Simultaneous determination of atmospheric sulfur and nitrogen oxides using a battery-operated portable filter pack sampler

Abstract

We developed a method to analyze atmospheric SO_x (particulate SO₄²⁻ + gaseous SO₂) and NO_x (NO + NO₂) simultaneously using a battery-operated portable filter pack sampler. NO_x determination using a filter pack method is new. SO_x and NO_x were collected on a Na₂CO₃ filter and PTIO (2-phenyl-4,4,5,5-tetramethylimidazoline-3-oxide-1-oxyl) + TEA (triethanolamine) filters (6 piled sheets), respectively. Aqueous solutions were then used to extract pollutants trapped by the filters and the resulting extracts were pre-cleaned (e.g. elimination of PTIO) and analyzed for sulfate and nitrite by ion chromatography. Recoveries of SO₂ and NO_x from standard pollutant gases and consistency of the field data with those from other instrumental methods were examined to evaluate our method. SO_x and NO_x could be analyzed accurately with determination limits of 0.2 ppbv and 1.0 ppbv (as daily average concentrations), respectively. The sampler can determine SO_x and NO_x concentrations at mountainous or remote sites without needing an electric power supply.

---

Introduction

Increases in atmospheric SO_x and NO_x cause acid depositions that have adverse effects on the forest ecosystem, as do ozone and water stress.^1–4 Measuring changes in SO_x and NO_x concentrations is very important for evaluating the effects of atmospheric pollutants on vegetation. However, in mountainous and remote areas it is usually difficult to measure these pollutants because of the lack of an electric power supply. Passive samplers, which are inexpensive non-electric devices, are exclusively used for extensive monitoring at such sites,^2,3,5 but they are not useful for short-term sampling because their slow collection rate is controlled by diffusion, and local meteorological conditions often affected the results.^3,6,7

Active samplers such as filter pack (FP) methods are widely used for sampling atmospheric pollutants.^8–10 FPs have been employed in acid deposition monitoring networks in various countries, such as in EANET,¹¹ CASTNet,¹² and EMEP.¹³ Although FPs provide sensitive measurements and are relatively inexpensive and simple, sampling artifacts often occur in distinguishing particulate (SO₄²⁻, NO₃⁻ and NH₄⁺) and gaseous (SO₂, HNO₃ and NH₃) forms.^14–16 To overcome these problems, annular denuder¹⁷ and impactor/honeycomb denuder/filter pack¹⁸ systems were introduced. Although these denuder systems are sophisticated and precise, they are relatively expensive, complex, and more labor-intensive than the more usual FPs.¹⁰

Active samplers are usually operated using a fixed power supply, and field-oriented samplers operated by batteries are limited;^19–21 although many personal samplers to monitor occupational atmospheric environments have been developed and are usefully employed.^22–27

We developed a battery-operated portable FP sampler to simultaneously collect atmospheric SO_x and NO_x at mountainous sites without the need for a power supply, and to obtain daily (or hourly if concentrations are relatively high) average concentrations. In mountainous areas, NO_x concentrations can rapidly fluctuate depending on mobile sources (automobiles), local meteorological (e.g. formation of a temperature inversion layer) and weather conditions.^1,2,28 They may also fluctuate as a result of NO, which is naturally emitted from the forest soils.^29,30 On the other hand, SO_x concentrations often change daily or hourly under the influence of volcanic gases (a specific condition in Japan). Forest health may be more seriously affected by short-term high concentrations of the pollutants than by long-term averages. Therefore, our FP sampler may provide a new tool for information gathering, one that would be available to monitor and diagnose forest health.

We analyzed SO_x and NO_x as total concentrations without distinguishing particulate SO₄²⁻ and gaseous SO₂, or NO and NO₂ due to the following reasons: (1) SO_x and NO_x deposited to vegetation are converted rapidly to SO₄²⁻ and NO₃⁻, respectively; (2) The analytical precision for SO_x is decreased by speciation, because of the very low SO_x concentrations in Japan and unexpected artifacts (partial adsorption of SO₂ on a pre-filter to collect particulate pollutants^14,31); (3) NO_x concentrations may be a useful index of air pollution, because most of NO_x from anthropogenic sources is emitted initially as NO, and then converted to NO₂. In Japan, NO_x emissions (especially from automobiles) have been controlled under effluent standards (e.g. NO_x and PM Act); and (4) All acidic pollutants including NO should be removed and so facilitate clean exhaust emission from our sampler to a small field chamber³² (details are to be published elsewhere).

