The Influence of Vertical Sorbed Phase Transport ... - ACS Publications

diffusion equation proposed by Jury et al. to describe organic chemical fate in a uniformly contaminated surface soil was modified to include vertical...
0 downloads 0 Views 359KB Size
Download PDF
Environ. Sci. Technol. 2002, 36, 4860-4867

The Influence of Vertical Sorbed Phase Transport on the Fate of Organic Chemicals in Surface Soils M I C H A E L S . M C L A C H L A N , * ,† GERTJE CZUB,† AND FRANK WANIA‡ Baltic Sea Research Institute, Seestrasse 15, D-18119 Rostock, Germany, and Department of Chemistry and Division of Physical Sciences, University of Toronto at Scarborough, 1265 Military Trail, Toronto, Ontario, Canada M1C 1A4

Gaseous exchange between surface soil and the atmosphere is an important process in the environmental fate of many chemicals. It was hypothesized that this process is influenced by vertical transport of chemicals sorbed to soil particles. Vertical sorbed phase transport in surface soils occurs by many processes such as bioturbation, cryoturbation, and erosion into cracks formed by soil drying. The solution of the advection/ diffusion equation proposed by Jury et al. to describe organic chemical fate in a uniformly contaminated surface soil was modified to include vertical sorbed phase transport. This process was modeled using a sorbed phase diffusion coefficient, the value of which was derived from soil carbon mass balances in the literature. The effective diffusivity of the chemical in a typical soil was greater in the modified model than in the model without sorbed phase transport for compounds with log KOW > 2 and log KOA > 6. Within this chemical partitioning space, the rate of volatilization from the surface soil was larger in the modified model than in the original model by up to a factor of 65. The volatilization rate was insensitive to the value of the sorbed phase diffusion coefficient throughout much of this chemical partitioning space, indicating that the surface soil layer was essentially well-mixed and that the mass transfer coefficient was determined by diffusion through the atmospheric boundary layer only. When this process was included in a non-steady-state regional multimedia chemical fate model running with a generic emissions scenario to air, the predicted soil concentrations increased by up to a factor of 25, while the air concentrations decreased by as much as a factor of ∼3. Vertical sorbed phase transport in the soil thus has a major impact on predicted air and soil concentrations, the state of equilibrium, and the direction and magnitude of the chemical flux between air and soil. It is a key process influencing the environmental fate of persistent organic pollutants (POPs).

Introduction Soil is the major repository of many organic chemicals in the environment, especially persistent lipophilic organic com* Corresponding author phone: +49 381 5197 300; fax: +49 381 5197 302; e-mail: [email protected]. † Baltic Sea Research Institute. ‡ University of Toronto at Scarborough. 4860

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 22, 2002

pounds that accumulate preferentially in organic matter compared with water and air (1-3). The multimedia behavior of such chemicals, and in particular their long range transport potential, is largely determined by soil-atmosphere exchange. In this paper we identify a major deficit in our current understanding of this process and explore its implications.

Background Many models of the exchange of organic contaminants between soils and the atmosphere are based on the landmark publication of Jury et al. (1). The evaluative model developed in this paper (hereafter referred to as the Jury model) treats soil as a mixture of air, water, and soil particles and assumes uniform soil properties, linear sorption isotherms, and equilibrium partitioning between the solid, air, and water phases. Following Jury eqs 2 and 3 (1), octanol is used as a surrogate for the sorption properties of the soil solids based on an equation from Rao and Davidson. Therefore the partitioning of a chemical between the three soil phases is defined by any two of the octanol-water partition coefficient (KOW), the octanol-air partition coefficient (KOA), and the air-water partition coefficient (KAW), whereby KOW ) KOA × KAW. Note that the Jury model does not consider the enhanced sorption to soil solids when the relative humidity in the soil drops below 100% (4, 5). The chemical concentration is assumed to be homogeneous in the horizontal plane, and a one-dimensional mass balance is conducted (i.e. in the vertical (Z) direction). Advection with porewater and diffusion through the soil air and the soil water phases are the transport processes considered, while sorbed phase transport and hydrodynamic dispersion are ignored. Degradation of the chemical in the soil is also accounted for by applying linear first-order kinetics

∂CT ∂2 C T ∂CT ) DE 2 - VE - µCT ∂t ∂Z ∂Z

(1)

where CT is the total concentration in the soil [mol‚m-3], t is time [d], DE is the effective diffusion coefficient [m2‚d-1], VE is the effective solute convection velocity [m‚d-1], and µ is the rate constant for degradation [d-1]. The effective diffusion coefficient is given by

DE ) DG/RG + DL/RL

(2)

where DG and DL are the soil-gas and soil-liquid diffusion coefficients [m2‚d-1] (each equal to the product of the molecular diffusivity multiplied by a tortuosity factor calculated from the Millington-Quirk equation), and RG and RL are the ratios of the total concentration in soil to the concentration in the gas phase and liquid phase, respectively. The effective solute convection velocity is given by

