Volatilization Modeling of Two Herbicides from Soil in a Wind Tunnel

Nov 6, 2012 - the permanent wilting point). This effect is caused by the hydrated mineral surfaces that become available as additional sorption sites ...
0 downloads 0 Views 474KB Size
Download PDF
Article pubs.acs.org/est

Volatilization Modeling of Two Herbicides from Soil in a Wind Tunnel Experiment under Varying Humidity Conditions Martina Schneider*,† and Kai-Uwe Goss*,†,‡ †

Helmholtz Centre for Environmental Research, Department of Analytical Environmental Chemistry, Permoserstrasse 15, 04318 Leipzig, Germany ‡ University of Halle Wittenberg, Institute of Chemistry, Kurt Mothes Strasse 2, 06120, Halle, Germany S Supporting Information *

ABSTRACT: Volatilization of pesticides from the bare soil surface is drastically reduced when the soil is under dry conditions (i.e., water content lower than the permanent wilting point). This effect is caused by the hydrated mineral surfaces that become available as additional sorption sites under dry conditions. However, established volatilization models do not explicitly consider the hydrated mineral surfaces as an independent sorption compartment and cannot correctly cover the moisture effect on volatilization. Here we integrated the existing mechanistic understanding of sorption of organic compounds to mineral surfaces and its dependence on the hydration status into a simple volatilization model. The resulting model was tested with reported experimental data for two herbicides from a wind tunnel experiment under various welldefined humidity conditions. The required equilibrium sorption coefficients of triallate and trifluralin to the mineral surfaces, Kmin/air, at 60% relative humidity were fitted to experimental data and extrapolated to other humidity conditions. The model captures the general trend of the volatilization in different humidity scenarios. The results reveal that it is essential to have high quality input data for Kmin/air, the available specific surface area (SSA), the penetration depth of the applied pesticide solution, and the humidity conditions in the soil. The model approach presented here in combination with an improved description of the humidity conditions under dry conditions can be integrated into existing volatilization models that already work well for humid conditions but still lack the mechanistically based description of the volatilization process under dry conditions.

1. INTRODUCTION The dramatic reduction of the volatilization of pesticides from soil surfaces under dry conditions has frequently been observed in field and wind tunnel experiments as described in the literature.1−4 Gish et al.5 observed a correlation of the volatilization of metolachlor with the soil surface temperature when the soil was wet, but not when the soil was dry which presented a contrast to the principles that had been developed so far. We have shown in a preceding paper6 that the “humidity effect” is caused by mineral surfaces that become available as additional sorption sites for organic compounds under dry conditions. So far the efforts to introduce this “humidity effect” into established volatilization models did not explicitly consider the hydrated mineral surfaces as an independent sorption compartment. Wolters et al.2 and Ferrari et al.7 introduced a fudge factor into the pesticide fate model PELMO in order to account for the increase of sorption under conditions drier than the permanent wilting point (PWP). This dimensionless factor, which increases the sorption coefficient, depends on the volumetric water content and a so-called maximum possible increase of soil sorption, when the soil is “air-dry”. For the investigated compounds from their study Wolters et al.2 arbitrarily set the possible maximum value for this factor to 100. This modeling approach did account for the humidity affect they had seen in their study and thus provides © 2012 American Chemical Society

an improvement compared to models that do not consider the humidity effect at all. However, this approach does not correctly describe the underlying processes and cannot be expected to work in other scenarios (see also Supporting Information). In general the sorption of organic compounds in soils is influenced by the soil water phase, the organic phase, and the hydrated mineral surfaces. Under humid conditions sorption of many pesticides is dominated by the organic phase in the soil (sorption to the water phase does not play an important role due to the low water solubility of most of the compounds). If the soil dries out (water content below the PWP), sorption capacity of the organic phase remains unaffected.8 The observed increase of sorption under dry conditions results from the hydrated mineral surfaces that become available as an independent sorption compartment. Thus under dry conditions one cannot, as done by Wolters et al.2 and Ferrari et al.,7 calculate the sorption coefficient from sorption under moist conditions because mineral sorption is a completely different mechanistic processes. The maximum increase of sorption under dry conditions depends on the number of sorption Received: Revised: Accepted: Published: 12527

