- Email: [email protected]
Cosolvent effects on sorption isotherm linearity
Journal of Contaminant Hydrology 56 (2002) 159 – 174 www.elsevier.com/locate/jconhyd
Cosolvent effects on sorption isotherm linearity $ Dermont C. Bouchard* US Environmental Protection Agency, Ecosystems Research Division, 960 College Station Road, Athens, GA 30605, USA Received 25 April 2001; received in revised form 10 August 2001; accepted 16 November 2001
Abstract Sorption – desorption hysteresis, slow desorption kinetics, and other nonideal phenomena have been attributed to the differing sorptive characteristics of the natural organic polymers associated with soils and sediments. In this study, aqueous and mixed solvent systems were used to investigate the effects of a cosolvent, methanol, on sorption isotherm linearity with natural organic matter (NOM), and to evaluate whether these results support, or weaken, the rubbery/glassy polymer conceptualization of NOM. All of the sorption isotherms displayed some nonlinear character. Our data indicates that all of the phenanthrene and atrazine isotherms were nonlinear up to the highest equilibrium solution concentration to solute solubility in water or cosolvent ratios (Ce/Sw,c) used, approximately 0.018 and 0.070, respectively. Isotherm linearity was also observed to increase with volumetric methanol content ( fc). This observation is consistent with the NOM rubbery/glassy polymer conceptualization: the presence of methanol in NOM increased isotherm linearity as do solvents in synthetic polymers, and suggests that methanol is interacting with the NOM, enhancing its homogeneity as a sorptive phase so that sorption is less bimodal as fc increases. When the equilibrium solution concentration was normalized for solute solubility in water or methanol – water solutions, greater relative sorption magnitude was observed for the methanol – water treatments. This observation, in conjunction with the faster sorption kinetics observed in the methanol – water sediment column systems, indicates that the increase in relative sorption magnitude with fc may be attributed to the faster sorption kinetics in the methanol – water systems, and hence, greater relative sorptive uptake for the rubbery polymer fraction of NOM at similar time scales. D 2002 Elsevier Science B.V. All rights reserved. Keywords: Sorption; Solubility; Transport; Mixed solvents; Kinetics
$ This paper has been reviewed in accordance with the US Environmental Protection Agency’s peer and administrative review policies and approved for publication. Mention of trade names or commercial products does not constitute endorsement or recommendation for use by the US EPA. * Tel.: +1-706-3558-333; fax: +1-706-3558-202. E-mail address: [email protected] (D.C. Bouchard).
0169-7722/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 9 - 7 7 2 2 ( 0 1 ) 0 0 2 1 3 - 3
160
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
1. Introduction Sorption of hydrophobic organic compounds (HOCs) to natural geosorbents is a major determinant of HOC transport, degradation and biological activity in the environment. For most soils and sediments, HOC sorption occurs primarily in the natural organic matter (NOM) fraction. Much research has indicated a linear relationship between the HOC sorbed and solution phase concentrations at equilibrium, i.e., linear sorption isotherms have been observed (Chiou et al., 1983; Kan et al., 1994; Chiou et al., 1998), thus supporting the concept of sorption as a partitioning process between the aqueous phase and a homogenous (with respect to partitioning) NOM phase. However, a growing body of work has indicated substantial sorption isotherm nonlinearity, particularly at low equilibrium solution concentration to solute solubility in water ratios (Ce/Sw), indicating that some processes, in addition to the linear partitioning between two homogeneous phases, are occurring (Spurlock and Biggar, 1994; Weber and Huang, 1996; Xia and Pignatello, 2001). Using a suite of substituted urea herbicides and phenolic compounds, Chiou and Kile (1998) observed significant sorption isotherm nonlinearity at Ce/Sw < 0.10– 0.13 for these polar solutes; however, above this concentration range, the isotherms were practically linear. Spurlock and Biggar (1994) also observed sorption isotherm nonlinearity for two substituted urea herbicides, and that the nonlinear sorption coefficient, n, increased with decreasing solute concentration. Sorption of polar solutes on NOM has also been observed to be competitive (Xing and Pignatello, 1998; Xing et al., 1996). The results of these latter studies support a specific-interaction model where polar solutes interact with, and compete for, specific NOM sites. Given the heterogeneous nature of NOM, these sites likely present a range of sorption energies, and the resulting sorption isotherms are nonlinear. The causes of sorption isotherm nonlinearity for nonpolar solutes at low Ce/Sw are less clear however. Sorption nonlinearity for nonpolar solutes appears to be restricted to lower relative concentrations than for polar solutes. Chiou and Kile (1998) observed significant sorption isotherm nonlinearity at Ce/Sw < 0.010 – 0.015 for nonpolar solutes, a linear range in order of magnitude lower than for the polar solutes. Again, as with the polar solutes, above this concentration range the isotherms were practically linear. Investigations of nonpolar solute sorption have indicated that nonlinearity increases with solute –sorbent contact time (Weber and Huang, 1996; Huang and Weber, 1998) and varies in magnitude for different types of NOM (Young and Weber, 1995; Xing et al., 1996; Chiou et al., 2000). In a study with 10 natural sorbents of differing geologic ages and organic matter compositions, Huang and Weber (1997) found that phenanthrene sorption isotherm nonlinearity was inversely correlated with the oxygen/carbon atomic ratios of the sorbent organic matter, with geologically older shales displaying greater nonlinearity than younger peat materials. These differences in nonlinearity between sorbents were interpreted in this and related papers using concepts from polymer chemistry. In this interpretation, the geologically younger, less condensed NOM is assumed to behave like a polymer in the rubbery state, where sorption is linear; and the geologically older, more condensed NOM is assumed to act like a polymer in the glassy state, where sorption is nonlinear. If NOM behaves similarly to less heterogeneous, synthetic polymers, with both a rubbery and glass-like nature, then one would expect that solvents would have similar effects on the sorptive properties of both NOM and synthetic polymers. Therefore, the
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
161
significance of solute sorption on NOM in the presence of cosolvents extends beyond the sorption and transport characteristics of a particular solute –sorbent –cosolvent system, and may be used to assess the accuracy of current conceptualizations of NOM and the sorption –desorption processes of HOCs. The influence of water-miscible organic solvents on HOC sorption on NOM was initially described by Rao et al. (1985) with subsequent investigations into differing aspects of the effects of cosolvents on contaminant sorption and transport (Wood et al., 1990; Brusseau et al., 1991; Nzengung et al., 1996; Bouchard, 1998). The primary organic cosolvent that has been used in these studies has been methanol. Methanol was also selected for this study because of the large database for its use in soils, its presence in the polymer science literature, and because of its high dissolution capacity in NOM (Chiou and Kile, 1994). In the mixed solvent studies cited above, sorption isotherm linearity was observed over a limited solute concentration range, or was assumed for transport model simplification even though curvilinear tendencies may sometimes be observed in the data. In either case, sorption isotherm linearity in mixed solvent systems has not been well described, particularly at low equilibrium solution concentration to solute solubility in water, or cosolvent, ratios (Ce/Sw,c), where nonlinearity would be most evident. Hence, the objectives of this study were to use aqueous and mixed solvent systems to investigate the effects of a cosolvent, methanol, on sorption isotherm linearity for NOM, and to evaluate whether these results support, or weaken the rubbery/ glassy polymer conceptualization of NOM.