VE ) JW/RL

(3)

where JW is the liquid water flux per unit soil surface area [m‚d-1]. Jury et al. present an analytical solution to this model for the case of an initially uniform soil concentration between the soil surface and a fixed soil depth L, an initial concentration of 0 at greater soil depths, a constant concentration of 0 in the atmosphere, a constant concentration of 0 at infinite soil depth, and aqueous phase convection by a steady water 10.1021/es025662y CCC: $22.00

 2002 American Chemical Society Published on Web 10/19/2002

TABLE 1: Properties Used in the Calculations air diffusion coefficient water diffusion coefficient sorbed phase diffusion coefficient (DS) soil porosity soil bulk density organic carbon mass fraction water content depth of incorporation (L) leaching rate (JW) atmosphere-soil mass transfer coeff (h)

0.43 m2‚d-1 4.3‚10-5 m2‚d-1 0; 5.5‚10-7; 1.09‚10-6 m2‚d-1 0.5 1350 kg‚m-3 0.025 0.3 0.01 (Figure 1); 0.1 m 0; 5‚10-3 m‚d-1 90 m‚d-1

is largely due to degradation. The chemical has however moved very little within the soil column and is still almost completely present in the top 1 cm. This stands in contrast to field observations. For instance, the PCB concentrations in an undisturbed grassland soil were reported to decrease only gradually from the surface down to a depth of 17 cm (9) (see Figure 1). Other persistent lipophilic organic compounds such as PCDD/Fs and PAHs are also present in subsurface soil layers at concentrations comparable to those found in the surface soil (10, 11). This suggests that the model is missing a key feature of PCB soil fate. In a companion paper to their soil profile study, Cousins et al. assembled a multilayer soil model to evaluate their data and concluded that bioturbation was the dominant transport mechanism for PCBs (12). The failure of current contaminant transport models to account for sorbed phase transport may thus be a major problem in modeling the fate of persistent lipophilic organic chemicals.

Vertical Sorbed Phase Transport in Soil

FIGURE 1. Distribution of PCB 101 in a soil 10 years after its uniform incorporation at unit concentration to a depth of 1 cm. The broken line shows the predicted concentration profile with no sorbed phase transport; the solid line shows the predicted concentration profile with a sorbed phase transport characterized by a sorbed phase diffusion coefficient of 5.5‚10-7 m2‚d-1. The triangles show PCB concentrations in an undisturbed grassland soil sampled in England in 1996 (from ref 12). flux (see eq 24 in ref 1). This scenario describes chemical elimination from a uniformly contaminated surface soil layer.

Application of the Jury Model to PCB 101 The analytical solution to this scenario was used to explore the fate of persistent lipophilic organic compounds assuming a shallow initially contaminated surface layer of 1 cm depth. A typical persistent lipophilic organic chemical, the polychlorinated biphenyl (PCB) congener 101 (IUPAC No.), was chosen as an example. PCBs entered soils throughout the world as a result of atmospheric deposition and are initially retained close to the soil surface. In industrial countries the levels of PCBs in air and soils have decreased markedly over the last 30 years (6). For such situations, the scenario that Jury treated is likely relevant. Equation 24 in ref 1 was used to calculate the concentration as a function of depth 10 years after the initial contamination (note that in ref 1 VEHE in line 3 of the equation should be VE/HE; note also the missing closing bracket after VE in the second last line (7)). The default soil properties and diffusivities in water and air proposed by Jury et al. (1) were employed. The mass transfer coefficient for the laminar boundary layer h was set to a value resulting from the stagnant air layer thickness of 4.75 × 10-3 m suggested in the Jury paper (see Table 1). The physical chemical properties used for PCB 101 were as follows: log KOW ) 6.31, log KAW (8 °C) ) -1.95; µ ) 0.000166 d-1 (8). The solution was programmed in MATLAB. To obtain a solution for the exponential and erfc functions, erfc(x) was approximated as 0.56 exp(-x2)/x for x > 6. The results are plotted in Figure 1. After 10 years the concentration of PCB 101 has decreased by about 55%. This