July 30, 2012 September 28, 2012 October 2, 2012 November 6, 2012 dx.doi.org/10.1021/es303001q | Environ. Sci. Technol. 2012, 46, 12527−12533

Environmental Science & Technology

■

MATERIALS AND METHODS Basic Model Description. To model the volatilization experiments we used a simple transport model similar to that from Jury et al.13 The contaminated soil compartment is considered as vertically and horizontally homogeneous and well mixed. Transport through the adjacent laminar layer in the gas phase is considered to be the kinetically controlling step. The mass transport of the pesticide from the soil surface into the air is calculated as

sites at the hydrated mineral surfaces and the properties of the investigated compound. For soils low in organic carbon and with a large specific surface area this disagreement would be even larger. Sorption of Organic Compounds to Hydrated Mineral Surfaces. The sorption of organic compounds to the organic phase in the soil is independent of the humidity state of the soil.8 This is different for the mineral surfaces, which become available as sorption sites only under conditions dryer than the PWP. The contribution of the mineral surfaces to the overall sorption is given by the adsorption coefficient Kads [m3air/kgdrysoil]: K ads = K min /air ·SSA

Article

F=

(1)

where Kmin/air [m3air/m2min] is the partitioning coefficient (sorption coefficient) of the organic compound between the hydrated mineral surfaces and air, and SSA [m2/g] is the specific surface area available for adsorption. Both parameters directly depend on the hydration status of the mineral surfaces. In the context of dry soils, the hydration status of the mineral surfaces is best described by the water activity, expressed as equilibrium relative humidity (RH) in the pore space of the soil, and not by the water content.9 Below the PWP even small pores in the soil are not filled with water, although all mineral surfaces are covered with a water film of several molecular layers. Due to the high affinity of water molecules for hydrophilic surfaces these water layers prevent the sorption of organic molecules directly at the mineral surfaces but adsorption of organic molecules on top of this water layer takes place.9 The thickness of this water layer strongly influences the strength of the sorption of the organic compound to the hydrated mineral surfaces, expressed by the sorption coefficient Kmin/air. Above 90% RH the water film is so thick that almost no molecular interaction between the compound and the minerals through the water layer occur and the sorption can be considered as identical to the sorption on a neat water surface (Kmin/air = Kwater surface/air). With decreasing RH, the thickness of the water layer reduces and the distance between the molecules and the mineral surfaces decreases. As a result the interactions with the mineral surface become more and more important and the overall sorption coefficients Kmin/air increase exponentially. In addition to the sorption coefficient, Kmin/air, the available surface area of the hydrated mineral surface also depends on the relative humidity in the soil. Below about 95% RH the water films on the minerals do not form a meniscus and the surface area of the hydrated minerals equals the maximum specific surface area of the dry soil (SSA). In this way a soil containing a high amount of clay can provide a significant contribution to the overall sorption. For RH > 95% the water films on the mineral surfaces rapidly grow thicker, start to get connected, and the very small pores, which have been responsible for the high surface area of the water film, finally get filled with water. Thus the total surface area of the water surface decreases dramatically until it becomes negligible close to 100% RH.10,11 Goss et al.12 showed that the sorption to hydrated mineral surfaces is not substantially influenced by the mineral type and at a certain relative humidity Kmin/air mainly depends on the compound properties. A more detailed description of sorption of organic compounds in soils is given elsewhere.6,9 The goal of this study was to integrate the process understanding of the sorption of pesticides to mineral surfaces and the influence of the hydration status into a volatilization model. The resulting model was tested with experimental data from a previous study.6 Based on this model we discuss the influence of the sorption coefficient Kmin/air and the specific surface area of the hydrated mineral surfaces on the volatilization of pesticides.

