Laboratory studies of the OH-initiated photooxidation of di-n-propyl ether

Article abstract of DOI:10.1002/1097-4601(2000)32:11<703::AID-KIN6>3.0.CO;2-Z

The OH-initiated photooxidation of di-n-propyl ether was investigated in this study. Di-n-propyl ether was mixed with nitric oxide and a hydroxyl radical precursor and irradiated using UV black lamps in a glass environmental chamber. Mass spectrometry was used as the primary analytical technique to monitor the reactants and products. FTIR spectroscopy was used to monitor formaldehyde. The products observed were propyl formate, acetaldehyde, propionaldehyde, and propyl propionate, with molar yields relative to di-n-propyl ether concentration loss of 0.61±0.044, 0.60±0.057, 0.15±0.062, and 0.043±0.015, respectively. Errors represent ±2σ. Nitrates could not be quantified because of a lack of commercially available standards. However, evidence exists for nitrate formation from the photooxidation of di-n-propyl ether. Formaldehyde concentrations were negligible. Mechanism predictions were performed on the di-n-propyl ether/OH system using the Carter kinetic software. Propyl formate and acetaldehyde yields were reasonably predicted (under 11.7% error). However, propionaldehyde and propyl propionate yields were vastly underpredicted, and examination of the experimental data suggested secondary production of both propionaldehyde and propyl propionate. Reactions were proposed for the photolysis and OH-initiated photooxidation of a primary nitrate product (1-propoxy propyl nitrate) that resulted in the formation of propionaldehyde and propyl propionate. Basic semiempirical computational chemistry calculations at the UHF/PM3 level of theory were performed using Hyperchem to investigate pathways for the secondary formation of propionaldehyde in particular.

Full text of DOI:10.1002/1097-4601(2000)32:11<703::AID-KIN6>3.0.CO;2-Z

Laboratory Studies
of the OH-Initiated
Photooxidation of
Di-n-propyl Ether
JOHN SEMPELES, JEAN M. ANDINO
Environmental Engineering Sciences Department, University of Florida, P.O. Box 116450, Gainesville, FL 32611-6450
Received 24 January 2000; accepted 15 June 2000
ABSTRACT: The OH-initiated photooxidation of di-n-propyl ether was investigated in this
study. Di-n-propyl ether was mixed with nitric oxide and a hydroxyl radical precursor and
irradiated using UV black lamps in a glass environmental chamber. Mass spectrometry was
used as the primary analytical technique to monitor the reactants and products. FTIR spec-
troscopy was used to monitor formaldehyde. The products observed were propyl formate,
acetaldehyde, propionaldehyde, and propyl propionate, with molar yields relative to di-n-
propyl ether concentration loss of 0.61 Ϯ 0.044, 0.60 Ϯ 0.057, 0.15 Ϯ 0.062, and 0.043 Ϯ 0.015,
respectively. Errors represent Ϯ2␴. Nitrates could not be quantified because of a lack of
commercially available standards. However, evidence exists for nitrate formation from the
photooxidation of di-n-propyl ether. Formaldehyde concentrations were negligible.
Mechanism predictions were performed on the di-n-propyl ether/OH system using the Car-
ter kinetic software. Propyl formate and acetaldehyde yields were reasonably predicted (under
1
1.7% error). However, propionaldehyde and propyl propionate yields were vastly underpre-
dicted, and examination of the experimental data suggested secondary production of both
propionaldehyde and propyl propionate. Reactions were proposed for the photolysis and OH-
initiated photooxidation of a primary nitrate product (1-propoxy propyl nitrate) that resulted
in the formation of propionaldehyde and propyl propionate. Basic semiempirical computa-
tional chemistry calculations at the UHF/PM3 level of theory were performed using Hyper-
chem௡ to investigate pathways for the secondary formation of propionaldehyde in particular.
᭧
2000 John Wiley & Sons, Inc. Int J Chem Kinet 32: 703–711, 2000
INTRODUCTION
and tert-amyl methyl ether (TAME) were proposed for
use in automotive fuels [2].
Ethers are currently used as industrial solvents and as
fuel additives.Oxygenated compounds such as methyl
tert-butyl ether (MTBE) and ethyl tert-butyl ether
Due to the high volatility of ethers, evaporation into
the troposphere is not uncommon.Given the increased
use of ethers in new fuels and in combustion pro-
cesses, it is important to assess the impact of these
compounds on the atmosphere.Investigations into the
atmospheric kinetics and mechanisms of ethers such
as MTBE, ETBE, and DIPE with .OH radicals have
already been published [3–6].In general, the mecha-
nisms of hydroxyl attack on ethers and the reactivity
of alkoxy radicals derived from ethers are not well
(ETBE) are added to gasoline to increase the octane
number and reduce the production of CO [1].In past
years, compounds such as di-isopropyl ether (DIPE)
Correspondence to: J.M.Andino (andino@ufl e. du).
