Acetonyl
Peroxy
and Hydroperoxy
Self- and Cross-Reactions:
Temperature-Dependent
Kinetic
Parameters,
Branching
Fractions,
and Chaperone
Effects
Published
as part of The Journal
of Physical
Chemistry
virtual
special
issue “Marsha
I. Lester
Festschrift”.
Kristen Zuraski, Fred J. Grieman,
Aileen O. Hui, Julia Cowen, Frank A. F. Winiberg,
Carl J. Percival,
Mitchio Okumura,
and Stanley P. Sander
*
Cite This:
J. Phys. Chem.
A
2023,
127,
7772−7792
Read
Online
ACCESS
Metrics
& More
Article
Recommendations
*
sı
Supporting
Information
ABSTRACT:
The
temperature-dependent
kinetic
parameters,
branching
fractions,
and
chaperone
effects
of the
self-
and
cross-reactions
between
acetonyl
peroxy
(CH
3
C(O)CH
2
O
2
) and
hydro
peroxy
(HO
2
) have
been
studied
using
pulsed
laser
photolysis
coupled
with
infrared
(IR)
wavelength-modulation
spectroscopy
and
ultraviolet
absorption
(UVA)
spectroscopy.
Two
IR lasers
simultaneously
monitored
HO
2
and
hydroxyl
(OH),
while
UVA
measurements
monitored
CH
3
C(O)CH
2
O
2
.
For
the CH
3
C(O)CH
2
O
2
self-reaction
(
T
= 270
−
330
K), the rate
parameters
were
determined
to be
A
= (1.5
−
0.3
+0.4
)
×
10
−
13
and
E
a
/
R
=
−
996
±
334
K and
the branching
fraction
to the alkoxy
channel,
k
2b
/
k
2
, showed
an inverse
temperature
dependence
following
the expression,
k
2b
/
k
2
= (2.27
±
0.62)
−
[(6.35
±
2.06)
×
10
−
3
]
T
(K).
For
the
reaction
between
CH
3
C(O)CH
2
O
2
and
HO
2
(
T
= 270
−
330
K),
the
rate
parameters
were
determined
to be
A
= (3.4
−
1.5
+2.5
)
×
10
−
13
and
E
a
/
R
=
−
547
±
415
K
for the
hydroperoxide
product
channel
and
A
= (6.23
−
4.4
+15.3
)
×
10
−
17
and
E
a
/
R
=
−
3100
±
870
K for the
OH
product
channel.
The
branching
fraction
for the
OH
channel,
k
1b
/
k
1
, follows
the
temperature-
dependent
expression,
k
1b
/
k
1
= (3.27
±
0.51)
−
[(9.6
±
1.7)
×
10
−
3
]
T
(K).
Determination
of these
parameters
required
an
extensive
reaction
kinetics
model
which
included
a re-evaluation
of the temperature
dependence
of the HO
2
self-reaction
chaperone
enhancement
parameters
due
to the methanol
−
hydroperoxy
complex.
The
second-law
thermodynamic
parameters
for
K
P,M
for the
formation
of the complex
were
found
to be
Δ
r
H
250K
°
=
−
38.6
±
3.3 kJ mol
−
1
and
Δ
r
S
250K
°
=
−
110.5
±
13.2
J mol
−
1
K
−
1
, with
the
third-law
analysis
yielding
Δ
r
H
250K
°
=
−
37.5
±
0.25
kJ mol
−
1
. The
HO
2
self-reaction
rate
coefficient
was
determined
to be
k
4
=
(3.34
−
0.80
+1.04
)
×
10
−
13
exp [(507
±
76)/
T
]cm
3
molecule
−
1
s
−
1
with
the enhancement
term
k
4,M
′′
= (2.7
−
1.7
+4.7
)
×
10
−
36
exp [(4700
±
255)/
T
]cm
6
molecule
−
2
s
−
1
, proportional
to [CH
3
OH],
over
T
= 220
−
280
K. The
equivalent
chaperone
enhancement
parameter
for the
acetone-hydroperoxy
complex
was
also
required
and
determined
to be
k
4,A
′′
= (5.0
×
10
−
38
−
1.4
×
10
−
41
) exp[(7396
±
1172)/
T
]
cm
6
molecule
−
2
s
−
1
, proportional
to [CH
3
C(O)CH
3
], over
T
= 270
−
296
K. From
these
parameters,
the rate
coefficients
for the
reactions
between
HO
2
and
the respective
complexes
over
the given
temperature
ranges
can be estimated:
for HO
2
·
CH
3
OH,
k
12
=
[(1.72
±
0.050)
×
10
−
11
] exp [(314
±
7.2)/T]
cm
3
molecule
−
1
s
−
1
and
for HO
2
·
CH
3
C(O)CH
3
,
k
15
= [(7.9
±
0.72)
×
10
−
17
] exp
[(3881
±
25)/
T
]
cm
3
molecule
−
1
s
−
1
. Lastly,
an estimate
of the
rate
coefficient
for the
HO
2
·
CH
3
OH
self-reaction
was
also
determined
to be
k
13
= (1.3
±
0.45)
×
10
−
10
cm
3
molecule
−
1
s
−
1
.
