Environ. Sci. Technol. 2003, 37, 542-547
Direct Evidence for Destruction of Polychlorobiphenyls by OH Radicals in the Subtropical Troposphere MANOLIS MANDALAKIS,† HARALD BERRESHEIM,‡ AND E U R I P I D E S G . S T E P H A N O U * ,† Environmental Chemical Processes Laboratory (ECPL), Department of Chemistry, University of Crete, GR-71409 Heraklion, Greece, and German Weather Service (DWD), Meteorological Observatory, Albin-Schwaiger-Weg 10, D-82383 Hohenpeissenberg, Germany
Although laboratory studies have indicated that OH radicals readily oxidize polychorobiphenyls (PCBs) in the gasphase, this mechanism has never been positively demonstrated under real atmospheric conditions. By applying elaborated sampling and analytical techniques we achieved for the first time simultaneously a field study of the diurnal atmospheric cycle of OH radical and PCBs in a remote site of eastern Mediterranean. In all cases, the concentration of ∑PCB (sum of 27 congeners) showed a characteristic depletion during daytime, while the concentration of OH radicals was at the maximum levels. By assuming that the depletion of PCBs was caused solely by the destruction from OH radicals, PCB-OH rate constants (KOH) of different PCB congeners were determined from field data by applying the relative rate method. Our field-determined KOH values were notably coherent with those previously measured in laboratory experiments. In all measurement periods, the KOH values consistently decreased in sequence for those compounds showing an increasing degree of chlorination on the biphenyl group. By taking into account KOH values and latitudedependent concentration of OH radicals, it was estimated that, near to tropical and subtropical regions, the atmospheric lifetimes of PCBs 8 and 110 should be substantially low (10 and 20 days, respectively). A significant fraction of PCBs should be destroyed during their residence over tropical/subtropical regions, due to the intensive destruction by OH.
Introduction Although the production and use of polychlorobiphenyls (PCBs) were banned by the mid-1970s, these chemicals are ubiquitous pollutants in nearly all environmental compartments (1, 2). Previous studies suggested that warm temperatures favor the volatilization of PCBs, from contaminated earth surfaces, which are subsequently transported and condensed to areas permanently or seasonally cold (2). Because of their high persistence and toxicity PCBs can pose toxic effects, on animals and humans, decades after their release into the environment. * Corresponding author phone: +30 2810 393628; fax: +30 2810 393678; e-mail:
[email protected]. † University of Crete. ‡ German Weather Service (DWD), Meteorological Observatory. 542
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 3, 2003
Laboratory studies have indicated that the reaction with OH radicals might be the dominant loss process of PCBs in the atmosphere (3). However, this mechanism has never been positively demonstrated under real atmospheric conditions. In fact, since the atmospheric concentration of OH radicals maximizes during daytime, a significant decrease of gasphase PCBs should be expected at the same time period. Nevertheless, such a decrease should be mostly observable in remote areas where the sources of PCBs are negligible and the daytime volatilization of PCBs to the atmosphere is minimal. The diurnal cycle of atmospheric PCBs has been investigated near to heavily with PCB contaminated areas of United Kingdom and United States (4-6), and it has been shown that the PCB air concentration increased during the day but decreased during the night. According to this observation, it was suggested that the temperature-controlled volatilization of PCBs from contaminated earth surfaces controls the air concentrations of these pollutants (4-6), while the destruction of PCBs by OH radicals should be negligible. In a recent study (7) a daytime depletion of PCBs was observed, and this was attributed to destruction by OH radicals. However, due to the low sampling frequency (two samples/day) and analytical problems (determination of PCBs in very low concentration) encountered in the abovementioned study as well as the significant assumptions about the concentration of OH radicals, this process could not be positively demonstrated. The in-situ determination of OH radicals is a difficult task, and for this reason very few measurements exist worldwide. In addition, the low concentration of PCBs in the atmosphere requires long sampling times, which do not allow attaining the suitable measurement resolution to study a detailed diurnal variation (4-6). By performing a field study, in which elaborated sampling and analytical techniques for both OH radicals (8) and PCBs (9) were applied, we achieved a simultaneous study of the diurnal atmospheric cycle of OH radical and PCBs in the atmosphere of eastern Mediterranean. It must be noted that, in this midlatitude area, volatilization of PCBs has been shown to be negligible (10), while enhanced daytime concentrations of OH radicals (11) should be expected.
