Article pubs.acs.org/est
Sulfate Aerosol as a Potential Transport Medium of Radiocesium from the Fukushima Nuclear Accident Naoki Kaneyasu,*,† Hideo Ohashi,‡ Fumie Suzuki,‡ Tomoaki Okuda,§ and Fumikazu Ikemori∥ †
National Institute of Advanced Industrial Science and Technology, 16-1 Onogawa, Tsukuba 305-8569, Japan Tokyo University of Marine Science and Technology, 4-5-7 Kounan, Minato-ku, Tokyo 108-8477, Japan § Keio University, 3-14-1 Hiyoshi, Kouhoku-ku, Yokohama 223-8522, Japan ∥ Nagoya City Institute for Environmental Sciences, 5-16-8 Toyoda, Nagoya 457-0841, Japan ‡
ABSTRACT: To date, areas contaminated by radionuclides discharged from the Fukushima Dai-ichi nuclear power plant accident have been mapped in detail. However, size of the radionuclides and their mixing state with other aerosol components, which are critical in their removal from the atmosphere, have not yet been revealed. We measured activity size distributions of 134Cs and 137Cs in aerosols collected 47 days after the accident at Tsukuba, Japan, and found that the activity median aerodynamic diameters of 134Cs and 137Cs in the first sample (April 28−May 12) were 0.54 and 0.53 μm, respectively, and those in the second sample (May 12−26) were both 0.63 μm. The activity size distributions of these radiocesium were within the accumulation mode size range and almost overlapped with the mass size distribution of nonsea-salt sulfate aerosol. From the analysis of other aerosol components, we found that sulfate was the potential transport medium for these radionuclides, and resuspended soil particles that attached radionuclides were not the major airborne radioactive substances at the time of measurement. This explains the relatively similar activity sizes of radiocesium measured at various sites during the Chernobyl accident. Our results can serve as basic data for modeling the transport/deposition of radionuclides.
■
INTRODUCTION The discharge of radioactive materials from the Fukushima Daiichi nuclear power plant (FDNPP) accident,1,2 triggered by the great Tohoku earthquake and tsunami on March 11, 2011, contaminated a wide area of northeastern Japan. From airborne monitorings3 and ground-based surveys,4,5 it has been revealed that the most contaminated area is localized in a narrow zone northwest of the power plant. Meanwhile, with the passage of a migrating cyclone system in the southern part of mainland Japan on March 15, 2011, a plume that included radioactive substances started to flow south−southwest of the power plant and penetrated the deep inland of the Kanto Plain.6 At Tsukuba, a city 170 km southwest of the FDNPP and 60 km northeast of Tokyo (Figure 1), the radiation dose in the air measured at the National Institute of Advanced Industrial Science and Technology (AIST) recorded a maximum value of 1.54 μSv h−1 at around 13:30 local time on March 15, 2011.7 By early April, the reported emission of radioactive substances from the reactor had decreased by about 4 orders of magnitude as compared to that during the massive discharge on March 15.1 However, the radiation dose in the surface air of the Kanto Plain had not decreased as such. For example, the reading at Tsukuba on April 28, 2011 (0.19 μSv h−1)7 was still 1/8 of the maximum value recorded on March 15, 2011. A question raised by the Japanese public was whether this was attributed solely to radiation from the ground surface that had © 2012 American Chemical Society
Figure 1. Map of Japan and the area that is concerned in the discussion of radioactive aerosol from the Fukushima Dai-ichi Nuclear Power Plant (FDNPP).
