Atmospheric Chemistry of FCOx Radicals: UV Spectra and Self

Making Sure That Hydrofluorocarbons Are “Ozone Friendly” ... the CF3CFO2CF3 Radical, Its Reactions with NO and NO2, and Fate of the CF3CFOCF3 Radi...
0 downloads 0 Views 1MB Size
2346

J . Phys. Chem. 1994, 98, 2346-2356

Atmospheric Chemistry of FCO, Radicals: UV Spectra and Self-Reaction Kinetics of FCO and FC(0)Oz and Kinetics of Some Reactions of FCO, with 0 2 , 0 3 , and NO at 296 K Timothy J. Wallington' Research StafJ SRL-E3083, Ford Motor Company, Dearborn, Michigan 481 21 -2053 Thomas Ellermann,+ Ole J. Nielsen,' and Jens Sehested Section for Chemical Reactivity, Environmental Science and Technology Department, R i m National Laboratory, DK-4000 Roskilde, Denmark Received: September 28, 1993; In Final Form: December 24, 1993'

A pulse radiolysis technique has been used to measure the UV spectra of FCO and F C ( 0 ) 0 2 radicals over the ranges 265-275 and 220-290 nm, respectively. At 269.6 nm, UFCO = (1.7 f 0.2) X cm2 molecule-'; at 250 nm, UFC(O)O~= (2.1 f 0.2) X cm2 molecule-'. The decay of UV absorption was used to study the kinetics of the self-reactions of FCO and F C ( 0 ) 0 2 radicals at 296 f 2 K. Observed self-reaction rate constants, defined as-d[FCO]/dt = 2k7[FC0I2and-d[FC(O)O2]/dt = 2kgoJFC(0)02I2, were k, = (1.6 f 0.2) X 10-11 and kgob = (6.0 f 0.7) X 10-12 cm3 molecule-' s-l. A rate constant, kl = (1.2 f 0.2) X 10-12 cm3 molecule-' s-l, was derived for the addition reaction FCO 0 2 F C ( 0 ) 0 2 in 1000 mbar of SF6 diluent. Rate constants for the reactions of FCO, FC(O)O, and F C ( 0 ) 0 2 radicals with NO were determined to be (1.0 f 0.2) X lo-',, (1.3 f 0.7) X 10-l0, and (2.5 f 0.8) X lo-", cm3 molecule-' s-l, respectively. An upper limit of k3 < 6 X cm3 molecule-l s-l was measured for the reaction of F C ( 0 ) O radicals with 0 3 . Finally, a rate constant of ( 5 . 5 f 0.7) X 10-12 cm3 molecule-' s-l was measured for the association reaction of F atoms to NO in 1000 mbar of SF6 diluent. All experiments were performed at 296 f 2 K. Results are discussed with respect to the atmospheric chemistry of hydrofluorocarbons.

+

Introduction An international effort is taking place to replace chlorofluorocarbons (CFCs) with environmentallyacceptablealternatives.'" Hydrofluorocarbons (HFCs) are one class of CFC replacements. For example, HFC-134a (CFsCFH2) is a replacement for CFC12 (CFZC12) in air conditioning and refrigeration units. Prior to their large-scale industrial use, the environmental consequences of HFC release into the atmosphere need to be considered. To define the environmental impact of the release of HFCs, two main questions need to be answered. First, what will be the effect on the ozone layer? Second, what will be the effect on potential global climate change? For each of these questions a subquestion needs to be considered: what effect will the atmospheric degradation products of HFCs have on the ozone layer and potential global climate change? The answer to the first question is straightforward; HFCs contain no chlorine and therefore are not expected to impact the ozone layer. The second question has been addressed by atmospheric modeling s t u d i e ~ . ~ . ~ Because of the ability of OH radicals to scavenge HFCs in the lower atmosphere, the atmospheric lifetimes of HFCs are, in general, much shorter than CFCs. With shorter lifetimes the HFCs have smaller global warming potentials. For example, the global warming potential of HFC- 134a is approximately an order of magnitude less than for CFC-12. Part of the subquestion has been addressed in our laboratories by identifyingthe atmospheric degradation products for a series of HFCs (HFC-23,6 HFC-32,7 HFC-41,* HFC-125: HFC-134,"J and HFC-134a11J2). From work performed in our laboratories, and elswhere,I3-l5 it is clear that either HC(0)F or COF2, or both, are formed during the atmosphericoxidationof every HFC. The atmospheric chemistries of HC(0)F and COFl are believed to be dominated by incorporation into rain-cloud-sea water followed by hydrolysis. Photolysis,while a minor loss process for both HC(0)F and COFz

* Authors to whom correspondence may be addressed.

7 Permanent 0

address: NERI, DK-4000, Roskilde, Denmark. Abstract published in Advance ACS Abstracts, February 1, 1994.