Contract grant sponsor: National Science Foundation
Contract grant number: Grant ATM-9702791
᭧
2000 John Wiley & Sons, Inc.

7
04
SEMPELES AND ANDINO
understood.Atkinson [7] has remarked that alkoxy
radicals produced from hydroxyl/ether reactions have
reaction rates that cannot currently be accurately pre-
dicted.Additional work is clearly required.
Ethers are degraded in the atmosphere predomi-
nately by hydroxyl radical reaction [8].Removal by
photolysis and reaction with ozone is considered slow
enough to be negligible [9,10].The hydroxyl radical
reaction with ethers proceeds via abstraction of a hy-
drogen atom from 9CH9, 9CH 9, or 9CH3
2
groups.Previous studies of ethers have suggested that
the oxygen in ethers has a long-range activating effect
on the hydrocarbon chain [11–15].Ethers display
9
CH 9 group reactivities approximately three
2
times higher than 9CH 9 groups in alkanes [11].
2
This inductive effect appears to decrease after the delta
or epsilon carbon.After this point, the reactivities of
9
CH 9 groups in ethers become indistinguishable
2
from those of 9CH 9 groups in alkanes.The acti-
2
vating effects for oxygen observed in previous studies
are not expected from either thermochemical data or
inductive effects [13].Nelson has suggested that the
increased reactivity must be due to other undetermined
mechanistic issues [13].
Figure 1 Diagram of the reaction chamber system.
Previous studies of di-n-propyl ether have focused
on the kinetics of the DOH-initiated reaction
light intensity within the chamber.The rate constant
of NO photolysis in the presence of N was used as
2
2
[
11,13,16,17].To date, product studies on this com-
a measure of the lamp intensity and was determined
Ϫ1
pound have not been reported in the literature.Product
studies of similar ethers such as di-methyl ether and
di-ethyl ether have been reported in the literature
to be approximately 0.5 min .Two black drapes cov-
ered the chamber to prevent outside light from enter-
ing the chamber.All experiments were carried out at
room temperature (298 Ϯ 2 K) and at 760 Torr.
Gas samples from the experiments were analyzed
primarily using a Finnigan MAT GCQ௡ system
[3,18,19].The study performed at the University of
Florida serves to expand the available database of re-
action mechanisms of ethers.
(Model 9001 gas chromatograph coupled to a mass
spectral detector).The samples were drawn into a
nickel sample loop (ca.1 ml) and were injected onto
a J&W Scientific DB-WAX GC column (0.25 mm ϫ
30 m ϫ 0.25 ␮m) using a heated 6-port variable res-
trictor metering valve.The GC temperature program
used was: 35ЊC for 5 min.followed by a ramp of 10 ЊC/
min.to a final temperature of 150 ЊC.Ultra-high-purity
helium (BITEC, Inc.) was used as the carrier gas. Mass
spectral analyses were performed in electron ioniza-
tion mode.
The contents of the chamber were also analyzed in
situ using a Fourier transform infrared (FTIR) spec-
trometer (Model M, MIDAC, Inc.). The MIDAC in-
strument initiated an IR beam that was directed into
the chamber through a 3-in.diameter, 3-mm thick KCl
transmission window (International Crystal).The
beam was reflected multiple times inside the chamber
using two opposite-facing silver-coated mirrors, which
were custom-manufactured by Infrared Analysis, Inc.
(Figure 1 only illustrates four passes of the IR beam
used in infrared detection and analysis.) The IR beam
EXPERIMENTAL METHODS
The reactions of di-n-propyl ether were conducted in
a 213-L borosilicate glass chamber that was custom-
manufactured by Infrared Analysis Inc.(Anaheim,
CA).Gas-phase reactants were introduced into the
chamber using a glass manifold system, which con-
sisted of a calibrated 1.87-L glass bulb and MKS Bar-
atron pressure gauges (MKS Inc.). Boyle’s law was
used to calculate the concentrations in parts per million
by volume.Dry, zero air (Strate Welding Inc ). was
used to flush the reactants into the chamber, and acted
as the diluent.Mixing was achieved using an inter-
nal brushless Teflon௡-coated mixing fan (Infrared
Analysis, Inc.). A schematic of the system appears in
Figure 1.
Photolysis reactions were initiated by 14 UV lamps
(GE F40BL).The lamps were mounted on a cylindri-
cal surface around the chamber and were surrounded
by a custom-designed metal reflector to maximize

OH-INITIATED PHOTOOXIDATION OF DI-n-PROPYL ETHER
705
exited through a 3-in.diameter, 3-mm thick KCl trans-
mission window and was collected by an external
MCT detector.The path length used for these studies
was 144 m, achieved by making 48 passes through the
compared with each other to confirm that reactions had
halted.Indeed, given the excellent agreement (3%) be-
tween duplicate samples analyzed via GC-MS for a
single photolysis event, the reactions had halted.
Separate studies were conducted using the GC-MS
on di-n-propyl ether and the products to determine the
extent of loss via photolysis and to the walls.All stud-
ies indicated negligible (Ͻ5% total) decrease in con-
centrations over time scales typical of those used in
the product studies.Aldehyde and ester loss due to
photolysis was not observed in the given experimental
system, as expected from consideration of absorption
cross section data [9].
