Atrazine Sorption Kinetics In A Characterized Soil: Predictive

Feb 1, 2008 - An experimental check and error analyses indicate that the type of model can be used for predictive calculations. The physical meaning o...
0 downloads 14 Views 122KB Size
Download PDF
Environ. Sci. Technol. 2008, 42, 1537–1541

Atrazine Sorption Kinetics In A Characterized Soil: Predictive Calculations DONALD S. GAMBLE* Department of Chemistry, Saint Mary’s University, Halifax, Nova Scotia, B3H 3C3 Canada

Received September 27, 2007. Revised manuscript received December 7, 2007. Accepted December 10, 2007.

The use and risks of agricultural pesticides will continue. It is proposed here that better control and possibly some prevention of environmental and health problems should replace the arbitrary standards and post- event monitoring that are still current practice. Mathematical models have been developed for atrazine in a characterized soil from outside Ottawa, Ontario. Experimental data obtained by the on line HPLC microextraction method were used for the development of the models. The labile sorption sites were treated as a reactant and the number of sites per gram of soil was used to define stoichiometry. This allowed a second-order kinetics integral rate law to be used for sorption from solution onto labile sorption sites, and a first order kinetics integral rate law to be used for bound residue formation. An experimental check and error analyses indicate that the type of model can be used for predictive calculations. The physical meaning of the distribution coefficient KD is also considered. The model suggests some practical implications for leaching through soil and for transport by storm runoff. The type of model would be best used for providing input data for fate and transport hydrology models.

Introduction As fuel crops begin to compete with food crops for agricultural land, the need for environmental protection from industrial scale pesticide use will increase. The development and testing of pesticide fate and transport hydrology models will become more important. For such models to be predictive, the input for chemical mechanisms will need to be as realistic as possible. Chemical reaction mechanism models could serve that purpose. The first step is the development and testing of sorption kinetics models. The next step would be the inclusion of chemical reactions in the models. Substantial progress on pesticide sorption kinetics and mechanisms by immersed soils has been reported by several authors, and this has supported some modeling (1-12). The basic facts reported by these and other authors are wellknown. A sorption from solution to wetted surfaces is governed by second order kinetics. This is followed by slower intraparticle diffusion from surface sites to interior sites. The second step is generally described by first-order kinetics. Sorption can continue for days or even months. Sorption rates and amounts are influenced by chemical structures of the pesticides, the physical structures of the soils, and the types and amounts of chemical materials in them. Humic materials have can have big effects, but clays and metal oxides can also influence the sorption. As expected, temperature * Telephone: (902) 667-1974; e-mail: [email protected]. 10.1021/es702432v CCC: $40.75

Published on Web 02/01/2008

 2008 American Chemical Society

influences sorption processes. A reason for developing models that are more predictive is the need for providing fate and transport hydrology models with more realistic input. The work of several laboratories including those cited here (1-12) has made available the information base that is necessary for the development of more predictive models. This has now been made more effective by additional concepts and experimental methods. It had previously been demonstrate that it is possible to make predictive calculations for the reactions of metal ions and organic chemicals with humic materials, which are complicated mixtures (13). With the introduction of an experimental method that allows total sorption to be resolved into labile and bound residue fractions, the same concepts and methods have been extended to whole soils, which include inorganic as well as humic components (14). The use of the numbers of occupied and empty sorption sites as reactants instead of empirical BET surface areas, the law of mass action instead of Freundlich isotherms, and the second-order kinetics integral rate law instead of pseudo first order kinetics can now make models more predictive. As Langford et al. have shown for humic mixtures, the natural mixture of a soil yields distributions of kinetic rate coefficients (15). In chromatography terms, a subsoil fate and transport model represents a sequence of theoretical plates. The sorption kinetics models represent individual theoretical plates in such a sequence. When a soil column has horizons, each horizon would need its own model for its theoretical plates. For each theoretical plate, t ) 0 would be the time at which the pesticide reached that theoretical plate, and the corresponding initial pesticide solution concentration MT would be determined by the transit time and sorption kinetics in the preceding theoretical plates. Depending on the relative rates of sorption and flow, equilibria might or might not be attained in a particular theoretical plate. Also, because distribution coefficients and kinetic rate coefficients would usually be functions of time, each theoretical plate needs a sorption kinetics model rather than simply a set of constant parameters. Obviously, theoretical plate models should include the kinetics of any chemical reactions that occur, but that will require future experimental work. Beyond that, for microbiological processes to be accounted for, the physical processes and chemical reactions will have to be quantitatively understood. Because some experimental data for pesticide sorption kinetics in soils is already available, work on the first step can begin. Although the fact that soil organic matter has a significant influence on the distribution coefficient KD is well-known, the effect is not included in the previously reported work and is not separately accounted for here. It has been deliberately omitted for four reasons. First, the KD and KOC have always been reported as constants for total sorption rather than for only the labile sorbed fraction. But it is only the labile sorbed fraction that will immediately leach, and both sorbed fractions are generally time dependent. The next reason is that the correction for the amount of soil organic matter to give KOC values is only an approximation. The chemical compositions of soil organic matter can have variable distributions of molecular weights and can therefore have different pH dependent solubilities. The number of carboxyl groups and the mole fraction of them that are protonated can vary (13) . Also, KD and KOC are sometimes regarded as representing equilibria, which might not exist. Finally, comparable questions can be raised about clays that are often present in somewhat larger amounts than organic matter. The impact on engineering practice of these apVOL. 42, NO. 5, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1537

