Journal Archive

The filter-pack sampling system in the present study is shown in Fig. 1. The sampling system consists of two filter-pack sets (Nilu filter packs supplied by Tokyo Dylec Co.). One set is for sampling of HNO₃ and HONO, and the other is for NO₂ and NO. FP1 is for HNO₃ and HONO sampling with a sampling flow rate of 4.0 dm³ min^-1. This flow rate is limited by an impactor for the PM_2.5 sampling requirement of F1 (the impactor is not indicated in Fig. 1). Therefore, the flow rate can be changed when PM_2.5 is not collected. The entire air flow was drawn by a Hiblow Air Pump from Techno Takatsuki Co., Ltd. (KP-5030S, 100 V and 0.26 A). After the particles are removed with a Teflon filter, F1, all HNO₃ is collected with a NaCl-impregnated filter, F2. Next, all HONO and a few percent of NO₂ are collected with a Na₂CO₃-impregnated filter, F3. Then, almost the same percent of NO₂ is absorbed on a Na₂CO₃-impregnated filter, F4 (Noguchi et al., 2007). After FP1, the sampling line is separated into two lines. One line is introduced to FP2 with a lower flow rate. There are two reasons for the lower flow rate of FP2. NO₂ cannot be collected completely at higher flow rates, such as FP1. Another reason is that NO₂ and NO concentrations are much higher than HNO₃ and HONO concentrations, and therefore, a high flow rate is not required for simultaneous sampling and possibly causes adsorption breakthrough on the impregnated filter. F5 and F6 are guaiacol/NaOH-impregnated filters for NO₂ collection. One filter is enough, but we used two filters to avoid adsorption breakthrough. F7 is a KMnO₄-impregnated filter for oxidation of NO. F8 and F9 are guaiacol/NaOH-impregnated filters to collect NO₂ produced by oxidizing NO. Each nitrogen oxide concentration can be calculated from the nitrate and nitrite concentrations explained later; the volume of extraction and the flow rate of the sample are as follows.

Fig. 1. 
Sampling setup for HNO₃, HONO, NO₂ and NO in the present method. FP1 (filter-pack set 1): for HNO₃ and HONO sampling, FP2 (filter-pack set 2): for NO₂ and NO sampling. F1: Teflon filter, F2: NaCl-impregnated filter, F3 and F4: Na₂CO₃-impregnated filters, F5, F6, F8 and F9: Guaiacol-impregnated filters, F7: KMnO₄-impregnated filter. MFC1, 2: mass flow controller, P: air pump. The flow rate of FP1 was controlled by MFC1 plus MFC2.

Here, subscripts indicate the filter number explained above. For the guaiacol filter, only nitrite was analyzed, since only nitrite was observed in our earlier experiment measured by ion chromatography. Additionally, it has been reported that almost all NO₂ is converted to nitrite in alkaline guaiacol solution (Nash, 1970).

To evaluate the performance of the present method, NO and NO₂ were measured by a Thermo Fisher Scientific, Model 42i chemiluminescence NO_x analyzer (CL). However, the NO₂ value obtained by the CL method includes NO₂, HNO₃, HONO and other nitrogen oxide and nitrogen-containing organics. In this study, nitrogen-containing organics were not considered, since the concentrations were generally not very high. Only HNO₃, HONO and NO₂ were evaluated.

Guaiacol, sodium hydroxide, potassium permanganate, sodium carbonate, sodium chloride, methanol, glycerine and other chemicals were reagent grade, obtained from Fujifilm Wako Pure Chemical Corporation and used as received. Ultrapure water was obtained by using a Direct-Q 3UV from Merck Millipore, Inc. (resistivity≥18.2 MΩ cm). Nitrogen dioxide (10.5 ppm, air balance) and nitric oxide (4.93 ppm, N₂ balance) were obtained from Taiyo Nippon Sanso Co. and diluted with purified air obtained by a Zero Air generator, Model 111, from Thermo Fisher Scientific, Inc. (passing through silica gel (Masuda Rika Kogyo, Ltd., 5-10 mesh), Purafil (Nippon Thermo Co., Ltd., No. 7075) and activated carbon).