accumulated the radioactive fallout or also to the resuspended soil particles that had adsorbed radioactive nuclides.8 However, since the occurrence of the FDNPP accident, there has been little observational data on the radioactivity of airborne soil Received: Revised: Accepted: Published: 5720
December 26, 2011 April 17, 2012 April 25, 2012 April 25, 2012 dx.doi.org/10.1021/es204667h | Environ. Sci. Technol. 2012, 46, 5720−5726
Environmental Science & Technology
Article
After the γ-ray spectrometry analysis, the ionic species in the aerosols such as chloride (Cl−), nitrate (NO3−), sulfate (SO42−), sodium (Na+), ammonium (NH4+), potassium (K+), magnesium (Mg2+), and water-soluble calcium (Ca2+) were extracted from the aliquots of the substrates (1/8 of the quartzfiber filter and 1/4 of the aluminum sheet) by using 20 mL of ultrapure water (resistivity >18.2 MΩ cm), filtered using a syringe filter (pore size, 0.45 μm), and determined by ion chromatography (DIONEX, IC-2000 and IC-1000). The analytical errors were estimated to be within 3% for SO42−, Na+, and NH4+, and within 6% for NO3−. The detection limits for SO42−, NO3−, Na+, and NH4+ were estimated to be 0.26, 0.58, 0.16, and 0.06 μg, respectively. Non-sea-salt (nss.) SO42− concentrations were calculated from the measured SO42− and Na+, assuming the SO42−/Na+ ratio of sea salt to be the same as that of the surface seawater. Silica (Si) and total calcium (Ca) in the aerosols were determined using an X-ray fluorescence spectrometer (EDXL300, Rigaku). Each aliquot of the filters was analyzed for 15 min to obtain a spectrum of X-ray counts versus photon energy, with the individual peak energies matching specific elements and the peak areas corresponding to the elemental concentrations. The amounts of the elements were calculated using a fundamental parameter method (Rigaku Profile Fitting Spectra Quant X). The analytical errors for Si and Ca were estimated to be within 16% and 8%, respectively. The detection limits for Si and Ca were 0.2 and 0.6 μg, respectively. Water-soluble organic carbon (WSOC) was ultrasonically extracted from other aliquots of the substrates (1/8 of the quartz-fiber filter and 1/4 of the aluminum sheet) with ultrapure water (10 mL). The extract was filtered through a syringe filter (pore size, 2.0 μm). The concentration of WSOC was determined using a total carbon analyzer (TOC-V CHP, Shimadzu). The analytical error was estimated to be within 5%, and the detection limit was to be 0.1 and 0.5 μg C for aluminum and quartz-fiber filter substrates, respectively. After the WSOC analysis, thermally separated elemental carbon (thermal-EC) and organic carbon (thermal-OC) were analyzed from the remaining substrates of the LPI-AIST-1 sample. Details of the analytical method are the same as those described elsewhere,12 except that OC and EC were separated at 340 °C in a pure oxygen flow for 2 h.13
particles, or more generally, the size distribution of radioactive aerosols. Data on the size distribution of radionuclides and their mixing state with other aerosol components are critical for evaluating the inhalation dose and for modeling their transition from the atmosphere to the ground or water surface (i.e., their deposition processes). In the dry deposition process, the aerodynamic diameter of the particle is an essential parameter. In detailed wet deposition model studies, which distinguish the in-cloud rainout and below-cloud washout processes, the mixing state of the radionuclides with other hygroscopic aerosol components is of chief concern. With these issues in mind, we have been collecting sizeresolved aerosols in Tsukuba since April 28, 2011. Although the data obtained are not from the first radioactive plumes that swept over Tsukuba on March 15, 2011, they should add insights into the state of fission product radionuclides in the surface air and the mechanism of their fallout. The results obtained in this study help interpret the activity sizes of radionuclides during the Chernobyl accident, in addition to the geographical distribution of the FDNPP derived radionuclides. They may also affect the dose risk evaluation for local residents via inhalation.