Experimental Section Measurements of atmospheric hydroxyl radical concentrations and sampling of gaseous PCBs were simultaneously performed at the marine background station of Finokalia (35°19′N, 25°40′E), a coastal site 70 km eastward of Heraklion (Island of Crete, Greece) at the top of a hilly elevation (130 m above sea level). Sixty (N ) 60) air samples were collected from August 18-21, 2001, using a high-volume air sampler located at the roof of the sampling station. The air was drawn through a polyurethane foam (PUF) plug (length 8.0 cm, diameter 7.5 cm) to collect particle and vapor-phase PCBs. The sampling frequency was 1 sample/h during daytime [7:00-20:00; Eastern European Summer Time (EEST)], while one sample was collected every 2 h during the nighttime [20:00-7:00 next morning; EEST]. Each sample corresponded to approximately 15 m3 of air. Extraction and analysis of the PUF samples have been described in detail elsewhere (9). A chromatographic peak was quantified as a PCB, if the following criteria were met: (a) the S/N ratio was higher than 3, (b) the isotope ratios for the two monitored product ions were within (15% of those obtained from reference standards, and (c) the retention time was within (2 s of those observed for reference standards. Two clean PUF plugs were used as blanks. Blanks were prepared, treated, and 10.1021/es020163i CCC: $25.00
2003 American Chemical Society Published on Web 12/20/2002
analyzed as the real samples. Blank values of PCBs were very low at both PUF plugs, and for most of the PCB congeners were not detectable. The maximum blank values were observed for PCBs 31 and 28 (34 and 28 pg, respectively). Method detection limits were derived from the blanks and quantified as the mean concentration in the blanks plus two times the standard deviation. The method detection limit of the PCB congeners not detected in field blanks was based on the instrumental detection limits. By considering the instrumental detection limit for PCBs (e.g. 600 fg µL-1), the recovery of the method (approximately 70%) (9), and the average volume of an air sample (15 m3), we calculated that the method detection limit of individual PCB congener will be approximately 1 pg m-3. The method detection limit for all PCB congeners ranged between 1 and 2 pg m-3. For all the samples collected during this campaign, the atmospheric concentration of individual PCBs was always higher than these values. Before the onset of the sampling program, PCB recovery studies were undertaken to verify the method. This involved spiking two precleaned PUFs with 5 ng of each PCB congener and treating them as real samples. Most of the congeners gave high recoveries ranging between 60 and 95%, while the average recovery of all congeners was 72%. As a quality assurance procedure, we also participated in an intercalibration study where six different laboratories analyzed the same sample. The concentrations we measured for PCBs 28, 52, 70, 90+101, 105, 110, 118, 149, 153, 160+158, 180, 194, and 199 were very close to the average values obtained by all laboratories. The average relative standard deviation between our results and the interlaboratory average values was 17%. In addition, duplicate samples were collected from the city of Heraklion in order to evaluate the reproducibility of our sampling method. The reproducibility of individual PCBs ranged between 40% and 2% (relative standard deviation) with an average value of 5%. Atmospheric OH concentrations were measured at the sampling station every 30 s and averaged over 5 min intervals using chemical ionization mass spectrometry (CIMS) based on methods previously developed by Eisele and co-workers (12, 13). The system used in the present study and its performance have been described in detail elsewhere (8). The OH detection limit was 2.5 × 105 molecules cm-3, with an estimated 2-sigma precision and accuracy of 32% and 38%, respectively. The PCB-OH rate constants of different PCB congeners were determined from field data by applying the relative rate method (3, 14-16). By assuming that the depletion of individual PCB congeners (PCB x) and reference compound (PCB 28) during daytime was caused solely by the destruction from OH radicals then
-d ln[PCB x]/dt ) k1[OH]
(1)
-d ln[PCB 28]/dt ) k2[OH]
(2)
where k1 and k2 are the OH radical rate constants for a individual PCB congener (PCB x) and for the reference compound (PCB 28), respectively. Hence,
(
ln
)
[PCB x]t0 [PCB x]t
)
(
)
k1 [PCB 28]t0 ln k2 [PCB 28]t
(3)
where [PCB x]t0, [PCB 28]t0 are the concentrations of x PCB congener under investigation and PCB 28, respectively, at time t0 (start time of daily depletion), and [PCB x]t, [PCB 28]t are the corresponding concentrations at time t (until the end of daytime depletion). The plots of ln([PCB x]t0/[PCB x]t) against ln([PCB 28]t0/[PCB 28]t) yielded straight lines with a zero intercept and a slope equal to k1/k2. Since the reaction rate constant of the reference compound (PCB 28) is known
TABLE 1. Summary of Meteorological Data for the Three Intensive Sampling Periodsa August 18, 2001
August 20, 2001
August 21, 2001
parameter
av value
SD
av value
SD
av value
SD
T (°C) RH (%) WS (m s-1) WD (deg) GLOB (W m-2) O3 (ppbv) JO(1D)
24.1 71 9.2 272 294 55 4.9
0.8 8 1.5 13 355 3 7.0
25.0 66 5.5 259 282 57 4.9
1.3 11 1.1 15 345 7 6.9
26.1 46 3.8 254 297 62 5.3
2.0 12 1.7 47 352 6 7.5
a T: air temperature (°C); RH: relative humidity (%); WS: wind speed (m s-1); WD: wind direction (deg); GLOB: global radiation (W m-2); O3: ozone (ppbv); JO(1D): photodissociation rate of ozone (×10-6 s-1); SD: standard deviation.