2. Material and methods 2.1. Solubility studies To allow the determination of the Ce/Sw,c values, measurement of phenanthrene and atrazine solubility at varying fc was necessary. Solubility studies were conducted with 12C phenanthrene and atrazine, while the batch sorption and column studies utilized the 14Clabelled compounds. All methanol (Fisher Scientific, Pittsburgh, PA) used was HPLC grade; all aqueous solution components were 0.01 N in CaCl2. Based on the solubility estimates generated by the SPARC properties calculator (Hilal et al., 1994), a 10 excess of solute was equilibrated with aqueous or water – methanol solutions in centrifuge tubes with Teflon-lined screw caps. Phenanthrene treatments were performed in duplicate and atrazine treatments in triplicate. The centrifuge tubes were placed on an orbital shaker for 48 h, then centrifuged at 600 RCF for 30 min to effect phase separation. An aliquot of the solution phase was then removed and phenanthrene and atrazine concentration determined using reverse-phase HPLC with UV detection at 250 and 222 nm, respectively. 2.2. Batch sorption studies The batch sorption techniques used were similar to those used previously by the author (Bouchard, 1999). All solutions used in the batch sorption and column studies contained 200 mg/l sodium azide as a biocide. The Call’s Creek and Upper Call’s Creek sediments were air-dried and passed through a 1-mm sieve prior to use, those fractions yielding total
162
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
organic carbon (TOC) contents of 0.10% and 0.75%, respectively. The particle size fractions of both sediments were dominated by the sand and silt fractions with < 1% clay content. Sediment and solution were equilibrated for 1 week in batch reactors; the phases were then separated by centrifugation at 600 RCF for 30 min, and the supernatant analyzed by liquid scintillation counting. The amount of phenanthrene or atrazine sorbed to sediment was determined by difference. 2.3. Column studies The experimental apparatus and methodology used in the sediment column miscible displacement experiments were similar to those used previously by the author (Bouchard, 1998). In the column system, the packed sediment column was oriented vertically and connected to two syringe pumps through an inert valve that allowed switching between the eluting solutions delivered by the pumps to the column. One pump contained eluting solutions of 0.01 N CaCl2, or 0.01 N CaCl2 and methanol, and the other pump contained the eluting solution with phenanthrene, atrazine, or 3H2O as solutes. The syringe pumps allowed for very accurate maintenance of the average pore water velocity at 32.5 cm h 1. Breakthrough curves (BTCs) were then conducted by introducing a pulse of the solute containing eluting solution into the column and monitoring column effluent until the solute effluent concentration, C (nmol ml 1), approached the influent concentration, C0 (nmol ml 1), i.e., C/C0 = 1. This solution was then displaced from the column with the same eluting solution without solute until C/C0 = 0 was approached. Tritiated water, 3H2O (Sigma, St. Louis, MO), was used as a nonsorbing tracer to characterize hydrodynamic dispersion in the column systems. Bulk density (q), saturated water content (h) and Peclet number (Pe) of the packed column were 1.55 g cm 3, 0.415 cm3 cm 3 and 78.0, respectively. 2.4. Data analyses For the batch experiments, the HOC equilibrium solution phase concentration (Ce, nmol ml 1), or Ce normalized by solubility in water or cosolvent (Ce/Sw,c), was plotted versus the HOC sorbed phase concentration (S, nmol g 1) and linear: S ¼ KCe
ð1Þ
and nonlinear (Freundlich): S ¼ Kf Cen
ð2Þ
S ¼ Kf norm ðCe =Sw,c Þn
ð3Þ
models used to describe the sorption data. A linear approximation of Kf, KL, was obtained by taking the average slope of the nonlinear isotherm: KL ¼ 1=ðC0 Ci ÞSCCi0 dS=dc dc
ð4Þ
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
163
which for Ci = 0, reduces to KL ¼ Kf C0n1
ð5Þ
A two-region, or bicontinuum, first-order mass transfer model was used to analyze the data generated in the BTC experiments. This model has been used with considerable success to describe solute transport through a variety of geosorbents, including soils, sediments and aquifer materials (Lee et al., 1988; Piatt et al., 1996). For the bicontinuum model, sorption in domain 1 is assumed to be instantaneous, and sorption in domain 2 is rate limited and described by first-order reversible kinetics. Five parameters (T, Pe, R, b, x) are required to run the bicontinuum model. The pulse width, in terms of pore volumes (T ), is known from measurement. The Peclet number was obtained from the 3H2O BTC data by using a nonlinear, least-squares optimization program (Van Genuchten, 1981) to solve the advective – dispersive local equilibrium solute transport model. The retardation factor was obtained by substituting in Eq. (6): R ¼ 1 þ ðq=hÞKL
ð6Þ
Values for the first-order desorption rate coefficient (k2, h 1) were calculated by substitution in Eq. (7): x ¼ k2 ð1 bÞRL=v
ð7Þ
where L is the column length (cm), and v is the pore water velocity (cm h 1). Values for b, the fraction of total retardation attributed to sorption in the instantaneous domain; and x, the Damkohler number which represents the ratio between hydrodynamic residence time and the characteristic time of sorption, were estimated by using a nonlinear, least-squares optimization program (Van Genuchten, 1981) for the bicontinuum model under flux-type boundary conditions.
3. Results and discussion 3.1. Solubility Proof of solution saturation was the presence of some visible phenanthrene and atrazine crystals following the equilibration period. Some undissolved analyte crystals were apparent in all of the treatments except for the phenanthrene treatments at fc > 0.60, indicating SPARC underestimation of phenanthrene solubility at the higher methanol concentrations. As a result, additional phenanthrene was added to the fc >0.60 treatments and these treatments were re-equilibrated for an additional 48 h. Due to the very similar densities of phenanthrene and water and water – methanol solutions at ambient laboratory temperatures, separation of the solid and liquid phases by centrifugation was problematic. In this study, phenanthrene was less dense than the aqueous and methanol –water solutions at fc < 0.20, and more dense than the methanol – water solutions at fc >0.20. To avoid any sampling of undissolved phenanthrene in the solubility determinations, the top 2-cm of the
164
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
supernatant of all treatments at fc < 0.30 was removed after the initial equilibration period, and the samples re-centrifuged. The low data scatter in Fig. 1 verifies the utility of this technique. Phenanthrene solubility over the full range of methanol volumetric fraction ( fc) displayed a biphasic nature with a linear segment below fc = 0.35 and a segment of different slope and greater nonlinearity above that value (Fig. 1). For phenanthrene, SPARC yielded solubility estimates within an order of magnitude of the measured values in the lower fc range; however, SPARC estimates showed increasing deviation from the measured values at fc >0.35. As noted above, these deviations were greater than an order of magnitude at fc >0.60. Deviations from the ideal log –linear relationship between solute solubility and fc have been observed in the other studies (Kimble and Chin, 1994; Li and Andren, 1994; Li et al., 1996). The observed deviations from ideality have been less for methanol than for some other solvents, but may still result in significant error when assuming linearity over a wide fc range. As the lower fc range was linear for phenanthrene, and because it represents more realistic environmental cosolvent concentrations, the methanol concentration in the remainder of this study was fc < 0.30 for both phenanthrene and atrazine. Fig. 2 depicts phenanthrene and atrazine solubility profiles at fc < 0.30. Somewhat surprisingly, SPARC yielded solubility estimates that were closer to the measured values for the more chemically heterogeneous compound, atrazine. In estimating the slopes (r) of the solubility profiles, the SPARC calculated r value for atrazine (95% C.L. = 1.581 –
Fig. 1. Phenanthrene solubility in methanol.
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
165
Fig. 2. Phenanthrene and atrazine solubility in methanol at low fc.
2.082) was lower than the r value derived from the measured data (95% C.L. = 2.129 – 2.269). However, for phenanthrene, SPARC calculated (95% C.L. = 3.148 – 3.647) and measured (95% C.L. = 3.159 – 3.484) r values were not significantly different. Values of r have been shown to be correlated with solute properties, such as molecular surface area, and to solvent properties such as dielectric constant and bulk surface tension (Rubino and Yalkowsky, 1987). They are measures of solute – cosolvent interactions and are an index of the solubilizing power of the cosolvent; they may be used to approximate the slope of the log K – fc relationship when cosolvent – sorbent interactions are not significant (Bouchard, 1998). 3.2. Batch sorption studies Fig. 3 contains representative batch sorption isotherms for atrazine on the Upper Call’s Creek sediment at fc = 0.00 and 0.30. Atrazine sorption on the Call’s Creek sediment was so low in the aqueous treatments (Table 1), that changes in the sorption isotherms with increasing fc were not measurable. As indicated in Fig. 3 and in Table 1, all of the other isotherms displayed some nonlinear character. The mean value of the exponent in Eq. (2), n, was less than one for all treatments, although n was not different from one for three of the treatments at the 95% level of significance (Table 1). By design, these isotherms were conducted at low Ce/Sw,c, which has been observed to yield greater nonlinearity than isotherms conducted at higher relative concentrations (Chiou and Kile, 1998; Chiou et al.,
166
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
Fig. 3. Atrazine sorption isotherms for Upper Call’s Creek sediment at fc = 0.00 and 0.30.
2000). Our data indicates that the phenanthrene and atrazine isotherms were nonlinear up to the highest Ce/Sw,c values used, approximately 0.018 and 0.070, respectively. Also evident in Fig. 3 and in Table 1 is that isotherm linearity increased with methanol content ( fc). In an earlier study, linear atrazine sorption isotherms were observed over a methanol range from fc = 0.00 to 0.50; however, isotherms of the substituted urea herbicide diuron were observed to increase in linearity with fc over the same range (Nkedi-Kizza et al., 1989). Most of the hypotheses that have been proposed to explain sorption isotherm nonlinearity have utilized a bimodal construct that differentiates NOM sorption capacity into two components, one linear and one nonlinear. Chiou and co-workers (Chiou and Kile, 1998; Chiou et al., 2000) have proposed that sorption occurs by linear partitioning to the bulk of NOM, but that the nonlinear sorption of nonpolar solutes is due to sorption onto a small amount of high-surface-area carbonaceous material, such as charcoal-like substances contained in soils and sediments, and that the greater nonlinear range observed for polar solutes is due to additional specific interactions with NOM. The nonlinear ranges of the nonpolar phenanthrene and more polar atrazine cannot be compared directly in this study as the nonlinear ranges were not exceeded, i.e., linear isotherm regions were not observed over the concentration ranges used in this study. However, the data in Table 1 indicate that the degree of phenanthrene and atrazine isotherm nonlinearity, as reflected in the n values, was similar within sediment and fc treatments. The increase in isotherm linearity with fc observed in this study suggests that methanol is interacting with the NOM, enhancing its homogeneity as a sorptive phase so that
Sediment Call’s Creek
Upper Call’s Creek
Chemical phenanthrene phenanthrene phenanthrene atrazine phenanthrene phenanthrene atrazine atrazine
fc 0.00 0.10 0.20 0.00 0.00 0.30 0.00 0.30
r2
K a
11.8 (11.2 – 12.4) 7.78 (7.26 – 8.29) 5.31 (4.91 – 5.71) 0.296 (0.288 – 0.304) 184 (170 – 198) 23.8 (22.7 – 24.9) 2.18 (2.05 – 2.30) 1.04 (1.00 – 1.07)
0.991 0.981 0.972 0.0996 0.981 0.989 0.982 0.995
Kf 8.07 6.18 4.73 0.398 104 23.1 3.56 1.14
n b
(6.89 – 9.26) (5.10 – 7.25) (3.91 – 5.56) (0.373 – 0.422) (86.0 – 122) (21.7 – 24.6) (3.20 – 3.92) (0.990 – 1.28)
0.849 0.867 0.910 0.891 0.775 0.918 0.800 0.967
(0.801 – 0.896) (0.786 – 0.948) (0.802 – 1.02) (0.866 – 0.916) (0.721 – 0.829) (0.824 – 1.01) (0.755 – 0.0844) (0.917 – 1.02)
r2
Kfnorm
0.995 0.985 0.974 0.998 0.991 0.990 0.944 0.995
29.7 42.8 73.7 39.4 341 755 221 726
(23.2 – 36.2) (27.7 – 57.8) (37.1 – 110) (36.6 – 42.1) (255 – 428) (446 – 1060) (191 – 251) (574 – 878)
It should be noted that the use of C.L.’s here is not rigorously correct since errors may not be normally distributed in nonlinear models; however, they still may be used for comparative purposes with the understanding that if the parameter is not normally distributed, there may be a less than 95% chance that the C.L. contains the true value. a 95% confidence limits from linear regression. b 95% confidence limits from nonlinear regression.