There are different mechanisms by which sorbed phases can be vertically transported in soils, including bioturbation, cryoturbation, and macropore transport. Mammals, ants, termites, wood lice, and worms can all contribute to bioturbation of surface soils (13). Moles have been reported to move 10-120 t‚ha-1‚yr-1 of soil in Russian forests, while ground squirrels transport 1.4-1.6 t‚ha-1‚yr-1 of soil from their burrows to the surface in semiarid regions around the Caspian Sea. Ants transport 50 t‚ha-1‚yr-1 of sand to the surface in forests in North Germany, and in Afghanistan wood lice move 1.5 t‚ha-1 of soil within a 3 month period. Hence, there are clearly many different actors contributing to bioturbation in diverse environments. Globally, the most significant contributor to bioturbation may be earthworms. They are common in pastureland and nonplowed agricultural land in temperate regions (note that in plowed agricultural land human activity dominates bioturbation) (13). The weight of the wormcasts thrown on the surface ranged from 1 to 25 t‚ha-1‚yr-1 in a study in England (14) and were as high as 250 t‚ha-1‚yr-1 in Germany (13), but most estimates for temperate pastures and grasslands are about 40-50 t‚ha-1‚yr-1 (15). Earthworms are believed to be the most important factor controlling the vertical transport of radionuclides in central European soils (16), and it is not surprising that earthworms have been shown to accelerate the transport of pesticides downward into the soil (17). In tropical soils the wormcast production ranges from that in temperate soils up to 1200 t ha-1‚yr-1 (15). In colder climates, where bioturbation is low, the freeze/ thaw process results in physical mixing of soil particles. This effect, called cryoturbation, has been shown to play an important role in carbon storage and distribution in tundra soils (18). Vertical mixing of soils also occurs via transport in macropores. These macropores can be formed by burrowing fauna, the decay of roots, or the drying and shrinking of the soil which can produce deep cracks. Physical (e.g. wind, water) and biological agents (e.g. animals) cause particles to fall down these macropores. This has been shown to contribute to the downward transport of sorbed soil constituents (19, 20). Interestingly, in sediments the influence of vertical sorbed phase transport on contaminant exchange with the water column has been the subject of considerable study (21, 22), but in soils the analogous process has not yet been addressed.

Modification of the Jury Model To Include Vertical Sorbed Phase Transport There are a range of models to describe bioturbation, particularly in sediments, some of which are quite sophisVOL. 36, NO. 22, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4861

ticated (23). However, for the illustrative application here we were interested in treating the aggregate of all processes contributing to vertical sorbed phase transport. To facilitate incorporation in the Jury modeling framework, vertical sorbed phase transport was treated as a diffusive process. Jury’s definition of the effective diffusivity DE was expanded to include sorbed phase diffusion where DEs is the modified

DEs ) DG/RG + DL/RL + DS/RS

(4)

effective diffusivity, DS is the diffusion coefficient for the soil sorbed phase, and RS is the ratio of the total soil concentration to the concentration in the sorbed phase (as defined in ref 1). Elzein and Balesdent modeled the transport of carbon in soils using this approach (24). They treated the transport of soil carbon as a combination of a diffusive and a vertical advective process. Due to their preferential sorption to soil organic matter, the transport of lipophilic organic chemicals is expected to be closely linked to the transport of soil carbon. In fitting their model to the carbon and ∆14C depth profiles of five forest soils from France, India, and Brazil, they obtained carbon diffusion coefficients ranging from 1 to 15 cm2‚yr-1 and convection coefficients ranging from 0.13 to 0.6 mm‚yr-1. Although it was expected that the diffusion coefficients would decrease with soil depth, the numerical analysis predicted constant diffusion coefficients throughout the surface soil profile. A bioturbation diffusion coefficient can also be approximated from the wormcast data using the concept of a characteristic velocity and a characteristic path length (25). An average annual surface accumulation of 40 t‚ha-1 (see above) corresponds to a soil accumulation rate of 0.3 cm‚yr-1 (assuming a soil bulk density of 1.3). Multiplying by an assumed typical vertical displacement of 5 cm yields a diffusion coefficient of 1.5 cm2‚yr-1, which lies in the range of carbon diffusion coefficients derived by Elzein and Balesdent. To ensure a conservative evaluation of the influence of vertical sorbed phase transport, a DS value from the low end of the range reported by Elzein and Balesdent (2 cm2‚yr-1 or 5.5‚10-7 m2‚d-1) was chosen for this paper. In their model, Elzein and Balesdent also considered vertical advection. The influence of this process on contaminant transport was tested by modifying the definition of the effective solute convection velocity VE in the Jury model

VEs ) JW/RL + JS/RS

(5)

where VEs is the modified effective solute convection velocity, and JS is the sorbed phase convection coefficient (m‚d-1). Using the values of JS reported by Elzein and Balesdent, it was found that this modification had a negligible impact on the model predictions compared to sorbed phase diffusion (results not shown), and it was discarded. Influence on the Effective Diffusivity in Soil. To evaluate the importance of sorbed phase transport for chemical transport in soil, the quotient of the effective diffusivity with and without sorbed phased diffusion (DEs/DE) was calculated. The chemical diffusion coefficients in air and water and the soil properties suggested by Jury were employed (see Table 1). The results are plotted as a function of the chemical partitioning properties KOA and KAW in Figure 2 (diagonal lines correspond to uniform log KOW values). Note the logarithmic scale for all three dimensions. In reading this figure, the influence of temperature on partitioning can be accounted for by using the KOA, KOW, and/or KAW value at the temperature of interest. For compounds with a low KOW and/ or a low KOA the quotient is 1, indicating that sorbed phase 4862

