The kinetics of the gas-phase decomposition of the 2-methyl-2-butoxyl and 2-methyl-2-pentoxyl radicals

Article abstract of DOI:10.1039/b501416f

The kinetics of the title reactions have been studied by relative-rate methods as a function of temperature. Relative-rate coefficients for the two decomposition channels of 2-methyl-2-butoxyl have been measured at five different temperatures between 283 and 345 K and the observed temperature dependence is consistent with the results of some previous experimental studies. The kinetics of the two decomposition channels of 2-methyl-2-pentoxyl have also been investigated, as a function of temperature, relative to the estimated rate of isomerisation of this radical. Room-temperature rate coefficient data for the two decomposition channels of both 2-methyl-2-pentoxyl and 2-methyl-2-butxoyl (after combining the relative rate coefficient for this latter with a value for the rate coefficient of the major channel, extrapolated from the data presented by Batt et al., Int. J. Chem. Kinet., 1978, 10, 931) are shown to be consistent with a non-linear kinetic correlation, for alkoxyl radical decomposition rate data, previously presented by this laboratory (Johnson et al., Atmos. Environ., 2004, 38, 1755-1765). The Owner Societies 2005.

Full text of DOI:10.1039/b501416f

View Article Online / Journal Homepage / Table of Contents for this issue
R E S E A R C H P A P E R
The kinetics of the gas-phase decomposition of the
-methyl-2-butoxyl and 2-methyl-2-pentoxyl radicals
2
David Johnson,*w Scott Carr and R. Anthony Cox
University Chemical Laboratory, Lensfield Road, Cambridge, UK CB2 1EW.
E-mail: d.johnson@imperial.ac.uk; Fax: þ44 (0) 207 594 2308
Received 27th January 2005, Accepted 31st March 2005
First published as an Advance Article on the web 15th April 2005
The kinetics of the title reactions have been studied by relative-rate methods as a function of temperature.
Relative-rate coefficients for the two decomposition channels of 2-methyl-2-butoxyl have been measured at five
different temperatures between 283 and 345 K and the observed temperature dependence is consistent with the
results of some previous experimental studies. The kinetics of the two decomposition channels of 2-methyl-2-
pentoxyl have also been investigated, as a function of temperature, relative to the estimated rate of isomerisation
of this radical. Room-temperature rate coefficient data for the two decomposition channels of both 2-methyl-2-
pentoxyl and 2-methyl-2-butxoyl (after combining the relative rate coefficient for this latter with a value for the
rate coefficient of the major channel, extrapolated from the data presented by Batt et al., Int. J. Chem. Kinet.,
1978, 10, 931) are shown to be consistent with a non-linear kinetic correlation, for alkoxyl radical decomposition
rate data, previously presented by this laboratory (Johnson et al., Atmos. Environ., 2004, 38, 1755–1765).
(
be temperature dependent.
i)) and hence the relative-rates of these reactions are expected to
6
Introduction
–10
Relative-rate coefficients have
Products of the atmospheric oxidation of volatile organic com-
pounds (VOCs)—which are released into the atmosphere from
biogenic and anthropogenic sources—affect not only the chem-
istry at the Earth’s surface (e.g. see refs. 1 and 2), but also it has
been reported for the unimolecular reactions (i.e. decomposition
and isomerisation) of a number of simple alkoxyl radicals but in
the majority of cases the data were obtained from measurements
made at room temperature (e.g. see refs. 10–13) or, in some
cases, at temperatures not representative of the Earth’s atmo-
sphere (e.g. see refs. 14–16). A number of absolute kinetic studies
3
recently been shown, that occurring in the upper troposphere.
For example, the photooxidant ozone constitutes a secondary
pollutant when it is produced following the interaction of
organic radicals, derived from VOC pollutants, with oxides of
ꢀ
have also been made of the decomposition of selected RO and
17–27
of their bimolecular reaction with O
2
.
Recent atmospheric
1
,2,4
nitrogen, NO
x
(NO
x
¼ NO þ NO
2
).
At ground level, ozone
formation contributes to urban ‘‘smog’’ and reduced air quality
measurements have revealed higher than expected concentra-
tions of partially oxidised hydrocarbons in the free and upper
4
in general—for example, ozone is a respiratory irritant. In the
upper troposphere/lower stratosphere (UTLS) region of the
3
troposphere which is suggestive of the occurrence of VOC
oxidation at higher altitudes where temperatures and pressures
are lower than at the Earth’s surface. It has thus become clear
that it is important to understand the relative-rates of processes
5
atmosphere ozone is an important greenhouse gas. The general
mechanisms of the atmospheric degradation of VOCs are fairly
6
well established. In the majority of cases, the oxidative degra-
(
i)–(iii) as a function of temperature—in a range pertinent to the
dation is initiated by reaction with the hydroxyl radical, OH,
and proceeds to the production of a first generation carbonyl
Earth’s atmosphere—for a variety of alkoxyl radicals, derived
not only from small but also larger VOCs, to assess the atmo-
spheric impact of the photooxidation of VOCs throughout the
troposphere. Further to this, the importance of identifying
patterns and trends in the reactivity of alkoxyl radicals of
various size and structure is becoming clear.
ꢀ
product through a chain of reactions mediated by alkyl (R ),
ꢀ
ꢀ
peroxyl (RO
urban atmospheric conditions (with 20% O
levels of NO
largely converted to RO radicals.
2
) and alkoxyl (RO ) radicals. Under polluted
2
and significant
ꢀ
x
from, primarily, combustion sources) R are
ꢀ
1,2,6
The atmospheric impact
Recent work in this laboratory was aimed at determining the
temperature-dependence of the relative-rates (relative to bimo-
of the oxidation chemistry of a given VOC is determined by the
nature of the first generation products and this is determined by
2
lecular reaction with O ) of isomerisation reactions of various
ꢀ
ꢀ
the chemistry of RO . In the atmosphere, RO exhibits three
6
ꢀ
28
model RO species. These relative-rate data were then used to
estimate absolute rate parameters and the effect of the structure
–9
2
modes of reaction: (i) bimolecular reaction with O (yielding a
carbonyl product, of the same carbon chain length as the parent
ꢀ
ꢀ
of RO upon Arrhenius activation barriers and A factors was
VOC, and a hydroperoxy radical, HO
(
2
); (ii) decomposition
yielding carbonyl and alkyl radical fragments, of smaller carbon
investigated. Derived Arrhenius activation barriers were shown
to correlate linearly with the strength of the C–H bond cleaved
in the isomerisation process and the A factors were found to
scale linearly with the number of abstractable H atoms in the
g-position to the alkoxyl functionality. The kinetics of the
decomposition of the 2-butoxyl radical have also been inves-
tigated in this laboratory using the same experimental techni-
chain length than the parent VOC) and (iii) isomerisation (a 1,5
H shift occurring through a six-membered cyclic transition
state). The energetic demands, and hence temperature-depen-
dencies, of these three processes are different (generally, the
activation barriers for the processes are in the order (ii) 4 (iii) 4
2
9
ꢀ
que.
The effect of the structure of RO upon the
decomposition kinetics of various alkoxyl radicals was also
investigated and a non-linear correlation between the loga-
rithm of the unimolecular rate coefficient and the average
w Present address: Department of Environmental Science and Tech-
nology, Imperial College London, The Manor House, Silwood Park,
Ascot, Berkshire, UK SL5 7PY.
2
182
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 2 1 8 2 – 2 1 9 0
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5

View Article Online
2
98K 298K
/k
ionisation potential (IPmean) of the product fragments was
respectively; giving k
1
2
¼ 85 and 123, respectively).
3
0
found for data obtained at 298 K. Correlations between
measured, estimated and theoretically calculated activation
barriers were also presented. During the course of this latter
investigation it was found that the kinetics of the decomposi-
tion of the 2-methyl-2-butoxyl (t-amyloxyl) radical to buta-
none and methyl radical products were anomalous in terms of
the identified correlation. 2-Methyl-2-butoxyl can decompose
by more than one pathway:
1 2
Other experimental determinations of k /k have been made by
measuring the relative yields of acetone and butanone.
