ARTICLE pubs.acs.org/est
Brominated Flame Retardants in the Atmosphere of E-Waste and Rural Sites in Southern China: Seasonal Variation, Temperature Dependence, and Gas-Particle Partitioning Mi Tian,†,‡ She-Jun Chen,*,† Jing Wang,†,‡ Xiao-Bo Zheng,†,‡ Xiao-Jun Luo,† and Bi-Xian Mai† †
State Key Laboratory of Organic Geochemistry, Guangzhou Institute of Geochemistry, Chinese Academy of Sciences, Guangzhou 510640, China and ‡ Graduate University of Chinese Academy of Sciences, Beijing 100049, China
bS Supporting Information ABSTRACT: The recycling of electrical and electronic waste (e-waste) in developing countries has attracted much attention as a significant source of brominated flame retardants (BFRs). Gaseous and particle-bound BFRs were measured in the atmosphere at e-waste and rural sites in southern China during 20072008. The annual average concentrations in the air were 3260 ( 3370 and 219 ( 192 pg/m3 for polybrominated diphenyl ethers (PBDEs) and were 546 ( 547 and 165 ( 144 pg/m3 for non-PBDE BFRs at the e-waste and rural sites, respectively. PBDEs had unusually high relative concentrations of di- and tribrominated congeners at the e-waste site. The ClausiusClapeyron (CC) plots showed that the gaseous concentrations of less brominated BFRs (di- through hexaBFRs) were strongly controlled by temperature-driven evaporation from contaminated surfaces (e.g., e-waste, soils, and recycled e-waste remains) except for winter. However, weak temperature dependence at the rural site suggests that regional or long-range atmospheric transport was largely responsible for the air concentrations. Gas-particle partitioning (KP) of PBDEs correlated well with the subcooled liquid vapor pressure (PLo) for most sampling events. The varied slopes of log KP versus log PLo plots for the e-waste site (0.59 to 1.29) indicated an influence of ambient temperature and atmospheric particle properties on the partitioning behavior of BFRs. The flat slopes (0.23 to 0.80) for the rural site implied an absorption-dominant partitioning. This paper suggests that e-waste recycling in Asian low-latitude regions is a significant source of less brominated BFRs and has important implications for their global transport from warm to colder climates.
’ INTRODUCTION Over the past two decades, there have been growing concerns about brominated flame retardants (BFRs). Many BFRs exhibit persistence and bioaccumulation properties similar to many persistent organic pollutants (POPs), and they are detected globally.1 Polybrominated diphenyl ethers (PBDEs) are one of the most widely used BFRs, for which three PBDE technical mixtures (penta-, octa-, and deca-BDE) have been or are manufactured. The main components (tetra- through heptaBDEs) of the former two mixtures have been included in the newly listed POPs of the 2009 Stockholm Convention because of their persistence, bioaccumulative capacity, toxicity, and adverse effects.2 More recently, there has been increasing evidence of the environmental occurrence of novel BFRs such as decabromodiphenyl ethane (DBDPE) and 1,2-bis(2,4,6-tribromophenoxy)ethane (BTBPE), in sediment, air, indoor dust, and aquatic and terrestrial species from different locations in the world.3,4 The recycling of electrical and electronic waste (e-waste) in developing countries has attracted significant attention as an r 2011 American Chemical Society
important source of many toxic chemicals. It has been reported that around 80% of the world’s e-waste is exported to mid- and low-latitude developing countries/regions of Asia, and 90% of this flows into China.5 The techniques for recycling e-waste in these regions are usually primitive and hazardous and lead to the emissions of high concentrations of heavy metals, polychlorinated biphenyls (PCBs), polychlorinated dibenzo-p-dioxin/furans (PCDD/Fs), and BFRs in the environment.69 The elevated body burdens of PBDEs, PCBs, and PCDD/Fs of residents and workers from an e-waste site in China suggest a substantial human exposure to these chemicals.10 The atmosphere plays a significant role in transporting BFRs from their sources to rural/remote locations. Studies on atmospheric BFRs have been conducted largely in mid- to high-latitude Received: July 3, 2011 Accepted: September 8, 2011 Revised: September 2, 2011 Published: September 08, 2011 8819
dx.doi.org/10.1021/es202284p | Environ. Sci. Technol. 2011, 45, 8819–8825
Environmental Science & Technology
ARTICLE
