The OH-initiated oxidation of hexylene glycol and diacetone alcohol

Article abstract of DOI:10.1021/es0264450

The OH-initiated oxidation of two VOCs directly emitted to the atmosphere through their use as industrial solvents, hexylene glycol (HG, (CH ₃)₂C(OH)CH₂CH(OH)CH₃) and diacetone alcohol (DA, (CH₃)₂C(OH)CH₂C(O)CH₃), has been studied in two photoreactors: a 140 L Teflon bag irradiated by lamps at CNRS-Orleans and the 200 m³ European photoreactor, EUPHORE, irradiated by sunlight. The rate constants for the reactions of HG and DA with OH radicals have been determined at (298 ± 3) K using a relative rate method: k_HG = (1.5 ± 0.4) x 10^-11 and k _DA (3.6 ± 0.6) x 10^-12 cm³ molecule ^-1 s^-1 and have been found in good agreement with estimations from structure-reactivity relationships. The study at Orleans and EUPHORE of the OH-initiated oxidation of hexylene glycol showed the formation of diacetone alcohol, acetone, and PAN as the principal products. The branching ratio of the H-atom abstraction from the > CH- group of HG has been estimated to be (47 ± 4)% corresponding to the measured formation yield of DA. The formation yields of acetone and PAN lead to the determination of a lower limit of (33 ± 7)% for the branching ratio of the H-atom abstraction of the -CH₂-group of HG. For diacetone alcohol, studies at EUPHORE have shown negligible photolysis under atmospheric conditions (J < 5 x 10^-6 s^-1) and the formation of acetone, PAN, HCHO, and CO in the OH-initiated oxidation experiments. The molar yield of acetone, close to 100%, corresponds to the branching ratio of the H-atom abstraction from the -CH₂- group of DA. The present study has allowed the identification of the nature and the fate of the oxy radicals as intermediates in the oxidation mechanism of both HG and DA. The atmospheric implication of these results, especially the ozone formation potential of HG and DA, is discussed.

Full text of DOI:10.1021/es0264450

Environ. Sci. Technol. 2003, 37, 4170-4181
paints, coatings, adhesives, and cleaning and as an inter-
The OH-Initiated Oxidation of
Hexylene Glycol and Diacetone
Alcohol
mediate for organic synthesis. These two compounds are
considered as high volume chemicals (HVC) with a produc-
6
tion exceeding 0.5 × 10 kg annually in the United States (1).
They can also be formed in the troposphere as intermediate
products in the chemical degradation of other VOCs. Indeed,
â-hydroxycarbonyls can be produced from the reaction of
diols or alcohols with the OH radicals (2, 3). The diacetone
alcohol is formed, as shown by Bethel et al. (4) and this work,
from the OH-initiated oxidation of hexylene glycol. Similarly
to other alcohols and ketones, the gas-phase atmospheric
oxidation of hexylene glycol and diacetone alcohol is initiated
mainly by reaction with OH radicals. The photolysis may not
be completely excluded as a loss process for diacetone
alcohol. The degradation of these two compounds may
contribute to the formation of components of the photo-
chemical smog in urban areas.
I . M A G N E R O N , V . B O S S O U T R O T ,
A . M E L L O U K I , * G . L A V E R D E T , A N D
G . L E B R A S
LCSR/CNRS, 1C Avenue de la Recherche Scientifique,
4
5071 Orl e´ ans Cedex 02 - France
The OH-initiated oxidation of two VOCs directly emitted to
the atmosphere through their use as industrial solvents,
hexylene glycol (HG, (CH3)2C(OH)CH2CH(OH)CH3) and diacetone
alcohol (DA, (CH3)2C(OH)CH2C(O)CH3), has been studied
in two photoreactors: a 140 L Teflon bag irradiated by lamps
The only existing data so far reported for the reaction of
OH with the title compounds are those from Bethel et al. and
from Atkinson and Aschmann for hexylene glycol and
diacetone alcohol, respectively (2, 3). Both measurements
were conducted using the relative rate method. In addition,
Bethel et al. reported that the major product of the OH-
initiated oxidation of HG was DA. No mechanistic study is
available in the literature on the OH-initiated oxidation of
DA. Therefore, further studies on the title compounds are
needed to complete the understanding of their atmospheric
fate. In addition, our work aims at better defining the overall
reactivity of oxygenated VOCs toward the hydroxyl radicals.
Using the relative rate method, we have determined the
rate constants for the reaction of the OH radical with hexylene
glycol and diacetone alcohol at (298 ( 3) K and 760 Torr of
air. We have also studied the OH-initiated oxidation mech-
anisms and determined the main oxidation products of these
two compounds using two different photoreactors at LCSR/
CNRS-Orl e´ ans and at Valencia, Spain (EUPHORE: European
Photoreactor). In addition, the UV-visible absorption spec-
trum and the photolysis rate were measured for diacetone
alcohol. The atmospheric implications of the data obtained
within this work are discussed.
3
at CNRS-Orl e´ ans and the 200 m European photoreactor,
EUPHORE, irradiated by sunlight. The rate constants for the
reactions of HG and DA with OH radicals have been
determined at (298 ( 3) K using a relative rate method:
kHG ) (1.5 ( 0.4) × 10 and kDA ) (3.6 ( 0.6) × 10 cm3
-
11
-12
-
1
-1
molecule
s
and have been found in good agreement
with estimations from structure-reactivity relationships. The
study at Orl e´ ans and EUPHORE of the OH-initiated
oxidation of hexylene glycol showed the formation of
diacetone alcohol, acetone, and PAN as the principal
products. The branching ratio of the H-atom abstraction
from the >CH- group of HG has been estimated to be (47
(
4)% corresponding to the measured formation yield of
DA. The formation yields of acetone and PAN lead to the
determination of a lower limit of (33 ( 7)% for the
branching ratio of the H-atom abstraction of the -CH2-
group of HG. For diacetone alcohol, studies at EUPHORE
have shown negligible photolysis under atmospheric
-
6 -1
conditions (J < 5 × 10
s ) and the formation of acetone,
PAN, HCHO, and CO in the OH-initiated oxidation
experiments. The molar yield of acetone, close to 100%,
corresponds to the branching ratio of the H-atom abstraction
from the -CH2- group of DA. The present study has
allowed the identification of the nature and the fate of
the oxy radicals as intermediates in the oxidation mechanism
of both HG and DA. The atmospheric implication of
these results, especially the ozone formation potential of
HG and DA, is discussed.
Experimental Section
The experimental systems and the procedures used in this
work are briefly described here; details can be found
elsewhere (5-8).
OH Reaction Rate Constant Measurem ents and OH-
Initiated Oxidation Investigation. The OH-initiated oxida-
tion of hexylene glycol and diacetone alcohol was investigated
at LCSR/ CNRS (Orl e´ ans) and at EUPHORE.
(
i) Experim ents Perform ed at LCSR/CNRS-Orl e´ ans. The
setup used at Orl e´ ans consists of a 140 L FEP Teflon bag
surrounded by six lamps emitting at 254 nm (Sylvania, G
3
0W), but only two of them were used. The bag and the
Introduction
lamps were positioned in a wooden box with internal faces
covered with aluminum foil. The number of lamps switched
on could be selected to vary the total intensity of light. Dry
air flowed around the reactor to stabilize the temperature at
(298 ( 3) K and to avoid water permeation through the FEP
foil. The compounds were introduced in the bag through a
stream of purified air by flushing a known amount of liquid
placed in a bubbler. The photoreactor was then filled to its
full capacity at atmospheric pressure with purified air.
Analysis of the reactants and products were performed using
a Nicolet FTIR spectrometer with a path-length of 10 m
The compounds of interest, hexylene glycol (2-methyl-2,4-
pentanediol), (HG, (CH
acetone alcohol (4-hydroxy-4-methyl-2-pentanone), (DA,
CH C(OH)CH C(O)CH ), are directly emitted to the at-
3 2 2 3
) C(OH)CH CH(OH)CH ), and di-
(
3
)
2
2
3
mosphere through their use as solvent in industry. For
example, hexylene glycol is used as a solvent for the perfume
and cosmetic industry and as an intermediate for organic
synthesis, while the diacetone alcohol is used as a solvent for
*
Corresponding author phone: (33) 238 25 76 12; fax: (33) 238 69
-
1
6
0 04; e-mail: mellouki@cnrs-orleans.fr.
(infrared spectra were taken at 1 cm resolution) and a gas-
4
1 7 0
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 18, 2003
10.1021/es0264450 CCC: $25.00