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
Table 1 Phenanthrene and atrazine batch sorption parameters
167
168
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
sorption is less bimodal as fc increases. Weber and co-workers (Weber and Huang, 1996; Leboeuf and Weber, 1997; Huang and Weber, 1998; Leboeuf and Weber, 2000) have drawn on synthetic polymer chemistry concepts in describing NOM as a rubbery/glassy polymer where sorption in the expanded NOM rubbery domain occurs by partitioning, which is linear with solute concentration; and where sorption in the more condensed NOM glassy domain is nonlinear due to HOC entrapment or hindered diffusion. Similarly, humic substances have also been described as having both condensed and expanded regions (Hayes and Himes, 1986), and humic polymers molecular weights are similar to those of many synthetic polymers. The rubbery/glassy conceptual model has been supported by other work that has suggested that HOC sorption nonlinearity may be attributed to solute retention in the nanometer-sized voids in the glassy region of NOM (Xing and Pignatello, 1998; Xia and Pignatello, 2001). Our observations are consistent with the NOM rubbery/glassy polymer conceptualization: the presence of methanol in NOM increased isotherm linearity as solvents do in synthetic polymers (Cain et al., 1991; Kamiya et al., 1992). Isotherm linearity has also been observed to increase above the posited NOM glass transition temperatures, which are the temperatures at which the glassy NOM regions transit to the rubbery states (Leboeuf and Weber, 1997). For the NOM, which can be expected to have a range of glass transition temperatures, the presence of solvents with solubility parameters similar to NOM would be expected to lower the glass transition temperatures of some of the NOM regions to below ambient temperatures, thus enhancing linearity. The solubility parameter of methanol, 14.5 cal1/2 cm 3/2 (Barton, 1991), is within the range of solubility parameters estimated for NOM (Freeman and Cheung, 1981; Chiou and Kile, 1994); hence, as observed in this study, methanol would be effective at lowering the NOM glass transition temperature and enhancing isotherm linearity. Figs. 4 – 6 contain phenanthrene and atrazine batch sorption isotherms with the equilibrium solution concentration, Ce, normalized for solute solubility. In each case, and in contrast to the Kf measurements discussed above, the mean partition coefficients from the normalized data (Kfnorm) increased with fc (Table 1), indicating higher sorptive uptake for the methanol – water treatments relative to solute solubility. To clarify, it should be noted that in no case did the actual mass of sorbed solute increase with fc, but rather, the mass of solute sorbed relative to solute solubility (which increases with fc) increased with fc. Though the 95% confidence limits of the mean Kfnorm value for the Call’s Creek sediment at fc = 0.10 overlapped the values at fc = 0.00 and at fc = 0.20, the values at fc = 0.00 and at fc = 0.20 were themselves statistically different. These data indicate that methanol is interacting with the NOM, in this case, to increase sorption magnitude relative to solute solubility. The original conceptualization of HOC sorption on NOM in mixed solvent systems did not account for solvent – sorbent effects (Rao et al., 1985), and many subsequent studies were not designed to measure such effects or they were assumed to be too small relative to the other phenomena being measured. However, prior work by this author and others has shown that organic solvents can increase the sorption – desorption kinetics of HOCs on soils and sediments (Brusseau et al., 1991; Bouchard, 1998). This implies that solvents change the sorbent character in some way that facilitates HOC exchange. The increase in magnitude of relative sorption, and of sorption – desorption kinetics may be interpreted using the rubbery/glassy polymer model.
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
169
Fig. 4. Atrazine sorption isotherms for Upper Call’s Creek sediment normalized for solubility at fc = 0.00 and 0.30.
Fig. 5. Phenanthrene sorption isotherms for Upper Call’s Creek sediment normalized for solubility at fc = 0.00 and 0.30.
170
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
Fig. 6. Phenanthrene sorption isotherms for Call’s Creek sediment normalized for solubility at fc = 0.00, 0.10 and 0.20.
Overall, Figs. 4– 6 and Table 1 indicate that relative sorption increased with methanol content; however, there are two phenomena having opposing effects on sorption that may be operative in these systems. First, by dissolving in the NOM polymer, methanol may reduce the glassy structure of NOM regions making these regions act more like a partitioning medium. Since sorption in the glassy regions is posited to be slow, and partitioning is known to be rapid, the end result may be greater relative sorption in the rubbery domain at similar time scales. However, synthetic polymers have lower sorption capacities in the rubbery than in the glassy state; therefore, it would be expected that at equilibrium, NOM glassy regions would actually yield greater relative sorption. Given that relative sorption increased with fc in this study, it is probable that sorption in the batch reactors was not at true thermodynamic equilibrium so that the faster diffusion in the rubbery than in the glassy NOM resulted in greater sorptive uptake in the rubbery regions at similar time scales. The column studies described below were designed to corroborate or refute this premise. 3.3. Column studies The 3H2O BTC was symmetric, and the data were described well by the advective – dispersive local equilibrium solute transport model (Fig. 7). Such symmetry and equilibrium model fit is indicative of hydrodynamic equilibrium during transport; that is, that diffusional mass transfer of the 3H2O solute into and out of the microporous structure of
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
171
Fig. 7. Atrazine breakthrough curves for Upper Call’s Creek sediment at fc = 0.00.
the sediment occurred rapidly enough relative to solute resident times in the column to be at equilibrium with bulk water transport. Fitting the advective –dispersive local equilibrium solute transport model to the 3H2O data yielded a Peclet number of 78.0; this measurement was consistent with the low dispersion expected for well-packed, unaggregated soils, and is comparable to those measured by other researchers using similar column apparatus (Lee et al., 1988; Nkedi-Kizza et al., 1989). The Peclet number was then used in the two-region nonequilibrium model for simulating HOC transport through the column. Figs. 7 and 8 depict atrazine BTCs at fc = 0.00 and 0.30 for the Upper Call’s Creek sediment. In both cases, the two-region model provided a good fit for the experimental data. When used as model input parameters, the linearized batch sorption values, KL, resulted in 95% confidence limits that captured most of the experimental data. The average KL values were 1.96 and 1.03 for atrazine at fc = 0.00 and 0.30, respectively. The k2 value for the best fit two-region simulation for the aqueous atrazine BTC fell within the range of values previously reported for atrazine in sediments having a range of TOC (Bouchard, 1999). In addition, the k2 values increased from 2.25 to 3.82 h 1 with fc, indicating more rapid sorption – desorption kinetics in the methanol –water systems and corroborating one of the conclusions drawn from the batch data. The poor fit of the equilibrium model to the data in Fig. 8 is a further indication that even though the sorption –desorption kinetics are faster at fc = 0.30, they are still not at equilibrium at these time scales. Looking at the batch and column data collectively, it would appear that the increase in relative sorption magnitude with fc may be attributed to the faster sorption kinetics in the
172
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
Fig. 8. Atrazine breakthrough curves for Upper Call’s Creek sediment at fc = 0.30.
methanol – water systems. If NOM acts like a glassy – rubbery polymer, then as the aqueous and methanol – water systems approach true thermodynamic equilibrium, it would be expected that the differences in relative sorption magnitude would decrease as the ‘‘kinetic advantage’’ of the methanol – water systems diminished. At equilibrium, the aqueous systems would display greater relative sorption as they would have a greater proportion of their NOM in the glassy state. These last two statements cannot be supported directly by the data set presented here, but rather provide some direction for future work using mixed solvents to describe the nature of the NOM sorption process.