0022-3654/94/2098-2346$04.50/0

-

overall, may be important in the stratosphere. Photolysis of HC( 0 ) F and COF, is expected to produce FCO radicals which will add 0 2 to give FC(0)Oz radicals. There has been speculation that FC(O)O, radicals could participate in catalytic ozone destruction cycles,l6 for example:

+ 0, F C ( 0 ) O + 0,

FC(O)O,

-

-

FC(0)O

+ 20,

FC(O)O,

+ 0,

Detailed information regarding the atmospheric chemistry of FCO, radicals, especially their reaction with 0 3 , is needed to place discussions of the environmental impact of HFC usage on a sound scientificbasis. Unfortunately,there is little data available concerning the atmospheric chemistry of FCO, radicals. We have used pulse radiolysis combined with transient UV absorption spectroscopy to provide such information. Results concerning the UV absorption spectra and self-reaction kinetics of FCO, FC(O)O, and FC(0)Oz radicals and the kinetics of reactions 1-4 are reported herein. FCO

+ + + -

+ 0, + M

FC(0)02

NO

FC(0)O FC(0)O

+M

(1)

+ NO,

(2)

FC(O)O,

FC(0)O

0,

NO

products

FNO

+ CO,

(3) (4)

Experimental Section The pulse radiolysistransient UV absorptionspectrometerused in the present experiments has been described previo~s1y.l~ FCO radicals were generated by the radiolysis O f SFs/CO gas mixtures in a 1-L stainless steel reactor with a 30-11s pulse of 2-MeV electrons from a Febetron 705B field emission accelerator. SF6

0 1994 American Chemical Society

The Journal of Physical Chemistry, Vol. 98, No. 9, 1994 2347

Atmospheric Chemistry of FCO, Radicals was always in great excess and was used to generate fluorine atoms:

SF, + e-

-+ F

products

(5)

I

g

0.20

7

F+CO-FCO (6) Five sets of experimentswere performed. First, the ultraviolet absorption spectra of FCO and FC(0)02 radicals were determined by observing the maximum in the transient UV absorption at short times (10-40 ps) following the pulse radiolysis of SF6/CO and SF6/C0/02 mixtures, respectively. Second, using a longer time scale (100-400 ps), the subsequent decay of the absorption was monitored to determine the kinetics of reactions 7 and 8: FCO FCO products (7)

-

+ FC(O)O, + FC(O)O,

-

products

(8)

Third, the rate of conversion of FCO into FC(0)Oz radicals was monitored, as a function of the concentration of 0 2 present, to derive a value for the rate constant of reaction 1. Fourth, the rate of NO2 formation and the behavior of FC(0)O following the pulse radiolysis of SF6/C0/O2/NO mixtures were used to measure kz and k4. Finally, the radiolysis of SF6/C0/02/03 mixtures was used to study the behavior of FC(0)O radicals in the presence of O3and, hence, to derive kinetic data for reaction 3. To monitor the transient UV absorption,the output of a pulsed 150-W xenon arc lamp with multipassed through the reaction cell using internal White cell optics (40-, 80-, or 120-cm path length). A McPhersongrating spectrometer,HamamatsuR 955 photomultiplier, and Biomation 8 100 waveform digitizer were used to detect and record the light intensity at the desired wavelength. The spectral resolution used was 0.8 nm. Data transfer and storaga were controlled by a PDPl 1 minicomputer. The following reagent concentrations were used: SF6,95-1000 mbar; 0 2 , 0-40 mbar; CO, 40 mbar; NO, 0-10 mbar (1 mbar = 2.46 X 1016 molecules cm-3). All experimentswere performed at 296 f 2 K. Ultrahigh-purity0 2 was supplied by L'Air Liquide, SF6 (99.97%) and CO (>99.9%) were supplied by Gerling and Holz, and NO (>99.8%) was supplied by Messer Griesheim. All reagents were used as received. Ozone was produced by flowing O2 through a conventional discharge ozonizer. The 02 was purified before entering the ozonizer by flowing through silica gel. Trace amounts of N02, N205,and H N 0 3were removed from the ozone/oxygen mixture by flowing through a silica gel trap cooled to -80 OC. Up to 1.0% O3 in 0 2 was produced in this way. Ozone concentrations in the reaction cell were determined by measurement of the absorption at 272 or 288 nm before and after filling with reaction mixtures. Ozone absorbs strongly at these wavelengths ( ~ ~ ( 2 nm) 7 2 = 6.92 X and uo1(288 nm) = 1.79 X 10-1s cm2molecule-l.18 No absorption at thesewavelengths was detected when the reaction cell was filled with reactant gases other than 03.Ozone concentrations used were in the range 0.25-0.39 mbar. Immediately prior to the conclusion of the experimental work described herein, a Princeton Applied Research OMA-I1 diode array was installed at the exit slit of the monochromator in place of the photomultiplierwhich has been routinely used to measure the transient absorptions. The system consisted of the diode array, animageamplifier (type 1420-1024HQ),acontroller (type 1421), and a conventional PC computer for data handling and storage. Spectral calibrationwas achievedusing Hg and Ne pen ray lamps. Diode array spectra of FCO and FC(0)O radicals were recorded and are presented herein.

Results and Discussion

Spectrum of FCO Radicals. Following the pulse radiolysis of SF6/CO mixtures a rapid increase in the UV absorption was

1 E

a, ffl

2

0.15

v

a 0

E

O.lO\

n i 0 ffl

n Q

0.05 :

0.00

M

Figure 1. Transient absorption at 269.6 nm following the radiolysis of a mixture of 40 mbar of CO and 960 mbar of SFs using a UV path length of 120 cm. The solid line is a second-order decay fit.