3
-m-long chamber.The minimum and maximum
achievable path lengths were 6 m and 360 m, respec-
tively.Experiments were performed with an IR reso-
Ϫ
1
lution of 0.5 cm and a scan range of 3000 to 500
Ϫ
1
cm .Infrared spectra were developed by coadding 16
interferogram scans.
Hydroxyl radicals were generated from the photol-
ysis of methyl or ethyl nitrite in air containing NO.
The mechanism is as follows:
Because of the ease in separating multicomponent
systems via GC, product peaks were identified by GC-
MS analysis using retention times and mass spectra.
The FTIR spectrometer was used only to quantify for-
maldehyde.Product concentrations were determined
from GC-MS analysis by using standards to find peak
area/concentration ratios.Formaldehyde concentra-
tions were determined from laboratory-calibrated
FTIR reference spectra available in our standard ref-
RCH ONO ϩ h␯ !: RCH OD ϩ NO
(1)
(2)
(3)
2
2
RCH OD ϩ O !: RCHO ϩ HO D
2
2
2
HO D ϩ NO !: DOH ϩ NO2
2
where R represents H for methyl nitrite and CH for
ethyl nitrite.
3
ϩ
When methyl nitrite is used to generate DOH radi-
cals, formaldehyde is also produced.Since formalde-
hyde is a potential product of di-n-propyl ether pho-
tooxidation, experiments were performed with ethyl
nitrite as the source of DOH radicals to specifically
search for formaldehyde as a product.In such exper-
iments, ethyl nitrite produces acetaldehyde instead of
formaldehyde.
Both hydroxyl radical precursors were prepared in
gram quantities using the method of Taylor et al.[20].
Methyl nitrite was stored at room temperature in a
stainless steel lecture bottle while ethyl nitrite was
stored at Ϫ80ЊC in a glass vial.The purity of methyl
and ethyl nitrite was verified by FTIR, and no contam-
inants were detected.
The product study consisted of placing di-n-propyl
ether, the hydroxyl radical precursor, and nitric oxide
in the chamber, photolyzing the mixture, and moni-
toring changes in reactant and product concentrations
over time.Typical starting concentrations for the re-
actants were 26–28 ppm of di-n-propyl ether, 41–43
ppm of methyl nitrite, and 19–20 ppm of NO.The
concentrations used for ethyl nitrite experiments were
erence library.Di- n-propyl ether (Aldrich, 99 %),
propyl formate (Aldrich, 97%), propionaldehyde (Al-
drich, 97%), propyl propionate (Aldrich, 99%), and
ϩ
acetaldehyde (Aldrich 99.5 %) were obtained com-
mercially and were further vacuum purified before use.
Product molar yields per reactant concentration loss
were determined from a linear least-squares fit of the
product concentrations plotted versus the amount of
reacted di-n-propyl ether.
RESULTS
The products of the DOH-initiated photooxidation
of di-n-propyl ether were propyl formate (HC(O)
OC H ), acetaldehyde (CH CHO), propionaldehyde
3
7
3
(CH CH CHO), and propyl propionate (CH CH C(O)
3
2
3
2
OCH CH CH ).A
peak
that
exhibited
2
2
3
characteristics of a nitrate (i.e., it contained a strong
fragment at m/z ϭ 46) was also observed in the chro-
matograms.Unfortunately, though, commercial sam-
ples of the nitrate were not available for comparison;
thus, a positive identification could not be ascertained.
However, the subject of nitrate formation will be dis-
cussed subsequently.A typical post-photolysis chro-
matogram is given in Figure 2 and illustrates that most
of the products were observable by GC-MS analysis.
As stated previously, since the GC-MS technique pro-
vided for easier separation and detection of com-
pounds, it was used as the primary analytical tech-
nique.In the figure, the terms TOT, 41, 43, and 57 in
the individual frames refer to the fragment ions that
were scanned in that frame.The term TOT, or Total
1
0–11 ppm of di-n-propyl ether, 13–14 ppm of ethyl
nitrite, and 7–8 ppm of NO.Sampling consisted of
taking three time-zero samples for GC-MS analyses,
turning on the UV lamps for 30–60 s intervals, and
then taking two duplicate samples for GC-MS analy-
ses after each successive photolysis.(In the experi-
ments with ethyl nitrite, each GC-MS analysis was
followed by an FTIR analysis.) Total photolysis times
were between 4 and 5 min for each experiment.The
duplicate samples taken after each photolysis were

7
06
SEMPELES AND ANDINO
spectively.The subscripts t and 0 refer to samples
taken at time t and time zero, respectively.Rate con-
stants for the observed products are given in Table I.
The calculated correction factor ranges for propyl for-
mate, acetaldehyde, propionaldehyde, and propyl pro-
pionate were 1.00–1.03, 1.03–1.25, 1.03–1.34, and
1
.01–1.06, respectively. The raw concentrations de-
termined from experiments were multiplied by the cor-
rection factor to obtain corrected yields.