Impregnated filters F2, F3 and F4 were prepared according to the literature (Toriyama et al., 2019). For the HNO₃-collecting filter (F2), 2 wt% NaCl and 1 wt% glycerine in methanol/water (1/1) were used, and for the HONO-collecting filter (F3, F4), 2 wt% Na₂CO₃ and 1 wt% glycerine in methanol/water (1/1) were used. Impregnated filters F5, F6, F8 and F9 were prepared with No. 51A cellulose filters from Toyo Roshi Kaisha, Ltd. (47 mmΦ) as follows. First, the filters were washed with ultrapure water three times and immersed for three minutes in 10-20 wt% guaiacol, 5 wt% NaOH, and 1 wt% glycerol in methanol/water (1/1) solutions three times in three separate bottles. Impregnated filter F7 was prepared with a QR-100 quartz filter from Advantec Co., Ltd. (47 mmΦ). The filter was immersed in a mixture solution of 0.16 mol dm^-3 KMnO₄, which is the same concentration of oxidant for the Saltzman method described in Japanese Standard Method (2004), and 0.51 mol dm^-3 H₂SO₄.

After immersion in each solution, the filters were dried at 60°C and then kept in sealed polypropylene bags until sampling. After sampling, each filter was stored in a sealed polyethylene bag and kept cool until analysis. The absorbed components on the filters were extracted by ultrapure water, and these extracts were analyzed by ion chromatography in the laboratory. The anion chromatograph system (883 basic IC plus, Metrom, Switzerland) had a guard column (SI-90G, Shodex), separation column (SI-90 4E, Shodex), and suppressor for anions, and the eluent was 9 mmol dm^-3 Na₂CO₃ solution at a flow rate of 1 mL min^-1. The detection limits of nitrite and nitrate that were defined as 3 times the standard deviation of five measurements for 20 μmol dm^-3 were 0.42 and 0.34 μmol dm^-3, respectively. Nitrite formed on the guaiacol-impregnated filter was extracted by ultrapure water, and the extract was analyzed by a Griess-Ilosvay method (US Standard methods) with a UV/Visible spectrophotometer (Shimazu Co., Ltd., UV-1800) at 545 nm. Three blank filters for each were also analyzed for correction. The detection limit of nitrite that was defined as 3 times the standard deviation of six measurements was 0.11 μmol dm^-3.

To determine the optimum concentration of guaiacol in the immersing solution for NO₂ sampling, NO₂ collection efficiency was investigated by comparing with NO₂ values measured with the chemiluminescence NO_x analyzer. The concentrations of glycerol and NaOH were fixed at 1 wt% and 5 wt%, respectively, and that of guaiacol was changed from 10 to 20 wt%. Here, the flow rate of the sampling was 0.1 dm m³ min^-1 for 20-24 hours. The results are shown in Fig. 2. At 10 wt% guaiacol, NO₂ was not completely absorbed on the filter. It is speculated that higher guaiacol was required. At 20 wt%, the scattering was large. Guaiacol was deposited on the filter, probably because some NO₂ was absorbed during filter preparation. Therefore, it is considered that the results were scattered. At 15 wt%, the NO₂ concentrations measured by the present method were in good agreement with those measured by the CL method.

Fig. 2. 
Absorption efficiencies of NO₂ by guaiacol-impregnated filters prepared with 10, 15 and 20 wt% guaiacol-impregnating solution. The flow rate was 1.0 dm³ min^-1. R is a correlation coefficient. □: 10 wt%, [NO₂]_FP=0.883[NO₂]_CL, R=0.995, ○: 15 wt%, [NO₂]_FP=0.963[NO₂]_CL, R=0.999, △: 20 wt%, [NO₂]_FP=1.27[NO₂]_CL, R=0.992.