regions of North America and Europe,1,4 while data from lowlatitude regions are very limited.11,12 Investigations from these regions are therefore needed, given that most worldwide e-waste is recycled in low-latitude regions where the tropical climate favors the release of BFRs into the atmosphere and their subsequent movement to colder regions of the earth.12,13 It has been known that the air concentrations of semivolatile organic compounds (SOCs) have a strong relationship to ambient temperature.14 The temperature dependence of SOCs in the air is conveniently described by the ClausiusClapeyron (CC) equation, which can be expressed as ln P ¼ m=T þ b
Table 1. Summary of Concentrations (pg/m3) of BFRs in the Atmosphere (Gas + Particle) at the E-Waste and Rural Sites in Southern Chinaa e-waste site BFRs
ð1Þ
where P is the partial pressure (atm), T is the temperature (K), and m and b are fitting parameters. The CC plot has been used to indicate the relative importance of evaporation from local surfaces versus long-range atmospheric transport (LRAT) or advection of background air mass in controlling air concentrations.14,15 Although a few studies have reported on the atmospheric concentrations of BFRs (in particular, PBDEs) from e-waste locations, emission mechanisms into the atmosphere have been poorly investigated.8,16,17 In the present study, we measured the concentrations of PBDEs and several non-PBDE BFRs in the atmosphere at an e-waste recycling site as well as a rural site in southern China. The emission profiles and seasonal variations of BFRs at the two sites were examined. This paper presents the temperature-dependence and gas-particle partitioning of BFRs to provide insights the emission mechanisms into the atmosphere and partitioning behaviors between the gas and particle phases.
’ EXPERIMENTAL SECTION Sample Collection. Details regarding the studied area and sampling technology are provided in the Supporting Information (SI). Briefly, the sampling was conducted at an e-waste site and a rural site in southern China. At the e-waste site, large quantities of e-waste (∼700 000 t) that is mostly imported from overseas is processed within an area of ∼330 km2 using primitive methods (e.g., mechanical shredding, acid processing, and open burning). The rural site (25 km northeast of the e-waste site) is located in an agricultural area, where BFRs in the air are mainly from the e-waste and urban areas (SI Figure S1).9 The average daily temperatures, wind speeds, and relative humidity during the sampling period varied from 7 to 33 °C, from 1.2 to 7.6 m/s, and from 44% to 90%. Air samples were taken by drawing air through glass fiber filters (GFFs) followed by polyurethane foam (PUF) plugs for 24 h, using a high-volume air sampler. A total of 60 pairs of samples (gas and particle) from each site were collected simultaneously on five consecutive days each month from July 2007 to June 2008. Sample Preparation and Instrumental Analysis. The PUF and GFF samples were Soxhlet extracted separately with a mixture of hexane and acetone (1:1) for 48 h. Prior to the extraction, BDE77, BDE181, and 13C-BDE209 (GFFs only) were added as surrogates to monitor the recoveries. The extracts were concentrated to 12 mL and then eluted through a silica/ alumina column with 1:1 hexane-dichloromethane. Following solvent exchange to hexane, BDE118 and BDE128 were added as quantitation standards. The target BFRs were analyzed using a gas chromatograph coupled to a mass spectrometer in electron capture negative ionization mode (GC-ECNI-MS). Di- through hepta-BDEs,
AM
GM
range
AM
GM
di-BDEs
2.324120
399
131
0.4164.4
8.92
4.67
tri-BDEs
3.983000
354
131
0.1529.3
6.05
4.07
tetra-BDEs
8.69762
220
128
0.7115.7
5.82
4.45
penta-BDEs 7.761170
284
160
0.9331.6
7.00
4.99
hexa-BDEs 3.25571 hepta-BDEs 3.12619
131 127
1.7019.1 nd17.8
4.29 5.50
3.60 3.94
octa-BDEs
nd491
154
nona-BDEs
nd767
187
67.8 67.5 94.2 139
nd64.9
10.3
1.6299.1
25.2
deca-BDEs
72.29700 1410
862
18.6804
142
PBDEs
10917900 3260
2080
37.0952
216
PBT
0.19125
PTBX
nd158
PBEB HBB
0.29867 4.47559
BTBPE
4.49398
DBDPE PBBs total a
range
rural site
nd2240 nd466
21.8 8.29 41.0 138 78.7 209 49.7
12019000 3810
5.96 18.3 91.8 165
9.64 0.213.57
0.93
0.76
0.64
nd3.43
0.33
0.09
0.104.80 0.4213.9
0.75 4.49
0.53 3.14
nd28.4
2.97
12.3 88.8 45.3 137 14.4 2490
3.971370 158 0.26.74
0.92
54.71710 384
1.65 80.6 0.50 287
AM: arithmetic mean; GM: geometric mean; nd: not detectable.