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 22, 2002

FIGURE 2. Influence of sorbed phase transport on the effective diffusivity DE of organic chemicals. The effective diffusivity was calculated using the soil properties in Table 1 with sorbed phase transport (DEs ; DS ) 5.5‚10-7 m2‚d-1) and without sorbed phase transport (DE). The quotient log(DEs/DE) is plotted as a function of log KOA and log KAW. transport does not influence the fate of these substances. However, for lipophilic compounds (high KOA and KOW), the quotient is much greater than 1. For these chemicals sorbed phase transport is very important. Because they are almost completely sorbed in soil, transport via the gas and liquid phases is very slow. A large number of environmental chemicals fall in this chemical partitioning space including all POPs. Their effective diffusion coefficient is severely underpredicted by the original model. Note that sorbed phase transport becomes important at relatively low log KOW values of about 2 (lower right of Figure 2). The diffusion coefficient for sorbed phase transport is of similar magnitude to the diffusion coefficient for transport in soil pore water. Therefore, the diffusive transport is comparable in the sorbed and aqueous phases when each contains similar fractions of the contaminant. For a soil with an organic carbon content of 2.5%, this occurs at a log KOW of ∼2. Note also that hydrodynamic dispersion, which is not included in the model, increases the soil-liquid diffusion coefficient when JW * 0. This will influence the relative importance of solid phase and liquid phase transport in soil, the former only becoming important at higher values of log KOW. Effect on Overall Fate in Soil. Having identified an influence of sorbed phase transport on the effective diffusivity of lipophilic chemicals in soil, its effect on overall chemical fate in soil was studied. For a persistent substance, the processes determining fate in surface soil are downward transport into deeper soil layers and volatilization. Both of these could be affected by vertical sorbed phase transport. However, the downward transport is not realistically reflected in the Jury screening model scenario because a constant effective diffusion coefficient with depth is assumed, whereas in most cases sorbed phase transport will decrease beneath a certain depth due to a reduction in bioturbation. Consequently, the investigation focused on the influence of sorbed phase transport on volatilization. Using eq 25 in ref 1, the time required for 10% of the chemical initially incorporated into the surface soil layer to volatilize was calculated. The chemical was assumed to be persistent (µ ) 0), and the properties listed in Table 1 were employed. The time for 10% volatilization was calculated without sorbed phase transport (tVo, DS ) 0) and with sorbed phase transport (tVs, DS ) 2 cm2‚yr-1). This calculation was carried out for the relevant chemical partitioning space identified in Figure 2 (log KAW > -8, log KOA > 6).

FIGURE 3. Influence of sorbed phase transport on the rate of volatilization of persistent organic chemicals from a contaminated soil surface layer 10 cm deep. The time for 10% of the chemical to be volatilized was calculated with sorbed phase transport (tVs) and without sorbed phase transport (tVo). The quotient tVo/tVs is plotted as a function of log KOA and log KAW for the case of (a) no leaching (JW ) 0) and (b) leaching (JW ) 0.005 m‚d-1). The quotients of the volatilization times without and with sorbed phase transport (tVo/tVs) are plotted in Figure 3a for the case of no leaching (JW ) 0). The ratio tVo/tVs is > 1 throughout most of the figure, which indicates that the rate of volatilization is larger when sorbed phase transport is included in the model. For values of log KOA less than 8.5, the effect of sorbed phase transport on the cumulative volatilization varies with the physical chemical properties in a manner similar to the effect on effective diffusivity (see Figure 2), with tVo/tVs increasing with KOA. However, in contrast to DEs/DE which increases continually with increasing KOA, tVo/ tVs reaches a maximum value of 60-70. This is due to the transport resistance of the stagnant air layer above the soil which places a limit on the rate of volatilization. For compounds in the plateau region of Figure 3a (log KOA > 8.5) the sorbed phase transport is so efficient that diffusion through the stagnant air layer becomes limiting. In a laboratory study of volatilization from sediments exposed to air, it was observed that reworking the sediments increased the volatilization of phenanthrene and dibenzofuran, which is in agreement with the impact of sorbed phase transport discussed above (26). Note that an analogous behavior is known for sediments, albeit for a different chemical partitioning space; in highly bioturbated sediments watersediment exchange of many chemicals is controlled by the water-side resistance (22). For even higher log KOA values (> 9.5), tVo/tVs decreases, approaching 0 at log KOA ) 10.5. For chemicals with KOA > 9.5 the time required for 10% volatilization exceeded 20 years. At these time scales the downward transport of chemical as a result of sorbed phase diffusion becomes significant. This reduces the concentrations in the surface soil and thus the rate of volatilization compared to the case if there was no downward sorbed phase transport out of the surface soil layer, leading to tVo/tVs values lower than expected. For log KOA > 10.5, tVs cannot be calculated as 10% volatilization is never achieved (> 90% of the chemical is transported downward). This result reflects the limitations of the screening model scenario; in reality the sorbed phase transport is likely to decrease with depth, and these low values of tVo/tVs would not be expected. As with DEs/DE, tVo/tVs is not elevated at values of log KOW < 2. Chemicals in this property range are primarily dissolved in the pore water and not sorbed to soil solids. Therefore

