Assessing the Fate of an Aromatic Hydrocarbon Fluid in Agricultural

Jul 31, 2015 - ... Applications Using the Three-Stage ADVOCATE Model Framework ... U.S. Highway 22 East Annandale, New Jersey 08801, United States...
0 downloads 0 Views 1MB Size
Article pubs.acs.org/JAFC

Assessing the Fate of an Aromatic Hydrocarbon Fluid in Agricultural Spray Applications Using the Three-Stage ADVOCATE Model Framework Liisa Toose,† Christopher Warren,# Donald Mackay,§ Thomas Parkerton,*,Δ Daniel Letinski,⊥ ̂ Ryan Manning,⊥ Martin Connelly,⊥ Arlean Rohde,⊗ Brad Fritz,Π and W. Clint HoffmannΠ †

LTEC, Wroxeter, Ontario, Canada ExxonMobil Research Qatar, Qatar Science & Technology Park, Doha, Qatar § Canadian Centre for Environmental Modelling and Chemistry, Trent University, Peterborough, Ontario, Canada Δ ExxonMobil Biomedical Sciences, Inc., 22777 Springwood Village Parkway, Spring, Texas 77339, United States ⊥ ExxonMobil Biomedical Sciences, Inc., 1545 U.S. Highway 22 East Annandale, New Jersey 08801, United States ⊗̂ CONCAWE, Boulevard du Souverain 165, B-1160 Brussels, Belgium Π Agricultural Research Service, U.S. Department of Agriculture, 2771 F&B Road, College Station, Texas 77845, United States #

S Supporting Information *

ABSTRACT: Components of emulsifiable concentrates (ECs) used in pesticide formulations may be emitted to air following application in agricultural use and contribute to ozone formation. A key consideration is the fraction of the ECs that is volatilized. This study is designed to provide a mechanistic model framework for estimating emissions of an aromatic hydrocarbon fluid used in ECs based on the results of spray chamber experiments that simulate fate as the fluids become subject to volatilization, sorption to soil, and biodegradation. The results indicate the need to treat the volatilization losses in three stages: (i) losses during spraying, (ii) losses up to 12 h after spraying in which the soil is coated with the ECs, and (iii) subsequent longer term losses in which the ECs become increasingly sorbed and subject to biodegradation. A mass balance model, the agrochemical derived volatile organic compound air transfer evaluation (ADVOCATE) tool, is developed, treating the ECs as seven hydrocarbon component groups, to estimate the volatilization and biodegradation losses using parameters fitted to empirical data. This enables losses to be estimated for each hydrocarbon component under field conditions, thereby providing a basis for improved estimation of ozone formation potential and for designing ECs that have lower emissions. KEYWORDS: pesticide formulation, ozone, emulsifiable concentrate, volatilization, biodegradation



INTRODUCTION Pesticide formulations are preparations of active and inert ingredients that allow simpler, safer, and more effective use of the resulting product for controlling target pests or plant diseases as compared to the active product alone.1 Inert ingredients do not act on the pest or disease but rather contribute to the performance of the formulation. The most frequently used products are liquid formulations that are mixed with water and then applied as sprays. The most common spray products are emulsifiable concentrates (ECs). An EC is composed of an immiscible solvent that is stabilized by emulsifying agents. The solvent serves to dissolve the active ingredient and provide a carrier for the other formulation components. Due to the favorable solubility behavior of many active ingredients in aromatic hydrocarbon fluids, the stability of the resulting emulsions, and the favorable cost, these substances are commonly selected as the solvent of choice in commercial ECs.2 Although the benefit of using pesticide formulations to reduce crop losses is obvious, the potential for adverse environmental impacts must be assessed and appropriately mitigated. Past efforts have often focused on understanding the © 2015 American Chemical Society

environmental fate and effects of active ingredients to reduce risks to nontarget organisms. However, given that inert ingredients can be present in pesticide formulations at higher concentrations, the potential environmental concerns that these ingredients pose warrants study. In the case of solvents used in ECs, a key concern is the emission of volatile organic compounds (VOCs) that may contribute to ozone formation after spraying.3 In Kumar et al.,3 VOCs in pesticide products were measured in prespray and postspray air and soil samples in four orchards. Elevated concentrations of tri- and tetraalkyl benzenes were found in both media after spraying but returned to prespray background levels in 1−2 days. Ozone measurements performed downwind of spray applications indicated a detectable increase of 10−15 ppb in concentration in two of the orchards investigated that was consistent with predictions of ozone formation obtained from a photochemical model. Received: Revised: Accepted: Published: 6866

February 27, 2015 June 6, 2015 July 7, 2015 July 31, 2015 DOI: 10.1021/acs.jafc.5b01076 J. Agric. Food Chem. 2015, 63, 6866−6875

Article

Journal of Agricultural and Food Chemistry Table 1. Calculated Vapor Pressures and Amounts of AF Components in the Tested EC Formulations

a

component

VP (Pa) at 25 °C

VP (Pa) at 20 °C

EC1 concentration (g in 4 L of solution)

EC2 concentration (g in 4 L of solution)

naphthalene C2 indanes C5/C6 benzene C1 naphthalenes C2 naphthalenes C3 naphthalenes C2−C4 biphenyls hexachlorobenzenea

11 23.85 24.05 9.08 3.9 0.645 0.471 0.000616

6.76 16.68 18.01 6.62 2.19 0.416 0.337 0.000424

8.94 2.88 5.36 41.31 22.74 8.34 9.73 1.08

17.87 5.76 10.72 82.62 45.48 16.68 19.46 2.18

Not present in commercial AFs but intentionally spiked to serve as a tracer in this study.

partitioning to soil and water decreases the mobility of substances with vapor pressures below approximately 10 Pa, thus lowering the fraction of the emitted mass available to react in the atmosphere and serve as ozone precursors.13 Nevertheless, the results of TGA or application of the calculation method yields a conservative EF estimate. The IR term reflects the differential reactivity of formulation components in producing ozone under set assumptions. Although this parameter depends on the reactivity scale selected, the intent is to provide a consistent basis for comparing relative ozone formation potentials of different products on a mass basis. The AMAF term has been included to account for the fact that not all VOCs associated with the product will necessarily be emitted to air. Application of multimedia fate models suggests other fate processes such as sorption and biodegradation may reduce VOC losses following spray application. However, because quantitative data are generally lacking, use of adjustment factors is often not considered. Thus, in the case of aromatic fluids with VP ≥ 0.05 Pa, a worst-case scenario is applied in which all of the mass that is applied in pesticide formulation is assumed to be VOCs (EF = 1) that are emitted to air and available to contribute to ozone formation (AMAF = 1). Further research is therefore needed to assess the validity of this assumption and incorporate refinements if justified. The objective of this study is to provide a model framework to estimate AMAF in spray application of aromatic fluid (AF) used in ECs. This objective is accomplished by (1) conducting spray chamber studies that simulate the fate of an EC containing aromatic fluid under field conditions and (2) using the resulting data to calibrate a predictive fate model, the agrochemical derived volatile organic compound air transfer evaluation (ADVOCATE) tool, to estimate VOC emissions that incorporates salient fate processes (volatilization, soil partitioning, and biodegradation) of constituents that comprise this fluid.