4. Conclusions The SPARC properties calculator was useful for estimating solute solubility’s in water and in water – methanol solutions at fc < 0.30, which would apply to most environmental scenarios. Results of the batch and column studies appear to support the rubbery/glassy polymer conceptualization of NOM. The observation that isotherm linearity increased with fc provided evidence that methanol was interacting with the NOM, perhaps having an effect similar to that in synthetic polymers where solvents can bring about a transition from the glassy (nonlinear) to the rubbery (linear) state. The glassy and rubbery polymer regions also differ with respect to sorption capacity and kinetics. Synthetic polymers have lower sorption capacities in the rubbery than in the glassy state; therefore, it would be expected that at equilibrium, sorption would be greater in the NOM glassy regions. However, when
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
173
the batch data in this study were normalized for solute solubility, the mean partition coefficients from the normalized data increased with methanol content, indicating higher sorptive uptake for the methanol – water treatments relative to solute solubility. This observation may be explained by the kinetics differences of the glassy and rubbery regions: sorption in the glassy regions is posited to be slow, and partitioning is known to be rapid. It is probable, therefore, that sorption in the batch reactors was not at true thermodynamic equilibrium and that the faster diffusion in the rubbery than in the glassy NOM regions resulted in greater sorptive uptake in the rubbery regions at similar time scales. In total, the data presented here provides further evidence that NOM may behave similarly to the glassy/rubbery regions of synthetic polymers.
References Barton, A.F.M., 1991. CRC Handbook of Solubility Parameters and Other Cohesion Parameters. CRC Press LLC, Boca Raton, FL, USA. Bouchard, D.C., 1998. Sorption kinetics of PAHs in methanol – water systems. J. Contam. Hydrol. 34 (1 – 2), 107 – 120. Bouchard, D.C., 1999. Sorption of vinclozolin and atrazine on four geosorbents. Pestic. Sci. 55, 1095 – 1102. Brusseau, M.L., Wood, A.L., Rao, P.S.C., 1991. Influence of organic cosolvents on the sorption kinetics of hydrophobic organic chemicals. Environ. Sci. Technol. 25 (5), 903 – 910. Cain, E.J., Wen, W.Y., Jones, A.A., Inglefield, P.T., Cauley, B.J., Bendler, J.T., 1991. A dual-mode interpretation of spin relaxation for 13CO2 sorbed in polycarbonate. J. Polym. Sci., Part B: Polym. Phys. 29, 1009 – 1020. Chiou, C.T., Kile, D.E., 1994. Effects of polar and nonpolar groups on the solubility of organic compounds in soil organic matter. Environ. Sci. Technol. 28 (6), 1139 – 1144. Chiou, C.T., Kile, D.E., 1998. Deviations from sorption linearity on soils of polar and nonpolar organic compounds at low relative concentrations. Environ. Sci. Technol. 32 (3), 338 – 343. Chiou, C.T., Porter, P.E., Schedding, D.W., 1983. Partition equilibria of nonionic organic compounds between soil organic matter and water. Environ. Sci. Technol. 17 (4), 227 – 231. Chiou, C.T., Mcgroddy, S.E., Kile, D.E., 1998. Partition characteristics of polycyclic aromatic hydrocarbons on soils and sediments. Environ. Sci. Technol. 32 (2), 264 – 269. Chiou, C.T., Kile, D.E., Rutherford, D.W., Sheng, G., Boyd, S.A., 2000. Sorption of selected organic compounds from water to a peat soil and its humic-acid and humin fractions: potential sources of the sorption nonlinearity. Environ. Sci. Technol. 34 (7), 1254 – 1258. Freeman, D.H., Cheung, L.S., 1981. A gel partition model for organic desorption from a pond sediment. Science 214, 790 – 792. Hayes, M.H.B., Himes, F.L., 1986. In: Huang, P.M., Schnitzer, M. (Eds.), Interaction of Soil Minerals with Natural Organics and Microbes. Soil Science Society of America, Madison, WI. Hilal, S.H., Carreira, L.A., Karickhoff, S.W., 1994. Estimation of chemical reactivity parameters and physical properties of organic molecules using SPARC. In: Politzer, P., Murray, J.S. (Eds.), Quantitative Treatment of Solute/Solvent Interactions. Theoretical and Computational Chemistry, vol. I. Elsevier, New York, pp. 291 – 353. Huang, W., Weber Jr., W.J., 1997. A distributed reactivity model for sorption by soils and sediments: 10. Relationship between desorption, hysteresis, and the chemical characteristics of organic domains. Environ. Sci. Technol. 31 (9), 2562 – 2569. Huang, W., Weber Jr., W.J., 1998. A distributed reactivity model for sorption by soils and sediments: 11. Slow concentration-dependent sorption rates. Environ. Sci. Technol. 32 (22), 3549 – 3555. Kamiya, Y., Bourbon, D., Mizoguchi, K., Naito, Y., 1992. Sorption, dilation, and isothermal glass transition of poly(ethyl methacrylate) – organic gas systems. Polym. J. 24, 443 – 449. Kan, T.K., Fu, G., Tomson, M.B., 1994. Adsorption/desorption hysteresis in organic pollutant and soil/sediment interaction. Environ. Sci. Technol. 28 (5), 859 – 867.
174
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
Kimble, K.D., Chin, Y.P., 1994. The sorption of polycyclic aromatic hydrocarbons by soils in low-methanol/water mixtures. J. Contam. Hydrol. 17 (1 – 2), 129 – 143. Leboeuf, E.J., Weber Jr., W.J., 1997. A distributed reactivity model for sorption by soils and sediments: 8. Sorbent organic domains: discovery of a humic acid glass transition and an argument for a polymer-based model. Environ. Sci. Technol. 31 (6), 1697 – 1702. Leboeuf, E.J., Weber Jr., W.J., 2000. Macromolecular characteristics of natural organic matter: 2. Sorption and desorption behavior. Environ. Sci. Technol. 34 (17), 3623 – 3631. Lee, L.S., Rao, P.S.C., Brusseau, M.L., Ogwada, R.A., 1988. Nonequilibrium sorption of organic contaminants during flow through columns of aquifer materials. Environ. Toxicol. Chem. 7, 779 – 793. Li, A., Andren, A.W., 1994. Solubility of polychlorinated biphenyls in water/alcohol mixtures: 1. Experimental data. Environ. Sci. Technol. 28 (1), 47 – 52. Li, A., Andren, A.W., Yalkowsky, S.H., 1996. Choosing a cosolvent: solubilization of naphthalene and cosolvent property. Environ. Toxicol. Chem. 15 (12), 2233 – 2239. Nkedi-Kizza, P., Brusseau, M.L., Rao, P.S.C., Hornsby, A.G., 1989. Nonequilibrium sorption during displacement of hydrophobic organic chemicals and 45Ca through soil columns with aqueous and mixed solvents. Environ. Sci. Technol. 23 (7), 814 – 820. Nzengung, V.A., Voudrias, E.A., Nkedi-Kizza, P., Wampler, J.M., Weaver, C.E., 1996. Organic cosolvent effects on the sorption equilibrium of hydrophobic organic chemicals by organoclays. Environ. Sci. Technol. 30 (1), 89 – 96. Piatt, J.J., Backhus, D.A., Capel, P.D., Eisenreich, S.J., 1996. Temperature dependent sorption of naphthalene, phenanthrene, and pyrene to low organic carbon aquifer sediments. Environ. Sci. Technol. 30 (3), 751 – 760. Rao, P.S.C., Hornsby, A.G., Kilcrease, D.P., Nkedi-Kizza, P., 1985. Sorption and transport of hydrophobic organic chemicals in aqueous and mixed solvent systems: model development and preliminary evaluation. J. Environ. Qual. 14 (3), 376 – 383. Rubino, J.T., Yalkowsky, S.H., 1987. Cosolvency and cosolvent polarity. Pharm. Res. 4, 220 – 230. Spurlock, F.C., Biggar, J.W., 1994. Thermodynamics of organic chemical partition in soils: 2. Nonlinear partition of substituted phenylureas from aqueous solution. Environ. Sci. Technol. 28 (6), 996 – 1002. Van Genuchten, M.Th., 1981. Research Report No. 119, US Department of Agriculture, US Salinity Laboratory, Riverside, CA. Weber Jr., W.J., Huang, W., 1996. A distributed reactivity model for sorption by soils and sediments: 4. Intraparticle heterogeneity and phase-distribution relationships under nonequilibrium conditions. Environ. Sci. Technol. 30 (3), 881 – 888. Wood, A.L., Bouchard, D.C., Brusseau, M.L., Rao, P.S.C., 1990. Cosolvent effects on sorption and mobility of organic chemicals in soils. Chemosphere 21 (4 – 5), 575 – 587. Xia, G., Pignatello, J.J., 2001. Detailed sorption isotherms of polar and apolar compounds in a high-organic soil. Environ. Sci. Technol. 35 (1), 84 – 94. Xing, B., Pignatello, J.J., 1998. Competitive sorption between 1,3-dichlorobenzene or 2,4-dichlorophenol and natural aromatic acids in soil organic matter. Environ. Sci. Technol. 32 (5), 614 – 619. Xing, B., Pignatello, J.J., Gigliotti, B., 1996. Competitive sorption between atrazine and other organic compounds in soil and model sorbents. Environ. Sci. Technol. 30 (8), 2432 – 2440. Young, T.M., Weber Jr., W.J., 1995. A distributed reactivity model for sorption by soils and sediments: 3. Effects of diagenetic processes on sorption energetics. Environ. Sci. Technol. 29 (1), 92 – 97.