Dair ·A (Csoil air − Cair flow ) x

(2)

here F is the volatilization rate, Dair is the diffusion coefficient of the compound in air, x is the thickness of the laminar boundary layer, A is the surface area of the soil, Cair flow is the concentration in the entering air flow, Csoil air is the concentration in the air within the contaminated soil. The soil air concentration Csoil air is calculated from the equilibrium partitioning of the compound between the different phases (water, air, organic matter, mineral surfaces) in the soil. We assume that the partitioning equilibrium between the different sorption compartments is reached immediately after application. The following equations consider the partition coefficients between air, water, organic phase, and hydrated mineral surfaces: Csoil air =

fair ·mtotal Vair

=

fair ·mtotal A ·d ·θair

(3a)

K OC/water ·fOC ·ρ θwater ρ 1 =1+ + + K min /air ·SSA· fair θair·K air/water K air/water ·θair θair (3b) Csoil air A mtotal fair Vair d θwater θair Kair/water

[kg/m3] [m2] [kg] [−] [m3] [m] [kgwater/kgdry soil] [m3air/m3bulk soil] [m3water/m3air]

ρ KOC/water

[kgsoil/m3bulk soil] [m3water/kgoc]

Kmin/air

[m3air/m2min]

SSA

[m2/kgsoil]

concentration of the pesticide in the soil air surface area of the soil in the experiment total mass of the chemical in the soil fraction of the compound in the soil air soil air volume within the penetration depth penetration depth of the pesticide solution water content of the soil air content of the soil air/water partitioning coefficient of the pesticide soil bulk density organic carbon/water partitioning coefficient of the pesticide sorption coefficient air/hydrated mineral surfaces of the pesticide surface area of the mineral surfaces available for sorption

The change in the total mass of chemical, mtotal, over time due to volatilization losses was solved numerically. The time steps in the model were adjusted according to the desired accuracy and ranged from 1 to 20 min. Note that additional degradation processes (e.g., photochemical or microbial) were not integrated into the model because the experiments were conducted in a dark hood and under very dry conditions. Model Parameter Determination. The top view surface area (2 × 15 cm2) was used as surfac area A in the calculation. Dair was calculated from the molecular volume and mass using the method of Fuller, Schettler, and Giddings.14 The concentration of the pesticide in the incoming air was set to zero. The thickness of the laminar boundary layer x was estimated by measuring the 12528

dx.doi.org/10.1021/es303001q | Environ. Sci. Technol. 2012, 46, 12527−12533

Environmental Science & Technology

Article

content if the water sorption isotherm (WSI) is known. The WSI displays the water content as a function of the equilibrium relative humidity in the soil. This relationship strongly depends on the clay content of the soil. For our modeling the WSIs for the respective soils were taken from Schneider and Goss.15 Before the application of the pesticide solution the soil humidity was in equilibrium with the constant air flow RH and as a result the water content was constant within the whole soil profile (60% RH equals 0.023g/g, 90% RH equals 0.045g/g for soil BL). Directly after application the equilibrium between the relative humidity in the air flow and the soil water content is disturbed by the water from the application solution which increases the water content within the upper soil layer. This amount of water is evaporated within a short time frame so that the soil goes back into its original equilibrium state. Evaporation of the water from the sprayed formulation is described in a simple way in the model (see Supporting Information). After its volatilization a constant water activity is assumed for the soil under constant RH conditions (humidity regimes I and II). Under humidity regime III the determination of the water activity in the soil is not as simple as under constant conditions. In Figure 1 we plotted the water content and the corresponding

