Environ. Sci. Technol. 2004, 38, 5863-5870
Effects of Competitor and Natural Organic Matter Characteristics on the Equilibrium Sorption of 1,2-Dichlorobenzene in Soil and Shale† DAEYOUNG JU‡ AND T H O M A S M . Y O U N G * ,‡,§ Agricultural and Environmental Chemistry Graduate Group and Department of Civil and Environmental Engineering, University of California, Davis, California 95616
The competitive sorption behaviors of 1,2-DCB in binary solute systems in four natural sorbents having natural organic matter (NOM) matrixes of different physicochemical characters were investigated in batch reactors. Specifically, the study focused on investigating how the extent of 1,2DCB competitive sorption depends on (i) the rigidity of NOM matrixes as assessed by the efficiency of chemical oxidation and (ii) the closeness of competitor structure to that of the primary solute. The chemical oxidation and elemental composition results suggest that the shale NOM is the most reduced and condensed, the peat was the most oxidized and amorphous, and two surface soils had intermediate NOM structures. Four chlorinated benzenes and phenanthrene were used as competing solutes. All five chemicals exhibited competition against 1,2-DCB in all sorbents, including the peat, but the extent of competition varied significantly. Little difference in the extent of competition with 1,2-DCB was observed for the various chlorinated benzenes even though some were liquids and some were solids at the experimental temperature. All of the chlorobenzenes were more effective competitors than phenanthrene. The shale showed markedly different competition features from the other sorbents, with a much smaller competitive effect at a given sorbed volume of competitor. However, normalizing sorbed competitor volumes by the capacity of the adsorption domain in the PolanyiManes single-solute partition-adsorption model (VO) produced qualitatively similar competitive behavior for each solute; displacement of 1,2-DCB increased with increasing sorbed competitor volumes up to VO, and little additional competition occurred beyond that point. The extent of competition was positively correlated with the maximum adsorption capacity and the fraction of “hard” and “soot” carbon contents as assessed by chemical and thermal oxidation methods. These findings indicate that competition is associated with voids in the NOM structure, that these voids are likely present within the condensed (“hard” plus “soot”) carbon domain, and therefore that diagenetic alteration of NOM plays a central role in †
This paper is part of the Walter J. Weber Jr. tribute issue. * Corresponding author phone: (530)754-9399: fax: (530)7527872: e-mail:
[email protected]. ‡ Agricultural and Environmental Chemistry Graduate Group. § Department of Civil and Environmental Engineering. 10.1021/es049668u CCC: $27.50 Published on Web 10/12/2004
2004 American Chemical Society
determining competitive sorption characteristics for hydrophobic contaminants.
Introduction The sorption/desorption behaviors of hydrophobic organic compounds (HOCs) in multi-solute systems can differ substantially from those in a single-solute system according to recent research reports (1-7). Specifically, the presence of a competing solute has been shown to reduce the equilibrium sorption of a primary solute of similar structure and/or speed its desorption, thereby enhancing its mobility, bioavailability, and toxicity (8-10). These findings directly contradict the independence of hydrophobic solute uptake that is presumed under the partitioning model of HOC sorption on soils and which has been demonstrated in a number of soil/solute systems (5, 11, 12). Single-component sorption data have not always proven useful for predicting pollutant transport rates in systems where multiple organic contaminants are present (9). The competitive sorption and displacement of HOCs in soils and sediments could result in more serious problems to human health and terrestrial and aquatic ecosystems than would be predicted with noncompetitive models. The enhanced leachability of HOCs in mixtures is likely to make the contaminated area more extensive and augment toxic effect of HOCs to aquatic and benthic organisms by enhancing the rate and extent of contaminant release (13). Efforts to classify soils exhibiting rate-limited release of