1. INTRODUCTION
Atmospheric
composition
in the
troposphere
is largely
influenced
by photochemically
generated
radical
species.
These
radicals
undergo
reactions
with
volatile
organic
compounds
(VOCs)
to generate
alkyl
radicals
that,
through
an oxygen
(O
2
) addition
reaction,
form
metastable
peroxy
radicals,
RO
2
. Photolysis
of carbonyl-containing
VOCs,
alkene
ozonolysis,
and
radical
recycling
reactions
also
generate
hydroperoxy
radicals
(HO
2
) in the troposphere.
In low-NO
x
(NO
+ NO
2
) environments,
radical
loss
reactions
between
HO
2
and
RO
2
play
a vital
role
in dictating
the HO
x
(HO
2
and
OH)
and
ozone
budgets
and,
consequently,
the
oxidizing
capacity
of the troposphere,
the Earth’s
radiative
balance,
and
future
changes
in climate.
1
−
13
Acetone
(CH
3
C(O)CH
3
) is one
of the
most
abundant
oxygenated
VOCs
emitted
into
the
atmosphere
in pristine
Received:
May
31, 2023
Revised:
August
1, 2023
Published:
September
8,
2023
Article
pubs.acs.org/JPCA
© 2023
. All rights
reserved.
Published
by American
Chemical
Society
7772
https://doi.org/10.1021/acs.jpca.3c03660
J. Phys. Chem.
A
2023,
127,
7772
−
7792
This article is licensed under CC-BY-NC-ND 4.0
environments
and
leads
to the formation
of the RO
2
: acetonyl
peroxy
(CH
3
C(O)CH
2
O
2
).
14
−
17
The
cross-reactions
between
CH
3
C(O)CH
2
O
2
and
HO
2
, R1,
as well
as the
CH
3
C(O)-
CH
2
O
2
self-reaction,
R2,
have
been
studied
at room
temper-
ature.
18
−
27
R1 serves
as either
a temporary
reservoir
reaction
in
the
atmosphere
through
the
formation
of hydroperoxides
(ROOH,
R1a)
or as a radical
propagation
pathway
by
generating
hydroxyl
radicals
(OH,
R1b).
For
R2,
there
are
three
available
pathways:
R2a
generates
hydroxyacetone
(CH
3
C(O)CH
2
OH)
and
methylglyoxal
(CH
3
C(O)CHO)
as
stable
products,
R2b
generates
the
acetoxy
radical
(CH
3
C-
(O)CH
2
O),
and
R2c
generates
the
higher
functionalized
accretion
product
(ROOR,
C
6
H
10
O
4
).
18,26
CH
C(O)CH
O
HO
CH
C(O)C
H
OOH
O
(R1a
)
CH
C(O)CH
O
OH
O
(R1b)
2CH
C(O)CH
O
CH
C(O)CH
OH
CH
C(O)CHO
O
(R2a)
2CH
C(O)CH
O
O
(R2b)
C
H
O
(ROOR)
O
(R2c)
3
2
2
2
3
2
2
3
2
2
3
2
2
3
2
3
2
3
2
2
6
10
4
2
+
+
+
+
+
+
+
+
Determining
the branching
fraction
R1b/R1
is motivated
by
the need
to accurately
account
for OH
production
pathways
in
atmospheric
models
to resolve
discrepancies
with
OH
field
measurements.
29
−
32
Previous
temperature
dependence
studies
of other
RO
2
+ HO
2
reactions
were
shown
to have
reaction
rate
coefficients
and
branching
fractions
for OH
production
that
are
inversely
dependent
on temperature.
33
R2
is an
additional
loss
pathway
for RO
2
radicals.
It is less
dominant
in
the
remote
atmosphere
compared
to R1 due
to the
higher
atmospheric
concentrations
of HO
2
relative
to RO
2
;
1,2
however,
determining
the
kinetics
and
branching
fractions
for R2 is important
for laboratory
studies.
