ISOTOPES AS TRACERS OF SOURCES OF LEAD AND STRONTIUM IN AEROSOLS ( TSP & PM 2 . 5 ) IN BEIJING

Even after its being phased out in gasoline in the late 90’s, lead (Pb) is still present at relatively high levels in the atmosphere of Beijing, China (0.10-0.18 μg.m). Its origin is subject to debate as several distinct sources may contribute to the observed pollution levels. This study proposes to constrain the origin(s) of Pb and strontium (Sr) in aerosols, by coupling both Pb and Sr isotope systematics. The characterisation of the main pollution sources (road traffic, smelters, metal refining plants, coal combustion, cement factories, and soil erosion) shows that they can unambiguously be discriminated by the multi-isotope approach (Pb/Pb and Sr/Sr). The study of total suspended particulates (TSP) and fine particles (PM2.5) from Beijing and its vicinity indicates that both size fractions are controlled by the same sources. Lead isotopes indicate that metal refining plants are the major source of atmospheric lead, followed by thermal power stations and other coal combustion processes. The role of this latter source is confirmed by the study of strontium isotopes. Occasionally, emissions from cement plants and/or input from soil alteration are isotopically detectable.


INTRODUCTION
High concentrations of fine particles are found in the air of big cities, which can be up to 300 µg.m -3 for PM 10 particles (Particulate Matter with a diameter <10 µm; Seinfeld and Pandis, 1997).
As fine atmospheric particles (e.g. PM 2.5 ) have a damaging effect on public health (e.g. Kappos et al., 2004), they have recently become a cause of major concern. Isotope compositions have proved to be reliable tracers of the origin of aerosols in the atmosphere, including urban air (e.g. Sturges et Barrie, 1989;Monna et al., 1997). Natural variations of selected isotope compositions (e.g. carbon (δ 13 C), nitrogen (δ 15 N)) and isotope ratios (e.g. 206 Pb/ 204 Pb,207 Pb/ 206 Pb,87 Sr/ 86 Sr) can provide clearer evidence for anthropogenic input compared to concentrations data alone, as the different anthropogenic sources, whether point-source or of regional scale, commonly show distinct characteristic isotope compositions compared to those found in natural sediments (e.g., Hamelin et al., 1990;Monna et al., 1997). The signatures of "heavy" isotope may ultimately be related to the isotope geochemistry of the ore deposit from which the industrial element was produced (Sangster et al., 2000). In the case of Pb air pollution, impact studies were supported, since the 1980s (e.g. Mukai et al., 1993), by the well defined isotopic composition of automobile emissions, the major source of lead in the ambient air (Nriagu, 1990).
However, since 1975 in the US, the mid-1980s in Europe and 2000 in China, legislative measures have passed to reduce and eventually eliminate Pb in gasoline. As a consequence of the phasing-out of Pb, the abundance of Pb in troposphere has decreased world-wide, but Pb remains essentially anthropogenic in atmospheric aerosols (e.g. Widory, 2006). Mukai et al. (2001) coupled sulphur and Pb isotopes to characterise the atmosphere of several Chinese urban sites. Lead isotope ratios suggested that coal combustion considerably contributed to atmospheric Pb in some cities in China. At the same time, influences by the emission of Chinese lead ores were also observed in northern Chinese cities. Seasonal variations of Pb isotope ratios indicated the existence of a certain amount of industrial sources other than coal combustion (Mukai et al., 2001).
Studies of atmospheric aerosols using Sr isotopes have mainly been carried out for the soluble components of aerosols in rainwater (Herut et al., 1993;Nakano and Tanaka, 1997). However, Sr is a soil constituent element, and water-soluble Sr is only a part of the total Sr content of an aerosol. Kanayama et al. (2002) studied 87 Sr/ 86 Sr ratios coupled with both major and trace elements chemistry to track the contribution of long-range-transported Asian dust (Kosa) on the overall atmosphere of Japan. Recently, Xu and Han (2009) monitored the monthly 87 Sr/ 86 Sr ratios of rainwater in Beijing over a period of one year to estimate the inputs of sea salt ( 87 Sr/ 86 Sr=0.70917; Dia et al., 1992) and terrestrial elements, as well as anthropogenic emissions.
