Temperature Dependence of the Characteristic Travel Distance

Here we explore how temperature influences the characteristic travel distance by using a ...... Environmental Science & Technology 2013 47 (1), 502-50...
0 downloads 0 Views 125KB Size
Download PDF
Environ. Sci. Technol. 2003, 37, 766-771

Temperature Dependence of the Characteristic Travel Distance A N D R E A S B E Y E R , * ,† F R A N K W A N I A , ‡ TODD GOUIN,§ DONALD MACKAY,§ AND MICHAEL MATTHIES† Institute of Environmental Systems Research, University of Osnabru ¨ ck, 49069 Osnabru ¨ ck, Germany, Division of Physical Sciences, University of Toronto at Scarborough, 1265 Military Trail, Scarborough, Ontario, Canada M1C 1A4, and Canadian Environmental Modelling Centre, Trent University, Peterborough, Ontario, Canada K9J 7B8

The effect of temperature variation on the environmental fate of organic chemicals can be evaluated in steady-state multimedia box models by expressing chemical partitioning data and reaction rate coefficients as functions of temperature. Using such a model the temperature dependence of the characteristic travel distance in air LA, which is a measure for the atmospheric long-range transport potential of organic chemicals, is calculated. Simulations are reported for a set of 40 chemicals of environmental interest. Increasing temperature is shown to have two opposing effects on LA. Rates of chemical transformations in the atmosphere (kair) and surface media are increased, which reduces LA. Rates of atmospheric deposition (kdep) are reduced leading to increased mobility and LA. Accordingly, LA can monotonically increase or decrease with increasing temperature, or it can have a maximum in the modeled temperature range, but it cannot have a minimum. For chemicals with a strong temperature dependence of kair relative to kdep, LA will increase with increasing temperature. Results for selected polychlorinated biphenyls are compared to monitoring data yielding qualitative agreement when chemical properties are adjusted to mean temperatures for the measurement period. The results demonstrate that the temperature dependence of the characteristic travel distance is highly dependent on chemical characteristics and can be counterintuitive. The use of mass balance models is thus essential. The difference between the LA values at 5 °C and 30 °C can be up to a factor of 6. Accordingly, chemical ranking with respect to LA can change significantly if performed at different temperatures. Implications of the different temperature dependencies on long-range transport to polar regions are discussed.

Introduction The potential for environmental transport of an organic chemical far from its location of release into the environment is one of the criteria for classifying persistent organic pollutants (1, 2). Scheringer (3, 4) has convincingly argued that the potential for widespread chemical dispersal is cause for concern independently of chemical toxicity. A readily * Corresponding author phone: +49-3641-65 6208; fax: +49-364165 6210; e-mail: [email protected]. † University of Osnabru ¨ ck. ‡ University of Toronto at Scarborough. § Trent University. 766

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 4, 2003

calculated and useful measure of a chemical’s long-range transport potential is the characteristic travel distance, which is defined as the simulated distance at which the initial concentration of a pollutant is reduced to 1/e (approximately 37%) in a plug-flow system (5-7). A characteristic travel distance can readily be derived from the results of multimedia mass balance models (6). Although it is well established that the environmental fate of persistent organic pollutants is strongly affected by temperature (8-14), to date temperature effects have not been accounted for in the calculation of the characteristic travel distance. If the characteristic travel distance is to be used in the screening for persistent organic pollutants, the potential effect of temperature should be understood. Further, when comparing calculated and empirical travel distances an adjustment to actual environmental temperatures is necessary. Temperature changes directly affect certain physicalchemical properties and can also indirectly influence environmental fate processes. A temperature increase can have the following effects on the environmental fate of organic chemicals: (1) Diffusion rates increase. (2) Partitioning equilibria shift from condensed phases to the gas phase. For instance, the particle bound fraction in the air decreases, which may increase the fraction available for degradation in the gas phase and reduce the efficiency of particle deposition. (3) Abiotic reaction rates increase. (4) Biotic reaction rates probably increase. (5) Hydroxyl radical concentrations in air will generally be elevated resulting in shorter atmospheric half-lives. Other environmental variables indirectly affected by temperature include rain, wind, or the convective mixing of the atmosphere. Qualitative changes of processes also occur, especially if temperatures drop below 0 °C. Precipitation may fall as snow rather than rain, while water bodies can become ice-covered retarding volatilization. Many of these process variabilities have been discussed in previous publications (9-12, 15) but are usually not implemented in simple compartment or “generic models”. Some sophisticated multimedia fate models account for temporal and spatial temperature differences by defining seasonally fluctuating temperatures in model compartments and deriving temperature-dependent partitioning and degradation parameters (16-18). Scheringer and co-workers similarly account for temperature differences in a “chain-of-boxes” model (19). However, simpler “unit-world” type multimedia models are rarely adjusted to explicitly account for varying temperatures. Here we explore how temperature influences the characteristic travel distance by using a multimedia chemical fate model, where the effects of the five points above are quantified. It is assumed that the same temperature applies to all compartments. Seasonal and diurnal variation of all other parameters such as emissions or wind speed is neglected in the interest of simplicity.