FIGURE 1. Titration of soil GB843 labile sorption sites with atrazine using on line HPLC microextraction. θC ) 0.397 × 10-6 (mol/g), std. dev. ) 0.02 × 10-6. Relative std error ) 7% (16, 17). proximations is unknown. Any attempt to replace the present KOC practice with better approximations would require a separate large project. It is therefore outside the scope of the present project. The previously reported physical and chemical analysis of the soil used for this research leaves that option open, however, for the future.

Previously Reported Experiments The solution, labile sorbed, and bound residue kinetics curves at 25° C for 8 × 10-6 (Mol/L) of atrazine with Soil GB843 have been previously reported (16, 17). Comprehensive physical and chemical analyses have also been reported for Soil GB843 (16, 17). Table S1 in the Supporting Information (18) summarizes part of the soil properties. The labile sorption capacity θC was used for the development of models. For atrazine and immersed Soil GB843, it had previously been measured by the titration of atrazine onto the labile sorption sites (17). That is, the total number of sites per gram of soil available for the reversible sorption of atrazine was determined. Although few laboratories have measured it, θC is a critical constant because it allows the numbers of occupied and empty labile sorption sites to be estimated. It then becomes possible to use chemical stoichiometry for conventional chemical kinetics and equilibria, instead of surface areas by BET or EGME, Freundlich isotherms, and pseudofirst-order kinetics. The titration curve in Figure 1 (17) has a plateau at 0.397 × 10-6 (mol/g). The relative standard error is 7%. The titration curve had two experimental requirements. First, the time dependent total sorption had to be resolved into the time-dependent labile sorbed and bound residue fractions. The equilibration time for the labile sorbed fraction had next to be identified on the labile sorption kinetics curve. As long as the labile sorption curve has reached equilibrium, the titration does not require that the bound residue also be at equilibrium. Bound residue equilibration is sometimes slow. The on-line HPLC microextraction method for resolving total sorption into labile sorption and bound residue kinetics curves has been use for several years (19). Bruckler et al. have recently reported two sorbed states identified by a radioisotope method (20), but it is not known whether or not chemical stoichiometry could be determined that way.

Theory The first sorption step, which is sorption of the pesticide from bulk solution onto the labile sorption sites, is known to be kinetically second order. But for pesticide sorption, it is usually described with pseudo-first-order kinetic rate coefficients. There has usually been no means for translating the ambiguous pseudo-first-order rate coefficients into more predictive second-order rate coefficients. The work of Ma et al. (8), however, highlights the importance of a second-order description. The difficulty has been the lack of information 1538

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 5, 2008

about the second reactant. For pesticides and immersed soils, the second reactant is the number of unoccupied labile sorption sites. For a few cases including the present example (16, 17), the difficulty has been resolved by the measurement of the total number of labile sorption sites as illustrated in Figure 1. The numbers of empty labile sorption sites have consequently been estimated by material balance to establish stoichiometry. For transfer from solution to labile sorption sites, the integral rate law for second-order kinetics as described by Laidler has been used. This is the first sorption step of the models. In eq 1, it has been copied with Laidler’s notation (21). This has been transformed into eqs 2-5 for use in spreadsheets. The ratio V/W represents L of solution/g of soil. The units conversion is required for the comparison of laboratory experiments and for the translation of bench scale laboratory tests into field conditions. The labile sorption sites gain atrazine sorbed from solution, but also lose atrazine by bound residue formation. This makes the calculations for labile sorption and bound residue interdependent, so that iterative calculations are necessary. b(a - x) 1 ln ) kt a-b a(b - x)