While
studying
R1
and
R2
in a laboratory
setting,
the
precursors,
methanol
(CH
3
OH)
and
CH
3
C(O)CH
3
, used
to
generate
the
starting
HO
2
and
CH
3
C(O)CH
2
O
2
radicals,
respectively,
form
the following
reactive
hydrogen-bonded
(H-
bonded)
adducts
with
HO
2
: HO
2
·
CH
3
OH
and
HO
2
·
CH
3
C-
(O)CH
3
.
18,34
−
38
In general,
the formation
of the radical
adduct
of HO
2
with
some
molecule,
X, is described
by the
rapid
equilibrium
reaction
HO
X
HO
X
2
2
+
·
F
(R3)
These
H-bonded
adducts
impact
the
observed
overall
kinetics
of HO
2
reactions
by accelerating
the
effective
HO
2
self-reaction,
R4,
via a chaperone
mechanism,
R5a
and
R5b,
which
increases
in rate
as the
temperature
is lowered.
Therefore,
this
effect
needs
to be quantitatively
considered
in
our
overall
kinetics
analysis
which
includes
R4.
HO
HO
H
O
O
2
2
2
2
2
+
+
(R4)
HO
HO
X
H
O
O
X
2
2
2
2
2
+
·
+
+
(R5a)
HO
X
HO
X
H
O
O
2X
2
2
2
2
2
·
+
·
+
+
(R5b)
R4 is an important
sink
for HO
x
in the clean
troposphere
and
is also
the
dominant
source
of stratospheric
hydrogen
peroxide,
H
2
O
2
, which
acts
as a temporary
reservoir
for HO
x
.
The
overall
rate
coefficient
of R4,
k
4
, is pressure-dependent
and
is expressed
as a sum
of two
terms
k
k
k
M
4
4,bi
4,te
r
=
+
[
]
(E1)
where
k
4,bi
is the
pressure-independent
bimolecular
rate
coefficient
and
k
4,ter
is the
pressure-dependent
termolecular
rate
coefficient.
Christensen
et al.
34
demonstrated
that
under
high
pressure
and
low
radical
concentrations,
the equilibrium
reaction
of R3 for X = CH
3
OH
is established
rapidly
on a
microsecond
timescale
as opposed
to the millisecond
timescale
of the HO
2
loss
rate
through
R1 and
R4. Their
work
revealed
that
HO
2
loss
followed
second-order
kinetic
behavior,
even
with
rate
enhancement
caused
by the H-bonded
adduct
with
CH
3
OH.
As a result,
any
unaccounted
loss
of HO
2
via R4
introduces
systematic
errors
that
propagate
into
errors
in the
determination
of the rate
coefficients
of HO
2
reactions
when
CH
3
OH
is used
as a radical
precursor.
Similarly,
when
CH
3
C(O)CH
3
is present,
the
possible
self-reaction
rate
enhancement
caused
by the
analogous
H-bonded
adduct
formed
with
CH
3
C(O)CH
3
needs
to be considered.
Since
equilibrium
concentrations
of these
adducts
increase
as
temperature
decreases,
the
rate
enhancement
effect
also
has
a temperature
dependence
that
needs
to be included
in
temperature-dependence
laboratory
studies
of HO
2
reacting
with
peroxy
radicals
using
these
precursors.
In order
to reconcile
discrepancies
in the current
literature
and
to analyze
the
data
measurements
for our
temperature
dependence
study
of R1a
and
R2a,
the equilibrium
constants
for
the
HO
2
·
CH
3
OH
and
HO
2
·
CH
3
C(O)CH
3
formation
reactions
and
the
temperature
dependence
of the
observed
kinetic
rate
coefficients
for R4 (which
includes
the enhance-
ment
caused
by
R5a)
as a function
of CH
3
OH
and
CH
3
C(O)CH
3
concentrations
were
investigated
over
the
temperature
range
T
= 220
−
296
K and
T
= 270
−
298
K,
respectively.
A reinvestigation
of the
thermodynamic
param-
eters
of the
equilibrium
reaction
with
CH
3
OH
was
accomplished
using
a van’t
Hoff
analysis
over
a wider
temperature
range
than
previously
studied.
34
Analogous
parameters
for the CH
3
C(O)CH
3
adduct
were
obtained
in a
previous
study
by Grieman
et al.
38
Using
these
results,
this
work
extends
the previous
room-temperature
determination
of
the
R1
and
R2
reaction
rate
coefficients
and
branching
fractions
18
to the temperature
range,
T
= 270
−
330
K, and
is
reported
here
for the first
time.