3
5–37
In
these studies, which are summarised in Table 1, 2-methyl-2-
butoxyl radicals were produced from both the thermolysis and
photolysis of various precursor compounds (alkyl nitrites and
dialkyl peroxides) under a variety of different conditions of
temperature, reactant concentration and in the presence and
absence of added NO. What is clear from the data presented in
Table 1 is that there is no clear consensus between the various
determinations of the ratio of decomposition rate coefficients
for 2-methyl-2-butoxyl. Similarly, the theoretically determined
rate coefficients for the major decomposition process (R1) are
significantly greater than the experimental determination of
ꢀ
ꢀ
CH
3
CH
2
C(O )(CH
3
)
2
- CH
3
C(O)CH
3
þ C
2
H
5
(R1)
(R2)
ꢀ
þ CH ꢀ
CH CH
3
2
C(O )(CH
3
)
2
- CH C(O)CH
3
2
CH
3
3
The room temperature rate coefficient data for (R1) and (R2)
3
included in the correlation plot were those presented by
0
31
Batt et al. and yet this latter value is consistent with the
7
Atkinson. These were values extrapolated (to 298 K) from
kinetic correlation identified previously in this laboratory.
The first aim of the present investigation, therefore, was to
reassess the relative kinetics of the decomposition channels of
2-methyl-2-butoxyl using the same methodology employed in
this laboratory to study the isomerisation of simple alkoxyl
9
radicals and the chemistry of the 1- and 2-butoxyl radicals.
This was achieved by measuring the relative-yields of acetone
and butanone from the steady-state, near-UV photolysis of
dilute mixtures of 2-methyl-2-butyl nitrite, as a function of
temperature, in a flow-reactor. The second aim of the investi-
gation was to attempt to place these relative-data onto a more
absolute basis by studying the chemistry of the 2-methyl-2-
pentoxyl radical. As discussed later, the decomposition kinetics
(R5,6) of this radical are expected to be similar to those of
2-methyl-2-butoxyl. In addition to decomposing, 2-methyl-
2-pentoxyl is able to isomerise (R7).
the higher temperature measurements (120–190 1C; 393–463 K)
3
1
of Batt et al. —rate coefficients, at four different temperatures
between 433 and 463 K, were determined relative to the
reaction of 2-methyl-2-butoxyl with NO.
2
8
2
ꢀ
CH CH C(O )(CH ) þNO-CH CH C(ONO)(CH ) (R3)
3
2
3 2
3
2
3 2
The source of alkoxyl radicals in this study was the thermal
decomposition of 2-methyl-2-butylnitrite:
ꢀ
CH
3
CH
2
C(ONO)(CH
3
)
2
-CH
3
CH
2
C(O )(CH
3
)
2
þNO (R4)
The rate of reaction (R1) was found to dominate over that of
R2) with a rate coefficient ratio (k /k ) of 80 at 433 K, and an
(
Arrhenius expression for (R1) was reported: k
1
2
1
4
1
¼ 5.0 ꢁ 10
ꢂ1
31
exp(ꢂ59.8 kJ mol /RT). These data were extrapolated to
7
98 K following the method described by Atkinson. A value
2
for k was calculated using the above Arrhenius expression of
ꢀ
1
CH CH CH C(O )(CH ) - CH C(O)CH
3
2
2
3 2
3
3
3
1
Batt et al. for T ¼ 448.2 K (the mid-point of the temperature
ꢀ
þ CH
3
CH
2
CH
2
(R5)
range employed in their experimental study) and then corrected
3
2
ꢀ
ꢀ
3
CH CH CH C(O )(CH ) -CH CH CH C(O)CH þCH
to reflect the recommendation of Atkinson for rate coeffi-
cients for the combination reactions of alkoxyl radicals with
NO (this was the reference reaction in the study). An Arrhenius
3
2
2
3 2
3
2
2
3
(R6)
ꢀ
ꢀ
CH
3
CH
2
CH
2
C(O )(CH
3
)
2
- CH
2
CH
2
CH
2
C(OH)(CH
3 2
)
1
3
ꢂ1
A factor of 1 ꢁ 10
s was then assumed for reaction (R1)—A
factors close to this value have been obtained in a number of
recent relative- and direct-kinetic studies of the decomposition
(
R7)
The rate of the isomerisation process (R7) (involving the
cleavage of primary C–H bond) can be estimated, as a function
of temperature, using the structure activity relationship re-
3
of alkoxyl radicals, as discussed by Johnson et al. —and an
0
activation barrier for (R1) was calculated to match the value of
4
48.2K
28
k
late k1
1
. This new Arrhenius expression was then used to calcu-
98K
cently developed in this laboratory and it is proposed that
2
4
ꢂ1
¼ 3.4 ꢁ 10 s . A similar extrapolation employing
this may be used as an estimated kinetic reference for the
decomposition reactions. Neither 2-methyl-2-butoxyl nor 2-
methyl-2-pentoxyl possesses H atoms in a position a- to the
4
33K 433K
298K
yields k
2
ꢂ1
the Batt et al. determination of k
1
/k
2
¼ 36 s
2
98K 298K
and hence k1
values for k
pared with the recent theoretical calculations of Me
/k2
E 950. These experimentally derived
(and, of course, k /k ) can be com-
reau
1
and k
2
1
2
alkoxyl functionality and hence reaction of the radicals with O
2
(which can yield useful diagnostic carbonyl products with fairly
´
3
3
34
298K
1
5
7–9
well established kinetics) does not occur and hence cannot
be used as a kinetic reference.
et al. and Peeters et al. (with values of k
6
¼ 9.4 ꢁ 10 and
ꢂ1
298K
2
4
4
ꢂ1
s ,
1
.6 ꢁ 10
s
; and k
¼ 1.1 ꢁ 10 and 1.3 ꢁ 10
1 2
Table 1 Summary of previous experimental determinations of k /k
ꢀ
RO Precursor
Experiment type
Conditions
T/K
k
1
/k
2
Ref.
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
a
3
3
3
3
3
3
3
3
3
3
3
3
3
–O–O–C(CH
3
)
2
CH
2
2
2
2
2
2
2
2
2
2
2
2
2
CH
CH
CH
3
2
2
Thermolysis
Thermolysis
Thermolysis
Thermolysis
Thermolysis
Thermolysis
Thermolysis
Photolysis
Peroxide : N molar ratio 2 : 3
20 Torr peroxide
20 Torr peroxide
2
468
387
8.4
16
35
36
36
37
37
37
31
36
36
36
37
37
37
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
2
2
2
2
2
2
2
2
2
2
2
2
–O–O–C(CH
–O–O–C(CH
C(ONO)(CH
C(ONO)(CH
–O–O–C(CH
C(ONO)(CH
–O–O–C(CH
–O–O–C(CH
–O–O–C(CH
C(ONO)(CH
C(ONO)(CH
C(ONO)(CH
3
)
3
)
3
)
3
)
3
)
3
)
3
)
3
)
3
)
3
)
3
)
3
)
CH
CH
3
3
410
15
26 Torr nitrite
26 Torr nitrite, 14 Torr NO
430–456
427–455
387
160
104
16
a
CH
2
CH
3
—
2.3 Torr nitrite
433
299
80
16
CH
CH
CH
2
2
2
CH
CH
CH
3
3
3
20 Torr peroxide, l ¼ 313 nm
Photolysis
Photolysis
20 Torr peroxide, l ¼ 313 nm, 15 Torr NO
20 Torr peroxide, l ¼ 313 nm, 15 Torr NO
26 Torr nitrite, 26–105 Torr NO, l ¼ 366 nm
26 Torr intrite, 20–85 Torr NO, l ¼ 313 nm
26 Torr nitrite, 0–85 Torr NO, l ¼ 253.7 nm
299
346
10
8.3
Photolysis
Photolysis
Photolysis
387
387
33–42
11.3
6.8
387
Details not given.
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 2 1 8 2 – 2 1 9 0
2183

View Article Online
Fig. 1 Schematic diagram of the steady-state, slow-flow, photochemical apparatus.
outlet of the flow-cell was connected to the first mass flow
00
Experimental
3
controller (calibration range 0–500 cm min ) with 1/4
ꢂ1
Experiments were performed in steady-state using a slow-flow
photochemical reactor and dedicated gas chromatograph with
flame ionisation detection (GC-FID). A schematic diagram of
the experimental apparatus is shown in Fig. 1.