■
SAMPLING AND MEASUREMENTS A low-pressure cascade impactor (12-stages with a backup filter, Tokyo Dylec LP-20) was operated on the fourth-floor balcony of a building (15 m from the ground surface) in the AIST west campus. The cutoff values for the impactor at 50% collection efficiency were 11, 7.8, 5.2, 3.5, 2.1, 1.2, 0.7, 0.49, 0.3, 0.2, 0.12, and 0.06 μm aerodynamic diameters. As impaction substrates, 12-μm-thick aluminum foil sheets were used for the first sample (LPI-AIST-1), and 0.5-mm-thick quartz fiber filters (Pallflex 2500 QAT-UP) were used for the second sample (LPI-AIST2). As a backup filter, a quartz fiber filter was placed after the 12th stage. The use of a fibrous filter for the impaction substrate may affect the apparent size distribution owing to the roughness of the surface and the penetration of the aerosol jet into the filter surface.9 The anticipated effect will be discussed later. For convenience in deriving the size distributions, the lower size limit of aerosol collection on the backup filter and the upper size limit of aerosol transmission through the impactor inlet were assumed to be 30 nm and 30 μm, respectively. The upper size limit was adopted from a previous study that used the same impactor,10 whereas the lower limit was assigned to a value close to that used in a data inversion algorithm study for a cascade impactor.11 The sampling periods for the LPI-AIST-1 and LPI-AIST-2 samples were from April 28 to May 12, 2011, and from May 12 to 26, 2011, respectively. The radioactivity of cesium isotopes 134Cs and 137Cs in the aerosols collected at each stage was determined by γ-ray spectrometry. Each sample was enclosed in a plastic (body: polystyrene; lid: polyethylene) container (16 mm diameter × 100 mm long) and was inserted into a well-type germanium detector (CANBERRA GCW3523). The photon counting time was typically 400 000 s, and the counting errors for 134Cs (604 keV) and 137Cs (661 keV) were less than 10%. A spectrum analyzing software (Spectrum Explorer, CANBERRA) was used for the quantification of 134Cs and 137Cs from the measured γray spectrum. Short-lived radionuclides such as iodine isotope 131 I were not detected because the radioactivity analysis was conducted after July, i.e., four months after the accident.
■
RESULTS AND DISCUSSION Activity Size Distributions of 134Cs and 137Cs. From the size-fractionated activities of 137Cs and 134Cs in the first LPIAIST-1 sample (Figure 2A), the radioactivity was found to be concentrated mostly in the accumulation mode size range (diameter between 0.1 and 2 μm), and it was found to be very weak in the coarse mode size range (diameter >2 μm). The radioactivity of 137Cs had a distinct maximum at the 0.49−0.7 μm stage, with a minor mode at the 0.2−0.3 μm stage. 134Cs also exhibited a major mode at the 0.49−0.7 μm stage, with a less prominent minor mode at the 0.2−0.3 μm stage. The activity median aerodynamic diameters (AMADs) of 134Cs and 137 Cs were 0.54 and 0.53 μm, respectively. In addition to the step-like size distributions, Twomey nonlinear iterative inversion11 was applied to the cascade impactor data to obtain smooth size distributions. In the retrieved curves (Figure 2A), the minor mode was more clearly visible. The aerodynamic diameters of the first and second activity modes in the retrieved curves were 0.23 and 0.67 μm for 134Cs and 137Cs, respectively. 5721
dx.doi.org/10.1021/es204667h | Environ. Sci. Technol. 2012, 46, 5720−5726
Environmental Science & Technology
Article
substrates used in the low-pressure impactor sampling tended to shift the apparent size distribution in the accumulation mode size range to a larger size side than that of the flat-surface substrate. Therefore, we cannot exclude the possibility that there was actually a smaller mode in the LPI-AIST-2 sample, but it was merged into the larger mode. If we convert the measured radioactivities of the collected nuclides at all stages to the mass of radiocesium, the resultant mass concentrations will become extremely low. For example, the radioactivity of 137Cs in the LPT-AIST-1 sample is equivalent to 2 fg m−3. Thus, the measured size distributions appear to substantiate that radiocesium was attached to or included in other aerosols (i.e., it was internally mixed with other aerosol components). Mass Size Distributions