(1)

Z1 ) ab[(exp(a - b)(V/W)kF1t) - 1]

(2)

Z2 ) a[(exp(a - b)(V/W)kF1t) - b]

(3)

ML ) x ) Z1/Z2

(4)

M1 ) MT - ML

(5)

In eq 1, a, b, and x are Laidler’s symbols for two reactants and the reaction product. k is the second order kinetic rate coefficient. The equation is defined for cases having a > b. In the present case, b is the concentration (mol/L) of dissolved pesticide. a ) (W/V)θa0 (mol/L) is the sample concentration of unoccupied labile sorption sites. θa0 (mol/g) is the number of empty labile sorption sites per mass of soil. kF1 g/(moles × days) is the second0order kinetic rate coefficient expressed in units that permit translation into the conditions of other cases including field conditions. MT and M1 are pesticide solution concentrations (mol/L) at reaction times t ) 0 and t ) ti. A desorption from the labile sites with first order kinetics was anticipated. This would have added extra terms to eq 1. But all attempts to observe it in the experimental data have failed. Assuming that it exists, it must be too small to be observed outside the measurement errors. The most likely explanation is that the experimentally observed labile site coverage remained small. It has generally been less than 10% of the labile sorption sites, partly because of bound residue formation. In desorption experiments, the concentration gradients are reversed so that desorption kinetics can be monitored as Cornelisson at al have recently done (9). Although diffusion theory should account for bound residue formation, other authors have found that first order kinetics describes it within experimental error (1- 14). This is experimentally confirmed by the present work and described by eq 6. θd1 (mol/g) is the amount (W/V)θd1 ) [ML - (W/V)θd1 ]{ 1 - exp[-kd1(tb - ta )]} (6) of bound residue. The reaction time for eq 6 is (tb - ta) because ta has been found to not be 0. For reaction times smaller than that, the labile sorption is still too small for bound residue formation to be experimentally observed. The labile sorbed curve in Figure 2 is a net accumulation curve. Labile sorption is therefore accounted for by the difference calculated by eq 7. This makes (W/V)θa1 ) q[ML - (W/V)θd1]

(7)

the bound residue and labile sorption calculations interdependent, so that iterative calculations are required. q is an

FIGURE 2. Mathematical model 1 for atrazine in soil GB843, from the Greenbelt Farm of the Central Experimental Farm, Ottawa Ontario (16, 17). Number of labile sorption sites from Figure 1. Second-order kinetics for labile sorption, followed by first order kinetics for bound residue formation. 46 data points. Relative standard errors: solution, 0.80%; labile sorbed, 13.%; bound residue, 0.70%.

TABLE 1. Experimental Parameters for Sorption Kinetic Models of Atrazine in Soil GB843 (triplicate runs for each experiment; 25 °C)a experimental parameter model experiment no. total no. of data points labile sorption capacity (mol/g) solution volume (mL) soil mass (g) initial concentration (mol/L)

experiment no.

model no.

1 1 46

2 2 121

3.97 × 10-7

3.97 × 10-7

24.943 0.508440 7.933 × 10-6

25.000 0.500000 1.000 × 10-6

a Extra digits were carried during computing to avoid truncation and roundoff errors. Rounding off should be done after computing.

empirical constant which is generally very close to 1 in the present case.

Development of Models The type of model presented here has three parts. The first part describes solution loss kinetics with Laidler’s secondorder kinetics, using an experimentally based data set of rate coefficients. The rate coefficients decrease with reaction time. The second part uses a first-order kinetics description of bound residue formation, also with a data set of experimentally based rate coefficients that decrease with reaction time. The third part produces a labile sorption curve by material balance. The mathematical interdependence of the three parts requires iterative calculations. Two kinetics models for sorption in two steps were created. They are based on eqs 1-7with the spreadsheet settings in Table S2 of the Supporting Information. Model 1 represents the experiment in Table S3 and model 2 corresponds to the experiment in Table S4 of the Supporting Information. The distinguishing experimental parameters are in Table 1. Model 1 was developed for the initial concentration of 8 × 10-6 mol/L and used for predictions at 1 × 10-6 mol/L. Model 2 was developed for the lower concentration and used for testing the predictions of model 1. The models look exactly like the experiments from which they were developed. Figure 2 for model 1 is an example. The differences in the solution volumes and soil weights were small compared with the large

