Anal. Chem. 2005, 77, 4423-4428
Analysis of Sulfur in Deposited Aerosols by Thermal Decomposition and Sulfur Dioxide Analyzer Masatoshi Yamamoto*
Hyogo Prefectural Institute of Public Health and Environmental Sciences, 3-1-27 Yukihira-cho, Suma-ku, Kobe, Japan, 654-0037
A thermal decomposition method that measures aerosol sulfur at the nanogram level directly from the collection substrate is described. A thermal decomposition apparatus was designed. A stainless steel strip was used as the aerosol collection substrate. A 0.1 mol/L MnCl2 solution was added as the thermal decomposition catalyst. Currents were passed through the strip where aerosol particles had been deposited. In this way, the strip was heated at 780 ( 10 °C, and particulate sulfur was evaporated. A sulfur dioxide analyzer (SDA) with flame photometric detector (FPD) was used to detect gaseous sulfur. High sulfur recoveries from (NH4)2SO4 and other inorganic sulfates, such as NH4HSO4, K2SO4, MgSO4, and CaSO4, were obtained. From the sulfur blank and the calibration, a lower limited detection of 0.2 ng of sulfur and the determination range of 3.3-167 ng of sulfur were estimated. The method is effective for measuring the sulfate size distributions of urban aerosols in a small sample air volume of 50-60 L. The method is applicable to measuring the sulfur in aqueous extracts of sizesegregated urban aerosols collected by impactor and comparing the results with the sulfate data measured by ion chromatography. The fine particles of atmospheric aerosols have been implicated in acid deposition, in visibility reduction in urban and rural areas, and in the perturbation of the earth’s radiation balance.1 These fine particles in the urban atmosphere create the so-called PM2.5 problem, which has been blamed for human health effects.2 A sulfate is the major inorganic constituent of these fine particles, and it is one of the key components in study of the behavior of these particles in the atmosphere. Currently, ion chromatography (IC) is the primary method for quantifying sulfate in atmospheric aerosol. With IC, a variety of related ions, such as nitrates, chlorides, and ammonium, can be measured together. However, for the chemical analysis of a small amount of aerosol, which might involve size-segregated particles and high time resolution samples, a much more accurate and simpler method than IC would be preferable. * E-mail:
[email protected]. (1) Seinfeld, J. H.; Pandis, S. N. Atmospheric chemistry and physics: from air pollution to climate change; Wiley-Interscience: New York, 1998. 10.1021/ac0481093 CCC: $30.25 Published on Web 06/03/2005
© 2005 American Chemical Society
Methods involving thermal decomposition and a sulfur dioxide analyzer (TD/SDA) to measure sulfur in atmospheric aerosols have been reported.3-11 To measure the size distribution of the sulfur or the sulfate, an aerosol sampling with a cascade impactor following a TD/SDA was reported by Roberts and Friedlander4 and Vossler et al.9 In these techniques, the sulfate in a deposited aerosol on the collection substrate of the impactor was directly heated and vaporized. Because this technique does not require the extracting operation, there is no possibility of loss or contamination during handling. In addition, other advantages are that, following sensitive gas analysis, such as flame photometry, the method has proven to be quite sensitive, accurate, simple, and rapid. These methods are not specific for sulfate, however, they are for sulfur. This is because sulfate is a dominant component of aerosol sulfur, and because of the advantages described above, the method still remains effective for studying the behavior of sulfate aerosol in the atmosphere. Recently, two techniques for continuous, short time resolution measurement of aerosol sulfate have been developed. One is a particle-into-liquid system,12 which uses IC after impaction collection, and the second is the techniques using the continuous sulfate instruments (CSI). Two types of CSI, the Harvard continuous thermal method (TEII, model 5020 SPA)13 and the Aerosol Dynamics Inc. method (R&P, model 8400S),14,15 have been (2) Dockery, D. W.; Pope, C. A., Ill; Xu, X.; Spengler, J. D.; Ware, J. H.; Fay, M. A.; Ferris, B. G., Jr.; Speizer, F. E. N. Engl. J. Med. 1993, 329, 1753-1759. (3) Husar, J. D.; Husar, R. B.; Stubits, P. K. Anal. Chem. 1975, 47, 20622065. (4) Roberts, P. T.; Friedlander, S. K. Atmos. Environ. 