ARTICLE pubs.acs.org/est
Nonsingular Adsorption/Desorption of Chlorpyrifos in Soils and Sediments: Experimental Results and Modeling Seyoum Yami Gebremariam,*,† Marc W. Beutel,† Markus Flury,‡ James B. Harsh,§ and David R. Yonge† †
Department of Civil and Environmental Engineering, Washington State University, Pullman, Washington 99164-2910, United States Department of Crop and Soil Sciences, Washington State University, Puyallup, Washington 98371-4900, United States § Department of Crop and Soil Sciences, Washington State University, Pullman, Washington 99164-6420, United States ‡
bS Supporting Information ABSTRACT: At environmentally relevant concentrations in soils and sediments, chlorpyrifos, a hydrophobic organic insecticide, showed strong adsorption that correlated significantly with organic matter content. Chlorpyrifos desorption followed a nonsingular falling desorption isotherm that was estimated using a memorydependent mathematical model. Desorption of chlorpyrifos was biphasic in nature, with a labile and nonlabile component. The labile component comprised 1828% of the original solid-phase concentration, and the residue was predicted to slowly partition to the aqueous phase, implying long-term desorption from contaminated soils or sediments. The newly proposed mechanism to explain sorption/desorption hysteresis and biphasic desorption is the unfavorable thermodynamic energy landscape arising from limitation of diffusivity of water molecules through the strongly hydrophobic domain of soils and sediments. Modeling results suggest that contaminated soils and sediments could be secondary long-term sources of pollution. Long-term desorption may explain the detection of chlorpyrifos and other hydrophobic organic compounds in aquatic systems far from application sites, an observation that contradicts conventional transport predictions.
’ INTRODUCTION Hydrophobic organic chemicals (HOCs) were deemed immobile in the environment by early studies that predicted low migration potential on the basis of the chemicals’ low aqueous solubility and high soil/water partition coefficients,1,2 yet HOCs have been widely detected in streams, rivers, groundwater, soil, and sediment far from application sites.3 The adsorption of HOCs to soil and sediment has largely been assumed to be reversible in contaminant transport models,4,5 but laboratory studies have indicated that HOCs can display adsorption/desorption nonsingularity, a phenomenon also known as hysteresis.6 Nonsingularity refers to a smaller sorbate release during desorption compared to initial adsorption.6,7 Two types of hysteresis have been identified in the literature: apparent and true. Apparent hysteresis is an experimental artifact arising from a range of mechanisms, including mass loss and solute transformations,8 difficulty in achieving equilibrium during adsorption or desorption,6 adsorption to suspended colloids,9 and repeated agitation of soil in batch experiments.10 True hysteresis, in which experimental artifacts have been accounted for or eliminated, has been documented for a handful of aromatic compounds.1114 While hysteresis is a pervasive and widely studied phenomenon, and adsorption/desorption hysteresis has been well-known for watersilica gel systems for over a century, HOC-related adsorption hysteresis is poorly understood. Other than the independentdomain theory proposed decades ago for gas-phase solutes,15 r 2011 American Chemical Society
mathematical models describing nonsingularity of liquid-phase solutes are unavailable and descriptions of HOC hysteresis are generally limited to quantification of hysteresis indices.12 Also, while the biphasic nature of HOC desorption is generally accepted,16 existing approaches to modeling biphasic desorption fail to adequately quantify the labile component or predict the long-term fate of the nonlabile fraction. The objective of this study was to evaluate adsorption/desorption hysteresis of a model HOC compound, radio-labeled chlorpyrifos, in soils and sediments at environmentally relevant concentrations. We developed a novel mathematical approach that estimates the magnitude of the labile faction and the fate of the nonlabile fraction of HOCs bound to soil and sediment, a critical limitation of current pesticide transport modeling, using data obtained from a limited number of chlorpyrifos desorption steps. Chlorpyrifos (O,O-diethyl O-3,5,6-trichloro-2-pyridyl phosphorothioate) is an organophosphorus pesticide (see the Supporting Information, Figure S1) used for a broad range of agricultural, industrial, and residential pest control applications. With an annual average global use of 25 million kg of active ingredient between 2002 and 2006, chlorpyrifos has been widely detected in Received: September 22, 2011 Accepted: November 16, 2011 Revised: November 14, 2011 Published: November 16, 2011 869