In 2012, 192 counties in the United States were designated as not attaining the 2008 ozone standard of 75 ppb over an 8 h averaging period.4 Proposed revisions would further reduce this standard.5 Given the Clean Air Act mandate to achieve attainment with the ozone standard, states must develop plans to meet this air quality objective. A key component of plans to reduce ground-level ozone often involves limiting VOC emissions from various sources, although oxides of nitrogen are precursors that can limit ozone formation. Understanding the nature and magnitude of VOC emissions is of particular concern in agricultural areas of California.6 In such regions, VOCs are evolved naturally from plants, from the use of farm equipment, and from the use of pesticides. To address this challenge, California has developed a framework to assess and manage the potential ozone formation associated with pesticide formulations.7 The ozone formation potential (OFP) of a pesticide formulation allows the reactivity weighted emissions to be estimated in terms of ozone equivalents. The OFP is calculated as OFP = M × EF × AMAF × IR

(1)

where OFP = ozone formation potential of pesticide formulation (kg ozone/day), M = mass of product applied (kg product applied/day), EF = mass emission fraction of product (kg VOC in product/kg product applied), AMAF = application method adjustment factor (kg VOC emitted/kg VOC in product), and IR = incremental reactivity (kg ozone formed/kg VOC emitted). The EF term is based on a thermogravimetric analysis (TGA) laboratory test, which involves heating a sample from 35 to 115 °C at a rate of 5 °C/min. The sample is then held at a final temperature of 115 °C to facilitate volatilization and loss of water in the formulation until the rate of sample mass loss drops below a defined threshold. The mean of three replicate loss measurements is used to estimate a product’s EF. When TGA data are not available, the EF can be estimated given the weight composition and estimated vapor pressure (VP) of ingredients having VP at 25 °C of ≥0.05 Pa.8 However, this vapor pressure cutoff is considerably lower than the value of 10 Pa that is commonly applied to define VOCs internationally9 as adopted by IUPAC.10 A key disadvantage of the TGA method is that the temperatures that the sample is exposed to are not representative of any naturally occurring environment; also, no chemical emission information is collected, and recent studies have shown that TGA can overestimate emissions quantified using GC-MS analysis.11 A more direct approach is to use evaporation data measured under standard test conditions or to estimate this rate from a simple linear correlation with vapor pressure.12 Furthermore, fate modeling has shown that



MATERIALS AND METHODS

Test Materials. Two ECs without an active ingredient were prepared from the following substances. A sample of AF (CAS Registry No. 64742-94-5) was obtained from ExxonMobil Chemical, Houston, TX, USA. This substance contains aromatic hydrocarbons predominantly with 10−13 carbons. TOXIMUL 3473, which is a blend of calcium sulfonate and nonionic surfactant, and TOXIMUL 3474, which is surfactant mixed with aromatic solvent, was obtained from the Stepan Co., Northfield, IL, USA. Hexachlorobenezene (CAS Registry No. 118-74-1) with a purity of 99% was purchased from Sigma-Aldrich and spiked into test EC formulations to serve as a conserved, involatile tracer. A fluorometric tracer dye was also included to determine spray recoveries. EC1 was prepared by mixing the above ingredients in water to yield the following composition on a 6867

DOI: 10.1021/acs.jafc.5b01076 J. Agric. Food Chem. 2015, 63, 6866−6875

Article

Journal of Agricultural and Food Chemistry volumetric basis; AF (2.5%); TOXIMUL 3473 (0.095%); TOXIMUL 3474 (0.12%); HCB (0.025%); and tracer dye (0.0125%). EC2 was prepared using twice these volumetric concentrations for all components. The two key substance properties determining the dissipation of ECs are vapor pressure (VP) and biodegradation rates in soil. The vapor pressures of the EC constituents were characterized using EPISuite MPBPWin v. 1.43 module14 and were then temperature corrected using the Clausius−Claperyon equation at 20 °C and study temperatures (Appendix S1). Predicted biodegradation half-lives were calculated using EPISuite BioHCWin v. 1.01a15 and compared to empirical estimates generated in this study. Table 1 shows the pertinent physical−chemical properties and amounts of EC constituents in the two spray formulations. Spray Chamber Studies. The fate study was performed in collaboration with the U.S. Department of Agriculture, Agricultural Research Service (USDA-ARS), in College Station, TX. Researchers at the USDA-ARS have expertise in aerial application of pesticide formulations and have developed a spray chamber used to mimic agricultural spray applications. The methodology is described in detail16 and is intended to simulate EC spraying under scenarios that are typical of ground application conditions. An agricultural sandyloam soil with an organic matter content of 1.2% collected from the Central Valley of California near Fresno, CA, USA, was used in these studies. Soil was placed in six aluminum containment boxes (0.61 L × 0.91 W × 0.15 H m3) with holes drilled in the bottom to allow for drainage. The filled boxes were placed in a greenhouse and watered over a 3 week period to return soil to conditions found in agricultural fields. Soil boxes were then placed on a table in a chamber that was equipped with automated spray nozzles that traverse the entire length of the chamber 1.4 m above the soil surface at a speed of 2.7 m/s. Air temperature at the time of spray application ranged from 16 to 19 °C. Two different spray nozzles were employed to deliver a high volume (188 L/ha of EC1) spray with larger droplets (volume median diameter of 227 μm) representative of residual soil applications and a lower volume spray (94 L/ha of EC2) that gave smaller droplets (volume median diameter of 166 μm) representative of foliar and insecticidal treatment applications. Both nozzles were operated at 30 psi and provided average nozzle flow rates for larger and smaller droplet spray of 25 and 14.3 mL/s, respectively. Good spray recoveries ranging from 93 to 99% were reported on deposition samplers that were previously reported as part of this study.16 Six soil containment boxes were sprayed with each of the two selected nozzles (three boxes for each nozzle). Four of the boxes were sprayed with each nozzle (1, 2) in duplicate (A, B) for a short-term study to quantify volatilization losses of AF after spraying. The remaining two boxes were used to provide a single replicate for each spray nozzle to investigate longer term losses of AF due to soil biodegradation (designated boxes C1 and C2). Soil samples were obtained from each location using a sharpened 1.6 cm diameter plug cutter inserted through the soil surface to a depth of 2.5 cm. The soil sample was then placed into a labeled 40 mL glass vial with an airtight screw-cap lid. The plug cutter was cleaned by rinsing with hexane between each sampling point. Four replicate soil samples were randomly collected from 20 gridded box locations at 0, 1, 4, 12, and 36 h for the volatilization study. Samples were frozen after collection to prevent biodegradation and then shipped for analysis. For the biodegradation study, samples were collected and frozen as above to serve as the day 0 time point. For the remaining samples, soil was collected immediately after spraying, placed in capped vials and shipped to the laboratory at ambient temperature to allow biodegradation to proceed in sealed vials. Samples were then analyzed after 0, 2, 4, 7, and 14 days of incubation taking into account elapsed transit time. Five replicates were included at time 0 and three replicates for subsequent sampling times. Control soil samples that were not sprayed were also collected and sent for analysis to determine background concentrations of AF components and HCB. All t = 0 samples were obtained while the containment boxes were on the spray table. Following initial soil sampling, the boxes were then moved to an adjacent greenhouse for subsequent sampling. Air temperature in the