contaminants as having reached an “environmentally acceptable end point” are also complicated by the possible displacement of such compounds by passage of future contaminant plumes. Developing a remediation strategy under such circumstances would consequently be more complicated, and cleanup targets would be more difficult to establish. Organic contaminants are almost always released to the environment as solute mixtures (e.g., underground fuel leaks, industrial hazardous waste releases, pesticide formulations). Therefore, it is critically important to understand the altered transport properties caused by competitive sorption and displacement phenomena and the underlying mechanisms of the processes to accurately predict the mobility of contaminants and their toxic effect on organisms. Nevertheless, the vast majority of previous studies investigates sorption and desorption processes under single contaminant conditions. Our understanding of factors controlling the competitive interaction among HOCs in geosorbents is still limited. The goal of this study was to systematically investigate the competition between 1,2-dichlorobenzene (1,2-DCB) and other chemicals in binary systems over a wide range of concentration in four natural geosorbents having natural organic matter (NOM) matrixes of different physicochemical characters. Specifically, the study sought to determine how the extent of 1,2-DCB competitive sorption depends on the nature of the soils, particularly in terms of whether their associated NOM matrixes are condensed as determined by chemical oxidation and elemental analysis, and the similarity of competitor and primary solute characteristics. The hypotheses underlying the study design were that competition would be most significant when the sorbents included a significant fraction of condensed NOM and when the competitor was similar in size and structure to the primary solute. Although previous studies have documented competitive sorption effects for solutes and soils of differing structures, the results presented here add to the underVOL. 38, NO. 22, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5863
prepared in three ways: untreated, chemically oxidized, and thermally oxidized. Carbon and nitrogen contents were determined for each prepared sample following acid treatment (10% HCl) to remove inorganic carbon. Hydrogen contents were measured (Huffman Labs, Golden, CO), and oxygen contents were calculated by the difference between organic matter (OM) content and carbon, hydrogen, and nitrogen contents under the assumption that the OM consists solely of carbon, oxygen, hydrogen, and nitrogen. OM contents were determined by loss-on-ignition at 430 °C in a muffle furnace for 24 h. Thermal oxidation was performed for ∼100 mg subsamples of individual sorbents in a muffle furnace under excess air for 24 h at 375 °C to measure soot carbon content (17). Chemical oxidation employed K2S2O8 because it is believed to selectively remove the easily oxidized organic matter (1, 18-20). Soil samples were mixed with K2S2O8 to organic matter at a ratio of 12 g/g, resulting in an aqueous persulfate concentration of 0.0357 g/mL (21), and the sample was heated in an autoclave at 121 °C for 2 h. Solids were washed and separated by conducting three cycles of centrifuge, decant, and refill. The fractional distribution among the soft, hard, and soot carbon domains was calculated based on differences in organic carbon contents before and after oxidation for each sorbent. To do so, the following equations were developed:
TABLE 1. Selected Physicochemical Properties of the Sorbates at 25 °C (14) name
MW
CB 1,2-DCB 1,4- DCB 1,2,4- TCB 1,2,3,4- TeCB PHN
112.6 147.0 147.0 181.5 215.90 178.22
molar vol solubility (mg/L) log (cm3/mol) (subcooled Csat) (Kow) 101.68 112.66 117.84 124.78 127.00 167.66
502.8 92.8 59.9 (114.1) 40.6 8.21 (13.6) 1.12 (6.18)
Tm (°C)
2.92 -45.6 3.38 -17.0 3.38 53.1 4.00 16.9 4.55 47.5 4.57 99.5
standing of competitive sorption in soils in several ways. Unique features of this work include: (i) the experimental matrix of five competitors and four soils is larger than in previous studies and (ii) both primary and competitor concentrations were monitored allowing total sorbed phase concentrations to be determined experimentally; this is shown below to help normalize competitive effects.