The novelty of our method is highlighted by the following points: (1) Determination of NO_x using a FP sampler is new, although such samplers (and personal samplers) have been applied for NO₂.^15,27,33; (2) Our FP sampler is a field-oriented type, which can operate continuously for periods of several days at mountainous sites; and (3) The system is completely air-tight from inlet to outlet; thus, clean exhaust at a constant flow rate can be supplied to a small field chamber, for example, for studying the material leaching from tree leaves.

The pollutants collected on the filters were analyzed by ion chromatography after extraction into aqueous solutions. The best sampling and analytical conditions to obtain the maximal and reproducible recoveries were investigated using standard pollutant gases. Then, field data obtained by our sampler were compared with those from other instrumental measuring methods to confirm performance.

Experimental

Sampler design

We developed battery-operated portable samplers (Fig. 1) and used them experimentally. SO_x and NO_x were collected on a Na₂CO₃ filter (1 sheet, 25 mm id, filter holder: Advantec PP-25) and PTIO (2-phenyl-4,4,5,5-tetramethylimidazoline-3-oxide-1-oxyl) + TEA (triethanolamine) filters (6 sheets in a pile, 47 mm id, filter holder: Advantec PP-47), respectively, by passing air sequentially through the filters. When sampling in the field, a Teflon net (127 mesh inch⁻¹, Sefar) was attached to the air intake to avoid trapping insects, and a polypropylene shade was attached as a rain shelter. A carbon vane pump (G12/01-4EB, Thomas) and an integrating mass flow meter (CMS20, Yamatake) were operated using two parallel rechargeable 12 V DC batteries (NP7-12, Yuasa). Although some of the other pumps (e.g., the PCXR-8, SKC Inc.), used in personal samplers (e.g., the IOM aerosol sampler²⁴), may be applicable to our sampler after slight modification, the pump used here was good in respect to its air-tightness, cleanliness (because it is oil-less), and torque (a constant flow rate was maintained after passing through 7 sheets of filters). Also, the flow meter was not affected by temperature or pressure; it recorded values at 20 °C and 1 atm. Silicone tubing (3 mm in inner diameter, 6 mm in outer diameter) was used to connect components. All the components were put in a plastic container (340 L, 230 W, 160 H mm), with the sampler weighing about 6 kg. The rapid decrease in the flow rate when the air intake or outtake was clogged verified that there was no air leakage. The sampler can be operated continuously for more than 24 h at a flow rate of ca. 1.0 L min⁻¹ (the decrease in the flow rate was <10%).

Fig. 1 Scheme (a) and picture (b) of a portable filter pack sampler for SO_x and NO_x.

Preparation of filters

Na₂CO₃ filter: Quartz fiber filters (Pallflex 2500QAST, 25 mm id) were washed with 0.1 M Na₂CO₃ by sonication for 30 min (28–100 kHz, Honda Electronics), and rinsed 3 times with deionized water. The filters were then soaked in 0.5 M Na₂CO₃ for 30 min (the solution was replaced with a fresh one during the soaking). After draining excess solution, the filters were dried in a vacuum desiccator, in which NaOH was used as the desiccant to prevent excessive drying.

PTIO + TEA filter: Cellulose filters (Advantec No. 5C, 47 mm id) were washed with deionized water by sonication for 30 min, and rinsed 3 times with deionized water. The filters were then dried for more than 12 h at 60 °C in an electric oven. 0.5 ml of 0.1 M PTIO and 0.5 M TEA (in acetone) was impregnated homogeneously into the filter. The filters were then shaken gently to evaporate the acetone, stored in an opaque desiccator containing NaOH (or silica gel) as the desiccant, and used within five days (the filters can be stored for more than 1 month in a refrigerator³⁴).

Post-sampling treatment of filters and analysis

After sampling, 30 ml of 0.3% H₂O₂ solution was used as an extractant on the Na₂CO₃ filter to oxidize the SO_x collected and to extract it by dissolving it as SO₄²⁻. The extract was filtered (pore size 0.45 μm, GL Science), and subjected to SO₄²⁻ analysis (and NO₂⁻ and NO₃⁻ analysis, if necessary). 30 ml of deionized water was applied to PTIO + TEA filters for more than 30 min with occasional shaking. Then 10 ml of chloroform (pre-washed with deionized water) was added to the extract, and the mixture shaken vigorously for 30 sec by hand. After centrifuging (800 G, 10 min), the filters were removed, and the mixture was shaken and centrifuged again. By this treatment, PTIO was removed from the extract into chloroform. Then, the extract (aqueous phase) was passed through a C₁₈-cartridge (Sep-Pak, Waters) to remove the remaining traces of chloroform and PTIO. The resulting solution was subjected to NO₂⁻ analysis (and NO₃⁻ analysis, if necessary). Anions (SO₄²⁻, NO₂⁻ and NO₃⁻) were analyzed by an ion chromatograph with a conductivity detector (DX-100, Dionex). The analytical conditions are summarized in Table 1. Filter blanks were also analyzed using unexposed filters, if necessary.