greenhouse during each soil sampling period ranged from 21 to 29 °C.16 Hydrocarbon Analysis. Two separate aliquots, equivalent in concentration, of emulsifiable concentrate (EC) amended with 1% HCB were received from USDA-ARS by the analytical chemistry laboratory of ExxonMobil Biomedical Sciences, Inc., Annandale, NJ, USA. Equal volumes of these EC aliquots were combined and diluted in acetone, yielding a single stock analytical standard solution. This stock solution was subsequently diluted in 1:1 methylene chloride/ acetone providing a series of analytical standards used for subsequent GC-MS analysis of treated soil extracts. The concentrations of the total AF in the standards were 89, 356, 890, and 3560 ng/mL. The corresponding HCB concentrations in the standards were 1, 4, 10, and 40 ng/mL. Each analytical standard also contained a constant 209 ng/ mL concentration of deuterated (d8) dichlorobenzene as an internal standard. Treated soil samples were extracted and analyzed by GC-MS. Briefly, 5 g samples of soil were dried with an approximately equal amount of Hydromatrix (Varian) and extracted on a Dionex Accelerated Solvent Extractor 200 at 1500 psi and 100 °C with a solvent 1:1 mixture of methylene chloride and acetone. The extracts were adjusted to a final volume of 20 mL of which a 1.0 mL aliquot was spiked with the equivalent of 209 ng/mL d8 dichlorobenzene internal standard prior to analysis. GC-MS analysis was performed in the selective ion monitoring mode (SIM) using an Agilent HP6890N GC and 5975 MSD with Gerstel MPS2 autosampler. A 30 m × 0.25 mm id Rtx-5Sil MS capillary column (Restek) with 0.25 μm film thickness preceded by an uncoated 10 m Integra guard column was used. Thirty microliter injections were made in the solvent vent mode into a Gerstel CIS 4 inlet. The inlet liner was packed with 10 mm of PDMS to trap the more volatile hydrocarbon components during the solvent vent process of the large-volume injections. The AF consists of one- and two-ring aromatic compounds with various methyl or other alkyl substitutions. Using constituent analysis provided by ExxonMobil Chemical Co., structures were assigned to component groups (e.g., methyl and dimethyl (C5/C6) benzenes, dimethylindanes (C2), naphthalene, methyl (C1) naphthalenes, dimethyl and ethyl (C2) naphthalenes, C3 naphthalenes, C2−C4 biphenyl) comprising the AF. The concentration in the analytical standard for each group accounted for the fact that the AF comprised only 90% of the EC with the balance comprising other formulation components, including surfactants and HCB. Specific retention time and SIM windows were established for each of the individual groups plus the HCB conserved marker and deuterated dichlorobenzene internal standard. This technique permitted internal standard normalized calibration curves for each component group and subsequent quantification in soil extracts. Data Analyses. The volatilization loss of each hydrocarbon component during spraying was estimated from the ratio of the measured concentrations of the AF components relative to HCB in the spray formulation and in the soil samples assuming that the relatively involatile and stable HCB (Table 1) is conserved. Direct measurement of the quantity of hydrocarbon sprayed and measured in soil samples is less accurate because of spray heterogeneity and possible spray drift. However, preliminary spray trials using the spray chamber and nozzle applied in this study indicated measured loss by spray drift was small (only 1−7%)16 and would not likley affect the hydrocarbon/HCB ratio. The predicted concentration of chemical in the soil at t = 0 and the fraction of the component volatilized Xi% can be estimated from eqs 2 and 3. The “predicted” concentrations assume the same ratio of hydrocarbon/HCB reaches the soil as is in the sprayed solution. The “observed” ratios are lower as a result of volatilization of the hydrocarbon components.

6868

Csoil pred,i = Csoil obs, HCB × CECi /CECHCB

(2)

Xi = (Csoil pred,i − Csoil obs,i)/Csoil pred,i × 100%

(3)

DOI: 10.1021/acs.jafc.5b01076 J. Agric. Food Chem. 2015, 63, 6866−6875

Article

Journal of Agricultural and Food Chemistry

Table 2. Average Measured Cumulative Mass Percent (Xi) ± the Standard Deviation of Each Hydrocarbon Component Lost during the Volatilization Studya component

VP (Pa) at 20 °C

C5/C6 benzenes C2 indanes naphthalene C1 naphthalenes C2 naphthalenes C3 naphthalenes C2−C4 biphenyls

18.01 16.68 6.76 6.62 2.19 0.42 0.34

AF (total)

0 h, n = 26 62.9 57.4 46.1 38.3 16.2 −2.5 −10.9

± ± ± ± ± ± ±

1 h, n = 16

16.4 13.2 20.5 14.1 12.4 9.6 8

79.3 74.3 66.2 57.1 26.5 −0.8 −11.4

28.3 ± 12

± ± ± ± ± ± ±

6.6 8.2 9.1 11 15.5 15.6 13.5

41.6 ± 11.7

4 h, n = 16 89.5 89 83.9 77.7 47.1 7.4 −13.2

± ± ± ± ± ± ±

5.6 5.3 6.8 9.7 20.5 21.4 17.2

57.6 ± 12.9

12 h, n = 16 92.7 92.8 86.8 83.7 59.8 20 −2.8

± ± ± ± ± ± ±

2.6 2.6 3.9 5.1 11.8 15.8 15

65.6 ± 7.8

36 h, n = 16 95.6 95.1 89 87.8 70.4 25.4 −5.6

± ± ± ± ± ± ±

2.6 1.9 3.9 4.3 11.9 20.1 18.7

70.4 ± 8.4

a

n is the number of analytical measurements taken from the sample boxes and used in analysis; for calculations at time zero, all six boxes were used; for subsequent times, analyses for boxes A1, A2, B1, and B2 were used.