FIGURE 3. Prediction: the solution concentration of atrazine for an initial concentration of 1.0 × 10-6 (mol/L). An experimental test of the model 1 prediction. Relative standard errors: prediction, 0.80%; experiment 2, 1.4%. difference in solution concentration. Test calculations showed that it was difficult to see any effects of these small differences in solution volumes and soil weights. Model 1 was set up by entering formulas for eqs 2-7 into a spreadsheet. Corel Quattro Pro 8 was used because the graphing was found to be convenient. Table S2 in the Supporting Information (18) lists the spreadsheet columns into which formulas for the equations were entered, together with the additional columns that are used for iterative calculations. The formulas have been copied into enough rows for all of the experimental data points. The spreadsheet has a graph of the model, Figure 2, which exactly resembles the graph of original experiment. Other graphs have been prepared with which predictive calculations can be monitored. Additional graphs can be added as required. Model 1 was calibrated by using data points for the kinetics curves for dissolved, labile bound, and bound residue atrazine of experiment 1. Second-order kinetic rate coefficients corresponding to the data points were first set so that the model reproduced the experimental solution curve to within six decimal places. The extra digits were produced for the particular purpose of reducing as much as possible the accumulation and propagation of errors. Kinetics calculations are known to be especially sensitive to calculation errors, and computing can cause truncation and roundoff errors. Rounding off of data should only be done after the computing and other calculations have been completed. Each of the rate coefficient values required an iterative calculation. Firstorder rate coefficients were next set to reproduce the bound residue and labile sorbed curves together. Iterative calculations were again required. As indicated in Table S2 of the Supporting Information (18), the iterative calculations were done by copying numerical values into columns AD and AE, from the formulas in columns DF and DG. The extra decimal places beyond the numbers of experimental significant figures were produced for the purpose of avoiding as much as possible the accumulation of rounding off and truncation errors. It is intended that final rounding off of numerical values should be done according to error estimates, after predictive calculations have been completed. Model 1 was set up for the initial atrazine concentration of MT ) 8 × 10-6 mol/L. Its ability to make predictions was then tested. As a direct test, it was used to predict kinetics curves for a case having an initial atrazine concentration of 1 × 10-6 mol/L. These predictions were checked against experiment 2, using estimated error limits. The temperature for both experiments was 25° C. To make the standard errors significantly smaller than the standard deviations, the numbers of data points listed in Table S2 of the Supporting Information were used (18). Total data points were assembled from triplicate experimental runs. Relative standard errors are reported for general use in comparisons. Model 2 was set VOL. 42, NO. 5, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1539

FIGURE 4. Prediction: bound residue of atrazine for an initial concentration of 1.0 × 10-6 (mol/L). An experimental test of the model 1 prediction. Relative standard errors: prediction, 0.70%; experiment 2, 1.9%.

FIGURE 5. Prediction: Labile sorption of atrazine for an initial concentration of 1.0 × 10-6 (mol/L). An experimental test of the model #1 prediction. Relative standard errors: prediction, 13.%; experiment 2, 11.%. up for comparison, using its parameters in Table 1 with the model 1 spreadsheet as a template. It was calibrated with the data points of experiment 2, for which the initial atrazine concentration was MT ) 1. × 10-6 mol/L. The second-order kinetic rate coefficients for labile sorption in models 1 and 2 are shown together in Figure S2 of the Supporting Information. For bound residue formation, the first-order rate coefficients of the two models are presented in Figure S3 of the Supporting Information. The kinetic rate coefficients are listed in Tables S3 and S4 of the Supporting Information (18) for future use. Extra decimal places are again included so that truncation and rounding off errors can be reduced. Rounding off should be done after the completion of calculations.

Results and Discussion The prerequisite first step was the determination of the stoichiometry. That is, the concentration (mol/g) of empty, occupied, and total soil sorption sites for atrazine were determined. This was done by titration and material balance

