Hydrogen peroxide and methylhydroperoxide distributions related to ozone and odd hydrogen over the North Pacific in the fall of 1991

. Hydrogen peroxide and methylhydroperoxide were measured in the troposphere over the western North Pacific as part of the airborne portion of NASA's Global Tropospheric Experiment/Pacific Exploratory Mission-West A field mission. The flights circled the North Pacific, focusing on the western Pacific, and extended from 300 to 13,000 rn altitude. The hydroperoxides were uniquely separated and quantified using a high-pressure liquid chromatography system in conjunction with a continuous enzyme fluorometric instrument. Results show a latitudinal gradient in both peroxides at all altitudes; for example, between 3 and 5 km, H20 2 median values decrease from 1700 to 500 parts per trillion by volume (pptv) in going from 0ø-15øN to 45ø-60øN, and the corresponding decrease in CH3OOH was 1100 to 200 pptv. Concentration maxima are observed in both species at altitudes of 2 to 3 km with H20 2 concentrations below 1 km lower by 30%, 10% for CH3OOH, and even lower, by a factor of 10, for both above 9 km. The H20 2 to CH3OOH ratio increased with altitude and latitude with ratios <1 in the tropical surface layer and >2 at midlatitude high altitude. Highest peroxide concentrations were encountered over the Celebes Sea in air which was impacted by aged biomass fire and urban pollutants. CH3OOH was below the level of detection in stratospheric air. H20 2 exceeded SO2 95% of the time, with the exceptions generally above 9 km. Above 3 km, 03 increases with decreasing H20 2 and CH3OOH. Below 3 km the O3-CH3OOH trend is the same but 03 increases with increasing H20 2. The measurements are compared with predictions based upon a photochemical steady state zero-dimensional model and a three-dimensional mesoscale time-dependent model. These models capture observed trends in H20 2 and CH3OOH, with the possible exception of H20 2 below 2 km where surface removal is important. A surface removal lifetime of 3.5 days brings the observed and zero-dimensional model-estimated H20 2 into agreement. The steady state model suggests a strong correlation between the ratios of NO/CO or HO2/HO and the ratio of H202/CH3OOH. The observed hydroperoxide ratios bracket the modeled relationship with occasionally much lower H20 2 than expected. tropical


Introduction
Hydrogen peroxide (H202) and methylhydroperoxide (CH3OOH) measurements were made from the NASA DC-8 aircraft throughout the western Pacific in the fall of 1991. These measurements were part of the suite of gas and aerosol chemical measurements which comprised the NASA Global  Kleinman, 1990]. The oxidative state of the atmosphere, determined by whether net ozone production is positive and further by whether there is an excess of peroxyl radicals to produce hydroperoxides, depends critically upon the relative concentration of NO to other constituents, namely, CO, CH4, and nonmethane hydrocarbons (NMHC) [Trainer et al., 1987;Lin et al., 1988;Kleinman et al., 1990;Jacob et al., 1993]. Observationally, the NO 2 to NO ratio under sunlight conditions has been used to qualitatively evaluate ozone production by how near the ratio is to that predicted assuming steady state among 03, NO, NO2, and NO 2 photolysis [e.g., Stedman and Jackson, 1975;Parrish et al., 1986;Carroll et al., 1990; Crawford et al., this issue; Davis et al., 1993]. A NO 2 to NO ratio higher than prescribed by simple photochemical steady state is suggestive of peroxyl radicals fueling 03 production.