of Major Aerosol Components. To obtain information on the mixing state of radiocesium with other components, the mass size distributions of major aerosol components were measured from the impactor samples. In the case of the mass size distribution of nss.SO42−, NO3−, Na+, NH4+, and Si in the LPI-AIST-1 sample (Figure 3A), the distinct maximum in the size distribution of nss.SO42− and its mass median aerodynamic diameter (MMAD, 0.53 μm) were almost identical to the maximum and AMADs, respectively, of the 137Cs and 134Cs activity size distributions. The size distribution of NH4+ follows that of nss.SO42−, indicating that nss.SO42− is in the form of ammonium salt. The mass size distributions of nss.SO42−, NO3−, Na+, NH4+, and Ca in the LPI-AIST-2 sample (Figure 3B) showed the same general features as those of LPI-AIST-1, except that the righthand side shoulder of the nss.SO42− and NH4+ maxima, i.e., the measured mass at the 0.7−1.2 μm stage, became more pronounced. In the case of the size distribution of NO3− in the LPI-AIST-1 sample, although a minor mode at the 0.49− 0.7 μm stage was identified, the mass of NO3− was mostly distributed within the coarse mode size range (Figure 3A). In addition, NO3− in the accumulation mode size range disappeared completely in the LPI-AIST-2 sample (Figure 3B). This implies that a relatively small amount of NH4NO3, which is in the accumulation mode size range, could exist when the LPI-AIST-1 sample was collected, whereas NH4NO3 could not be formed at the increased ambient temperature when the LPI-AIST-2 sample was collected. The mass size distributions of Si and Na+ in the LPI-AIST-1 sample were mainly in the coarse mode, suggesting that 137Cs and 134Cs are not associated with these substances. In other words, very little radiocesium was attached to or included in the soil and sea-salt particles. This characteristic is the same for Ca and Na+ in the LPI-AIST-2 sample. It should be noted that as a tracer of soil particles, Ca was used as the surrogate of Si for the LPI-AIST-2 sample owing to the large blank value of Si in the substrate material. We thus conclude that, at least around Tsukuba, resuspension of soil particles that had adsorbed radiocesium was not the dominant source of airborne
Figure 2. Activity size distributions of radiocesium at Tsukuba. (A) Distribution of 134Cs and 137Cs determined from LPI-AIST-1 samples collected from April 28 to May 12, 2011. (B) Distribution in the LPIAIST-2 samples collected from May 12 to 26, 2011. Smooth distributions retrieved by the Twomey nonlinear iterative algorithm11 are also shown in each panel with thin lines, indicating the bimodal structure of the activity in the accumulation mode size range for the LPI-AIST-1 sample.
The activities of radiocesium in the second LPI-AIST-2 sample (Figure 2B) showed the same maximum size in the 0.49−0.7 μm stage, although the minor mode disappeared. The modes of the retrieved curves and AMADs were both 0.63 μm for 134Cs and 137Cs, respectively. These size parameters are summarized in Table 1, with standard deviations of the retrieved modes that are fitted to the log-normal functions. With regard to the difference in the activity size distributions between LPI-AIST-1 and -2, a possible effect due to the adoption of different impaction substrates must be considered. From the previous studies,14,15 it was found that filter-type
Table 1. Size Parameters of the Radiocesium and Non-Sea-Salt (nss.) SO42−a retrieved first mode Da (SD)
AMAD sample ID
sampling period
LPI-AIST-1 LPI-AIST-2
Apr. 28 − May 12 May12 − May 26
134
Cs
0.54 0.63
137
Cs
0.53 0.63
134
Cs
0.23 (1.4)
137
Cs
0.23 (1.4)
retrieved second mode Da (SD) 134
Cs
0.67 (1.3) 0.63 (1.4)
137
Cs
0.67 (1.3) 0.63 (1.4)
nss.SO42−MMAD 0.53 0.63
a Unit: μm. Abbreviations: AMAD = activity median aerodynamic diameter; Da = aerodynamic diameter; MMAD = mass median aerodynamic diameter. Standard deviations (SD) of the modes in the retrieved curves are calculated by the fitting of log-normal functions.
5722
dx.doi.org/10.1021/es204667h | Environ. Sci. Technol. 2012, 46, 5720−5726
Environmental Science & Technology
Article
Figure 4. Daily maximum and daily maximum momentary wind velocities (left Y-axis) and daily precipitation (right Y-axis) at Tsukuba (Tateno station) from April 27 to May 26, 2011.