Figure 1. HCB-normalized soil concentrations of hydrocarbon components in the short-term volatilization study, showing the initial fast loss during the first 12 h followed by a slower rate of loss. Csoil pred,i is the predicted soil concentration (μg/g dry wt) of component i, where i can be the selected hydrocarbon component or the sum of all AF components, C soil obs,HCB is the measured HCB soil concentration (μg/g dry wt), and C EC,i/ C EC, HCB is the concentration ratio of component i to HCB in the EC formulation, which is known on the basis of analytical characterization of the EC formulation (unitless). Biodegradation rates were estimated by measuring concentration ratios in samples from box C maintained in closed vials to prevent volatilization, again assuming HCB is conserved. The use of HCB as a nondegradable benchmark chemical that facilitates derivation of biodegradation half-lives has been reported in previous studies.17 The rate constant kbio was estimated by fitting the observed concentration ratios of component(s) to HCB over time using eq 4: ln[(Csoil obs,i /Csoil obs,HCB) = − k bio × t

quantity of AF sprayed is sufficient to saturate only a fraction of a millimeter depth of soil. Thus, subsaturation conditions must exist in the 2.5 cm deep soil samples, indicating that component concentrations are not homogeneously distributed in the soil compartment. Volatilization Study. HCB-normalized concentrations for total AF hydrocarbons as well as component groups were used to calculate cumulative percent losses from 0 to 36 h for each spray nozzle using eq 3 (Table S1). Given limited data and lack of clear differences in calculated losses between the two nozzle types (Tables S2−S7), average values were determined as summarized in Table 2 and plotted in Figure 1. For t = 0, all six boxes were used; for subsequent times only the four boxes (A and B) comprising the volatilization study were used. At time 0, immediately after spraying, a significant fraction of the AF is lost into the air by volatilization (Table 2). Most of this loss is attributable to the more volatile, higher vapor pressure components. Following this loss there appears to be a further initial fast drop in concentration within the first 12 h followed by a slower reduction to 36 h (Figure 1). This behavior suggests a change in loss mechanism over time and indicates that a simple first-order volatilization model cannot be used to adequately simulate this empirical loss profile. This observation casts doubt on the practice of assigning a single half-life to pesticides and carrier fluids that are applied to soil by spraying. Figure 1 also shows that the HCB-normalized concentrations of C3 naphthalenes and C2−C4 biphenyls

(4)

Corresponding half-lives were calculated as τ1/2bio = 0.693/kbio.



RESULTS HCB Concentrations in Soil. The initial concentrations of HCB in soil were found to be close to the expected concentration in soil based on mass balance calculations assuming no losses. Furthermore, whereas absolute concentrations were variable over time, no significant decline in measured HCB soil concentrations was observed over the study period (Figure S1). These data support the assumption that HCB served as a conserved tracer in this study. On the basis of mass balance calculations described in Appendix S2, the 6869

DOI: 10.1021/acs.jafc.5b01076 J. Agric. Food Chem. 2015, 63, 6866−6875

Article

Journal of Agricultural and Food Chemistry appear to increase at t = 1 h, then level out, showing very little loss over the entire 36 h of the volatilization study. The apparent increase in concentration results in negative calculated losses and is likely an artifact of the small changes observed (±1−14% as per Table 2) and the large variation within those observations in comparison to the lighter hydrocarbons. Biodegradation Study. As the microbial community becomes acclimated to the presence of the AF hydrocarbons in soil, biodegradation processes are expected to contribute to observed losses. Figure 2 shows the decline in total AF

reported in this study are up to an order of magnitude lower (i.e., faster rates) than assumed in earlier modeling studies predicting the multimedia fate of hydrocarbons.18,19 However, past laboratory studies have reported shorter, measured biodegradation half-lives for naphthalene and methyl naphthalene in agricultural soil of 1−2 days,20,21 which is in agreement with values observed in this study. A modest agreement between empirical and predicted half-lives is observed. Differences of about a factor of 4 indicate HCBioWin predictions are within the correct order of magnitude but less precise in discriminating between hydrocarbon structures. Model Formulation. It is hypothesized that volatilization of AF components during and after spraying is best treated in three distinct stages: stage 1, during the spray application (ending at t = 0 h); stage 2, volatilization of EC coating the surface of the soil (t = 0−12 h); and stage 3, combined transport into soil vapor from chemical sorbed to soil solids with subsequent volatilization and biodegradation (t = 12−36 h and beyond). Stage 1: Volatilization during Spraying. During the spray event, a fraction of the AF evaporates or drifts to a downwind location. This fraction is likely dependent on the vapor pressure of the component and the dynamics of the spray process. The relationship of the percent loss of chemical with its vapor pressure can be determined and used to predict the percent mass lost to the air during such spray events. We assume a spray droplet to contain V mol of hydrocarbon, each individual component having an initial mole fraction concentration x0. As the droplet falls some of the hydrocarbon evaporates and the mole fraction x falls according to

Figure 2. Regression of HCB-normalized soil concentrations of total hydrocarbons with time for the two spray nozzles in replicate boxes during biodegradation study.

hydrocarbons normalized to HCB over the 14 day (330 h) test period. The source of the increased variability in normalized values at the end of the test in nozzle 2 box C is not known. However, exclusion of these data in the analysis has little influence on the regression analysis results. The average half-life for total AF hydrocarbons derived from results with both spray nozzles was 180 h (7.5 days), corresponding to a mean biodegradation rate constant (kbio) of 0.00385 h−1. For model calibration, the biodegradation rate of each component is required. Analogous plots and observed regressions for each of the AF components are included in Figure S2. Data for C5/C6 benzenes were highly variable, and estimated biodegradation rates are less certain. As the estimated biodegradation rates derived from tests with the two nozzles were similar across the different constituents as indicated by the individual regression equations included in Figure S2, values were averaged for input to model simulations as summarized in Table 3. Measured half-lives varied by a factor of 5 across the different components. The soil biodegradation half-lives

dx /dt = − k V × A × CA mol/h

where kV is a mass transfer coefficient, A is the surface area of the droplet, and CA is the equilibrium concentration in the air intimately in contact with the drop. CA can be estimated as xPM/RT, where P is the component’s vapor pressure (Pa) and R, T, and M are the gas constant (8.314 m3·Pa/mol·K), temperature (K), and molar mass (g/mol), respectively. This assumes that Raoult’s law applies and that the partial pressure of each component is proportional to its mole fraction in a fluid of constant volume and composition; thus, the prediction may not be highly accurate, but the structure of the correlation equation is believed to be satisfactory. A rigorous equation would represent the batch or Rayleigh volatilization of a multicomponent mixture of changing volume and would be excessively complex for the present purposes. Integration from x = x0 at t = 0 gives x = x0·exp[− Y ·P]

component

measured t1/2 (h)

predicted t1/2 (h)

ratio

C5/C6 benzenes C2 indanes naphthalene C1 naphthalenes C2 naphthalenes C3 naphthalenes C2−C4 biphenyls

0.0020 0.0040 0.0077 0.0085 0.0044 0.0014 0.0023

355 173 90 82 159 495 308

128 48 133 213 341 233 718

2.77 3.62 0.67 0.38 0.47 2.12 0.43

(6)

where

Table 3. Estimated Biodegradation Rates from Tests with Both Spray Nozzles and Corresponding Measured and Predicted Half-Lives Using HCBioWin measured kbio (h−1)

(5)

Y = k V ·A ·M ·t /(V ·R·T )

Because of the complexity of the mass transfer process, it is impossible to estimate the parameter Y a priori, but it can be estimated from empirical data quantifying losses during spraying. Furthermore, the process is more complex than is assumed here because V changes with time and there is a spectrum of droplet sizes. Thus, the parameter Y consolidates this information on droplet size, dynamics, and spray time into a single fitted parameter. It is preferable to express the percentage of the original mass remaining in the soil at each stage as Mit/Mi0. 6870