However, before an air mass can be characterized in terms of its net 03 production, additional information is required concerning 03 sinks related to water vapor and odd hydrogen radicals [e.g., Davis et al., this issue; Liu et al., 1992]. The hydroperoxide measurements reported here provide insight on the activity of peroxyl radicals, complementing the NO-NO: hydrocarbon-based assessment of tropospheric 0 3 chemistry in the study region. The hydroperoxides are directly related to atmospheric odd hydrogen (sum of HO, HO2, and CH3OO). HO is the most significant oxidant and cleansing agent of the troposphere [National Research Council (NRC), 1984]. H202, CH3OOH, and nitric acid (HNO3) comprise odd hydrogen reservoirs with their photolysis returning odd hydrogen. The removal of the reservoir compounds through precipitation or dry deposition is one of the principal sinks of odd hydrogen. The other major sink is the reaction of HO with these same reservoir species [Logan et al., 1981]. Measurements of H202 which are greater than model predictions are suggestive of either higher HO 2 concentrations or slower removal and vice versa [Liu et al., 1992]. However, the sensitivity of modeled H202 to parameterizations of its removal, at the surface, in-cloud reaction, or by precipitation, obscures its utility near the surface as a constraint on odd hydrogen photochemistry in the lower troposphere [Heikes, 1992;Thompson et al., 1993]. However, H202 in conjunction with CH3OOH does place bounds on these same parameterizations. The lower solubility of CH3OOH in water Kok, 1986, 1994] makes it less dependent on precipitation and surface removal. H202 rates of production and loss and those for CH3OOH (lifetimes of the order of days) make them ideally suited to studies of diurnally averaged photochemistry in the middle to upper troposphere but somewhat less applicable to investigations of fast photochemical processes [Davis et al., this issue; J. Rodriguez et al., unpublished material, 1995].
Prior measurements of H202 and CH3OOH in the remote Pacific midtroposphere are at variance with photochemical theory. Liu et al. [1992] have discussed the contradictions in odd hydrogen radical chemistry posed by the MLOPEX 1988 H202, CH3OOH, and CH20 data of Heikes [1992], the nitric acid and nitrate data of Norton et al. [1992], and the NOx to NOy ratio [Hublet et al., 1992;Carroll et al., 1992]. Most significant was the lower than expected levels of CH20 and CH3OOH, both by factors of 2 or greater. H202 was also smaller than expected but by less than a factor of 2. It will be shown, by the PEM-West A data which follow and by the H202, CH3OOH, and CH20 data of Heikes et al. [1993] from MLOPEX 2, that the CH20 measurements remain at odds with models but that CH3OOH and H202 are more in accord with theoretical levels, as was the case for the equatorial Pacific surface layer [Thompson et al., 1993].
The second PEM-West A mission goal concerned the atmospheric sulfur cycle. The conversion of sulfur dioxide (SO2) and dimethylsulfide to sulfuric acid proceeds through reactions of SO2 with HO, 03, H202, and CH3OOH in the gas and aqueous phases [e.g., Penkett et al., 1978;Calvert et al., 1985;Bandy et al., 1992;Kriedenweis and Seinfeld, 1988]. Measurements of H202 and CH3OOH are directly coupled to aqueous sulfur chemistry and indirectly coupled to gas phase sulfur chemistry through their involvement with odd oxygen and odd hydrogen. Hence measurements of peroxides facilitated understanding the sulfur cycle over the western North Pacific.
In the following sections, the hydrogen peroxide and methylhydroperoxide data set is presented. The collection procedures and analytical methods are briefly described. Geographical distributions will be summarized. Hydroperoxides are interpreted with respect to photochemical theory using the results of photostationary steady state and time-dependent point models (Davis et al., [Lee, 1995], and with respect to a continuous diagnostic measure of air mass processing, the ethyne to carbon monoxide ratio [Smyth et al., this issue].

Experiment
Hydrogen peroxide and methylhydroperoxide were collected in aqueous solution using concurrent flow glass coils and analyzed using two different techniques: the continuous flow procedure ofLazrus et al. [1986] modified by Heikes [1992] and the high-pressure liquid chromatography (HPLC) method described by Lee et al. [1995]. Heikes [1992] describes a continuous flow-modulated serial coil method, with the only significant change for PEM-West A being an increase to four channels, which permitted both coils in series to be monitored continuously. In this manner the organic hydroperoxide concentration of the air could be assessed relative to hydrogen peroxide in three ways: using differential catalase enzyme activity, using differential solubility, and using the HPLC system.
The HPLC method was intended to identify and quantify the organic hydroperoxide species collected by the aqueous collection coils of Lazrus et al. [1986] and in so doing to verify the common assumption that only methylhydroperoxide be considered in interpreting the catalase enzyme channel in remote atmospheres. Collected samples were to be analyzed post flight for specific hydroperoxides (e.g., H202, CH3OOH, or hydroxymethylhydroperoxide). However, hydroxy-methylhydroperoxide and hydroxy-ethylhydroperoxide were found to decompose after an hour or several minutes, respectively, even when stored in an ice-water bath . Consequently, the HPLC analyses were performed in flight.