Cosolvent effects on sorption isotherm linearity $ Dermont C. Bouchard* US Environmental Protection Agency, Ecosystems Research Division, 960 College Station Road, Athens, GA 30605, USA Received 25 April 2001; received in revised form 10 August 2001; accepted 16 November 2001
Abstract Sorption – desorption hysteresis, slow desorption kinetics, and other nonideal phenomena have been attributed to the differing sorptive characteristics of the natural organic polymers associated with soils and sediments. In this study, aqueous and mixed solvent systems were used to investigate the effects of a cosolvent, methanol, on sorption isotherm linearity with natural organic matter (NOM), and to evaluate whether these results support, or weaken, the rubbery/glassy polymer conceptualization of NOM. All of the sorption isotherms displayed some nonlinear character. Our data indicates that all of the phenanthrene and atrazine isotherms were nonlinear up to the highest equilibrium solution concentration to solute solubility in water or cosolvent ratios (Ce/Sw,c) used, approximately 0.018 and 0.070, respectively. Isotherm linearity was also observed to increase with volumetric methanol content ( fc). This observation is consistent with the NOM rubbery/glassy polymer conceptualization: the presence of methanol in NOM increased isotherm linearity as do solvents in synthetic polymers, and suggests that methanol is interacting with the NOM, enhancing its homogeneity as a sorptive phase so that sorption is less bimodal as fc increases. When the equilibrium solution concentration was normalized for solute solubility in water or methanol – water solutions, greater relative sorption magnitude was observed for the methanol – water treatments. This observation, in conjunction with the faster sorption kinetics observed in the methanol – water sediment column systems, indicates that the increase in relative sorption magnitude with fc may be attributed to the faster sorption kinetics in the methanol – water systems, and hence, greater relative sorptive uptake for the rubbery polymer fraction of NOM at similar time scales. D 2002 Elsevier Science B.V. All rights reserved. Keywords: Sorption; Solubility; Transport; Mixed solvents; Kinetics
$ This paper has been reviewed in accordance with the US Environmental Protection Agency’s peer and administrative review policies and approved for publication. Mention of trade names or commercial products does not constitute endorsement or recommendation for use by the US EPA. * Tel.: +1-706-3558-333; fax: +1-706-3558-202. E-mail address: [email protected] (D.C. Bouchard).
0169-7722/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 9 - 7 7 2 2 ( 0 1 ) 0 0 2 1 3 - 3
160
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
1. Introduction Sorption of hydrophobic organic compounds (HOCs) to natural geosorbents is a major determinant of HOC transport, degradation and biological activity in the environment. For most soils and sediments, HOC sorption occurs primarily in the natural organic matter (NOM) fraction. Much research has indicated a linear relationship between the HOC sorbed and solution phase concentrations at equilibrium, i.e., linear sorption isotherms have been observed (Chiou et al., 1983; Kan et al., 1994; Chiou et al., 1998), thus supporting the concept of sorption as a partitioning process between the aqueous phase and a homogenous (with respect to partitioning) NOM phase. However, a growing body of work has indicated substantial sorption isotherm nonlinearity, particularly at low equilibrium solution concentration to solute solubility in water ratios (Ce/Sw), indicating that some processes, in addition to the linear partitioning between two homogeneous phases, are occurring (Spurlock and Biggar, 1994; Weber and Huang, 1996; Xia and Pignatello, 2001). Using a suite of substituted urea herbicides and phenolic compounds, Chiou and Kile (1998) observed significant sorption isotherm nonlinearity at Ce/Sw < 0.10– 0.13 for these polar solutes; however, above this concentration range, the isotherms were practically linear. Spurlock and Biggar (1994) also observed sorption isotherm nonlinearity for two substituted urea herbicides, and that the nonlinear sorption coefficient, n, increased with decreasing solute concentration. Sorption of polar solutes on NOM has also been observed to be competitive (Xing and Pignatello, 1998; Xing et al., 1996). The results of these latter studies support a specific-interaction model where polar solutes interact with, and compete for, specific NOM sites. Given the heterogeneous nature of NOM, these sites likely present a range of sorption energies, and the resulting sorption isotherms are nonlinear. The causes of sorption isotherm nonlinearity for nonpolar solutes at low Ce/Sw are less clear however. Sorption nonlinearity for nonpolar solutes appears to be restricted to lower relative concentrations than for polar solutes. Chiou and Kile (1998) observed significant sorption isotherm nonlinearity at Ce/Sw < 0.010 – 0.015 for nonpolar solutes, a linear range in order of magnitude lower than for the polar solutes. Again, as with the polar solutes, above this concentration range the isotherms were practically linear. Investigations of nonpolar solute sorption have indicated that nonlinearity increases with solute –sorbent contact time (Weber and Huang, 1996; Huang and Weber, 1998) and varies in magnitude for different types of NOM (Young and Weber, 1995; Xing et al., 1996; Chiou et al., 2000). In a study with 10 natural sorbents of differing geologic ages and organic matter compositions, Huang and Weber (1997) found that phenanthrene sorption isotherm nonlinearity was inversely correlated with the oxygen/carbon atomic ratios of the sorbent organic matter, with geologically older shales displaying greater nonlinearity than younger peat materials. These differences in nonlinearity between sorbents were interpreted in this and related papers using concepts from polymer chemistry. In this interpretation, the geologically younger, less condensed NOM is assumed to behave like a polymer in the rubbery state, where sorption is linear; and the geologically older, more condensed NOM is assumed to act like a polymer in the glassy state, where sorption is nonlinear. If NOM behaves similarly to less heterogeneous, synthetic polymers, with both a rubbery and glass-like nature, then one would expect that solvents would have similar effects on the sorptive properties of both NOM and synthetic polymers. Therefore, the
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
161
significance of solute sorption on NOM in the presence of cosolvents extends beyond the sorption and transport characteristics of a particular solute –sorbent –cosolvent system, and may be used to assess the accuracy of current conceptualizations of NOM and the sorption –desorption processes of HOCs. The influence of water-miscible organic solvents on HOC sorption on NOM was initially described by Rao et al. (1985) with subsequent investigations into differing aspects of the effects of cosolvents on contaminant sorption and transport (Wood et al., 1990; Brusseau et al., 1991; Nzengung et al., 1996; Bouchard, 1998). The primary organic cosolvent that has been used in these studies has been methanol. Methanol was also selected for this study because of the large database for its use in soils, its presence in the polymer science literature, and because of its high dissolution capacity in NOM (Chiou and Kile, 1994). In the mixed solvent studies cited above, sorption isotherm linearity was observed over a limited solute concentration range, or was assumed for transport model simplification even though curvilinear tendencies may sometimes be observed in the data. In either case, sorption isotherm linearity in mixed solvent systems has not been well described, particularly at low equilibrium solution concentration to solute solubility in water, or cosolvent, ratios (Ce/Sw,c), where nonlinearity would be most evident. Hence, the objectives of this study were to use aqueous and mixed solvent systems to investigate the effects of a cosolvent, methanol, on sorption isotherm linearity for NOM, and to evaluate whether these results support, or weaken the rubbery/ glassy polymer conceptualization of NOM.