1976, 10, 403-408. (5) Husar, J. D.; Husar, R. B.; Macias, E. A.; Wilson, W. E.; Durhan, J. L.; Shephered, W. K.; Anderson, F. A. Atmos. Environ. 1976b, 10, 591-595. (6) Huntzicker, J. J.; Hoffman, R. S.; Ling C. S. Atmos. Environ. 1978, 12, 8388. (7) Cobourn, W. G.; Husar, R. B.; Husar, J. D. Atmos. Environ. 1978, 12, 8998. (8) Kittelson, D. B.; McKenzie R.; Vermeersch, M.; Dorman, F.; Pui, D.; Linn, M.; Whitby, K. Atmos. Environ. 1987, 12, 105-111. (9) Vossler, T. L.; Macias, E. S. Environ. Sci. Technol. 1986, 20, 1235-1243. (10) Boehm, R.; Israel, G. W. J. Aerosol Sci. 1987, 18, 857-860. (11) Kim, M. G.; Yagawa, K.; Inoue, H.; Shirai, T. Anal. Appl. Pyrolysis 1991, 20, 263-273. (12) Weber, R. J.; Orsini, D.; Daun, Y.; Lee, Y. N.; Klotz, P. J.; Brechtel, F. Aerosol. Sci. Technol. 2001, 35, 718-727. (13) Allen, G. A.; Harrison, D.; Koutrakis, P. 20th Ann. Conf. Am. Assoc. Aerosol. Res. Portland, OR, 2001.. (14) Stolzenburg, M. R.; Hering, S. V. Environ. Sci. Technol. 2000, 34, 907914.
Analytical Chemistry, Vol. 77, No. 14, July 15, 2005 4423
Figure 1. Schematic diagram of the analytical system for the sulfur analysis. The system consists of a thermal decomposition apparatus (inside of the broken frame) and a gas analytical system (outside of the broken frame).
commercialized. However, these instruments have not been designed for the size distribution measurement, and it is thought that, to obtain the precise sulfur size distribution data using a multistage impactor, the TD/SDA is an effective method for sulfur determination. On the other hand, thermal decomposition techniques for measuring a nitrate aerosol have been reported.14,16,17 Yamamoto and Kosaka16 improved a method for thermal decomposition and chemiluminescence (TD/CL) that had been reported by Moscowitz,17 who applied a TD/SDA method developed by Roberts and Friedlander,4 using a chemiluminescent NOx detector instead of SDA. The Yamamoto and Kosaka method,16 which required the use of a thermal decomposition catalyst and an aerosol collection strip of unbaked stainless steel, allowed high time resolution measurements (sampling rate 1 L/min; sampling time 30-60 min) of the nitrate size distribution of the ambient aerosol. In this report, in a manner similar to the TD/CL reports provided above,16 previous work3,4,9 on TD/SDA is reexamined, and the improved method is investigated using the FPD for the SDA and the newly selected catalyst, the unbaked strip, and a new type of thermal decomposition apparatus are examined. EXPERIMENTAL SECTION Apparatus. A schematic diagram of the analytical system is shown in Figure 1. The system consists of a thermal decomposition apparatus (inside the dotted frame) and a gas analytical system (outside the dotted frame). The thermal decomposition apparatus consists of a vaporization cell (shaded area), a vaporized gas introduction tube for, a timer-controlled source of electricity, an infrared thermometer, and an air jack system. The vaporization cell consists of an upper and a lower cell. The lower cell is made of polypropylene and is equipped with two metal electrodes that (15) Hering, S. V.; Kirby, B. In Standard operating procedure for the routine operation of the R&P 8400S Ambient Particulate Sulfate Monitor, 2001. (16) Yamamoto, M.; Kosaka, H. Anal. Chem. 1994, 66, 362-367. (17) Moscowitz, A. H. EPA-600/3-77-053, 1977.