The HPLC system is relatively new and instrumental details may be found in the work of Lee et al. [1995]. A pH = 6 phthalate buffer solution was used to minimize artifacts due to O3 and SO2 during collection [Lazrus et al., 1986;Lee et al., 1995]. Analysis was performed immediately after collection (collection time was typically 5 min). Sample frequency was limited by the difference in elution times for H202 and CH3OOH. These were the only hydroperoxides detected by the HPLC system in PEM-West A. Other hydroperoxides would have had to exceed -100 parts per trillion by volume (pptv) to be quantitatively observed, based upon the detection limits given by Lee et al. [1995].
Air samples were brought into the aircraft through a forward facing inlet of 3/8 in. OD (1/4 in. ID) PFA Teflon. An excess of ultra-high-purity (UHP) zero air (liquid carbonics) flowed through the inlet during operation on the ground and at takeoff to minimize contamination of the inlet. The inlet was periodically cleaned with methanol and water. The air sample flow rate for the four-channel system was 3 standard liters per minute (slpm) (STP = 1 atm and 273 K). The aqueous collection solution flow rate was nominally 0.9 mL/min. The HPLC air sample flow rate was 2 slpm and aqueous collection solution flow rate was nominally 0.4 mL/min. Collection solution flow rates were calibrated volumetrically. The air sample flow rates were regulated using low-pressure drop mass flow controllers (MKS Instruments) and remained constant up to a pressure altitude of 10 km; by approximately 12.5 km the flow rates had decreased to 1.8 and 1.3 slpm for the four-channel and HPLC systems, respectively. The decrease in flow was caused by a decrease in pumping efficiency at low pressure.
Collection solution and water blanks were taken by passing UHP zero air through the inlet and collection coil or by diverting the air sample stream through a Hopcolite (Mine Safety Appliance) trap and then into the collection coil. H202 blanks can be appreciable and variable depending on water quality. A CH3OOH background in reagent waters has not been observed to date.
Calibrations were performed using at least four different aqueous concentrations of H202 and CH3OOH. Standards were prepared by serially diluting primary stock H202 and CH3OOH standards. The H202 stock was prepared by dilution of Ultrex 30% H202 (J. T. Baker). The primary H202 stock was standardized by titration using KMnO4 and by UV absorption spectroscopy [Miller, 1990]. CH3OOH was synthesized from dimethylsulfate (Aldrich Chemical Company) and H20 2 (J. T. Baker, 30%). The CH3OOH produced was standardized against the H202 primary standard using the continuous enzyme fluorescence system and by titration using an iodinestarch procedure [Lee, 1995]. The collection efficiency of the coils for H202 and CH3OOH was empirically determined by substituting aqueous standards and blanks for the collection solution in the coils. The H202 collection efficiency was always greater than 98%. The HPLC CH3OOH collection efficiency ranged from 50 to 65% and the serial coil CH3OOH collection efficiency was between 60 and 75%. Measured and theoretically derived collection efficiencies for CH3OOH [see Lee et al., 1995], agreed to within +10%. A corrected value of the Henry's law constant (1.3 times the value of Kok [1986, 1994] Figures 2 and 3). This was evident at all altitudes. There was a prominent maximum in H20: at altitude, typically 1-3 km, for all latitudes (Figure 2). CH3OOH maxima also occurred at lower altitudes, typically at 0.5-1 km (Figure 3).
The ratio of H202 to CH3OOH is shown in Figure 4. The ratio is plotted as a log•o value to give equal visual weight to ratios less than and greater than 1; for example, ratios of 0.5 and 2 are equally distant from a ratio of 1. There was a tendency for greater amounts of CH3OOH relative to H20 2 in the tropics than at higher latitudes. The latitude trend in the ratio occurred at all altitudes. CH,•OOH decreased relative to H20: with height. The smallest ratios of H20 2 to CH3OOH were found in the tropical boundary layer and the highest ratios were found at altitude in the 30ø-45øN latitude belt. The increased spread in the ratio at altitudes greater than 9 km results, in part, from vertical transport (discussed below) and from both species being nearer their respective detection limits at the highest altitudes.