DOI: 10.1021/acs.jafc.5b01076 J. Agric. Food Chem. 2015, 63, 6866−6875

Article

Journal of Agricultural and Food Chemistry

vapor from the fluid surface and is best expressed as being first order in hydrocarbon concentration and will be proportional to the vapor pressure. It is assumed that there is no biodegradation during this stage. During the 12 h following spraying the concentration of AF fell from ∼6 to ∼3.2 μg/g, that is, a loss of 47%, suggesting an overall half-life of approximately 13 h. Figure 4 shows that

The percentage loss (L1) of a specific hydrocarbon during spraying is L1 = 100[1 − Mi1/Mi0] = 100[1 − exp( −YP)]

(7)

Specifically Mi0 is the original mass of component i. Mi1 is the mass deposited by the spray; thus, the spray loss is Mi0 − Mi1. Similarly Mi2 is the mass after volatilization during stage 2, and Mi3 is the mass after volatilization and biodegradation during stage 3. It follows that Mi1 = Mi0 exp( − YP) = (1 − L /100) or YP = − ln(Mi1/Mi0)

(8)

In this study, the HCB normalized data are used to reduce the noise associated with spatial variation in the spray pattern. The most reliable data are those of the five most volatile components for which Y is shown to average 0.066 Pa−1 (Figure 3). There is no significant difference in losses between

Figure 4. Regression of −ln(Mi2/Mi1) against the product of vapor pressure at 20 °C and 12 h for the two nozzles tested in duplicate spray boxes during stage 2.

dependence of the total rate of loss is stronger for the less volatile components during stage 2; components with vapor pressures of >10 Pa are likely to volatilize quickly during the first stage, thus showing less dectectable loss by stage 2. Assuming that volatilization rates are first order, the volatilization loss during stage 2 (L2) for the less volatile components (VP < 10 Pa at 25 C) can be expressed as a function of vapor pressure P (Pa), time period of the stage, t (h), and a volatilization rate parameter that is common to all components, kv2 (h−1·Pa−1), as follows:

Figure 3. Regression of −ln(Mi1/Mi0) against vapor pressure at 20 °C for the two spray nozzles in replicate boxes during spray loss for the five most volatile EC components. Differences are within the standard errors. The highest and lowest rates of loss were observed for A2 and C2.

L 2 = 100[1 − Mi2 /Mi1] = 100[1 − exp( −k v2 × P × t )] (9)

the nozzles during stage 1. Box A2 shows the largest value of Y (0.0873 Pa−1), and C1 shows the smallest value (0.0346 Pa−1). The box C results from the degradation study for both nozzles are less consistent with boxes A and B from the volatilization study than they are within that study, but are included in the t = 0 spray loss analysis to increase the degrees of freedom in the analysis from n − 1 = 15 to n − 1 = 25. Because AF is a mixture, it is impossible to assign a single vapor pressure; however, assuming Y to be 0.066 Pa−1, a mean effective vapor pressure of the AF used in this study during spraying is 4.96 Pa. The model provides a simple method for predicting percent loss following spraying, as a function of vapor pressure during the stage 1 spray event. Volatilization during Stage 2. When the sprayed fluid reaches the soil surface, it coats the soil particles and may not homogeneously distribute over the surface area depending on the nature of the spraying process. Volatilization then occurs directly from the deposited fluid to the atmosphere. In their studies of pesticide volatilization, Woodrow and Seiber22 and Woodrow et al.23 have suggested that this period may last up to 24 h, but it is observed here to be shorter, for example, 8−12 h for some of the more volatile hydrocarbons present in the EC applied. During stage 2 when free or pure phase hydrocarbon is present at the surface, the volatilization rate will likely be controlled by diffusive mass transport of the near-saturated

kv2 × P is the rate constant (h−1)) for volatilization. This volatilization rate constant for stage 2 was deduced as for stage 1 as 0.019 and can be used to predict the volatilization of chemical as a function of vapor pressures up to 10 Pa (commonly used to define substances as VOCs) and for times up to 12 h postspray, assuming there is limited sorption of the AF to the soil. At a VP of 10 Pa the rate of change with vapor pressure levels off. Components with VP > 10 Pa at 25 °C exhibit a 36−70% loss during stage 1 and have much lower concentrations prior to stage 2. During stage 2 (between 0 and 12 h postspray), these more volatile components experience a 69−86% loss with an average of 79%, independent of vapor pressure (Figure 4). Because the y-intercept is constant at 1.56, the predicted percent stage 2 loss for hydrocarbons with VP > 10 Pa simplifies to 100[1 − exp(−1.56)]. The y-intercept of 1.56 can be prorated for a time to estimate losses between 0 and 12 h. For example, at 1 h, the y-intercept is 1.56 × 1/12 = 0.13, which corresponds to a stage 2 loss of 100[1 − exp(−0.13)] or 12%. Thus, 100 g of an AF component with a vapor pressure 10 Pa (6.8 Pa at 20 °C) will experience a 40 g loss during stage 1 and a further 47 g loss during stage 2, making a total cumulative loss of 87 g. One hundred grams of a hydrocarbon with a vapor pressure of 0.05 Pa (0.034 Pa at 20 °C) will experience a loss of 6871

DOI: 10.1021/acs.jafc.5b01076 J. Agric. Food Chem. 2015, 63, 6866−6875

Article

Journal of Agricultural and Food Chemistry

h−1 and the half-time is 0.693/(kv3P) h. This is an order of magnitude slower rate than applies to stage 2. The total loss for stage 3, L3, is then calculated as

0.25 g during stage 1 and a further 0.77 g loss during stage 2, thus making the total cumulative loss about 1 g by 12 h. Volatilization and Biodegradation in Stage 3. The volatilization study results provide an estimate of the rate of volatilization during stage 3 from approximately 12 to 36 h. During stage 3, the fluid increasingly sorbs into the soil particles and diffuses into the soil matrix, especially into the soil organic matter. Volatilization then occurs as a result of transport of soil vapor in equilibrium with the fluid present primarily in a sorbed state through the pore air space and a boundary layer to the atmosphere. The partial pressure or fugacity of the fluid component in stage 2 likely corresponds to that of the sprayed fluid, but in stage 3 it drops to a lower value corresponding to that of the sorbed state. In addition, as the hydrocarbon penetrates more deeply into the soil matrix the rate of mass transfer slows because of the increased path length for diffusion. The net result is relatively fast stage 2 volatilization, followed by a slower rate in stage 3 that is also assumed to be first-order in chemical concentration. During these periods the rates are controlled by the different air−fluid equilibrium relationships of the components and kinetic or mass transport terms. Estimating Volatilization and Biodegradation in Stage 3. During stage 3, an attempt was made to regress the total loss (which includes both volatilization and biodegradation) against the product of vapor pressure and the 24 h duration (12−36 h) to obtain a total loss rate constant. However, no discernible pattern of loss during stage 3 was observed for spray nozzle 2. In fact, data show an increase in concentration for many components (evidenced by points below the x-axis), so results cannot be reliably modeled (Figure 5). In contrast, a decrease in concentration was observed for all