dx.doi.org/10.1021/es203341b | Environ. Sci. Technol. 2012, 46, 869–875
Environmental Science & Technology
ARTICLE
Table 1. Physical and Chemical Properties of Soils and Sediments
a
EC,
CEC,
sand concn,
silt concn,
clay concn,
area,
TN concn,a
OC concn,b
soil/sediment source
pH
mS/cm
cmol/kg
wt %
wt %
wt %
m2/g
wt %
wt %
θc
Grayland cranberry field
3.80
1.65
13
94
2
4
0.90
0.13
2.90
107
Long Beach cranberry field
3.87
0.48
8
94
3
3
1.25
0.13
2.40
97
Moscow bulrush wetland
6.46
0.79
20
7
66
27
20.71
0.14
2.13
45
Moscow cattail wetland
6.77
0.78
18
7
70
23
16.99
0.12
1.77
58
Paradise Creek
6.23
0.68
18
29
55
16
11.43
0.11
1.57
58
vegetated filter strip
7.24
2.18
15
18
74
8
14.49
0.08
1.23
65
Sunnyside cattail wetland Spring Valley Reservoir
6.04 5.31
1.91 0.20
10 20
63 10
27 75
10 15
9.96 9.20
0.07 0.03
0.90 0.63
95 59
Total nitrogen. b Organic carbon. c Airwatersolid contact angles in degrees.
the environment and poses a widespread public health concern.17 The detection of chlorpyrifos away from application sites, contrary to predictions based on its low water solubility ( 0) (both endothermic and spontaneous). Unlike exothermic processes that spontaneously spread heat, entropy-driven endothermic processes become spontaneous because they lead to the spreading of mass,40 thereby leading to a continuous release of chlorpyrifos with every step of the desorption cycle. Therefore, as can be inferred from eq 5, the extent of desorption as determined by the energy of desorption (ΔG° < 0) is dependent on temperature and the interaction between polar water molecules and hydrophobic organic matter to which chlorpyrifos was partitioned, rather than sorbent-related diffusion processes. This was evidenced partly by the fact that desorption increased with increasing temperature. Results from other studies also support
Figure 3. Experimental (a, c) and simulated (b, d) descending desorption isotherms for Grayland cranberry field soil (upper panels) and Spring Valley Reservoir sediment (lower panels).
biphenyls,33 but less than estimates (>50%) for chlorinated benzenes 34,35 in sediments. The dynamics of biphasic desorption was elucidated further using descending desorption curves (Figure 3). Experimental results (n = 5) and model simulation based on eqs 4a and 4b (n = 2000) showed differences between soils and sediments. For example, desorption was much more prominent from the loworganic Spring Valley Reservoir sediment compared to the highorganic Grayland cranberry field soil. Desorption equilibrium concentrations for Grayland soil at the early stages of desorption were 4 times less than those of the reservoir sediment, and it would require over 5 times as many desorption steps before the same labile fraction was removed from Grayland soil when compared to the reservoir sediment (Figure 3). About 48% of the labile fraction was desorbed from the reservoir sediment in four consecutive desorption cycles compared to 8% of the labile fraction from the highly organic Grayland cranberry field soil. The labile fractions estimated for the two samples, 24% for the reservoir sediment and 18% for the cranberry soil, were also substantially different (Table 2). Simulated desorption falling isotherms for the Grayland soil (Figure 3b) more closely approached the ordinate than for the reservoir sediment (Figure 3d), suggesting that desorption was more resistant in the high organic soils and that hysteresis is more pronounced in soils with high affinity for chlorpyrifos. Our model predicts a continuous release of the nonlabile chlorpyrifos in small amounts in contrast to others32,34,35 who propose that the nonlabile component of HOCs is irreversible. The performance of the modeling approach in this study was assessed using a parity plot for over a dozen desorption steps performed to generate a separate experimental data set not used in model development (see the Supporting Information, Figure S3). Model-generated data significantly correlated with experimental p < 0.001). Our effort is different from previous approaches of modeling of two-stage desorption29,31,32 in that the model minimizes uncertainties by depending on measurable initial conditions (Vo, Co, Ws, r) and two parameters, Km and b. The standard error associated with the estimates of Km and b was generally smaller than 5% of the estimated value (Table 2), and p values of the model fit were
