Observing the Transition From NO x -Limited to NO x -Saturated O 3 Production J. A. Thornton 1, P. J. Wooldridge 1, R. C. Cohen 1, M. Martinez 2, H. Harder 2, W. H. Brune 2, E. J. Williams 3, S. R. Hall 4, R. E. Shetter 4, B. P. Wert 4, B. Henry 4, A. Fried 4, F. E. Fehsenfeld 3 1 Department of Chemistry; University of California, Berkeley; Berkeley, CA 94720; 2 Department of Meteorology; Pennsylvania State University; 3 Aeronomy Laboratory, NOAA; Boulder, CO; 4 Atmospheric Chemistry Division, NCAR; Boulder CO Tropospheric O 3 concentrations are functions of the chain lengths of NO x (NO x  NO + NO 2 ) and HO x (HO x  OH + HO 2 + RO 2 ) radical catalytic cycles. For a fixed HO x source at low NO x concentrations, kinetic models indicate the rate of O 3 production increases linearly with increases in NO x concentrations (NO x -limited). At higher NO x concentrations, kinetic models predict ozone production rates decrease with increasing NO x (NO x -saturated). We present observations of NO, NO 2, O 3, OH, HO 2, H 2 CO, actinic flux, and temperature obtained during the 1999 Southern Oxidant Study from June 15 – July 15, 1999 at Cornelia Fort Airpark, Nashville, TN. The observations are used to evaluate the instantaneous ozone production rate (P O3 ) as a function of NO abundances and the primary HO x production rate (P HOx ). These observations provide quantitative evidence for the response of P O3 to NO x. For high P HOx (0.5 < P HOx < 0.7 ppt/s), O 3 production at this site increases linearly with NO to ~ 500 ppt. P O3 levels out in the range ppt NO, and decreases for NO above 1000 ppt. An analysis along chemical coordinates indicates that models of chemistry controlling peroxy radical abundances, and consequently P O3, have a large error in the rate or product yield of the RO 2 + HO 2 reaction for the classes of RO 2 that predominate in Nashville. Photochemical models and our measurements can be forced into agreement if the product of the branching ratio and rate constant for organic peroxide formation, via RO 2 + HO 2  ROOH + O 2, is reduced by a factor of Alternatively, these peroxides could be rapidly photolyzed under atmospheric conditions making them at best a temporary HO x reservoir. This result implies that O 3 production in or near urban areas with similar hydrocarbon reactivity and HO x production rates may be NO x -saturated more often than current models suggest. I. Introduction II. Measurements and Site Description: SpeciesMethodTotal Uncertainty* NO 2 Laser-Induced Fluorescence10% NOChemiluminescence10% O3O3 UV Absorbance<10% OHLaser-Induced Fluorescence~20% HO 2 Titration to OH by NO Followed by Laser-Induced Fluorescence ~5% H 2 COTunable Diode Laser Absorption Spectroscopy ~10% Solar Actinic FluxScanning Actinic Flux Spectral Radiometer 10% Relative Humidity and Temperature Commercial Probe5% * Includes both accuracy and precision The extensive suite of measurements made at Cornelia Fort Airpark (CFA) over the period June 15 – July 15, 1999 as part of the Southern Oxidant Study (SOS 99) provides one of the most detailed characterizations of an urban environment to date. CFA is located 8 km, NE of downtown Nashville, TN in the flood plain of the Cumberland River. The measured species used here, the methods used to measure them, and the reported uncertainties are shown in the table above. For the purposes of this study, all species were averaged to 1-minute intervals and none of the measurements required interpolation to this time base. III. Photochemical O 3 Production Eqn (2) J NO2 is the photolysis rate constant for NO 2 derived from solar actinic flux measurements. Reaction rate constants are taken from DeMore, et al, 2000, DeMore, et al., 1997, and Atkinson, V. The Dependence of P O3 on Primary Radical Production and NO Eqn (1) Eqn (6) Figure 1 ( left) shows 1-minute averaged P O3 calcuated using Eqn. 2 plotted versus Hour of Day. All of the data obtained at CFA is shown. VI. NO x -limited versus NO x -saturated O 3 Production at CFA IV. The Photo-Stationary State Assumption Is the atmosphere in steady state?: The closest large NO x sources were ~ 15 minutes away for typical wind speeds of 4-5 m/s. This is several e-folds in the NO x intra-conversion lifetime (~ 100 sec). The effect of potential surface emissions of NO on the stationary state were estimated by subtracting typical nighttime values of NO (~20-50 ppt) from the observed daytime concentrations. NO emissions lead to a potential bias of at most 5-10% in the P O3 calculated from the steady state assumption. Is the PSS-calculated P O3 accurate and precise?: The experiment