observed, followed by a slower decay. Figure 1 shows the transient absorption at 269.6 nm observed for the first 180 ps following the radiolysis of a mixture of 40 mbar of CO and 960 mbar of SF6 (1 mbar = 2.46 X 10l6 molecules cm-3). The UV path length was 120 cm. No absorption was observed when either 40 mbar of CO or 960 mbar O f SF6 was radiolyzed separately. We ascribe the absorptionshown in Figure 1 to the formationof FCO radicals and their subsequent loss by self-reaction. Measurement of the absolute absorption spectrum of FCO radicals requires calibration of the initial F atom concentration. Additionally, experimental conditions are needed to ensure that there is 100%conversion of F atoms to FCO radicals. The yield of F atoms was established by monitoring the transient absorption at 260 nm due to methylperoxy radicals produced by radiolysis of SF6/CH4/02 mixtures as described previou~ly.~~ Using a value of 3.18 X 10-1s cm2 molecule-' for u(CH302) at 260 nm,M the yield of F atoms at 1000 mbar of SF6 was measured to be (2.92 f 0.15) X 10l5 ~ m at- full ~ irradiation dose. The quoted error represents statistical uncertainty (2 standard deviations). Potential systematic error associated with the uncertainty in u(CH3O2) is estimated to be *lo%. Combining the statistical and potential systematic uncertainties, we arrive at a calibrated F atom yield of (2.92 f 0.33) X 1015~ m - ~ . To work under conditions where the F atoms are converted stoichiometricallyinto FCO, it is necessary to consider potential interfering secondary chemistry such as reaction 9.

F

+ FCO

-

products

(9)

To check for such a complication, the maximum transient absorption at 269.6 nm was measured in experimentsusing [CO] = 40 mbar and [SF,] = 960 mbar, with the radiolysis dose varied over a wide range. Figure 2 shows the maximum absorption at 269.6 nm as a function of the radiolysis dose. (The UV path length was 80 cm.) For doses less than one-half of maximum, the absorption increased linearly with radiolysis dose (and hence initial radical concentration),suggesting that unwanted secondary radical-radical reactions are unimportant. The absorption observed using maximum dose was approximately 10%less than that expected from a linear extrapolation of the low-dose data. We ascribe this behavior to incomplete conversion of F atoms into FCO radicals at the highest initial F atom concentration used. Linear least-squares analysis of the data in Figure 2 (full dose experiment excepted) gives a slope of 0.168 f 0.005. Combining this value with the calibrated yield of F atoms of (2.92 f 0.33) X loi5molecules ~ m (full - ~dose and [SFs] = 1000 mbar) gives u(FC0) = (1.73 f 0.20) X 10-18 om2 molecule-' at

2348 The Journal of Physical Chemistry, Vol. 98, No. 9, 1994

Wallington et al. 250

.

.

I

,

.

I

I

,

I

.

I

,

.

.

l

,

.

.

.

,

.

.

FCO

c

I

.

2

200

3

u

a, 0 E

150

N

E

0

100

N 0

I

0 r-

d

50

o ~ " ' " " " ' " ' ' " ' ' " ' ' ~ > k8 each FC(0)O radical formed in reaction 8a will consume an additional FC(0)Ozradical via reaction 14. Hence, keo& could be up to a factor of 1.8 greater than the true bimolecular rate constant ks. Kinetic data for reaction 14and information regarding the lifetime of the trioxide are needed to fully assess the role played by reaction 14. We will present kinetic data for reaction 14 in a subsequent section dealing with the study of the reaction of FC(0)O radicals with 03.Unfortunately, there are no available data concerning the lifetime of the trioxide. Thus, it is not possible to fully assess