Additional experiments were performed to examine
secondary sources of the observed products.Each
commercially available product was placed in the
chamber with methyl nitrite and NO and was photo-
lyzed for times comparable to those used in the orig-
inal product study.GC-MS analyses revealed no sec-
ondary sources of propyl formate, propionaldehyde, or
propyl propionate from the compounds tested.How-
ever, acetaldehyde was produced from DOH reactions
with propionaldehyde and propyl propionate.The ac-
etaldehyde concentrations measured in the di-n-propyl
ether experiments were therefore corrected to take
these secondary sources into account.
Figure 2 Chromatogram of a chamber sample after 300 s
total of photolysis.Compounds detected include (1) di- n-
propyl ether, (2) propyl formate, (3) acetaldehyde, (4) pro-
pionaldehyde, (5) unidentified nitrate, (6) propyl propionate,
and (7) air.
Plots of the corrected product concentration data
versus the change in di-n-propyl ether concentrations
appear in Figure 3.Linear least-squares fit of the data
were used to obtain molar yields for the products.
These yields were 0.61 Ϯ 0.044 for propyl formate,
0
.60 Ϯ 0.057 for acetaldehyde, 0.15 Ϯ 0.062 for pro-
pionaldehyde, and 0.043 Ϯ 0.015 for propyl propio-
nate.Quoted errors represent Ϯ2␴.It is important to
note the curvature in the propionaldehyde and propyl
propionate data in Figure 3B.This curvature contrib-
uted to the relatively large errors in the yields for pro-
pionaldehyde and propyl propionate, and is indicative
of secondary formation of these products, as will be
discussed subsequently.
Ion Count, indicates that all of the ion fragments
within the measured mass-to-charge ratio range of 34–
1
20 were scanned.
The product concentrations were corrected for sec-
ondary hydroxyl radical reaction (reactions of prod-
ucts with hydroxyl radicals) using the correction fac-
tor, F [21]:
FTIR analysis of the ethyl nitrite/di-n-propyl ether
experiments was used solely for the detection and
quantification of formaldehyde.Formaldehyde was
quantified using an FTIR spectral range of 2769–2765
(
k Ϫ k )
(test compound)t
(test compound)0
1
2
ϫ
1 Ϫ
ͩ
ͪ
k1
F ϭ
,
1 2
k /k
(
test compound)
(test compound)
t
t
Ϫ
ͩ
ͩ ͪ ͪ
cm .Formaldehyde production from di- n-propyl
ether/DOH reactions was negligible.
Ϫ
1
(
test compound)0
(test compound)0
4)
(
The observed products account for approximately
73% of the carbon from the reacted di-n-propyl ether.
It is possible that nitrates or isomerization products
where k and k are rate constants for the test com-
pound (di-n-propyl ether) and product compound, re-
1
2
Table I Rate Constants for Product/DOH reactions
k ϫ 1012 cm molecule
3
Ϫ
1
s
Ϫ
1
Source
Products
Proply formate
Acetaldehyde
Propionaldehyde
Propyl propionate
2.38 Ϯ 0.27 @ 296
16.2 Ϯ 1.0 @ 298
22.2 Ϯ 0.9 @ 298 Ϯ 2
4.02 Ϯ 0.32 @ 296
Wallington et al.[22]
Scollard et al. [23]
Niki et al.[24]
Wallington et al. [22]

OH-INITIATED PHOTOOXIDATION OF DI-n-PROPYL ETHER
707
duce the overall rate constant.The probability of at-
tack at each carbon site is determined by dividing each
carbon’s reactivity by the overall rate constant.The
SAR method suggested the most likely site of hy-
droxyl radical attack on di-n-propyl ether was on the
secondary carbon at site III:
I
III
III
I
O
II
II
with a total probability of 88.0%. The probability of
attack on the remaining sites were 10.5% and 1.50%
for sites II and I, respectively.
Using the SAR, the most likely site of DOH attack
on di-n-propyl ether is at carbon site III.Hydrogen
abstraction at this site forms a radical (reaction 5) that
can then react with O (reaction 6) to form a peroxy
2
radical.This peroxy radical can then react with NO to
produce either an alkoxy radical (reaction 7) or 1-pro-
poxypropyl nitrate (reaction 8).
CH CH CH OCH CH CH ϩ DOH !:
3
2
2
2
2
3
CH CH CDHOCH CH CH (5)
3
2
2
2
3
CH CH CDHOCH CH CH ϩ O !:
3
2
2
2
3
2
Figure 3 (A) Plots of propyl formate (ϫ) and acetaldehyde
ٗ) concentrations versus the change in di-n-propyl ether
CH CH CH(O D)OCH CH CH (6)
3
2
2
2
2
3
(
concentration.The dotted and solid lines represents linear
least squares fit of the propyl formate and acetaldehyde data,
respectively.(B) Plots of propyl propionate ( ᭺) and propi-
onaldehyde (ϫ) concentration versus the change in di-n-pro-
pyl ether concentration.The solid lines represent linear least-
square fits to the data.The dotted lines are presented to show
the curvature in the propyl propionate and propionaldehyde
data.
