Environ. Sei. Technol. 1983, 17, 94-99
(11) Spigarelli,S. A.; Thommes, M. M. J. Great Lakes Res. 1976, 2,99-110. (12) Spigarelli, S. A.; Thommes, M. M.; Prepejchal, W.; Gold-
stein, R. M. Environ. Biol. Fishes, in press.
(13) Spigarelli, S. A; Smith, D. W. In CONF-750425;Gibbons, J. W., Sharitz, R. R., Eds.; 1976; NTIS: Springfield,VA, pp 100-105. (14) Spigarelli, S. A. Health Phys. 1976, 30, 411-413. (15) Romberg, G. P.;Spigarelli, S. A.; Prepejchal, W.; Thommes, M. M. R o c . Conf. Great Lakes Res. 1974,17, 68-77. (16) Elliott, J. M. J. Anim. Ecol. 1975, 44, 805-821. (17) Gesser, H. D.; Chow, A.; Davis, F. C.; Uthe, J. F.; Reinke, J. Anal. Lett. 1971, 4 , 883-886. (18) U.S. Dept. Health, Education,Welfare “PesticideAnalytical Manual”; 1977; Vol. 1. (19) Griffiths, J. S. Ph.D. Thesis, University of Toronto, Toronto, Ontario, 1977. (20) Reinert, R. E.; Stone, L. J.; Willford, W. A. J. Fish. Res. Board Can. 1974, 31, 1649-1652. (21) Edgren, M.; Olsson, M.; Renburg, L. Ambio 1979, 8, 270-272.
(22) Boudou, A.; Delarche, A.; Ribeyre, F.; Marty, R. Bull. Enuiron. Contam. Toxicol. 1979, 22, 813-818. (23) Brett, J. R.; Shelbourn, J. E.; Shoop, C. T. J. Fish. Res. Board Can. 1969,26, 2363-2394. (24) Atherton, W. D.; Aitken, A. Comp. Biochem. Physiol. 1970, 36, 719-747. (25) Niimi, A. J.; Beamish, F. W. H. Can. J. 2001.1974, 52, 447-456. (26) Shul’man, G. E. “Life Cycles of Fish”;Wiley: New York, 1974. (27) Neely, W. B.; Branson, D. R.; Blau, G. E. Enuiron. Sci. Technol. 1974,8,1113-1115. (28) Veith, G. D.; DeFoe, D. L.; Bergstedt, B. V. J.Fish. Res. Board Can. 1979,36, 1040-1048. (29) Ribeyre, F.; Boudou, A.; Delarche, A. Ecotoxicol. Enuiron. Safety 1979, 3, 411-427. (30) Hamelink, J. L. Ann. Rev. Pharmacol. Toxicol. 1977,17, 167-177. (31) Sheffy, T. B. Wisc. Dept. Nat. Res. Report, 1977, 1.
Received for review February 1,1982. Accepted October 18,1982.
Correlation of the Ozone Formation Rates with Hydroxyl Radical Concentrations in the Propylene-Nitrogen Oxide-Dry Air System: Effective Ozone Formation Rate Constant Hajlme Akimoto* and Fumlo Sakamaki Division of Atmospheric Environment, The National Institute for Environmental Studies, P.O.Tsukuba-gakuen, Ibaraki 305, Japan Ozone formation rates obtained in smog chamber experiments for the C&gNO,-dry air system were analyzed with the aid of computer simulation using a detailed reaction model. In the region of [C3H6],/[N0,], 2 5, the ozone formation rate was found to be approximately proportional to the product of the OH-radical concentration and the initial concentration of C3H6in the earlier stage of photooxidation until d[03]/dt reaches a maximum. The proportionality constant was defined as an effective ozone formation rate constant and is proposed to be a useful parameter to represent photochemical reactivity of hydrocarbon mixtures on the basis of ozone formation rate. The effective ozone formation rate constant for C3Hs in the dry air-NO, mixture was determined to be (6.2 f 1.1) x io4 ppm-l min-’. Quantitative characterization of photochemical ozone formation in organics-nitrogen oxide-air mixtures is of critical importance in planning ozone control strategy based on smog chamber data and computer simulation. From this viewpoint, an “ozone formation potential” was proposed as one of the generalized reaction parameters in our previous studies (1-5). While the “ozone formation potential” governs the maximum amount of ozone formed ultimately after prolonged irradiation, the “ozone formation rate” is another even more important parameter which characterizes the photochemical ozone formation in organics-nitrogen oxide-air mixtures. The purpose of this study is to present a method for analyzing the ozone formation rate and to propose a phenomenological rate parameter that would be useful for representing reactivity of mixtures of hydrocarbons and other organics whose components are not necessarily known. As for the rate parameters that could be representative of ozone formation rate, the NO oxidation rate has been recognized as a useful measure of the photochemical re94