Atmospheric Chemistry of FCO, Radicals

The Journal of Physical Chemistry, Vol. 98, No. 9, 1994 2351

of the possible interference of reaction 14, the value of ksob measured in the present work may be an overestimateof the true bimolecular rate constant k8. Hence, 2kso~[FC(0)02]02 may be 0 30 an overestimate of the rate of production of FC(0)O radicals. 0 Consequently, the value of u(FC(0)O) derived above is a lower limit. In principle, the u(FC(0)O) could be up to a factor of 1.8 greater than derived above. According, we choose to quote u2 0.20 O 2 I (FC(0)O) = (2.0 f 0.6) X 10-17cm2molecule-1. Maricq et a1.28 w have reported absolute absorption cross sections for FC(0)O 0.15measured using a technique which was similar to that described . I 9 above. In their analysis Maricq et al. assumed there were no 1 ," 0.10 - 1 interferences caused by reaction 14. As stated above, this n 6 assumption is not necessarily justified. Nevertheless, a direct 005comparison of the results obtained in our work and that of Maricq et a1.28can be made by using u(FC(0)O) = 1.4 X l e 1 ' cm2 0 00 molecule-' at 696.1 nm to calibrate our absorption spectrum. As 0 50 100 150 seen from Figure 9 the agreement is excellent. From a detailed consideration of the formation and decay of the transient Time, ps absorptionat 696.1 nm, kinetic data concerning reactions 13 and Figure 8. Transient absorption at 696.1 nm following the radiolysis of 14 can be derived. Discussion of these results can be found in a mixture of 40 mbar of CO, 40 mbar of 02,and 920 mbar of SFs (full dose, UV path length = 80 cm). thesectiondescribingthestudy ofthereactionof FC(0)Oradicals with 03. 1500 Finally, the diode array system was used to obtain the spectrum of FC(0)O radicals given in Figure 9. The conditions were as I follows: spectral resolution, 0.13 nm; gate, 7 ps; delay, 50 ps; [CO], 40 mbar; [Oz], 40 mbar; [SF6], 920 mbar, full radiolysis 3 0 dose. As seen from this figure, the band positions measured in 2 1000 the present work are in good agreement with the results from ref 28. N Kinetic Studies of the Reactions F NO FNO and FCO E 0 NO Products. As a preliminary exercise prior to the 0 investigation of the reaction of FC(0)02 radicals with NO, a N I 500 series of experimentswere performed to investigatethe reactions 0 of NO with both F atoms and FCO radicals. The reaction of F with NO was measured by observing the first-order rise of the d absorption of FNO at 3 10.5nm following the radiolysis of mixtures of 1000 mbar of SF6 and 2-10 mbar of NO. The available literature UV spectra of FN032,33 do not report absolute absorption 690 700 710 cross sections. To provide such, the maximum absorption Wavelength, nm following the radiolysis of 1000 mbar of SF6 and 2 mbar of NO Figure 9. Absorption cross-section data for FC(0)O radicals measured was measured as a function of the radiolysis dose (UV path length in this work ( 0 ) . The solid line A is the spectrum reported by Maricq = 80 cm); results are shown in Figure 2, from which we derive et a1.28 The solid line B is the diode array spectrum recorded in this work. u(FN0) at 310.5 nm = (4.73 f 0.66) X lO-lg cm2 molecule-'. The formation of FNO followed first-order kinetics, allowing a the impact of reaction 14 on kinetic studies of k8. This point is value of k(F+NO) = (5.5 f 0.7) X 10-12 cm3 molecule-1 s-1 to discussed further in the following section. be measured. Variation of the radiolysis dose by an order of Study of FC(0)O Radicals. The self-reaction of FC(0)02 magnitude and the NO concentration by a factor of 5 had no radicals produces FC(0)O radicals in a yield of 80 f 20%.31 discernible effect on the value of k(F+NO). FC(0)O radicals absorb strongly in the visible at 500-700 nm.28 The reaction of FCO radicals with NO is expected to produce The formation and decay of FC(0)O radicals can be studied by either CO + FNO or the adduct FCONO, or both. FNO exhibits radiolyzing SFa/C0/02 mixtures and monitoring the increase in a structured UV absorption in the wavelength region 275-350 absorption in the visible. Figure 8 shows the observed transient nm,32.33as does FCO. A monitoring wavelength of 303 nm was absorption at 696.1 nm following the radiolysis of a mixture of chosen to measure the decay of FCO in the presence of NO 40 mbar of CO, 40 mbar of 02,and 920 mbar of SF6 (full dose, because at this wavelength the differencein the absorption cross UV path length = 80 cm). Control experimentswere performed sections of FCO and FNO radicals is large. Following the in which either SF6, SF6/02,or SFb/CO mixtures were radiolyzed radiolysis of mixtures of 960 mbar of SF6,40 mbar of CO, and separately. No absorption at 696.1 nm was detected. We ascribe 0.5-2.2 mbar of NO, a rapid initial increase in absorption at 303 the transient absorption shown in Figure 8 to the formation of nm was observed followed by a slow decay. The decay of FC(0)O radicals via reaction 8a. To map out the spectrum of absorptionfollowed first-order kinetics. The observed first-order FC(0)O radicals, the monitoring wavelength was increased in loss rate, kl*t, is plotted as a function of the NO concentration 0.2-0.5-nm steps over the range 694-697 nm. As shown in Figure in Figure 10. The y-axis intercept is not statistically significant. 9, the shape of the resulting spectrum was in good agreement Linear least-squaresanalysis gives k(FCO+NO) = (1.03 f 0.22) with that given in ref 28. X 10-12 cm3 molecule-1 s-l at 1000 mbar of SF6 diluent. To the To place our measured spectrum on an absolute basis, the best of our knowledge, there are no literature data with which initial rate of increase in absorption at 696.1 nm was combined with the calculated FC(0)02 loss rate ( ~ ~ s ~ ~ [ F C ( O ) Oand Z ] O ~ )to compare this result. Kinetic Study of the Reaction FC(0)Oa NO FC(0)O f the branching ratio kss/k8 = 0.831 to give u(FC(0)O). For the N02. The kinetics of reaction 2 were studied by monitoring the data shown in Figure 8 a value of u(FC(0)O) = 1.4 X l e 1 ' cm2 rate of increase in absorption at 400 nm (attributed to the molecule-1 is derived. As discussed in the previous section, because

-

/'

?

E

- i 1

-

Q)

E

-

+

-

+

7

+

-

Wallington et al.

2352 The Journal of Physical Chemistry, Vol. 98, No. 9, 1994

t

n

/ FC(O)O,+NO

I 0.0

0.5

1.0

2.0

1.5

2.5

3.0

[NO] , m b a r

--_

-b"to-"

Figure 10. Plot of the pseudo-first-order decay rate of FCO radicals in the reaction of FCO with NO (A)and the pseudo-first-orderappearance

-0.04 .

I

:

rate of NO2 in the reaction of FC(0)02 with NO (a),as functions of the NO concentrations.

-

0.05

0 7

$

0.04

0

n

-0.08 1

'

0

1000

2000

3000

4 IO

Time, ps

-10

0

10

20

30

40

Time, ps Figure 11. Transient absorption at 400 nm observed following the pulse radiolysis (full dose) of a mixture of 0.75 mbar of NO, 40 mbar of CO, 40 mbar of 02,and 920 mbar of SF6. The solid line is a first-order rise fit which gives klIt = 4.05 X lo5 s-l.

formation of NO2) following the radiolysis of mixtures of 0.3-1 .O mbar of NO, 40 mbar of CO, 40 mbar of 02, and 920 mbar of SF6. This method of measuring the kinetics of the reaction of peroxy radicals with NO has been used extensively in our laboratory and is discussed in detail elsewhere."J4 Figure 1 1 shows the results from a mixture with [NO] = 0.75 mbar. The smooth curve in Figure 11 is a first-order fit using the expression Abs(t) = (Ainf- C)[1 - exp(-k1s't)] C, where Abs(t) is the absorbance as a function of time, Ainfisthe absorbance at infinite time, kl8t is the pseudo-first-order appearance rate of N02, and C is the extrapolated absorbance a t t = 0. For the data shown in Figure 11, kist = 4.05 X 105 s-I. In all cases, the rise in absorption followed first-order kinetics. Control experimentswere performed in which SF6/CO, sFb/co/02, SF6/02, or just sF6 was radiolyzed; no change in absorption at 400 nm was observed. It seems reasonable toconclude that N02is the species responsible for the absorption change following radiolysis of SFs/C0/02/ N O mixtures. As seen from Figure 10, the pseudo-first-order rate constant, klst, increased linearly with [NO]. The y-axis intercept is not statistically significant. Linear least-squares analysis gives k2 = (2.5 f 0.8) X 10-11 cm3 molecule-' s-1. Errors are 2 standard deviations.