CH CH CH(O D)OCH CH CH ϩ NO !:
3
2
2
2
2
3
CH CH CH(OD)OCH CH CH ϩ NO (7)
3
2
2
2
3
2
CH CH CH(O D)OCH CH CH ϩ NO !:
3
2
2
2
2
3
CH CH CH(ONO )OCH CH CH (8)
3
2
2
2
2
3
The di-propoxy radical formed in reaction 7 will either
react with O (reaction 9), decompose at the C9C
2
bond (reaction 10), decompose at the C9O bond (re-
action 11), or undergo isomerization through 1,5-hy-
drogen atom abstraction (reaction 12):
were formed in the chamber that were not observed or
could not be quantified.Losses of these products to
the chamber walls and GC/MS equipment were also
possible.However, standard samples were not avail-
able to confirm these possibilities.
CH CH CH(OD)OCH CH CH ϩ O !:
3
2
2
2
3
2
CH CH C(O)OCH CH CH ϩ HO D (9)
3
2
2
2
3
2
CH CH CH(OD)OCH CH CH !:
3
3
3
2
2
2
2
2
3
DISCUSSION
CH
3
CDH
2
ϩ HC(O)OC
3
H
7
(10)
The probabilities for hydroxyl radical attack on each
of di-n-propyl ether’s six carbon sites were determined
using the structure activity relationship (SAR) [25].
Based on experimental data, the SAR is used to predict
rate constants for organic reactions with DOH radicals
by assigning a reactivity to each carbon in a hydro-
carbon chain and then adding these reactivities to pro-
CH
CH
CH(OD)OCH
CH CH
2
CH
CHO ϩ CH
2
CH
3
!:
CH
CH
OD (11)
3
2
3
2
2
CH
CH
CH(OD)OCH
CH CH
CH
CH !:
2 3
2
CH(OH)OCH
CDHCH
(12)
3
2
2
3
The observed yield of propyl formate indicated that

7
08
SEMPELES AND ANDINO
5
Ϫ3
decomposition via the C9C bond (reaction 10) was
a major pathway.Propionaldehyde was formed from
decomposition at the C9O bond (reaction 11) and
also by the subsequent reaction of CH CH CH OD
2 ϫ 10 molecule cm [19].Acetaldehyde has a life-
time of 3.6 days with respect to DOH attack, and can
also photolyze in the troposphere to produce carbon
monoxide [9].Therefore, both propyl formate and ac-
etaldehyde could contribute significantly to photo-
chemical smog formation in the troposphere.
3
2
2
with O .Propyl propionate was formed from reaction
2
with O (reaction 9).Products of the isomerization re-
2
action were not observed.However, commercially
available standards were not available for comparison.
Acetaldehyde is formed from further reaction of the
ethyl radical produced in reaction 10:
MECHANISM PREDICTIONS
Experimental yields were compared to mechanism
predictions developed using the Carter kinetic soft-
ware [26].Rate constants were estimated using the
SAR [25] method and values from Atkinson [7].The
mechanisms considered attack on all three sets of car-
bon sites and included isomerization and nitrate reac-
tions.Secondary reactions of the primary products
were not included so as to allow for direct comparison
of the corrected product yields and the predicted data.
Decomposition enthalpies were calculated using NIST
values [27].Decomposition reaction rate values were
determined using the following equations [7,28]:
CH CDH ϩ O !: CH CH O D
(13)
3
2
2
3
2
2
CH CH O D ϩ NO ϩ M !:
3
2
2
CH CH ONO ϩ M (14)
3
2
2
CH CH O D ϩ NO !: CH CH OD ϩ NO
(15)
(16)
3
2
2
3
2
2
CH CH OD !: DCH ϩ HCHO
3
2
3
CH CH OD ϩ O !: CH CHO ϩ HO D (17)
3
2
2
3
2
Acetaldehyde could also be formed from decom-
position of the alkoxy radical produced in reaction
Ϫ(E
d
/RT)
k ϭ A ϫ e
(23)
d
d
(
11):
where
CH CH CH OD !: CH CDH ϩ HCHO (18)
3
2
2
3
2
Ed ϭ a ϩ 0.36 ϫ ⌬Hd
(24)
(25)
The CH CDH radical will then follow reactions 13–
3
2
ϭ (2.0 ϫ 1014 d) s 1
Ϫ
1
7 to eventually lead to acetaldehyde formation.Since
A
d
the corrected acetaldehyde and propyl formate yields
were nearly identical, they appeared to form mainly
along the same pathway (reaction 10).The fact that
the acetaldehyde yield was the same as the propyl for-
mate yield coupled with the insignificant yield of
HCHO suggested that reaction (17) was more likely
to occur than reaction (16).
The experimental results in this study support the
findings of previous studies on straight-chain ethers.
Previous studies with di-ethyl ether [18,19] also de-
termined that decomposition at the C9C bond was
more likely than decomposition at the C9O bond.In
the di-ethyl ether studies, ethyl formate was the major
product and formaldehyde accompanied it with the
same yield.The current study indicates that acetalde-
hyde is produced in the same yield as propyl formate.