Model Description Different mathematical definitions of the characteristic travel distance in air LA exist (6). Here, we use the definition that is deemed to be most suitable for understanding the influence of temperature on LA, namely

LA )

u kair + kdep‚F

(1)

where u is the (average) wind speed (in m/d), kair is the reaction rate constant in air (in 1/d), kdep is the deposition 10.1021/es025717w CCC: $25.00

 2003 American Chemical Society Published on Web 01/09/2003

rate constant (in 1/d), and F is the overall stickiness. The stickiness F is the fraction of chemical remaining in a surface compartment (e.g. soil or water) after deposition (6). The overall stickiness accounts for the revolatilization from all surface compartments (i.e. water and soil together) and is defined as the ratio of the net-flux from air to surface divided by the gross-flux to the surface, i.e., it is the fraction retained in surface media. When calculating LA it is assumed that 100% of the emission is to the air compartment. We use the ELPOS-multimedia model (20), which is a steady state (level III) box model. ELPOS is based on the regional EUSES-SimpleBox model (21), which is part of European legislation (22). Two modifications of the original SimpleBox-model used in EUSES were necessary for this study, namely: (i) a substance specific soil depth and (ii) a variable hydroxyl radical concentration in air. A substance-specific soil depth is important for appropriately describing the exchange of semivolatile chemicals between air and soil. Bennett et al. (5) previously used a variable soil depth in calculations of LA. In a box model it is assumed that instant and complete mixing within each compartment occurs. However, organic chemicals sorb to soil particles to a variable extent and are thus transported to deeper soil layers at variable rates. The chemicaldependent soil depth yields surface concentrations that account for the velocity of vertical transport. Since complete mixing is still assumed, this does not necessarily yield more realistic concentrations in deeper layers, but the surface concentration is most relevant to the exchange between air and soil. Mathematically we use the same approach as in SimpleBox 2.0 (23) to account for vertical water transport in soil. Additionally we consider vertical solid-phase transport in the soil (24) when calculating the effective soil depth. McLachlan et al. (24) argue that bioturbation and particle transport in macropores can yield significantly faster mixing particularly for hydrophobic, particle bound chemicals. This additional transport process can easily be included by adding another dispersion term (DS) when calculating the vertical transport in soil. We set DS to 5.5‚10-7 m2/d based on ref 24. SimpleBox distinguishes natural soil, agricultural soil, and industrial/urban soil, with fixed soil depths of 5, 20, and 5 cm, respectively. In our study the depth of the agricultural soil compartment remains fixed at 20 cm, whereas the depths of the other two soil compartments are calculated as suggested by Brandes et al. (23). A minimum depth of 1 cm is used. The diffusion path length in soil is set equal to the effective soil depth. Atmospheric hydroxyl radical concentrations COH (molecules/cm3) are indirectly related to temperature, because the production of OH-radicals depends on sunlight (25). Temperature and COH thus tend to co-vary. Since OH reactions are the most important reactive removal pathway in air for most organic chemicals, the correlation of COH with temperature should be included in model simulations. Using data for surface concentrations of OH-radicals from ref 26 we have derived a simple equation to express COH as a function of temperature, T (K), in the atmospheric compartment between 0 and 30 °C:

COH ) (0.5 + 0.4‚(T - 273.15))‚105

(2)