2003 Am erican Chem ical Society
Published on Web 08/12/2003

phase chromatograph coupled with a flame ionization
detection (GC/ FID), both connected to the chamber through
a stainless steel line. For the GC-FID analysis, gas samples
were automatically collected from the chamber every 12 min
and introduced via a heated sampling loop of 0.25 mL capacity
onto a DB1 column at 150 °C for HG and 120 °C for DA (J&W
Scientific; 30 m/ 0.32 mm/ 5 µm) mounted in a Varian 3600
CX instrument. OH radicals were produced from the pho-
tolysis of H
2 2
O using two lamps emitting at 254 nm.
(
ii) Experim ents Perform ed at EUPHORE. A detailed
description of the EUPHORE facility and the existing
analytical equipment can be found in refs 7 and 8; however,
a brief description is given here. The EUPHORE facility
consists of two independent hemispherical outdoor simula-
tion chambers, made of fluorene-ethene-propene (FEP)
3
foil, with a volume of 200 m each. The used FEP foil has
more than 80% transmission of the solar radiation in the
wavelength range between 290 and 520 nm. Both chambers
are equipped with FTIR spectrometers coupled with White-
type multipath mirror systems for in-situ analysis. The optical
path lengths were 326.8 m in one chamber and 553.5 m in
the other one. The IR spectra were recorded every 10 min by
FIGURE 1. Stability of hexylene glycol (HG) and diacetone alcohol
(DA) in the chamber.
-
1
co-adding 550 interferograms with a resolution of 1 cm
.
Analyses were also accomplished using gas chromatographs
equipped with different detectors (flame ionization (FID),
the reference compounds using the chamber of LCSR.
photoionization (PID), and electron capture (ECD)). O
3
, CO,
and NO were analyzed using specific monitors, Monitor
x
OH + HG f products kHG
Labs 9810, Thermo Environment 48C, Monitor Labs 9841A
and ECO-Physics CLD770 AL ppt with PLC 760 photolytic
converter, respectively. Reactant and product concentrations
were determined using calibrated reference spectra. Known
amounts of reactants (HG or DA) were introduced into the
OH + DA f products kDA
OH + reference f products kref
6
chambers at the ppm level along with SF used to measure
Assuming that the compounds are only consumed by
reaction with OH, it can be shown that
the dilution rate caused by minor leaks, the thermal expansion
of the reaction mixture, and the sampling of the connected
instruments. In the photolysis experiment performed to
determine the photolysis rate, cyclohexane was also added
to scavenge OH radicals. During the OH-initiated oxidation
experiments, NO was added to initiate the OH radical
formation under sunlight conditions. The actinic flux was
measured by using a calibrated filter radiometer specific to
ln([reactant] / [reactant] ) ) (kreactant/ k ) ×
0
t
ref
(
ln([ref] / [ref] )) (I)
0 t
where the subscripts 0 and t indicate the initial concen-
trations and those at time t, respectively.
the photolysis frequency of NO
2
. For each experiment, to
The rate constant of OH reaction with HG was measured
relative to that of OH with 1,3-dioxolane (kOH ) (1.11 ( 0.09)
check for any dark effect, the analysis of the gas mixture was
started at least 30 min before exposing the mixture to the
sunlight.
-
11
-11
× 10 (10)), acetaldehyde (kOH ) (1.53 ( 0.08) × 10 (11)),
-
12
n-hexane (kOH ) (5.45 ( 0.16) × 10
(12)), and n-octane
-
12
3
-1 -1
UV Absorption Cross Sections Measurem ents. UV-
visible absorption spectrum of diacetone alcohol was de-
termined at room temperature at LCSR/ CNRS-Orl e´ ans.
Detailed description of the set up can be found in previous
papers from our group (e.g. ref 9). The collimated output of
a deuterium lamp passed through an absorption cell and
was focused onto the entrance slit of a spectrograph equipped
with a 1800 grooves/ mm grating, which dispersed the
radiation onto a 1024 element diode array detector. The
absorption cell, equipped with quartz windows, was made
of Pyrex and was 100 cm long with an internal diameter of
(kOH ) (8.67 ( 0.17) × 10 cm molecule
s
(13)), while
that of OH with DA was determined relative to that of OH
with 1,3-dioxolane, n-butyl formate (kOH ) (3.54 ( 0.52) ×
-
12
3
-1 -1
10 cm molecule
s
(14)), and n-hexane. The photolysis
2 2
of H O at 254 nm in air was used as the OH source at
atmospheric pressure and (298 ( 3) K. The initial concentra-
tions of reactants and references were in the range 17-48
ppm. The compounds showed a good stability when they
were mixed in the chamber with H
for about 1 h as shown in Figure 1. Moreover, in the absence
of H , the photolysis of HG in air for more than 2 h did not
2 2
O and kept in the dark
2
O
2
2
2
.5 cm. Measurements were made over the wavelength region
34-330 nm. The wavelength scale was calibrated using lines
show any decrease in the concentrations. On the other hand,
the loss due to the photolysis of DA was about 6% per hour
using two lamps at 254 nm. A correction for this loss has
been applied in the kinetic analysis using the following
formula
from Zn-lamp (213.8 nm) and Hg Penray lamp (253.65, 296.73,
and 365.02 nm).
Chem icals. The chemicals used and their stated purities
were as follows: hexylene glycol (99%) and diacetone al-
cohol (99%). They were obtained from Aldrich. All samples
were degassed by several freeze-pump-thaw cycles before
use.
ln([DA]o/ [DA]t) - kpt ) kDA/ kref ln([ref]o/ [ref]t)
(II)
where kp represents the photolysis rate of DA in our
experimental conditions. At least two measurements were
performed with each reference, except with acetaldehyde,
to check the reproducibility of the results. Analysis of the
species were performed using a FTIR spectrometer with 10
m optical path length and GC/ FID. However, the obtained
Results and Discussion
Kinetic Studies. Relative rate constants were measured by
comparing the OH reaction rate with HG and DA to that with
VOL. 37, NO. 18, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4 1 7 1