Table 1 Analytical conditions for ion chromatography

a Dionex. b Hitachi.
Analyzer:	DX-100^a
Column:	IonPac AS12A/AG12A (4 mm id)^a
Eluent:	2.7 mM Na2CO3 and 0.3 mM NaHCO3
Flow rate:	1.5 ml min^−1
Suppressor:	ASRS-ULTRA or ASRS-ULTRA II^a
	Autosuppression recycle mode
Sample injection volume:	250 μl
Detector:	Conductivity detector^a
Integrator:	D-2500 chromato-integrator^b

---

All chemicals were obtained from Wako Pure Chemical Industries, except for PTIO (Tokyo Kasei Kogyo). Deionized water was obtained from a Milli-Q system (Millipore).

Recovery tests

SO₂, NO and/or NO₂ in artificially polluted air samples (described below) were collected on various filters at flow rates of ca. 1.0 L min⁻¹. Although recovery tests are usually conducted using dilute standard gases,³³ relatively concentrated gases prepared in Tedlar bags were used because we wanted to accurately understand the loadings, and to quickly and inexpensively carry out the tests. Recoveries of pollutants were calculated using the following equation: recovery (%) = 100 × [SO₄²⁻ or NO₂⁻ (+NO₃⁻) found/SO₂ or NO_x loaded]. Since temperature was not controlled, it was assumed to be 20 °C (average room temperature), and deviations in recovery due to possible temperature changes were expected to be <3%. Because they were much lower than the loadings, corrections for filter blanks were not performed in this experiment.

Preparation of artificially polluted air samples

Artificially polluted air samples were prepared in 1 L or 10 L Tedlar bags (GL Science) (thoroughly pre-cleaned with N₂) with small Teflon balls (10 balls, 3 mm id, GL Science). The bag was filled with N₂ (or air), and then appropriate small amounts of certified standard gases (SO₂: 4178 ppmv at 20 °C, N₂ balance, Takachiho Chemical; NO: 8150 ppmv at 20 °C, N₂ balance, Takachiho Chemical; NO₂: 2018 ppmv at 20 °C, N₂ balance, Sumitomo Seika) were injected into the bag using a gastight syringe (1050TLL, Hamilton). The bag was then shaken vigorously to homogenize the gases. The amounts of pollutants in the bags corresponded approximately to those for 48 h sampling periods at polluted urban areas, i.e. SO₂ was ca. 100 ppmv (in a 1 L bag) or 10 ppmv (in a 10 L bag), NO was 400 ppmv (1 L bag) or 40 ppmv (10 L bag), and NO₂ was 200 ppmv (1 L bag) or 20 ppmv (10 L bag). The sampling duration for each test was ca. 1 min (for the 1 L bags) or 10 min (10 L bags). To prevent reaction and adsorption of gases in the bags, air samples were used just after preparation.

Comparison of our data with that obtained by other instrumental methods

Daily average concentrations of SO_x and NO_x in suburban air were analyzed by this method at and around (within 10 m of) the Air Monitoring Station of the National Institute for Environmental Studies (NIES) (in Tsukuba, Ibaraki, Japan). The results were compared with those obtained by other instrumental methods;³⁵i.e., concentrations of SO₂ obtained by UV fluorescence (GFS-32, DKK) and those of NO_x obtained by chemiluminescence (GLN-254, DKK). The comparisons were conducted for 56 sets of data obtained at a frequency of 1–3 times per month from January 2004 to February 2005.

Statistical analysis

Statistical calculations (linear regression, correlation, and the Student t-test using paired data) were performed with StatView 4.5j (Abacus Concepts).

Results and discussion

Recovery of SO₂ from various alkaline filters

SO_2, from the artificially polluted air samples containing SO₂ alone or SO₂ and NO_x, was collected on various alkaline filters including NaOH, Na₂CO₃, K₂CO₃, tetrabutylammonium hydroxide, and on PTIO + TEA filters (Table 2). All filters were prepared by a method similar to that described above. Collection by an aqueous absorbent³⁶ was also carried out, as an indication of the quantitative recovery.