Unlike di-ethyl ether, the presence of an extra
d is the path degeneracy for the alkoxy radical decom-
position reaction, R is the gas constant (0.00198 kcal
mol K ), T represents the temperature in Kelvin,
Ϫ
1
Ϫ1
Ϫ
1
and ⌬H is the decomposition enthalpy in kcal mol
d
Ϫ
1
[7,28].The parameter a (kcal mol ) is dependent on
the alkyl leaving-group and has the following values:
11.1 for a primary alkyl radical leaving-group
(RCDH2), 9.3 for a secondary alkyl radical leaving-
group (R
ing-group (R
decomposition pathways were calculated to be 9.04 ϫ
R
1
CDH), and 7.9 for a tertiary radical leav-
2
R
1
R
CD) [28].The rate constants for the
2
3
6
Ϫ
1
Ϫ1
10 s and 697 s for reactions (10) and (11), re-
Ϫ
15
3
spectively.Rate constants of 8 ϫ 10
cm mo-
Ϫ
1
Ϫ
1
6
Ϫ1
lecule
s
and 1.6 ϫ 10 s were used for reactions
(9) and (12), as recommended by Atkinson [7,28].The
rate constant used for di-n-propyl ether was 17.2 ϫ
Ϫ12
3
Ϫ
1
Ϫ1
9
CH 9 group in the di-n-propyl ether chain leads
10 cm molecule
form of reaction (26):
s
[7].Nitrate reactions in the
2
to the production of ethyl radicals rather than methyl
radicals from the decomposition of the di-propoxy rad-
ical.
One of the two most abundant products, propyl for-
mate, has a tropospheric lifetime of 24.3 days in the
atmosphere with respect to reaction with DOH radicals,
assuming a tropospheric DOH radical concentration of
RO
D ϩ NO !: RONO
(26)
2
2
were assigned rate constants according to the method
given by Atkinson [7].
Table II provides a comparison between predicted

OH-INITIATED PHOTOOXIDATION OF DI-n-PROPYL ETHER
709
and observed experimental product yields.Propyl for-
CH CH CH(ONO )OCDHCH CH ϩ O !:
3
2
2
2
3
2
mate and acetaldehyde were predicted as the most
abundant products.The mechanism predicted propyl
formate’s yield within experimental error; however,
acetaldehyde’s yield was over predicted by 11.7%.
The mechanism vastly underpredicted the observed
yields of propionaldehyde and propyl propionate.
Product studies of the commercially available primary
products revealed no other secondary source of pro-
pionaldehyde or propyl propionate.However, the cur-
vatures of the propionaldehyde and propyl propionate
plots in Figure 3B support the conclusion of secondary
formation of each of these products.It is possible that
these products may have been produced from either
the OH radical or photolysis reactions of primary ni-
trate products.As indicated earlier, a product that
could not be positively identified because of the lack
of a commercially available standard, but had char-
acteristics of a nitrate compound, was observed in the
experiments.This nitrate is believed to be 1-propoxy
propyl nitrate (formed in reaction 8).Model predic-
tions estimate the 1-propoxy propyl nitrate yield at
CH CH CH(ONO )OC(O D)HCH CH (28)
3
2
2
2
2
3
CH CH CH(ONO )OC(O D)HCH CH ϩ NO !:
3
2
2
2
2
3
CH CH CH(ONO )OC(OD)HCH CH ϩ NO (29)
3
2
2
2
3
2
CH CH CH(ONO )OC(OD)HCH CH !:
3
2
2
2
3
CH CH CH(ONO )OD ϩ CH CH CHO (30)
3
2
2
3
2
CH CH CH(ONO )OC(OD)HCH CH !:
3
2
2
2
3
CH CH CH(ONO )OCHO ϩ CH CH (31)
3
2
2
2
3
Pathway 2:
CH CH CH(ONO )OCH CH CH ϩ DOH !:
3
2
2
2
2
3
CH CH CH(ONO )OCDHCH CH ϩ H O (27)
3
2
2
2
3
2
CH CH CH(ONO )OCDHCH CH !:
3
2
2
2
3
CH CH CDHONO ϩ CH CH CHO (32)
3
2
2
3
2
Thus, the radical formed as a result of DOH abstrac-
tion in reaction (27) could either directly decompose
to form propionaldehyde (reaction (32)) or undergo O2
addition (reaction (28)) followed by NO reduction (re-
action (29)) and decomposition to eventually form
propionaldehyde (reaction 30).Note that the radical in
reaction (32) (CH CH CDHONO ) might also decom-
0
.18, accounting for 18% of the reacted carbon.
The estimated OH radical rate constant of reaction
with 1-propoxy propyl nitrate (using the SAR method)
Ϫ
12
3
Ϫ
1
Ϫ1
was 12.1 ϫ 10 cm molecule s .Hydroxyl rad-
ical attack at site IV of 1-propoxy propyl nitrate
3
2
2
pose to yield propionaldehyde and NO .
