Sci.
Technol. 1994, 28, 715.
(11) Atkinson, R. J. Phys. Chem. Ref. Data, Monogr. 1994, 2, 1.
(12) Shi, J.; Wallington, T. J.; Kaiser, E. W. J. Phys. Chem. 1993, 97,
6184.
(13) Taylor, W. D.; Allston, T. D.; Moscato, M. J.; Fazekas, G. B.;
Kozlowski, R.; Takacs, G. A. Int. J. Chem. Kinet. 1980, 12, 231.
(14) Stevens, W.; Van Es, A. Recueil 1964, 83, 1287.
(15) Thomas, W. Ph.D. Thesis, Bergische University of Wuppertal,
Germany, 1998.
(16) Tsalkani, N.; Toupance, G. Atmos. EnViron. 1988, 23, 1849.
(17) Barnes, I.; Donner, B. Tropospheric research programm results
1996-1997. GSF-Forschungszentrum fu¨r Umwelt und Gesundheit GmbH
1997, 69.
(18) Wallington, T. J.; Dagaut, P.; Liu, R. Kurylo, M. J. Int. J. Chem.
Kinet. 1988, 20, 177.
(19) Le Calve`, S.; Le Bras, G.; Mellouki, A. J. Phys. Chem. A 1997,
101, 5489.
(20) Notario, A.; Le Bras, G.; Mellouki, A. J. Phys. Chem. A 1998,
102, 3112.
(21) Wallington, T. J.; Andino, J. M.; Ball, J. C.; Japar, S. M. J. Atmos.
Chem. 1990, 10, 301.
(22) Olah, G. A.; Vankar, Y. D.; Arvanaghi, M.; Sommer, J. Angew.
Chem. 1979, 91, 649.
(23) Tuazon. E. C.; Aschmann, S. M.; Atkinson, R.; Carter, W. P. L. J.
Phys. Chem. A 1998, 102, 2316.
(24) Bilde, M.; Wallington, T. J.; Ferronato, C.; Orlando, J. J.; Tyndall,
The oxidation of DOX and cyclic diethers in general will
lead to formation of alkyl diformates. The major atmospheric
loss processes for these alkyl diformates will be gas phase
reaction with OH radicals and wet precipitation. Both of these
processes will result in acid formation, mainly formic acid, and
the widespread use of ethers could, therefore, lead on the long
term to an increase in the atmospheric acid burden.
G. S.; Estupinan, E.; Haberkorn, S. J. Phys. Chem. 1998, 102, 1976.
(25) Prinn, R. G.; Weiss, R. F.; Miller, B. R..; Huang, J. Alyea, F. N.;
Cunnold, D. M.; Fraser, P. J.; Hartley, D. E., Simmonds, P. J. Science 1995,
269, 187.