Figure 3. Mass size distributions of major aerosol components. (A) Non-sea-salt (nss.) SO42−, NO3−, Na+, and NH4+ determined by ion chromatography (left Y-axis scale) and Si determined by X-ray fluorescence spectroscopy (right Y-axis scale), from the LPI-AIST-1 aerosol sample. (B) Same as (A) except that Ca is used instead of Si determined from the LPI-AIST-2 aerosol sample. Ca is used because of the large blank values of Si attributed to the use of quartz-fiber filters as impaction substrates.
Figure 5. Mass size distributions of WSOC in the LPI-AIST-1 (solid line) and LPI-AIST-2 (broken line) aerosol samples.
The bimodal nature of the WSOC mass size distribution, i.e., the main peak in the accumulation mode size range and the minor peak in the coarse mode size range, has been reported in previous studies.17,18 Thermal-EC and OC were only analyzed for the LPI-AIST-1 sample, because all of the remaining sections of the LPI-AIST-2 sample substrates were used for the analysis of WSOC. In the case of the LPI-AIST-1 aerosol sample, thermal-EC and OC both show local modes in the accumulation mode size range (Figure 6). It should be noted that the large value of thermal-OC in the smallest size should include the adsorption of gaseous organic species onto the backup quartz-fiber filter, and the local mode of the thermal-EC in the coarse mode size range is interpreted as the contribution of carbonate (CO3) in the soil particles. Potential Transport Medium of Radiocesium from the FDNPP Accident. The coincidence of the activity size distributions of radiocesium and the mass size distribution for nss.SO42− (and its counterion NH4+) are both observed in the LPI-AIST-1 and -2 samples. On the other hand, a small peak of NO3− in the accumulation mode size range was only identified in the LPI-AIST-1 sample, and the maximum of WSOC in the accumulation mode size range was apparent only in the LPIAIST-2 sample. From these findings, we suggest that these radionuclides are mainly attached to or included in the sulfate salt (ammonium sulfate and ammonium bisulfate) aerosol. In other words, sulfate salt aerosol can be regarded as a potential transport medium of radiocesium from the FDNPP accident.
radiocesium during the sampling period between April 28 and May 27. If we collected the aerosol samples at a lower height, e.g., 1.5 m from the ground surface, the concentration of the soil particles could be higher by several factors. Even in that case, radiocesium in the coarse mode size range could merely be a minor contributor to the airborne radioactive particles in view of the measured amount. Nevertheless, if the concentration of the soil particles increased by about 2 orders of magnitude, as in the case of the local dust-storm events in spring, the resuspension of soil dust could be a main contributor to the airborne radiocesium. During the sampling period, such local dust-storm events did not occur owing to the intermittent precipitation and the resultant relatively wet surface condition, irrespective of the occasional high surface wind velocities (Figure 4). Carbonaceous components are known to comprise a curtain fraction of the aerosol mass. Among those components, WSOC is often referred to have a close relation with secondary formed organic aerosols (SOA)16 and is expected to exist ubiquitously. WSOC in LPI-AIST-1 and LPI-AIST-2 showed different size distributions (Figure 5); that of LPI-AIST-1 showed no obvious mode, whereas that of LPI-AIST-2 exhibited a bimodal shape. 5723
dx.doi.org/10.1021/es204667h | Environ. Sci. Technol. 2012, 46, 5720−5726
Environmental Science & Technology
Article
EC (or black carbon: BC), which is a typical combustionderived primary particle, has been studied in view of the removal/deposition mechanism of EC and the enhancement of its light-absorbing ability. However, EC itself is not likely the transport medium of radiocesium from the FDNPP accident. This is because flaming fires did not continue during the FDNPP accident, except for a few small fires at the time of hydrogen explosions of containment buildings. This course of progression differs from that observed in the case of the Chernobyl accident, where the reactor core made of graphite was continuously burning in the open air during the accident. In addition, large-scale forest fires were not reported around the FDNPP and in the northern Kanto Plain, and the local residents have refrained from the burning of firewood, fallen leaves, and branches of trees in fear of the re-emission of radionuclides25 since the accident. If the measured thermal-OC mainly consists