volatilization of water from a moist soil surface (for a more detailed description see Supporting Information). The penetration depth of the applied pesticide solution into the soil has a direct influence on the concentration in the soil and therefore on the concentration gradient between the soil air and the atmosphere. In preliminary experiments of the previously presented work,6 we observed that our application procedure did cause a color change of the dry soil surface that reached to about 1 mm in depth with a rather sharp borderline. Experiments in which soils had been equilibrated with various relative humidities taught us that such color changes indicate a water activity above the PWP.15 We assume that this depth of 1 mm is identical with the penetration depth and that almost none of the applied pesticide solution penetrated any deeper into the soil because of the very small hydraulic conductivity that is observed for soil under dry conditions.16,17 Therefore we also neglected any further downward transport of the pesticides in modeling of the remainder of the experiment. The compound properties and specific soil parameters used for the calculation can be found in Table 1. Table 1. Model Parameters, Physico−Chemical Properties of Trifluralin and Triallate (Data Taken from Refs 3 and 18) and Properties of the Soil Used in This Study model parameters laminar boundary layer [cm] 0.22 gas volume flow [L/min] 17.1 0.003 penetration depth [cm] 0.1 soil surface area [m2] compound properties trifluralin triallate soil properties BL 6S water solubility [mg/L] 0.3 4 clay [%] 21 42 vapor pressure [mPa] 13.7 16 silt [%] 68 55 KOC [L/kg] 8000 2400 org. C [%] 2.1 1.6 Dair [cm2/s] 0.05 0.048 SSAsoil [m2/g] 18.3 37.6

Volatilization Data Set. The data to test the model were taken from Schneider et al.6 who measured volatilization with a bench-scale wind tunnel system. The system allowed the establishment of well controlled humidity conditions in the soil and the volatilization rate was measured with a high time resolution. The data revealed the influence of different humidity conditions on the volatilization of two pesticides (triallate and trifluralin) from the bare soil surfaces. In the beginning of each experiment the water activity in the soil was homogeneous and in equilibrium with the relative humidity in the air phase. The pesticides were applied in a formulation to a bare soil surface under the following humidity regimes: Regime I: soil in equilibrium with 60% RH. Regime II: soil in equilibrium with 90% RH. Regime III: soil in equilibrium with 60% RH at the start, increase of RH up to 85% after 2 h, simulation of a rain event after 6 h. A table with more information about the single experiments (application and rain amount, etc.), that is needed for the modeling, can be found in the Supporting Information. A detailed description of the complete experimental setup, sample analysis, and a discussion of the data can be found in Schneider et al.6 Note that for humidity regime III we present additional data points after 24 h that had not been included in Schneider et al.6 Water Content, RH, and SSA in Relation to Humidity Conditions. The water activity in the soil expressed as equilibrium relative humidity is a key parameter for the description of sorption to mineral surfaces. It can be derived from the water

Figure 1. Water content, relative humidity, and specific surface area of the hydrated mineral surfaces in the top soil layer (1 mm) in moisture regime III for soil BL. At points 1−5 the water content is well-known due to the experimental procedure (e.g., addition of a specified amount of water or equilibration with a specified humidity) and our knowledge of the WSI.

relative humidity against time to illustrate the different humidity states within the experiment. The time points for which the water content was known are marked in the figure (black squares). Between these points the water content had to be interpolated. We did not put much emphasis on optimizing this interpolation because the major features of the time curve of the moisture related parameters are already presented by the well-defined points 1−5 in Figure 1 (for more information about the interpolation and results for the second soil 6S see Supporting Information). We stress here that in field situations, where the boundary conditions are much less well-defined than in the wind tunnel experiments modeled here, it will be very challenging to get accurate data for the moisture-related input parameters. Figure 1 also shows the resulting specific surface area that is available for adsorption of pesticides. The specific surface area was calculated based on the following assumptions: Under dry conditions (RH < 95%) the hydrated mineral surface area 12529

