In Situ Measurements of Hydrogen Peroxide, Nitric Acid and Reactive Nitrogen to Assess the Ozone Sensitivity in Pingtung County, Taiwan

This study was aimed to investigate the concentration of hydrogen peroxide (H2O2), nitric acid (HNO3) and reactive nitrogen (NOy) at three sites in Southern Taiwan during 2003–2004. The mean H2O2 concentrations were 1.0–2.7 ppbv, 1.6–2.8 ppbv and 1.0–2.6 ppbv at Pingtung, Chao-Chou and Kenting, respectively. Meanwhile, the mean HNO3 concentrations were 2.7–5.5 ppbv at Pingtung, 1.9–4.6 ppbv at Chao-Chou and 1.2–3.0 ppbv at Kenting. The NOy mean concentrations were 61.0–85.0 ppbv, 16.0–75.0 ppbv and 4.2–12.0 ppbv at Pingtung, Chao-Chou and Kenting, respectively. The hourly data of H2O2, HNO3, NOy and O3 were then used to determine the ratio of H2O2 to HNO3, O3 to HNO3, O3 to NOy and the extent of reaction (E) in Smog Production Model (SPM) to evaluate the NOx and VOC sensitivity of ozone formation in Southern Taiwan. The results show that Pingtung city was primarily VOC-sensitive regime in four seasons. At Chao-Chou, the percentages of VOC-sensitive regime exceed those in NOx-sensitive regime in H2O2/HNO3 and O3/NOy in four seasons; but O3/HNO3 indicates NOx-sensitive regime in autumn, winter and spring. Kenting was mainly dominated by NOx-sensitive by use of three indicators in four seasons. The SPM results are consistent with those determined by the photochemical indicator ratios although the percentages of dominance may vary among these analyses.


INTRODUCTION
Ground-level ozone is a secondary pollutant produced from its precursors of nitrogen oxides (NO x ) and volatile organic compounds (VOC) via complex photochemical reactions in sunlight.Former studies have shown that the relationships between ozone and its precursors, VOC and NO x , are non-linear (Liu, et al., 1987;Milford et al., 1994;Sillman, 1999).The ozone-VOC-NO x sensitivity can be determined directly from measurements and from models.(Peng et al., 2006;Kumar et al., 2008;Tseng et al., 2009a;Tseng et al., 2009b;Lee et al., 2010) However, threedimensional photochemical model simulations for evaluations of ozone-VOC-NO x sensitivity are difficult because of the uncertainties of model assumptions, e.g., emission inventory and meteorology (Fujita et al., 1992;Sillman, 1995;Bishoi et al., 2009).Several alternative approaches have been developed to determine the ozone sensitivity.One approach is to use photochemical indicator species or species ratios, such as hydrogen peroxide and reactive nitrogen species (Milford et al., 1994;Sillman et al., 1997).Comparisons between model predictions and measured values for the indicator species would also provide a test of the accuracy of model sensitivity predictions (Sillman, 1995, Peng et al., 2006).Another approach is to use the ambient data of ozone and reactive nitrogen oxides to estimate the extent of reaction, the so-called smog production model (SPM), to predict whether ozone formation is VOC-or NO x -sensitive (Johnson, 1984;Chang and Suzio, 1995;Chang et al., 1997;Blanchard, 2000;Blanchard and Fairley, 2001).
Atmospheric chemistry of H 2 O 2 and its contribution to the formation of free radicals have been studied intensively over the pass years (Sakugawa and Kaplan, 1990;Das and Aneja, 1994;Jackson and Hewitt, 1996;Sauer et al., 2003).The major routes to H 2 O 2 are: (a) recombination of HO 2 • (Kleinman, 1991;Clavert and Stockwell, 1983), (b) reaction of ozone and biogenic hydrocarbons like isoprene and isobutane in water droplets, without radicals (Sauer et al., 1999).HNO 3 is an odd nitrogen trace specie and the end product of reactive nitrogen compounds such as NO, NO 2 , and N 2 O 5 .In the atmosphere, HNO 3 may originate from (a) the homogeneous reaction of NO 2 with •OH, (b) Hydrolysis of N 2 O 5 and (c) hydrogen abstraction by NO 3 • from aldehydes and hydrocarbons.Reactions a and b constitute the dominant route to HNO 3 during the daytime and night time, respectively (Kitto and Harrison, 1992).
Pingtung County is in the southern end of Taiwan with around 0.91 million inhabitants and an area of 2,775 km 2 (Fig. 1).It is mainly agricultural, with touring and sightseeing (e.g.Kenting town).The air pollutants of Pingtung County most come from traffic sources (Lin et al., 2005;Lin et al., 2007;Lin et al., 2008;Lin et al., 2009) and some densely populated areas in Pingtung city (Northern part) and Chao-Chou town (Central part), with several small-sized industrial parks.Pingtung County belongs to Kao-Ping air basin which has the worst air quality, in Taiwan.This is mainly because the northern and central parts of Pingtung County are south of, and thus downwind of, the Kaohsiung areas (Kaohsiung City and Kaohsiung County), whenever a northerly or north-easterly wind prevails, such as in autumn and winter (Chen et al., 2003(Chen et al., , 2004)).Photochemical simulations by Chen et al. (2003) showed that around 49-57% of the ambient ozone in Pingtung County were transported from and/or contributed by Kaohsiung area, depending on the season.The simulations also showed that most of concentrations of the ambient ozone in Kaohsiung City and Pingtung City were sensitive to the reduction of VOC emissions.However, no direct measurement has been made to evaluate the ozone-VOC-NO x sensitivity in southern Taiwan.
This work measures the concentrations of hydrogen peroxide, nitric acid and reactive nitrogen in the ambient air in Pingtung County in four seasons during 2003-2004.The three photochemical indictor ratios were calculated and the extent of reaction in the SPM was estimated.The VOC-or NO x -sensitive regime to the formation of ozone was identified in each region.