L3 = 100[1 − Mi3/Mi2] = 100[1 − exp( − (k bio + k v3 × P)t )] (10)

where t is the length of time (24 h) between sampling at the beginning and end of stage 3 (12−36 h). It is likely that this rate constant for volatilization, kv3, decreases with time as the AF components becomes more deeply sorbed into the soil matrix. The relative contribution of volatilization to the total loss of the AF during stage 3 can be determined. The ratio of rates of biodegradation to volatilization is kbio/kv3P; thus, the total quantities of hydrocarbon components after 12 h, Mi2, that are volatilized and biodegraded can be estimated as respectively Mi2 × (kv3P/(kv3P + kbio) and Mi2 × (kbio/(kv3P + kbio). The losses of low-volatility chemicals are dominated by biodegradation, and the ratio kbio/kv3P is >1. This ratio is quite small for the two most volatile components; thus, volatilization dominates the overall loss processes. Table 4 shows that biodegradation is Table 4. Relative Importance of Biodegradation and Volatilization to Overall Loss of Components during Stage 3 component

biodegradation kbio (h−1)

volatilization kv3P (h−1)

ratio kbio/kv3P

C5/C6 benzenes C2 indanes naphthalene C1 naphthalenes C2 naphthalenes C3 naphthalenes C2−C4 biphenyls

0.0020 0.0040 0.0077 0.0085 0.0044 0.0014 0.0023

0.0418 0.0388 0.0157 0.0154 0.0051 0.0010 0.0008

0.05 0.10 0.49 0.55 0.86 1.40 2.88

more significant for the more methylated naphthalenes and the predominant loss process for the biphenyls. Clearly biodegradation occurs and it is likely responsible for long-term removal of the less volatile hydrocarbons, but it is inconsequential for the more volatile components because for these components the rate of biodegradation cannot effectively compete with volatilization. To a first approximation, the biodegradation half-lives of the components are similar in magnitude, but the volatilization rates differ by almost 2 orders of magnitude depending on component vapor pressure. Combined Model of All Three Stages. The regressions described above have been combined in a spreadsheet model that estimates the fate of the hydrocarbon components and the AF for the sequence of stages 1, 2, and 3. It can be used, with appropriate caution, to provide a conservative estimate of the cumulative quantity of hydrocarbon lost and remaining for longer times on the order of weeks using eqs 7, 9, and 10 as shown in eqs 11 and 12. For VP < 10 Pa at 25 °C

Figure 5. Regression of −ln(Mi3/Mi2) against vapor pressure at 20 °C for both nozzles in replicate spray boxes during stage 3. A1, kv = 0.0022 and R2 = −0.239; B1, kv = 0.0024 and R2 = 0.2602; A2, kv = 0.0002 and R2 = 0.1696; B2, kv = −0.0001 and R2 = −1.481.

components with nozzle 1. Further observed losses showed a positive relationship with vapor pressure. These findings suggest AF sprayed at a lower spray rate and smaller droplet size is better retained by the soil after 36 h; that is, there is less volatilization. To provide a conservative model parametrization of stage 3 losses an estimate of k3total was derived using only nozzle 1 data. The biodegradation rate constant was then subtracted from this value to give the rate constant for volatilization (that depends on vapor pressure). The stage 3 volatilization rate parameter kv3 common to all components in the EC for spray nozzle 1 is then estimated to be 0.0023 Pa−1 h−1 (Figure 5). The rate constant for volatilization is then kv3P

LT = 100 × [1 − exp( − 0.066 × P) × exp( − 0.019P × 12) × exp( − (k bio + 0.0023 × P) × t )]

(11)

For P ≥ 10 Pa at 25 °C LT = 100 × [1 − exp( − 0.066) × exp( − 1.56) × exp( − (k bio + 0.0023 × P) × t ))] 6872

(12)

DOI: 10.1021/acs.jafc.5b01076 J. Agric. Food Chem. 2015, 63, 6866−6875

Article

Journal of Agricultural and Food Chemistry

Table 5. Estimated Cumulative Percent Dissipated (Volatilized and Degraded) Using the Three-Stage Model for t = 0−14 Days (336 h)a stage 1

stage 2

stage 3

component

0h

1h

4h

12 h

C5/C6 benzenes naphthalene C2 indanes C1 naphthalenes C2 naphthalenes C3 naphthalenes C2−C4 biphenyls total AF

69.61 66.82 36.03 35.44 13.47 2.72 2.21 27.21

73.33 70.88 43.85 43.08 17.00 3.48 2.83 32.34

81.96 80.30 62.02 60.98 26.74 5.75 4.69 44.77

93.64 93.05 86.61 85.75 47.49 11.54 9.46 63.99

36 h 97.78 97.51 92.34 91.61 58.12 16.43 15.74 70.76

(0.18) (0.42) (1.87) (1.78) (4.9) (2.9) (4.64) (2.75)

120 h 99.94 99.93 98.92 98.69 81.03 31.53 34.49 82.84

(0.28) (0.64) (4.02) (3.93) (15.45) (11.84) (18.48) (8.38)

168 h

336 h

99.99 (0.28) 99.99 (0.65) 99.65 (4.25) 99.55 (4.2) 87.93 (18.63) 38.9 (16.21) 43.27 (24.96) 86.32 (10.24)

100 (0.28) 100 (0.65) 99.99 (4.37) 99.99 (4.33) 97.52 (23.05) 58.99 (28.1) 65.71 (41.53) 92.62 (13.94)

a

Cumulative percent degraded in parentheses for stage 3. Bold values indicate cumulative percent degraded is greater than the cumulative percent volatilized.

where P is VP in Pa at 20 C, t is time in hours during stage 3 (i.e., time from spraying minus 12 h), and kbio is the first-order substance-specific soil biodegrdation rate constant in h−1 (Table 3). The model input is the AF composition treated as seven groups or “pseudo-components”. Other hydrocarbon groups can be included as desired. For each group a temperaturedependent vapor pressure is assigned and used to estimate rate constants for all stages. The biodegradation half-life in Table 3 is assigned and used in stage 3. The model calculates masses remaining after stages 1 (spray loss), 2 (12 h), and 3 (36 h) as described earlier. It also estimates the masses remaining in soil for up to 2 weeks. The masses volatilized and biodegraded are also calculated. The altered AF compositions are calculated at all stages. The spreadsheet is “open”; thus, any desired times and hydrocarbon properties can be input and simulated and can be accessed in Appendix S3 as an MSExcel spreadsheet. Table 5 gives the mass balance estimated by the model for the test conditions and comparison with the experimental data (given below in parentheses). For the AF, 27% is predicted to be lost in stage 1 (28%). After stage 2, a loss of 64% is predicted (66%). After 36 h, the predicted loss is 71% (70%). After 1 week, the predicted percentage lost is 86%, and at 2 weeks the loss is >92% as biodegradation removes the residual hydrocarbons. It is shown that for C2−4 biphenyls, biodegradation becomes the dominant loss process by 120 h (5 days). Figure 6 shows the measured versus the modeled predicted soil concentrations for each of the AF components for 0−36 h.