Table 2 Recoveries of SO₂ from alkaline filters

Alkaline solutions impregnated into filters			Recovery of SO2 (%)
Alkali	Solvent	Volume/ml	From SO2 alone^b	From SO2 + NOx^c
Values: means (standard deviations), replicates. The cellulose filters (Advantec No. 5C, 47 mm id) were used.a Tetrabutylammonium hydroxide.b 100 ppmv SO2 in 1 L N2.c 100 ppmv SO2 + 400 ppmv NO + 200 ppmv NO2 in 1 L N2.d Not tested.
1 M Na2CO3	Water	0.3	98.7 (0.3), 3	25.1 (3.5), 3
1 M Na2CO3·1.5H2O2	Water	0.3	—^d	56.7 (6.4), 3
1 M NaOH	Water	0.3	98.8 (5.5), 3	16.7 (2.1), 3
1 M NaOH + 0.2%Ca(OH)2	Water	0.3	98.2 (0.9), 3	12.7 (0.7), 3
6% K2CO3 + 2% glycerol	Water	0.3	101.8 (2.1), 3	52.0 (1.9), 3
0.5 M TEA + 0.1 M PTIO	Acetone	0.5	87.5 (13.8), 13	48.7 (2.4), 5
10% TBAH^a	Methanol	0.5	100.4 (2.5), 3	67.2 (3.7), 3
10% TBAH + 0.5 M TEA + 0.1 M PTIO	Methanol	0.5	—	44.9 (2.3), 3
Aqueous absorbend (0.1 M NaOH)	0.6% H2O2	10.2	96.6 (3.0), 13	103.1 (0.7), 3

---

The recoveries of SO₂ from air samples with SO₂ alone (without NO_x) were quantitative, irrespective of the filter types, except for the PTIO + TEA filters, with which recoveries varied from 63 to 99% (this variation was not improved by changing the sample volume (1 L or 10 L) or the matrix (N₂ or air)). However, when SO₂ coexisted with NO_x in the air samples, SO₂ recoveries decreased in all cases, although SO₂ detected in the 2nd alkaline filters, which were installed behind the 1st filters, was just trace (<5%).

These results suggested that SO₂ was collected quantitatively on all alkaline filters, but when SO₂ coexisted with NO_x it was partly transformed to S-species, which were not recoverable by the analytical method (ion chromatography) used. Such a reaction might be promoted when concentrated standard gases prepared in dry N₂ matrix are used. Sickles and Hodson¹⁶ also reported that a portion of the SO₂ retained by nylon filters was unrecoverable. To convert the unrecoverable S-species on alkaline filters to SO₄²⁻, the oxidative digestion method³⁷ was applied as the post-sampling treatment. When the air samples contained SO₂ and NO_x, SO₂ recoveries were low (38–69%) by the H₂O₂ extraction method (described above), but improved to 72–100% when the oxidative digestion method was employed (Table 3). On the other hand, recoveries from air containing SO₂ alone (without NO_x) were quantitative in both methods. These results suggested that the intermediate oxidized S-species are formed on the alkaline filters during sampling.

Table 3 Recoveries of SO₂ by different post-sampling treatments of filters

Alkaline solutions impregnated into filters	Recovery of SO2 (%)
	From SO2 alone^a		From SO2 + NOx^b
	H2O2 extraction	Oxidative digestion	H2O2 extraction	Oxidative digestion
Values: means (standard deviations), replicates. The quartz fiber filters (Pallflex 2500QAST, 47 mm id) were used.a 10 ppmv SO2 in 10 L N2.b 10 ppmv SO2 + 40 ppmv NO + 20 ppmv NO2 in 10 L N2.c Not tested.
0.5 M Na2CO3	96.7 (2.1), 5	96.8 (2.1), 3	49.7 (8.3), 4	83.7 (3.0), 10
1 M Na2CO3	—^c	—	64.0 (3.9), 3	84.6 (10.6), 3
1 M NaOH	99.4 (—), 2	—	64.6 (6.9), 3	90.2 (7.8), 10

---

Effect of humidity on SO₂ recovery

Recovery of SO₂ from air with different humidities was also examined (Fig. 2). Humid air samples were prepared as follows: ambient air (humid; 10 L) was cleaned by passing it through the PTIO + TEA filters, and then mixed with the artificially polluted air (dry; 10 L, N₂ matrix) through a Y-type joint tube. The resulting air samples had relative humidities of 18–35% (temperature: 27–29 °C), on the assumption that the humidities were half those of the ambient air.