2
Hyperchem௡ [29] was used to investigate the for-
mation routes for propionaldehyde from DOH reaction
with 1-propoxy propyl nitrate.Unrestricted Hartree
Fock (UHF) geometry optimization calculations were
performed at the PM3 level for each of the species in
reactions (28) and (32).In addition, all species in re-
actions (30) and (31) were subjected to geometry op-
timization calculations to investigate the C9O versus
C9C decomposition of these two reactions.The
heats of formation obtained from the UHF/PM3 cal-
culations were used to determine overall enthalpy
changes for the reactions.Although these results are
based on semiempirical calculations, they do provide
basic insight as to whether or not the overall reactions
are energetically favored.The change in enthalpies of
ONO2
I
VI
O
II III
IV V
formation from the nitrate reaction with OH radicals
would then exist:
Pathway 1:
CH CH CH(ONO )OCH CH CH ϩ DOH !:
3
2
2
2
2
3
CH CH CH(ONO )OCDHCH CH ϩ H O (27)
3
2
2
2
3
2
Table II Predicated and Experimental Yields of Detected Products from Di-n-propyl Ether Photooxidation
Yields
Compound
Predicted
Experimental (Corrected)
Propyl formate
Acetaldehyde
Propyl propionate
Propionaldehyde
Formaldehyde
0.60
0.67
0.0027
1 ϫ 10
4 ϫ 10
0.61 Ϯ 0.044
0.60 Ϯ 0.057
0.043 Ϯ 0.015
0.15 Ϯ 0.062
not detected
Ϫ
4
4
Ϫ

7
10
SEMPELES AND ANDINO
reactions (28) and (32) were Ϫ16.2 kcal/mol and
Ϫ32.6 kcal/mol, respectively. Thus, reaction (32) is
favored energetically over reaction (28).In addition,
if one considers the decomposition pathways of reac-
tions (30) and (31), it is apparent that decomposition
at the C9C bond (reaction (31)) is favored over de-
composition at the C9O bond (reaction (30)).The
enthalpy changes for reactions (30) and (31) are ϩ3.62
and Ϫ7.99 kcal/mol, respectively. Thus, even if re-
action (28) was favored over reaction (32), the for-
mation of propionaldehyde via reaction (30) would
most likely be unfavored.However, based on the re-
sults, it seems that the most likely fate for alkoxy alkyl
nitrate radicals formed from OH abstraction of a hy-
drogen from the corresponding nitrate is decomposi-
mate and acetaldehyde formation proceeded by the
C9C decomposition pathway and had the highest
yields.Other products, such as propyl propionate and
propionaldehyde, were observed in much lower yields.
The lower yields of these products indicate that reac-
tion of the dipropoxy radical with O and decompo-
2
sition at the ether C9O bond, respectively, are minor
pathways.
The experimentally determined propionaldehyde
and propyl propionate yields observed in this study
provide insight into the reactions of nitrates formed
from ethers.The results presented suggest secondary
formation routes for propionaldehyde and propyl pro-
pionate through both the photolysis and DOH-initiated
photooxidation of 1-propoxypropyl nitrate.However,
additional experimental and/or computational work is
needed to fully elucidate the reactions of alkoxy alkyl
nitrates.
tion (reaction (32)) rather than O reaction (reaction
2
(28)).
It is well known that alkyl nitrates in the atmo-
sphere are capable of undergoing photolysis as well as
reaction with OH radicals [7].Thus, it is possible that
the alkoxy alkyl nitrate, 1-propoxypropyl nitrate,
could photolyze via reaction (33).The subsequent re-
actions would likely lead to the formation of second-
ary stable products, including propyl propionate via
reaction (34) and two molecules of propionaldehyde
via reaction (35):
The authors thank the National Science Foundation for sup-
porting this research (Grant ATM-9702791).
BIBLIOGRAPHY
CH CH CH(ONO )OCH CH CH ϩ h␯ !:
3
2
2
2
2
3
1
.Guidance on Estimated Motor Vehicle Emission
Reductions from the Use of Alternate Fuels and
Fuel Blends; Report No.EPA-AA-TSS-PA-87-4;
U.S.E.P.A.: Ann Arbor, MI, 1988.
CH CH CH(OD)OCH CH CH ϩ NO (33)
3
2
2
2
3
2
CH CH CH(OD)OCH CH CH ϩ O !:
3
2
2
2
3
2
HO D ϩ CH CH C(O)OCH CH CH (34)
2.Alcohols and Ethers: A Technical Assessment of their
Application as Fuels and Fuel Components; Publ.No.
2
3
2
2
2
3
4
261; American Petroleum Institute: Washington, DC,
CH CH CH(OD)OCH CH CH !:
3
2
2
2
3
July 1988.
CH CH CHO ϩ CH CH CHO (35)
3
2
3
2
3
4
5
6
. Japar, S. M.; Wallington, T. J.; Richert, J. F. O.; Ball,
J.C.Int J Chem Kinet 1990, 22, 1257–1269.