pentabromoethylbenzene (PBEB), 2,3,5,6-tetrabromo-p-xylene (pTBX), pentabromotoluene (PBT), hexabromobenzene (HBB), and 2,20 ,4,40 ,5,50 -polybrominated biphenyl (BB153) were separated with a DB-XLB (30 m 0.25 mm i.d., 0.25 μm film thickness) capillary column. For octa- through deca-BDEs, DBDPE, BTBPT, and BB209, a DB-5HT (15 m 0.25 mm i.d., 0.10 μm film thickness) column was used. More information on the procedures is given in the SI. Quality Control. The breakthrough of the gas-phase BFRs was tested twice at each sampling site using a second PUF plug (2.5 cm thick) in series with the first one. Field blanks (clean PUF plugs and GFFs) were prepared identically to that of the real samples, except with no air drawn through. Twenty-four procedural blanks were run sequentially with the samples. Breakthrough less than 10% occurred for several di- through hexaBDEs at the e-waste site. BDE28, 47, 99, 206, 207, 208, and 209 were found in the procedural blanks, but their amounts were all less than 5% of those in the corresponding sample extracts. The concentrations in the samples were blank corrected accordingly. The amounts of PBDEs found in the field blanks were not significantly higher than the procedural blanks. The recoveries of the surrogate standard (mean ( standard deviation) were 94.3 ( 15.2% for BDE77, 87.2 ( 11.3% for BDE181, and 107 ( 18.0% for 13C-BDE209. The recoveries of the target compounds in the six matrix-spiked samples were 67.2%122% (standard deviations pTBX > BTBPE > PBBs > PBEB > PBT for both sites. This BFR profile was similar to the atmospheric deposition profile observed at the same sites in our previous study.9 SI Figure S3 depicts the congener profiles of the atmospheric PBDEs at the two sites. BDE209 was the most dominant congener in almost all of the samples. It contributed 47 ( 17% and 59 ( 16% of the total PBDEs at the e-waste and rural sites, respectively. This is consistent with the fact that the deca-BDE mixture is one of the most frequently used flame retardants across the world, especially in electronic/electric products.3,18 On the other hand, it is surprising to observe high contributions of diand tri-BDE congeners in the air (especially of the e-waste site) because these congeners are not present, or present only in minor amounts, in the technical PBDE mixtures.19 Highest geometric mean concentrations of di- and tri-BDEs of 4120 and 3000 pg/m3, respectively, were found at the e-waste site and were comparable to the concentrations of tetra-, penta-, and nona-BDEs (Table 1). The atmospheric PBDE congener profiles in this study were quite different from the profiles that have typically been observed in the air from other locations, in which di- and tri-BDEs accounted for very small proportions of the PBDEs.2023 In fact, tetra-, penta-, and hexa-BDEs, which are mainly derived from technical penta-BDE mixtures, were also present in significant proportions in the studied area compared to those in the urban air in southern China from an early study.11 This finding provides evidence that e-waste recycling is a significant emission source of less brominated PBDEs (di- through hexa-BDEs) to the environment. These PBDEs (in particular those absent from the technical PBDE products) could originate from photolytical degradation of highly brominated BDEs in the environment. However, they are more likely to originate from pyrolytic degradation during the recycling processes (e-waste burning) due to the shielding effects of carbonaceous aerosols to photodegradation indicated in a previous study.24 Additionally, high ambient temperatures facilitate the volatilization of less brominated congeners (with relatively high vapor pressures) from contaminated environmental compartments, massive e-waste, and recycled e-waste remains stacked in the fields. The total PBDE concentrations (∑36PBDEs) at the e-waste site in this study were lower than those detected in the air from another e-waste site (Guiyu) in China, with average concentrations of 21 500 pg/m3 in 2004 (∑22PBDEs without BDE209) and 8760 pg/m3 in 2005 (∑11PBDEs).16,17 These differences could
