Environ. Sci. Technol. 2001, 35, 2375-2380
Chlorothalonil and Its 4-Hydroxy Derivative in Simple Quartz Sand Soils: A Comparison of Sorption Processes DONALD S. GAMBLE,* ELSPETH LINDSAY, ALDO G. BRUCCOLERI, AND COOPER H. LANGFORD Department of Chemistry, University of Calgary, Calgary, Alberta T2N 1N4 GREGORY A. LEYES Ricerca Incorporated, 7528 Auburn Road, P.O. Box 1000, Painesville, Ohio 44077-1000
Quartz sandy soils from Simcoe, Ontario, Canada and North Carolina had sorption properties for chlorothalonil that were nearly the same. For labile surface sorption kinetics, the Simcoe soil gave a pseudo first-order rate constant of kS1 ) (7.4 ( 0.7) × 10-2 days-1. At equilibrium, the labile surface sorption capacity θC of Simcoe soil for chlorothalonil was 23.8 × 10-6 (mol/g). The sorption properties of the 4-hydroxy derivative of chlorothalonil were different in two important respects. They were larger by an order of magnitude, and they were substantially different for the two soils. Sorption by the Simcoe soil was too fast for kinetics measurements by the on-line HPLC micro extraction method, but for the North Carolina soil kS1 ) (1.15 ( 0.01) days-1 was recorded. For the Simcoe and North Carolina soils, respectively, θC > 200 (µmol/g) and θC ≈113 (µmol/ g). Two conclusions can be drawn. First, the replacement of the Cl by OH on the 4 position of chlorothalonil makes the sorption effects much greater. Second, the stronger interactions are associated with a greater sensitivity to small differences in the chemical compositions of the soils. Subtle soil properties causing significant effects might include small amounts and physical structures of organic matter and metal oxides. This implies that, for predictive computer models, mechanism parameters will have to be correlated in two dimensions: chemical structure, and the composition and amounts of chemical materials in soils.
Introduction In their report of a pesticide transport survey over extensive areas of two states, Blanchard and Lerch (1) concluded that “The chemistry of the contaminant determines the potential hydrologic transport pathways for that chemical to be lost from soil.” They found this to be the most dominant of three factors, with land use the least important. For sorption processes, chemical structures of the pesticides and chemical compositions of the soils are implicated. Predictive engineering calculations will require that the molecular level mechanisms be quantitatively determined. Data from moni* Corresponding author phone: (902) + 667-1974; fax: (902) + 667-1984; e-mail:
[email protected]. 10.1021/es001864n CCC: $20.00 Published on Web 04/27/2001
2001 American Chemical Society
toring activities frequently only document consequences after events, but are unable to support predictions for management and prevention (2). The test protocols used for commercial product registration are, in contrast, intended to be used for the safe management of practical field operations (3, 4). The intention is to prevent problems, especially uncontrolled persistence and leaching in soils. From this it follows that a higher priority should be given to the development of more predictive computer models and the test protocols that support them, with a lower priority given to monitoring activities. Lerch et al. (5) have recently commented that agronomy management practices have less influence on the persistence of atrazine than soil chemical and hydrologic processes do. An example is Sharom and Stephenson’s observation (6) that movement of metribuzin in soil tends to be inversely related to the ability of soil to bind it. Van Zoonen et al. (7) were more specific. They stated that “Pesticide mobility in soils is critically dependent on pesticide species, the soil properties, and other environmental factors.” Hamaker and Thompson were the first to suggest, as verified by others, that intraparticle diffusion of pesticides in soils is central to the control of sorption processes (8, 9). Apparent sorption hysteresis and other slow processes are known to result (10-13). A difficulty for the program of developing predictive parameters for model development, mentioned for example by Brunauer, is that the complex nature of soils contributes to an incomplete understanding of sorption mechanisms (14). This problem needs to be addressed so that the results of test protocols are not too site specific. This is why the factors that influence the numerical values of the physicalchemical parameters include the types and amounts of chemical materials in soils, and the chemical structures of the pesticides. Therefore, the physical-chemical processes of pesticides in soils should be characterized by correlations