Fig. 2 Recoveries of SO₂ at different humidities. Relative humidity: dry, ca. 0%; humid, 25.2 ± 5.4% (18–35%). Concentrations of pollutants: SO₂, 5–10 ppmv; NO, 20–40 ppmv; NO₂, 10–20 ppmv. Filter: Na₂CO₃ filter of 47 mm id.

Recoveries of SO₂ from dry air with SO₂ and NO_x were ca. 50%, whereas those from humid air were ca. 87% (range: 83–92%). This could indicate that moisture accelerates the oxidation of SO₂ collected on the Na₂CO₃ filters to SO₄²⁻, even when NO_x coexists. Lewin³⁸ and Sickles et al.¹⁰ also reported that humidity was an important parameter in the determination of SO₂ by the FP method.

Recovery of NO_x by PTIO + TEA filters

Recovery of NO_x from artificially polluted air samples was studied as a function of NO_x concentration, species of matrix gas, and number and size of filters (Table 4). Recoveries were calculated from the amounts of NO₂⁻ and also from the total amounts of NO₂⁻ and NO₃⁻ that were analyzed.

Table 4 Recoveries of NO_x from PTIO + TEA filters

Artificially polluted air samples				Filters		Recovery of NOx (%)
Pollutants	Conc./ppmv	Volume/L	Matrix	Size/mm id	Sheets	As NO2^−	As NO2^− + NO3^−	Replicates
Values: means (standard deviations).a 40 ppmv NO + 20 ppmv NO2 + 10 ppmv SO2.b 400 ppmv NO + 200 ppmv NO2 + 100 ppmv SO2.c Each filter was impregnated with 0.5 ml of 0.1 M PTIO + 0.5 M TEA (acetone solution).d Each filter was impregnated with 0.1 ml of 0.1 M PTIO + 0.5 M TEA (acetone solution).
NO	40	10	N2	47^c	6	87.6 (3.6)	97.4 (5.3)	11
NO	40	10	Air	47	6	103.0 (4.2)	104.2 (4.5)	8
NO	20	10	Air	47	5	101.3 (2.8)	103.5 (2.7)	4
NO	20	10	Air	47	4	91.5 (5.6)	93.6 (5.4)	4
NO	20	10	Air	25^d	4	70.7 (9.5)	73.8 (9.7)	22
NO2	20	10	N2	47	6	87.7 (5.1)	104.7 (9.9)	10
NO2	200	1	N2	47	6	101.2 (2.3)	104.7 (2.0)	6
NOx + SO2	Low^a	10	N2	47	6	103.3 (1.9)	107.4 (1.2)	4
NOx + SO2	High^b	1	N2	47	6	99.3 (6.1)	102.3 (7.4)	5

---

NO_x was recovered quantitatively using the 6 piled sheets of PTIO + TEA filters. When 40 ppmv NO in the 10 L bag with N₂ was sampled, the amount of NO collected on each filter decreased exponentially from the 1st to 6th filters, with 90% of the NO being recovered as NO₂⁻ and the remainder as NO₃⁻. However, most (ca. 99%) NO was recovered as NO₂⁻ when NO was prepared in an air matrix (with moisture). As shown in Fig. 3, NO is oxidized to NO₂ by PTIO, and then the NO₂ is disproportionated with TEA and H₂O to form NO₂⁻ and NO₃⁻.^39,40 Therefore, conversion of NO to NO₂⁻ may be promoted by moisture. At the lower NO loadings, 5 filter sheets were sufficient for quantitative collection, but the recoveries decreased if the number of filters was reduced further or if smaller filter sizes were used. This may be attributed to the slow reaction between NO and PTIO, and thus a portion of NO may have escaped without oxidative collection. Recoveries were also decreased when coarse filters were used.

Fig. 3 Successive reactions of NO_x with PTIO, TEA and H₂O.

NO₂ was collected more easily, and thus by less filters than NO, probably because the reaction between NO₂ and TEA was rapid. For example, NO₂ in air samples (20 ppmv in 10 L N₂) was collected quantitatively on the 1st to 3rd filters. Sickles et al.³³ also recovered NO₂ almost quantitatively (87 ± 9%, n = 33) on glass fiber filters with TEA from air containing 5–400 ppbv NO₂.