calculation. The titration of atrazine onto the labile sorption sites has been plotted in Figure 1. On line HPLC micro extraction resolved total sorption into the labile and bound residue fractions. Equilibrium data for the labile sorption were then plotted against solution concentration. The site saturation plateau determined the labile sorption capacity θC. The second step was the experimental production of kinetics curves for all three physical states. These are the dissolved, labile sorbed, and bound residue states. On-line HPLC micro extraction with material balance calculations produced the kinetics curves in Figure 2, for the three physical states of model 1. The labile sorption decreased from a maximum because of bound residue formation. Surfacesolution equilibrium was maintained while the bound residue continued to increase. The model 2 experiment had a comparable set of kinetics curves. Testing predictive calculations was the third step. Model 1 was used to predict the solution kinetics curve for a case having an initial solution concentration of 1 × 10-6 (Mol/L). Model 2, which was based on an independent data set, has tested this prediction in Figure 3. In Figures 4 and 5, the bound residue and labile sorption predictions have been tested in the same way. The amplification of the measurement errors is seen here, especially for the labile sorption which had small differences between the relatively larger solution and bound residue curves. For cases having much larger labile sorption, the effects of measurement errors would be much smaller. Figure S1 in the Supporting Information shows a similar test of a prediction, for total recoverable atrazine. The two experiments used for the models have given kinetic rate coefficients with the same behavior. Both kF1 curves in Figure S2 of the Supporting Information had the usual initial fast decrease within the first day. Because an initial decrease caused by the establishment of a concentration gradient at the solution-solid interface is expected to be faster than that, some other cause is suspected. Structure changes in the sorbed layer of water and solute molecules could be speculated. More experiments would be required to distinguish between bias errors and systematic trends. The soil must have a mixture of labile sorption site acting as a mixture of reactants for the sorption. Distributions of weighted average kinetic rate coefficients for small reagents with mixtures have been reported by other authors, for example Langford et al. (15). Figure S3 in the Supporting Information shows a similar comparison for the bound residue rate coefficients of the two models. The total recovery in Figure S1 of the Supporting Information indicates the total amounts that could be leached or otherwise transported by water. Figure 3 has a similar prediction for the amount of atrazine in solution. It shows what might happen with a storm runoff, if only the sediment were trapped while the water runoff continued. The predicted bound residue in Figure 4 can only be transported if the soil or sediment that has trapped it is moved by water. If it has been trapped by intra particle diffusion, the pesticide is

FIGURE 6. Distribution coefficient in model 1 from experiment 1: the effect of sorpotion type. b, Defined in terms of labile sorbed atrazine, relative standard error 15.%. , Defined in terms of total sorbed atrazine, relative standard error 14.%. 1540

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 5, 2008

protected from bio degradation by microorganisms that are too big for molecular sized pores. But when the concentration gradients reverse, it can slowly diffuse back out into solution. That contributes to a persistence problem. The small bias discrepancy in the predicted bound residue is within the estimated error limits. The labile sorption curves in Figure 5 have the largest amount of error propagation. The calculation of the labile sorption as a difference between bigger numbers tends to amplify the measurement errors. But on one hand if the labile sorption is a big enough fraction of total sorption to be important, the error amplification is limited. On the other hand if the labile sorption is small enough to have a significantly increased error, its contribution to total sorption is less important. Either way, the problem does not get seriously out of control if the measurement errors are small enough. The experimental error in the labile sorption capacity θC is the same for both experiments, so that the comparison would tend to cancel it out. But because the potential for quantitative predictions has been demonstrate, more careful measurements of labile sorption capacities are justified. Errors smaller than 7% should be possible. The experimental curves in Figure 6 reveal some questions about the definition and practical use of the distribution coefficient KD. The resolution of total sorption into labile sorbed and bound residue fractions raises the question about whether KD should be defined in terms of the quickly reversible sorption, or in terms of the total sorption. Related to that are the uses in fate and transport hydrology models of sink and retardation parameters. Also for this particular example neither of the curves reaches constant values, and this is likely true for other examples. In general, the potential of models for predictive calculations could have two types of limitations to be considered. One is the realism of the model. The other is the propagation of measurement errors. The two possibilities have different implications for subsequent research. In the present case, the results are consistent with the limitation being the measurement errors. There are some possibilities for the refinement of the experimental method. For example, solid phase micro extraction fibers might be used for solution measurements instead of phase separation methods. Better data logging systems might produce more measurements for the reduction of standard errors. A reviewer has pointed out that the fate ant transport models with which the sorption-reaction models might be used also have some remaining problems under investigation. Especially in clay soils, macropores and cracks cause flow characteristics not accounted for in idealized chromatography theory.

Supporting Information Available Figures S1-S3 and Tables S1-S4 (PDF). This material is available via the Internet at http://pubs.acs.org.

Literature Cited (1) Karickhoff, S. W.; Morris, K R. Sorption Dynamics Of Hydrophonic Pollutants In Sediment Suspension. Environ. Toxicol. Chem. 1985, 4, 469–479.