Environ. Scl. Technol., Vol. 17, No. 2, 1983
activity of various hydrocarbons and most extensively studied by Glasson and Tuesday (6). More recently, Darnall et al. (7) have proposed the OH-hydrocarbon reaction rate constant as a measure of hydrocarbon reactivity. However, although these parameters have been used for classification of numerous hydrocarbons based on a reactivity scale, no direct general relationship between these parameters and the actual ozone formation rate observed in a photochemical run of a selected organicsnitrogen oxide mixture has been proposed on an absolute rate basis. This study presents an analysis of ozone formation rate data for the propylene-nitrogen oxide-dry air system carried out in an evacuable photochemical smog chamber. An approximately proportional relationship between the ozone formation rate and the product of the maximum OH-radical concentration and initial concentration of propylene was found to hold. The relationship was further confirmed in terms of computer simulation using a detailed kinetic reaction model. Experimental Section and Computations All the experimental data used in the present study were obtained in C&-NO,-dry air runs by using the evacuable and bakable photochemical smog chamber at NIES. Experimental procedures and details of these runs have been reported previously (1). The experiments were carried out at 30 “C. The initial conditions and light intensity expressed as Itl (NOzphotolysis rate) are given in Table I. Typical wall loss rate of -0.04 ppm of ozone in the chamber was 0.07 f 0.01 h-l. The experimental maximum ozone formation rate, (d[03]/dt),mob*d,is defined as the maximum slope of the plot of ozone concentration vs. irradiation time. The experimental maximum OH-radical concentration, [OH]marobSd, was obtained from the maximum slope of the semi-log plot of the decay of propylene
0013-936Xl83/0917-0094$01.50/0
0 1983 American Chemical Society
Table I. Experimental and Calculated Ozone Formation Rates and OH-Radical Concentrations in the C,H6-NO,-Dry Air System
k run 1 2 3 4 5 7 8
s1
s2 10 s3 14 15 16 6 17 18 19 20 54 9 21 22 11 13 12 23 24 25 26 27
PPm 0.10 0.10 0.10 0.10 0.10 0.10 0.10 0.50 0.50 0.50 0.50 0.50 0.50 0.05 0.10 0.15 0.20 0.30 0.40 0.50 0.10 0.20 0.33 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50
(d[O31/dt)2,dI IO-, ppm min-'
PPm 0.009 0.020 0.026 0.034 0.036 0.052 0.063
0,010 0.020 0.045 0.150 0.187 0.291 0.038 0.043 0.039 0.040 0.039 0.039 0.040 0.086 0.086 0.091 0.090 0.090 0.089 0.085 0,090 0.083 0.088 0.089
0.16(" 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.16 0.37 0.31 0.25 0.19 0.13
0.12 0.30 0.33 0.37 0.41 0.38 0.35 2.31 3.40 2.48 0.13 0.35 0.71
1.00 1.40 1.74 0.19 0.78 2.10 2.97 2.59 2.50 7.27 6.25 4.38 3.50 2.82
0.38 0.43 0.42 0.41 0.39 0.38 0.33 0.84 1.46 2.37 2.78 2.58 1.84 0.16 0.38 0.64 0.92 1.45 1.90 2.22 0.30 0.77 1.73 2.86 2.65 2.60 5.26 4.60 3.94 3.22 2.47
[OH I$!, lo-' ppm
[oHl&&d,
lo-'
0.56 0.79 0.73 0.94 0.78 0.95 0.84 1.11 1.25 0.42 0.88 0.7 1 0.69 0.70 0.99 0.96 0.97 0.74 2.50 1.97 1.40 1.23 0.80
ppm
0.64 0.80 0.83 0.87 0.86 0.90 0.91 0.28 0.48 0.80 1.10
1.10 0.96 1.07 0.88 0.86 0.89 0.86 0.80 0.75 0.96 0.88 0.99 1.06
1.01 1.00 1.91 1.69 1.42 1.18 0.91
a Typical uncertainty in h , is 0.16 t 0.02 min-I.