dx.doi.org/10.1021/es303001q | Environ. Sci. Technol. 2012, 46, 12527−12533

Environmental Science & Technology

Article

quantitatively predicted by the model. Under both humidity conditions the elevated volatilization directly after application is reproduced by the model, but in the following (until 5 h) the model shows too quick a decrease in the volatilization rate. Possible reasons for that may be that the model overestimates the evaporation rate of the water from the initially applied pesticide solution and the assumption of immediate sorption equilibrium of the compound may also not hold. After 5 h at 60% RH the volatilization reaches a quasi steady state; the volatilized amount is small compared to the total amount in the soil so that the concentration gradient does not change substantially over the next 20 h. A similar explanation applies at 90% RH although here the higher volatilization rate results in a slight depletion of triallate which, in turn, leads to a small but observable decrease in the volatilization rates. The characteristic trends of the data at 60 and 90% RH are both adequately presented by the model predictions. The model results for volatilization of triallate under humidity regime III for both soils are presented in Figure 3a and b. The increase of volatilization after increasing the relative humidity is qualitatively well presented but for both soils and both humidification steps (increase of RH and simulated rain event) the maximum observed volatilization rate is not reached by the model. This is mainly due to the simplifying assumption of an instantaneous re-equilibration of the partition system after the externally applied humidity changes. The calculated equilibrium distribution of the total amount of compound between the different sorption compartments (hydrated mineral surfaces, organic carbon. and water) is presented in Figure S5 (Supporting Information). The model reveals that in the beginning of the experiment for soil BL under humidity condition III only 2% of the total amount of triallate is absorbed in the organic phase and up to 97% is adsorbed to the hydrated mineral surfaces. At 85% RH the amount absorbed in the organic phase increases up to about 20% and then after the rain event increases to 99%. Our model is based on the assumption that the compound experiences an instantaneous reallocation between the different sorption compartments according to the changes in Kmin/air and the available SSA as the humidity changes. In reality, compound molecules desorbed from the hydrated minerals will need to pass the air phase and diffuse into the humic matter in order to achieve the new equilibrium partitioning. Thereby in reality the concentration in the soil air will temporarily increase more than estimated from the equilibrium distribution and the overall volatilization rate increases above the maximum calculated value. Along with the rain event the pores in the soil get filled with water, the available surface area of the hydrated minerals decreases dramatically, and a maximum in the volatilization rate is observed in the experiments. This is correctly simulated by the model. The model also correctly accounts for the subsequent decrease of the volatilization rate. In the model this effect results from two processes: (a) depletion of the chemical reservoir in the soil and (b) drying of the soil surface due to a relative humidity in the air stream of 85% so that additional sorption sites at the hydrated mineral surfaces become available again. In reality the volatilization decrease could also be partly caused by the dislocation of the compound into the soil profile after the rain event, which was not considered in the model (see Supporting Information Figure S6). The model predictions for trifluralin and additional data for triallate are presented in the Supporting Information and show very similar results. Comparing the results for the different soils, the model shows a higher volatilization rate under dry conditions for soil BL, which

available for sorption of organic compounds, SSA, equals the specific surface area of the oven-dry soil, SSAsoil: RH < 95%: SSA = SSA soil

(4a)

Under moist conditions (100% RH) the available surface area of the hydrated mineral surface SSA is zero. Between 95% and 100% RH, we chose a linear interpolation between the maximum value of SSAsoil and zero:19 95% > RH < 100%: −

SSA soil SSA soil ·RH + ·100% 100% − 95% 100% − 95% (4b)

Estimation of the Sorption Coefficient Kmin/air. The sorption coefficient to the mineral surfaces, Kmin/air, as a function of relative humidity is a key parameter within the prediction of the volatilization of pesticides under dry conditions in the soil. Goss9 presented a pp-LFER model to predict Kmin/air. The pesticides studied here do not fall into the application domain of this model though, because they contain several functional groups that may not all lie in the sorption plane of the compound.9 However, from this earlier work an empirical correlation can be derived that allows for prediction of the slope of the exponential relationship of the sorption coefficient with RH for any compound20 (see Figure S1 in Supporting Information). The predicted slope of the exponential RH relationship combined with a single value for Kmin/air at any relevant RH provides the input information needed for modeling. To obtain this Kmin/air value we optimize the volatilization model to the quasi steadystate situation at 60% RH in the experiments from Schneider et al. using the solver method in Excel. We stress that besides this one value no other fitting to the volatilization data was used in the model results presented below. The possibility of integrating temperature dependence into the model and the estimation of ΔHads of the sorption processes to the different phases is presented in the Supporting Information.