in at least two dimensions. One of these consists of the correlations of physical-chemical mechanism constants with the chemical structures of the pesticides. The other is the correlation of these same constants with the types and amounts of chemical materials in soils. The categories of chemical materials in soils are well-known. They include clays, other silicates, metal oxides, and organic matter. Each of these categories has several types of materials. The chemical structure factors of organic chemical pesticides include the sizes and shapes of the molecules, and the distributions of electric charges over them. Zero and nonzero net charge, charge separation to form dipoles, and polarizability are among the electric charge properties of the molecules. A database of experimental values is necessary for the correlations. Because of the large number of possible cases, a first step would be correlations with typical or important representatives of each category of soil chemical materials. A baseline for this could be the comparisons of a pair of related chemical structures in very simple similar soils dominated by the simple mineral phase of quartz sand. In order for predictive tools such as computer models (15, 16) to be more reliable and less site specific, both the models and their input data must more effectively account for the molecular level mechanisms that link effects to the causes that drive them. The molecular level physicalchemical mechanism parameters include labile surface sorption capacities, sorption-desorption kinetic rate constants, and law of mass action equilibrium functions, as well as the kinetics and equilibria of any chemical reactions that occur (3, 4, 17-26). These are the mechanism parameters VOL. 35, NO. 11, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2375
for which the correlations mentioned above are required. As a first step, the purpose of the present research was to compare the interactions of two compounds that are structurally related, in two similar, very simple, quartz sand soils. Future research could then use this information as a reference point or origin for correlations in the two dimensions as mentioned above. The comparisons of chlorothalonil and its 4-hydroxy derivative in sand soils that have 0.2% (North carolina) and 5% (Simcoe) nonquartz components might provide a useful starting point. The necessary analytical chemical methods, supporting theory, and test results for the present project have been described previously (18-23, 25). The chemical compositions of the soils have been documented (3, 4).
sorption sites that often proves to be related to rapid extractability by organic solvents. The other neglected issue is the possibility of interior particle sorption sites that do not bind strongly. Labile interior sorption capacities are to be expected but tortuous diffusion means access on a longer time scale. For that reason such sites have not yet been measured. Qualitatively, it is anticipated that sorption effects should depend on both the chemical structures of pesticide molecules, and the chemical compositions of soils. The quantitative descriptions of such effects require systematic investigations with different chemical structures and known chemical compositions of soils. The purpose of the present work is to add to a database of structure and soil composition effects for future correlations with mechanism parameters.
Theory
Experimental Section
A number of authors have contributed over the course of several years to an understanding of a minimum two-stage model for sorption of pesticides in soils. At least some of them can be cited here (3, 4, 8-10, 15, 2, 17, 21, 24, 26-32). The kinetics and equilibrium of labile surface sorption are outlined by equations 1 and 2. In some cases, the reactions are too fast for detailed kinetic analysis by the on-line HPLC micro extraction method. Eqs 2 and 3 define an empirical labile surface sorption equilibrium function and capacity, θC, in terms of the empty and filled surface sorption sites θ0 and θ1, respectively (3, 4, 2, 31). In those cases where pesticide water solubility allows a sufficiently wide range of concentration variation, the equilibrium function may even be measured as has been done in some cases (e.g., ref. 22). If additional research can achieve the more difficult equilibrium titration of surface sorption sites, then θC and K1 values might be obtainable for some of the other cases. M1 is the solution