2. Material and methods 2.1. Solubility studies To allow the determination of the Ce/Sw,c values, measurement of phenanthrene and atrazine solubility at varying fc was necessary. Solubility studies were conducted with 12C phenanthrene and atrazine, while the batch sorption and column studies utilized the 14Clabelled compounds. All methanol (Fisher Scientific, Pittsburgh, PA) used was HPLC grade; all aqueous solution components were 0.01 N in CaCl2. Based on the solubility estimates generated by the SPARC properties calculator (Hilal et al., 1994), a 10 excess of solute was equilibrated with aqueous or water – methanol solutions in centrifuge tubes with Teflon-lined screw caps. Phenanthrene treatments were performed in duplicate and atrazine treatments in triplicate. The centrifuge tubes were placed on an orbital shaker for 48 h, then centrifuged at 600 RCF for 30 min to effect phase separation. An aliquot of the solution phase was then removed and phenanthrene and atrazine concentration determined using reverse-phase HPLC with UV detection at 250 and 222 nm, respectively. 2.2. Batch sorption studies The batch sorption techniques used were similar to those used previously by the author (Bouchard, 1999). All solutions used in the batch sorption and column studies contained 200 mg/l sodium azide as a biocide. The Call’s Creek and Upper Call’s Creek sediments were air-dried and passed through a 1-mm sieve prior to use, those fractions yielding total
162
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
organic carbon (TOC) contents of 0.10% and 0.75%, respectively. The particle size fractions of both sediments were dominated by the sand and silt fractions with < 1% clay content. Sediment and solution were equilibrated for 1 week in batch reactors; the phases were then separated by centrifugation at 600 RCF for 30 min, and the supernatant analyzed by liquid scintillation counting. The amount of phenanthrene or atrazine sorbed to sediment was determined by difference. 2.3. Column studies The experimental apparatus and methodology used in the sediment column miscible displacement experiments were similar to those used previously by the author (Bouchard, 1998). In the column system, the packed sediment column was oriented vertically and connected to two syringe pumps through an inert valve that allowed switching between the eluting solutions delivered by the pumps to the column. One pump contained eluting solutions of 0.01 N CaCl2, or 0.01 N CaCl2 and methanol, and the other pump contained the eluting solution with phenanthrene, atrazine, or 3H2O as solutes. The syringe pumps allowed for very accurate maintenance of the average pore water velocity at 32.5 cm h 1. Breakthrough curves (BTCs) were then conducted by introducing a pulse of the solute containing eluting solution into the column and monitoring column effluent until the solute effluent concentration, C (nmol ml 1), approached the influent concentration, C0 (nmol ml 1), i.e., C/C0 = 1. This solution was then displaced from the column with the same eluting solution without solute until C/C0 = 0 was approached. Tritiated water, 3H2O (Sigma, St. Louis, MO), was used as a nonsorbing tracer to characterize hydrodynamic dispersion in the column systems. Bulk density (q), saturated water content (h) and Peclet number (Pe) of the packed column were 1.55 g cm 3, 0.415 cm3 cm 3 and 78.0, respectively. 2.4. Data analyses For the batch experiments, the HOC equilibrium solution phase concentration (Ce, nmol ml 1), or Ce normalized by solubility in water or cosolvent (Ce/Sw,c), was plotted versus the HOC sorbed phase concentration (S, nmol g 1) and linear: S ¼ KCe
ð1Þ
and nonlinear (Freundlich): S ¼ Kf Cen
ð2Þ
S ¼ Kf norm ðCe =Sw,c Þn
ð3Þ
models used to describe the sorption data. A linear approximation of Kf, KL, was obtained by taking the average slope of the nonlinear isotherm: KL ¼ 1=ðC0 Ci ÞSCCi0 dS=dc dc
ð4Þ
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
163
which for Ci = 0, reduces to KL ¼ Kf C0n1
ð5Þ
A two-region, or bicontinuum, first-order mass transfer model was used to analyze the data generated in the BTC experiments. This model has been used with considerable success to describe solute transport through a variety of geosorbents, including soils, sediments and aquifer materials (Lee et al., 1988; Piatt et al., 1996). For the bicontinuum model, sorption in domain 1 is assumed to be instantaneous, and sorption in domain 2 is rate limited and described by first-order reversible kinetics. Five parameters (T, Pe, R, b, x) are required to run the bicontinuum model. The pulse width, in terms of pore volumes (T ), is known from measurement. The Peclet number was obtained from the 3H2O BTC data by using a nonlinear, least-squares optimization program (Van Genuchten, 1981) to solve the advective – dispersive local equilibrium solute transport model. The retardation factor was obtained by substituting in Eq. (6): R ¼ 1 þ ðq=hÞKL
ð6Þ
Values for the first-order desorption rate coefficient (k2, h 1) were calculated by substitution in Eq. (7): x ¼ k2 ð1 bÞRL=v
ð7Þ
where L is the column length (cm), and v is the pore water velocity (cm h 1). Values for b, the fraction of total retardation attributed to sorption in the instantaneous domain; and x, the Damkohler number which represents the ratio between hydrodynamic residence time and the characteristic time of sorption, were estimated by using a nonlinear, least-squares optimization program (Van Genuchten, 1981) for the bicontinuum model under flux-type boundary conditions.
3. Results and discussion 3.1. Solubility Proof of solution saturation was the presence of some visible phenanthrene and atrazine crystals following the equilibration period. Some undissolved analyte crystals were apparent in all of the treatments except for the phenanthrene treatments at fc > 0.60, indicating SPARC underestimation of phenanthrene solubility at the higher methanol concentrations. As a result, additional phenanthrene was added to the fc >0.60 treatments and these treatments were re-equilibrated for an additional 48 h. Due to the very similar densities of phenanthrene and water and water – methanol solutions at ambient laboratory temperatures, separation of the solid and liquid phases by centrifugation was problematic. In this study, phenanthrene was less dense than the aqueous and methanol –water solutions at fc < 0.20, and more dense than the methanol – water solutions at fc >0.20. To avoid any sampling of undissolved phenanthrene in the solubility determinations, the top 2-cm of the
164
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
supernatant of all treatments at fc < 0.30 was removed after the initial equilibration period, and the samples re-centrifuged. The low data scatter in Fig. 1 verifies the utility of this technique. Phenanthrene solubility over the full range of methanol volumetric fraction ( fc) displayed a biphasic nature with a linear segment below fc = 0.35 and a segment of different slope and greater nonlinearity above that value (Fig. 1). For phenanthrene, SPARC yielded solubility estimates within an order of magnitude of the measured values in the lower fc range; however, SPARC estimates showed increasing deviation from the measured values at fc >0.35. As noted above, these deviations were greater than an order of magnitude at fc >0.60. Deviations from the ideal log –linear relationship between solute solubility and fc have been observed in the other studies (Kimble and Chin, 1994; Li and Andren, 1994; Li et al., 1996). The observed deviations from ideality have been less for methanol than for some other solvents, but may still result in significant error when assuming linearity over a wide fc range. As the lower fc range was linear for phenanthrene, and because it represents more realistic environmental cosolvent concentrations, the methanol concentration in the remainder of this study was fc < 0.30 for both phenanthrene and atrazine. Fig. 2 depicts phenanthrene and atrazine solubility profiles at fc < 0.30. Somewhat surprisingly, SPARC yielded solubility estimates that were closer to the measured values for the more chemically heterogeneous compound, atrazine. In estimating the slopes (r) of the solubility profiles, the SPARC calculated r value for atrazine (95% C.L. = 1.581 –
Fig. 1. Phenanthrene solubility in methanol.
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
165
Fig. 2. Phenanthrene and atrazine solubility in methanol at low fc.
2.082) was lower than the r value derived from the measured data (95% C.L. = 2.129 – 2.269). However, for phenanthrene, SPARC calculated (95% C.L. = 3.148 – 3.647) and measured (95% C.L. = 3.159 – 3.484) r values were not significantly different. Values of r have been shown to be correlated with solute properties, such as molecular surface area, and to solvent properties such as dielectric constant and bulk surface tension (Rubino and Yalkowsky, 1987). They are measures of solute – cosolvent interactions and are an index of the solubilizing power of the cosolvent; they may be used to approximate the slope of the log K – fc relationship when cosolvent – sorbent interactions are not significant (Bouchard, 1998). 3.2. Batch sorption studies Fig. 3 contains representative batch sorption isotherms for atrazine on the Upper Call’s Creek sediment at fc = 0.00 and 0.30. Atrazine sorption on the Call’s Creek sediment was so low in the aqueous treatments (Table 1), that changes in the sorption isotherms with increasing fc were not measurable. As indicated in Fig. 3 and in Table 1, all of the other isotherms displayed some nonlinear character. The mean value of the exponent in Eq. (2), n, was less than one for all treatments, although n was not different from one for three of the treatments at the 95% level of significance (Table 1). By design, these isotherms were conducted at low Ce/Sw,c, which has been observed to yield greater nonlinearity than isotherms conducted at higher relative concentrations (Chiou and Kile, 1998; Chiou et al.,
166
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
Fig. 3. Atrazine sorption isotherms for Upper Call’s Creek sediment at fc = 0.00 and 0.30.
2000). Our data indicates that the phenanthrene and atrazine isotherms were nonlinear up to the highest Ce/Sw,c values used, approximately 0.018 and 0.070, respectively. Also evident in Fig. 3 and in Table 1 is that isotherm linearity increased with methanol content ( fc). In an earlier study, linear atrazine sorption isotherms were observed over a methanol range from fc = 0.00 to 0.50; however, isotherms of the substituted urea herbicide diuron were observed to increase in linearity with fc over the same range (Nkedi-Kizza et al., 1989). Most of the hypotheses that have been proposed to explain sorption isotherm nonlinearity have utilized a bimodal construct that differentiates NOM sorption capacity into two components, one linear and one nonlinear. Chiou and co-workers (Chiou and Kile, 1998; Chiou et al., 2000) have proposed that sorption occurs by linear partitioning to the bulk of NOM, but that the nonlinear sorption of nonpolar solutes is due to sorption onto a small amount of high-surface-area carbonaceous material, such as charcoal-like substances contained in soils and sediments, and that the greater nonlinear range observed for polar solutes is due to additional specific interactions with NOM. The nonlinear ranges of the nonpolar phenanthrene and more polar atrazine cannot be compared directly in this study as the nonlinear ranges were not exceeded, i.e., linear isotherm regions were not observed over the concentration ranges used in this study. However, the data in Table 1 indicate that the degree of phenanthrene and atrazine isotherm nonlinearity, as reflected in the n values, was similar within sediment and fc treatments. The increase in isotherm linearity with fc observed in this study suggests that methanol is interacting with the NOM, enhancing its homogeneity as a sorptive phase so that
Sediment Call’s Creek
Upper Call’s Creek
Chemical phenanthrene phenanthrene phenanthrene atrazine phenanthrene phenanthrene atrazine atrazine
fc 0.00 0.10 0.20 0.00 0.00 0.30 0.00 0.30
r2
K a
11.8 (11.2 – 12.4) 7.78 (7.26 – 8.29) 5.31 (4.91 – 5.71) 0.296 (0.288 – 0.304) 184 (170 – 198) 23.8 (22.7 – 24.9) 2.18 (2.05 – 2.30) 1.04 (1.00 – 1.07)
0.991 0.981 0.972 0.0996 0.981 0.989 0.982 0.995
Kf 8.07 6.18 4.73 0.398 104 23.1 3.56 1.14
n b
(6.89 – 9.26) (5.10 – 7.25) (3.91 – 5.56) (0.373 – 0.422) (86.0 – 122) (21.7 – 24.6) (3.20 – 3.92) (0.990 – 1.28)
0.849 0.867 0.910 0.891 0.775 0.918 0.800 0.967
(0.801 – 0.896) (0.786 – 0.948) (0.802 – 1.02) (0.866 – 0.916) (0.721 – 0.829) (0.824 – 1.01) (0.755 – 0.0844) (0.917 – 1.02)
r2
Kfnorm
0.995 0.985 0.974 0.998 0.991 0.990 0.944 0.995
29.7 42.8 73.7 39.4 341 755 221 726
(23.2 – 36.2) (27.7 – 57.8) (37.1 – 110) (36.6 – 42.1) (255 – 428) (446 – 1060) (191 – 251) (574 – 878)
It should be noted that the use of C.L.’s here is not rigorously correct since errors may not be normally distributed in nonlinear models; however, they still may be used for comparative purposes with the understanding that if the parameter is not normally distributed, there may be a less than 95% chance that the C.L. contains the true value. a 95% confidence limits from linear regression. b 95% confidence limits from nonlinear regression.