4424
Analytical Chemistry, Vol. 77, No. 14, July 15, 2005
are connected to the electric source by cables and with a clean air inlet. A mirror placed in the bottom of the cell reflects a beam emitted from a flashed stainless steel strip. The reflected beam is introduced to a no-contact-type infrared thermometer (LEC, Ltd., model KTL 520B), which instantaneously measures the temperature of the heated spot in the strip. The upper stainless steel cell is connected to the bottom of the vaporized gas introduction tube. A silicon rubber packing is mounted to the bottom of the upper cell. The lower cell moves up and down with an air jack that is controlled by an electromagnetic valve for compressed air. A stainless steel strip with a deposit of standard sulfur or an aerosol sample is placed over two electrodes. While the lower cell is being lifted to the upper cell, slipping the silicon packing, the inner room of the cell is air-proofed; in this way, each end of the strip is tightly pressed to each electrode. The gas analytical system consisted of a flame photometric detector (FPD, CSI Corp. model SA260), a strip chart recorder, and an electronic integrator (Shimadzu model CR4A). Carrier air was introduced into the vaporization cell from a clean air inlet by an FPD suction pump. The clean air was generated by passing the air through a column packed with an activated carbon. The sulfur analyzer was calibrated using a standard SO2 calibration gas produced by a dilution system (Standard Technology, Inc., SGGU-14). A 185.6 ppm SO2 in N2 standard gas, which was referenced to a primary standard of the Chemical Inspection and Testing Institute, was used. Thermal decomposition was achieved by rapidly heating the stainless steel strip with direct current, using an electric source equipped with a step-down transformer and a timer-controlled switch. The vaporized gas was introduced through the vaporized gas introduction tube that was heated by a ribbon heater and connected with the gas analytical system. The suction rate of the carrier was 240 mL/min. Reagents. Reagent grade chemicals (Wako Pure Chemical) were used throughout without further purification. Ultrapurified water obtained by passing distilled and ion-exchanged water through a water purification system (Milli-QTM, Millipore Corp.) was used to clean the strips and prepare the reagents. Stainless Steel Strip. The stainless steel strips were cut from the stainless steel film of type AISI No. 304. The size was 5 mm in width, 25 mm in length, and 0.03 mm in thickness. The strip was cleaned by sonication in hot detergent, washed with ultrapurified water, and dried in a furnace for 1 h at 180 °C. The baked strip, which was only used in the preliminary experiment, was prepared by baking a cleaned strip at 700 °C for 30 min. These baking conditions were milder than those (900 °C 1 h) reported.4,9,13 To prevent particles from bouncing off the strips, each stainless steel strip was coated with a 0.2-µL silicone grease solution with a microsyringe onto an impact area, and the solvent was evaporated using cleaned air. The 0.2% silicone grease solution was prepared by emulsifying it with ethanol. Analytical Procedure. The measured sulfur (ng) was obtained by dividing the measured peak area (counts) by the K (8.909 × 103 counts/sulfur ng, described in the next section), and the recovery (%) was calculated from the measured sulfate and the amount of standard sulfur placed onto the strip.The 1-µL Hamilton 7001-NCH syringes were used to add the analytical reagents and the standard solutions.
Atmospheric Aerosol Sampling. To simultaneously measure the sulfur and nitrate size distribution of the atmospheric aerosol, two impactors, which were designed to be identical, were used. The impactor was a single-jet, eight-stage, low-pressure impactor (LPI) with dimensions as designed by Hering et al.18 Despite its advantage of wide range cutoff diameter (0.01-10 µm), the Moudi impactor (Moudi-II, Nanomoudi-II) was not used, because the aerosol collection plate of the Moudi impactor is too large to use for thermal decomposition. The 50% cutoff aerodynamic diameters of each stage of the Hering LPI are as follows: (1) 4, (2) 2, (3) 1, (4) 0.5, (5) 0.25, (6) 0.11, (7) 0.75, and (8) 0.05 µm. The aerosol collection substrates of the stainless steel strip fixed with two clips onto a glass strip holder (26-mm diameter) were mounted in the stages of the impactor. The aerosol sampling rate was 1 L/min. Each sampler was preceded by an acid gas denuder coated with Na2CO3 and glycerine,19 which eliminates the interference from atmospheric acidic gases, such as SO2, NO2, and HNO3. The denuder was a single cylindrical glass tube with an internal diameter of 7 mm and a length of 50 cm. The denuder has a theoretical denuding efficiency of 99% for HNO3 and SO2; experimental values of 98% for HNO3 and 99% for SO2 were obtained. Sulfur in Aqueous Extracts of Impactor Samples. To measure the sulfur size distribution of the ambient aerosol by TD/ SDA, an air sample of only 30-60 L of air is required (sampling periods 30-60 min); however, an IC measurement requires a much larger volume (more than 