NO_x was also recovered quantitatively from air samples with NO_x and SO₂, and most (>96%) of the NO_x was recovered as NO₂⁻. SO₂ did not interfere with the NO_x determination. Similar results have also been reported by Sickles et al.³³

Elimination of PTIO from the PTIO + TEA filters extract

In NO_x analysis using PTIO, extracts are usually determined by colorimetric or flow injection methods.^5,34 However, after elimination of the co-extracted PTIO, they can be analyzed more conveniently by ion chromatography. Saito⁵ removed PTIO from the extracts by extraction with ethyl ether, whereas we used chloroform extraction followed by filtration through a C₁₈-cartridge. As a result, 10 mmol L⁻¹ of PTIO in the extracts was reduced to 40 ± 2 nmol L⁻¹ after extraction, and then to 21 ± 1 nmol L⁻¹ after filtration (n = 15, elimination rate: 99.8%; PTIO concentration was determined colorimetrically at 345 nm). Recoveries of 3.0 mg L⁻¹ of NO₂⁻, NO₃⁻ and SO₄²⁻, spiked in the filter extracts, were 100%, 105% and 99%, respectively (n = 5), after treatment. Smaller amounts (0.3 mg L⁻¹) of NO₂⁻ and SO₄²⁻ were also recovered quantitatively without any effect from the treatment, although NO₃⁻ recovery was affected by the impurities of PTIO when NO₃⁻ was present at a low concentration.

Field measurements

Atmospheric SO_x and NO_x were collected simultaneously on Na₂CO₃ (1 sheet) and the PTIO + TEA (6 sheets) filters, respectively, installed in that order, at NIES and Mt. Tsukuba (ca. 20 km away from NIES). Concentrations of SO₂ recovered from Na₂CO₃ filters that were placed behind an additional quartz fiber filter (2500QAST, Pallflex) (this filter can remove particulate SO₄²⁻) were 85 ± 12% (n = 5, SO₂ concentrations were 0.9–1.9 ppbv) of those measured by the UV fluorescence method, though the difference was not significant (p > 0.05). A portion of the SO₂ might have been adsorbed on the quartz fiber filter.^14,31 In addition, although the conversion of SO₂ to unrecoverable S-species on the Na₂CO₃ filter seems to be insignificant when moist and dilute ambient air is sampled (see Fig. 2), the reaction might have occurred to a slight extent, because the sampling periods were very dry. However, this small underestimation may not be significant in field measurements.

Most of the NO_x collected on the PTIO + TEA filters was recovered as NO₂⁻ (NO₃⁻ was trace), probably because the ambient air contained moisture. Distributions of NO₂⁻ recovered from the 1st to 7th filters are shown at different concentrations of NO_x in ambient air (Fig. 4). More than 98% of the NO₂⁻ was detected in the PTIO + TEA filters (the 2nd to 7th filters). Trace NO₂⁻ and NO₃⁻, recovered from the Na₂CO₃ filter (the 1st filter), might have been nitrous acid and particulate nitrate + nitric acid, respectively.^10,33 In other FP and passive sampling methods using TEA as absorbent, most NO₂ is retained as NO₂⁻ (with trace NO₃⁻) on the filters.^33,39,41

Fig. 4 Distribution of NO_x recovered from each filter at different atmospheric concentrations of NO_x. 1st filter: Na₂CO₃ filter, 2nd–7th filters: PTIO + TEA filters. Sampling sites: NIES and Mt. Tsukuba. Period: January–April 2004.

When ambient SO_x was sampled, it was not detected from the PTIO + TEA filters (these can also collect SO_x) behind the Na₂CO₃ filter, or from the other Na₂CO₃ filter which was tentatively placed behind the first Na₂CO₃ filter. The side reaction (i.e., from SO₂ to unrecoverable S-species) on the Na₂CO₃ filter was usually negligible under moist conditions. Also, almost quantitative recoveries of ambient SO₂ using similar alkaline filters have been reported.⁹ In the case of ambient NO_x, it was recovered mostly from the first PTIO + TEA filter, and recoveries decreased exponentially at the subsequent filters (those from the last filter were negligible) (see Fig. 4). In addition, as described later (Fig. 5), good correlation was observed between the ambient pollutant concentrations obtained by our FP method and those by other instrumental methods. Therefore, even under field diluted conditions we expected good recoveries of SO_x and NO_x, similar to those observed for standard gases with high pollutants.

Fig. 5 Relationships between (a) daily average concentrations of SO_x obtained by the current FP method and those of SO₂ obtained by UV fluorescence, and (b) daily average concentrations of NO_x obtained by the current FP method and those obtained by chemiluminescence. Values of slope and intercept: mean ±95% confidence interval. Period: January 2004–February 2005.