FIGURE 2. Plots of eq I for the reaction of hexylene glycol (HG) with OH with 1,3-dioxolane (a), acetaldehyde (b), n-hexane (c), and n-octane
d) used as reference compounds.
(
rate constants for the OH reaction with HG using the GC
analysis were found to be scattered and dependent on the
reference used. For example, using 1,3-dioxolane, n-hexane,
and n-octane, the obtained kHG were in the ranges (2.6-4.7)
The quoted errors represent twice the standard deviation.
The obtained kHG value is compared with that reported
recently by Bethel et al. using relative method and n-octane
-11
cm3 molecule
-1
as a reference (kHG ) (2.77 ( 0.61) × 10
-
11
-11
-11
3
-1
×
10 , (1.3-2) × 10 , and (2-3.9) × 10 cm molecule
-1
s
) (4). This value is almost a factor of 2 higher than that we
-
1
s
, respectively. For this reason we have used only the data
obtained in this work; however, these authors used GC/ FID
for analysis, and as mentioned above the use of this analytical
technique was not conclusive in our experiments. The GC
analyses used in this work were significantly different to the
method used by Bethel et al. These authors collected gas
samples onto solid adsorbent with thermal desorption
directly onto the GC column, while in the present work
samples passed through a gas sampling valve. However, it
has to be noticed that the measured kHG in this work using
GC/ FID and n-octane as reference is in the range (2-3.9) ×
obtained by FTIR analysis, which showed a good reproduc-
ibility independently on the reference used, to derive the
recommended rate constant kHG. Figures 2 and 3 show the
relative losses of HG and DA obtained versus that of the
reference compounds in the presence of OH radicals. The IR
-
1
spectral features used for analysis were 2770.8-3044 cm
-
1
-1
for HG, 2743.8-3141.5 cm for DA, 2599.1-3100.5 cm for
-
1
1
3
,3-dioxolane, 992.7-1218.8 cm for acetaldehyde, 2764-
1
-
-1
059.1 cm for n-hexane, 2783.3-3067.3 cm for butyl
-
1
formate, and 2768.3-3058.6 cm for n-octane. The con-
centrations of reactants were obtained by using a subtraction
procedure. The experimental conditions and the rate constant
obtained for both compounds are summarized in Table 1.
The average values derived from different experiments are
-11
3
-1 -1
1
0
cm molecule
s , which is not in disagreement with
that reported by Bethel et al., but we have more confidence
in the results derived using FTIR analysis for the reason
described above.
The kDA value obtained in this work, kDA ) (3.6 ( 0.6) ×
10- cm molecule
12
3
-1 -1
s , is in agreement with that reported
-
11
cm molecule s-1
3
-1
kHG ) (1.5 ( 0.4) × 10
kDA ) (3.6 ( 0.6) × 10
previously by Atkinson and Aschmann using a relative
-
12
cm3 molecule
-1 -1
method ((4.0 ( 1.2) × 10
s ) (2). These
-
12
cm molecule s-1
3
-1
authors have used n-butane as the reference (k(OH +
4
1 7 2
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 18, 2003