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
Table 1 Phenanthrene and atrazine batch sorption parameters
167
168
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
sorption is less bimodal as fc increases. Weber and co-workers (Weber and Huang, 1996; Leboeuf and Weber, 1997; Huang and Weber, 1998; Leboeuf and Weber, 2000) have drawn on synthetic polymer chemistry concepts in describing NOM as a rubbery/glassy polymer where sorption in the expanded NOM rubbery domain occurs by partitioning, which is linear with solute concentration; and where sorption in the more condensed NOM glassy domain is nonlinear due to HOC entrapment or hindered diffusion. Similarly, humic substances have also been described as having both condensed and expanded regions (Hayes and Himes, 1986), and humic polymers molecular weights are similar to those of many synthetic polymers. The rubbery/glassy conceptual model has been supported by other work that has suggested that HOC sorption nonlinearity may be attributed to solute retention in the nanometer-sized voids in the glassy region of NOM (Xing and Pignatello, 1998; Xia and Pignatello, 2001). Our observations are consistent with the NOM rubbery/glassy polymer conceptualization: the presence of methanol in NOM increased isotherm linearity as solvents do in synthetic polymers (Cain et al., 1991; Kamiya et al., 1992). Isotherm linearity has also been observed to increase above the posited NOM glass transition temperatures, which are the temperatures at which the glassy NOM regions transit to the rubbery states (Leboeuf and Weber, 1997). For the NOM, which can be expected to have a range of glass transition temperatures, the presence of solvents with solubility parameters similar to NOM would be expected to lower the glass transition temperatures of some of the NOM regions to below ambient temperatures, thus enhancing linearity. The solubility parameter of methanol, 14.5 cal1/2 cm 3/2 (Barton, 1991), is within the range of solubility parameters estimated for NOM (Freeman and Cheung, 1981; Chiou and Kile, 1994); hence, as observed in this study, methanol would be effective at lowering the NOM glass transition temperature and enhancing isotherm linearity. Figs. 4 – 6 contain phenanthrene and atrazine batch sorption isotherms with the equilibrium solution concentration, Ce, normalized for solute solubility. In each case, and in contrast to the Kf measurements discussed above, the mean partition coefficients from the normalized data (Kfnorm) increased with fc (Table 1), indicating higher sorptive uptake for the methanol – water treatments relative to solute solubility. To clarify, it should be noted that in no case did the actual mass of sorbed solute increase with fc, but rather, the mass of solute sorbed relative to solute solubility (which increases with fc) increased with fc. Though the 95% confidence limits of the mean Kfnorm value for the Call’s Creek sediment at fc = 0.10 overlapped the values at fc = 0.00 and at fc = 0.20, the values at fc = 0.00 and at fc = 0.20 were themselves statistically different. These data indicate that methanol is interacting with the NOM, in this case, to increase sorption magnitude relative to solute solubility. The original conceptualization of HOC sorption on NOM in mixed solvent systems did not account for solvent – sorbent effects (Rao et al., 1985), and many subsequent studies were not designed to measure such effects or they were assumed to be too small relative to the other phenomena being measured. However, prior work by this author and others has shown that organic solvents can increase the sorption – desorption kinetics of HOCs on soils and sediments (Brusseau et al., 1991; Bouchard, 1998). This implies that solvents change the sorbent character in some way that facilitates HOC exchange. The increase in magnitude of relative sorption, and of sorption – desorption kinetics may be interpreted using the rubbery/glassy polymer model.
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
169
Fig. 4. Atrazine sorption isotherms for Upper Call’s Creek sediment normalized for solubility at fc = 0.00 and 0.30.
Fig. 5. Phenanthrene sorption isotherms for Upper Call’s Creek sediment normalized for solubility at fc = 0.00 and 0.30.
170
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
Fig. 6. Phenanthrene sorption isotherms for Call’s Creek sediment normalized for solubility at fc = 0.00, 0.10 and 0.20.
Overall, Figs. 4– 6 and Table 1 indicate that relative sorption increased with methanol content; however, there are two phenomena having opposing effects on sorption that may be operative in these systems. First, by dissolving in the NOM polymer, methanol may reduce the glassy structure of NOM regions making these regions act more like a partitioning medium. Since sorption in the glassy regions is posited to be slow, and partitioning is known to be rapid, the end result may be greater relative sorption in the rubbery domain at similar time scales. However, synthetic polymers have lower sorption capacities in the rubbery than in the glassy state; therefore, it would be expected that at equilibrium, NOM glassy regions would actually yield greater relative sorption. Given that relative sorption increased with fc in this study, it is probable that sorption in the batch reactors was not at true thermodynamic equilibrium so that the faster diffusion in the rubbery than in the glassy NOM resulted in greater sorptive uptake in the rubbery regions at similar time scales. The column studies described below were designed to corroborate or refute this premise. 3.3. Column studies The 3H2O BTC was symmetric, and the data were described well by the advective – dispersive local equilibrium solute transport model (Fig. 7). Such symmetry and equilibrium model fit is indicative of hydrodynamic equilibrium during transport; that is, that diffusional mass transfer of the 3H2O solute into and out of the microporous structure of
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
171
Fig. 7. Atrazine breakthrough curves for Upper Call’s Creek sediment at fc = 0.00.
the sediment occurred rapidly enough relative to solute resident times in the column to be at equilibrium with bulk water transport. Fitting the advective –dispersive local equilibrium solute transport model to the 3H2O data yielded a Peclet number of 78.0; this measurement was consistent with the low dispersion expected for well-packed, unaggregated soils, and is comparable to those measured by other researchers using similar column apparatus (Lee et al., 1988; Nkedi-Kizza et al., 1989). The Peclet number was then used in the two-region nonequilibrium model for simulating HOC transport through the column. Figs. 7 and 8 depict atrazine BTCs at fc = 0.00 and 0.30 for the Upper Call’s Creek sediment. In both cases, the two-region model provided a good fit for the experimental data. When used as model input parameters, the linearized batch sorption values, KL, resulted in 95% confidence limits that captured most of the experimental data. The average KL values were 1.96 and 1.03 for atrazine at fc = 0.00 and 0.30, respectively. The k2 value for the best fit two-region simulation for the aqueous atrazine BTC fell within the range of values previously reported for atrazine in sediments having a range of TOC (Bouchard, 1999). In addition, the k2 values increased from 2.25 to 3.82 h 1 with fc, indicating more rapid sorption – desorption kinetics in the methanol –water systems and corroborating one of the conclusions drawn from the batch data. The poor fit of the equilibrium model to the data in Fig. 8 is a further indication that even though the sorption –desorption kinetics are faster at fc = 0.30, they are still not at equilibrium at these time scales. Looking at the batch and column data collectively, it would appear that the increase in relative sorption magnitude with fc may be attributed to the faster sorption kinetics in the
172
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
Fig. 8. Atrazine breakthrough curves for Upper Call’s Creek sediment at fc = 0.30.
methanol – water systems. If NOM acts like a glassy – rubbery polymer, then as the aqueous and methanol – water systems approach true thermodynamic equilibrium, it would be expected that the differences in relative sorption magnitude would decrease as the ‘‘kinetic advantage’’ of the methanol – water systems diminished. At equilibrium, the aqueous systems would display greater relative sorption as they would have a greater proportion of their NOM in the glassy state. These last two statements cannot be supported directly by the data set presented here, but rather provide some direction for future work using mixed solvents to describe the nature of the NOM sorption process.