This yields a COH of 0.5‚105 cm-3 at 0 °C and 12.5‚105 cm-3 at 30 °C. These concentrations refer to 24-h averages in the lower troposphere of the Northern Hemisphere. The particle bound fraction in air is calculated using an approach based on the octanol-air partition coefficient KOA (27-29). Calculating the particle bound fraction based on the vapor pressure (30, 31) does not change the temperature dependence of LA significantly. The temperature dependence of the particle bound fraction is similar for both models,

because the energies of phase transfer between octanol and the gas phase and between the pure liquid and the gas phase are similar for most POPs (32). A detailed discussion of ELPOS’ limitations can be found in ref 20. ELPOS has not been designed to simulate actual environmental concentrations, but rather for distinguishing chemicals with high and low long-range transport potentials under fixed environmental conditions. Since ELPOS assumes only one box for each compartment, no differentiation of spatially varying environmental properties is possible. Hence, ELPOS is applicable only on a relatively small subcontinental scale, with a box length of a few hundred up to a few thousand kilometers (20). Substance Property Data. Calculations of LA were done for a set of 40 selected chemicals containing polychlorinated biphenyls (PCB), chlorinated benzenes, chlorinated pesticides, polyaromatic hydrocarbons (PAH), and one polychlorinated dibenzo-p-dioxin (1,2,3,4-TetraCDD). Necessary data for the model calculations are the air-water partition coefficient KAW, the octanol-water partition coefficient KOW, and the octanol-air partition coefficient KOA as well as degradation rate constants in air, water, soil, and sediment. Temperature-dependent partitioning data were taken from ref 32. This data set is almost entirely based on reported measurements and is thermodynamically consistent. The temperature dependence of the partitioning data is calculated according to (32)

Kx(T) ) Kx0‚exp(∆Ux/R(1/T0 - 1/T))

(3)

where Kx is the chemical property (KAW, KOW, KOA), T is the absolute temperature, Kx0 is the chemical property at reference temperature T0 (298.15 K), ∆Ux is the internal energy change of Kx (in J mol-1), and R is the universal gas constant (8.314 J mol-1 K-1). For most chemicals, degradation half-lives in the various surface compartments were obtained from ref 33 and represent “reactivity classes” rather than discrete quantities. Half-lives for the hexachlorocyclohexanes were taken from ref 6. A fixed activation energy Ea of the first-order reaction of 30 kJ mol-1, which corresponds to a doubling of the halflife between 10 and 25 °C, is assumed and degradation rate coefficients k in d-1 are calculated by

k(T) ) k0‚exp(Ea/R(1/T0 - 1/T))

(4)

where k0 is the degradation rate constant at the reference temperature T0, which is 298.15 K. Atmospheric rate constants only include OH-radical reactions since this reaction pathway is deemed most important for the chemicals under investigation (34, 35). The degradation rates in air along with their temperature dependence are reported in the Supporting Information (Table S1). Measured degradation rate constants in air were preferentially selected (34, 36-38) or when necessary were derived from molecular structure using the computer program AOPWIN (Syracuse Research Corporation, version 1.90). However, AOPWIN does not estimate activation energies, which are necessary to calculate degradation rate constants at different temperatures. Since the effect of temperature on the reaction of PAHs with OH is ambiguous (Table S1), no temperature dependence was assumed for the calculated PAH degradation rate constants in air. For the PCBs a significant increase of Ea with increasing degree of chlorination is apparent from Table S1

Ea ) 2.460‚N + 0.625

(5)

where N is the number of chlorines. Although the coefficient of determination is small (R2 ) 0.67), the slope is significantly different from zero (two-sided confidence range R ) 0.01). VOL. 37, NO. 4, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

767

TABLE 1. Simulated Characteristic Travel Distance in Air (LA) at 5 °C, 15 °C, and 30 °C Assuming Variable OH-Radical Concentration and Observed (obs) LA Derived from Data in Ref 40 for Selected Substances LA (km) at 15°C

at 30°C

PCB-15 7700 3200 PCB-28 14700 6500 PCB-29 10200 4600 PCB-30 11600 4400 PCB-31 10800 5200 PCB-33 13700 6200 PCB-44 15300 8100 PCB-47 9700 5900 PCB-52 19500 9200 PCB-61 17200 8200 PCB-77 4900 5900 PCB-95 11700 13600 PCB-101 9900 13100 PCB-105 1700 3700 PCB-110 8000 8300 PCB-116 5400 5900 PCB-118 2700 5800 PCB-126 1300 2500 PCB-138 3000 7300 PCB-153 3600 8700 PCB-180 2000 4000 R-HCH 4700 6900 γ-HCH 2600 4200 chlorobenzene 20600 8000 1,2,4,5-tetrachlorobenzene 127100 65300 1,2,3,5-tetrachlorobenzene 65700 29200 pentachlorobenzene 137900 84500 hexachlorobenzene 227500 187100 naphthalene 700 300 biphenyl 800 300 acenaphthene 300 100 fluorene 1200 500 phenanthrene 500 200 anthracene 100 < 50 fluoranthene 1300 600 pyrene 300 100 benzo[a]pyrene 600 300 benz[a]anthracene 400 100 chrysene 400 200 1,2,3,4-TetraCDD 2200 3300