In all experiments,
the kinetics
of the key reactants
and
products
species
were
monitored
using
the
infrared
(IR)
kinetics
spectroscopy
(IRKS)
instrument
which
employs
simultaneous
time-resolved
near-IR
(NIR),
mid-IR
(MIR),
and
ultraviolet
absorption
spectroscopic
(UVA)
detection
of HO
2
, OH,
and
CH
3
C(O)CH
2
O
2
,
respectively.
2. METHODS
2.1. IRKS
Instrument.
The
IRKS
instrument
and
the
chemical
mechanism
used
for fitting
the data
to determine
the
temperature
dependence
for the rates,
branching
fractions,
and
chaperone
effects
of the
self-
and
cross-reactions
of CH
3
C-
(O)CH
2
O
2
and
HO
2
have
been
described
in the recent
room-
temperature
kinetics
publication
17
as well
as in previous
works.
31
−
33
The
methodologies
are
described
in brief
here
with
an emphasis
on details
that
pertain
to this
work.
All
experiments
were
initiated
inside
a temperature-controlled
(
T
= 220
−
330
K, 2
σ
= 1 K) flow
cell
(175
cm
long,
5 cm
diameter).
The
temperature
was
controlled
by flowing
a
silicone-based
fluid
(Syltherm)
through
the cell
jacket
from
a
chiller
capable
of both
cooling
and
heating
(Thermo
Neslab
ULT-95).
In all experiments,
the temperature
was
monitored
with
a calibrated
type-T
thermocouple
(Omega),
which
was
The Journal
of Physical
Chemistry
A
pubs.acs.org/JPCA
Article
https://doi.org/10.1021/acs.jpca.3c03660
J. Phys. Chem.
A
2023,
127,
7772
−
7792
7773
inserted
into
the
jacket
and
was
in contact
with
the
temperature-controlling
fluid.
2.2. Radical
Generation.
Pulsed
laser
photolysis
by a XeF
excimer
laser
(Lambda
Physik
Compex
301,
351
nm,
110
mJ/
pulse
in constant
energy
mode,
0.2 Hz repetition
rate)
was
used
to generate
the starting
radical,
Cl, from
Cl
2
(air products,
9.99%
in He,
[Cl
2
] = (0.8
−
10)
×
10
15
molecule
cm
−
3
). The
range
of total
starting
radical
concentrations,
[Cl]
0
, following
photolysis
was
(0.2
−
2.2)
×
10
14
molecule
cm
−
3
. Depending
on
the
experiment,
nitrogen
carrier
gas
was
bubbled
through
CH
3
OH
(Fisher
Optima
A454-1,
>99.9%,
bubbler
temperature
= 0
°
C)
and/or
CH
3
C(O)CH
3
(Fisher
Optima
A929-1,
>99.9%,
bubbler
temperature
=
−
25
°
C)
to entrain
these
reagent
species
in the gas phase.
Concentrations
ranging
from
[CH
3
C(O)CH
3
] = 1.7
−
2.8
×
10
16
and
[CH
3
OH]
= (1.0
−
25)
×
10
15
molecule
cm
−
3
were
determined
manometrically
based
on the measured
pressures
(absolute
capacitance
manometers,
MKS
Baratron)
and
regulated
flow
rates
(mass
flow
controllers,
MKS
Instruments).
Within
the
continuous
flow
cell,
R6
−
R9
resulted
in the
generation
of HO
2
, CH
3
C(O)-
CH
2
O
2
, or HO
2
+ CH
3
C(O)CH
2
O
2
, showing
the
self-
reactions
and
cross-reactions,
respectively.
Cl
CH
OH
CH
O
H
HCl
3
2
+
+
(R6)
CH
OH
O
HO
CH
O
2
2
2
2
+
+
(R7)
Cl
CH
C(O)C
H
CH
C(O)CH
HCl
3
3
3
2
+
+
(R8)
CH
C(O)CH
O
CH
C(O)CH
O
3
2
2
3
2
2
+
(R9)
The
gas-phase
Cl
2
, CH
3
OH,
and
CH
3
(O)CH
3
were
pre-
mixed
with
O
2
(Airgas
Corps.,
99.996%,
[O
2
] = 1.6
×
10
18
molecule
cm
−
3
) and
N
2
bath
gas (Airgas
Corps.,
99.997%)
in a
jacketed
Pyrex
manifold
and
thermally
equilibrated
to the
selected
reaction
temperature
prior
to being
introduced
to the
flow
cell.