The instrumentation and experimental procedure for the on-line HPLC micro extraction method used in this work were the same as previously used (2, 4, 19-23, 25). The experimental samples are slurries consisting of soil solids suspended in aqueous solutions of chemicals such as pesticides. Briefly, the method involves two HPLC runs for each time point. In one run, the whole soil slurry is injected and trapped by the on-line filters ahead of the guard column. The mobile phase extracts the “labile” sorbed pesticide so that the HPLC gives the sum of dissolved and labile sorbed. In the second run, the slurry is filtered off-line so that only dissolved pesticide is measured. The difference gives the labile sorbed. Mass balance allows calculation of the nonlabile sorbed. Background level and material balance tests confirmed the previously demonstrated validity of the method (2, 21). As before, the Nylon 66 micro filters used for solution phase analyses were calibrated for chlorothalonil sorption (3, 4). Chemical analyses of the two simple quartz sand soils designated Simcoe and North Carolina have been described in separate reports (3, 4). Powder X- ray patterns of the North Carolina soil indicated 99-100% quartz sand (4). Organic C was detected only at a low level. In contrast, about 2.4% of lepidocrocite, FeO(OH), was noted in the Simcoe soil. The full chemical composition of the Simcoe sand soil is shown in Table 1(3). Analyses produced no additional data for a North Carolina soil table. The other notable difference from the North Carolina soil is the presence of 1.4% C by weight, implying almost 3% organic matter in Simcoe. The chemical compositions therefore differed by about 5-10%. The lepidocrocite used separately for sorption kinetics experiments was a natural mineral received from the Department of Geology of the University of Calgary. It was a soft, brittle crystalline material having a dark brown color. It was easily ground to a fine powder and used without further treatment. X-ray diffraction confirmed the crystal structure. Experimental samples were prepared by slurrying the soil in water for 2 days. Samples were then spiked with chlorothalonil or its 4-hydroxy derivative, and were then made up to volume. Each sample had 25 mg of soil or lepidocrocite slurried in 25 mL of aqueous solution. The chlorothalonil experiments were conducted as previously reported (3, 4, 18, 19, 21, 22). For 4-hydroxy derivative experiments, slurry samples had an initial concentration of 2.73 × 10-5 M. The 22-µm Nylon filters used for off-line filtration were calibrated for the sorption of the 4-hydroxy derivative of chlorothalonil. The calibration curve in eq 5 for the 22-µm Nylon filters was used for all of the solution phase analysis curves.
-
( )
dM1 ) kB1 θ0M1 - kS2 θ1 dt K1 )
θ1 M 1 θ0
θC ) θ1 + θ0
(1)
(2) (3)
concentration in molarity. kB1 is a second-order rate constant for labile sorption, and kS2 is a first-order rate constant for desorption. Labile surface sorption equilibrium is governed by the equilibrium function K1. Eq 4 is widely used (2) for the first-order kinetics approximation to the intra particle diffusion process.
( )
dθD ) kDθ1 dt
(4)
θD is the coverage of sorption sites from which the pesticide may not be readily extracted (nonlabile or “bound” sites). Often, such sites are those in the interiors of soil particles. The kinetics of uptake to the nonlabile sites is usually successfully described by a first-order rate constant, kD. Two related issues have received little attention. The first is that at the molecular level of sizes, the geometric concept of “surface” is ambiguous. With pores and cracks communicating between the outsides and insides of particles, any attempt to distinguish between surfaces and interiors produces arbitrary definitions. An experimental part of the problem is that different surface area measurement methods such as the ethylene glycol monoethyl ether (EGME) and Brunauer, Emmett & Teller (BET) surface measurement methods, give different numerical values for surface area. An operational definition of the distinction between the surfaces and interiors of soil particles can be based on the kinetics of access to the 2376
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 11, 2001
M1 ) 3.81 × 10-6 + 1.010MF
(5)
The corrections of 18.5% or less did not contribute significantly to the estimated experimental errors. In preparation
TABLE 1. Analyses of Soils from Simcoe, Ontario, and North Carolina method
component
symbol/formula
result
Simcoe, Ontario soil elemental analysis chemical analysis powder X-ray diffraction
Powder powder X-ray diffraction and chemical analysis
carbon hydrogen nitrogen iron phosphorus quartz calcite albite brushite
C H N Fe P SiO2 CaCO3 NaAlSi3O8 CaPO3(OH)•2H2O
1.4 wt % 0.14 wt % 0.03 wt % 1.66 wt % 0.047 wt % 90-95 wt %
lepidocrocite
FeO(OH)
> 2.4.wt %
dufrenite
Fe5(PO4)3(OH)5•2H2O
< 0.35wt %
C H N SiO2