(k2 ) 1.1 10-12 cm3 s-1 molecule-1 (3)), then the reaction rate constant k1 of PCB x can be easily calculated from the slope of eq 3. Meteorological data, including air temperature, wind speed, wind direction, solar radiation, ozone levels, and photodissociation rate of O3, were measured at the Finokalia station during sampling (Table 1.). To investigate the influence of the origin of air masses arriving at Finokalia on the measured PCB concentrations, the 850-hPa isobaric air mass back-trajectories were calculated. All the backtrajectories extended back over 72 h (or 3 days) and were obtained from HYSPLIT4 model of NOAA (17).
Results and Discussion Figure 1a,b,c presents the diurnal variation corresponding to 3-points moving average of the total PCB concentration (∑PCB, sum of 27 congeners). The simultaneous variation of the OH radical concentration for each one of the three intensive sampling periods is also shown in the same figures. In all cases, the variation of ∑PCB concentration showed a diurnal pattern inversely related to that of OH. In particular, the highest OH radical concentrations were observed approximately between 10:00 and 15:00 EEST, while the lowest ∑PCB concentrations were measured during the same daytime period (Figure 1a,b,c). Figure 1d,e,f shows the corresponding diurnal variation of the ambient temperature. As for OH radical concentration, an opposite trend between ambient temperature and ∑PCB concentration was observed. This observation is in contrast to observations reported by previous studies performed at northern Minnesota (4), Indiana (5), and northern England (6). In these studies, PCB air concentrations exhibited a strong positive correlation with diurnal temperature profiles, with daytime PCB concentrations exceeding the corresponding nighttime values by a factor of 2 -4. In the above-mentioned studies, it was suggested that the observed diurnal cycle of vapor-phase PCBs arises from the rapid, temperature-controlled exchange of these compounds between terrestrial surfaces and the corresponding boundary-layer atmosphere (4-6). The diurnal variation of PCBs observed in our study suggests that the volatilization/exchange of PCBs from contaminated surfaces was rather of minor importance. This conclusion was drawn from a previous study at the same area, when the seasonal variation of PCB atmospheric concentration was investigated (10). In addition, it has been established (19) that substantially lower amounts of PCBs were used in the eastern Mediterranean region compared to areas in the United Kingdom or United States, where previous studies took place. We examined in detail the effect of meteorological parameters (Table 1). Air mass back-trajectory analysis during VOL. 37, NO. 3, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
543
544 9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 3, 2003
FIGURE 1. Diurnal variation of the atmospheric concentration of ΣPCB (solid line) and hydroxyl radicals (circles) (a,b,c) and of the ambient temperature (d,e,f).
FIGURE 3. Variation of ln([PCB x]t0/[PCB x]t) against ln([PCB 28]t0/[PCB 28]t) for congener 8 (dichlorinated biphenyl) and 44 (tetrachlorinated biphenyl) observed in August 21, 2001. FIGURE 2. Back-trajectories of air masses arriving at the sampling station throughout the sampling period are given plots of 3-days air mass back-trajectories for every 12 h between August 18, 2001 (6:00 EEEST) and August 22, 2001 (6:00 EEST). The number and the date printed along the trajectories indicate the sequence of time. the measurement period (August 18-21, 2001) showed that air masses reaching the sampling station of Finokalia were of similar origin, predominantly from north easterly directions (Figure 2). Consequently, observed diurnal changes in the concentration of PCBs could not be explained by significant changes in air mass origin. Previous studies suggested (4) that increasing wind speed, in remote areas, may cause a dilution of PCB concentrations due to higher atmospheric turbulence and vertical mixing. Wind speed was measured continuously in our study, during sampling, and no significant inverse correlation with ∑PCB was found. Thus, the observed decline of ∑PCB during daytime could also not be attributed to dilution by higher wind speeds, especially on August, 21 (Figure 1c), where wind speed was the lowest measured (