Teflon tubing and the outlet of the mass flow controller was
connected to a glass, gas-handling line, again with Teflon
tubing, which in turn was connected to the rotary pump. The
total flow-rate through the cell was controlled between 125 and
The flow-reactor comprised a double-jacketed, 1 m long,
3
3
ꢂ1
3
25 cm min giving a residence time in the flow-cell of
cylindrical quartz cell with an internal volume of 220 ꢃ 10 cm .
between 108 and 42 s. This resulted in fractional photolytic
consumption of the RONO precursors of between 75 and 25%.
Also connected between the outlet of the flow-cell and the gas-
handling line (and hence the rotary pump) was a small
sampling loop, which consisted of a broken short length of
The internal surfaces of the cell were coated with halocarbon
wax to reduce their reactivity. The temperature of the cell was
controlled, between 283 and 345 K, by flowing a thermostatted
water/ethylene glycol mixture (equal concentrations by vol-
ume) from a circulating bath (Neslab RTE-140) through the
middle jacket. The outer jacket was evacuated to provide
thermal insulation. Surrounding the cell were four, metre-long,
fluorescent black-lamps (F40-BLB-EX) which provided con-
tinuous near-UV radiation in the range 350 to 400 nm, with
0
0
0
0
1
/4 Teflon tubing with 1/4 PFA compression fittings, and a
by-pass loop, both of which were connected up-line of the
second calibrated mass flow controller (see Fig. 1). The PFA
00
compression fittings allowed an adsorbant tube (1/4 outside
diameter, stainless steel tube containing Tenax TA) to be
connected in-line with the photochemical flow-cell and the
l
max E 365 nm. The lamps were mounted equidistantly around
and parallel to the flow-cell, and the lamps and cell were
housed in an aluminium box. The alkoxyl radicals studied
were produced directly from the photodissociation of alkyl
nitrite (RONO) precursors using the radiation from the black-
lamps. Pure samples of RONO were synthesised from the
parent alcohol (ROH) by reaction with nitrous acid (HONO),
3
second mass flow controller (calibration range 0–100 cm
ꢂ1
3
ꢂ1
min ). A controlled flow, of 30 cm min , from the outlet
of the photochemical cell was maintained through the second
flow controller throughout each experiment and this came
either via the sample loop (with adsorbant tube in place) or
the by-pass loop. The ‘‘route’’ of this 30 cm min flow was
determined using three PFA valves (again, see Fig. 1). Gas-flow
through the sample loop (with adsorbant tube in place) was
3
ꢂ1
3
8
at 0 1C, and stored under vacuum at 4 1C. The purity of each
RONO sample was assessed by GC-FID and proton nuclear
magnetic resonance spectroscopy, and was found to be Z 95%
in all cases. The major impurity in all samples was unreacted
ROH. Measured UV/visible absorption spectra of 2-methyl-2-
butyl nitrite and 2-methyl-2-pentyl nitrite are shown in Fig. 2.
These spectra are typical of those for simple alkyl nitrites
and compare well, in terms of general shape and maximal
absorption cross-section, with spectra reported in the literature
3
used to immobilise, and hence concentrate, 150 cm gas
samples after passage through the photochemical flow-cell. A
3
9,40
2 4
for other simple C –C alkyl nitrites.
A mixture of the RONO precursor (100–140 ppmv) and
sufficient cyclohexane (2500 ppmv) to scavenge 495% of any
OH formed during the experiments (from the reaction of any
HO2 formed with NO) was prepared in N2 (Air Products,
grade 5.0) and O
collapsible Tedlar bag (Andersen Instruments, Nutech 219).
The diluent gas mixture was 80% N and 20% O by volume.
2
(Air Products, grade 5.0) in a 300 L
2
2
During the photochemical experiments, the contents of the
Tedlar reactant reservoir were drawn first through a jacketed
glass mixing-manifold (which was thermostatted using the
same circulating water/ethylene glycol mixture as the flow-cell)
before passing through the flow-cell. The flow of gases was
achieved using a rotary vacuum pump and was regulated using
two calibrated mass flow controllers (Tylan) which were con-
nected before the pump (see Fig. 1). Connections between the
reactant reservoir, the mixing manifold and the inlet of the
0
0
flow-cell were made with 1/4 Teflon tubing which was ex-
ternally, thermally insulated with polyurethane foam. The
Fig. 2 UV/visible absorption spectra of 2-methyl-2-butyl nitrite (solid
line) and 2-methyl-2-pentyl nitrite (dotted line).
2
184
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 2 1 8 2 – 2 1 9 0
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5

View Article Online
clean adsorbant tube was connected in-line with the sample
loop, the two sample-loop PFA valves were opened and the by-
pass loop PFA valve was closed for 5 min. After this time, the
sample-loop was isolated, the by-pass PFA valve was reopened
and the adsorbant tube was removed and placed into a thermal
desorber (Markes International Ltd UNITY thermal desorber)
which was connected to the GC (Agilent 6890N). The immo-
bilised hydrocarbons were then flash desorbed and flushed into
the capillary column (50 m, CP-Sil 5CB) of the GC for
subsequent separation and detection. An example temperature
programme, for experiments involving 2-methyl-pentyl nitrite,
started with the column at a temperature of 10 1C for 5 min
Table 2 Measured relative-yields of acetone and butanone from the
near-UV photolysis of 2-methyl-2-butyl nitrite
T/K
[2-Methyl-2-
butyl nitrite]
ppmv
No. of
determinations
D[acetone]/
D[butanone]
(k /k )
1 2
a
0
/
2
2
3
3
83.2
20.6
9.6
4
4
5
5
5
71.0 (ꢃ7.9)
48.7 (ꢃ6.9)
98.2
08.2
23.2
12.9
11.8
11.2
42.8 (ꢃ10.0)
35.0 (ꢃ6.9)
30.7 (ꢃ7.0)
338.2
a
ꢂ1
ꢃ2s from replicate determinations.
before being raised to 110 1C at a rate of 25 1C min . A carrier
gas pressure of 18 psi was used for these experiments which
3
ꢂ1
gave a carrier gas flow-rate of 2.8 cm min through the
capillary column.
ꢀ
GC peak areas were converted to volume mixing ratios after
calibration with pure standards. Standard mixtures were pre-
pared by flushing a known pressure of pure vapour in a
calibrated glass volume into an 80 L, collapsible Tedlart bag
with a known volume of GC grade nitrogen. The standard
mixture was then drawn through the mixing manifold and
flow-cell, and subsequently sampled and analysed in the same
way as described above for photolysis experiment samples.
Five or six separate mixtures for each reactant and product of
interest, of varying volume mixing ratio, were prepared and
analysed in this way to construct calibration curves. The linear
regression statistics of these calibration curves gave an indica-
tion of the repeatability of the analyses, 5% (2s). The total
uncertainty in measured mixing ratios, accounting for systema-
tic uncertainties in the preparation of calibration mixtures (i.e.
uncertainty in pressure measurements and, in the case of alkyl
nitrites, the purity of standards), was of the order of 7–10%. In
this way, yields of diagnostic carbonyl reaction products were
determined relative to the amount of alkyl nitrite precursor
consumed. All experiments were carried out at atmospheric
pressure (760 ꢃ 10 Torr). All condensed-phase reagents under-
went two freeze–pump–thaw cycles before being used. UV/
visible absorption spectra of the synthesised alkyl nitrites were
measured and recorded, with a spectral resolution of 0.5 nm,
using a spectrograph (Acton Research Corporation), with
broadband CCD detector, connected to a personal computer.
The pyrex-filtered radiation from a deuterium discharge lamp
was passed longitudinally through a second quartz cell (60 cm
length) before being directed, with a fibre optic, into the
entrance slit of the spectrograph. The absorption cell was
connected to the glass gas-handling line and this allowed
absorption spectra to be measured as a function of the partial
pressure of the alkyl nitrite in the absorption cell, in the range
tion of ‘‘hot’’ RO produced from the 360 nm photolysis of
alkyl nitrites is of the order of 5–10%.