at 5°C

1600 2700 2200 2000 2500 2900 3400 2900 4000 3700 3800 7500 8200 6200 4600 3200 7100 4300 12400 13300 10800 8700 6100 4200 37100 15800 51000 88700 100 200 100 200 100 < 50 300 100 100 100 100 3700

obs

2900

1400

1100 1000 1000 900

When no measured data were available, the activation energy Ea (kJ/mol) for PCBs was calculated using eq 5. We consider estimating Ea this way to be more realistic than assuming it independent of chlorine number. The measured as well as the calculated degradation rate constants in air refer to reactions in the gas-phase only. Since little is known about the degradation of semivolatile organic compounds bound to particles (35), it is assumed that no degradation on aerosols occurs, which is a conservative assumption in that it may overestimate LA.

Results and Discussion The temperature dependence of LA in eq 1 can be understood as the combined effect of the temperature dependence of reactivity in air (kair), total deposition (kdep), and interaction with the surface compartments (stickiness F). Characteristic travel distances were calculated for the 40 chemicals between 0 °C and 30 °C in steps of 5 °C. Selected results are presented in Table 1 and in Figure 1. Full results are given in the Supporting Information. To make the figures easier to read and to show the sensitivity of LA to the temperature change, only representative chemicals were selected, and all values were normalized to the mean value of each chemical. Magnitude and Direction of Temperature Effects. A first observation is that the characteristic travel distance of some 768

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 4, 2003

FIGURE 1. Characteristic travel distance in air LA as a function of temperature for selected substances. The relative deviation from the mean characteristic travel distance of each substance is displayed. Substance parameters are assumed to be temperature dependent as described in the text. Solid lines: hydroxyl radical concentration in air (COH) is simulated as a function of temperature using eq 2. Dashed lines: a constant COH of 5‚105 cm-3 is assumed. chemicals is strongly influenced by temperature, by up to a factor of 6 within the 25 K range (Table 1, e.g. PCB-180, 2,2′,3,4,4′,5,5′-heptachlorobiphenyl). This confirms the need to account for the role of temperature in the assessment of long-range transport potential. The characteristic travel distance, and thus the long-range transport, of these chemicals clearly depends on both seasonal and climatic conditions. A chemical may travel six times as far in winter than in summer or vice versa. Similarly, it may travel six times further if emitted in a Northern country than if emitted in a tropical country or vice versa. Chemical ranking can also significantly change with temperature. For instance chlorobenzene has an LA significantly larger than that of γ-HCH (γ-hexachlorocyclohexane) at 5 °C, whereas the reverse applies at 30 °C (Table 1). Second, we note that the characteristic travel distance can increase, decrease, or remain relatively constant with increasing temperature (Figure 1). Such divergent LA versus temperature relationships even occur within a single substance class such as the PCBs. The PCB-153 congener (2,2′,4,4′,5,5′-hexachlorobiphenyl) has a longer characteristic travel distance at higher temperatures, whereas PCB-15 (4,4′dichlorobiphenyl) displays a decreasing characteristic travel distance. Deposition Rate. The different dependencies of LA on temperature for different chemicals can be explained if viewed as the result of opposing effects related to the terms appearing in eq 1. Since the wind speed is fixed in the model, only the denominator in eq 1 changes with temperature. Figure 2 displays the deposition rate constant kdep as a function of the air-water partition coefficient KAW and the octanol-air partition coefficient KOA. The position of a chemical in this parameter space gives insights into its environmental fate (39). A chemical in the top left corner (i.e. low KOA and high KAW) is mainly present in the gas-phase facilitating atmospheric transport. Deposition rates are slow (kdep < 0.1 d-1), and the atmospheric residence time is largely determined by chemical transformation rates. Chemicals on the right-hand side are strongly particle bound and are efficiently removed by wet and dry particle deposition. Their atmospheric residence time is less dependent on kair. For chemicals that are entirely bound to particles LA approaches a limit at which changes in KOA and KAW as well as kair have little effect on LA (6). Chemicals at the bottom of Figure 2 (low KAW) are highly water soluble and tend to be removed by rain (wet gas-phase deposition). Changes in kair have little effect on the atmospheric residence time of these polar chemicals.