■

RESULTS AND DISCUSSION Volatilization under Different Humidity Conditions. Figure 2 presents the experimental data and the model results for volatilization of triallate under constant humidity conditions from the soil BL. Over the whole time period the generally higher volatilization under 90% RH compared to 60% RH is

Figure 2. Measured and predicted triallate volatilization rates after spray application to soil BL under constant humidity conditions at 60 and 90% RH. 12530

dx.doi.org/10.1021/es303001q | Environ. Sci. Technol. 2012, 46, 12527−12533

Environmental Science & Technology

Article

Figure 3. Measured and predicted triallate volatilization rates after spray application to soil BL (a) and 6S (b) under humidity conditions III (60 → 85% RH, rain event).

Figure 4. Sensitivity study for the volatilization under humidity regime III for the parameters organic carbon, application rate, SSA, and the penetration depth.

reflects its lower specific surface area compared to soil 6S. The underestimation of the volatilization directly after the rain is even larger for soil 6S. This is consistent with the fact that the model does not consider increased losses to the gas phase during the reallocation of the compound by establishing a new equilibrium conditions (under 85% RH only 10% of the total amount of triallate is estimated to be absorbed in the organic carbon of soil 6S, so a larger amount of the compound needs to be reallocated, see Figure S5 in the Supporting Information). Sensitivity Study. Above we have shown that the model correctly predicts the influence of humidity on volatilization of

pesticides in different settings. In an additional sensitivity study we used the model to show the influences of various input parameters organic carbon content f OC, application rate (respectively mtotal), SSAsoil, and the penetration depth of the applied formulation on the volatilization under humidity regime III (Figure 4). For that purpose we calculated the volatilization rate of triallate from soil BL under humidity regime III while increasing and decreasing each parameter by 30% and keeping the rest of the parameters unchanged. Note that SSA highly correlates with the clay content of the soil.21 Therefore the clay content that influences the shape of the WSI was changed accordingly. 12531

dx.doi.org/10.1021/es303001q | Environ. Sci. Technol. 2012, 46, 12527−12533

Environmental Science & Technology

Article

The penetration depth and the application rate proportionally influence the volatilization rate over the whole humidity range. As expected the influence of SSAsoil and the organic carbon is only visible under specific humidity conditions. According to theory the influence of SSAsoil only shows under dry conditions. The specific surface area of the soil directly influences the maximum number of sorption sites on the hydrated mineral surfaces that are available under dry conditions and dominate the volatilization. A change in the organic carbon content affects the volatilization over the whole humidity range. But it becomes visible only under humid conditions, where sorption to organic carbon dominates the overall sorption process. Model Application. What does all this mean for the agricultural practice? At first glance one might expect that an application to a dry soil surface might be favorable because elevated adsorption reduces the overall volatilization. However, under these circumstances the pesticide can not penetrate deeper into the soil. The next rain event would then result in an increase of the volatilization and a loss of the compound that otherwise would have happened right away if the pesticide had been applied to a moist soil (model prediction see Figure S9). An application to a dry soil will likely only postpone the volatilization loss of the pesticide unless it is mechanically incorporated before the next rain event. This comparison is based on the assumption that the penetration depth of the applied formulation is the same for spraying on a dry or a wet soil. However, the application solution should penetrate deeper into a moist soil than into a dry soil due to the much higher hydraulic conductivity of the former. For a hypothetical experimental scenario (triallate applied to soil BL, water content 0.13 g/g, 100% RH) and an assumed penetration depth that is 3 times deeper for a wet soil than for the dry soil our model predicts a volatilization rate that is similar to the one that we observed and predicted for application to dry soil. Thus there is a chance that application to a wet soil might even be favorable in terms of the minimization of volatilization losses, because the deeper penetration might compensate for the lower sorption and further losses during the next rain event. A specific study of the effect of soil moisture on the penetration depth of applied pesticides would be desirable. Also interesting is the effect of temperature in combination with the humidity effect. On air-dry soils it has been observed that a temperature increase does not result in increased volatilization.4,5 This may seem counterintuitive because a temperature increase always shifts sorption equilibrium toward the gas phase. It has to be considered though that in a diurnal cycle the absolute water content in the atmosphere typically stays constant so that a temperature increase goes along with a decrease in relative humidity. Thus two opposing effects have to be considered: an increase in sorption due to a decrease in relative humidity and a decrease in sorption due to an increase in temperature. Here we have used our model to evaluate quantitatively which of both effects prevails (Figure 5). The calculations were done for triallate on soil BL with a temperature curve taken from Gish et al.5 We assumed that the absolute amount of water in the systems stayed constant and that the sorption equilibrium of water in the soil followed the RH curve instantaneously. The results demonstrate that the drying effect dominates over the direct temperature effect on sorption so that the volatilization rate is highest when the temperature is lowest and vice versa. In contrast, volatilization under completely humid conditions (100% RH) that we have calculated as a reference scenario shows an increase in volatilization with increasing temperature. These calculations are in excellent agreement with field data which