after subtracting the decay of propylene due to ozone reaction. The rate constants used for the C3H6-OH and C3He-03 reactions were 2.51 X lo-'' (8)and 1.30 X lo-'' (9)cm3 molecule-' s-l, respectively. The data reductioh techniques used to obtain the maximum OH-radical concentration have been reported in our earlier paper (IO). The obtained [OH]-Obad for each run is cited in Table I. The estimated error in [OH],,Obsd is f25% for the runs with [C3H6], = 0.50 ppm. For lower [C3H6],runs, the error tends to get larger due to the larger scattering error in the C3& concentration measurement and a larger contribution of the correction term of C3H6-03 reaction. Computer simulations were performed for the C3H6NO,-dry air runs by using the reaction model reported before (4). The model consists of 158 chemical equations and 89 species and was used without modification. A wall loss rate of O3 of 0.058 h-' was used throughout the computation runs. Calculated maximum ozone formation rate, (d[03]/dt)c*'cd, and maximum OH concentration, were obtained directly from the computer output of these values. The integration program used was the same as described previously (4, 11). The initial conditions of the computer runs were the same as those of corresponding experimental runs. A few supplemental computer runs were also performed under initial conditions for which experimental runs were not available. Results Table I summarizes the observed and calculated maximum ozone formation rates and maximum OH concentrations for all the C3H6-N0,-dry air runs studied in this
work. The deviation of the calculated values from the observed ones is largest for the runs with high light intensity (runs 23 and 24). Although the agreement between the calculated and observed values is thought to be satisfactory except for these runs, it should be noted that the present reaction model predicts a much slower initial oxidation rate than observed experimentally. For example, the simulation for run 1 predicts the maximum ozone formation rate at 120 min, whereas the experimental run gives the maximum rate at 80 min after the irradiation began. Although this deviation could be due to the presence of unknown radical sources as discussed by Carter et al. (12,13), this problem was not pursued further in the present study since the following discussion on the relationship between (d[03]/dt),, and [OH],, will not be affected by the delay of photooxidation. Since it was generally found both experimentally and by computer simulation that the maximum ozone formation rate and maximum OH concentration are observed at nearly the same time after irradiation in most of the runs, it is expected that the (d[03]/dt),, is correlated to [OH],, quantitatively. Figure 1 shows a plot of (d[03]/dt),, vs. [OH],, for the runs with various [NO,], and k, but with the same initial concentration of propylene ([c3H6l0= 0.50 pprn). The linear plot implies that (d[03]/dt), is linearly proportional to [OH], even though (d[O,]/dt),, and [OH],, vary with the [NO,],, and kl. The proportionality was also confirmed by computer simulation as demonstrated in Figure 1. Deviation from the linear line can be noted for the run with a low [C&&/ [NO,], ratio. Thus, both the experimental and calculated points for run 15 are much lower than the linear line. This Envlron. Sci. Technol., Vol. 17,
No. 2, 1983 95
--IC
6-
E
6 -
m
'p 4 v
0.5
1.0
1.5
( l d 7 ppm
O [ H,I
2.0
2.5
0
04
02
IO
08
0.6
O [ H,]
12
[ C ~ H ~ I O(10-7PPrn2)
)
Figure 1. Plot of (d[O,]/dt),, vs. [OH],. Filled and open symbols are for observed and cafculated values, respectively; [C,H,], = 0.50 ppm. Variable [NO,], runs (A,A),k l = 0.16 mln-'; variable k , runs (0,0),[NO,], = 0.09 ppm.
Figure 3. Plot of (d[03]ldt),, vs. [OH],,[C3He],. Filled and open symbols are for observed and calculated values, respectively. Variable [C3H6],runs for [NO,], N 0.04 (w, 0)and 0.09 ppm (+, O ) , k , = 0.16 mln-I; variable [NO,], runs for [C,H,], = 0.10 (A,A) and 0.50 ppm (V, V);variable k lruns for [C3HeJ0 = 0.50, [NOXI,= 0.09 pprn (0,0). i
"
"
"
'
I
0
- I G
I
I
E
L 0
,
2
I
I
I
I
I
I
4
6
8
10
12
14
j 16
[ C ~ H ~ I ONOxlo
[OHlt
ppm)
Figure 2. Plot of calculated (d[03]/dt) vs. [OH], In the computer simulation for runs 10 (A,A), 11 (8,0),and 25 (W, 0).Filled and open symbols correspond to values before and after the d[03]ldt values reach the maximum.