Sampling Sites and Periods
Three sampling sites, Pingtung, Chao-Chou and Kenting in Pingtung County were chosen because they had airquality monitoring stations operated by the Taiwan-EPA (Environmental Protection Administration).Therefore, hourly air quality and meteorological data, on O 3 , NO x , temperature, and/or wind, were available.The distances between the monitoring sites are approximately 18 km between Pingtung and Chao-Chou and 70 km between Chao-Chou and Kenting (Fig. 1).
Samplings were taken on five days in four seasons in 2003 and 2004.Samples of H 2 O 2 , HNO 3 and NO y were collected concurrently for eight 1-h periods between 09:00 and 17:00 at Pingtung, Chao-Chou and Kenting in Pingtung County.Table 1 presents the meteorological conditions at the sampling sites, including temperature, wind speed, relative humidity and period of sunshine.

Sampling Methods H 2 O 2
Fig. 2(a) shows the H 2 O 2 sampling apparatus.Gaseous samples were collected in two identical midget fritted glass bubblers , each containing 15 mL of TiOSO 4 collecting solution.The inlet of the first bubbler had two ports: one port was connected to the inlet of the second bubbler, and the other was connected to the outlet  8.9 8.9 67.3 64.9 64.9 Note: 1. Autumn samplings in Chao-Chou were made on 20-24 November in 2003 due to the EPA monitoring site in Chao-Chou was undergoing maintenance on 9-13 November in 2003. of an air sampler, via silica gel tubes.Desiccants were packed in the tube between the first bubbler and an air pump (SKC Model 210-1000s) operated at a flow rate of 500 mL/min.The sampling and analytical procedures complied with US-OSHA (Occupational Safety and Health Administration) Method VI-6 (OSHA, 1978).The Ti-H 2 O 2 solution was analyzed in the laboratory using a spectrophotometer (Shimadzu UV-160) at a wavelength of 410 nm.No breakthrough of the species occurred in the test tubes.Calibrations were conducted using seven concentrations of standard samples, with a coefficient of determination, R 2 , of above 0.997.The detection limit was 0.12 ppbv.