The
flow
cell pressure
was
held
constant
at 100
Torr,
the
total
flow
rate
was
set to 2160
sccm,
and
the
flow
cell
residence
time
was
9.7
s. For
experiments
focusing
on the
cross-reaction,
secondary
chemistry
from
the CH
3
C(O)CH
2
O
2
self-reaction
was
minimized
by keeping
[HO
2
] in excess
of
[CH
3
C(O)CH
2
O
2
] using
ratios
of [HO
2
]/[CH
3
C(O)CH
2
O
2
]
= 4
−
6.
2.3. Detection
of CH
3
C(O)CH
2
O
2
, HO
2
, and OH.
UVA
and
IR-wavelength
modulation
spectroscopy
(WMS)
techni-
ques
were
used
to monitor
the time-dependent
concentrations
of CH
3
C(O)CH
2
O
2
, HO
2
, and
the
product
OH
radicals.
CH
3
C(O)CH
2
O
2
concentrations
were
monitored
using
312
nm UV
light,
where
CH
3
C(O)CH
2
O
2
has spectral
absorption
distinct
from
all of the
other
radical
species
present
in the
reactions
studied.
18
Two
independent
continuous-wave
distributed
feedback
IR lasers
(NASA
JPL
Microdevices
Laboratory)
monitored
the
concentrations
of HO
2
and
OH
via rovibrational
lines
at 6638.2
(2
ν
1
) and
3407.6
cm
−
1
(
ν
1
),
respectively.
Typical
experiments
recorded
the time-dependent
UV
kinetic
trace
and
the two
IR kinetic
traces
simultaneously
via absorbance
following
the excimer
photolysis
pulse.
All three
signals
were
digitized
and
averaged
{60
shot
averaging
for the
HO
2
self-reaction
and
≥
800
shot
averaging
for the reactions
with
CH
3
C(O)CH
2
O
2
} while
being
recorded
using
NI
LabVIEW
software.
Nitrogen-purged
aluminum
boxes
at each
end
of the
flow
tube
behind
the gas exit
ports
contained
custom-coated,
half-
moon-shaped,
Herriot
mirrors
(Rocky
Mountain
Instrument
Co.).
The
pulsed
photolysis
beam
and
continuous
collimated
broadband
UV light
from
a laser-driven
light
source
(Energetiq
EQ-99XFC)
entered
and
exited
the cell above
and
below
these
mirrors
to each
make
a single
counterpropagating
pass
through
the cell.
The
UV light
was
then
isolated
from
the excimer
beam
outside
the flow
cell using
dichroic
mirrors
and
dispersed
using
a monochromator
(Acton
Research
Corporation
Spectra
Pro-
300i,
1200
grooves/mm)
slit width
∼
160
μ
m)
coupled
to a
photomultiplier
tube
(EMI
9781A).
The
UV
absorption
path
length
was
determined
empirically
to be 148
±
10 cm long
by
measuring
Cl
2
absorption
at 320
nm
(
σ
Cl2
= 2.37
×
10
−
19
cm
2
).
39
The
two
IR lasers,
each
wavelength-modulated
at 6.8 MHz,
entered
the cell through
a hole
in one
of the mirrors
and
each
made
30
passes
through
the
cell
in a Herriot
optical
arrangement
resulting
in a total
IR effective
path
length
of
approximately
27 meters.
The
IR beams
exited
the cell through
the
same
hole
that
they
entered
and
were
detected
independently
after
being
separated
by dichroic
optics
using
an indium
gallium
arsenide
detector
(InGaAs,
New
Focus
1811)
and
a liquid
nitrogen-cooled
indium
antimonide
detector
(InSb,
IR Associates
IS-0.25)
for the near-
and
MIR
wavelengths,
respectively.
The
signals
were
demodulated
at
twice
the
modulation
frequency
(WMS,
2f-heterodyne
detection,
13.6
MHz)
and
amplified
by a factor
of 200.
The
normalized
noise-equivalent
sensitivity
concentrations
for the
detection
of HO
2
and
OH
radicals
were
on the order
of 10
8
molecule
cm
−
3
Hz
−
1/2
(10
9
molecule
cm
−
3
for
typical
experiments).
Calibration
experiments
determining
the
con-
version
factor
between
the 2f signals
and
the absolute
HO
2
and
OH
concentrations
were
conducted
daily
at room
temperature
(see
the
Supporting
Information
from
our
previous
publica-
tion
18
for further
details).
2.4. Analysis
of Experimental
Data.