at CFA provided some of the most accurate measurements of NO 2. The reported total uncertainty of the measurements combine to give a total uncertainty in P O3 of  34%. However, we note that for examining trends P O3 and our model of the crossover point, precision is more important than the total uncertainty or accuracy, and the precision of the calculated P O3 is approximately 10%. These conflicting results imply that the rate of peroxide formation (HHLoss) is either too fast or the rate of nitrate formation (NHLoss) is too slow, or some combination of the two. VII. Chemical Coordinate Analysis Figure 4 The observationally constrained ratio P HOx /L HOx is plotted versus P HOx (left), the fraction of HO x loss due to ROOH formation (center), and the fraction of HO x loss due to HNO 3 formation. In this initial model, we assume alkyl nitrate formation is negligible and set  = 0. From this analysis, it is apparent that the chemistry describing ROOH formation is in error by nearly a factor of 10. Figure 5 The panels show P HOx /L HOx with a new model of HO x loss plotted versus the same chemical coordinates as in Figure 4. Reducing ROOH formation by a factor of 10 and including 3% organic nitrate formation improves the radical budget and removes most of the trends of P HOx /L HOx versus chemical coordinates observed in Figure 4. We use this improved model to calculate the crossover point between NO x - limited and NO x -saturated O 3 production below in Figure 6. Figure 6 (right) shows the two fractions, HHLoss/L HOx and NHLoss/L HOx plotted versus NO where we have reduced the organic peroxide formation rate by a factor of ~ 10, and assumed a 3% organic nitrate yield. As opposed to the results shown in the lower right panel of Figure 3, the new model suggests a crossover at approximately ppt NO. This result is more consistent with the P O3 derived from observations. OHRO 2 RO HO 2 NO RH O2O2 NO RORO O3O3 NO 2 RO 2 HO 2 h O O2O2 O 3 + h H 2 CO + h HO x Cycle NO x Cycle H2OH2O O2O2 Figure 2 (left) shows P O3 for three consecutive days. While there is a clear diurnal trend with rates peaking near 12pm, there is evidence of significant day-to-day variation and variation on the time scale of minutes to hours as well. HO 2 + HO 2 + M H 2 O 2 + O 2 HO 2 + RO 2 ROOH + O 2 HO 2 + OH H 2 O + O 2 HO x -HO x NO 2 + OH + M HNO 3 RO 2 + NO RONO 2 M RO + NO 2 not terminating HO x -NO x The dependence of P O3 on NO x arises from the competition between two categories of chain termination reactions: Implications for RO 2 +HO 2 Chemistry, any combination of: k HO2+RO2 10x’s Smaller RO 2 + HO 2 ROOH + O 2 A RO 2 + HO 2 ROOH + O 2 RO + OH + O 2 B RO 2 + HO 2 ROOH + O 2 ROOH + h RO + OH C Isoprene-type ROOH NOT Scavenged 10% 90% Figure 3 The upper panels, illustrate the predictive behavior of this chemical system in a simple model. The Left panel shows P O3 versus NO for three different regimes of P HOx The right panel shows P O3 (red), and the rates of HHLoss (blue) and NHLoss relative to the total plotted versus NO. A wide range of parameters and conditions were used in the model. The point where the two fractions HHLoss/L HOx and NHLoss/L HOx were equal always occurred at an NO concentration that was ~ 25% less than that where the peak in P O3 occurred, however, the model is not expected to reproduce the observations in an absolute sense. The lower panels show the same quantities as the upper panels but now derived from the observations. P O3 derived from observations exhibits, qualitatively, the expected behavior as a function of P HOx and NO (lower left). However, the two fractions HHLoss/L HOx and NHLoss/L HOx (lower right) are equal at approximately ppt NO, a higher concentration than where P O3 peaks. This conflicts with the expected behavior shown in the upper right panel. L HOx = HHLoss + NHLoss VS P HOx [HO x ] P O3 The role of the HO x production rate, P HOx, on P O3 for constant NO x : We calculate P HOx using observations and Equation 6. Examine the balance between P HOx and L HOx P HOx /L HOx ~ 1  HOx ~ 30sec VIII. Conclusions and Implications [O 3 ] are the integral of P O3 (and L O3 ) Our new model of HO x loss predicts that O 3 production will be NO x -saturated more often than current models predict. Acknowledgements NASA Earth Systems Science Graduate Fellowship NOAA Office of Global Programs Questions for the Future Is Nashville comparable to other urban areas? VOC/NO x RO 2 /OH HO x -HO x vs. HO x -NO x Do RO 2 + HO 2 conclusions depend on identity of R? ?? isoprene-RO 2 ?? How to extrapolate to larger spatial scales? old new Eqn (5) Eqn (3) Eqn (4)