FIGURE 2. Total deposition rate constant kdep in d-1 as function of KOA and KAW. Deposition includes wet and dry particle deposition, wet gaseous deposition, and diffusive uptake into surface media. KOA and KAW of sample chemicals are varied within 5 and 30 °C. A chemical is represented in this space as a trajectory that follows the change of KOA and KAW as function of temperature (Figure 2). Changes along the x-axis and y-axis are determined by the corresponding phase transfer energies ∆UOA and ∆UAW. As temperature increases, a chemical will generally “move down the hill” toward regions with reduced wet and dry deposition rates, thus increasing LA as a result of reduced kdep. Temperature changes will have a strong effect on the deposition when a chemical moves over a large range in this parameter space (provided it does not move perpendicular to the gradient) or along a steep gradient. In the first case the chemical’s partitioning properties are sensitive to temperature changes, and in the second case the deposition rate is sensitive to changes of the partitioning properties. Thus, the sensitivity of the deposition rate with respect to temperature depends on a chemical’s position in Figure 2 and the relative magnitude of ∆UOA and ∆UAW. The characteristic travel distance is also determined by the degradation rates in air and the surface media, with degradation in air having the greatest impact. As a result of both faster gas-phase reactions and higher mass fractions in the gas phase, degradability in air kair always increases with increasing temperature, thus lowering LA. We may thus conclude that LA of a given substance may increase or decrease with temperature increases depending on whether deposition or degradation dominates the overall loss from air and how sensitive the two processes are to temperature changes. Stickiness. Temperature has opposing effects on the stickiness F. Increasing temperature elevates the volatility of the chemicals, which reduces F. On the other hand, it accelerates the rate of degradation in the surface media, thereby increasing F, because a smaller fraction of the chemical remains available for evaporation. A larger F results in more efficient net-deposition (i.e. the term kdep‚F in eq 1 increases) and thus reduces LA. Changes in F have little effect on LA if deposition is insignificant, i.e., if kdep is small compared to kair. Figure 3 illustrates the relationship between LA and stickiness for a hypothetical substance with a log KOA of 8 that does not degrade. Also shown is LA as a function of stickiness and temperature for selected compounds. If a substance does not degrade, i.e., it is infinitely persistent, the LA will increase to an infinitely large value as the stickiness approaches zero, since most of the substance will be found in the atmosphere. When the stickiness approaches 1.0, LA will be limited by the deposition velocity and will consequently experience a minimum. For substances that react in the environment three distinct patterns emerge from Figure 3. First are substances that have a high stickiness and low LA at low temperatures and which

FIGURE 3. Characteristic travel distance in air as a function of total stickiness for a substance that does not degrade in the environment and as a function of both total stickiness and temperature for selected compounds including OH-radical concentration in air simulated as function of temperature.

FIGURE 4. Temperature dependence of the characteristic travel distance in air LA of a hypothetical chemical with stickiness F of 1.0. The characteristic travel distance is proportional to the inverse of the sum of the degradation rate constant kair and the deposition rate constant kdep. This “hill-shaped” relationship is shifted to the left or right for other chemicals, thus minima cannot occur. display an increase in LA with increasing temperature. This is characteristic of γ-HCH or PCB-153 and is typical of substances which have relatively long atmospheric half-lives, that are only marginally affected by temperature. Second are those substances for which stickiness is strongly influenced by temperature and experience decreasing LA with increasing temperature. This is characteristic of the behavior of PCB-15, which has a relatively short half-life in air, that is strongly influenced by temperature. For these substances LA is strongly influenced by the rate of degradation, whereas changes to the rate of deposition are less important. Third are substances that experience an increasing stickiness with increasing temperature. This is characteristic of hexachlorobenzene, which becomes more soluble in water with increasing temperature, causing an increase in the stickiness in water. All substances will encounter a temperature at which they experience a maximum LA as illustrated in Figure 4. The temperature dependence of the reaction rate kair and the deposition rate kdep is modeled such that kair will monotonically increase with rising temperature, whereas kdep will monotonically decrease (Figure 4). The sum of these rates will thus always have a global minimum. This could only be VOL. 37, NO. 4, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