Comparison of our data with that obtained by other instrumental methods

Daily average concentrations of SO_x and NO_x obtained by our FP method were compared with those obtained by other instrumental methods (Fig. 5). The data were plotted without weighting because: (1) They had been already averaged (i.e., daily average values); and (2) The regression lines in Fig. 5 were almost the same as the new regression lines, calculated using the averages of the data allocated to the sub-divided concentration ranges.

In Fig. 5(a), the pollutants analyzed were different between our FP and the instrumental methods (i.e., SO_x in the former and SO₂ in the latter); thus, linear regression that did not pass the origin was employed. SO_x concentrations obtained by our method correlated well with those obtained for SO₂ (not SO_x) by the UV fluorescence method, but were always ca. 30% higher, probably due to particulate SO₄²⁻ also being accounted for in this method, as our sampler was not equipped with a particle collection filter and/or an impactor. The intercept did not differ significantly from zero; suggesting that the SO₄²⁻ fractions were relatively small and changed with SO₂ (i.e., were absent when SO₂ was absent). High concentrations of SO_x, affected by volcanic gases discharged and transported from Miyakejima Island (located 240 km SSW of NIES) were sometimes observed. This was supported by back trajectory analysis using METEX.⁴²

In Fig. 5(b), the pollutants analyzed were the same between both methods; therefore, linear regression including the origin was employed. NO_x concentrations obtained by our method were in good agreement with those obtained by chemiluminescence, showing a regression line with a slope of about one.

During the study period, temperature and relative humidity varied widely over ranges of 2.8–29.5 °C (average: 13.0 ± 8.7 °C) and 29–83% (59 ± 15%), respectively. Therefore, our FP method may enable simultaneous determination of atmospheric concentrations of SO_x and NO_x, without significant effects from temperature or humidity.

Reproducibility and determination limits

Mean relative standard deviations of the results (13 sets) obtained by the three FP samplers collocated at the same sites were 6 ± 3% for SO_x and 5 ± 4% for NO_x. Those for nitrous acid and particulate nitrate + nitric acid (NO₂⁻ and NO₃⁻, respectively, recovered from the Na₂CO₃ filter) were 26 ± 27% and 13 ± 12%, respectively, and increased with decreasing pollutant concentrations. The determination limit of SO_x was defined as that of ion chromatography (0.05 mg L⁻¹ as SO₄²⁻; corresponding to 0.2 ppbv in air) because the filter blanks were small (0.02 ± 0.04 ppbv). On the other hand, the filter blanks for NO_x were relatively high (0.6 ± 0.3 ppbv), and the determination limit was defined as the value (1.0 ppbv), which was about three times as large as the standard deviation. Since battery power and the efficacy of the PTIO reagent can last for 2 days, and the determination limits of SO_x and NO_x can be lowered approximately to 0.1 ppbv and 0.5 ppbv, respectively; over a 2-day sampling period, we should be able to obtain reasonable data for SO_x and NO_x levels using our FP method at all sites in Japan.

Acknowledgements

The authors thank Dr M. Nishikawa of NIES for providing data from the Air Monitoring Station.