The city of Beijing in general, and particularly during the preparation for the 2008 Olympics, has been trying hard to improve theist air quality. Leaded gasoline in the city has been phased-out since 1998, but even if a slight decrease in its atmospheric concentrations has been observed, the levels remain critically high (0.10-0.18 μg.m -3 in [2005][2006]. Industrial emissions, particularly from the non-ferrous industry as well as coal-combustion are the usual suspected vectors of pollution (Xiao et al., 2008;Li et al., 2008), but so far the classical chemical methods for assessing Pb levels have proven limited in their ability to determine the respective sources' contributions.
The present study assesses the use of coupling both lead ( 206 Pb/ 204 Pb) and strontium ( 87 Sr/ 86 Sr) isotope systematics to help determinate the origin of aerosols (TSP and PM 2.5 ) in the atmosphere of Beijing (China). Potential sources of pollution in the atmosphere of Beijing, such as cement factories, coal combustion and Pb refining, were first chemically and isotopically characterized. Then, through the isotope characterisation of ambient aerosols, the major sources of pollution were identified and, when possible, their respective contributions to particulate-matter contents appraised.

MATERIALS AND METHODS
The major sources of aerosol pollution in the atmosphere of Beijing, analysed during this study, include: coal combustion (4 samples), cement factories (3 samples) and smelters (1 sample). Chemical and isotope characteristics from road traffic were taken from the literature (Chen et al., 2005;Widory et al., 2007b). Wind transport of aerosols from surrounding deserts and non-polluted rural areas were also taken into account (2 standardised loess and 3 soils from different locations in China were sampled and analysed). A total of 63 samples of ambient PM 2.5 were collected at the Sino-Japan Beijing and 3) Liangxiang (31 km southwest of Beijing). These 23 TSP samples were part of a larger of particles was made on quartz filters (QMA). PM 2.5 were sampled during runs of 24 hours at a constant flow of 1 m 3 h -1 , using a low volume sampler, while TSP samples were collected using a high volume sampler (runs of 24 hours).
In the laboratory (in a class-10 000 clean room with class-100 laminar flow hood), samples (filters in the case of PM 2.5 and TSP, and solids and particles in the case of the different inputs end-members) were leached with Supra-pure hydrobromic acid (HBr). The efficiency of Pb (& Sr) complexation by HBr is generally better than 90 %, confirming that the major part of these elements is contained in the labile fraction (e.g. Widory, 2004b and. Pb & Sr concentrations were measured with an Inductively Coupled Plasma Mass Spectrometer (ICP-MS; precision 5%). Blank measurements made on quartz sampling filters yielded relatively low Pb and Sr contents (3.9±0.7 and 0.7±0.05 ng, respectively), compared to those obtained from ambient air samples and pollution sources (lower by at least a factor 15).
Pb-isotope measurements were measured on a Neptune Multiple Collector Inductively Coupled Plasma Mass Spectrometer (MC-ICP-MS; Thermo-Finnigan). The procedure, described in White et al. (2000), was slightly modified in order to certify some key features of the instrument's doublefocusing system (Motellica-Heno et al., 2003). Due to the physical principles of this machine, a rather high mass fractionation is observed (~0.7%/amu). Use of thallium (Tl), an element of similar mass, as an internal standard (Longrich et al., 1987) provides an opportunity for correction of mass dependency of the instrumental bias. Despite the fact that the fractionation coefficients are not strictly identical for Tl and Pb, the ratio of these coefficients remains constant for a given analytical session (White et al., 2000). Furthermore, Tl normalisation seems to reduce a potential matrix effect. In addition, samplestandard bracketing was applied to control potential random instability of the instrument as a whole, resulting in the following sequence: Blank -NIST 981 -Cleaning solution HNO 3 , 3% -Blank -Sample -Cleaning solution HNO 3 , 3% -Blank -etc. The blank level obtained is very low (<1 mV on 208 Pb) and is subtracted from the next sample in the sequence. For this environmental study, samples were diluted to 50 to 100 ppb of Pb in solution, and 20 to 30 ng of Tl were added to 1 ml of solution.
Baselines were measured at mass 100 at the beginning and end of each analysing sequence. A correction for 204 Hg used a 202 Hg measurement, applying a 202 Hg/ 204 Hg ratio of 4.35. For routine analyses following this procedure, avoiding any chemical lead purification, the uncertainty level was 0.01% for 208 Pb/ 206 Pb and 207 Pb/ 206 Pb, and 0.1% for 206 Pb/ 204 Pb, 207 Pb/ 204 Pb and 208 Pb/ 204 Pb.