HNO 3
HNO 3 sampling apparatus is shown in Fig. 2(b) Gaseous samples were collected using silica gel adsorption tubes (SKC No. 226-10-03) of length 11 cm and outer diameter 7 mm, containing a 400-mg front section and a 200-mg backup section of washed silica gel, with flamed-sealed ends with plastic caps.The adsorption tube was connected to a pump (SKC No. 210-1002MH) operated at an airflow rate of 250 mL/min.No breakthrough of the compounds occurred in the tubes.The sampling and analytical procedures complied with US-NIOSH (National Institute for Occupational Safety and Health) Method 7903 (NIOSH, 1994).The samplers were analyzed in the laboratory using ion chromatography (DIONEX Model DX-100) with an AS-4ASC separator column of 4 mm.Calibrations were conducted using seven concentrations of standard samples, with a coefficient of determination, R 2 , of above 0.997.The detection limit was 1.55 ppbv.

NO y
NO y sampling apparatus is shown as Fig. 2(c) NO y refers to total reactive nitrogen including NO, NO 2 , HNO 3 , HNO 2 , PAN and alky nitrates.Ambient NO y was converted to NO via molybdenum catalyst converter at 350°C with low influence from NH 3 or other species (Williams, 1995;Honrath and Jaffe, 1990;Fehsenfeld et al., 1987).The instrument detection limit was 0.1 parts per billion volume (ppbv) for both NO and NO y (Delany et al., 1982;Dickerson et al., 1984).The ground molybdenum was put in a glass tube (length = 15 cm, D = 3 cm) with quartz wool in both end and heated to 350°C by temperature controller.Inlet NO y will be converted to NO under the molybdenum catalysis.Outlet NO gaseous was connecting to an ambient NO detector (Ecotech, 1070).The convertibility of NO y to NO in this study is 93.8%.Blanchard et al. (1999) reformulated the algorithm to compute the "smog produced" (SP), originally proposed by Johnson (1984) according to two alternative equations: where DO 3 is the accumulated deposition losses of ozone.Furthermore, the maximum potential SP is given by

Data Analysis Using Smog Production Model
The extent of reaction, E(t)  SP/SP max , can thus be obtained by or, (5) In evaluating E, the instantaneous data of O 3 (t), NO(t), and/or NO x (t) must be measured.Also, those unmeasured quantities, including DO 3 (t), O 3 (0), NO(t), NO x (i), and others must be estimated.Detailed descriptions on estimating relevant parameters are given in Blanchard et al. (1999), in which the default values were used here.
Furthermore, the intermediate values of E = 0.6 -0.8 used in Blanchard and Stoeckenius (2001) were adopted here such that a value of E lower than 0.6 indicates a VOC-sensitive regime, while a value of E higher than 0.8 indicates a NO x -sensitive regime.Table 2 statistically summarizes the H 2 O 2 , HNO 3 and NO y concentration in mean, maximum, minimum and standard deviation (S.D.) at three sites over the entire study period.The mean H 2 O 2 concentrations were 1.0-2.7 ppbv, 1.6-2.8ppbv and 1.0-2.6 ppbv at Pingtung, Chao-Chou and Kenting, respectively.According to the S.D concentrations, we can realize that the H 2 O 2 concentration varies slightly in four seasons.Peng et al. (2006)   1.40-1.76ppbv at Pingtung and Chao-Chou in Pingtung County.The present results are in the range of the reported ambient H 2 O 2 concentrations of 0.1 to 3.5 ppbv.For example, the concentration of H 2 O 2 ranged in 0.1 to 0.3 ppbv in downtown Raleigh of North Carolina (Das and Aneja, 1994), at a forest site in Portugal (Jackson and Hewitt, 1996), and in downtown Seoul, Korea (Kang et al., 2002), whereas the concentrations of other pollutants such as O 3 and NO x were also much lower than those here.Additionally, the present results are close to 0.86-2.04ppbv observed in Los Angeles (Sakugawa andKaplan, 1989 and1990).