Geographic distributions of the hydroperoxides are shown in  I  I  I   I  I  I  I  I  I  I  I  I  I  120 I  I  I  I  I  I  I  I  I  I  I  I   0 I  I  I  I  I  I  I  I  I  •  I  I  of H202 by precipitation. The highest H202 (>5000 pptv) and CH3OOH (>2500 pptv) concentrations for PEM-West A were observed just above the marine boundary layer over the Celebes Sea south of the Philippines (-125øE). Portions of data were missing from flights 15 to 19 out of Guam because of difficulties with the HPLC system. Gas bubbles formed in the reagents even after they were sparged with helium and allowed to sit for 12 or more hours. The NaOH reagent did not develop bubbles. The gas bubbles were thought to be due to CO2 outgassing after adding acid to the carbonaterich waters of Guam, after exposure to high temperatures (T > 30øC) during preflight testing, takeoff, and low-altitude sampling, and from the reduction in cabin pressure during ascent.

Discussion
The PEM-West A H202, CH3OOH, and ratio data presented above can be compared with prior measurements over the

Smyth e! al. [this issue] have presented an air mass characterization scheme based upon hydrocarbon ratios and Browell et al. [this issue] have characterized the atmosphere based upon aerosol backscatter, ozone, and potential vorticity [Newell et al., this issue (b)]. H202 was at or below our detection limit and CH3OOH was always below our detection limit in air that was classified as stratospheric air [Browell e! al., this issue]. North of 20øN, H202 was 370 pptv greater in continental outflow than in marine air, from the data presented by Gregory et al. [this issue] and Talbot e! al. [this issue
] for altitudes above the boundary layer. South of 20øN, H20 2 was 310 pptv higher in continental outflow than in aged marine air. The differences noted between continental and marine air for CH3OOH were smaller than those for H20 2 and were +90 and -70 pptv for the cases north and south of 20øN, respectively. The H202/ CH3OOH ratio was about 2.25 in continental outflow and is 1.8 in north marine air and 0.9 in south marine air. Smyth et al.
[this issue] presented the ethyne to carbon monoxide ratio, C2H2/CO , as a measure of air mass processing, be it through photochemical oxidation (principally HO) or the mixing of fresher air (continentally influenced air masses) with wellprocessed air. The higher the ratio the more recent an air mass was exposed to surface emissions. Their processing factor is fundamentally similar to the CaH8/C2H 6 ratio used by Kondo et al. [this issue]. Observed H20 2 was independent of the C2H2/CO ratio or processing factor. On the other hand, CH3OOH increased with atmospheric processing or decreased C2H2/CO ratio. Consequently, the H202/CH3OOH ratio also decreased with processing, consistent with the evaluated trends and, more importantly, interspecies ratios can be used to examine the chemical functionality of the models as they are related to hydroperoxide concentrations. The evaluation of complex photochemical models with direct field measurements is often confounded by model initial and boundary conditions and by the limited spatial and temporal extent of the data, but the characteristics of the model (e.g., 03, H202, or CH3OOH as a function of NOx and hydrocarbons or products of each other) can be explored by such data [e.g., National Center for Atmospheric Research (NCAR) 1986]. The ratio of NO 2 to NO, while clearly dependent upon 03, actinic flux and temperature, should also be dependent upon peroxyl radical concentrations on timescales of a few minutes to tens of minutes. H20, 2 and CH3OOH concentrations are also related to peroxyl radicals and through them to the NO2fNO ratio. Figure 8 shows the relationship between H20 2 or CH3OOH and the NO 2 to NO ratio for all altitudes and for altitudes greater than 3 km. The measurement trend line, labeled m, is computed using a local weighted smoothed scatterplot routine, LOWESS (Splus, Statistical Sciences Incorporated, Seattle). NO 2 data are taken from the model of Crawford et al. [this issue] for reasons given therein. A subset of the entire data set was selected using these criteria: NO > 3 times the standard deviation of its signal and solar elevation angle >30 ø . Two plots are given for all data meeting these criteria and for those data from altitudes greater than 3 km. From an examination of Figure 2 it was believed H202 data from above this height would be relatively free of the effects of surface removal and more clearly reflect photolytic sources and sinks. The range and trend line for the steady state point model predictions are given by the lines labeled u, 1, and t, respectively. Direct model-measurement point comparisons are presented by Davis et al. [this issue] and shown below. The model does a good job of capturing the data ranges and trends above 3 km for both H202 and CH3OOH and for CH3OOH at all altitudes. It overpredicts H20 2 at NO2/NO ratios >4 in the lower altitudes and this was considered to be the result of surface deposition. The underprediction of CH3OOH was most likely due to an underestimate of CH3OO , since H202 and HO 2 appeared to be well represented.