360-720 L). To provide sufficient aqueous extracts of LPI samples for a comparison of the two methods, the aerosols were collected over 36 h. Small pieces of about 5 × 5 mm length, on which the aerosols were deposited, were cut from the collection substrates of each of the eight stages. They were put into centrifuge tubes, and the aerosol deposited on each stage was dissolved ultrasonically for 10 min with 1 mL of water. The sample solutions were separated by centrifugation into soluble and insoluble fractions. The supernatant (water-soluble fraction) was used as an aqueous extract sample. The extracts used were 750 µL for the IC and 25 µL for the TD/SDA. The IC was performed with a Dionex Qic analyzer model QIC-2. RESULTS AND DISCUSSION SDA Calibration by Standard SO2. The responses of the analytical system (sulfur analyzer-integrator) to a known quantity of standard SO2 were examined. Twenty-five ppm standard SO2 in N2 was directly injected with a syringe into the bottom of the vaporized gas introduction tube. A linear curve from 0 to 10 mL of standard SO2 (equivalent to 0-357 ng of sulfur) that intersected at the origin was obtained, with a correlation coefficient of 0.999. A K of 8.909 × 103 (counts/sulfur ng), the constant of the response of the analytical system calculated from the slope of the calibration line, was obtained. The linearity of the response over the concentrations studied indicated that the transfer of the sulfurcontaining gas produced in the reaction cell into the analyzer, the sulfur measurement, and the peak area counting were essentially quantitative. Preliminary Study. Standard (NH4)2SO4 (AS) and Na2SO4 (SS) solutions, both containing 33.3 ng of sulfur, were placed on (18) Hering, S. V.; Flagan, R. C.; Friedlander, S. K. Environ. Sci. Technol. 1978, 12, 667-673.
Table 1. Sulfur Recoveries from (NH4)2SO4 and Na2SO4 on a Baked and Unbaked Stripa reaction vessel
baked stainless
steelc
stainless steel stainless steel + MnCl2 sold
sulfur recoveryb (%) (NH4)2SO4
Na2SO4
38.5 ( 3.1 (CV ) 8.0%) 26.8 ( 2.4 (CV ) 8.8%) 97.0 ( 4.9 (CV ) 5.0%)
40.5 ( 3.0 (CV ) 7.4%) 33.7 ( 8.0 (CV ) 23.8%) 95.3 ( 5.1 (CV ) 5.4%)
a A total of 33.3 ng of sulfur added; number of replicate was 10. Average ( standard deviation (coefficient of variation). c Roberts and Friedlander’s method.4 d 0.01 N 0.3-0.75 µL.
b
the two types of strips, and their sulfur recoveries were examined (Table 1). In the baked strip, a sulfur recovery of 38.5 ( 3.1% for AS and 40.5 ( 3.0% for SS was obtained. These recoveries were slightly higher than those of 26.8 ( 2.4 and 33.7 ( 8.0% obtained on the unbaked strip, respectively. Under any of the heating conditions described earlier,4,9 values over ∼50% could not be obtained for these sulfates and for these types of strips. The recovery was higher in the baked strip than in the unbaked strip because of the catalytic effect of the surface of the strip; there, some of the metal oxides produced during the baking process may have acted as a catalyst during the thermal decomposition. On the other hand, we assumed that the sulfur analysis obtained by TD/SDA demonstrated that the baked strip was unsuitable for the aerosol collection substrate because of the likelihood of SO2 interference. Therefore, we attempted to use an unbaked strip and a catalyst instead of the baked strip. In crude experiments, a variety of metal (Cr, Mn, Ni, Cu, Fe) oxides and chlorides were estimated as the thermal decomposition catalyst, and good reproducibility and high recovery were obtained by adding an aqueous MnCl2 for the sulfur recovery of SS. Most metal oxides and chlorides, except for MnCl2, needed a higher vaporizing temperature and yielded a noisy spectrum and low sulfur recovery. Using MnCl2, the optimum heating temperature was determined by measuring the sulfur recovery from standard SS. The highest recovery was obtained by passing currents at 2.2 V (∼1.2 A) three times for 0.1 s with an interval of 1 s. Under these conditions, the temperature of the heated spot measured with the infrared thermometer was 780 ( 10 °C. Unless otherwise noted, these conditions were used throughout this study. The 780 ( 10 °C temperature is much lower than 1200 °C for the SS thermal decomposition reported earlier.3,4,9 The relationship between the quantity of MnCl2 and sulfur recovery was examined. In the range of 0.8-1.5 mL of 0.1 mol/L MnCl2, 93-103% recovery was obtained from 200 and 2000 ng of sulfur in SS solutions. A 1.0-µL aliquot of 0.1 mol/L MnCl2 was optimal and was used in the following experiments. Higher recoveries of 97.0 ( 4.9% for AS and of 95.3 ( 5.1% for SS were obtained, as shown in Table 1. Sulfur Blank. The sulfur blank of TD/SDA was measured. A 1-µL aliquot of purified water and 1.0 µL of MnCl2 solution were placed on a stainless steel strip coated with silicon grease. Several experiments of ∼10 replicates were conducted on arbitrarily selected days. All solutions and stainless steel strips were freshly prepared, but a small level of background signals could not be Analytical Chemistry, Vol. 77, No. 14, July 15, 2005
4425
Table 2. Blank of Stainless Steel Strip by TD/SDA Methoda
a
exp. no.