The data in Table 2 are plotted in Fig. 3a in Arrhenius
form—i.e. the natural logarithm of the relative rate coefficients
vs. the reciprocal of the absolute temperature—although there
does appear to be some curvature in these data. If the rate
coefficients for reactions (R1) and (R2) are expressed by
Arrhenius expressions—i.e. k
1
¼ A
1
exp(ꢂE
1
/RT) and k
2
¼
A exp (ꢂE /RT)—then the intercept of the plot corresponds
2
2
to the logarithm of the ratio of the A factors of the two
decomposition pathways (E1).
ln(k /k ) ¼ ln(A /A ) þ (E —E )/RT
(E1)
1
2
1
2
2
1
ꢀ
It is expected that Arrhenius A factors for a single RO
decomposition channel are similar for any given alkoxyl
radical—this is to say that the electronic rearrangements com-
mensurate with a given decomposition processes (i.e. a b C–C
scission giving rise to carbonyl and alkyl radical fragments) are
similar. For the decomposition of 2-methyl-2-butoxyl, how-
ever, via reaction (R2) there are two possible carbon atoms that
can be ejected as methyl radicals and hence, statistically, we
expect A to be twice the value of A and thus the intercept of
Fig. 3a to be ln(1/2) (ꢂ0.69, two decimal places). From a least
squares linear regression analysis of the data in Fig. 3a,
weighted by the uncertainties in the data, the intercept is
ꢂ1.16 ꢃ 2.24 (uncertainty ¼ 2 ꢁ standard error) and thus
the relative-rate data obtained in the present investigation are
consistent with the expected relative degeneracies of pathways
(R1) and (R2). That said, however, the data are also consistent
with the values of A and A being equal and with the value of
2
1
1
2
A being twice that of A
the data is that the values of A and A are similar and that this
1
2
—all that can really be inferred from
1
2
0.5–3 Torr. The absorption cross-sections indicated in Fig. 2
were determined in this way.
is consistent with the measured carbonyls being produced by
similar decomposition mechanisms.
Plotted in Fig. 3b are the various relative-rate coefficient
determinations listed in Table 1. It is interesting to divide these
additional data into four groups depending on the type of
Results and discussion
ꢀ
precursor used and the method of RO production from that
2
-Methyl-2-butoxyl
precursor, viz.: thermolysis of organic peroxide, thermolysis of
organic nitrite, photolysis of organic peroxide and photolysis
of organic nitrite. It can be seen in Fig. 3b that the relative-rate
determinations from peroxide thermolysis studies are consis-
tent with the results obtained in the present investigation and
that the apparent curvature in Fig. 3a does not appear to be as
significant over the extended temperature range of the com-
bined two datasets. Relative-rate coefficients from peroxide
photolysis studies lie, unsurprisingly, below the solid line in the
figure and this can be explained in terms of the effects of
chemical activation in the alkoxyl radicals resulting from
the 313 nm photolysis of organic peroxide precursors. Rela-
tively more butanone (and methyl) will result from the
decomposition of ‘‘hot’’ 2-methyl-2-butoxyl radicals than from
the decomposition of thermalised radicals and thus the value of
The results of experiments using the photolysis of 2-methyl-2-
butyl nitrite as a source of 2-methyl-2-butoxyl radicals are
summarised in Table 2. The ratios of the measured yields of
acetone and butanone are assumed to be equal to the relative
rate coefficient k /k . Implicit with this analysis is the assump-
1
2
tion that no significant proportion of activated, or ‘‘hot’’,
alkoxyl radicals is produced from the near-UV photolysis of
the alkyl nitrite precursor. This assumption is consistent with
the results of other experimental studies made in this
4
0
laboratoryz and the conclusion of Heicklen that the propor-
z Unpublished experimental results of studies of the near-UV photo-
lysis of 1- and 2-butyl nitrite are consistent with a production of ca.
10% ‘‘hot’’ alkoxyl radicals in both systems.
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 2 1 8 2 – 2 1 9 0
2185

View Article Online
tion of added NO was also investigated by these authors. An
initial rapid decrease of the quantum yield sum with the
amount of added NO (up to ca. 20 Torr NO) was attributed
to the scavenging of thermalised alkoxyl radicals by NO and a
subsequent slower decrease (35 Torr o NO o 100 Torr) was
ꢀ
attributed either to the deactivation of excited RO by NO or
ꢀ
some other reaction of excited RO with NO giving products
other than ketones. These conclusions are not, however, con-
sistent with the results of other studies in which alkoxyl
radicals were produced from the near-UV photolysis of alkyl
nitrites where evidence for very much smaller proportions of
ꢀ
28,29,40
excited RO were obtained.
The effect of added NO upon the ratio of the yields acetone
and butanone in thermolysis and photolysis experiments was
3
also investigated by Durant and McMillan and in both
7
experiment types D[acetone]/D[butanone] decreased with added
3
1
NO. In the experiments of Batt et al. around 1 atm of NO
was added to around 2 Torr 2-methyl-2-butyl nitrite and no
butanone could be observed at all. One of the major differences
between the experiments carried out in the present investigation
and the alkyl nitrite thermolysis and photolysis experiments
indicated in Table 1 are the starting concentrations of the
radical precursor. In the present investigation, mixing
ratios of around 10 ppmv 2-methyl-2-butyl nitrite were em-
ꢂ2
ployed which equate to partial pressures of around 1 ꢁ 10
Torr. This compares with pressures of 2.3 and 26 Torr of alkyl
nitrite employed in the studies in Table 1. It may be possible
that, under the conditions of these latter experiments, a large
proportion of the ethyl (and methyl) radicals formed from
the decomposition of alkoxyl radicals can react with the alkyl
nitrite.
ꢀ
C H þ CH CH C(ONO)(CH ) - Products
(R8)
2
5
3
2
3 2
If this reaction promotes the formation of acetone over that
of butanone then this will result in an ‘‘artificially’’ high value
of k /k being determined from measurements of D[acetone]/
1 2
D[butanone]. The effect of adding NO to such systems can now,
qualitatively, be explained in terms of a scavenging of the ethyl
4
1
33K
radicals. From our relative rate data we can estimate k
4
/
33K
k
2
¼ 11.5 which compares with the ratio of 80 reported
3
1
by Batt et al. If we consider the limiting case where every
ethyl radical produced reacts with a molecule of 2-methyl-2-
butyl nitrite and leads to the formation of a molecule of
acetone then, according to the relative-rate data obtained in
the present investigation, D[acetone]/D[butanone] ¼ 23. This is
still a factor of four smaller than the carbonyl yield (and hence
rate coefficient) ratio reported by Batt et al. A possible
explanation for this discrepancy may involve the oxidation of
1 2
Fig. 3 (a) Arrhenius plot of relative rate coefficients (k /k ) for the
decomposition of 2-methyl-2-butoxyl to acetone and ethyl (R1), and to
butanone and methyl (R2). The indicated error bars (2s) were calcu-
lated from replicate determinations of D[acetone]/D[butanone] (see
Table 2). The solid line represents the weighted least squares regression
line for these data. (b) Arrhenius plot of relative rate coefficients for the
decomposition of 2-methyl-2-butoxyl. The filled diamonds correspond
to the data measured in the present investigation; also included are
2
NO to NO by nitrosoethane (which is formed in the reaction
of ethyl radicals with NO) with the reproduction of ethyl
radicals.
ꢀ
data from experiments in which RO was produced from organic
peroxide precursors (by thermolysis (open circles) and photolysis (open
squares)) and from alkyl nitrite precursors (by thermolysis (open
triangles) and photolysis (crosses)). The solid line is reproduced from
Fig. 2a and is extrapolated to 1/T ¼ 0. The short horizontal line
indicates a value on the y-axis of ln(1/2).
ꢀ
C
2
H
5
þ NO - C
2
H
5
NO
(R9)
2
C H
5
NOþNOþNO (þNO)-C
2
H
5
ꢀþN
2
þNO
2
(þNO
2
)
(
R10)
4
Christie et al. observed NO quantum yields of up to 110
1
ꢀ
2
k
containing a significant excess of energy, will be more indis-
1
/k
2
will be reduced—i.e. the decomposition of these RO ,
from the photolysis of t-butyl nitrite in the presence of high
pressures of NO. If the product of reaction (R8) is a C-centred
radical, resulting from the abstraction of a secondary-H atom,
then it is possible that the reaction of this radical with NO2
may produce an alkoxyl radical, the subsequent decomposition
of which can yield acetone, acetaldehyde and NO:
3
6
criminate. The effect of activation is discussed by McMillan.
The relative rate coefficients obtained from experiments using
-methyl-2-butyl nitrite are, however, more difficult to explain.