Figure 5. Diurnal temperature of the soil surface and the respective relative humidity under the assumption of a constant absolute humidity. The volatilization rates calculated with the model for two scenarios are shown: volatilization from a dry soil where relative humidity is a direct function of temperature, and volatilization from a wet soil where RH is constantly 100%.

show the same trends for wet and dry soils and thus provide a convincing explanation for an otherwise unexpected observation. In general we conclude that the presented model was able to correctly predict volatilization rates of two pesticides on two soils under varying humidity conditions by considering mineral surfaces as an independent sorption compartment. The model approach presented here in combination with an improved description of the WSI under dry conditions can be integrated into existing volatilization models that already work well for humid conditions but still lack the mechanistically based description of the volatilization process under dry conditions. In addition to the appropriate mechanistic approach it is also essential to have good quality input data in order to obtain good model results. The simple estimation of the humidity dependence of SSA and Kmin/air that we proposed here should work in other cases as well. However, there still is the need for a reliable predictive method of Kmin/air for a reference humidity. The rather accurate information on the humidity state of the soil that we could extract here from the well-defined wind tunnel experiments might be very difficult to get in a field situation. Only recently are tools beginning to be developed that allow the prediction of the water activity and the hydraulic conductivity in soils under dry conditions.16,17

■

ASSOCIATED CONTENT

* Supporting Information S

More information for the experimental methods and the water sorption isotherm of the soils used in this study. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected], [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was kindly supported by Helmholtz Impulse and Networking Fund through Helmholtz Interdisciplinary Graduate School for Environmental Research (HIGRADE). Special thanks to Satoshi Endo and Trevor Brown for a careful review of the manuscript. 12532