count
sulfur (ng)
1 2 3 4 5 6 7 8 9 10
12 584 13 956 12 745 12 989 13 301 12 726 14 119 12 955 13 068 12 937
1.41 1.57 1.43 1.46 1.49 1.43 1.58 1.45 1.47 1.45
average std dev CV (%)
13 138 515.5 3.9
1.47 0.06 3.9
Detection limit (3 × SD) ) 0.17 S ng.
reduced any further. Typical results of the experiment are shown in Table 2, with averages and standard deviations of 1.47 ( 0.06 ng of sulfur. In earlier reports,4 strips with a blank value of 0.9 ( 0.4 ng of sulfur have been reported. However, it is our conclusion that the blank value of our strip cannot be reduced any further without baking the strip. It was then considered that these backgrounds could not be attributed to the contamination of the analytical process in the laboratory but, rather, to the heating process of the stainless steel strip and to traces of the regents used (MnCL2 and silicon grease). Calibration Curve. Predominant sulfates in urban aerosols, such as SS, AS, and H2SO4 (SA), were examined for calibration. The 0.2-2.0-µL samples of these standard solutions of 25 and 250 ppm sulfate, equivalent to 3.3-167 ng of sulfur, were placed on the strips, and 1 µL of 0.1 mol/L MnCl2 was added to each standard solution spot. The strip was then placed in the reaction cell, dried under a carrier, and heated. These calibrations were conducted with five replicates for each sulfate. As shown in Figure 2, the net integrated area (after subtracting the blank value of 1.31 × 104 counts) is plotted versus the quantity of sulfur placed on the strip; the error bars represent the standard deviation (SD) of each sulfate. Nonlinearity is observed in the lower portion of the calibration curve. This nonlinearity is similar to that obtained by Husar et al.,3 who assumed that the nonlinearity may be attributed to a “tailing effect” at higher concentrations. A dashed line in Figure 2 shows a 100% response of sulfur for the added standard sulfates. In the range of 7.5-167 ng of sulfur of added sulfates, 3.6 ( 0.4 ng of sulfur of approximately constant discrepancy between two lines is observed. In our chromatogram, longer peak tails at a higher concentration range (50-150 ng of sulfur) were also observed; however, the peak was automatically integrated until the chromatogram reaches a baseline level. Then, to interpret the constant discrepancy, it is assumed that there is another process, i.e., an evaporation process relating to the thermal conversion efficiency of particulate sulfur to gaseous sulfur. The other is an after-evaporation process, i.e., the absorption of gaseous sulfur onto an inner surface of the cell or unheated parts of the strip in a dilution or transformation process by a carrier. However, the cause of the nonlinearity is still inexplicable. 4426 Analytical Chemistry, Vol. 77, No. 14, July 15, 2005
Figure 2. Calibration curve for Na2SO4 (O), (NH4)2SO4 (4), and H2SO4 (0). Table 3. Sulfur Recoveries of Sulfates by TD/SDA Method (%) Added Std Sulfur
(NH4)2SO4 Na2SO4 NH4HSO4 K2SO4 MgSO4 CaSO4
16 ng
33 ng
83 ng
167 ng
76 ( 3.5 79 ( 2.9 75 ( 3.3 79 ( 6.0 78 ( 7.0 73 ( 5.3
90 ( 6.9 92 ( 7.0 88 ( 6.3 85 ( 5.4 89 ( 6.2 90 ( 4.7
96 ( 4.2 94 ( 5.2 98 ( 1.6 99 ( 4.0 97 ( 4.7 83 ( 3.1
97 ( 6.4 97 ( 6.0
Added Std Sulfur
H2SO4
15 ng
33 ng
98 ng
163 ng
76 ( 3.5
88 ( 5.8
96 ( 4.0
100 ( 3.5