is because the initial concentration ratio, [C3H6]0/[NO,], = 1.7, for run 15 does not fall into the hydrocarbon-excess region as will be discussed later. A proportionality between (d[O,]/dt), and [OH], within a single run was next checked by computer simulation. Figure 2 depicts examples for runs 10, 11, and 26. (d[O,]/dt), is, in general, proportional to [OH], until d[O,]/dt and [OH] reach their maximums, except during the very early stage of photooxidation. After d[O,]/dt and [OH] reach their maximums, the ozone formation rate decreases faster than the decrease of [OH]. For the runs with different [C3H6],, a plot of (d[03]/ dt), vs. [OH],,[C3H610 was attempted. Figure 3 shows the plot for all the runs with different [C3H6I0,[NOXI,,and kl. The plots of most of the runs except for those with a 1oW [C3H6],/[N0,], ratio fell on a single linear line going through the origin, verifying an approximate proportionality between (d[O,l/dt),, and [OHlrn,[C3H61~:
96
Environ. Sci. Technol., Vol. 17, No. 2, 1983
Figure 4. Plot of (d[0,]ldt),/([OH],,[C3He]o) vs. [C3H3lo/[NO,I0: filled symbols, experimental; open symbols, calculated; variable [NO,], runs for [C3HB],= 0.10 (A,A)and 0.50 ppm (V,V),k , = 0.16 min-'; variable [C3HB],runs for [NO,], = 0.04 ppm (B, 0)and 0.09 ppm (+, 0),k , = 0.16 min-l; varlable k , runs for [C,He], = 0.50, [NO,], N 0.09 ppm (0,0). Dashed line shows the ratio of (d[O,]/dt),,/ ([OH1,JC3HeIO) (see text).
where k e C ais the effective rate constant of photochemical ozone formation for propylene. From a least-squares fit to all of the experimental points, the value of k 3 H 6 is obtained to be 6.0 X lo4ppm-l min-' as the slope of Figure 3. The relationship can be confirmed by the computer simulation as demonstrated in Figure 3, and the calculated keCSH6 value agreed with the experimental value within 10%. In order to check the validity of k3H6,we plotted the experimental and calculated ratios (d[03]/dt)rnax/ ([OH],,[C,H,],) for all runs as a function of [C&,],/ [NO,], in Figure 4. The solid curve is drawn through the calculated data points. However, since it was found by the computer modeling that at a low [C3H6]o/[N0,]~ratio, time for d[O,]/dt and [OH] to reach their maximum does not coincide, but the latter is delayed; the ratio of (d[03]/dt),, to [OH]tmlu[C3H6]0 is also shown by a dashed curve. Here, [OH],, is the OH radical concentration at the time when d[O,]qdt reaches the maximum. The deviation of the two curves is apparent at a low [C,H&/ [NO,], ratio. As shown in Figure 4, both ratios decrease
as the [C3H6]o/[N0,]oratio decreases. Although the experimental points scatters appreciably, they also tend to decrease as the [C3H6]o/[NoX],ratio decreases in accordance with the prediction of the computer simulation. The region where the ratio gives nearly a constant value, which can be defied as k , w , can be called a hydrocarbon-excess region for the ozone formation rate. The hydrocarbonexcess region for the C3H6-N0,-dry air system may be defined as [C3H6],/[N0,], 2 5 from the plot of Figure 4, if one takes the 90% line of the calculated limiting value. The average of experimental data only in the region of [C3H6],/[N0,], 2 5 gives the observed k , value of 6.2 f 1.1 ppm-l min-' (2).
Discussion The importance of OH radicals in the reaction of photochemical air pollution is well recognized (14-21), and efforts have been made to correlate OH-radical rate constants with ozone formation in both experimental (23) and assessment studies (24). However, no direct correlation between the OH-radical concentration and ozone formation rate has been verified experimentally. The results of the present study shown in Figures 1-3 reveal that the ozone formation rate and the OH-radical concentration can be correlated in the hydrocarbon excess region by the equation (d[O,I/dt)t = ~eC3He[OHlt[C3H610
d[031/dt = k2[Ol[O~l- &[NO1 [O,] - CLi(O3) (3) and -d[NO]/dt = -k,[NO2] + k~[NOl[O3]+ C ~ [ R O Z I [ N O+I Ckj[RO][NO] (4) where k1, kz, and k3 are the rate constanta of the reactions NO, + hv-+ NO + 0 (i), 0 O2 + M O3 M (ii), and NO + O3--c NOz + Oz(iii),respectively. The Li(03)stands for every loss term of O3 except reaction iii, and ROz and RO represent all peroxy- and oxy-type radicals that can react with NO, respectively. At the time when the maximum ozone formation rate is observed, the inequalities d[03]/dt >> d[NOl/dt and Ckj[ROl[NOl