The
data
used
to
determine
kinetics
parameters
and
the equilibrium
constant/
rate
enhancement
effects
from
CH
3
OH
adduct
formation
on
R4 for the
HO
2
-self-reaction
were
fit using
a Python-based
kinetics
modeling
algorithm
41
that
included
the reactions
from
Table
1. The
rate
enhancement
of R4
by radical
adducts
formed
from
the reaction
of HO
2
with
CH
3
OH
required
only
the NIR
probe
operating
under
WMS
conditions
to measure
the
loss
of species
due
to diffusion,
determine
the
starting
radical
concentrations,
and
measure
the kinetic
decay
of the
HO
2
reactant.
The
UVA
probe
{
σ
λ
(225
nm)
= 2.88
×
10
−
18
cm
2
for HO
2
absorbance,
with
correction
for H
2
O
2
product
absorbance
at 225
nm
42
} was
used
only
to calibrate
the NIR
laser
absorbance.
The
equilibrium
constant
of R3,
K
c,X
, for X =
Table
1. Chemical
Mechanism
Used
in Fitting
Kinetics
Data
for the HO
2
Self-Reaction
Data
a
reaction
number
chemical
reaction
rate
coefficient
(cm
3
molecule
−
1
s
−
1
)
R6
Cl + CH
3
OH
→
CH
2
OH
+ HCl
5.5
×
10
−
11
R7
CH
2
OH
+ O
2
→
HO
2
+ CH
2
O
9.1
×
10
−
12
R4
HO
2
+ HO
2
→
H
2
O
2
+ O
2
this
work
Cl + HO
2
→
OH
+ ClO
3.6
×
10
−
11
exp(
−
375/
T
)
→
O
2
+ HCl
1.4
×
10
−
11
exp(270/
T
)
HO
2
+ CH
2
O
→
HOCH
2
O
2
6.7
×
10
−
15
exp(
−
600/
T
)
a
Rate
coefficients
are taken
from
the recommended
values
in the JPL
Data
Evaluation
19-5;
42
R4 is the exception,
which
is from
this
work.
The Journal
of Physical
Chemistry
A
pubs.acs.org/JPCA
Article
https://doi.org/10.1021/acs.jpca.3c03660
J. Phys. Chem.
A
2023,
127,
7772
−
7792
7774
CH
3
OH
was
measured
at 100
Torr
over
the temperature
range
T
= 220
−
280
K, where
K
HO
X
HO
X
c,X
2
eq
2
eq
eq
=
[
· ]
[
]
[ ]
(E2)
and
the
kinetics
parameters
related
to R4 were
determined
over
the
range
T
= 220
−
296
K. Within
this
temperature
regime,
only
the lower
CH
3
OH
concentrations
(where
linear
regression
fits well-represented
the kinetic
data)
were
used
to
exclude
higher-order
effects
on the kinetics.
Experiments
involving
the
CH
3
C(O)CH
2
O
2
radical
were
more
complex
and
used
the
kinetics
mechanism
and
fitting
algorithm
reported
by Zuraski
et al.
18
in the room-temperature
measurement
for
this
same
chemistry
with
temperature-
dependent
rate
coefficients
and
branching
ratios
from
references
included
therein.
Temperatures
below
270
K were
not used
due
to an observed
increase
in absorption
in the UV
kinetic
trace
ascribed
to possible
aerosol
formation.
Additional
reactions
suggested
to be
important
for
this
chemical
mechanism
in the
recent
work
by Assali
and
Fittschen
28
were
tested
in a sensitivity
analysis
but were
not
found
to be
relevant
under
our
experimental
conditions
(see
the Support-
ing Information).
For
R1 and
R2, the CH
3
C(O)CH
2
O
2
, HO
2
,
and
OH
kinetic
data
were
fit simultaneously
using
a
Levenberg
−
Marquardt
algorithm
40,43,44
to optimize
the kinetic
rate
coefficients,
branching
fractions,
and
rate
enhancement
terms
for R4.
Consistent
with
the room-temperature
analysis,
the
fits
were
iterated
1000
times
per
experimental
run
following
a Markov
Chain
Monte
Carlo
(MCMC)
algorithm.
This
method
randomly
sampled
all parameters
and
systematic
uncertainties
(reaction
rate
coefficients
and
branching
fractions,
concentrations,
calibration
constants,
cell
path
length,
Poisson
counting
in the
data,
and
absorption
cross
sections)
within
each
respective
uncertainty.