+

Figure 12. Transient absorptions at 696.1 (A), 288 (B), and 272 nm (C) following the pulse radiolysis of SF6/CO/O& mixtures; see text for details. The path length for the analysis light was 80 cm. There are three traces in panel A. The dotted smooth curve in (A) is that predicted using the chemical mechanism given in Table 2; see text. The two experimentaltraces shown in (A) are transient absorptionsat 696.1 nm observed in the presence and absence of 0 3 . The trace with the largest maximum absorbance was recorded in the absence of 03. In panels B and C the dotted lines are the behavior predicted using the chemical model given in Table 2 with k4 = 0,3 X 10-14,and 5 X 10-14cm3molecule-1 S I .

The increase in absorbance at 400 nm can be combined with the literature value of UNO* (400 nm) = 5.9 X lO-I9 cm2 molecule-' l8 to calculate the yield of NO2 in this system. The yield of NO2 in the five experiments given in Figure 10, expressed as moles of NO2 produced per mole of FC(0)02 radicals consumed, was 91 f 13%, suggesting that the majority of reaction 2 proceeds to give NO2 and, by implication, FC(0)O radicals. In this calculation allowance was made for loss of F atoms via reaction with 0 2 , and loss of FCO radicals via reaction with NO, using rate data measured in the present work. Kinetic Study of the Reaction FC(0)O 00. As discussed above, the self-reaction of FC(0)02 radicals is believed to give FC(0)O radicals as the major, if not sole, product.31 The selfreaction of FC(0)02 radicals can be used to prepare FC(0)O radicals in the presence or absence, of 0,.FC(0)O radicals can be readily monitored using their absorption in the visible; see Figure 9. The difference in the temporal behavior of FC(0)O radicals with, and without, O3provides information on the kinetics of reaction 3.

+

FC(0)O + 0,

-

products

(3)

Figure 12A shows the transient absorptionsat 696.1nm following the pulse radiolysis of 40 mbar of CO, 40 mbar of 02, and 920 mbar of SF6 with, and without, 0.26mbar of 0,.As seen from Figure 12A,themaximum FC(0)Oconcentrationobservedwhen O3is present is slightly less than that when O3is absent. However,

The Journal of Physical Chemistry, Vol. 98, No. 9, 1994 2353

Atmospheric Chemistry of FCO, Radicals there is little or no discernibledifferencein the decay of FC(0)O radicals in the presence, or absence, of 03.Experiments were also performed using a UV monitoring wavelength of 288 and 272 nm; see panels B and C of Figure 12, respectively. Ozone absorbs strongly at these wavelengths (m3(288 um) = 1.79 X 10-18 and ~ ( 2 7 nm) 2 = 6.92 X 10-18cm2 molecule-' 18), and the transientabsorption provides information on the temporal behavior of O3in the presence of FCO, radicals. As seen from Figure 12, there is a small decrease in the absorption at 272 and 288 nm following radiolysis of C0/02/03/SF6 mixtures. The effect is most pronounced at 272 nm with a change of 0.021 absorbance units; at 288 nm the change is 0.006 units. At both wavelengths the change occurs very shortly after the radiolysis pulse (within 50 ps), and subsequently there is little or no observable loss of Os. The change in absorption at 288 and 272 nm can be equated with an ozone loss of 9.6 X 10" and 8.7 X 1013cm-3, respectively. The initial ozone concentrations used in the experiments given in Figure 12B,C were 0.39 and 0.30 mbar. The initial F atom concentration employed was 2.69 X 1015 cm-3. In the present experiments there is competition for available F atoms between reactions 6, 11, and 15. F

+ CO

--*

FCO

F + 0, + M --* FO, F

(6)

+M

+ 0, -,FO + 0,

(11)

(15)

TABLE 2

Reaction Mechanism reaction rate constant" Base Model F + CO + M -.FCO + Me 1.8 X

-

+

F + 0 2 + M F02 M FCO FCO COF2 + CO FCO 0 2 + M FC(0)02 + M FC(0)Oz + FC(0)Oz FC(0)O FC(0)O 02 FC(0)02 FC(0)02 FC(O)OOC(O)F 0 2 FC(OjO.+ FC(Oj0 M-. FC(O)OOC(O)F + M FC(0)O FC(0)02 M FC(O)OOC(O)F M

+ +

+

+

+

+

+ +

+

+

FC(0)O + 03 F + 0 3 * FO

-+

-

+ +

+

products

0 2

1.9 X lW3b 1.6 X 1.2 X l0-I2

+ + +

-

+

+

29 30 this work

this work

4.8 X

this work, 31

1.2 X 1 0-12

this work, 3 1

7.0 X 10-13 8.0 X

b

this work this work

03 Chemistry d 1.0 x 10-11

NO, Chemistry F + NO M-.FNO M 5.5 X 10-12b FCO NO 4 FNO CO 1.0 X 10-12 FC(0)02 NO 2.5 X 10-" FC(0)O + NO2 FC(0)O NO FNO C02 d

+

reference

35

this work this work this work

+

Units of cm3 molecule-l s-I. Pseudo-second-order rate constant, value appropriatefor lo00 mbarof SF6. M represents a third body (SF6 in the present system). Varied, see text.