The C2−C4 biphenyls and C3 naphthalenes experience the least loss (lowest spread) in the mid-concentration range. The lighter C1 and C2 naphthalenes experience relatively fast loss, mainly through volatilization. The model appears to overpredict soil concentrations in the beginning of stage 2 by approximately 25%, but then brings the predictions closer to measured values by stage 3 (Appendix S4). The R2 obtained by regressing the predicted against the observed losses for all components and sampling times of 0.92 is regarded as an indication of satisfactory model performance.



DISCUSSION On the basis of the model framework presented, it is instructive to consider how different conditions will influence the potential for AFs included in ECs to be emitted to air and potentially contribute to ozone formation. Application Rate Per Unit Area. In this study the EC application rate was 4.7 kg/ha or 0.47 g/m2. The model estimates the fraction of the mass of hydrocarbon sprayed that volatilizes, not the actual volatilization rate, although this can be calculated from the results. To a first approximation, an increase in application rate by, for example, a factor of 2 should not affect the volatilized fraction in stage 1, and it should have little effect on the fractions in stage 2, although the duration of the stage 2 period may be prolonged as more hydrocarbon must volatilize from a thicker coating on the soil before the onset of stage 3. The spray rates (L/Ha) were varied in this study, but the EC application rates (kg/Ha) were the same (the higher EC concentration solution was sprayed at a lower rate). Variations in Organic Matter (OM) Content of the Soil. The OM content of the tested soil was 1.2%. No experimental determinations of the effect of OM content were performed; however, there is little doubt that increasing OM content may reduce volatilization rates. Doubling the OM content to 2.4% may not have an effect on stage 2, but the increased sorptive capacity of the soil is likely to reduce the volatilization rate in stage 3 and subsequently at longer times, probably by a factor of approximately 2. The model can thus directionally simulate the effect of soil OM content. Other Hydrocarbon Components. The volatilization characteristics of other hydrocarbon components or groups can be estimated by input of the appropriate vapor pressure and biodegradation half-lives obtained experimentally or from estimation methods. Relative Losses of Hydrocarbons. The model tends to overpredict the volatilization of the more volatile hydrocarbons. This can be accounted for in part by reducing the volatilization

Figure 6. Comparison of measured and model-predicted soil concentrations for all sampling times during volatilization study (0, 1, 4, 12, 36 h). R2 = 0.92. 6873

DOI: 10.1021/acs.jafc.5b01076 J. Agric. Food Chem. 2015, 63, 6866−6875

Article

Journal of Agricultural and Food Chemistry

Table 6. Estimated AMAF (Cumulative Percent Volatilized) for Hypothetical Chemicals Using the Three-Stage Model for t = 0−14 Days (336 h) stage 1

stage2

stage 3

chemical VP (Pa)

0h

1h

4h

12 h

36 h

120 h

168 h

336 h

30 20 10 5 1 0.5 0.1 0.05 0.01

74.31 59.58 36.43 20.27 4.43 2.24 0.45 0.23 0.05

77.45 64.52 44.19 25.30 5.67 2.87 0.58 0.29 0.06

84.74 76.00 62.25 38.56 9.28 4.75 0.97 0.49 0.10

94.62 91.54 86.69 63.52 18.26 9.59 2.00 1.00 0.20

98.05 95.75 90.60 69.38 21.09 11.17 2.34 1.18 0.24

99.31 98.38 95.18 79.11 27.45 14.86 3.17 1.60 0.32

99.32 98.47 95.70 81.11 29.49 16.11 3.47 1.75 0.35

99.32 98.49 95.96 82.86 32.56 18.14 3.98 2.01 0.41

rate constant as was done between stages 2 and 3. It appears that the most volatile hydrocarbons fail to achieve the expected losses at longer exposures. It is believed that this is due to some of the volatile hydrocarbons being “trapped” in the bulk of the EC mixture and being less available for volatilization. Essentially there appears to be a diffusive resistance to mass transfer when concentrations are low. The model inherently assumes wellmixed conditions in the EC mixture. In reality, there is likely extensive volatilization of water and AF components, thereby increasing the density and viscosity of the mixture. There may be some phase separation of the solid hydrocarbons as the “solvent effects” of the more volatile components are dissipated. To accommodate such effects would greatly increase the complexity of the model; thus, the simple and conservative assumptions used here are regarded as justified. Implications for Estimating the AMAF. Table 5 shows that for hydrocarbons with vapor pressures of 0.337−18 Pa at 20 °C, 38.9−99.99% of the applied substance is predicted to be volatilized by 7 days postapplication. Several hypothetical components with 0.01−30 Pa vapor pressures (at 25 °C, adjusted to 20 °C for modeling predicted losses) and a degradation half-life of 100 h were also simulated to estimate the cumulative percent volatilized over time (Table 6). This model provides a practical method to quantify the application method adjustment factor (AMAF, kg VOC volatilized/kg VOC in product) term in the calculation of the ozone formation potential (OFP) included in eq 1. This work shows that the regulatory assumption that AMAF is 1 for all hydrocarbons with a vapor pressure >0.05 Pa is overly conservative as only about 2% is predicted to have volatilized by 2 weeks postspray for a hypothetical chemical of a vapor pressure of 0.05 Pa and a degradation half-life of 100 h. Factors such as the degradation rate constant and the time post application as well as the vapor pressure are influential in determining the magnitude of the AMAF. For example, 7 days after application, only 18% of C2−4 biphenyl (VP = 0.34 Pa) losses are attributed to volatilization and 25% to degradation; after 14 days, only 24% is lost to volatilization, and the majority of the loss, 41%, to biodegradation. The model was used for determining cutoff values reflecting 67% losses by volatilization after 12, 24, 96, and 168 h, based on a typical biodegradation half-life of 100 h. The resulting cutoff values range between 2 and 3.5 Pa, nearly 2 orders of magnitude larger than the proposed cutoff (Appendix S5). Summary and Future Work. After spraying, three stages of AF loss in an EC formulation can be characterized: initial loss during the spray event, which is related to the component vapor pressure; then loss from the pure phase of fluid coating

the soil directly after the spray event; and, third, the volatilization and biodegradation of the remaining fluid that has been absorbed by the soil beyond 12 h postapplication. The mechanisms of volatilization and the corresponding models developed here give an adequate quantitative description of the fate processes of the seven components comprising the AF investigated during the complex series of events that occur following spraying. The model is simple, robust, and consistent with the experimental results and is intuitively in accord with expectations based on conventional descriptions of the salient equilibrium and kinetic processes; uncertainty analysis could further refine and add confidence to the framework. This model has practical utility in refining current default regulatory assumptions for estimating the ozone formation potential of ECs used in agricultural sprays. A logical focus of future modeling work would be incorporation of photodegradation process at the soil surface and one or more vegetation compartments to account for substance bioaccumulation and degradation by plants as these processes are known to influence fate behavior of substances released to soil.24−27 The present study also provides an initial model framework that can potentially be extended to other applications such as evaluating substances used in dust suppression control or petroleum substance spills to soils.