FIGURE 3. Plots of eq I for the reaction of diacetone alcohol (DA) with OH with 1,3-dioxolane (a), n-butyl formate (b), and n-hexane (c)
used as reference compounds.
TABLE 1. Initial Concentrations of Reactants, Slopes k/kref, and the Obtained Rate Constants for the Reaction of OH with HG and
DA at (298 ( 3) K
k (cm molecule s-1
)
3
-1
compound
[reactant]0 (ppm)
reference
[reference]0 (ppm)
k/kref
11
hexylene glycol (HG)
22
1,3-dioxolane
47
28
28
32
35
39
30
30
30
1.23 ( 0.03
(1.35 ( 0.28) × 10-
2
2
2
2
3
2
2
2
8
6
0
9
9
0
3
3
-
11
11
acetaldehyde
n-hexane
1.12 ( 0.12
2.00 ( 0.08
(1.72 ( 0.18) × 10
-
(1.09 ( 0.14) × 10
11
n-octane
1.90 ( 0.07
(1.65 ( 0.20) × 10-
-
11
average
0.35 ( 0.01
(1.5 ( 0.4) × 10
-
12
12
12
diacetone alcohol (DA)
23
1,3-dioxolane
n-butyl form ate
n-hexane
32
34
36
29
32
31
(3.88 ( 0.47) × 10
2
3
2
3
1
0
9
0
0
7
1.10 ( 0.01
0.64 ( 0.01
average
(3.89 ( 0.12) × 10-
(3.49 ( 0.16) × 10-
(3.6 ( 0.6) × 10-
12
-
12
cm3 molecule-1 s-1) and the
ONO) in the presence of NO
n-butane) ) 2.51 × 10
The measured values of kDA and kHG have been compared
to those calculated using the structure-activity relationship
(SAR) of Atkinson (15, 16) revised recently by Bethel et al. (4)
photolysis of methyl nitrite (CH
as the OH source.
3
VOL. 37, NO. 18, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4 1 7 3

FIGURE 4. OH-initiated oxidation of HG: FTIR spectra before (t ) 0) and after irradiation (t ) 90 min) of the mixture (Orl e´ ans).
for alcohols and diols. Bethel et al. considered the effects of
the OH substituent(s) on H-atom abstraction at the R- and
â-positions, while in the previous SAR only the effect at
R-position was considered (15, 16). Using the new substituent
factors derived from the updated estimation, F(-OH) ) 2.9
and F(-CH OH) ) F(>CHOH) ) F(>COH) ) 2.6, a good
2
agreement is obtained for kHG between the calculated and
-
11
-11
the experimental values, 1.46 × 10 and (1.5 ( 0.4) × 10
3
-1 -1
cm molecule
s
, respectively. For kDA, the calculated value
-
12
is ≈25% lower than the experimental one, 2.8 × 10
and
-
12
3
-1 -1
(
3.6 ( 0.6) × 10
(in cm molecule ), respectively.
s
However, as it was mentioned by Bethel et al., even with the
use of the new subtituent factors a disagreement still exists
between the calculated and estimated rate constants for OH
reactions with some oxygenated compounds.
Product Studies. Experiments on the OH-initiated oxida-
tion of hexylene glycol and diacetone alcohol were carried
out in different conditions using two photoreactors at Orl e´ ans
and at EUPHORE as described in the Experimental Section.
Hexylene Glycol (HG). At Orl e´ ans, the experiments were
performed in the 140 L bag using H
source, in the presence and in the absence of NO
8-42 ppm, [HG] ) 19-46 ppm). The mixtures were
irradiated for a period of time ranging between 2 and 3 h
leading to more than 50% consumption of HG. Figure 4 shows
an example of typical IR spectra of the gas mixture recorded
before and after 90 min of irradiation. The main primary
2
O
2
as the OH radicals
FIGURE 5. HG and corrected products concentrations versus reaction
time in the OH-initiated oxidation of HG in air in the presence of
NO at EUPHORE.
x
x
([NO] )
0
3
0
derived from this work for kHG and kDA and from ref 18 for
the other OH reactions.
At EUPHORE, a single experiment was conducted where
products observed, in the absence or in the presence of NO
were diacetone alcohol (DA) with yields of (49 ( 4)% and (44
4)%, respectively, and acetone with yields of (27 ( 6)% and
35 ( 4)%, respectively. CO was observed as a secondary
product in the absence and in the presence of NO , while
PAN was produced in the presence of NO also as secondary
x
,
7
50 ppb of HG were introduced into the chamber in a stream
of dry air in the presence of NO (<120 ppb). The main
observed products were similar to those observed in the
laboratory (DA: (51 ( 4)% and acetone: (22 ( 1)%). A small
amount of PAN was also detected and estimated to be (1.0
(
(
x
x
(
0.1)% at the end of the experiment.
product. The IR spectral features used for analysis product
studies were the same as those used for the kinetic studies
It has to be noticed that the carbon balance in both types
-
1
-1
(
2770.8-3044 cm for HG and 2743.8-3141.5 cm for DA).
of experiments is only about 65% which may indicate the
presence of unidentified and not quantified products such
as hydroxyaldehydes. The concentration-time profiles of
HG and the products were obtained from integration of the
The concentration of reactants and products were obtained
by using a subtraction procedure. Identification and quan-
tification of the reactants and the reaction products were
made either by introducing known concentrations of au-
thentic samples (HG, DA, acetone, acetaldehyde) or by using
our IR library (formaldehyde, PAN, methyl glyoxal, CO).
Corrections for the photolysis of DA (6%/ h) and acetone
-
1
IR spectra between 2771 and 3044 cm for HG, 2744 and
-
1
-1
3142 cm for DA, 1150 and 1280 cm for acetone, 1103 and
-
1
1215 cm for PAN. The concentration-time profiles of HG
and its oxidation products in the presence of NO at
x
(
10%/ h) were made to derive the products yields. The
EUPHORE are shown in Figure 5, and the product yield plots
are presented in Figure 6.
The obtained products and their yields indicate that the
OH reaction with hexylene glycol (in the absence or in the
measured concentrations of the observed products were
corrected for their secondary reaction with OH radicals as
described by Atkinson et al. (17) using the rate constants
4
1 7 4
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 18, 2003