Experimental Section Sorbates and Sorbents. Analytical standard grade 1,2-DCB (Pestanal; >99.9%; Sigma-Aldrich, Inc.) was used as the primary solute. Four chlorinated benzenes [chlorobenzene (CB), 1,4-dichlorobenzene (1,4-DCB), 1,2,4-trichlorobenzene (1,2,4-TCB), and 1,2,3,4-tetrachlorobenzene (1,2,3,4-TeCB)] and a polycyclic aromatic hydrocarbon [phenanthrene (PHN)] were used as competing solutes against 1,2-DCB in this study. All of the competitors were analytical standard grade from Sigma-Aldrich and were used as received. Some important characteristics of the sorbates investigated are summarized in Table 1. Four natural sorbents (Yolo and Forbes soils, Ohio shale, and Pahokee peat) known to have great variation in many of their important physical and chemical properties were selected (Table 2). Yolo and Forbes soils were collected below the litter layer at an agricultural site in Yolo County, California, and at a forested site in Placer County, California, in the Sierra Nevada Foothills, respectively (15). Pahokee peat is a reference sample of the International Humic Substances Society (IHSS), collected from Everglades, FL. Ohio shale is a sedimentary rock sampled in northwestern Ohio (16). Prior to experiments, Yolo and Forbes soil and Pahokee peat were carefully air-dried, passed through a 2 mm sieve, and split into suitable subsamples using a riffle-splitter. Ohio shale had been crushed and passed through a no. 60 (250 µm) sieve before the study and was used without any further processing. The sorbents were stored at 4 °C under dark conditions. Carbon and nitrogen contents were determined using high-temperature dry combustion at 1020 °C (Carlo Erba NA1500, CE Elantech, Inc., Lakewood, NJ). Each sorbent was
soot carbon (%) ) OCtherm_ox (%) × (100 - OM (%)) (1) OCtherm_ox 100 × OM (%) OCuntrt
(
)
hard carbon (%) ) OCchem_ox (%) × (100 - OM (%)) - soot carbon (%) (2) OCchem_ox 100 × OM (%) OCuntrt
(
)
soft carbon (%) ) OCuntrt (%) - hard carbon (%) soot carbon (%) (3) where OCtherm_ox is OC content (%) measured after thermal combustion at 375 °C for 24 h; OCchem_ox is OC content (%) measured after persulfate oxidation; and OM and OCuntrt are the organic matter and organic carbon contents (%) of the untreated sorbents, respectively. The second term on the right side in eqs 1 and 2 compensates for the changes in sorbent mass due to the loss of organic matter during the thermal and chemical oxidation processes. It is assumed that the ratio between OM and OC in the denominator of the second term remains constant before and after oxidation processes for a given sorbent.
TABLE 2. Characteristics of the Sorbentsa organic carbon (%) sorbent Forbes soil Ohio shale Pahokee peat Yolo soil
organic matter (%) 13.9 (13.5)c 5.3 (5.68)a 89.88 (93.1)b (86.4)d 3.8 (4.06)c
total OC
hard OC
soot C
domain distribution (soft -hard soot carbon ratio)
H/O
5.52
3.78
0.03
33.2-66.3-0.5
3.06
2.49 (2.44)a 51.8 (44.6)b (57.1)d 1.08
2.26
0.05
9.2-89.0-1.73
5.18
1.21
49.8-50.0-0.24
1.89
0.01
39.3-60.0-0.64
3.77
55.8 0.65
a Values in parentheses represent previously reported measurements for the same materials from the following: a, ref 26; b, ref 27; c, ref 28; and d, ref 29. All data are reported on a moisture-free basis.
5864
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 22, 2004
The definitions of soot, hard, and soft carbon fractions (eqs 1-3) are highly empirical and suffer from a number of limitations that have been identified in recent studies (2224). In particular, differences among black carbon contents of up to 2 orders of magnitude have been identified for eight soils using six different techniques (23), and the thermal combustion method used in this work was found to provide highly variable results for wood chars, which are likely to constitute a significant fraction of the black carbon in many soil samples (24). Despite the ambiguity that remains in black carbon concentration measurements in soils, the operational measures of carbon “hardness” used in this work have proven to be useful correlating variables for a number of sorption/ desorption descriptors, and their use here is intended to provide only qualitative indications of carbon rigidity. Single Solute Sorption Experiments. Single solute sorption isotherms of five chlorinated benzenes (CB, 1,2-DCB, 1,4-DCB, 1,2,4-TCB, and 1,2,3,4-TeCB) in the four geosorbents were measured over wide concentration ranges with a 30-d contact time using 40-mL glass tubes with silver-foil lined (25) Teflon caps at room temperature (23 ( 2 °C). The background solution contained CaCl2 (5 × 10-3M) as an electrolyte and 200 mg/L NaN3 as an inhibitor for microbial decomposition. Preliminary