2
Evidence for the production of ‘‘hot’’ radicals from the shorter
wavelength (l ¼ 253.7, 313 nm) photolysis of 2-methyl-2-butyl-
3
7
ꢀ
nitrite was presented by Durant and McMillan who observed
an increased randomisation in the bond breaking of 2-methyl-
CH CHC(ONO)(CH ) þ NO -
3
3 2
2
ꢀ
CH
3
CH(O )C(ONO)(CH
3
)
2
þ NO
(R11)
(R12)
2
-butoxyl at these shorter wavelengths—this is to say that the
relative rate of the more endothermic decomposition process
R2) increased. The dependence of the (366 nm) photolysis
quantum yield sum (Facetone þ Fbutanone) upon the concentra-
ꢀ
CH CH(O )C(ONO)(CH ) - CH CHO
3
3 2
3
(
þ CH C(O)CH þ NO
3
3
2
186
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 2 1 8 2 – 2 1 9 0
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5

View Article Online
Table 3 Rate data for the decomposition of 2-methyl-2-pentoxyl derived from measurements of the change in concentration of 2-methyl-2-pentyl
nitrite, acetone and pentanone (see text and eqn. (E3)). The initial mixing ratio of alkyl nitrite in each experiment was 100–140 ppmv. Between five
and seven separate measurements of reactant and product concentrations were made at each different temperature
ab
cd ꢂ1
decomp /s
e
ꢂ1d
ꢂ1d
6
/s
T/K
(kisom/kdecomp
)
k
k
5
/k
6
k
5
/s
k
3
3
2
2
2
3
3
3
3
a
83.2
24.2 (ꢃ9.4)
19.3 (ꢃ2.6)
18.1 (ꢃ2.7)
12.2 (ꢃ3.1)
9.53 (þ17.0, ꢂ8.0) ꢁ 10
21.2 (ꢃ6.9)
22.2 (ꢃ4.1)
12.0 (ꢃ0.76)
18.2 (ꢃ3.8)
16.7 (ꢃ2.6)
12.1 (ꢃ2.2)
9.1 (þ16.6, ꢂ8.2) ꢁ 10
4.3 (þ7.8, ꢂ3.9) ꢁ 10
4
4
2
98.2
05.2
20.2
33.2
45.2
2.30 (þ4.0, ꢂ1.8) ꢁ 10
2.2 (þ3.9, ꢂ1.7) ꢁ 10
9.9 (þ17.5, ꢂ7.8) ꢁ 10
4
4
3
3.25 (þ5.7, ꢂ2.5) ꢁ 10
3.0 (þ5.3, ꢂ2.3) ꢁ 10
2.5 (4.4, ꢂ1.9) ꢁ 10
4
4
3
8.44 (þ14.9, ꢂ6.7) ꢁ 10
8.0 (þ14.2, ꢂ6.6) ꢁ 10
4.4 (þ7.8, ꢂ3.6) ꢁ 10
5
5
4
7.52 (ꢃ0.94)
7.36 (ꢃ1.4)
2.12 (þ3.7, ꢂ1.6) ꢁ 10
2.0 (þ3.5, ꢂ1.6) ꢁ 10
1.2 (þ2.1, ꢂ0.9) ꢁ 10
5
5
4
3.14 (þ5.5, ꢂ2.4) ꢁ 10
2.9 (þ5.1, ꢂ2.3) ꢁ 10
2.4 (þ4.2, ꢂ1.9) ꢁ 10
b
c
kisom/kdecomp ¼ k
7
/((D[nitrite]/(D[acetone] þ D[pentanone]) ꢂ 1); ꢃ2s from replicate determinations;
k
decomp (¼ k
5
þ k
6
) values derived by
30
10
ꢂ1
estimating kisom (i.e. k
7
) using the following Arrhenius expression from Johnson et al. –kisom ¼ 9.0 ꢁ 10 exp(–30.3 kJ mol /RT). The uncertainty
d
ascribed to these estimated reference rate coefficients is þ175%, –75% (see text); Combined uncertainty from replicate determinations of carbonyl
e
yields and estimated kinetics of reference reaction; Values and uncertainties calculated from replicate measurements of D[acetone]/D[pentanone].
A reaction such as (R11) was proposed to be occurring for the
further chemistry of the d-hydroxyalkyl radical formed) were
not measured and thus the proportions of RO chemistry
4
n-propyl radical by Jaffe and Wan in order to explain
2
ꢀ
products observed in both thermal and photochemical reac-
tions of NO with butanal. For the purposes of the present
2
proceeding via this route were calculated from the difference
between the change in nitrite concentration (i.e. assumed to be
equal to the total concentration of alkoxyl radicals) and the
sum of the carbonyl yields (i.e. the total proportion of reaction
going via the decomposition channels). The total decomposi-
discussion it was not attempted to numerically simulate the
kinetics of the experiments of Batt et al. due to the catalytic
nature of the chemistry and the unknown kinetics of the
various potential chain-terminating reactions. What does seem
clear, however, is that the presence of high concentrations of
alkyl nitrite molecules and the potential catalytic oxidation of
NO to NO2 will add significant complexity to the reaction
system studied, and to the kinetic interpretation of product
yield data.
tion rate coefficient (kdecomp ¼ k
5
þ k
6
) was then simply
calculated from the following expression:
D[nitrite]/(D[acetone]þD[2-pentanone]) ¼ 1þ(k /kdecomp) (E3)
7
where nitrite corresponds to 2-methyl-2-pentyl nitrite. Average
values of kdecomp, k /k (and hence, k and k ) determined from
5
6
5
6
1
Using a value for k (at 298 K) obtained by extrapolation of
(E3) are listed in Table 3. These temperature-dependent data
for k and k are plotted in Arrhenius form in Fig. 4 where the
solid lines correspond to the following expressions:
3
1
4
ꢂ1
298K
1
the data of Batt et al. (3.9 ꢁ 10 s ) and the value of k
/
determined in the present investigation we propose a 298
4
5
2
98K
k
K value for k
2
ꢂ1
2
of 800 s . Both of these values are significantly
3
smaller than the theoretical calculations of Mereau et al.
´
3
12
¼ 4.5 ꢁ 10 exp(ꢂ47.4 (ꢃ16.2) kJ mol /RT)
ꢂ1
k
k
5
6
2
1
98K
5
ꢂ1
298K
2
4
ꢂ1
4
(
et al. (k
k
¼ 9.4 ꢁ 10 s , k
¼ 1.1 ꢁ 10 s ) and Peeters
12
¼ 7.1 ꢁ 10 exp(ꢂ53.0 (ꢃ16.1) kJ mol /RT)
ꢂ1
2
8
298K
1
6
ꢂ1 298K
ꢂ1
¼ 1.6 ꢁ 10 s , k
2
¼ 1.3 ꢁ 10 s ) and yet
the relative-rate coefficient measured at 298 K in the present
investigation is in reasonable agreement—within a factor of
two—with the results of both theoretical studies. Certainly, this
which were obtained by a weighted least squares regression
analysis. The uncertainty associated with the Arrhenius A
factors for (R5) and (R6) are very large, with the following
2
98K
7
18 ꢂ1
s
latter agreement is much better than with the values of k
2
ranges of values: A ¼ 1.7 ꢁ 10 to 1.2 ꢁ 10
and A ¼ 9.6 ꢁ
1
5
6
98K
31
7
17 ꢂ1
and k
2
derived from the data of Batt et al.
10 to 5.9 ꢁ 10
s (these ranges are 2 ꢁ the standard error in
the intercept from the regression analyses).
2
-Methyl-2-pentoxyl
The results of the experiments using 2-methyl-2-pentyl nitrite
are summarised in Table 3. As already mentioned, the kinetics
of the isomerisation reaction (R7) were used as an estimated
reference in order to estimate the kinetics of the two decom-
position channels (R5,6). Rate coefficients for (R7) were
estimated using the expression recently proposed by Johnson
2
8
et al. for the isomerisation of a simple alkoxyl radical
involving the cleavage of a primary C–H bond.