In the remote free troposphere the production and cycling of odd hydrogen between HO, HO 2 and CH3OO are dependent upon the spectral intensity of light, temperature, 03, H20, CO, CH4, and the reactive oxides of nitrogen (NOy). The production rate of HO is proportional to O3-to-O•D photolysis rate, 03, and H20 , and its loss is proportional to CO and CH 4. For the purpose of discussing HO, CH 4 in PEM-West A was effectively constant and [HO] -• will be proportional to the ratio [O3'H20/CO] -•. The reaction of HO with CO yields HO 2 and the reaction of HO with CH 4 yields CH3OO. The peroxyl radicals can react with NO or react with each other to produce hydroperoxides. The former gives rise to NO 2 and increased odd oxygen. Some odd hydrogen will be sequestered in HNO3 and some in hydroperoxides. Both the data and the models were expected to yield positive correlations between H202 or CH3OOH and the O3'H20/CO or O3*H20/NOy ratios. These relationships are shown in Figures 9 and 10 Figure 11. The modeled CH3OOH showed a better linear fit to the measurements (r 2 = 0.72) than did H202 (r 2 = 0.58) with, however, a tendency to more greatly underpredict its concentration, slope = 0.72. The H202 slope was 0.85. Slopes were calculated using the line of organic correlation method [Hirsch and Gilroy, 1984]. NO 2 was also underestimated by the model and this is discussed more fully by Crawford  Surface removal for H202 and CH3OOH was represented by a first-order loss rate. Model and measured H202 concentrations could be made equal by using a surface loss e-fold time of about 3.5 days. CH3OOH modeled and measured values could not be resolved by adjusting its surface removal rate, with measured values being approximately 20% higher than model predictions. One scenario to bring both species into better agreement would involve increasing model CH3OO, HO2, and HO while increasing the surface deposition of H202 relative to CH3OOH. Higher peroxyl radical concentrations would also bring the higher-altitude measurements and modeled values into better agreement. Heikes et al. [1995] have argued for H202 and CH3OOH deposition velocities of 0.8 and 0.6 cm/s over the South Atlantic which yielded surface deposition loss time constants of the order of 1-2 days, more rapid rates of loss for both species than indicated above for H202 alone.
The ratio of HO to HO 2 and the ratio of HO to CH3OO should be proportional to the ratio of NO to CO to a first approximation in a remote atmosphere. The resultant steady state ratio of H202 to CH3OOH may also depend upon the NO to CO ratio. These trends are examined in Figure 12, where a linear relationship is seen between the above species ratios and the ratio of NO to CO for steady state model conditions. Figure 12d shows the measured ratio of H202 to CH3OOH versus the measured NO to CO ratio and Figure 12e shows the peroxide ratio plotted against the modeled HO2/HO ratio. There is considerably more scatter in the measured H202/CH3OOH ratios than in the modeled ratios, Figures 12c and 12e. Most striking was the measured ratios less than 1 (log,() [ratio] = 0) with values approaching 0.1. None of the modeled ratios are below 1. One possible explanation for this involved vertical transport with H202 removal by precipitation or in-cloud chemistry.
The measurements of H202, CH3OOH , and their ratio provided insight on the recent processing of sampled air masses by cloud and precipitation. There are three possible fates for H202 when transported through cloud; removal by precipitation, reaction with SO2 in cloud particles, or transport through the cloud unaltered. The occurrence of high H202 above 5 km ( Figure 6c) and its increase relative to CH3OOH (Figure 6d) appeared to be associated with convective activity. Dibb et al.