Seasonal
During the whole measurements, HNO 3 concentration ranged from 2.7-5.5, 1.9-4.6 and 1.2-3.0 at Pingtung, Chao-Chou and Kenting, respectively.The concentrations of gaseous HNO 3 here are lower than 6.11-7.79ppbv in Kaohsiung City and 3.78-4.92ppbv in Pingtung County (Peng et al., 2006) but much higher than 1.55 ppbv at a rural site in Pennsylvania (Buhr et al., 1990), 0.67 ppbv in the central Piedmont region of North Carolina (Aneja et al., 1996), and 0.02-2 ppbv in the upper troposphere above Ellington Field in Texas (Neuman et al., 2001).
NO y mean concentrations were relatively high (all above 61.0ppbv) at Pingtung and Chao-Chou, but were low at Chao-Chou during the summer times (16.0 ppbv) and at Kenting in four measurement periods (4.2-12.0ppbv).This may due to Pingtung is a sub-urban city with more vehicles exhausts than that at Chao-Chou and Kenting.The range of NO y concentrations was 14.2-136.0ppbv, 6.0-142.0ppbv and 2.3-38.0ppbv at Pingtung, Chao-Chou and Kenting, respectively.NO y concentrations, that are similar to other air pollutants in Kenting, were low in four seasons because of  (Aneja et al., 1997) but higher than 0.56-9.29 ppb in eastern Canada (Zhang et al., 2008).The present results are somewhat lower than 156-222 pptv in the Azores Island, Portugal where the mean NO x concentrations were ranged from 26-37 ppbv during the measurement period (Martin et al., 2008).Table 3 shows the results of H 2 O 2 , HNO 3 and NO y measurements found in the literatures together with the results from this paper.These ozone relative species were then used to evaluate the sensitivity of ozone formation in the ambient air, namely photochemical indicators, discussed in the coming section.

Ozone Sensitivity Determined by Indicator Ratios
The ranges of H 2 O 2 /HNO 3 , O 3 /HNO 3 and O 3 /NO y based on the present measurements were used to assess the conditions of NO x -sensitive and VOC-sensitive regimes to ozone formation.Table 4 summaries the results of NO x -sensitive and VOC-sensitive regimes to ozone at the four sites in four seasons.Here, the threshold ratios of H 2 O 2 /HNO 3 = 0.3-0.6,O 3 /HNO 3 = 12-16 and O 3 /NO y = 6-7 were adopted (Sillman 1995;Sillman et al., 1997), in which a lower value than the threshold ratio indicates a VOC-sensitive regime, a higher value than the threshold ratio indicates a NO x -sensitive regime, and otherwise indicates an intermediate (or a transition) regime.
Table 4 shows the percentages of VOC-sensitive and NO x -sensitive regimes to ozone formation at the three sites in four seasons in 2003 and 2004, based on measurements of photochemical indicators.Pingtung city was primarily VOC sensitive regime in four seasons by the percentage of 44.6-72.2%,60.7-88.7%and 100% in H 2 O 2 /HNO 3 , O 3 /HNO 3 and O 3 /NO y , respectively.At Chao-Chou, the percentages of VOC-sensitive regime exceed those in NO x -sensitive regime in H 2 O 2 /HNO 3 and O 3 /NO y in four seasons; but O 3 /HNO 3 indicates NO x -sensitive regime in autumn, winter and spring.Kenting site was mainly dominated by NO x -sensitive by use of three indicators in four seasons.Notably, Pingtung City is an urban area and close to the pollution sources in the Kaohsiung area, whereas Chao-Chou and Kenting are rural regions that are farther away from the source regions.Therefore, the present results are consistent with the fact that freshly emitted pollutants are typically characterized by VOCsensitive chemistry (such as in Pingtung City) and evolve towards NO x -sensitive chemistry (as in Chao-Chou town and Kenting town) as the air mass moves downwind (Sillman, 1999;Peng et al., 2006).