Figure 1. Seasonal variations (from July 2007 to June 2008) of the major BFRs in the gaseous phase at the two sites in southern China. E-waste site: A (sum of di-, tri-, and tetra-BDEs, PBT, PEEB, and HBB) and B (sum of penta- and hexa-BDEs). Rural site: C (sum of di- and triBDEs, PBT, PEEB, and HBB) and D (sum of tetra-, penta-, and hexaBDEs).
be ascribed to the distance from the sampling sites to the e-waste recycling facilities and/or the recycling manners. There has been a substantial decrease in the activities of open burning e-waste because of the increasingly stringent restrictions. For instance, the highest concentrations in Guiyu were measured right at a building where there are open burning operations.16 It has been shown in previous studies that the average PBDE concentrations in urban air were on an order of 50150 pg/m3, and those in rural/remote air were generally in the range of 515 pg/m3 in North America, Europe, and Asia.1,20,23,25,26 A few studies have reported higher average concentrations in China (2450 pg/m3 in urban air and 220 pg/m3 in rural air)11,27 and in Ontario, Canada (300 pg/m3 in rural air)28 as well as lower concentrations in the Arctic (18 pg/m3).22,29 These levels are generally lower than the average concentrations of ∑16PBDEs (that have been typically reported in the literature) at the e-waste site (2220 pg/m3) and rural site (192 pg/m3) in the present study (SI Table S2). Atmospheric measurements of non-PBDE BFRs in other studies are very limited. The average concentrations of BTBPE in rural air were 3.4 pg/m3 in the east-central U.S. and 0.51.2 pg/m3 near the Great Lakes (U.S.), which were comparable to that in this study.21,25 The average concentrations of DBDPE in the air near the Great Lakes were 122 pg/m3, much lower than the concentrations in the present study.25 PBEB, with a relatively high concentration of 550 pg/m3, was measured near the Great Lakes.25 Seasonal Variations. The seasonal variations of the total air BFR concentrations during the sampling year (July 2007June 2008) at the e-waste and rural sites are shown in SI Figure S4. The monthly air BFR concentrations (five day averaged concentration) varied from 912 to 10 300 pg/m3 at the e-waste site. Seasonal variations of the total BFR concentrations in the gaseous and particle phases were generally similar (r = 0.85, p = 0.001). The monthly air BFR concentrations at the rural site ranged from 144 to 1160 pg/m3. Different seasonal variations were observed for the gaseous and particle-bound BFRs. 8821
dx.doi.org/10.1021/es202284p |Environ. Sci. Technol. 2011, 45, 8819–8825
Environmental Science & Technology
Figure 2. Seasonal variations (from July 2007 to June 2008) of the major BFRs in the particle phase at the two sites in southern China. E-waste site: A (sum of di-, tri-, and tetra-BDEs, PBT, PEEB, and HBB), B (sum of penta-, hexa-, and hepta-BDEs), and C (sum of octa-, nona-, and deca-, BTBPE, DBDPE). Rural site: D (sum of tri-, tetra-, and pentaBDEs), E (sum of hexa-, hepta-, and octa-BDEs), and F (nona- and decaBDEs and DBDPE).