(
CH ) C(OH)CH C°(OH)CH + O f
3 2 2 3 2
(
CH ) C(OH)CH C(O)CH + HO
3 2 2 3 2
Since no other reaction of this dihydroxy alkyl radical is
expected, the formation yield of DA of (47 ( 4)% obtained
experimentally also corresponds to the branching ratio of
channel 1b of the OH + HG reaction.
The branching ratio of channel 1a can be derived from
the measured yields of acetone (31 ( 6)% (channel 1a1) and
PAN (2.4 ( 0.9)% (channel 1a2), this latter being formed
from the degradation of acetaldehyde, which leads to (33 (
7
)% of the reaction. However, this branching ratio has to be
considered as a lower limit because of the possible reactions
involving the peroxyacetyl radicals (CH C(O)O ) which will
not lead ultimately to the formation of PAN.
3
2
The measured products yields and derived branching
ratios are consistent with those calculated using the SAR
method. Effectively, the SAR method predicts a branching
ratio of 43 and 47% for channels 1a and 1b, respectively, the
H-atom abstraction from the -CH
3
and -OH groups
representing less than 10% of the reaction. The DA formation
yield obtained in this work is also in good agreement with
that reported by Bethel et al. (4) who observed diacetone
alcohol as the major product of OH-initiated oxidation of
hexylene glycol with a yield of (47 ( 9)%. The relatively poor
carbon balance (65%) in our experiments may be due to the
difficulties in detecting the hydroxyaldehydes formed in
channel 1a.
FIGURE 6. Corrected concentrations of products versus consumption
of HG in the OH-initiated oxidation of HG in air at EUPHORE.
presence of NO
x
) proceeds mainly by H-atom abstraction
- and >CH- groups through the following
from -CH
2
channels (see mechanism in Scheme 1):
Diacetone Alcohol (DA). At Orl e´ ans, experiments were
OH + (CH ) C(OH)CH CH(OH)CH f
x
performed, in the absence and in the presence of NO , in
3
2
2
3
similar conditions to those for HG. The concentrations of
DA and NO were in the range 22-34 and 38-42 ppm,
respectively. An example of the obtained IR spectra before
and after 90 min irradiation of the mixture is given in Figure
(
CH ) C(OH)C°HCH(OH)CH + H O (1a)
3 2 3 2
OH + (CH ) C(OH)CH CH(OH)CH f
3
2
2
3
(
CH ) C(OH)CH C°(OH)CH + H O (1b)
3 2 2 3 2
7
. The main primary oxidation products obtained in the
experiments performed in the absence of NO were acetone
x
The dihydroxyalkyl radicals formed in reaction 1a, cor-
responding to the abstraction of H-atom from -CH - group
of HG, react with O to form dihydroxyalkyl peroxy radicals
CH C(OH)CH(OO°)CH(OH)CH that react with HO radi-
) radicals or predominantly
(
80 ( 10)%, formaldehyde (21 ( 3)%, and CO. In the presence
2
of NO , the main observed products were acetone (94 ( 14)%,
x
2
formaldehyde (not quantified), and PAN (10 ( 2)%. Cor-
rections were made to take into account the photolysis of DA
and of acetone as well as for the reaction with OH radicals
as described for HG product studies. In addition to the above
products, GC/ MS analysis indicated the presence of mass
(
3
)
2
3
2
cals and other organic peroxy (RO
with NO (when present):
2
+
+
(
(
CH ) C(OH)CH(OO°)CH(OH)CH + HO f
fragments corresponding to 28 (CO ), 43 (CH
3
C(O) ), 59
3
2
3
2
+
+
+
(
(CH
3
)
2
C(OH) ), 87 ((CH
3
)
2
C(OH)C(O) ), and 113 (C
5
H
9
O
3
)
(
CH ) C(OH)CH(OOH)CH(OH)CH + O
3
2
3
2
2
which led us to assume that the corresponding parent
compound would be (CH C(OH)C(O)C(O)CH . The exist-
3
)
2
3
CH ) C(OH)CH(OO°)CH(OH)CH + RO f
3
2
3
2
ence of this latter compound could not be proven since we
could not check it with an authentic sample. GC analysis
indicated also the presence of a number of small peaks which
could not be identified.
At EUPHORE, two experiments were carried out where,
respectively, 616 and 968 ppb of DA were introduced into
the chamber in a stream of dry air in the presence of NO
(<120 ppb). The main observed products were similar to
those observed in the laboratory: acetone (88 ( 12)%,
formaldehyde (48 ( 7)%, and PAN (11 ( 2)%.
(
CH ) C(OH)CH(O°)CH(OH)CH + RO° + O
3
2
3
(
(
CH ) C(OH)CH(OO°)CH(OH)CH + RO f
3 2 3 2
other products
CH ) C(OH)CH(OO°)CH(OH)CH + NO f
3
2
3
(
CH ) C(OH)CH(O°)CH(OH)CH + NO
3 2 3 2
The dihydroxyalkoxy radicals ((CH
OH)CH ) formed through the reactions with RO
or decompose by C-C bond scission to
form an hydroxyaldehyde (CH CH(OH)CHO) and acetone
) and/ or another hydroxyaldehyde ((CH C-
OH)CHO) and acetaldehyde (CH CHO). The reaction with
may lead to (CH C(OH)C(O)CH(OH)CH as a stable
compound, but no evidence was found for its formation.
The second channel (1b) proceeding by H-atom abstrac-
tion from the >CH-group leads to the formation of diacetone
3
)
2
C(OH)CH(O°)CH-
It has to be mentioned that for all experiments, the carbon
balance was about 53% which indicates the presence of other
oxidation products that we could not detect or quantify such
(
3
2
and NO
may react with O
2
3
3 2 3
as hydroxyaldehydes and (CH ) C(OH)C(O)C(O)CH . The
(
(
O
CH
3
C(O)CH
3
3
)
2
concentration-time profiles of DA and of the products were
3
obtained from integration of the IR spectra between 2744
-
1
-1
2
3
)
2
3
and 3142 cm for DA, 1150 and 1280 cm for acetone, 1103
-
1
-1
and 1215 cm for PAN, 2025 and 2225 cm for CO, and 2638
-
1
and 3079 cm
for formaldehyde. Examples of typical
concentration-time profiles of DA and its oxidation products
as well as the derived product yield plots are shown in Figures
8 and 9, respectively.
alcohol through the reaction of (CH
with O
3 2 2 3
) C(OH)CH C°(OH)CH
2
:
VOL. 37, NO. 18, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4 1 7 5