[this issue] showed similar features in the near-Asia 21øpb, 7Be, aerosol, and acidic gases data. The enhancement in H202 was contrary to expectations based upon measurements in high SO2 environments (eastern North America), in which aqueous reactions between H202 and SO2 and precipitation removal would reduce peroxide levels during convection in cloud (B. G. Heikes, unpublished data, 1988). However, these findings were in agreement with a recent numerical cloud experiment using SO2 levels consistent with those reported for PEM-West A by Thornton  In subtle contrast, it can be seen from Figure 4 that above 3 km, hydroperoxide ratios less than 0.7 were found only at the higher altitudes, above 7 km. Transport alone should enhance both H202 and CH3OOH with their ratio expected to remain equal to the low-altitude value or to increase upon mixing with middle and upper level air containing relatively more H202 than CH3OOH. The SO2 data [Thornton et al., this issue] aided in interpreting the H202 data. If SO2 is present in air passing through cloud, it will react with H202 and reduce its concentration relative to CH3OOH. In the cases, in which H202 was depleted (<200 pptv) relative to CH3OOH (>250 pptv), SO2 concentrations were in excess of 100 pptv and at times near 300 pptv, concentrations which were near or in excess of the observed H202 concentration. In the earlier convective cases when H202 was elevated, CH3OOH was also elevated and SO2 was below 100 pptv and often below 50 pptv. This indicated H202 was well in excess of SO2 and a significant fraction of the H202 could have passed through cloud without depletion.
Additional indications on convective transport was gleaned from the oxides of nitrogen data. HNO 3 is soluble and will be removed from air which has experienced precipitation. Limited HNO 3 data were available at the times of the H202 cases of interest and it was necessary to invoke a surrogate measure of HNO 3 in order to examine these cases. The surrogate was calculated by subtracting the observed concentrations of NO2, NO, and PAN from NOy. When H202 was high, the HNO3 surrogate was between 300 and 500 pptv. Conversely, when H202 appeared to be depleted with respect to CH3OOH , the HNO 3 surrogate was between 200 and 300 pptv. For the few cases with measured HNO3, HNO 3 was 40 pptv when the H202/CH3OOH ratio was less than 1 and it was 40-300 pptv when H202 was high. Formic acid showed a trend similar to HNO 3 or its surrogate, but acetic acid concentrations were the same in both sets of cases.
The behavior of the soluble species and their less soluble cohorts suggested their future use as indicators of recent vertical motion with and without reaction and with and without precipitation. Deviation from photochemical steady state will be a transient phenomenon, and in the cases of H202 and CH3OOH they should return to steady state values within a couple of days as their photochemical lifetimes were both of the order of a day during PEM-West A. Hence their diagnostic utility will be temporally restricted to times less than this.

Conclusions
H202 and CH3OOH distributions have been presented for the North Pacific in the fall of 1991. Both species decreased with latitude. H202 exhibited a strong lower to middle altitude maximum, with concentrations decreasing with lower altitudes due to surface removal and decreasing with increasing altitude due to a decrease in radical precursors and H20. CH3OOH also showed a lower-to middle-altitude maximum with a slight decrease with decreasing altitude due to surface deposition or a reduction in source strength (HO titrated by hydrocarbons and nitrogen dioxide and with less HO produced due to lower ozone), and the decrease with increasing altitude being steeper relative to H202. CH3OOH concentrations were directly coupled to HO, whereas H202 is more strongly coupled to HO loss and the consequent formation of HO2.
Photochemical models (steady state point, time-dependent point, and three-dimensional time dependent), represented well the values, trends, and photochemical relationships between the hydroperoxides and other measured parameters, notably the oxides of nitrogen, nonmethane hydrocarbons, ozone, water, methane, and carbon monoxide. Exceptions were noted in the boundary layer where surface removal of H202 is important and in the upper altitude regions thought to be impacted by recent vertical transport. These regions where characterized by either enriched or depleted concentrations of H202 relative to CH3OOH depending upon whether SO2 was present at high levels or precipitation was inferred.