sorbed phase transport has little influence on the volatilization rate. Figure 3b shows tVo/tVs when pore water leaching is included in the model (JW ) 0.005 m‚d-1). The influence of sorbed phase transport on volatilization for log KAW > -0.5 is the same as for the case with no pore water leaching (Figure 3a). However, for compounds with -2 < log KAW < -0.5, the effect of sorbed phase transport on volatilization is much more pronounced. In this property range a small fraction of the chemical partitions from the sorbed phase into pore water. On the time scale of years, the porewater leaching results in a small downward flux of chemical that depletes the concentration at the surface, reducing volatilization. When sorbed phase transport is included, upward mixing of soil particles compensates for the downward flux of chemical with porewater, and the rate of volatilization is not reduced. For chemicals with log KAW < -2, 10% volatilization was not achieved because > 90% of the chemical was removed by leaching. Other simulations showed that for values of log KOW < 3 the influence of sorbed phase transport is no longer observed because the fraction of the chemical present in the sorbed phase is so small compared to the fraction present in the pore water that the upward flux due to sorbed phase transport is negligible compared to the leaching flux. The modified model was used to re-estimate the soil concentration of PCB 101 for the scenario described above (see Figure 1). The predicted depth profile after 10 years is very different from the profile obtained with the original model. Instead of a sharp peak confined to the originally contaminated surface layer, a broad peak spread down to a depth of > 15 cm is observed. This is more consistent with the soil depth distribution of PCB 101 observed in undisturbed grassland soil (Figure 1). The model predicts that sorbed phase transport has a large influence on the fate of PCBs in soil, which concurs with the findings of ref 12. Sensitivity Analysis. The earlier discussion indicated that there is considerable variability in sorbed phase transport in soils. Choosing an appropriate value of DS is therefore a potential difficulty in assembling models of contaminant behavior. To evaluate this problem, the sensitivity of the model predictions to the values of DS was evaluated. Figure 4 shows the quotient of tVs calculated with DS ) 4 cm2‚yr-1 and DS ) 2 cm2‚yr-1. Doubling the value of the sorbed phase diffusion coefficient increased the rate of VOL. 36, NO. 22, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4863

and kSA is the overall mass transfer coefficient for soil-air exchange (m‚d-1), h is the mass transfer coefficient for transport through the boundary layer above the soil (m‚d-1), and L is the depth of the surface soil layer (m). This approach has been used in a wide range of multimedia models (2831), although it is recognized that it is problematic for lipophilic substances (32). This multimedia submodel was modified to include the effects of sorbed phase diffusion. DE in eq 7 was redefined according to eq 4. To evaluate the ability of the modified model to predict soil-air exchange, the time required for 10% volatilization of chemical was calculated using the equation

FIGURE 4. Sensitivity of the volatilization loss from soil to the value of the sorbed phase diffusion coefficient DS. The chemicals were assumed to be persistent, and the leaching rate was set to 0.005 m‚d-1. The time for 10% of the chemical to volatilize was calculated for DS ) 5.5‚10-7 m2‚d-1 (tVs) and DS ) 1.09‚10-6 m2‚d-1 (tV2s). The quotient tVs/tV2s is plotted as a function of log KOA and log KAW. volatilization by less than one-third throughout much of the chemical partitioning space (log KOA > 8). The decrease in the rate of volatilization observed at high KOA values is due to the nonrealistic aspect of the screening model scenario discussed above, namely downward sorbed phase diffusion out of the surface layer. A marked effect of DS on tVs is observed only along the band of log KOA ∼ 6-7. The insensitivity throughout much of the chemical partitioning space can be explained by the fact that the transport resistance of the stagnant air layer above the soil surface limits volatilization or, in other words, the soil is essentially well-mixed. For most chemicals, a highly accurate value of DS is not needed to obtain a reasonable estimate of volatilization.

The Influence of Vertical Sorbed Phase Transport on the Multimedia Fate of Organic Chemicals Soil-air exchange plays a major role in the multimedia fate of many persistent organic chemicals. Given its pronounced impact on soil-air exchange, sorbed phase transport can also be expected to influence the overall behavior of chemicals in the environment. To investigate this question, an existing multimedia fate model was modified to include the effects of sorbed phase transport. Comparison of a Multimedia Submodel with the Jury Model. The description of chemical fate in surface soil used in many multimedia fate models is based on a simplified implementation of eqs 1-3 (25, 27). The surface soil is assumed to be a single well-mixed compartment of fixed depth (e.g. 10 cm). Leaching is treated as a constant advective flow of soil porewater as in eq 1. Air-soil gas exchange is described using a two resistance model, whereby the major difference to the Jury model is that the soil-side resistance is estimated from the quotient of the effective diffusivity (in the soil-air and porewater) and a constant diffusion path length which is some fraction of the depth of the soil compartment (e.g. half). Assuming the concentration of the chemical in the atmosphere to be zero and neglecting degradation, the mass balance equation for the surface soil compartment is given by

∂CT JW ) - kSACT - CT ∂t RL

L

(6)

where

kSA ) 4864

9

[

]