10
kisom(Primary) ¼ 9.0 ꢁ 10 exp(ꢂ30.3 kJ molꢂ1/RT) (E2)
Expression (E2) itself was derived from a series of relative-rate
experiments in which the reference reaction was that of a model
alkoxyl radical (2-pentoxyl) with O
uncertainty of rate coefficients estimated using expression
E2) (the uncertainty in the reference rate coefficient (e.g. see
ref. 9) combined with the uncertainty in the relative rate
2
. Thus, the combined
(
2
8
coefficients measured in the study of Johnson et al. ) is of
the order of þ175%, ꢂ75%. Thus by assuming that reactions
(
R5–R7) were the only ones available to 2-methyl-2-pentoxyl,
and that the quantum efficiency for thermalised alkoxyl radical
production from alkyl nitrite photolysis was unity, the total
rate of decomposition of 2-methyl-2-pentoxyl could be deter-
mined. The relative-rates of (R5) and (R6) were measured
directly (as D[acetone]/D[2-pentanone]). It should be noted that
the products of the isomerisation reaction (R7) (i.e. from the
Fig. 4 Arrhenius plot of rate coefficients determined for the decom-
position of 2-methyl-2-pentoxyl to acetone and n-propyl ((R5), open
circles), and to pentanone and methyl ((R6), open diamonds). The
indicated error bars were calculated from replicate measurements (see
Table 3) and include the estimated uncertainty in the reference rate
coefficient (see text).
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 2 1 8 2 – 2 1 9 0
2187

View Article Online
Correlation-type structure activity relationship (SAR)
As already stated, the first aim of the present investigation was
to re-determine the relative rate coefficient k
motivation for this was the fact that a value for k
1 2
/k and the
2
98K
1
, extra-
polated by Atkinson from the higher-temperature data of Batt
et al., was consistent with a kinetic correlation identified in this
298K
3
0
laboratory and yet a similarly extrapolated value for k
2
was not.
3
0
In the SAR of Johnson et al. the logarithm of 298 K
decomposition rate coefficients are correlated with the average
ionisation potential (IPmean) of the decomposition reaction
products, for a variety of simple alkoxyl radicals for which
rate data are available. The observed quadratic dependence
4
3
was rationalised in terms of simple Marcus theory in which
the activation barrier for the endothermic decomposition
reaction is non-linearly related to the enthalpy of the reaction.
The enthalpy of reaction was simply related to the difference in
ꢀ
IP of the reactant RO and the average IP of the reaction
products. If the former value is assumed to be constant
(
calculations suggest that the spin density is mostly centred
4
4
on the O atom of the radical) then the reaction enthalpy can
simply be related to the average IP of the reaction products.
Previously, Baldwin et al. proposed a dependence of Arrhe-
nius decomposition energies (E ) upon (estimated) reaction
4
5
ꢂ1
Fig. 5 Correlation plot of ln(k/y/s ) vs. the average ionisation
potential of the reaction products (where k is the 298 K rate coefficient
for a given decomposition channel and y is the reaction path degen-
d
enthalpies (DH ). This dependence was, however, subsequently
d
24
eracy for that channel) reproduced from Johnson et al. Decomposi-
tion rate coefficient data are included for the following radicals (open
circles): ethoxyl, n-propoxyl, i-propoxyl, 2-butoxyl (2 channels),
t-butoxyl, 3-pentoxyl, i-butoxyl and 2-pentoxyl. Data are also included
for 2-methyl-2-butoxyl corresponding to (R1) and (R2) extrapolated
amended to account for the fact that the correlation only
worked for reactions with a common radical product and thus
4
6
expression (E4) was proposed by Choo and Benson
E ¼ a þ bDHd
(E4)
25
d
from the measurements of Batt et al. (filled circle and open diamond,
respectively), and (R2) calculated using the relative rate coefficient
where b is a constant and a is linearly dependent upon the
ionisation potential of the alkyl radical fragment. The depen-
dence of a upon ionisation potential was attributed to polari-
sation in the late, tight transition state with a partial positive
charge residing on the carbon atom of the nascent alkyl radical
and a partial negative charge on the O atom (which now
resembles a carbonyl O atom).
1 2
(k /k ) determined in the present investigation (filled diamond). Data
are also included for the decomposition of 2-methyl-2-pentoxyl, (R5,6),
as determined in the present investigation (filled square and filled
triangle, respectively). The indicated error bars correspond to a varia-
tion in each rate coefficient by a factor of three.
will hence be easier if the ionisation potentials of both products
are relatively low. Thus the energy of the transition state will be
lower with respect to the energy of the alkyl radical and
carbonyl, regardless of whether it is accessed from the
b-scission of an alkoxyl radical or from the addition of an
3
0
alkyl radical to a carbonyl. The data used by Johnson et al. to
construct the original correlation are re-plotted in Fig. 5 along
with values for k and k y derived from the studies of Batt
1
2
3
1
If ionisation of the alkyl radical is relatively facile, the charge
transfer required to produce the product-like transition state
will be relatively easier and the reaction barrier relatively
lower. Observation of a large solvent effect in the decomposi-
tion of t-butoxyl is consistent with a charge separation in the
2
et al. and a second value for k obtained using the same value
for k and the rate coefficient ratio, k /k , determined in the
1
1
2
present investigation.
The new value for k is clearly seen to be consistent with the
2
quadratic line plotted through the data which provides some
further support for an additional source of acetone, in experi-
ments employing high concentrations of 2-methyl-2-butyl ni-
trite, other than the thermal decomposition of 2-methyl-2-
butoxyl radicals. Although this additional acetone source will
4
7
transition state. Expression (E4) was later updated by At-
7
kinson who proposed the following values for the coefficients:
4
8
a ¼ (2.4 ꢁ IP ꢂ 8.1) and b ¼ 0.36. Somnitz and Zellner,
however, observed of the development of expression (E4), as
new and additional experimentally determined rate data be-
came available, the importance of the reaction enthalpy in
determining the Arrhenius activation energy had decreased on
significantly affect the value of k derived from the carbonyl
2
3
1
1
yield ratio measurements of Batt et al. the value of k derived
from their data is still valid as—according to the relative rate
coefficients measured in the present investigation—the minor
decomposition channel still only constitutes o5%, at 298 K, of
the total decomposition of the radical. The kinetic correlation
reproduced in Fig. 5 suggests that the 298 K rate coefficient for
the decomposition of 2-methyl-2-butoxyl and 2-methyl-2-pen-
toxyl to acetone and alkyl radicals (ethyl and propyl, respec-
tively) should be very similar—this is to say that the reaction
products are alike with very similar average product ionisation
potentials (IP(acetone) ¼ 9.70 eV; IP(ethyl) ¼ 8.12 eV; IP
4
5
going from the original expression of Baldwin, to the expres-
4
6
sion of Choo and Benson (with b ¼ 0.58) and to the updated
7
version of expression (E4) of Atkinson (with b ¼ 0.36). They
further showed that the use of a constant value in place of the
DH
not degrade the accuracy of the predictions of activation
energies and thus that E has a purely structural dependence
on the decomposition fragments. This is consistent with the
d
term (i.e. independent of the alkoxyl radical reaction) did
d
3
0
correlation-type SAR of Johnson et al. which further takes
into account the ionisation potential of the carbonyl product.
In terms of the production of the polarised transition state
from the combination of the products this will require a shift of
electron density from the alkyl radical and the carbonyl and
4
9
(propyl) ¼ 8.10 eV) and this is seen to be the case. This also
y In Fig. 5 the rate coefficient has been divided by the reaction path
degeneracy, y.
2
188
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 2 1 8 2 – 2 1 9 0
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5

View Article Online
Table 4 Measured and estimated rate coefficients for the decomposition of simple alkoxyl radicals
ꢂ1
k
decomp 298 K/s
a
b
c
d
e
g
Alkoxyl radical
Experimental
Atkinson
´
Mereau et al.
Somnitz and Zellner
Peeters et al.
Johnson et al.