769

contravened if F were to increase substantially with rising temperature. However, in our calculations temperature variation of the stickiness generally had a much smaller effect on LA than variability of the deposition rate. Since LA is proportional to the inverse of the sum kair and F‚kdep, its temperature dependence will display a maximum. Changes in F will have little effect on this shape. The position of the maximum of this relationship will lie at different temperatures for different chemicals and may not occur in an environmentally relevant range. Hydroxyl Radical Concentration. The OH-radical concentration COH has a strong effect on kair of all chemicals independent of their Ea, because kair is assumed to be linearly related to COH. However, the effect on LA when varying COH with temperature also depends on the relative importance of deposition versus degradation. The dashed lines in Figure 1 were produced assuming a fixed COH. Thus, in this scenario only substance parameters were assumed to be temperature dependent. For instance, with a fixed COH the relatively volatile PCB-15 has a much shorter LA at low temperatures, which is related to the decreasing OH-concentration assumed in the first scenario. Due to its volatility PCB-15 is mainly removed from the atmosphere by reaction with hydroxyl radicals. The PCB-153 congener, on the other hand, is less affected by this difference, because its removal from air is dominated by deposition. The temperature dependence of benzo[a]pyrene’s LA is less affected by COH although it is similarly sensitive to changes in kair as PCB-15 (not shown). The reason for this difference is the changing particle bound fraction, which also affects the removal by chemical transformation. Since PCB-15 is almost entirely in the gas phase even at 5 °C, it is relatively insensitive to temperature if a constant OH-radical concentration is assumed. Hence, the temperature dependence of benzo[a]pyrene’s LA is due to an increasing fraction of chemical in the gas phase which makes it available for degradation at high temperatures. Comparison to Monitoring Data. Regrettably, it is very difficult to obtain monitoring data over a transect with variable temperature to validate these assertions, since the monitoring data must satisfy several criteria. The substances should ideally have been emitted from a single source at one end of the transect at rates with similar time profiles. The predominant wind direction should be along the transect. Environmental characteristics such as the nature of soils and vegetation should be similar along the transect. The simulated temperature should reflect average conditions. It is unlikely that all these criteria can be satisfied and the model cannot be expected to simulate transport distances quantitatively. It should, however, be capable of simulating qualitatively the ratio of concentrations of substances with differing LA along the transect (20). Here, we compare the observed long-range transport behavior of PCBs (40) with the corresponding simulated values from Table 1. Ockenden et al. (40) measured PCBs in air along a transect from England in the south to Northern Norway using passive samplers, which yield time averaged concentrations. Assuming that most of the primary emission of PCBs occurred in the south, concentration gradients were calculated by log-linear regression of the concentrations versus distance from monitoring site C (Lancaster, see ref 40). A log-linear regression is consistent with the assumption of an exponential concentration decrease underlying eq 1 (6). Details of this regression procedure are given in ref 20, and the resulting concentration gradients are reported in Table 1. A direct quantitative agreement between the values in Table 1 cannot be expected for the reasons stated above, as well as the arbitrarily assumed wind speed and that dilution due to atmospheric eddy diffusion (dispersion) is not taken into account. However, neither factor affects the ranking 770