(2) Wu, S.-C.; Gschwend, P. M. Sorption kinetics of hydrophobic organic compounds to natural sediments and soils. Environ. Sci. Technol. 1986, 20, 717–725. (3) Weber, W. J., Jr.; Smith, E. H. Simulation and Design Models For Adsorption Processes. Environ. Sci. Technol. 1987, 1040– 1050. (4) Weber, W. J., Jr.; McGinley, P. M.; Katz, L. E. Sorption Phenomena In Subsurface Systems: Concepts, Models And Effects On Contaminant Fate And Transport. Water Res. 1987, 25, 499– 528. (5) Sparks, D. L.; Fendorf, S. E.; Zhang, P. C.; Tang, L. In Migration and Fate of Pollutants in Soils and Subsoils; Petruzzeli, D., Helfferich, F. G., Eds.; NATO ASI Series G; Springer-Verlag: Berlin, 1993; Vol. 32. (6) Walker, A.; Barnes, A. Simulation of herbicide persistence in soil; a revised computer model. Pestic. Sci. 1981, 12, 123–132. (7) Miller, C. T.; Pedit, J. A. Use of a reactive surface-diffusion model to describe apparent sorption-desorption hysteresis and abiotic degradation of lindane in a subsurface material. Environ. Sci. Technol. 1992, 26, 1417–1427. (8) Ma, L.; Selim, H. M. Predicting atrazine adsorption desorption in soils: A modified second-order kinetic model. Water Resour. Res. 1994, 30, 447–456. (9) Cornelissen, G; van Noort, P. C. M.; Parsons, J. R.; Govers, H. A. J. Temperature dependence of slow adsorption and desorption kinetics of organic compounds in sediments. Environ. Sci. Technol. 1997, 31, 454–460. (10) Pignatello, J. J.; Xing, B. Mechanisms of slow sorption of organic chemicals to natural particles. Environ. Sci. Technol. 1996, 30, 1–11. (11) Ball, W. P.; Roberts, P. VLong-Term Sorption Of Halogenated Organic Chemicals By Aquifer Material. 2. Intraparticle Diffusion. Environ. Sci. Technol. 1991b, 25, 1237–1249. (12) Reichenberger, S.; Laabs, V. Kinetic evaluation of pesticide sorption in two contrasting Tropical soils. XII Symposium Pesticide Chemistry 2006, 309–528. (13) Gamble, D. S.; Langford, C. H.; Bruccoleri, A. G. Chemical Stoichiometry and Molecular Level Mechanisms As Support for Future Predictive Engineering. In Use of Humic Substances to Remediate Polluted Environments: From Theory to Practice; Perminova, I. V., Herkorn, N., Baveye, P., Eds.; NATO Science Series; Kluwer Academic: Dordrecht, The Netherlands, 2005; Chapter 6. (14) Gamble, D. S.; Lamoureux, M. Analytical chemical methods for determining reaction mechanisms in soils and sediments. Refereed Proceedings of EnviroAnalysis 2006. The 6th Biennial Conference on Monitoring and Measurement of the Environment,Toronto, Canada, May 15–17, 2006. (15) Mak, M. K. S.; Langford, C. H. A kinetic study of the interaction of hydrous aluminum oxide colloids with a well-characterized soil fulvic acid. Can. J. Chem. 1982, 60 (15), 2023–2028. (16) Li, J. Equilibrium and Kinetics Studies of Atrazine and Lindane Uptake By Soils and Soil Components. Doctoral Thesis, Concordia University, Montreal, Canada, 1993. (17) Li, J.; Langford, C. H.; Gamble, D. S. Atrazine sorption by a mineral soil: Processes of labile and nonlabile uptake. J. Agric. Food Chem. 1996, 44, 3672–3679. (18) Supporting Information. (19) Gamble, D. S.; Bruccoleri, A. G.; Lindesay, E.; Leyes, G. A.; Langford, C. H. Chlorothalonil in a quartz sand soil: Speciation and kinetics. Environ. Sci. Technol. 2000, 34, 120–124. (20) Saffih-Hdadi, K.; Bruckler, L.; Lafolie, F.; Barriuso, E. A Model for Linking the Effects of Parathion in Soil to its Degradation and Bioavailability Kinetic. J. Environ. Qual. 2006, 35, 253–267. (21) Laidler, K. J. Chemical Kinetics, 1st ed.; Mcgraw-Hill: New York, 1950, p 10, eq 32.

ES702432V

VOL. 42, NO. 5, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1541