Comparisons of Ozone Sensitivity between Indicator Ratios and SPM Results
Fig. 4 illustrated the SPM result which shows VOC sensitive, intermediate, and NO x -sensitive regimes at the four sites in summer, autumn, winter, and spring.The VOC-sensitive regime dominated at Pingtung in all four seasons.Chao-Chou was primarily VOC-sensitive regime except for summer season.The percentages of NO xsensitive occupy more than 94.65% at Kenting in all the analyzed episodes.Castell et al. (2009) applied MM5-CAMx model system to explore the relationships between various photochemical indicators and ozone sensitivity.Results show that O 3 /NO y and E have similar behavior on evaluating ozone sensitivity, i.e., E parameter shows good correlation in VOC sensitive conditions (Blanchard and Stoeckenius, 2001;Sillman and He, 2002;Castell et al., 2009).In this study, the SPM results are consistent with those determined by the indicator ratios, discussed in the preceding section, although the percentages of dominance may vary among these analyses.

Fig. 1 .
Fig. 1.Locations of the three measurement sites in Pingtung County, Southern Taiwan.
Variations of H 2 O 2 , HNO 3 and NO y ConcentrationsFig.3shows the box plots of measured ozone sensitive species concentrations, namely H 2 O 2 , HNO 3 and NO y , at Pingtung, Chao-Chou and Kenting in four seasons.The box plots indicate the percentiles (5th, 10th, 25th, 50th, 75th, 90th and 95th) and means of the data.The distance between 25th and 75th in the box plot represents the data divergence.In addition to spring, H 2 O 2 concentration trends show Chou-Chao > Pingtung > Kenting in summer, autumn and winter.Typically, H 2 O 2 is lower inside in an urban plume relative to the rural area(Weinstein-Lloyd et al., 1998).The concentration trends of HNO 3 are consistent with the NO x , show Pingtung > paperChao-Chou > Kenting in four seasons.Summertime box plot shows NO y concentration at Pingtung was much higher than that in Chao-Chou and Kenting.

Fig 4 .
Fig 4. Percentages of VOC-sensitive and NO x -sensitive regimes to ozone formation at Pingtung, Chao-Chou and Kenting in 2003 and 2004, based on SPM results 2-3.0 ppbv at Kenting.The NO y mean concentrations were 61.0-85.0ppbv, 16.0-75.0ppbv and 4.2-12.0ppbv at Pingtung, Chao-Chou and Kenting, respectively.The hourly data were then used to evaluate VOC and NO x sensitivity of ozone formation.The threshold values of H 2 O 2 /HNO 3 = 0.3-0.6,O 3 /HNO 3 = 12-16 and O 3 /NO y = 6-7 indicate that Pingtung was dominant by VOC-sensitive in four seasons.Chao-Chou was VOC-sensitive chemistry dominant by use of both O 3 /HNO 3 and O 3 /NO y indicators, but NO x -sensitive chemistry was dominant by use of H 2 O 2 /HNO 3 indicator in four seasons.Meanwhile, Kenting was primarily NO xsensitive in all seasons.The VOC-sensitive and NO xsensitive regimes determined using the indicator ratios agree well with those obtained from the SPM using the threshold value E = 0.6-0.8.

Table 2 .
Statistical data of H 2 O 2 , HNO 3 and NO y measured at Pingtung,Chao-Chou and Kenting in 2003-2004.
no emission sources surrounded, and with strong wind speed (see Table1.).The NO y concentrations in southern Taiwan, Pingtung County, were similar to 0.3-110 ppbv on the Augustine's campus near the downtown Raleigh, North Carolina

Table 3 .
Comparison of measurements of H 2 O 2 , HNO 3 and NO x in this study and previous studies.

Table 4 .
Percentages of VOC-sensitive and NO x -sensitive regimes to ozone formation at Pingtung, Chao-Chou and Kenting in four seasons in2003-2004.