From
this
method,
the
value
and
uncertainty
of each
fitted
parameter
were
determined
from
the geometric
mean
and
full width
at
half-maximum
(
σ
)
of its resulting
distribution
(uncertainties
reported
as 2
σ
unless
otherwise
stated).
Calculation
of the
geometric
mean
to constrain
the peak
value
of the Gaussian
fit
was
used
to avoid
the misrepresentation
of the median
value
of
the
MCMC
outputs
(otherwise
under-
or over-estimated
by
the arithmetic
mean)
due
to the asymmetry
of the histogram
distributions.
45
The
CH
3
C(O)CH
2
O
2
self-reaction,
R2,
data
were
analyzed
first,
where
the fitting
algorithm
determined
the
R2 reaction
rate
coefficient
and
branching
fractions.
These
results
were
then
incorporated
into
the R1 analysis,
where
the
fitting
algorithm
was
used
to determine
the R1 rate
coefficient,
branching
fractions,
and
R4
enhancement
term
for
the
CH
3
C(O)CH
3
precursor.
3. RESULTS
3.1.
CH
3
C(O)CH
2
O
2
+ CH
3
C(O)CH
2
O
2
.
Experiments
monitoring
the
CH
3
C(O)CH
2
O
2
self-reaction,
R2,
simulta-
neously
observed
the CH
3
C(O)CH
2
O
2
reactant
radicals
and
the
HO
2
and
OH
product
species
that
are
generated
from
secondary
chemistry
following
the subsequent
alkoxy
channel,
R2b.
The
CH
3
C(O)CH
2
O product
rapidly
decomposes
to
acetyl,
CH
3
CO,
which
undergoes
O
2
addition
to form
acetyl
peroxy,
OH,
and
HO
2
.
18
Figure
1 shows
the observed
kinetics
for CH
3
C(O)CH
2
O
2
for
T
= 270
−
330
K at
P
= 100
Torr.
Representative
fits
for
[CH
3
C(O)CH
2
O
2
], [HO
2
], and
[OH]
kinetics
traces
for
T
= 330
K are shown
in Figure
2.
For
each
temperature
used
over
the
range
considered,
the
resulting
distributions
from
the
MCMC
simulations
for the
rate
coefficient
k
2
and
the branching
fraction
k
2b
/
k
2
are shown
in Figure
3. The
geometric
mean
values
and
uncertainties
derived
from
the Gaussian
fits for each
temperature
are given
in Table
2. At
T
= 270,
280,
and
290
K, the branching
fraction,
k
2b
/
k
2
, showed
bimodal
distributions
that
may
indicate
that
the
fits
are
no
longer
constrained
sufficiently
to accurately
determine
the
branching
fraction
or that
the
model
is not
accurately
representing
all of the OH
and
HO
2
production
and
loss
channels.
Increasing
the
number
of iterations
in the
MCMC
calculations
did not
resolve
the observed
asymmetry
in the distributions.
The
branching
fractions
centering
on the
lower
values
for
T
= 270
−
290
K have
broader
(higher
uncertainty)
distributions
and
are inconsistent
with
the trend
observed
at the
higher
temperature
values.
Therefore,
the
distributions
with
the
higher-value
branching
fractions
are
considered
in this
work
to represent
the reaction
kinetics
and
are reported
in Table
2 (see
the Supporting
Information
for
further
discussion
and
numerical
values
for
the
other
distributions).
Figure
4 shows
the weighted
fits to the outputs
of both
the
k
2
and
k
2b
/
k
2
, which
are inversely
related
to temperature.
The
fit in the Arrhenius
plot
(Figure
4a) for the rate
coefficient,
k
2
,
represents
the
rate
parameters:
A
= (1.5
−
0.4
+0.3
)
×
10
−
13
cm
3
molecule
−
1
s
−
1
and
E
a
/
R
=
−
996
±
334
K. The
branching
fraction,
k
2b
/
k
2
, follows
the temperature-dependent
expression:
k
2b
/
k
2
= (2.27
±
0.62)
−
[(6.35
±
2.06)
×
10
−
3
]
T
(K).
3.2. CH
3
C(O)CH
2
O
2
+ HO
2
.
Figure
5 shows
representative
kinetics
for CH
3
C(O)CH
2
O
2
, HO
2
, and
the product
OH
for
R1a
at
T
= 270
−
330
K and
P
= 100
Torr.
For clarity,
only
270,
298,
and
330
K are shown
for the
peroxy
and
OH
kinetic
traces.