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 4, 2003

FIGURE 5. Observed and simulated characteristic travel distances (LA) for selected PCB congeners. All LA are normalized by the mean LA of all PCBs shown in this figure to depict the deviation between congeners. Observed LA are derived from values reported in ref 40 by log-linear regression. between the chemicals and thus, a comparison of the relative differences between the PCB congeners should be feasible. We therefore normalized the observed and the simulated spatial gradients by the mean values of all PCBs (Figure 5). Figure 5 shows that the scenario at 5 °C best reproduces the observed relative differences between the PCBs by suggesting a longer LA for the lighter PCBs 52 and 101. The observed and calculated ranking at 5 °C are in agreement. Contrary to the observations, the calculations for 15 °C and 30 °C predict that PCB-153 and PCB-138 have the highest long-range transport potential among the investigated PCBs. Also the 0 °C calculations agree less with the observations than the 5 °C scenario, i.e., the 5 °C scenario is closest to the observed pattern. Average ambient temperatures along this transect are close to 5 °C (40), which highlights that adjusting chemical properties to actual environmental temperatures can improve the environmental fidelity of model results. In addition these findings show the model’s capability of revealing qualitative differences between chemicals. Implications for Environmental Transport. The different temperature dependencies of LA have implications for longrange transport to cold regions such as the Arctic or Antarctic. Our results suggest that as a chemical is transported to a cooler region its transport potential may increase or decrease and further atmospheric transport can be inhibited or accelerated. According to the model, atmospheric concentrations will always decrease as a chemical is transported into a cooler region, because deposition will become more efficient. However, the fraction of the chemical removed by deposition, as opposed to being transported further, can decrease. For example, LA of PCB-15 would increase as T decreases. For such a compound, deposition would not be very efficient, even at lower temperatures, since most PCB15 would still be in the gas phase. Coupled with the slow degradation in air under low T in the Arctic, the fraction being removed by deposition compared to that being transported further would decrease as T decreases. This divergent behavior of the chemicals with respect to temperature changes can therefore increase the global fractionation of semivolatile organic chemicals (8, 41). Whereas it can be stated quite generally that the persistence of a chemical will increase when temperature drops, a correspondingly general statement is not possible for the atmospheric transport potential. The actual fate strongly depends on the properties of the chemical. In view of the magnitude of its effects, temperature must be considered in multimedia fate models, particularly when using models to assess long-range transport potential. The characteristic travel distance may behave nonintuitively because of the interplay of many competing factors, and the effect of temperature cannot be anticipated without using a model that accounts for these factors. The results further underline the importance of measuring physical-chemical

properties and degradation rate constants as functions of temperature.

Acknowledgments Funding by the German Federal Environmental Agency, Berlin (R&D Project FKZ 299 65 402), the Canadian Natural Sciences and Engineering Research Council, the Northern Contaminants Program of the Canadian Department of Indian Affairs and Northern Development, and the consortium of chemical companies that support the Canadian Environmental Modeling Centre is gratefully acknowledged.

Supporting Information Available Tables with degradation rates that were used for this work, tables reporting detailed results, and a figure showing the regression of PCB activation energy versus homologue group. This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Vallack, H. W. et al. Environ. Toxic. Pharmacol. 1998, 6, 143175. (2) Rodan, B. D.; Pennington, D. W.; Eckley, N.; Boethling, R. S. Environ. Sci. Technol. 1999, 33, 3482-3488. (3) Scheringer, M. Environ. Sci. Technol. 1996, 30, 1652-1659. (4) Scheringer, M. Persistence and Spatial Range of Environmental Chemicals; Wiley-VCH: Weinheim, Germany, 2002. (5) Bennett, D. H.; McKone, T. E.; Matthies, M.; Kastenberg, W. E. Environ. Sci. Technol. 1998, 32, 4023-4030. (6) Beyer, A.; Mackay, D.; Matthies, M.; Wania, F.; Webster, E. Environ. Sci. Technol. 2000, 34, 699-703. (7) Mackay, D.; Webster, E.; Beyer, A.; Matthies, M.; Wania. F. In Persistent, bioaccumulative, and toxic chemicals II: Assessment and new chemicals; Lipnick, R. L., Jansson, B., Mackay, D., Petreas, M., Eds.; American Chemical Society: Washington, DC, 2001. (8) Wania, F.; Mackay, D. Ambio 1993, 22, 10-18. (9) Wania, F.; Haugen, J. E.; Lei, Y. D.; Mackay, D. Environ. Sci. Technol. 1998, 32, 1013-1021. (10) Wania, F.; Axelman, J.; Broman, D. Environ. Pollut. 1998, 102, 3-23. (11) Wania, F.; Hoff, J. T.; Jia, C. Q.; Mackay, D. Environ. Pollut. 1998, 102, 25-41. (12) Wania, F.; Mackay, D.; Hoff, J. T. Environ. Sci. Technol. 1999, 33, 195-197. (13) Halsall, C. J.; Gevao, B.; Howsam, M.; Lee, R. G. M.; Ockenden, W. A.; Jones, K. C. Atmos. Environ. 1999, 33, 541-552. (14) Simcik, M. F.; Basu, I.; Sweet, C. W.; Hites, R. A. Environ. Sci. Technol. 1999, 33, 1991-1995. (15) Lee, R. G. M.; Hung, H.; Mackay, D.; Jones, K. C. Environ. Sci. Technol. 1998, 32, 2172-2179. (16) Wania, F.; Mackay, D.; Li, Y. F.; Bidleman, T. F.; Strand, A. Environ. Toxicol. Chem. 1999, 18, 1390-1399. (17) Wania, F.; Persson, J.; Di Guardo, A.; McLachlan, M. S. The POPCYCLING-Baltic model. A non-steady-state multicompartment mass balance model of the fate of persistent organic pollutants in the Baltic Sea environment; Technical Report; Norwegian Institute for Air Research (NILU): Kjeller, Norway, 1999.