Using 15 wt% guaiacol immersing solution, the dependence of the absorption efficiency of NO₂ on the flow rate was investigated. The sampling times were 20-24, 6-7, 3-4 and 2.4-4 hours at 0.1, 0.2, 0.3, 0.4 and 0.5 dm³ min^-1, respectively. The samplings were conducted with two filters in series, and the concentration was calculated from the sum of nitrite and nitrate concentrations in the two filters. The results are shown in Fig. 3. From Fig. 3, it was found that at a flow rate lower than 0.4 dm³ min^-1, almost all NO₂ could be collected. Therefore, we decided that 0.3 dm³ min^-1 was to be used in the sampling for safety. The ratios of HONO/NO₂ were reported to be a few percent to 13.4%, and therefore, 4 dm³ min^-1 for HONO sampling (FP1) and 0.3 dm³ min^-1 for NO₂ sampling (FP2) were well balanced in the simultaneous sampling. One filter in FP2 could absorb NO₂ of approximately 80% on average, and therefore, a total of approximately 96% of NO₂ could be absorbed.

Fig. 3. 
Change in collection efficiencies of NO₂ with sampling flow rate. A 15 wt% guaiacol-impregnated filter was used. The circles and error bars show the average of 3-5 samples and one standard deviation, respectively. The sample air was outdoor air at the A5 building of Osaka Prefecture University.

In the sample air passing through FP1 and two guaiacol-impregnated filters, NO is still present. If NO is oxidized to NO₂, the NO concentration can be obtained since NO₂ can be absorbed by the two guaiacol filters. Then, the oxidation efficiencies of NO were investigated. NO was generally oxidized by chromium oxide (Hilliard and Wheeler, 1977), PTIO (Hauser et al., 2009; Watanabe et al., 2006), KMnO₄ (Japanese Industrial Standards, 2004) or ozone. In Japan, the use of chromium is socially problematic. PTIO requires an additional separation process to remove PTIO from the extraction (Hauser et al., 2009; Watanabe et al., 2006). To produce ozone, a UV lamp is required. In field measurements, an additional electric source for the UV lamp to produce ozone is not favorable. Therefore, in this study, KMnO₄ was used to oxidize NO. First, the dependence of the concentration of KMnO₄ in the immersing solution of the oxidation filter (quartz filter) on the oxidation efficiencies of NO was investigated. The mole ratio of KMnO₄ and H₂SO₄ was constant at 0.158 mol KMnO₄ and 0.51 mol H₂SO₄, respectively. The results are shown in Fig. 4. The oxidation efficiencies were higher than 80% from 0.02-0.2 mol dm^-3 KMnO₄ and stable at 0.16 mol dm^-3 KMnO₄. The oxidation efficiency of NO to NO₂ at 70% is used in the Japanese Industrial Standard (2004), but almost 100% was obtained in our results. The reason is not clear. The difference is that the Japanese Industrial Standard method uses an aqueous solution, and the present method uses a wet solid on a filter. The solubility of NO in water is not very high, and therefore, it is speculated that the contact of NO with KMnO₄ is not effective in aqueous solution. Furthermore, the oxidation efficiency was 95% at 320 ppb NO with a 0.16 mol dm^-3 KMnO₄-impregnated filter.

Fig. 4. 
Oxidation efficiencies of NO by KMnO₄-impregnated filters prepared with various concentrations of KMnO₄. The mole ratio of KMnO₄ and H₂SO₄ was 0.158 mol KMnO₄ and 0.51 mol H₂SO₄, respectively. The NO concentration was 80 ppb. The sampling was conducted at 0.3 dm³ min^-1 for 20-40 minutes. The error bars show one standard deviation of 3-4 samples.

The oxidation efficiency of 300 ppb NO was maintained at a flow rate of at least 2.75 hours at 4.0 dm³ min^-1. This indicates that a total of 8.24×10^-6 mol of NO could be oxidized continuously. When the flow rate of the sampling was set at 0.3 dm³ min^-1, the KMnO₄-impregnated filter could be used for 183 hours in the case of 60 ppb NO. It is concluded that this KMnO₄-impregnated filter prepared with 0.16 mol dm^-3 KMnO₄ can be used for field sampling sufficiently under normal ambient air conditions.

The optimum conditions for NO₂ and NO sampling (FP2) are summarized below.

• - Concentration of the immersing solution (1/1; methanol/ water) of guaiacol;Guaiacol: 15 wt%Na₂OH: 5 wt%Glycerol: 1 wt%
• - Concentration of immersing solution (water) of KMnO₄ solution;KMnO₄: 0.16 mol dm^-3H₂SO₄: 0.51 mol dm^-3
• - Flow rate: 0.3 dm³ min^-1.