References

1. A. Bytnerowicz, M. Tausz, R. Alonso, D. Jones, R. Johnson and N. Grulke, Environ. Pollut., 2002, 118, 187.
2. A. Bytnerowicz, O. Badeab, F. Popescuc, R. Musselmand, M. Tanaseb, I. Barbue, I. Fraczekf, N. Gembasub, A. Surdub, F. Danescub, D. Postelnicug, R. Cenusae and C. Vasile, Environ. Pollut., 2005, 137, 546.
3. R. M. Cox, Environ. Pollut., 2003, 126, 301.
4. T. Takamatsu, H. Sase and J. Takada, Can. J. For. Res., 2001, 31, 663.
5. K. Saito, J. Jpn. Soc. Atmos. Environ., 2003, 38, 145 (in Japanese).
6. R. H. Brown, J. Environ. Monit., 2000, 2, 1.
7. S. V. Krupa and A. H. Legge, Environ. Pollut., 2000, 107, 31.
8. M. S. Erduran and S. G. Tuncel, J. Environ. Monit., 2001, 3, 661.
9. S. Leppänen, P. Anttile, H. Lättilä and U. Makkonen, Atmos. Environ., 2005, 39, 2683.
10. J. E. Sickles II, L. L. Hodson and L. M. Vorburger, Atmos. Environ., 1999, 33, 2187.
11. ADORC (ADORC prepared for EANET), Technical document for filter pack method in East Asia, Acid Deposition and Oxidant Research Center, Niigata, Japan, 2003.
12. U.S. EPA (MACTEC Engineering and Consulting, Inc. prepared for U.S. EPA), Clean air status and trends network (CASTNet) 2003 annual report, U.S. EPA Office of Air and Radiation Clean Air Markets Division, Washington, DC, USA, 2004.
13. EMEP, EMEP/CCC-Report 1/95 (revision 2001), EMEP manual for sampling and chemical analysis, Norwegian Institute for Air Research, Kjeller, Norway, 2001.
14. B. R. Appel, Y. Tokiwa, M. Haik and E. L. Kothny, Atmos. Environ., 1984, 18, 409.
15. T. A. Pakkanen, R. E. Hillamo, M. Aurela, H. V. Andersen, L. Grundahl, M. Ferm, K. Persson, V. Karlsson, A. Reissell, O. Royset, I. Floisand, P. Oyola and T. Ganko, J. Aerosol Sci., 1999, 30, 247.
16. J. E. Sickles II and L. L. Hodson, Atmos. Environ., 1999, 33, 2423.
17. M. Possanzini, A. Febo and A. Liberti, Atmos. Environ., 1983, 12, 2605.
18. P. Koutrakis, C. Sioutas, S. T. Ferguson, J. M. Wolfson, J. D. Mulik and R. M. Burton, Environ. Sci. Technol., 1993, 27, 2497.
19. P. Beddoes, Atmos. Environ., 1967, 1, 595.
20. K. W. Brown, G. B. Wiersma and C. W. Frank, J. Environ. Qual., 1981, 10, 52.
21. A. Gupta, R. Kumar, K. M. Kumari and S. S. Srivastava, Atmos. Environ., 2003, 37, 4837.
22. P. Demokritou, I. G. Kavouras, S. T. Ferguson and P. Koutrakis, Aerosol Sci. Technol., 2001, 35, 741.
23. M. Harper and B. S. Muller, J. Environ. Monit., 2002, 4, 648.
24. D. Mark and J. H. Vincent, Ann. Occup. Hyg., 1986, 30, 89.
25. C. Misra, M. Singh, S. Shen, C. Sioutas and P. A. Hall, J. Aerosol Sci., 2002, 33, 1027.
26. C. J. Tsai, C. H. Huang, S. H. Wang and T. S. Shih, Aerosol Sci. Technol., 2001, 35, 611.
27. Y. Wang, A. Allen, D. Mark and R. M. Harrison, J. Environ. Monit., 1999, 1, 423.
28. A. Naemura, A. Tsuchiya, Y. Fukuoka, K. Nakane, H. Sakugawa and H. Takahashi, Jpn. J. Biometeor., 1996, 33, 131.
29. R. Gasche and H. Papen, J. Geophys. Res., 1999, 104, 18505.
30. X. K. Xu and K. Inubushi, Soil Sci. Plant Nutr., 2005, 51, 683.
31. S. Batterman, I. Osak and C. Gelman, Atmos. Environ., 1997, 31, 1041.
32. A. Bytnerowicz, D. M. Olszyk, P. J. Dawson and L. Morrison, J. Environ. Qual., 1989, 18, 268.
33. J. E. Sickles II, P. M. Grohse, L. L. Hodson, C. A. Salmons, K. W. Cox, A. R. Turner and E. D. Estes, Anal. Chem., 1990, 62, 338.
34. Ogawa & Company, USA, Inc., NO, NO₂, NO_x and SO₂ sampling protocol using the Ogawa passive sampler, 1998.
35. M. Nishikawa, personal communication.
36. T. Takamatsu, Trans. Res. Inst. Oceanochem., 1999, 12, 43 (in Japanese).
37. T. Takamatsu, Int. J. Environ. Anal. Chem., 1996, 62, 303.
38. E. Lewin and B. Zachau-Christiansen, Atmos. Environ., 1977, 11, 861.
39. R. Cee and J. C. Ku, Analyst, 1994, 119, 57.
40. K. Hirano, H. Maeda and K. Matsuda, Simultaneous determination method of NO, NO₂ and SO₂ in ambient air by use of diffusional sampling device for short-term integrated samplers, YERI-Report 128, Yokohama Environmental Science Research Institute, Yokohama, Japan, 1997 (in Japanese).
41. D. Krochmal and A. Kalina, Atmos. Environ., 1997, 31, 3473.
42. J. Zeng, M. Katsumoto, R. Ide, M. Inagaki, H. Mukai and Y. Fujinuma, Data analysis and graphic display system for atmospheric research using PC, CGER Report M014-2003, National Institute for Environmental Studies, Tsukuba, Japan, 2003.

---