ꢀ
Ethoxyl (CH
3
O )
4.8
1.0 ꢁ 10
0.31
3.4
2.5 ꢁ 10
1.3 ꢁ 10
1.2
2.6
6.5
ꢀ
ꢀ
3
2
2
2
2
2
3
1
2
2
2
-Propoxyl (CH
-Propoxyl (CH
3
3
O )
O )
3.5 ꢁ 10
3.5 ꢁ 10
4.9 ꢁ 10
1.9 ꢁ 10
2
2
1.4 ꢁ 10
7.3
4.7 ꢁ 10
91
91
69
ꢀ
2
2
-Butoxyl (CH
-Butoxyl (C
3
O )
ꢀ
1.3 ꢁ 10
14
2.3 ꢁ 10
140
6.5 ꢁ 10
100
31
36
1.7 ꢁ 10
4
4
4
4
3
5
4
4
4
4
4
4
4
2
H
5
O )
ꢀ
1.8 ꢁ 10
3.5 ꢁ 10
3.0 ꢁ 10
9.4 ꢁ 10
1.1 ꢁ 10
5.7 ꢁ 10
2.2 ꢁ 10
3.4 ꢁ 10
1.6 ꢁ 10
1.7 ꢁ 10
1.0 ꢁ 10
3
2f
3
3
t-Butoxyl (CH O )
3
2.0 ꢁ 10
2.0 ꢁ 10
9.5 ꢁ 10
1.2 ꢁ 10
ꢀ
4
4f
5
4
2
2
-Methyl-2-butoxyl (C
2
H
5
O )
ꢀ
3.4 ꢁ 10
1.1 ꢁ 10
5.9 ꢁ 10
3.2 ꢁ 10
2
2f
3
3
-Methyl-2-butoxyl (CH O )
3
7.0 ꢁ 10
2.0 ꢁ 10
6.3 ꢁ 10
1.3 ꢁ 10
ꢀ
4
4
4
4
2f
4
4
i-Butoxyl (i-C
3
H
7
O )
1.2 ꢁ 10
3.4 ꢁ 10
3.3 ꢁ 10
1.6 ꢁ 10
b
2.0 ꢁ 10
9.3 ꢁ 10
1.3 ꢁ 10
ꢀ
3
3
4
4
2
3
a
-Pentoxyl (n-C
3
H
7
ꢀ
O )
7.5 ꢁ 10
9.0 ꢁ 10
1.7 ꢁ 10
1.1 ꢁ 10
4
4
4
4
-Pentoxyl (C
2
H
5
O )
3.3 ꢁ 10
3.9 ꢁ 10
3.4 ꢁ 10
Ref. 7; Ref. 33; Ref. 48, rate
3.7 ꢁ 10
30 c
d
e
Radical decomposition product shown in parentheses; Data taken from Table 1 of Johnson et al.;
f
coefficients calculated using SAR to estimated decomposition barrier (298 K, 760 Torr) and A factors reported by the authors; Rate coefficients
1
3
ꢂ1
g
per degenerate reaction channel; Ref. 34;
calculated using SAR to estimated decomposition barrier (298 K, 760 Torr) and A factors of 1 ꢁ 10
s
Rate coefficients calculated using expression (E5) and multiplying by reaction path degeneracy.
provides some validation of our choice of reference reaction—
i.e. a reaction whose kinetics were estimated. A 298 K value for
et al., Somnitz and Zellner, and Peeters et al. also predict rate
data to within around a factor of three of their measured values
for the majority of reactions considered. There are, however,
some notable exceptions. The method of Somnitz and Zellner
underestimates the rate coefficients for the decomposition of
t-butoxyl and i-butoxyl by factors of 10 and 60, respectively,
and the reasons for this are not clear, although their SAR was
formulated using theoretically calculated data for straight-
4
ꢂ1
k
2
5
of 2.2 ꢁ 10 s , estimated relative to the isomerisation of
-methyl-2-pentoxyl, is consistent with the correlation plot in
Fig. 5, as is the 298 K value for k . The solid line in Fig. 5
represented below by expression (E5)) is not that previously
6
(
3
0
presented by Johnson et al. but represents a re-calculated
second order polynomial taking account of the revised value of
k
2
and including the values of k
5
and k
6
.
5
chain alkoxyl radicals up to C . The methods of Mereau
´
et al. and Peeters et al. overestimate the rate coefficients for
both decomposition channels of 2-methyl-2-butoxyl by factors
of around 20 to 30. Again the reasons for this are unclear, but
the SAR predictions merely reflect the fact that, as described in
the Introduction section, the theoretically calculated rate data
also overestimate the experimental determinations. Thus, on
the evidence of the agreement between predictions and avail-
able experimental determinations, the correlation-type SAR
updated in the present paper appears to be the most reliable
method for estimating room temperature rate coefficients for
the decomposition of simple alkoxyl radicals.
ln(k/y/sꢂ1) ¼ ꢂ1.22 (IPmean) þ 17.64 (IPmean) ꢂ 49.93(
2
E5)
Here, y is the reaction path degeneracy for a given decomposi-
tion channel and IPmean is the average ionisation potential of
the carbonyl and alkyl radical fragments. In all but one case
(
the kinetics of the decomposition of t-butoxyl) the plotted rate
data agree with expression (E5) within a factor of two of the
experimentally determined rate coefficients.
Other SARs for alkoxyl radical decomposition kinetics have
3
3
recently been presented by Mereau et al. (who re-parame-
´
4
6
terised the Choo and Benson -type expression by replacing the
reaction enthalpy with a term which takes into account the
‘
ondary or tertiary substitution at the alkoxyl C atom), Somnitz
‘class’’ of alkoxyl radical—‘‘class’’ referring to primary, sec-
4
8
and Zellner (who related reaction barriers to differences in
carbon chain lengths of the carbonyl and radical fragments),
3
4
and Peeters et al. (who correlated reaction barriers with the
numbers of alkyl-, hydroxyl- and/or oxo-substituents on the
a- and b-carbon atoms of the breaking bond). As a compar-
ison, decomposition rate coefficients (for 298 K and 760 Torr)
7
33
have been estimated using the SARs of Atkinson, Mereau,
´
4
8
34
30
Somnitz and Zellner, Peeters et al. and Johnson et al. for
the following radicals for which the experimental rate data are
3
0
available (see Table 1 of Johnson et al. ): ethoxyl, n-propoxyl,
i-propoxyl, 2-butoxyl (2 channels), t-butoxyl, i-butoxyl, 2-
methyl-2-butoxyl (2 channels), 2-pentoxyl, and 3-pentoxyl.
Experimental and estimated rate coefficients are listed in Table
4, and estimated rate coefficients are plotted against the
measured values in Fig. 6 (as a log–log plot).
As can be seen in this figure each of the SARs captures the
general variation observed in decomposition rate coefficients—
which span some five orders of magnitude—but with varying
degrees of accuracy. The Atkinson SAR appears to be the least
reliable whereas the Johnson et al. SAR is able to predict rate
coefficients that are within a factor of three of the measured
Fig. 6 Log–log plot of estimated (298 K) alkoxyl radical decomposi-
tion rate coefficients vs. experimentally determined rate coefficients for
a series of simple alkoxyl radicals. Data are included for the decom-
position of ethoxyl, n-propoxyl, i-propoxyl, 2-butoxyl (2 channels),
t-butoxyl, i-butoxyl, 2-methyl-2-butoxyl (2 channels), 2-pentoxyl, and
3-pentoxyl radicals (see text).
values in all cases for the data in Fig. 6. The SARs of Mereau
´
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 2 1 8 2 – 2 1 9 0
2189

View Article Online
6
7
8
9
R. Atkinson and J. Arey, Chem. Rev., 2003, 103, 4605–4638.
R. Atkinson, Int. J. Chem. Kinet., 1997, 29, 99–111.
P. Devolder, J. Photochem. Photobiol. A, 2003, 157, 137–147.
J. J. Orlando, G. S. Tyndall and T. J. Wallington, Chem. Rev.,
Conclusions
The relative-kinetics of the decomposition of 2-methyl-2-bu-
toxyl have been re-investigated using the near-UV photolysis
of dilute (ppmv) mixtures of 2-methyl-2-butyl nitrite as the
source of alkoxyl radicals. Yields of acetone and butanone
were measured at a total pressure of 1 atm and as a function of
temperature between 280 and 340 K. The relative rate coeffi-
cient (k /k ) obtained at 298 K is around a factor of 20 smaller
2
003, 103, 4657–4689.
1
0
1
W. P. L. Carter, A. C. Lloyd, J. L. Sprung and J. N. Pitts Jr., Int.
J. Chem. Kinet., 1979, 11, 45.
R. A. Cox, K. F. Patrick and S. A. Chant, Environ. Sci. Technol.,
1981, 15, 587.
1
12 R. Atkinson, E. S. C Kwok, J. Arey and S. M. Aschmann,
Faraday Discuss., 1995, 100, 23.
13 N. Meunier, J. F. Doussin, E. Chevallier, R. Durand-Jolibois, B.
Picquet-Varrault and P. Carlier, Phys. Chem. Chem. Phys., 2003,
1
2
7
than that estimated by Atkinson by extrapolation of the
3
higher temperature data of Batt et al. Previous experimental
determinations of k /k in which 2-methyl-2-butoxyl was pro-
duced from the thermal decomposition of dialkyl peroxide
precursors are consistent with the temperature-dependence of
the relative rate coefficients measured in the present investiga-
tion. Results from other studies in which the alkoxyl radical
was produced from the photolysis of dialkyl peroxides or alkyl
nitrites, or from the thermal decomposition alkyl nitrite pre-
cursors are not consistent with the temperature-dependent
1
1
2
5
, 4834–4839.