4. Conclusions The SPARC properties calculator was useful for estimating solute solubility’s in water and in water – methanol solutions at fc < 0.30, which would apply to most environmental scenarios. Results of the batch and column studies appear to support the rubbery/glassy polymer conceptualization of NOM. The observation that isotherm linearity increased with fc provided evidence that methanol was interacting with the NOM, perhaps having an effect similar to that in synthetic polymers where solvents can bring about a transition from the glassy (nonlinear) to the rubbery (linear) state. The glassy and rubbery polymer regions also differ with respect to sorption capacity and kinetics. Synthetic polymers have lower sorption capacities in the rubbery than in the glassy state; therefore, it would be expected that at equilibrium, sorption would be greater in the NOM glassy regions. However, when
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
173
the batch data in this study were normalized for solute solubility, the mean partition coefficients from the normalized data increased with methanol content, indicating higher sorptive uptake for the methanol – water treatments relative to solute solubility. This observation may be explained by the kinetics differences of the glassy and rubbery regions: sorption in the glassy regions is posited to be slow, and partitioning is known to be rapid. It is probable, therefore, that sorption in the batch reactors was not at true thermodynamic equilibrium and that the faster diffusion in the rubbery than in the glassy NOM regions resulted in greater sorptive uptake in the rubbery regions at similar time scales. In total, the data presented here provides further evidence that NOM may behave similarly to the glassy/rubbery regions of synthetic polymers.
References Barton, A.F.M., 1991. CRC Handbook of Solubility Parameters and Other Cohesion Parameters. CRC Press LLC, Boca Raton, FL, USA. Bouchard, D.C., 1998. Sorption kinetics of PAHs in methanol – water systems. J. Contam. Hydrol. 34 (1 – 2), 107 – 120. Bouchard, D.C., 1999. Sorption of vinclozolin and atrazine on four geosorbents. Pestic. Sci. 55, 1095 – 1102. Brusseau, M.L., Wood, A.L., Rao, P.S.C., 1991. Influence of organic cosolvents on the sorption kinetics of hydrophobic organic chemicals. Environ. Sci. Technol. 25 (5), 903 – 910. Cain, E.J., Wen, W.Y., Jones, A.A., Inglefield, P.T., Cauley, B.J., Bendler, J.T., 1991. A dual-mode interpretation of spin relaxation for 13CO2 sorbed in polycarbonate. J. Polym. Sci., Part B: Polym. Phys. 29, 1009 – 1020. Chiou, C.T., Kile, D.E., 1994. Effects of polar and nonpolar groups on the solubility of organic compounds in soil organic matter. Environ. Sci. Technol. 28 (6), 1139 – 1144. Chiou, C.T., Kile, D.E., 1998. Deviations from sorption linearity on soils of polar and nonpolar organic compounds at low relative concentrations. Environ. Sci. Technol. 32 (3), 338 – 343. Chiou, C.T., Porter, P.E., Schedding, D.W., 1983. Partition equilibria of nonionic organic compounds between soil organic matter and water. Environ. Sci. Technol. 17 (4), 227 – 231. Chiou, C.T., Mcgroddy, S.E., Kile, D.E., 1998. Partition characteristics of polycyclic aromatic hydrocarbons on soils and sediments. Environ. Sci. Technol. 32 (2), 264 – 269. Chiou, C.T., Kile, D.E., Rutherford, D.W., Sheng, G., Boyd, S.A., 2000. Sorption of selected organic compounds from water to a peat soil and its humic-acid and humin fractions: potential sources of the sorption nonlinearity. Environ. Sci. Technol. 34 (7), 1254 – 1258. Freeman, D.H., Cheung, L.S., 1981. A gel partition model for organic desorption from a pond sediment. Science 214, 790 – 792. Hayes, M.H.B., Himes, F.L., 1986. In: Huang, P.M., Schnitzer, M. (Eds.), Interaction of Soil Minerals with Natural Organics and Microbes. Soil Science Society of America, Madison, WI. Hilal, S.H., Carreira, L.A., Karickhoff, S.W., 1994. Estimation of chemical reactivity parameters and physical properties of organic molecules using SPARC. In: Politzer, P., Murray, J.S. (Eds.), Quantitative Treatment of Solute/Solvent Interactions. Theoretical and Computational Chemistry, vol. I. Elsevier, New York, pp. 291 – 353. Huang, W., Weber Jr., W.J., 1997. A distributed reactivity model for sorption by soils and sediments: 10. Relationship between desorption, hysteresis, and the chemical characteristics of organic domains. Environ. Sci. Technol. 31 (9), 2562 – 2569. Huang, W., Weber Jr., W.J., 1998. A distributed reactivity model for sorption by soils and sediments: 11. Slow concentration-dependent sorption rates. Environ. Sci. Technol. 32 (22), 3549 – 3555. Kamiya, Y., Bourbon, D., Mizoguchi, K., Naito, Y., 1992. Sorption, dilation, and isothermal glass transition of poly(ethyl methacrylate) – organic gas systems. Polym. J. 24, 443 – 449. Kan, T.K., Fu, G., Tomson, M.B., 1994. Adsorption/desorption hysteresis in organic pollutant and soil/sediment interaction. Environ. Sci. Technol. 28 (5), 859 – 867.
174
D.C. Bouchard / Journal of Contaminant Hydrology 56 (2002) 159–174
Kimble, K.D., Chin, Y.P., 1994. The sorption of polycyclic aromatic hydrocarbons by soils in low-methanol/water mixtures. J. Contam. Hydrol. 17 (1 – 2), 129 – 143. Leboeuf, E.J., Weber Jr., W.J., 1997. A distributed reactivity model for sorption by soils and sediments: 8. Sorbent organic domains: discovery of a humic acid glass transition and an argument for a polymer-based model. Environ. Sci. Technol. 31 (6), 1697 – 1702. Leboeuf, E.J., Weber Jr., W.J., 2000. Macromolecular characteristics of natural organic matter: 2. Sorption and desorption behavior. Environ. Sci. Technol. 34 (17), 3623 – 3631. Lee, L.S., Rao, P.S.C., Brusseau, M.L., Ogwada, R.A., 1988. Nonequilibrium sorption of organic contaminants during flow through columns of aquifer materials. Environ. Toxicol. Chem. 7, 779 – 793. Li, A., Andren, A.W., 1994. Solubility of polychlorinated biphenyls in water/alcohol mixtures: 1. Experimental data. Environ. Sci. Technol. 28 (1), 47 – 52. Li, A., Andren, A.W., Yalkowsky, S.H., 1996. Choosing a cosolvent: solubilization of naphthalene and cosolvent property. Environ. Toxicol. Chem. 15 (12), 2233 – 2239. Nkedi-Kizza, P., Brusseau, M.L., Rao, P.S.C., Hornsby, A.G., 1989. Nonequilibrium sorption during displacement of hydrophobic organic chemicals and 45Ca through soil columns with aqueous and mixed solvents. Environ. Sci. Technol. 23 (7), 814 – 820. Nzengung, V.A., Voudrias, E.A., Nkedi-Kizza, P., Wampler, J.M., Weaver, C.E., 1996. Organic cosolvent effects on the sorption equilibrium of hydrophobic organic chemicals by organoclays. Environ. Sci. Technol. 30 (1), 89 – 96. Piatt, J.J., Backhus, D.A., Capel, P.D., Eisenreich, S.J., 1996. Temperature dependent sorption of naphthalene, phenanthrene, and pyrene to low organic carbon aquifer sediments. Environ. Sci. Technol. 30 (3), 751 – 760. Rao, P.S.C., Hornsby, A.G., Kilcrease, D.P., Nkedi-Kizza, P., 1985. Sorption and transport of hydrophobic organic chemicals in aqueous and mixed solvent systems: model development and preliminary evaluation. J. Environ. Qual. 14 (3), 376 – 383. Rubino, J.T., Yalkowsky, S.H., 1987. Cosolvency and cosolvent polarity. Pharm. Res. 4, 220 – 230. Spurlock, F.C., Biggar, J.W., 1994. Thermodynamics of organic chemical partition in soils: 2. Nonlinear partition of substituted phenylureas from aqueous solution. Environ. Sci. Technol. 28 (6), 996 – 1002. Van Genuchten, M.Th., 1981. Research Report No. 119, US Department of Agriculture, US Salinity Laboratory, Riverside, CA. Weber Jr., W.J., Huang, W., 1996. A distributed reactivity model for sorption by soils and sediments: 4. Intraparticle heterogeneity and phase-distribution relationships under nonequilibrium conditions. Environ. Sci. Technol. 30 (3), 881 – 888. Wood, A.L., Bouchard, D.C., Brusseau, M.L., Rao, P.S.C., 1990. Cosolvent effects on sorption and mobility of organic chemicals in soils. Chemosphere 21 (4 – 5), 575 – 587. Xia, G., Pignatello, J.J., 2001. Detailed sorption isotherms of polar and apolar compounds in a high-organic soil. Environ. Sci. Technol. 35 (1), 84 – 94. Xing, B., Pignatello, J.J., 1998. Competitive sorption between 1,3-dichlorobenzene or 2,4-dichlorophenol and natural aromatic acids in soil organic matter. Environ. Sci. Technol. 32 (5), 614 – 619. Xing, B., Pignatello, J.J., Gigliotti, B., 1996. Competitive sorption between atrazine and other organic compounds in soil and model sorbents. Environ. Sci. Technol. 30 (8), 2432 – 2440. Young, T.M., Weber Jr., W.J., 1995. A distributed reactivity model for sorption by soils and sediments: 3. Effects of diagenetic processes on sorption energetics. Environ. Sci. Technol. 29 (1), 92 – 97.