RG 0.5L + h DE

-1

(7)

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 22, 2002

tV ) -

[ (

)]

kSA + JW/RL L 1 - ln 0.1 kSA + JW/RL kSA

(8)

employing the properties in Table 1 and setting DS ) 2 cm2‚yr-1. In Figure 5a the quotient of this time and the corresponding time calculated with the Jury model is plotted for the case of no leaching (JW ) 0). In general the quotients are greater than one, indicating that the multimedia submodel underpredicts the rate of volatilization. Good agreement is observed between the two models for log KOA > 8. Although the quotients drop below 1 for log KOA > 9, this is again just a reflection of the downward mixing of the sorbed phase in the Jury model as discussed for Figure 4. Without this effect the agreement between the two models would be good as the volatilization is governed by the resistance of the stagnant boundary layer which both models treat identically. For lower values of KOA the agreement deteriorates somewhat, with the multimedia submodel underpredicting the rate of volatilization by up to a factor of 5. In this portion of the chemical partitioning space the multimedia submodel overestimates the soil-side diffusion resistance due to the arbitrarily long diffusion depth. Particularly interesting is the more severe deviation between the models for log KOW < 3 (lower left of Figure 5a). In this region the sorbed phase transport has little influence on DE (see Figure 2), while the diffusive transport in the soil still limits the rate of volatilization. The multimedia submodel with its arbitrary long diffusion depth performs poorly under these conditions. At yet lower values of KAW the agreement between the models is better, which can be explained by the fact that for chemicals with very low KAW values the resistance of the stagnant boundary layer again limits volatilization (33). The calculations were repeated including leaching (JW ) 0.005 m‚d-1; see Figure 5b). Once again, a volatilization of 10% was only achieved for chemicals with relatively high KAW values. In this chemical partitioning space the differences between the two models were very similar to the case with no leaching (Figure 5a). It is concluded that the multimedia submodel gives reasonable estimates of soil-air exchange for much of the chemical partitioning space in which sorbed phase transport influences the effective diffusivity in soil DE (see Figure 2), particularly for chemicals with log KOA > 8. A Comparison of Multimedia Behavior with and without Sorbed Phase Transport. To evaluate the effect of sorbed phase transport on the overall fate of POPs, the behavior of persistent chemicals with partitioning properties in the relevant range (5 < log KOA < 11, -3 < log KAW < 3) was calculated using the non-steady-state multimedia fate model CoZMo-POP (34). The forest environment was made negligibly small, and the soil parameters (porosity, water content, OC fraction, depth, MTC for atmospheric boundary layer) were taken from Table 1. Otherwise all the default values supplied by the model were employed. Generic emissions to air increasing from 1950 to 1975 and decreasing from 1975 to 2000 were assumed, as was a closed system, i.e., the

FIGURE 5. Comparison of the volatilization predicted by the modified Jury model and the modified multimedia soil submodel. The chemicals were assumed to be persistent. The time for 10% of the chemical to volatilize was calculated for DS ) 5.5‚10-7 m2‚d-1. The quotient of the result of the multimedia submodel to the result of the Jury model is plotted as a function of log KOA and log KAW for (a) no leaching (JW ) 0) and (b) leaching (JW ) 0.005 m‚d-1).

FIGURE 6. Comparison of the air and soil fugacities calculated by the CoZMo-POP model with (DS ) 5.5‚10-7 m2‚d-1) and without solid-phase diffusion (DS ) 0 m2‚d-1). Fugacity quotients are plotted as a function of log KOA and log KAW for the year of maximum emission (1975) and the last year (1999) of a 50 year generic emission scenario. For other model assumptions see text. boundary conditions require that what goes out comes in. The chemicals were further assumed to have a degradation half-life in all compartments except air of 10 years, a gasphase reaction rate with OH radicals of 3‚10-13 cm3‚molecules-1‚s-1, and energies of phase transfer between air and water and octanol and air of 60 and -80 kJ‚mol-1, respectively. These are typical values for POPs. Calculations were performed with sorbed phase diffusion (DS ) 2 cm2‚yr-1) and without sorbed phase diffusion (DS ) 0 cm2‚yr-1). CoZMoPOP does not include sorbed phase convection. The calculated fugacities in air and soil for the years 1975 (peak emission

year) and 1999 (last simulation year) were averaged, and their quotient plotted for the investigated chemical partitioning space in Figure 6. Overall the results show that a scenario without sorbed phase transport significantly underestimates the extent of air-soil exchange for chemicals of log KOA > 6. During the period of increasing emissions, the model without sorbed phase diffusion greatly underestimates the uptake of the chemicals into the soil, leading to air fugacities (and thus air concentrations) in 1975 that are up to a factor of 3 higher than those calculated by the model with sorbed phase VOL. 36, NO. 22, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4865