Seasonal variations of the major gaseous and particle-bound BFRs (BFR concentrations with close variations were combined) were compared in Figures 1 and 2. At both the sites, the concentrations of more volatile BFRs (di-, tri, and tetra-BDEs, PBT, PBEB, and HBB) in the gaseous phase were relatively high in the summer months. Gaseous penta- and hexa-BDEs showed a similar seasonal variation to the more volatile BFRs at the e-waste site; whereas variation of gaseous tetra-, penta-, and hexa-BDEs differed from the more volatile BFRs at the rural site (Figure 1). The particle-bound BFR concentrations showed different seasonal variations at both the sites (Figure 2). It is interesting to find that, at the e-waste site, penta-, hexa-, and hepta-BDEs exhibited variations similar to the less volatile BFRs (octa-, nona-, and decaBDEs, DBDPE, and BTBPE) in warm season (April to September) and to the more volatile BFRs in colder season (October to March). This may suggest an influence of gas-particle partitioning behavior of these BFRs (e.g., lower temperature favors deposition of gaseous BFRs to particles). At the rural site, particle-bound hexa-, hepta-, and octa-BDEs and BTBPE showed a seasonal variation similar to more volatile tri-, tetra-, and penta-BDEs rather than less volatile nonaand deca-BDEs and DBDPE. These less volatile compounds are the dominant ingredients of technical deca-BDE and DBDPE products that are used in large quantities the nearby urban region; while trithrough octa-BDEs and BTBPE are more important pollutants at the e-waste site than in the urban region.9 The results suggest differences in the source and/or behavior of the BFRs in the atmosphere at the two sites.
ARTICLE
The seasonal variations are believed to be affected by the meteorological conditions. However, significant relationships were not found between the air concentrations (individual or total BFRs) and wind speed or relative humidity. Temperature is one of the major factors controlling seasonality of SOCs in air. The influence of ambient temperature on BFRs at the sites is discussed below. Influences of Ambient Temperature. The CC equation was applied to the major gaseous BFRs at the two sites, and the regression results are summarized in SI Table S3. The CC plots show apparent differences in temperature dependence of the BFR concentrations at high and low temperatures at the e-waste site (Figure 3). At high temperatures (1930 °C, average daily temperature), the gaseous concentrations (except for hepta-BDEs) were positively correlated with temperatures (p < 0.001), which explained approximately 44%78% (r2) of the variations. The mean slope of the CC plots in the present study was 20 840 ( 3250, which are steeper than those for PBDEs reported from Europe (9110 ( 3940) and North America (6400 ( 600 for BDE47 and 5300 ( 960 for BDE99).25 To date, the temperature dependence of air concentrations associated with e-waste has not been examined. This finding suggests that gaseous concentrations of BFRs in high-temperature seasons at the e-waste site are strongly controlled by temperature-driven evaporation from contaminated surfaces (e-waste, soils, and recycled e-waste remains as mentioned above) in the local surroundings of this site.14,15 This result also has important implications for the global transport from warm climates to colder climates of these chemicals resulting from e-waste recycling (grass-hopper effect). At low temperatures (819 °C) during the winter, many BFRs showed increasing concentrations with declining temperatures, although the linear relationships were not all statistically significant (SI Table S3). The differences in temperature relationships for SOCs in winter and other seasons have been reported in early studies, although the winter temperatures were quite different.14,15,30 These studies attributed the lack of a temperature dependence of SOCs at lower temperatures to LRAT being the dominant sources of the air concentrations. Despite the correlations with temperature for many BFRs in winter in the present study, we think temperature may explain a small fraction at best of the variability of these BFRs in the air. Their occurrence was also unlikely to be a result of LRAT because of the obvious presence of local sources. Instead, we speculated that the elevated concentrations were due to the increasing e-waste recycling activities in winter of the sampling year, which resulted in an increased emission of pollutants. Similar CC plots for PBDEs have also been observed in Chilton, UK, where the high concentrations in winter resulted from seasonal combustion sources.20 At the rural site, BFR concentrations had significant temperature dependence for di-, tri, and tetra-BDEs and PBT but were weak or lacking in temperature dependence for other BFRs (SI Table S3 and Figure S5). The slopes of the CC plots (7185 ( 2500) were significantly flatter than those observed at the e-waste site. A clear decline in slope with increasing distance from the suspected sources have been observed for PCBs.14 A flat slope or low temperature dependence indicates that regional or longrange atmospheric transport controls atmospheric levels at a sampling site.14,15 The e-waste and urban areas are the main sources of BFRs in the rural air as indicated in our recent finding.9 Gas-Particle Partitioning. Gas-particle partitioning is an important process governing the atmospheric fate of SOCs. 8822
dx.doi.org/10.1021/es202284p |Environ. Sci. Technol. 2011, 45, 8819–8825
Environmental Science & Technology
ARTICLE
Figure 3. ClausiusClapeyron (CC) plots of BFRs at the e-waste site. Data of all samples from September 2007 and a sample from February 2008 were excluded because of the exceptionally low and high concentrations, respectively. The open circles represent data from winter.