FIGURE 7. OH reaction with DA: an example of the obtained FTIR spectra before (t ) 0) and after irradiation (t ) 90 min) of the mixture
Orl e´ ans).
(
SCHEME 1: OH-Initiated Oxidation Mechanism of Hexylene Glycol in Air in the Presence or Absence of NO
x
The primary radicals formed in reactions 2a, 2b, and 2c
will react with O followed by reactions with NO or RO (in
the absence of NO) to lead to the corresponding RO radicals.
The H-atom abstraction from the -CH - group (2a) leads
to the oxy radical (CH C(OH)CH(O°)C(O)CH which may
react with O or decompose (Scheme 2). The reaction with
may lead to (CH C(OH)C(O)C(O)CH (4-hydroxy-4-
The OH reaction with diacetone alcohol proceeds mainly
groups
through the following channels (see Schemes 2 and 3):
by H-atom abstraction from -CH
2
- and -CH
3
2
2
2
3
)
2
3
OH + (CH ) C(OH)CH C(O)CH f
3
2
2
3
2
(
CH ) C(OH)C°HC(O)CH + H O (2a)
3 2 3 2
O
2
3
)
2
3
methyl-2,3-pentanedione) as it has been suggested from GC/
MS analysis, but this could not be confirmed. The decom-
f C°H (CH )C(OH)CH C(O)CH + H O
(2b)
(2c)
2
3
2
3
2
position m ay form m ethylglyoxal (CH
3
C(O)CHO) and
C°(O). Both
f (CH ) C(OH)CH C(O)C°H + H O
3
2
2
2
2
(CH C°(OH) and/ or ((CH C(OH)CHO) and CH
3
)
2
3
)
2
3
4
1 7 6
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 18, 2003

proceeds mainly through channels 2a1 and 2c as shown in
Schemes 2 and 3. However, considering that the H-atom
abstraction from a -CH
3
group in an R-position of a carbonyl
group is not significant compared to that from a -CH
2
-
group in the same position, we conclude that the reaction
channel 2a1 is the dominant one. Concerning the reaction
pathway 2a2, it must not be neglected since we have detected
PAN as primary product with a yield of ≈11%. The slight
difference between the acetone yield in the absence and in
the presence of NO
involving peroxyl radicals such as (CH
O)CH and RO or HO to form hydroxyperoxides which
x
could be due to the existence of reactions
3 2
)
C(OH)CH(OO°)C-
(
3
2
2
were not detected in our experiments. The obtained yield of
acetone compared to that of PAN from channels 2a1 and 2a2
indicate that (CH
pose mainly through C-C cleavage from -C(OH)CH(O°)-,
and the breaking of the C-C bond from -CH(O°)C(O)CH
3 2 3
) C(OH)CH(O°)C(O)CH radicals decom-
3
is minor. It has to be mentioned that carbon balance during
these experiments was relatively poor (≈50%) which indicates
the formation of a number of nondetected products.
UV-Absorption Spectrum and Photolysis Rate of Di-
acetone Alcohol. UV-Absorption Spectrum. The obtained UV-
absorption spectrum of diacetone alcohol is shown in Figure
FIGURE 8. DA and corrected products concentrations versus reaction
time in the OH-initiated oxidation of DA in air in the presence of
NO at EUPHORE.
x
1
2
0 and the measured values of cross sections averaged over
nm intervals and given in Table 2. Absorption cross sections
at different wavelengths (λ, nm) were obtained using the
Beer-Lambert law (in cm2 molecule
)
-1
σ(λ) ) -ln[I(λ)/ I (λ)]/ LC
(III)
0
where I and I
empty cell, respectively, L is the length of the absorption cell
in cm), and C is the concentration of the compound in the
0
are the light intensities for the filled and the
(
-
3
cell (molecule cm ). The measurements were performed at
0 and 50 °C, in static conditions and in a reduced pressure
3
range (0.3-0.8) Torr due to the low vapor pressure of DA.
The pressure was measured using a 0-10 Torr capacitance
manometer. No difference was observed between the cross
sections obtained at these two temperatures. We have also
measured the cross sections at 213.8 and 254 nm using Zn
and Hg pen-ray lamps. For these latter measurements, the
deuterium lamp was replaced by Zn (213.8 nm) or Hg (254
nm) lamps, and the transmitted light was detected by a
photodiode using a filter at 214 or 254 nm depending on the
lamp used. The obtained value at 254 nm using this meth-
od is in agreement with that obtained using the deuterium
-
20
-20
cm2 molecule
,
-1
FIGURE 9. Corrected concentrations of products versus consumption
of DA in the OH-initiated oxidation of DA in air at EUPHORE.
lamp (8.46 × 10
respectively). At 213.8 nm, the obtained value is σ ) 2.12 ×
cm2 molecule . The absorption spectrum given in
Figure 10 shows a maximum around 283 nm (σ ) 8.91 ×
cm molecule ) and indicates that the diacetone alcohol
and 7.57 × 10
-
19
-1
1
0
radicals (CH
form CH C(O)CH
quently, CH C(O)O
NO with CH C°(O)O
react with NO and lead ultimately to HCHO and CO
depending on the NO/ NO ratio. It has to be mentioned that
(CH C(OH)CHO) could also be formed in the HG oxidation
3
)
2
C°(OH) and CH
and CH C°(O)O
NO (PAN) will be formed by reaction of
. The peroxy acetyl radical could also
3
C°(O) will react with O
2
to
-20
2
-1
1
0
3
3
3
2
, respectively. Subse-
may be photolyzed in the atmosphere considering the
absorption cross sections values at wavelength higher than
3
2
2
2
3
2
2
90 nm.
Measurem ent of the Photolysis Rate Constant for DA
2
2
under Atm ospheric Conditions. One photolysis experiment
of diacetone alcohol was conducted using the EUPHORE
(
3 2
)
as shown in Scheme 1. The residual IR spectra in both
outdoor smog chamber. A known amount of DA ([DA]
.45 ppm) was introduced into the chamber in a stream of
dry air. The experiment was performed in the presence of
o
)
experiments, with HG and DA, showed bands around 1150-
1
-
1
1
200 and 1050 cm regions; it cannot be excluded that these
bands correspond to (CH
Channel 2b proceeding by H-atom abstraction from -CH
groups in R-position of the >C(OH)- group, presented in
Scheme 3, leads to the formation of HCHO and CH C(O)-
CH C(O)CH which has not been detected indicating that if
formed its concentration is below the detection limit. Channel
3 2
) C(OH)CHO.
SF
6
, to derive the dilution rate (kSF6) which is given by
3
ln([SF ] / [SF ] ) ) k × t
(IV)
6
0
6 t
SF6
3
2
3
where [SF
6
]
0
and [SF
6
]
t
are the initial SF
6
concentration and
2
c leads also to the formation of HCHO and to acetone.
The high experimental yields of acetone (80 ( 10)% and
at reaction time t, respectively. This experiment has been
conducted in air in the presence of cyclohexane in order to
scavenge any OH radicals formed during the photolysis. In
these conditions, DA is lost by photolysis and by dilution.
(
x
91 ( 8)% in the absence and in the presence of NO ,
respectively, indicates that the reaction of OH with DA
VOL. 37, NO. 18, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4 1 7 7

