Baratron Type 122A pressure transducer. A gas cell of 8.0(1)
cm length with quartz windows was used.
Relative Rate Measurem ents. The reaction rate coef-
ficients were determined by the relative rate method
ter ions with m/ z ) M-19 resulting from HF elimination
from the MH+-ions. The relative concentrations of the
reference compounds CH3OH, CH3CH2OH, CH2Cl2, CHCl3,
CH2ClCH2Cl, CH2FCHF2, CF3CH2OCH3, and C4F9OC2H5 were
determined from the m/ z signals at 33 [CH3OH]H+, 47 [CH3-
CH2OH]H+, 49 ([CH2Cl2]H+ - HCl), 83 ([CHCl3]H+ - HCl),
63 ([CH2ClCH2Cl]H + - HCl), 65 ([CH2FCHF2]H+ - HF), 115
[CF3CH2OCH3]H+, and 245 ([C4F9C2H5]H+ - HF). The PCI-
MS spectra of all compounds used are given as Supporting
Information.
Chem icals. Hydroxyl radicals were generated by pho-
tolysis of O3 in the presence of H2 employing a Philips TUV
30W lamp (λmax ∼ 254 nm) mounted in a quartz tube in the
smog chamber. The lamp was turned off during recording
of the spectra.
kS
S + X
R + X
98 Prod
kR
98 Prod
(2)
where S is the substrate of interest, R is the reference
compound, X is the radical, and kS and kR are the reaction
rate coefficients. Assuming that the substrate and reference
compounds are lost solely via reaction with the radical species
of interest and that they are not reformed in any process, the
relative rate coefficient, krel, can be obtained by the following
relation
O3 + hν (λmax ∼ 254 nm) f O(1D) + O2
O(1D) + H2 f OH + H
(4)
(5)
[S]0
[S]t
[R]0
[R]t
kS
kR
ln
) krel‚ln
;
krel
)
(3)
{
}
{
}
in which [S]0, [R]0, [S]t, and [R]t denote the concentrations of
S and R at time zero and t, respectively. A plot of ln{[S]0/ [S]t}
vs ln{[R]0/ [R]t} will give krel as the slope. Data from inde-
pendent experiments were analyzed jointly according to eq
3 using a weighted least squares procedure including
uncertainties in both reactant concentrations (6); the un-
certainties in the reactant concentrations were taken as the
variance in three consecutive measurements but not less
than 1%. The quoted errors in this work represent the 3σ
statistical errors and include the uncertainty in the reference
reaction rate coefficients, but not any possible systematic
errors.
This OH production scheme produces not only OH radicals
in the ground-state but also in excited vibrational states (7-
9). However, the collisional quenching rate coefficient of OH
by O2 and N2 is of the order of 10-10 cm3 molecule-1 s-1 (10),
that is 2 to 3 orders of magnitude faster than the OH reaction
rate coefficients of the HFEs and the reference compounds.
In addition, the mixing ratios of O2 and N2 are 5 orders of
magnitude larger than those of the HFEs and the reference
compounds, and one may therefore safely assume that the
HFEs and the reference compounds react exclusively with
OH in the vibrational ground state.
Ozone was produced from oxygen by using a TRI-OX
Ozone Generator model T-200 that converts approximately
2% of the oxygen gas flow to ozone. Cl atoms were generated
by photolysis of Cl2 using two Philips TL 18W/ 08 fluorescence
lamps (λmax ∼ 375 nm); photolysis was carried out in time
intervals of 1 to 20 min. Typical mixing ratios were as
follows: HFEs and reference compounds, 2-6 ppm; Cl2, 5-10
ppm; H2, 1000 ppm; O3, 100-400 ppm. Synthetic air (CO +
NOx <100 ppb, CnHm <1 ppm), helium (99.9999%), hydrogen
(99%), and oxygen gas (99.95%) were delivered from AGA.
CF3CH2OCHO was synthesized from CF3CH2OH (Fluorochem
Ltd.) and concentrated HCOOH and purified by standard
methods. (CF3)2CHOCH3 (purity 97%), CHF2CF2CH2OCH3
(purity 97%), and CF3CF2CH2OCH3 (purity 97%) originated
from Fluorochem Ltd., CF3CH2OCH2CF3 (purity 99%) from
Aldrich, and C4F9OC2H5 from 3M, while the other reference
compounds were standard laboratory chemicals. The samples
were distilled in a vacuum prior to use. The only impurity
observed in the gaseous HFEs by IR and headspace GC-MS
was CF2O, which was removed by single-plate vacuum
distillation at -78 °C.
The measurements were performed at 1013 ( 15 hPa and
298 ( 2 K in synthetic air in a 250 L smog chamber of
electropolished stainless steel. In situ air analyses were
obtained with an Agilent 6890/ 5973 GC-MS employing
chemical ionization (CI). The GC was operated under
isothermal conditions at 40 °C. A constant overpressure of
ca. 5 hPa was applied to the reactor to ensure a steady flow
of ca. 20 mL/ min through a 0.5 mL GC sampling loop the
content of which was injected into the GC in a 1:50 split
mode using helium as the carrier gas. The column used for
the separation was DB-Waxetr with a length of 30 m, internal
diameter of 0.25 mm, and film coating of 0.25 µm. The DB-
Waxetr column was coated with a poly(ethylene glycol). The
inlet and the sample loop temperatures were kept constant
at 100 °C.
An inert tracer (1,2-perfluorodimethylcyclohexane) was
added in initial experiments to monitor the dilution of the
reactants due to the constant flow out of the reactorsthe
dilution was in all cases negligible. The positive chemical
+
ionization (PCI) mode employing CH5 was used, as it is a
soft method of ionization with little fragmentation of the
reactants and the products. As a first step, the full scan mode
was selected to obtain complete mass spectra to identify the
reactants and the reaction products in the gas-chromato-
grams. The compounds studied, the reference compounds,
and the reaction products all had unique mass peaks, which
made it possible to operate the MS in the Selective Ion Mode,
SIM, in which only chosen m/ z numbers are followed. The
SIM mode was then employed for the quantification of the
individual compounds, the advantage being a reduced
background noise and the elimination of overlap in cases of
incomplete gas-chromatographic separation.
The relative concentrations of the HFEs (CF3)2CHOCH3,
CF3CH2OCH2CF3, CHF2CF2CH2OCH3, and CF3CF2CH2OCH3,
and the ester CF3CH2OCHO were determined from the m/ z
signals at 183 [(CF3)2CHOCH3]H+, 183 [CF3CH2OCH2CF3]H+,
147 [CH3OCH2CF2CHF2]H+, 165 [CF3CF2CH2OCH3]H+, and
129 [CF3CH2OCHO]H+, respectively, and/ or from the daugh-
Results and Discussion
IR Absorption Cross-Sections. Infrared absorption cross-
sections were determined from the absorbance spectra
assuming that the gas was ideal and applying a baseline
correction. The latter was performed by subtracting a
polynomial function, obtained by fitting the regions of the
spectrum where no absorptions were expected. The integra-
tions over the absorption bands were carried out using a
method that defines the baseline from an average of two
points on one side of the band and the average of two points
on the other side of the band.
The integrated absorption cross-section of the absorption
bands, or regions of overlapping bands, was determined by
plotting the integrated absorbance against the product of
the number density and the path length. Since none of the
regression lines had a y-intercept significantly different from
9
5 5 6 8 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 21, 2004