. Smith, D. F.; Kleindienst, T. E.; Hudgens, E. E.; Mc-
Iver, C.D.Int J Chem Kinet 1991, 23, 907–924.
. Smith, D. F.; Kleindienst, T. E.; Hudgens, E. E.; Mc-
Iver, C.D.Int J Chem Kinet 1992, 24, 199–215.
. Wallington, T. J.; Andino, J. M.; Potts, A. R.; Rudy,
S.J ;. Siegl, W.O.Environ Sci Technol 1993, 27, 98–
104.
Therefore, several pathways may exist for the second-
ary formation of propionaldehyde and propyl propio-
nate from 1-propoxypropyl nitrate.This result has far-
reaching implications for nitrates formed from ethers
and their potential to act as reservoirs for highly re-
active carbonyl-containing compounds.Clearly, ad-
ditional experimental work on the nitrates formed
from the reactions of ethers is warranted.
7.Atkinson, R.J.J Phys Chem Ref Data 1994, Mono-
graph 2, 1–216.
8
.Finlayson-Pitts, B. J; . Pitts, J.N,. Jr.Atmospheric
Chemistry; Wiley: New York, 1986, 405–521.
. Calvert, J.; Pitts, J. N., Jr. Photochemistry; Wiley: New
York, 1966.
CONCLUSIONS
9
The present study has determined that the major path-
way of di-n-propyl ether photooxidation is via decom-
position at the di-propoxy radical C-C bond.This has
been confirmed as the major pathway both by analogy
with previous studies of di-ethyl ether and from mech-
anism predictions performed in this study.Propyl for-
1
1
0. Atkinson, R .; Carter, W.P.L.Chem Rev 1984, 84, 437–
70.
1. Wallington, T. J.; Liu, R.; Dagaut, P.; Kurylo, M. J. Int
J Chem Kinet 1988, 20, 41–49.
4
12. Wallington, T. J.; Dagaut, P.; Liu, R.; Kurylo, M. J. Int
J Chem Kinet 1988, 20, 541–547.

OH-INITIATED PHOTOOXIDATION OF DI-n-PROPYL ETHER
711
1
1
1
1
3. Nelson, L.; Rattigan, O.; Neavyn, R.; Sidebottom, H.
Int J Chem Kinet 1990, 22, 1111–1126.
4. Bennet, P.J ;. Kerr, J.A.J Atmos Chem 1989, 8, 87–
23. Scollard, D.; Corrigan, M.; Sidebottom, H.; Treacy, J.
Kinetics and Mechanisms for the Reactions of Haloge-
nated Organic Compounds in the Troposphere.Pro-
ceedings of the STEP-HALOCSIDE/AFEAS Work-
shop; University College, Dublin, Eire, May 14–16,
1991.
9
4.
5. Semandeni, M.; Stocker, D. W.; Kerr, J. A. J Atmos
Chem 1993, 16, 79–93.
6. Wallington, T. J.; Andino, J. M.; Skewes, L. M.; Siegl,
W.O ;. Japar, S.M.Int J Chem Kinet 1989, 21, 993–
24. Niki, H.; Maker, P. D.; Savage, C. M.; Breitenbach,
L.P.J Phys Chem 1978, 82, 132–134.
1
001.
7. Bennett, P.J ;. Kerr, J.A.J Atmos Chem 1990, 10, 29–
8.
8.Wallington, T. J; . Japar, S.M.Environ Sci Technol
991, 25, 410–415.
25. Kwok, E.S.C ;. Atkinson, R.Atmos Environ 1995, 29,
1685–1695.
1
1
1
2
3
26.Carter, W.P.L.Documentation for the SAPRC At-
mospheric Photochemical Mechanism Preparation
and Emissions Processing Programs for Implementation
in Air-Shed Models; Contract No.A5-122-32; Califor-
nia Air Resources Board: Riverside, CA, October
1988.
27. Stein, S. E.; Rukkers, J.; Brown, R. NIST Standard Ref-
erence Database 25.Structures and Properties Database
Estimation Program, Version 2.0; National Institute of
Standards and Technology: Gaithersburg, MD, 1994.
28.Atkinson, R .I nt J Chem Kinet 1997, 29, 99–111.
29.Hyperchem Release 5; Hyprcube In c: .Gainesville, FL.
1
9. Eberhard, J.; Muller, C.; Stocker, D. W.; Kerr, J. A. Int
J Chem Kinet 1993, 25, 639–649.
0. Taylor, W. D.; Allston, D.; Moscato, M. J.; Fazckas,
G. D.; Kozlowski, R.; Takacs, G. A. Int J Chem Kinet
1
980, 12, 231–240.
2
2
1. Atkinson, R.; Aschmann, S. M.; Carter, W. P. L.; Winer,
A.M ;. Pitts, J.N ,. Jr.J Phys Chem 1982, 86, 4563–
4
569.
2. Wallington, T. J.; Dagant, P.; Liu, R.; Kurylo, M. J. Int
J Chem Kinet 1988, 20, 177–186.