SCHEME 2: OH-Initiated Oxidation Mechanism for the Reaction of OH with Diacetone Alcohol in Air in the Presence or Absence
of NO (H-Atom Abstraction from the -CH - Group)
x
2
Therefore, it can be shown that
ln([DA] / [DA] ) ) (J + kSF6)t ) ktotal × t
2
The photolysis rate of NO measured during the experi-
) ) 9.3 × 10 s-1. The data obtained showed
-3
ment was J(NO
2
(V)
that the decay of DA during that experiment was only due
to dilution. The decay rate was then determined from the
0
t
where J is the photolysis coefficient of diacetone alcohol,
and kSF6 the dilution rate.
first-order decay of SF
limit of the photolysis rate of DA: Jexp < 5 × 10
6
. This enables us to derive an upper
-
6
-1
s
.
4
1 7 8
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 18, 2003

SCHEME 3: OH-Initiated Oxidation Mechanism for the Reaction of OH with Diacetone Alcohol in Air in the Presence or Absence
of NO (H-Atom Abstraction from the -CH - Groups)
x
3
The photolysis rate can be also calculated using to the
following formula: Jcalc ) ∫ σ(λ) × φ(λ) × I(λ) × dλ, where σ(λ)
are the cross sections measured in this work, I(λ) (in photons
the experiment, and φ(λ) is the quantum yield assumed to
be equal to unity. The calculated photolysis rate was found
-
5
-1
to be Jcalc ) 2.1 × 10
s . Effective quantum yield, defined
-
2
-1
cm
s ) is the actinic flux measured at EUPHORE during
as φeff ) Jexp/ Jcal, was then derived (φeff e 0.2) indicating that
VOL. 37, NO. 18, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4 1 7 9