For predominant sulfates, a least-squares fit of the data between 7.5 and 167 ng resulted in correlation coefficients of 0.9996 and the regression of Y ) 0.0895X - 0.3588 indicated in Figure 2. Response from Various Sulfates. Other sulfates, such as NH4HSO4 (ABS), K2SO4 (PS), MgSO4 (MS), and CaSO4 (CS), are likely to be found in ambient aerosol. Thus, the response of TD/ SDA for standard ABS, PS, MS, and CS containing 17, 33, and 83 ng of sulfur was examined, and the response of the predominant sulfates was compared. The ratios of the measured response versus the calculated response are shown in Table 3. The calculated responses were evaluated, plotting the added standard sulfur to the calibration curve of Figure 2. Except for the high concentration (83 ng) of CS, which was in the range of 17-83 ng of sulfur for these sulfates, the ratio of 0.95-1.04 was obtained. To achieve a higher recovery of sulfates, such as SS and CS, which have high decomposition temperatures, a vaporization temperature in excess of 1150-1200 °C was needed.3,4,7 Our results showed that 780 ( 10 °C is enough to decompose such sulfates to gaseous sulfur with high recoveries and good reproducibility (83 ( 3.1% for CS and 94 ( 5.2% for SS). The addition
Figure 3. Hourly variations of particulate sulfates and nitrates size distribution in the urban atmosphere. Samples were collected in winter season 1994 at Kobe Suma.
of MnCL2 may act as a catalyst and contribute to the high conversion efficiency of the particle sulfur to gaseous sulfur. The recovery experiments for these sulfates showed that TD/SDA can determine sulfur across a very wide range of concentration and with high sensitivity. Determination Range and Detection Limit. Defining the range showing less than 10% of the coefficient of variation (CV) as the determination range, a range of 3.3-167 ng of sulfur for sulfates with 1.8-10.3% CV is obtained. The lowest detection limit (3 times SD of blank) is 0.2 ng of sulfur. At a rate of atmospheric air sampling for LPI of 1 L/min in a 1-h sampling period, the determination range would be 0.06-2.8 µg/m3, and the detection limit would be 0.003 µg/m3 for each impactor stage. The methods make it possible to conduct a photochemical aerosol formation experiment,20 when the sulfur aerosol concentrations are higher than those observed in the urban atmosphere. In this experiment, the profile of the sulfur aerosol formation and growth in the condensation mode and the droplet mode were observed with the high time resolution of 5 min. Measurement of the Sulfate and Nitrate Size Distributions. The sulfate and the nitrate size distributions of the aerosol in the Kobe atmosphere were measured simultaneously. The sulfate concentration is converted from sulfur concentration determined by the TD/SDA, assuming that all of the water-soluble sulfur in urban aerosol is sulfate. An example on 26 January 1994 is shown in Figure 3. During 5 h, three pairs of sulfate and nitrate data were obtained. Sampling time was 50 min at an LPI flow rate of 1 L/min, and then the determination range and the detection limit of each stage is estimated 0.2-10.0 and 0.01 µg/m3 for sulfate and 0.1-40.0 and 0.02 µg/m3 for nitrate.16. In Figure 3, the sulfate and nitrate concentrations (solid line) are normalized by the factor of ∆logDp (width of cutoff diameter of each LPI stage) and indicated together with the lower limit of the determination range (dashed line). The measured sulfate and nitrate concentration levels were sufficiently above the lower limit of each determination range. (19) Appel, B. R.; Winter, A. M.; Tokiwa, Y.; Biermann, H. W. Atmos. Environ. 1990, 24A, 611-616. (20) Yamamoto, M.; Kosaka, H. Proc. 10th Symp. Aerosol Sci. Technol., Muroran, Japan, 1993; Vol. 10, pp 76-78.