Chemical purification of Sr (~3 µg) was performed, on the HBr leachate, using an ion-exchange column (Sr-spec) before mass analysis according to a method adapted from Pin and Bassin (1992), with a total blank of less than 1 ng for the entire chemical procedure. After chemical separation, around 150 ng of Sr was loaded onto a tungsten filament with tantalum activator and analysed on a MAT 262 Thermo-Ionisation Mass Spectrometer (TIMS; Finnigan).The 87 Sr/ 86 Sr ratios were normalised to a 87 Sr/ 86 Sr ratio of 0.1194. An average internal precision of ±10 ppm (2σ) was obtained and the reproducibility of the 87 Sr/ 86 Sr measurements was tested through repeated analyses of the NBS987 international standard, for which a mean value of 0.710243±0.000022 (2σ; n=13) was obtained during the period of analysis.

Characterization of potential end-members
Both Pb and Sr concentrations from pollution sources display large fluctuations (Table 1) (Chen et al., 2005). The use of leaded gasoline was first banned in Beijing in 1997, and then in Shanghai, Guangzhou, Tianjin and other big cities (Sun et al., 2006). Chen et al. (2005) report Pb concentrations (77 to 399 ppm) and 207 Pb/ 206 Pb (0.862 to 0.879) in unleaded vehicle exhausts. However, the authors were not able to measure the 204 Pb content, using their particular ICP-MS technique.
The "natural" sources of aerosols measured in this study include loess and soils collected at various locations in China (Table 1). Slight variations are observed in both their Pb and Sr concentrations, from 24 to 41 ppm (average of 32±7 ppm) and 242 to 333 ppm (average of 286±36 ppm), respectively.
No clear chemical discrimination is apparent between the loess and the soils. Their corresponding 206 Pb/ 204 Pb ratios form a somehow compact range, from 18.312 to 18.693, with an average of 18.459±0.17. This is consistent with the values of 18.671±0.006 reported for soils by Jones et al. (2000). 87 Sr/ 86 Sr, in loess and soils, range from 0.7118 to 0.7148 (average of 0.7136±0.0012), similar to previous studies. Zhang et al. (1995) reported 87 Sr/ 86 Sr ratios of 0.7111-0.7154 from rivers around the Taklimakan Desert in northwest China, similar to those measured at both the Huanghe river (Yellow River;0.7111;Palmer and Edmond, 1989), and the Central Loess Plateau (0.7111; Yokoo et al. 2004).

Characterisation of ambient air samples
While certainly not representative of its overall budget in Beijing, data for TSP concentrations in the air (Figure 2 Table 1, they fall into the range of emissions from coal combustion, lead refining plants and cement factories. The PM 2.5 and TSP samples taken on the 3 rd of January yield 206 Pb/ 204 Pb ratios that are closer to those measured in the emissions from the smelter. Strontium, on the other hand, is in significantly greater abundance in the coarser mode ( Figure   4A), with an average concentration of 215±196 ppm (variations from 90 to 958 ppm). In PM 2.5 , the Sr concentrations fluctuate from 11 to 185 ppm, with an average of 90±49 ppm. This pattern is similar to what has been previously observed in other cities (e.g. Widory et al., 2007b). The corresponding 87 Sr/ 86 Sr ratios ( Figure 4B) vary from 0.7086 to 0.7107 (average of 0.7097±0.0005) in the TSP, and from 0.7092 to 0.7101 (average of 0.7097±0.0003) in the PM 2.5 . This overlap between the ranges found in both size fractions may argue for similar sources, but the isotope disparity clearly indicates at least two distinct sources. The comparison with results from Table 1 shows that emissions from both coal combustion and cement factories have similar 87 Sr/ 86 Sr ratios compared to those measured in the aerosols. Still, for the lowest Sr isotope ratios measured (close to 0.7086), the input from another unidentified source is required (see below). Variations of Sr and Pb concentrations in both TSP and PM 2.5 seem to be roughly linear (with opposite slopes for PM 2.5 and TSP, negative and positive, respectively; Figure 5). While at low Pb concentrations the trends for TSP and PM 2.5 merge, perhaps indicating a single origin for both size fractions in the atmosphere, at high Pb concentrations, at least two distinct sources seem to be required. At high Pb concentrations, the source of PM 2.5 is characterised by a high Pb content coupled with a relatively low Sr content (<100 ppm), while the source of TSP is characterised by high Pb and Sr concentrations. This conclusion is in agreement with the fact that PM 2.5 samples are expected to be mainly influenced by activities such