The partitioning is usually defined by the partition coefficient (KP) KP ¼ ðF=TSPÞ=A
ð2Þ
where F and A are the particle-bound and gaseous concentrations (pg/m3), respectively, and TSP is the concentration of total suspended particles (μg/m3). Two mechanisms (adsorption onto the particle surface and absorption into the organic matter of particles) have been used to interpret the partitioning, both of which lead to a linear relationship between log KP and log PoL: log KP ¼ mlog PLo þ b
ð3Þ
where PoL is subcooled liquid vapor pressure.31,32 Although it has been suggested that the slope m should be near 1 for true equilibrium partitioning for both adsorption and absorption processes, deviations of m for field data do not necessarily indicate nonequilibrium conditions. Instead, the slopes for a given set of compounds rely largely on sorption properties of particles, ambient temperature, and relative humidity.33 The relationships between log KP and temperature-corrected log PoL were investigated separately for PBDEs from different sampling events at the two sites (Table 2 and SI Figures S67). The slopes of Junge-Pankow plots ranged widely from 0.59 to 1.29 at the e-waste site. The influence of temperature and
relative humidity on the varied slopes was evaluated by investigating the correlations between them. The slopes correlated positively with temperature (p = 0.04) (slopes tended to be flat with increasing temperature) but not with relative humidity (SI Figure S8). While possible blow-off sampling artifact could cause this relationship with temperature,34 a likely interpretation for this is that more BFRs with a high vapor pressure would volatilize into the air as ambient temperature increased, and their subsequent partitioning onto/into particles would make the slopes flatter. This interpretation also implies a nonequilibrium of the partitioning. However, temperatures explained only ∼36% of the total variation of the slopes. Another factor was probably attributed to different sorptive properties of the atmospheric particles as a result of their various origins at this site. The particles could originate from e-waste shredding and burning, resuspension of soil and dust, and secondary aerosols formed in the atmosphere. The observed slopes at the e-waste site were not able to indicate adsorption or absorption mechanisms.33 Pankow calculated that intercept values between 7.3 and 8.9 would specify absorptive partitioning. 31 The intercepts b (4.42 to 7.72) were mostly outside this range suggesting a primary adsorption mechanism. The slopes (0.23 to 0.80) for data from the rural site clearly deviated from 1 (except for those with no linear relationships for three events), indicating that absorption into 8823
dx.doi.org/10.1021/es202284p |Environ. Sci. Technol. 2011, 45, 8819–8825
Environmental Science & Technology
ARTICLE
Table 2. Slopes (m), Intercepts (b), r2, p-Values, and Number of Data Points (n) of the Linear Regression of Log KP (m3/μg) versus Log PoL (Pa) of PBDEs in Different Sampling Eventsa sampling event
e-waste site m ( SE
b
b ( SE
rural site r
2
p-value
n
m ( SE
b ( SE
r2
p-value
n
July 2007
0.83 ( 0.05
5.80 ( 0.15
0.82