(18) Paraiba, L. C.; Bru, R.; Carrasco, J. M. Environ. Model. 2002, 147, 221-232. (19) Scheringer, M.; Wegmann, F.; Fenner, K.; Hungerbu ¨ hler, K. Environ. Sci. Technol. 2000, 34, 1842-1850. (20) Beyer, A.; Matthies, M. Criteria for atmospheric long-range transport potential and persistence of pesticides and industrial chemicals; Erich Schmidt: Berlin, Germany, 2002. (21) European Community. EUSES - the European Union system for the evaluation of substances; Institute of Public Health and the Environment (RIVM): Bilthoven, The Netherlands, 1996. (22) European Community. Technical guidance document in support of the commission directive 93/67/EEC on risk assessment for new notified substances and the commission regulation (EC) 1488/94 on risk assessment for existing substances, Parts I-IV; Office for Publications of the European Communities: Luxemburg, 1996. (23) Brandes, L. J.; den Hollander H.; van de Meent, D. SimpleBox 2.0: a nested multimedia fate model for evaluating the environmental fate of chemicals; Report No. 719101029; National Institute of Public Health and the Environment (RIVM): Bilthoven, The Netherlands, 1996. (24) McLachlan, M. S.; Czub, G.; Wania, F. Environ. Sci. Technol. 2002, 36, 4860-4867. (25) Altshuller, A. P. JAPCA 1989, 39, 704-708. (26) Rodriguez, J. M.; Ko, M. K. W.; Sze N. D.; Heisey, C. W. In Global biomass burning; Levine, J. S., Ed.; The MIT Press: Cambridge, MA, 1992; pp 351-359. (27) Finizio, A.; Mackay, D.; Bidleman, T.; Harner, T. Atmos. Environ. 1997, 31, 2289-2296. (28) Harner, T.; Bidleman, T. F. Environ. Sci. Technol. 1998, 32, 14941502. (29) Falconer, R. L.; Harner, T. Atmos. Environ. 2000, 34, 40434046. (30) Junge, C. E. In Fate of pollutants in the air and water environments; Suffet, I. H., Ed.; Wiley: New York, 1977. (31) Pankow, J. F. Atmos. Environ. 1994, 28, 185-188. (32) Beyer, A.; Wania, F.; Gouin, T.; Mackay, D.; Matthies, M. Environ. Toxicol. Chem. 2002, 5, 941-953. (33) Mackay, D.; Shiu, W. Y.; Ma, K. C. Illustrated handbook of physical-chemical properties and environmental fate, (CD-ROM); Chapman & Hall/CRCnetBase: Boca Raton, FL, 2000. (34) Anderson, P. N.; Hites, R. A. Environ. Sci. Technol. 1996, 30, 1756-1763. (35) Atkinson, R.; Guicherit, R.; Hites, R. A.; Palm, W. U.; Seiber, J. N.; de Voogt, P. Water, Air, Soil Pollut. 1999, 115, 219-243. (36) Brubaker, W. W.; Hites, R. A. Environ. Sci. Technol. 1997, 31, 1805-1810. (37) Brubaker, W. W.; Hites, R. A. Environ. Sci. Technol. 1998, 32, 766-769. (38) Brubaker, W. W.; Hites, R. A. J. Phys. Chem. 1998, 102, 915-921. (39) Gouin, T.; Mackay, D.; Webster, E.; Wania, F. Environ. Sci. Technol. 2000, 34, 881-884. (40) Ockenden, W. A.; Sweetman, A. J.; Prest, H. F.; Steinnes, E.; Jones, K. C. Environ. Sci. Technol. 1998, 32, 2795-2803. (41) Wania, F.; Mackay, D. Environ. Sci. Technol. 1996, 30, 390A396A.

Received for review April 15, 2002. Revised manuscript received October 22, 2002. Accepted November 20, 2002. ES025717W

VOL. 37, NO. 4, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

771