of SOA, it is plausible that it is one of the transport media of radiocesium. It is unlikely for the primary particles to serve as the transport media unless strong phenomena such as the burning of materials occur concurrently. By comparing Figures 5 and 6, we can observe that a majority of the fraction of the thermal-OC in the LPI-AIST-1 sample is in the water-insoluble state. Although SOA has been discussed with regard to its water-soluble characteristic,16 the formation of water-insoluble SOA was reported in a dry condition.26 At present, we cannot exclude the possibility that water-insoluble OC also acted as a transport medium of radiocesium. An issue to be taken into account in this measurement is that the sampling periods of each size-fractionated aerosol were fairly long. At present, we have no data on the temporal change in the size and concentration of the aerosol components during each of the 2-week-long sampling periods. The measured size distributions of sulfate were probably the overlapped ones of areal air pollution in the Kanto Plain and those from the direction of FDNPP. In fact, the double peak structure shown in Figure 2A is not clearly seen in the accumulation mode sulfate shown in Figure 3A. Therefore, we withhold a decisive conclusion that sulfate aerosol is the only transport medium of radiocesium until the time-series data for both the radionuclides and the aerosol chemical components, if they exist, are reported. Comparison with the Chernobyl Accident Data. Following the Chernobyl nuclear accident, the activity size distributions of 137Cs and several other radionuclides measured at Zurich, Switzerland, showed concentration patterns very similar to those of NH4+, SO42−, and NO3− (data not shown).27 Our data agree with this result, except that the size distribution of NO3− at Tsukuba was not similar to that of radiocesium. In the case of Chernobyl radionuclides measured at Zurich, we presume that NO3− was added locally as an urban air pollutant in the form of NH4NO3 that existed in the accumulation mode size range. The reported parameters of the activity size distributions of radiocesium, namely the AMAD, that experienced long-range transport at the time of the Chernobyl accident, were in a relatively narrow range. For example, the AMADs of 137Cs were 0.71 μm at Zurich, Switzerland,27 0.83 μm at Göttingen, Germany,28 0.63 μm at Helsinki, Finland,29 and 0.68 μm in a parametrization study30 (Figure 7). Time series measurements at Tennessee (arithmetic mean of four values: 0.57 μm)31 and at Tsukuba (arithmetic mean of four values: 0.51 μm)32 include smaller AMADs than those in our study. These sizes appeared
Figure 6. Mass size distributions of thermal-EC (solid line) and OC (dotted line) in the LPI-AIST-1 aerosol sample.
The bimodal shape observed in the retrieved curves of the Cs and 134Cs activity size distribution (Figure 2A) can be explained by the commonly noted characteristics of differently aged sulfate aerosols. That is, sulfate aerosols sometimes show a bimodal distribution in the accumulation mode size range; the larger (i.e., aged) mode is known as the “droplet mode”.19,20 The formation of this mode has been explained by the growth of pre-existing sulfate aerosols through inhomogeneous reactions, i.e., their activation to a cloud or fog droplet, absorption of sulfur dioxide (SO2) and its aqueous reaction to form SO42−, and subsequent evaporation. It is apparent that the size distributions of the sulfate aerosols collected in Tsukuba are dominated by the droplet mode size particles during the sampling period. This means that the cloud/fog processing of the pre-existing sulfate played a substantial role in maintaining the size distribution of the sulfate aerosol at this time of the year in the eastern Kanto Plain. In a previous study on the size distribution of the radon (222Rn) decay long-lived product 210Pb and ambient aerosols, it was found that the AMAD of 210Pb was significantly less than the MMAD of ambient aerosols, and it was greater than or equal to their surface median aerodynamic diameter.21 Our results on the activity distribution of fission products are different from that of such naturally originating long-lived radionuclides, which generally follow the surface area distribution of pre-existing aerosols. If the sulfate aerosol acted as the main transport medium of FDNPP-derived radiocesium to Tsukuba, the incorporation of radiocesium into the sulfate aerosol should have occurred in the early formation stage of sulfate, i.e., in the Aitken (or nucleation) mode size range (diameter