(O)OOOC(O)F at ambient temperature. It is possible that M represents a third body (SF6 in our case). Using k6 = 1.8 X reaction -14 occurs on the millisecond time scale of the present lO-12,29 kll = 1.9 X lO-'3,3Oand kls = 1.0 X 10-11cm3 molecule-' pulse radiolysisexperiments, and if so, then thevalue of kl3 derived s-1,35 the loss of O3through reaction 15 following the radiolysis from the base model given in Table 2 is a lower limit. In the of mixtures containing 40 mbar of CO, 40 mbar of 0 2 , and either absence of data concerning k-14 we choose to exclude this reaction 0.39 or 0.30 mbar of O3is expected to be 1.26 X 1014and 9.77 from the model. The aim of the present study is to measure the X 1013 ~ m - respectively. ~, The expected ozone losses in the reactivity of FC(0)O radicals toward 03.Inadequacies in the experimentsgiven in Figure 12B,C are then somewhat more than model in Table 2 caused by ignoring reaction -14 should have those experimentallyobserved. However, after considering the a negligible effect on conclusions relating to k(FC(O)O+O3). experimentaldifficulties associated with the measurement of the The lower limit of kl3 > 7 X l @ I 3 cm3molecule-' s-I obtained small absorption changes evident in Figure 12B,C, we conclude in the present work can be compared to the value of k13 = (2.3that the observed loss of ozone is indistinguishable, within the 6.5) X 10-12cm3molecule-'^-^ reported by Maricqet a1.28 While experimental uncertainties, from that expected from the above the twovalues areconsistent,this may be fortuitousas the chemical analysis. The fact that the O3loss evident in Figure 12B,C can mechanisms used to derive kl3 differ in two important respects. be entirely explained by reaction 15 shows that loss of Os by First, in the study of Maricq et a1.28reaction 14 was not included reaction with FC(0)O radicals (or indeed by any other process) in the mechanism. Second, to fit their data, Maricq et a1.28 found is negligible in the present work. To derive an upper limit for it necessary to include reaction -13 with k-13 = 700 s-1. As k3, the experimental traces given in Figure 12 were simulated discussed above, such a rate is inconsistent with the stability of using the CHEMSIMUL program36 with the mechanism given FC(O)OOC(O)F reported in ref 31. In both the present work in Table 2. and that of Maricq et al.,28 the FC(0)O radicals are formed The first task was to simulate the behavior of FC(0)O radicals indirectly in complex chemical systems containing FC(O)O* in the absence of Os. As shown in Figure 12A, the base model radicals. Further studies of the kinetics of reaction 13 are required given in Table 2 reproduces the observed behavior adequately. using more direct methods to produce FC(0)O radicals in the In this model k13 and k14 were varied to provide a satisfactory absence of FC(0)02 radicals, thereby removing complications fit to the data, k6 and kll were taken from previous ~ o r k ,and ~ ~ * ~caused ~ by reactions 14 and -14. k7,kl, and k8 were taken from results obtained in the present As shown in Figure 12A, the decay of the transient absorption work. Reactions 16, -13, and -14 were not included in the attributed to FC(0)O radicals is unaffected by the presence of mechanism but merit discussion. 03.This observation suggests that reaction 3 is not a significant loss process for FC(0)O radicals. Figure 12B,C shows that in FC(0)O F CO, (16) the time period from 0.1 to 4.0 ms after the radiolysis pulse the concentration of O3is essentially constant, and yet as evidenced in Figure 12A during this time there is a significant exposure of FC(O)OOC(O)F- F C ( 0 ) O F C ( 0 ) O (-13) the 0 3 to FC(0)O radicals. The dashed lines in Figure 12B,C show the expected behavior when reactions 3 and 15 are added FC(O)OOOC(O)F F C ( 0 ) O FC(O)O2 (-14) to the base model. As seen from Figure 12, the lack of any discernible 0 3 loss in the period 0.1-4.0 ms enables an upper At ambient temperature, FC(0)O radicals and the peroxide, limit of k3 < 3 X FC(O)OOC(O)F, have lifetimes with respect to thermal decm3 molecule-' s-l to be established. At this point we need to consider the impact that uncertainties in composition of at least 3 s and 800 min, re~pectively.~'Hence, u(FC(0)O) have on k3. The dotted lines in Figure 12B,C were reactions 16 and -13 can be excluded from the mechanism. derived assuming u(FC(0)O) = 1.4 X The absence of the trioxide, FC(O)OOOC(O)F, as a stable cm2 molecule-'. As discussed previously,this value is actually a lower limit; in principle product in the FTIR-Smog chamber study of the FC(0)02 selfu(FC(0)O) could be as much as a factor of 1.8 larger. If ureaction31 places an upper limit of 2 min for the lifetime of FC-

-+ -

+

+

The Journal of Physical Chemistry, Vol. 98, No. 9, 1994

2354

.. , .