dx.doi.org/10.1021/es303001q | Environ. Sci. Technol. 2012, 46, 12527−12533

Environmental Science & Technology

■

Article

REFERENCES

(1) Majewski, M.; Desjardins, R.; Rochette, P.; Pattey, E.; Seiber, J.; Glotfelty, D. Field Comparison of an Eddy Accumulation and an Aerodynamic-Gradient System for Measuring Pesticide Volatilization Fluxes. Environ. Sci. Technol. 1993, 27, 121−128. (2) Wolters, A.; Klein, M.; Vereecken, H. An improved description of pesticide volatilization: Refinement of the Pesticide Leaching Model (PELMO). J. Environ. Qual. 2004, 33, 1629−1637. (3) Bedos, C.; Rousseau-Djabri, M. F.; Gabrielle, B.; Flura, D.; Durand, B.; Barriuso, E.; Cellier, P. Measurement of trifluralin volatilization in the field: Relation to soil residue and effect of soil incorporation. Environ. Pollut. 2006, 144, 958−966. (4) Reichman, R.; Rolston, D. E.; Yates, S. R.; Skaggs, T. H. Diurnal Variation of Diazinon Volatilization: Soil Moisture Effects. Environ. Sci. Technol. 2011, 45, 2144−2149. (5) Gish, T. J.; Prueger, J. H.; Kustas, W. P.; Daughtry, C. S. T.; McKee, L. G.; Russ, A.; Hatfield, J. L. Soil Moisture and Metolachlor Volatilization Observations over Three Years. J. Environ. Qual. 2009, 38, 1785−1795. (6) Schneider, M.; Endo, S.; Goss, K.-U. Volatilization of pesticides from the bare soil surface: Evaluation of the humidity effect. J. Environ. Qual. 2012, under review. (7) Ferrari, F.; Klein, M.; Capri, E.; Trevisan, M. Prediction of pesticide volatilization with PELMO 3.31. Chemosphere 2005, 60, 705−713. (8) Niederer, C.; Goss, K.-U.; Schwarzenbach, R. P. Sorption equilibrium of a wide spectrum of organic vapors in Leonardite humic acid: Experimental setup and experimental data. Environ. Sci. Technol. 2006, 40, 5368−5373. (9) Goss, K. The air/surface adsorption equilibrium of organic compounds under ambient conditions. Crit. Rev. Environ. Sci. Technol. 2004, 34, 339−389. (10) Costanza, M.; Brusseau, M. Contaminant vapor adsorption at the gas-water interface in soils. Environ. Sci. Technol. 2000, 34, 1−11. (11) Costanza-Robinson, M.; Brusseau, M. Air-water interfacial areas in unsaturated soils: Evaluation of interfacial domains. Water Resour. Res. 2002, 38. (12) Goss, K.; Buschmann, J.; Schwarzenbach, R. Adsorption of organic vapors to air-dry soils: Model predictions and experimental validation. Environ. Sci. Technol. 2004, 38, 3667−3673. (13) Jury, W. A.; Spencer, W. F.; Farmer, W. J. Behavior Assessment Model for trace organics in soil. 0.1. Model Description. J. Environ. Qual. 1983, 12, 558−564. (14) Lyman, W. J.; Rosenblatt, D. H.; Reehl, W. F. Handbook of Chemical Property Estimation Methods; American Chemical Society: Washington, DC, 1990; p 17-9 (15) Schneider, M.; Goss, K.-U. Prediction of the water sorption isotherm in air dry soils. Geoderma 2012, 170, 64−69. (16) Lebeau, M.; Konrad, J.-M. A new capillary and thin film flow model for predicting the hydraulic conductivity of unsaturated porous media. Water Resour. Res. 2010, 46. (17) Zhang, Z. F. Soil Water Retention and Relative Permeability for Conditions from Oven-Dry to Full Saturation. Vadose Zone J. 2011, 10, 1299−1308. (18) Tabernero, M.; Alvarez-Benedi, J.; Atienza, J.; Herguedas, A. Influence of temperature on the volatilization of triallate and terbutryn from two soils. Pest Manage. Sci. 2000, 56, 175−180. (19) Garcia, L. Volatilisation de l’ammoniac et des pesticides au champ Analyse et modélisation de l’influence des propriétés et des conditions physiques de la surface à l’aide du modèle Volt’Air. Ph.D. thesis, Pierre and Marie Curie University, Paris, France, 2010. (20) Goss, K.; Buschmann, J.; Schwarzenbach, R. Determination of the surface sorption properties of talc, different salts, and clay minerals at various relative humidities using adsorption data of a diverse set of organic vapors. Environ. Toxicol. Chem. 2003, 22, 2667−2672. (21) Sokolowska, Z.; Borowko, M.; Reszko-Zygmunt, J.; Sokolowski, S. Adsorption of nitrogen and water vapor by alluvial soils. Geoderma 2002, 107, 33−54.

12533

dx.doi.org/10.1021/es303001q | Environ. Sci. Technol. 2012, 46, 12527−12533