FIGURE 7. Time profile of the air and soil concentrations, air-soil fugacity fractions, and air-soil exchange fluxes calculated for a chemical with log KOA ) 8 and log KAW ) 0 using the CoZMo-POP model with (DS ) 5.5‚10-7 m2‚d-1) and without solid-phase diffusion (DS ) 0 m2‚d-1). diffusion. The dependence on partitioning properties is very similar to that found for the Jury model (Figure 3) with the strongest effect noted for chemicals with a log KOA between 7 and 11 and a log KAW > -3. The impact is much stronger on the levels in soil, which are up to 25 times higher when sorbed phase transport is included. At the end of the period of declining emissions, a model without solid-phase diffusion overestimates the atmospheric levels of substances with log KOA less than 8, while continuing to underestimate those for chemicals with a higher KOA. In the scenario with solid-phase diffusion, faster soil uptake during the contamination phase reduces air concentrations, whereas faster soil release during the decontamination phase slows down the decline in air concentrations. Figure 7 exemplifies the temporal trends for a substance at the center of the investigated partitioning space (log KOA ) 8, log KAW ) 0). In the scenario without solid phase transport, air concentrations rise and decline rapidly in response to the changes in emissions. A simulation in which the diffusive air-soil exchange was shut off completely by setting h ) 0 yielded exactly the same air concentrations over the entire 50 year time period, indicating that the process of gaseous air-soil exchange has no impact on the air concentrations calculated with a multimedia models without sorbed phase transport. The soil concentrations only increase slowly, mostly as a result of wet and particle-bound deposition processes, and at no time during the simulation period is an equilibrium with the atmosphere established. The fugacity fractions fA(/fA+fS) remain well above 0.7, indicating net deposition. The soil remains a sink for atmospheric chemical even after the air concentrations have dropped by an order of magnitude. Intensified air-soil exchange processes caused by the inclusion of solid phase transport lead to air concentrations that rise and fall less rapidly than in the scenario without solid phase transport, balancing out such that the same air concentrations are established at the end of the 50 year simulation period. The faster rate of air-soil exchange further leads to much higher soil concentrations. They approach equilibrium with the atmosphere during the 1980s, as indicated by fugacity fractions that fluctuate around 0.5 in response to seasonal temperature fluctuations. At that time the soil starts to become a source of the chemical to the atmosphere during summer and a sink during winter. Sorbed 4866

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 22, 2002

phase transport clearly has a significant impact on the multimedia fate of persistent lipophilic chemicals. What form it takes, namely whether it leads to higher or lower air and soil concentrations, is dependent on the particular emission situation. Instigated by the recognition of the limitations of the multimedia soil submodel described here, a second approach to modeling soil-air exchange has found widespread use in multimedia models in recent years (35-37). Instead of assuming a constant depth of the soil layer for all chemicals, the depth is calculated based on a modified Damko¨hler Number. In effect, depth is chosen such that the rate of chemical reaction-disappearance in the surface compartment is equal to the rate of movement into this compartment by diffusion and advection (32). The soil compartment depths determined in this way are smaller for some chemicals, i.e., hydrophobic substances with significant degradation rates. However, by not including sorbed phase diffusion the calculated soil depths for lipophilic compounds are far too low and do not reflect the soil depth distribution of these compounds observed in the environment. In the multimedia model the storage capacity of soil for the chemical is underpredicted, distorting fate predictions. Furthermore, the fundamental problem of the soil submodel for this kind of chemical, namely the low mass transfer coefficient for volatilization arising from the high soil-side resistance, is not addressed, since the diffusion path length in soil remains constant for all chemicals (38). It is clear that vertical sorbed phase transport not only affects contaminant transport in surface soil and gaseous soil-air exchange but also strongly influences their overall environmental fate, most particularly for persistent lipophilic compounds such as POPs. To be able to quantitatively describe the environmental fate of these chemicals, it will be necessary to assemble a good understanding of vertical sorbed phase transport as a function of depth in the myriad of surface soil environments encountered on the globe.

Acknowledgments Financial assistance of the CEFIC Long-range Research Initiative and the Scientific and Technical Cooperation Program of the German Federal Ministry of Education and Research is gratefully acknowledged.