x
a significant source of HO radicals which may influence the
ozone budget (e.g. ref 19). However, in addition to the gas-
phase loss by OH reaction other possible loss processes of
HG and DA such as reactions in solutions in the troposphere
have to be taken into account to assess the overall atmo-
spheric impact.
Considering only the gas-phase loss by OH reaction, the
impact that could have the oxidation of HG and DA on the
ozone formation in the troposphere can be estimated for
instance by calculating their photochemical ozone creation
potential (POCP). The POCP characterizes the potential of
a VOC to form ozone in the tropospheric boundary layer
through Europe during 5 days along a given trajectory. The
POCP is derived from chemistry-transport models (20), but
it has been shown by Jenkin et al. (21) that it can also be
reliably estimated for a large variety of VOC from parameters
defining the structure of the VOC, its reactivity (kOH), the
yields and the reactivity of its oxidation products not highly
-
12
cm3 molecule
-1 -1
reactive (kOH e 10
s ).
POCP of HG and DA have been estimated using their
reaction rate constants with OH measured in this work as
well as the obtained products and their yields. The only
unreactive compound considered in these calculations was
acetone which was observed as a reaction product (30% in
the OH-initiated oxidation of hexylene glycol and 80% in
that of diacetone alcohol). The estimated POCP of hexylene
glycol and diacetone alcohol are 47 and 17, respectively (the
POCP of the reference compound, ethylene, is by definition
FIGURE 10. UV-absorption spectrum of diacetone alcohol.
TABLE 2. Absorption Cross Sections of Diacetone Alcohol at
303 ( 1 K
λ
σ
λ
σ
2
20
2
(
nm) (×10- cm molecule-1) (nm) (×10-20 cm molecule-1)
1
00). The POCP of HG, 47, is similar to that of other diols
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
34
36
38
40
42
44
46
48
50
52
54
56
58
60
62
64
66
68
70
72
74
76
78
80
82
14.8
13.6
12.2
11
284
286
288
290
292
294
296
298
300
302
304
306
308
310
312
314
316
318
320
322
324
326
328
330
8.89
8.66
8.55
8.22
7.92
7.37
6.89
6.44
5.96
5.34
4.70
4.13
3.56
3.04
2.53
2.01
1.60
1.22
0.949
0.679
0.490
0.346
0.238
0.184
such as propylene glycol (HOCH CH CH OH) which has an
2
2
2
estimated POCP of 45 (21), the obtained POCP by modeling
being 51 (20). Compared to propylene glycol, only a small
difference can be expected between the estimated POCPs
and those obtained by models for other glycols such as HG.
For diacetone alcohol the estimated POCP is low (17) because
of the low reactivity of DA and that of acetone, its main
oxidation product. The POCP of DA has also been calculated
from a model and found to be 26 (20).
9.97
9.07
8.38
8.02
7.84
7.65
7.57
7.44
7.37
7.33
7.29
6.99
7.07
7.20
7.50
7.73
8.17
8.46
8.68
8.84
8.90
Acknowledgments
We thank TOTALFINAELF company and CNRS through the
PNCA program for financial support and Dr. K. Wirtz and
co-workers for fruitful cooperation at EUPHORE labora-
tories.
Literature Cited
(
(
1) http:/ / www.scorecard.org/ .
2) Atkinson, R.; Aschmann, S. M. Environ. Sci. Technol. 1995, 29,
5
28.
3) Aschmann, S. M.; Arey, J.; Atkinson, R. Environ. Sci. Technol.
000, 34, 1702.
(4) Bethel, H. L.; Atkinson, R.; Arey, J. Int. J. Chem. Kinet. 2001, 33,
10.
5) Th e´ venet, R.; Mellouki, A.; Le Bras, G. Int. J. Chem. Kinet. 2000,
2, 676.
(
2
photolysis is not an important process in the tropospheric
degradation of diacetone alcohol at least under our experi-
mental conditions.
Atm ospheric Im plications. Similarly to other simple
alcohols and ketones, the reactions of hexylene glycol and
diacetone alcohol with other atmospheric oxidants such as
3
(
(
3
6) Magneron, I.; Th e´ venet, R.; Mellouki, A.; Le Bras, G. J. Phys.
Chem. A 2002, 106, 2526.
(7) The European photoreactor EUPHORE; Becker, K. H., Ed.;
O
3
and NO
3
are expected to be negligible. The experiment
Final Report of the EC project EV5V-CT92-0059; Wuppertal,
1
996.
carried out at EUPHORE on the photolysis of diacetone
alcohol showed that this process is negligible in the tropo-
sphere. Therefore, the OH reaction is the major tropospheric
loss process for HG and DA in the gas phase. The tropospheric
lifetimes of HG and DA with respect to the reaction with OH
were calculated to be around 20 h and 3 days, respectively,
(
(
8) EUPHORE report 1997; Barnes, I., Wenger, J., Ed.; Institute of
Physical Chemistry, Bergische Universit a¨ t: Wuppertal, 1998.
9) Mu, Y.; Mellouki, A. J. Photochem. Photobio. A: Chem. 2000,
1
34, 31.
(10) Le Calv e´ , S.; Le Bras, G.; Mellouki, A. Phys. Chem. Chem. Phys.
2002, 4, 5622.
(11) Sivakumaran, V.; Crowley, J. Phys. Chem. Chem. Phys. In press.
(12) Donahue, N. M.; Anderson, J. G.; Demerjian, K. L. J. Phys. Chem.
A 1998, 102, 3121.
6
using the measured rate constants and [OH] ) 10 molecule
-
3
cm . Both compounds will be oxidized at a local or regional
scale. Their atmospheric oxidation leads to high yield of
acetone, this latter has been recently recognized to play an
important role in the upper troposphere where it provides
(
13) Atkinson, R. J. Phys. Chem. Ref. Data 1997, 26, 215.
(14) Le Calv e´ , S.; Le Bras, G.; Mellouki, A. J. Phys. Chem. A 1997, 101,
5489.
4
1 8 0
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 18, 2003

(
(
(
15) Atkinson, R. Chem. Rev. 1986, 86, 69.
(20) Derwent, R. G.; Jenkin, M. E.; Saunders: S. M.; Pilling, M. J.
Atmos. Environ. 1998, 32, 2429.
(21) Jenkin, M. E. Report of contract 1/3/70; September 1998.
16) Kwok, E. S. C.; Atkinson, R. Atm. Environ. 1995, 29, 1685.
17) Atkinson, R.; Aschmann, S. M.; Carter, W. P. L.; Winer, A. M.;
Pitts, J. N., Jr. J. Phys. Chem. 1982, 86, 4563.
(
(
18) Atkinson, R.; Baulch, D. L.; Cox, R. A.; Hampson, R. F., Jr.; Kerr,
J. A.; Rossi, J. A.; Troe, J. J. Phys. Chem. Ref. Data 1999, 28, 191.
19) Jaegl e´ , L.; Jacob, D. J.; Brune, W. H.; Wennberg, P. O. Atmos.
Environ. 2001, 35, 469.
Received for review December 19, 2002. Accepted June 30, 2003.
ES0264450
VOL. 37, NO. 18, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4 1 8 1