0

5 ./ n

0.10

Z

0.06

0.08

-z n

i ,

1

h

t

I

-10

0

0.05

0.04

0.05

0

n

::

0.02

0.00 10

20

30

Time, ps

Figure 13. Transient absorptionsat 696.1 (A) and 310.5 nm (B) observed following the pulsed radiolysis (full dose) of mixtures of 40 mbar of CO, 40 mbar of 02,and 920 mbar of sF6 with 1.0 (A) or 0.5 mbar (B) of NO. The solid smooth curves are those predicted using the chemical mechanism given in Table 2; see text. The UV path length was 80 cm.

cm2 molecule-', then the exposure of

(FC(0)O) was 2.5 X

03 to FC(0)O radicals would be about one-half of that used to derive the dotted lines in Figure 12B,C. Hence, we chose to increase our upper limit by a factor of 2, giving k3 < 6 X cm3 molecule-' s-l. There have been no other studies of this reaction with which we can compare this result. Kinetic Study of the Reaction FC(0)O NO. The reaction of FC(0)02 radicals with N O produces FC(0)O radicals as the major (91 f 13%), if not sole, carbon-containing product and thus provides a convenient method of preparing FC(0)O radicals in the presence of NO. By monitoring the temporal behavior of FC(0)O radicals in the presence of NO, kinetic information regarding the reaction of FC(0)O with N O can be derived. Experiments were performed using the radiolysis of SFs/CO/ 02/NO mixtures with the transient absorption monitored at 696.1 and 310.5 nm. The temporal profiles of FC(0)O radicals were found to be significantly perturbed in the presence of NO. Experiments were performed with N O concentrations varied over the range 0.211.45 mbar. As the N O concentration was increased, the maximum in the transient absorption at 696.1 nm attributed to FC(0)O radicals was found to decrease and shift to shorter times. The rateofdecay of the absorption (and hence FC(0)Oconcentration) increased with increasing NO concentration. A comparison of the transient absorption given in Figure 13A (obtained in the presence of 1 mbar of NO) with that given in Figure 8 (obtained in the absence of NO) illustrates these effects. It is worth emphasizing that with the exception of the addition of 1.O mbar of NO in the mixture used to obtain the trace given in Figure 13A, all experimental conditions used to derive the traces shown in Figures 8 and 13A were identical. Clearly, additional loss processes for FC(0)O radicals are present when N O is added to the reaction mixture. There are three possibilities: FC(0)O radicals could react with NO, NO2 (produced in reaction 2), or both. To investigate the relative importance of reactions 4 and 17, the transient absorption a t 696.1 nm was simulated using the model given in Table 2.

+

FC(0)O FC(0)O

+ NO

+ NO2

-

-

FNO + C 0 2 FNO,

+ C02

(4)

(17)

Wallington et al. As shown in Table 2, the reactions of N O with F, FCO, FC(O)O, and FC(0)02 were included in the model. All rate constant in the model were fixed, while k4 was then varied to provide the best fit with the experimental data. The solid line in Figure 13A represents the predicted behavior using the mechanism in Table 2 with k4= 9.0 X 10-ll cm3molecule-l~-~ and providesa reasonable fit to the experimental data. The dotted lines in Figure 13A show the effect of varying k4 by f20%. Experiments were performed using initial N O concentrations in the range 0.2 11.45 mbar, and in all experiments the observed FC(0)O temporal profile was well described by the mechanism given in Table 2 using k4 = (1.0 f 0.3) X 10-l0 cm3 molecule-' s-l. The chemical mechanism given in Table 2 does not contain reactions of NO2 with FCO, radicals or with F atoms. Reaction of F atoms and FCO radicals with NO2 can be neglected because NO2 is only formed by the reaction of FC(O)O2 radicals with NO, and the concentrations of F atoms and FCO radicals have decayed to low levels before significant production of NO2 has occurred. The fact that the yield of NO2 following reaction of FC(0)02 with NO is 9 1 f 13%suggests that NO2 is not effectively scavenged by FC(0)02 radicals. There are two experimental observations that suggest that reaction of FC(0)O radicals with NO2 is not a major complicating factor in our data analysis. First, the value of k4 derived from fitting the observed transients at 696.1 nm to the mechanism in Table 2 was found to be independent of the initial NO concentration over the range 0.211.45 mbar at a constant initial radical concentration. With a constant initial radical concentration the NO2 produced via reaction 2 will also be constant. If reaction 17 augmented the loss of FC(0)O radicals, then values of k4 derived in the above analysis should appear to increase as the initial N O concentration is decreased; no such effect was observed. Second, the value of k4 derived in the present work is within a factor of 3 of the gas kinetic limit. The NO2 concentration present during the FC(0)Odecay was up to a factor of 14 times less than that of NO. For reaction with NO2 to account for a substantial fraction of the observed FC(0)O decay would require a rate constant k(FC(O)O+NO,) in excess of the gas kinetic limit which is not physically reasonable. While it is clear that the reaction of FC(0)Owith NO2 does not play a major role in the observed FC(0)Odecay, we cannot exclude a small contribution from this reaction. Accordingly, we choose to add an additional 10% uncertainty range to k4 to give k4 = (1.0 f 0.4) X 10-lO cm3 molecule-1 s-l. Finally, the impact of uncertainties in u(FC( 0 ) O )on kq needs consideration. The modeled traces in Figure 13A werederivedusingo(FC(0)O) = 1.4 X 10-17cm2molecule-l. As discussed previously, this value is actually a lower limit; in principle, a(FC(0)O) could be as much as a factor of 1.8 larger. To assess the maximum possible impact that uncertainties in o(FC(0)O) could haveupon thederived valuesof k4, themodeling procedure was repeated using a(FC(0)O) = 2.5 X 10-l' cmz molecule-l. With this larger absorption cross section, values of k4 of (1.5-2.0) X 10-l0 cm3molecule-' s-l were needed to fit the experimental traces. Toaccount for uncertaintiesin a(FC(O)O), we choose to quote kq = (0.6-2.0) X 10-lo cm3molecule-' s-l, Le., k4 = (1.3 f 0.7) X 10-lo cm3 molecule-1 s-l. There have been no other studies of this reaction with which we can compare this result. The chemistry following the pulsed radiolysis of SFs/CO/ NO/O2 mixtures was also probed at a monitoring wavelength of 3 10.5 nm. This wavelength was selected because it is where FNO absorbs strongly32J3 (u(FN0) at 310.5 nm = (4.73 f 0.66) X 10-19 cm2 molecule-1 (this work)), and by analogy to the reaction of C F 3 0 radicals with NO,37 we expect that FNO is a likely product of the reaction of FC(0)O radicals with NO. Figure 13B shows the transient absorption at 310.5 nm following the radiolysis (full dose) of a mixture of 40 mbar of CO, 40 mbar of 02,O.S mbar of NO, and 920 mbar of SF6. The UVabsorption