Literature Cited (1) Jury, W. A.; Spencer, W. F.; Farmer, W. J. J. Environ. Qual. 1983, 12, 558-564. (2) Harrad, S. J.; Sewart, A.; Alcock, R. E.; Boumphrey, R.; Burnett, V.; Duarte-Davidson, R.; Halsall, C. J.; Sanders, G.; Waterhouse, K.; Wild, S. R.; Jones, K. C. Environ. Pollut. 1994, 85, 131-146. (3) Axelman, J.; Broman, D. Tellus 2001, 53B, 235-259. (4) Chiou, C. T.; Shoup, T. D. Environ. Sci. Technol. 1985, 19, 11961200. (5) Hippelein, M.; McLachlan, M. S. Environ. Sci. Technol. 2000, 34, 3521-3526. (6) Alcock, R. E.; Johnston, A. E.; McGrath, S. P.; Berrow, M. L.; Jones, K. C. Environ. Sci. Technol. 1993, 27, 1918-1923. (7) Jury, W. A.; Spencer, W. F.; Farmer, W. J. J. Environ. Qual. 1987, 16, 448. (8) Li, N.; Wania, F.; Lei, Y. D.; Daly, G. L. Submitted to J. Phys. Chem. Ref. Data, July 22, 2002. (9) Cousins, I. T.; Gevao, B.; Jones, K. C. Chemosphere 1999, 39, 2507-2518. (10) Rotard, W.; Christmann, W.; Knoth, W. Chemosphere 1994, 29, 2193-2200. (11) Krauss, M.; Wilcke, W.; Zech, W. Environ. Pollut. 2000, 110, 79-88. (12) Cousins, I. T.; Mackay, D.; Jones, K. C. Chemosphere 1999, 39, 2519-2534. (13) Graff, O.; Makeschin, F. Z. Pflanzenernaehr. Bodenk. 1979, 142, 476-491. (14) Evans, A. C. App. Biol. 1948, 35, 1-13. (15) Anderson, J. M. Agriculture, Ecosystems Environ. 1988, 24, 5-19. (16) Mu ¨ ller-Lemans, H.; van Dorp, F. J. Environ. Radioactivity 1996, 31, 7-20. (17) Farenhorst, A.; Topp, E.; Bowman, B. T.; Tomlin, A. D. Soil Biol. Biochemistry 2000, 32, 23-33. (18) Michaelson, G. J.; Ping, C. L.; Kimble, J. M. Arctic Alpine Res. 1996, 28, 414-424. (19) Litaor, M. I.; Barth, G.; Zika, E. M.; Litus, G.; Moffitt, J.; Daniels, H. J. Environ. Radioactivity 1998, 38, 17-46. (20) Wiechen, A.; Tait, D.; Haase, G. Kieler Milchwirtschaftliche Forschungsberichte 1993, 45, 101-108. (21) Thibodeaux, L. J. Chemodynamics; John Wiley: New York, 1996. (22) Thibodeaux, L. J.; Valsaraj, K. T.; Reible, D. D. Environ. Eng. Sci. 2001, 18, 215-223. (23) Reible, D.; Mohanty, S. Environ. Toxicol. Chem. 2002, 21, 875881.

(24) Elzein, A.; Balesdent, J. Soil Sci. Soc. Am. J. 1995, 59, 1328-1335. (25) Mackay, D. Multimedia Environmental Models, The Fugacity Approach; Lewis: Boca Raton, 2001. (26) Valsaraj, K. T.; Ravikrishna, R.; Choy, B.; Reible, D. D.; Thibodeaux, L. J.; Price, C. B.; Yost, S.; Brannon, J. M.; Myers, T. E. Environ. Sci. Technol. 1999, 33, 142-149. (27) Mackay, D.; Stiver, W. In Environmental Chemistry of Herbicides Vol. II; Grover, R., Cessna, A. J., Eds.; CRC Press: Boca Raton, 1991; pp 281-297. (28) Di Guardo, A.; Calamari, D.; Zanin, G.; Consalter, A.; Mackay, D. Chemosphere 1994, 28, 511-531. (29) Harner, T.; Mackay, D.; Jones, K. C. Environ. Sci. Technol. 1995, 29, 1200-1209. (30) Mackay, D.; Paterson, S.; Di Guardo, A.; Cowan, C. E. Environ. Toxicol. Chem. 1996, 15, 1627-1637. (31) Scheringer, M. Environ. Sci. Technol. 1996, 30, 1652-1659. (32) Cowan, C. E.; Mackay, D.; Feijtel, T. C. M.; van de Meent, D.; Di Guardo, A.; Davies, J.; Mackay, N. The Multi-media Fate Model: a Vital Tool for Predicting the Fate of Chemicals; SETAC Press: Pensacola, 1994; pp 47-51. (33) Jury, W. A.; Spencer, W. F.; Farmer, W. J. J. Environ. Qual. 1984, 13, 567-572. (34) Wania, F.; Persson, N. J.; Di Guardo, A.; McLachlan, M. S. 2000, WECC Report 2000/1. CoZMo-PoP. A fugacity-based multicompartmental mass balance model of the fate of persistent pollutants in the coastal zone. (35) Bennett, D. H.; McKone, T. E.; Matthies, M.; Kastenberg, W. E. Environ. Sci. Technol. 1998, 32, 4023-4030. (36) 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; National Institute for Public Health and the Environment (RIVM): Bilthoven, The Netherlands, 1996; RIVM report no. 719101029. (37) Beyer, A.; Matthies, M. Criteria for Atmospheric Long-range Transport Potential and Persistence of Pesticides and Industrial Chemicals; Umweltbundesamt: Berlin, Germany, 2001; Report No. 29965402. (38) Beyer, A. Personal communication, 2001.

Received for review March 21, 2002. Revised manuscript received September 3, 2002. Accepted September 10, 2002. ES025662Y

VOL. 36, NO. 22, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4867