Figure
6 displays
the outputs
of the MCMC
fits,
and
Table
3 gives
the
values
for
the
geometric
means
of the
MCMC
outputs
for the
k
1
rate
coefficients
and
the
k
1b
/
k
1
branching
fraction
with
their
respective
uncertainties
derived
from
the Gaussian
fits of the distributions
at each
temperature.
Figure
1.
Acetonyl
peroxy
kinetics
following
the acetonyl
peroxy
self-
reaction
at
T
(K)
= 330
(red),
320
(orange),
310
(yellow),
290
(gray),
280
(cyan),
and
270
(blue).
Experimental
results
from
Zuraski
et al.
18
for the acetonyl
peroxy
self-reaction
under
similar
conditions
at
T
=
298
K are
shown
in black.
Reactant
concentrations
were
approximately
1.5
×
10
14
, 3.3
×
10
16
, and
1.6
×
10
18
molecule
cm
−
3
for [Cl]
0
, [CH
3
C(O)CH
3
]
0
, and
[O
2
]
0
, respectively.
The Journal
of Physical
Chemistry
A
pubs.acs.org/JPCA
Article
https://doi.org/10.1021/acs.jpca.3c03660
J. Phys. Chem.
A
2023,
127,
7772
−
7792
7775
The
inverse
correlations
between
the temperature
and
k
1a,
k
1b
,
and
k
1b
/
k
1
are shown
in Figure
7. The
weighted
fits in the
Arrhenius
plot
were
used
to calculate
the rate
parameters:
A
=
(3.4
−
1.5
+2.5
)
×
10
−
13
cm
3
molecule
−
1
s
−
1
and
E
a
/
R
=
−
547
±
415
K for
k
1a
and
A
= (6.23
−
4.4
+15.3
)
×
10
−
17
cm
3
molecule
−
1
s
−
1
and
E
a
/
R
=
−
3100
±
870
K for
k
1b
. The
branching
fraction,
k
1b
/
k
1
,
follows
the temperature-dependent
expression:
k
1b
/
k
1
= (3.27
±
0.51)
−
[(9.6
±
1.7)
×
10
−
3
]
T
(K).
3.3. HO
2
Self-Reaction
Rate Enhancement
by Radical
Adducts.
3.3.1. Temperature
Dependence
of the Equili-
brium
Constant
for HO
2
+ CH
3
OH
⇌
HO
2
·
CH
3
OH.
Following
photolysis,
the
initial
Cl atoms,
[Cl]
0
, were
completely
converted
to HO
2
. The
total
initial
HO
2
concentration,
[HO
2
]
0
, is expressed
as
HO
HO
HO
CH
O
H
Cl
2
0
2
eq
2
3
eq
0
[
]
=
[
]
+
[
·
]
=
[
]
(E3)
where
[HO
2
]
eq
and
[HO
2
·
CH
3
OH]
eq
are the concentrations
of
the
remaining
HO
2
and
the
complex
HO
2
·
CH
3
OH,
respectively,
following
the rapid
equilibrium
reaction
HO
CH
OH
HO
C
H
OH
2
3
2
3
+
·
F
(R10)
The
equilibrium
constants
for R10
were
determined
at each
temperature
by measuring
the loss
of HO
2
in the first
∼
20
−
50
μ
s after
photolysis
using
the
NIR
probe.
This
approach
assumes
that
the timescale
for reaching
equilibrium
is much
shorter
than
that
for HO
2
loss
by R4.
This
assumption
was
validated
by the observed
drop
in the peak
HO
2
signal
with
increasing
[CH
3
OH]
0
(CH
3
OH
is in excess)
at early
times
(
t
<
50
μ
s).
Because
the NIR
probe
only
measures
the amount
of
non-complexed
HO
2
, the
observed
difference
in the
peak
[HO
2
] signal
with
various
concentrations
of CH
3
OH
provided
an indirect
measurement
of the equilibrium
concentration
of
the complex
formed.
Figure
2.
(a) Acetonyl
peroxy,
(b) HO
2
, and
(c) OH
observed
(red)
kinetics
following
the acetonyl
peroxy
self-reaction
at
T
= 330
K and
P
= 100
Torr.
Simulations
from
the fit results
are shown
in black.
The
fits for the acetonylperoxy
kinetics
follow
the primary
reactant
for the R2a
reaction.
OH
and
HO
2
fits are the results
of secondary
chemistry.
The Journal
of Physical
Chemistry
A
pubs.acs.org/JPCA
Article
https://doi.org/10.1021/acs.jpca.3c03660
J. Phys. Chem.
A
2023,
127,
7772
−
7792
7776