ASSOCIATED CONTENT

S Supporting Information *

Supplementary appendices, tables, and figures. The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jafc.5b01076.



AUTHOR INFORMATION

Corresponding Author

*(T.P.) E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors would like to thank Pierre-Yves Guyomar, ExxonMobil Chemical Europe, for a critical review and helpful comments for improving an early draft of this paper.



REFERENCES

(1) Knowles, D. Chemistry and Technology of Agricultural Formulations; Kluwer Academic: London, UK, 1998. (2) Krenek, M.; Rohde, W. Pesticide formulations and application systems. In An Overview − Solvents for Agricultural Chemicals; Hovde,

6874

DOI: 10.1021/acs.jafc.5b01076 J. Agric. Food Chem. 2015, 63, 6866−6875

Article

Journal of Agricultural and Food Chemistry D., Beestman, G., Eds.; American Society for Testing and Materials: Philadelphia, PA, USA, 1989; Vol. 8, pp 113−127. (3) Kumar, A.; Howard, C.; Derrick, D.; Malkina, I.; Mitloehner, F.; Kleeman, M.; Alaimo, C.; Flocchini, R.; Green, P. Determination of volatile organic compound emissions and ozone formation from spraying solvent-based pesticides. J. Environ. Qual. . 2001, 40, 1423− 1431. (4) U.S. EPA. Nonattainment Designations for the 2008 Ozone Standards − Counties by State. U.S. Evironmental Protection Agency − Area Designations for 2008 Ground-Level Ozone Standards, 2012. (5) U.S. EPA. Regulatory Actions. U.S. Environmental Protection Agency − Ground-Level Ozone, 2011. (6) CARB. Area Designations for the Federal 8-h Ozone Standard, California Environmental Protection Agency Air Resources Board, 2013. (7) CPDR. Pilot Project Proposal: Estimating Pesticide Product Volatile Organic Compound Emission Speciation and Reactivity Based on Product, California Department of Pesticide Regulations, 2011. (8) CPDR. Estimating the Pesticide Product volatile Organic Compounds Ozone Reactivity, Part 2 Reactivity Based Emissions, California Department of Pesticide Regulations, 2011. (9) CEC. Proposal for a Council Directive on limitation of emissions of volatile organic compounds due to the use of organic solvents in certain industrial activities, article 2, 96/0276 (SYN); Commission of the European Communities, 1996. (10) Heinrich-Ramm, R.; Jakubowski, M.; Heinzow, B.; Christensen, J.; Olsen, E.; Hertel, O. Biological monitoring for exposure to volatile organic compounds (VOCs). Pure Appl. Chem. 2000, 72, 385−436. (11) Zeinali, M.; McConnell, L. L.; Hapeman, C. J.; Nguyen, A.; Schmidt, W. F.; Howard, C. J. Volatile organic compounds in pesticide formulations: methods to estimate ozone formation potential. Atmos. Environ. 2011, 45, 2404−2412. (12) Mackay, D.; van Wesenbeeck, I. Correlation of chemical evaporation rate with vapor pressure. Environ. Sci. Technol. 2014, 48, 10259−10263. (13) Foster, K.; Sharpe, S.; Webster, E.; Mackay, D.; Maddalena, R. The role of multimedia mass balance models for assessing the effects of volatile organic compound emissions on urban air quality. Atmos. Environ. 2006, 40, 2986−2994. (14) U.S. EPA. EPiSuite MPBPWin v1.43, 1.43, 2000. (15) U.S. EPA. EpiSuite BioHCWin v 1.01a, 2008. (16) Fritz, B. K.; Hoffmann, W. C.; Rohde, A.; Warren, C.; Faulkner, W. B. Simulating and characterizing agricultural ground applications for soil VOC deposition studies. J. ASTM Int. 2010, 7.10277610.1520/ JAI102776 (17) Prince, R.; Haitmanek, C.; Lee, C. The primary aerobic biodegradation of biodiesel B20. Chemosphere 2008, 71, 1446−1451. (18) Foster, K. L.; Mackay, D.; Parkerton, T. F.; Webster, E.; Milford, L. Five-Stage Environmental exposure assessment strategy for mixtures: gasoline as a case study. Environ. Sci. Technol. 2005, 39, 2711−2718. (19) Coulon, F.; Whelan, M.; Patton, G.; Semple, K.; Villa, R.; Pollard, S. Multimedia fate of petroleum hydrocarbons in the soil: oil matrix of constructed biopiles. Chemosphere 2010, 81, 1454−1462. (20) Park, K. S.; Sims, R. C.; DuPont, R.; Doucette, W. J.; Matthews, J. E. Fate of PAH compounds in two soil types: Influence of volatilization, abiotic loss and biological activity. Environ. Toxicol. Chem. 1990, 9, 187−195. (21) Wang, C.; Wang, F.; Wang, T.; Tao; Bian, Y.; Jiang, X. PAHs biodegradation potential of indigenous consortia from agricultural soil and contaminated soil in two-liquid-phase bioreactor (TLPB). J. Hazard. Mater. 2010, 176, 41−47. (22) Woodrow, J. E.; Seiber, J. N.; Baker, L. W. Correlation techniques for estimating pesticide volatilization flux and downwind concentrations. Environ. Sci. Technol. 1997, 31, 523−529. (23) Woodrow, J. E.; Seiber, J. N.; Dary, C. Predicting pesticide emissions and downwind concentrations using correlations with estimated vapor pressures. J. Agric. Food Chem. 2001, 49, 3841−3846.

(24) Wang, D.; Chen, J.; Xu, Z.; Qiao, X.; Huang, L. Disappearance of polycyclic aromatic hydrocarbons sorbed on surfaces of pine [Pinua thunbergii] needles under irradiation of sunlight: volatilization and photolysis. Atmos. Environ. 2005, 39, 4583−4591. (25) Steyaert, N. L.; Hauck, M.; Van Hulle, S.; Hendricks, A. Modelling bioaccumulation of semi-volatile organic compounds (SOCs) from air in plants based on allometric principles. Chemosphere 2009, 77, 727−732. (26) Cropp, R. A.; Hawker, D. W.; Boonsaner, M. Predicting the accumulation of organic contaminants from soil by plants. Bull. Environ. Contam. Toxicol. 2010, 85, 525−529. (27) Xu, C.; Dong, D.; Meng, X.; Su, X.; Zheng, X.; Li, Y. Photolysis of polycyclic aromatic hydrocarbons on soil surfaces under UV irradiation. J. Environ. Sci. 2013, 25, 569−575.

6875

DOI: 10.1021/acs.jafc.5b01076 J. Agric. Food Chem. 2015, 63, 6866−6875