Figure 4. Comparison between the sulfur (TD/SDA, Y axis) and the sulfate (IC, X axis) concentration in aqueous extracts of impactor samples. The number indicates the LPI stage number.
It can be seen that both sulfur and nitrate data show two distinct modes of 0.075-0.1 and 0.5-1.0 µm in an accumulation range.1 These two modes may correspond to the so-called condensation mode (smaller diameter) and the droplet mode (larger diameter). The smaller mode we obtained seems to be a rather smaller diameter and relatively higher peak than those reported by Wall et al.21 and John et al.22 A short-term sampling or the meteorological conditions, such as low temperature and low humidity, may be responsible for the differences; however, the reasons are still inexplicable. Sulfur in Aqueous Extracts of Impactor Samples. The TD/ SDA technique for the measurement of sulfur in aqueous extracts of atmospheric aerosol collected with LPI was examined. These size-segregated extracts were also used for sulfate analysis by IC. To provide sufficient amounts of aqueous extracts for both methods, the aerosols were collected over a period of 34 h. Sampling was conducted at Suma, Kobe, on 27 and 28 February 1995. A small piece (about 5 × 5 mm length) of the stainless steel strip, where the aerosols were deposited, was cut with scissors from the strip from each of the eight stages and put into a centrifuge tube. The aerosol deposited on the piece was dissolved ultrasonically for 10 min with 1 mL of water. The sample solutions were separated by centrifugation, and the supernatant (watersoluble fraction) was used as the aqueous extract sample. IC and TD/SDA needed 750 and 10 µL of the extracts needed, respectively. IC was performed with a Dionex Qic analyzer model QIC2. The results are shown in Figure 4; the X axis indicates sulfate (µg/mL), and the Y axis indicates the sulfur (µg/mL), both in an extract of 1 mL. A 1:1 line indicates an equivalent line. Numbers indicate the LPI stage numbers. All samples from the eight stages are scattered above the 1:1 line. These results indicate that the sulfur components may exist, except for the sulfate in the urban aerosol. However, in fine particle size range of below 1.0 µm (21) Wall, S. M.; John, W.; Ondo, J. L. Atmos. Environ. 1988, 22, 1649-1656. (22) John, W.; Wall, S. M.; Ondo, J. L.; Winklmayr, W. Atmos. Environ. 1990, 24A, 2349-2359.
Analytical Chemistry, Vol. 77, No. 14, July 15, 2005
4427
(stages 4-8), sulfur and sulfate concentrations are distributed near the equivalent line, indicating that sulfate is the predominant sulfur (96% of sulfur at stage 4 and 91% at stages 4-8). Kim et al.,11 using pyrolysis-gas chromatography with the FPD method, found that some sulfur-containing components, but not ammonium sulfate, i.e., thiophene-2,2-methylthiophene, 3-methylthiophene, and benzothiazole, originated from asphalt pavement and tire tread. The excess of sulfur rather than sulfate in a coarse particles size range (stages 1-3) may be due to those sulfur components from mechanically generated aerosols. SDA. In this report, all experiments were conducted using FPD as similar as the earlier works.3-11 However, it is considered that any type of SDA is available for determination of the gaseous sulfur, as long as the whole vaporized gases can be continuously measured with detection limit of ∼1 ppb sulfur dioxide, and considered that a pulse UV fluorescent SDA (e.g., TEII model 43CTL), which has a high sensitivity and a simplicity of the operation, should be substituted for the FPD.
4428
Analytical Chemistry, Vol. 77, No. 14, July 15, 2005
CONCLUSIONS Efforts were made to solve problems in the former TD/SDA method and improve the technique. The method has several advantages, such as the potential for rapid and small-size analysis, simple handling, and a low detection limit of 0.2 ng of sulfur. The method is effective for the simultaneous measurement of the sulfate size distribution with the nitrate method16 at a time resolution of about 30-60 min as well as for measuring sulfur in aqueous extracts of size-segregated urban aerosols collected by an impactor. ACKNOWLEDGMENT The author thanks Hiroshi Kosaka, at HIPES, for the worthwhile advice in this work. Received for review December 22, 2004. Accepted May 10, 2005. AC0481093