L. Batt and R. T. Milne, Int. J. Chem. Kinet., 1977, 9, 549.
15 L. Batt and R. T. Milne, Int. J. Chem. Kinet., 1977, 9, 141.
16 S. Dobe, T. Berces and F. Marta, Int. J. Chem. Kinet., 1986, 18,
29.
R. J. Balla, H. H. Nelson and J. R. McDonald, Chem. Phys., 1985,
9, 323–335.
1
4
´
´
´
3
1
1
7
8
9
C. Mund, C. Fockenberg and R. Zellner, Ber. Bunsen. Phys.
Chem., 1998, 102, 709–715.
values of k
1
/k
2
measured in the present investigation. It seems
19 F. Caralp, P. Devolder, C. Fittschen, N. Gomez, H. Hippler, R.
Mereau, M. T. Rayez, F. Striebel and B. Viscolcz, Phys. Chem.
Chem. Phys., 1999, 1, 2935–2944.
likely that these differences are attributable to the effects of
chemical activation in alkoxyl radicals (when formed from the
UV photolysis of organic peroxide precursors) or to additional
carbonyl-forming chemistry occurring in the presence of high
´
2
0
1
P. Devolder, C. Fittschen, A. Frenzel, H. Hippler, G. Poskreby-
shev, F. Striebel and B. Viscolcz, Phys. Chem. Chem. Phys., 1999,
1, 675.
concentrations of alkyl nitrite precursors. The 298 K value of k
1
2
M. Blitz, M. J. Pilling, S. H. Robertson and P. W. Seakins, Phys.
Chem. Chem. Phys., 1999, 1, 73–80.
3
1
derived form the data of Batt et al. is consistent with a non-
linear correlation, between the logarithm of the rate coefficient
and the average IP of the reaction products, previously pre-
22 C. Fittschen, A. Frenkel, K. Imrik and P. Devolder, Int. J. Chem.
Kinet., 1999, 31, 860–866.
23 C. Fittschen, H. Hippler and B. Viscolcz, Phys. Chem. Chem.
Phys., 2000, 2, 1677–1683.
3
0
2
sented by this laboratory. The 298 K value of k —derived
from this latter value for k1 and the relative rate coefficient
determined in the present investigation—is also consistent with
the kinetic correlation plot. Rate coefficients have also been
estimated for the two decomposition channels of the 2-methyl-
2
4
W. Deng, C. J. Wang, D. R. Katz, G. R. Gawinski, A. J. Davis
and T. S. Dibble, Chem. Phys. Lett., 2000, 330, 541–546.
25 W. Deng, A. J. Davis, L. Zhang, D. R. Katz and T. S. Dibble,
J. Phys. Chem. A, 2001, 105, 8985–8990.
2
2
2
6
7
8
L. Zhang, K. A. Kitney, M. A. Ferenac, W. Deng and T. S.
Dibble, J. Phys. Chem. A, 2004, 108, 447–454.
G. Falgayrac, F. Caralp, N. Sokolowski-Gomez, P. Devolder and
C. Fittschen, Phys. Chem. Chem. Phys., 2004, 6, 4127–4132.
D. Johnson, P. Cassanelli and R. A. Cox, J. Phys. Chem. A, 2004,
108, 519–523.
2
as an estimated kinetic reference. Once again, the estimated 298
-pentoxyl radical (R5,6) using the isomerisation of this radical
K values of k and k are consistent with the non-linear kinetic
5
6
correlation which lends support to the choice of an estimated
rate coefficient as the kinetic reference. The 298 K rate coeffi-
cient for the decomposition of 2-methyl-2-pentoxyl to acetone
29 P. Cassanelli, D. Johnson and R. A. Cox, 2005, manuscript in
preparation.
and n-propyl (k ) is similar to that for the decomposition of 2-
5
3
0
D. Johnson, P. Casanelli and R. A. Cox, Atmos. Environ., 2004,
8, 1755–1765.
L. Batt, T. S. A. Islam and G. N. Rattray, Int. J. Chem. Kinet.,
978, 10, 931.
1
methyl-2-butoxyl to acetone and ethyl (k ) which is not un-
3
expected as both reactions give similar products (with similar
3
1
30
values for IPmean). The expression presented by Johnson et al.
1
for the estimation of 298 K rate coefficients for the decomposi-
tion of simple alkoxyl radicals has been updated to incorporate
the revised and previously unreported rate data. This correla-
tion-type SAR appears to be the most accurate, currently
reported, SAR method for estimating (298 K) decomposition
rate coefficients for simple alkoxyl radicals, for which experi-
mental rate data are available at this time.
32 R. Atkinson, D. L. Baulch, R. A. Cox, R. F. Hampson, Jr, J. A.
Kerr, M. J. Rossi and J. Troe, J. Chem. Phys. Ref. Data, 1997, 26,
521–1011.
R. Mereau, M. T. Rayez, F. Caralp and J. C. Rayez, Phys. Chem.
3
3
4
´
Chem. Phys., 2000, 2, 3765–3772.
J. Peeters, G. Fantechi and L. Vereecken, J. Atmos. Chem., 2004,
3
4
8, 59–80.
35 J. H. Raley and D. O. Collamer, J. Am. Chem. Soc., 1952, 74,
606.
36 G. R. McMillan, J. Am. Chem. Soc., 1962, 84, 2514.
1
Acknowledgements
3
3
7
8
D. Durant and G. R. McMillan, J. Phys. Chem., 1966, 70, 2709.
W. A. Noyes, in Organic Syntheses, ed. A. H. Blatt, Wiley, New
York, 1943, vol. 2, pp. 108–109.
The authors would like to acknowledge the NERC and the EU
for funding under the UTLS thematic programme and UTO-
PIHAN-ACT project (EVK2-CT2001-00099), respectively.
39 J. Heicklen, Adv. Photochem., 1988, 14, 177.
4
0
J. G. Calvert and J. N. Pitts, Jr, Photochemistry, J. Wiley and
Sons, New York, 1966, pp. 452–455.
References
4
1
M. I. Christie, C. Gillbert and M. Voisey, J. Chem. Soc., 1964,
3147–3155.
1
2
3
R. P. Wayne, Chemistry of Atmospheres, Oxford University Press,
Oxford, 2000.
42 S. Jaffe and E. Wan, Environ. Sci. Technol., 1974, 8, 1024–1025.
43 F. Jensen, Introduction to Computational Chemistry, Wiley, New
York, 1999, pp. 364–368.
´
44 R. Mereau, M. T. Rayez, F. Caralp and J. C. Rayez, Phys. Chem.
Chem. Phys., 2000, 2, 1919–1928.
45 A. C. Baldwin, J. R. Barker, D. M. Golden and D. G. Hendry, J.
Phys. Chem., 1977, 81, 2483.
46 K. Y. Choo and S. W. Benson, Int. J. Chem. Kinet., 1981, 13, 13.
47 M. Weber and H. Fischer, J. Am. Chem. Soc., 1999, 121, 7381.
48 H. Somnitz and R. Zellner, Phys. Chem. Chem. Phys., 2000, 2,
4319–4325.
B. Finlayson-Pitts and J. N. Pitts, Jr, Chemistry of the Upper and
Lower Atmosphere, Academic Press, New York, 1999.
H. Singh, Y. Chen, A. Tabazadeh, Y. Fukui, I. Bey, R. Yantosca,
D. Jacob, F. Arnold, K. Wohlfrom, E. Atlas, F. Flocke, D. Blake,
N. Blake, B. Heikes, J. Snow, R. Talbot, G. Gregory, G. Sachse,
S. Vay and Y. Kondo, J. Geophys. Res., 2000, 105, 3795–3805.
PORG, Ozone in the UK, Fourth Report of the Photooxidants
Review Group, 1997, Department of the Environment, Transport
and Regions.
4
5
IPCC, Aviation and the Global Atmosphere, Intergovernmental
Panel on Climate Change, 1999, Cambridge University Press,
Cambridge.
49 NIST, Chemistry WebBook, NIST Standard Reference Database
No. 69, March 2003 Release.
2
190
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 2 1 8 2 – 2 1 9 0
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5