Under the optimum conditions, the concentrations of NO and NO₂ in the ambient air were measured by the present method and compared with those by the CL method. Fig. 5A and 5B shows the results of NO and NO₂, respectively. Here, NO₂ concentrations measured by the CL method include NO₂, HNO₃, HONO and other nitrogen compounds. Therefore, those by the present method are expressed with the sum of NO₂, HNO₃, and HONO concentrations. Both lines showed a good linear relationship with a slope of almost 1.0; that is, for NO, [NO]_FP=0.987[NO]_CL, R=0.998, and for NO₂, [NO₂]_FP=1.04[NO₂]_CL, R=0.997.

Fig. 5. 
Comparison of NO and NO₂ concentrations measured by the present method and the CL method. (A) NO: [NO]_FP= 0.987[NO]_CL, R=0.995. (B): NO₂, [NO₂]_FP=1.04[NO₂]_CL, R=0.997 (R is a correlation coefficient).

The detection limits of these gases depend on the sampling time. If the detection limits of HNO₃, HONO, NO₂ and NO were calculated in the case of 4 hours of sampling by considering only errors from measurements of ions and error progression including three blank filters, those were 0.17, 0.31, 4.1 and 4.1 ppb, respectively.

Three sets of the present system were applied for simultaneous measurements of HNO₃, HONO, NO₂ and NO. The sampling was conducted on September 15, 2020, from 7:30-12:15 and 12:45-17:45. One set was automatically controlled with a timer and electromagnetic valve. The sampling points were 1) Osaka Prefecture University in Sakai City, Osaka (OPU); 2) Izumiotsu City, Osaka (IZM); and 3) Sango Town, Nara (SAN). OPU (34.548°N, 135.506°E) is located in an urban and residential area, and national road #310 (160 m apart) runs beside OPU. The sampling point of IZM (34.514°N, 135.414°E) is near the cross-section (5 meters from the road) of the Osaka Prefectural Road #29 and Osaka Rinkai Line, which is a three-lane road with heavy truck traffic on each side. SAN (34.595°N, 135.682°E) is in a residential area in a suburban area, and there is not much traffic. The results are shown in Table 1.

The concentrations of nitric acid were very low at OPU and SAN. At IZM, these values were 1.3 ppb in the morning and 1.0 ppb in the afternoon. The disadvantage of filter-pack sampling is well known and artifacts come from evaporation of nitric acid from particles on the filter. However, the concentrations of HNO₃ were not very high in this sampling period, and the artifacts were not very important. It is considered that the other gases were not affected by the artifacts of particles. The concentrations of HONO were not very different at the three points. For NO₂, the highest concentration, 83.8 ppb, was observed in the afternoon at IZM. At this point, there are many traffic sources, and high NO₂ concentrations have been reported in the past (Morioka et al., 2000). The NO₂ concentration at SAN was lower than that at the other two points. This was due to the lower traffic at SAN. The concentrations of NO were highest at IZM due to heavy traffic. In this sampling, the numbers of traffic at IZM were 2,500-4,300 vehicles/hour from 7 AM to 12 AM, including passenger cars, heavy-duty cars and motorbikes, and 2,300-3,500 vehicles/hour from 13 PM to 17 PM. Therefore, the NO and NO₂ concentrations were very high. However, in Table 1, the NO concentration in the morning at IZM must have been undercalculated. After sampling, the oxidation filter for NO was decolorized. This indicates that most KMnO₄ was reduced. This phenomenon was never observed at OPU. Sufficient amounts of KMnO₄ compared to NO were coated on the filter as mentioned above. It is speculated that high concentrations of reductants, such as organic compounds, were emitted from vehicles that reacted with KMnO₄. Therefore, in the afternoon sampling, the KMnO₄-impregnated filter was changed every two hours before the color of KMnO₄ became lighter. The conditions in the morning sampling at IZM were very special. In the case of sampling at such highly polluted places, several KMnO₄ filters should be connected in series.