Atmospheric Chemistry of FCO, Radicals path length was 80cm. The transientabsorption has an interesting shape. There is an initial rapid rise which is complete within the first 0.3 ps. This is ascribed to the rapid formation of FCO radicals which absorb at 310.5 nm (u(FC0) at 310.5 nm = 1.1 X 10-18 cm2 molecule-' 21). Over the next 2 ps the absorption decays fairly rapidly, which can be ascribed to the conversion of FCO into FC(0)02 radicals via reaction 1. FC(0)02 radicals do not absorb significantly at 310.5 nm. At times greater than 2.5 ps the absorption slowly increases because of the reaction of FC(0)02 radicals with NO to produceNOzand FC(0)Oradicals. NO2 absorbs weakly at 310.5 nm (u = 1.9 X 10-19 cm2 molecule-' 18). FC(0)O radicals then react with NO to give FNO, which absorbs more strongly at 310.5 nm. The solid line shown in Figure 13B is the behavior predicted using the CHEMSIMUL program with the chemical mechanism given in cm3 molecule-' s-I). As seen from Table 2 (k4 = 8.0 X Figure 13B, the mechanism in Table 2 describes the observed transient absorption at 310.5 nm adequately. It should be noted that the reaction of FC(0)O radicals with NO is assumed in Table 2 to yield FNO and CO, as the sole products. The agreement evident in Figure 13B supports this assumption. Four experimentswere performed with NO varied over the range 0.31.0. In all cases the transient absorption at 310.5 nm was well described by the chemical mechanism in Table 2. The final absorption, when corrected for the contribution from N02, provides a yield of 78 k 4%for the yield of FNO (moles of FNO produced per mole of FC(0)O radicals lost) from reaction 4. Quoted errors reflect statistical uncertainties (2 standard deviations), and we estimate that potential systematic errors could add an additional 20% range. It is interesting to compare the kinetic and mechanistic data for the reaction of FC(0)O radicals with NO derived in the present work with the available literature data concerning the analogous reaction of C F 3 0 radicals with NO. While FC(0)O and C F 3 0 radicals both react with NO to give FNO as a major product,37 FC(0)O radicals react significantly more rapidly (k(CFjO+NO) 3 X 10-II cm3 molecule-' s-I 384. That FC(0)Oradicals are more reactive toward NO than CF30is probably a reflection of the relatively weak F-C bond energy in FC(0)O compared to C F 3 0 radicals.

-

Implications for Atmospheric Chemistry The results presented here substantially improve our understanding of the atmospheric chemistry of FCO, radicals. As demonstrated here and elsewhere,28 FCO radicals react rapidly with 0 2 to give FC(0)02 radicals. Using kl = 1 X 10-12 cm3 molecule-' s-l, we calculate a lifetime of 0.2 ps for FCO radicals with respect to conversion into FC(0)02 in 760 Torr of air. Reaction with 0 2 is the sole atmospheric fate of FCO radicals. We have shown that FC(0)02 radicals react rapidly with NO to give FC(0)O radicals and NO2. Using k2 = 2.5 X 10-" cm3 molecule-' s-l together with a background troposphericNO level of2.5 X 108cm-3, thelifetimeofFC(0)02withrespect toreaction with NO is 3 min. Reaction with NO is likely to be a major fate of FC(0)02 radicals. It has been shown in the present work that FC(0)O radicals react rapidly with NO. Using k4 = 1.3 X 10-Io cm3 molecule-' s-1 and [NO] = 2.5 X 108 cm-3, the lifetime of FC(0)O radicals with respect to reaction with NO is calculated to be 30 s. This lifetime is of the same order of magnitude as that for thermal decomposition of FC(0)O radicals into F atoms and C02.31941Under ambient conditions it is possible that both thermal decomposition and reaction with NO are important fates of FC(0)O radicals. From the viewpoint of establishing the atmosphericchemistry of FC(0)O radicals, the issue of whether thermal decomposition or reaction with NO dominates the loss of FC(0)O radicals is moot as the product of reaction with NO, FNO, will photolyze to give F atoms and NO. F atoms will abstract H atoms from hydrogen-containingcompounds (e.g., H20 or CH,) to give HF, which will then be rained out.

The Journal of Physical Chemistry, Vol. 98, No. 9, 1994 2355 In contrast to the reactivity of FC(0)O radicals toward NO, we have shown here that FC(0)O radicals react slowly (if at all) with 03.While the kinetic data reported here were measured at ambient temperature and.not at the low temperatures applicable to the stratosphere, the data presented serve as a guide to the relative importance of these reactions at lower temperatures. At 20 km the concentrations of O3 and NO are 7 X 10'2 and 5 X 108 cm-3 (24-h average), respectively.18 Multiplying the concentrations by the room temperature rate constants measured here (