as lead refining, while the coarser TSP fraction incorporates inputs from activities such as coal combustion or emissions from cement plants (Zhang et al., 2007). Two TSP samples taken on the 28 th of January 2006 plot outside these trends (one is from downtown Beijing, while the other is from Changping, suburb of Beijing). They coincide with the Chinese New Year's Eve festivities, which would explain the high Sr concentrations measured (>600 ppm), as Sr is commonly used as a colouring compound in fireworks. Even though Pb has been phased-out in Beijing since 1997, it is still present (in low quantities) in the unleaded gasoline burnt today in the city, and thus may represent a potential source of contamination (Wang et al., 2003). Figure 6A compares the 207 Pb/ 206 Pb measured in both TSP and PM 2.5 with those characterising the different sources of particles in the air (including ranges of road traffic emissions measured by Chen et al., 2005). For the majority of samples, results confirm that, road traffic is no longer a major source of atmospheric Pb, which was expected following its phasingout in gasoline. Only two samples, taken consecutively, one TSP sample taken in Changping  Figure 6B) shows that all Pb isotope ratios measured in the ambient air samples (both TSP and PM 2.5 ) plot within the ranges obtained on aerosols from the main pollution sources noted above (Table 1), indicating that no other pollution source is required for explaining our aerosols sample set.
Still, the use of the sole Pb isotopes is not sufficient for precisely tracing the origin of these elements in the aerosols (both TSP and PM 2.5 ). Figure 6C plots the more discriminating 206 Pb/ 204 Pb ratio versus the reciprocal of Pb concentrations in the samples. Two distinct emitters appear to play a major role in the observed Pb pollution levels: i) Pb refining plants (to which most of the aerosols samples are closely associated), and to a lesser extent ii) the Jingneng coal-fired power station, located southeast of Beijing. Two samples taken on the 28 th of January 2006, both TSP (downtown Beijing and Changping sites) and PM 2.5 (Beijing site) display the most radiogenic 206 Pb/ 204 Pb ratios ( Figure 6C). Their date of sampling and Sr contents (>600 ppm) may corroborate the influence of fireworks residues, used during the New Year's Eve festivities. This is consistent with conclusions drawn from Figure 5, but this has yet to be confirmed by measurements, as to our knowledge no studies have ever reported Pbisotope ratios for lead nitrate (Pb(NO 3 ) 2 ), used in fireworks and other pyrotechnics. As discussed above, samples taken on the 4 th (PM 2.5 ) and 5 th (TSP Changping) of January may indicate an input from road traffic ( Figure 6A), or more probably, the influence of smelter emissions due to their relatively high Pb concentrations (1145 and 621 ppm, respectively). Finally, one TSP sample taken in Changping (22 nd of April) suggests that emissions from cement plants are the corresponding source of pollution. Figure 7 shows that most samples in both size fractions plot within the range of 87 Sr/ 86 Sr from the analysed end-members (Table 1), and are consistent with a mixing between 1) a major endmember, emissions from coal combustion, and 2) "secondary" contributors, such as cement plants and smelters. Four PM 2.5 samples (4 th of March, 21 st of June, 1 st and 19 th of September), as well as two TSP samples from the Changping and downtown Beijing sites (both taken on the 28 th of January) plot on a binary mixing line (r²=0.97; Figure 7) that yields a 87 Sr/ 86 Sr ratio of 0.7086 and a Sr concentration of at least 103 ppm for its corresponding source. This Sr isotope ratio is significantly lower than the range measured on potential end-members (Table 1)

CONCLUSIONS
China, and particularly its megacities such as Shanghai or Beijing, is working towards improving air quality, especially regarding airborne particulate matter. Measures have already been taken, including the phasing-out of Pb in gasoline (the major vector of atmospheric lead until the end of the 90's).
However, critically high concentrations of this toxic element are observed. We studied here the possibility of using Pb isotope systematics coupled with Sr isotopes systematics, to help decipher the origin of these elements (and ultimately the origin(s) of both PM 2.5 and TSP) in the atmosphere of Beijing and its vicinity. Results lead to the following conclusions:          Widory et al., 2007b). Note that the X-axis is under a logarithmic scale.