experiments indicated that sorbent-water distribution coefficients were approximately constant after the 30-d contact time. Stock solutions of each sorbate were prepared by dissolving a predetermined mass of sorbate in methanol. Specified amounts of stock solution were delivered into the aqueous phase of each reactor through the Teflon septa using a microsyringe. The cap was quickly replaced. Methanol volume was controlled so that it did not exceed 0.1% v/v of the total aqueous phase to minimize cosolvent effects. Initial concentration ranges for each solute were determined based on aqueous solubility (from 0.08% up to 15-20% of solubility) to enable more appropriate comparison of the sorption isotherm model parameters. The soil-to-water ratio was selected to attain 20-80% solute uptake by the soils and the peat and 15-90% by the shale. Control reactors containing no sorbent were prepared in triplicate at each concentration for all experiments to evaluate solute loss to reactor components. Average solute losses in control reactors was less than 6%. All reactors were filled until no headspace remained and tumbled end-over-end at 10 rpm. After the 30-d contact time, the reactors were centrifuged at 3000 rpm for 30 min. Predetermined volumes of supernatant were removed and extracted with hexane. The volumes for supernatant and hexane were determined depending on the anticipated range of solute concentration. Quantification of solute concentrations in the aqueous phase was performed using a GC-ECD (Agilent 6890 gas chromatograph with a micro-cell electron capture detector and a G2613 autosampler) for 1,2-DCB, 1,4DCB, 1,2,4-TCB, and 1,2,3,4-TeCB and a GC-MS (Agilent 6890 GC interfaced to a 5973 network mass selective detector) for CB. DB-VRX (30.0 m × 0.45 mm × 2.55 µm) and DB-624 (30.0 m × 0.25 mm × 1.4 µm) capillary columns (J&W Scientific, Folsom, CA) were used for GC-ECD and GC-MS, respectively. The adsorbed sorbate mass was calculated by difference between the total solute mass added and the mass remaining in the solution phase at equilibrium. The average solute mass remaining in corresponding no-soil control tubes after the specified contact time was used as an initial aqueous loading, assuming that solute loss in each reactor was proportional to the final solute concentration (30). The Freundlich isotherm model and the adsorptionpartitioning model based on the Polanyi-Manes theory (P-M model) were used to fit the experimental data. The logtransformed Freundlich model takes the following form:
log(qe) ) log(KF) + n log(Ce)
(4)
where qe is the solid-phase concentration (µmol/kg OC), Ce is the aqueous-phase concentration (µmol/L), KF is the Freundlich coefficient, and n is the Freundlich exponent. Polanyi-Manes theory has been applied to describe HOC sorption on various sorbents such as activated carbon, peat, and soil (7, 31-35). Incorporating a partitioning term into the P-M isotherm model, uptake of the sorbates investigated are simulated using the following model (34):
{ ( )}
qe ) KPCe + VOF exp R
sw Vm
β
(5)
where KP is the partition coefficient (L/kg OC), VO is the maximum adsorption capacity (cm3/kg OC), F is the density of sorbate, R and β are fitting parameters, Vm is the molar volume of the solute (L/mol), and sw is the solute adsorption potential () RT ln(Sw/Ce)) where Sw is the solute’s aqueous solubility (subcooled solubility in the case of compounds that are solids at the temperature of interest). To decrease the possibility of multiple solutions, we reduced the number of parameters to three by setting β values in eq 5 to 2.0 as used by Kleineidam et al. (32) in their study. This restriction produced more consistent VO values and little change in model error percent (36). The fitting parameters for the two modelssn and KF for the Freundlich model and KP, VO, and R for the P-M models were selected by minimizing the mean relative squared percent error. The error is calculated using the following equation, and the optimization processes for fitting parameters were conducted using the solver function of Microsoft Excel:
[
N
∑(q
error (%) ) 100 ×
]
0.5
o
- qm)2/qo2
i)1
(N - ν)
(6)
where N is the number of sample points, ν is the number of fitting parameters, and qo and qm are the observed and modeled solute concentrations in the solid phase. 1,2-DCB Sorption in Binary Solute Systems. The initial aqueous phase concentration of 1,2-DCB, the primary solute, was fixed at 0.1 mg/L. This is assumed to be low enough not to cause severe conformational change to NOM matrixes (i.e., below the glass transition concentration; 37). Predetermined amounts of 1,2-DCB and a competitor were dissolved in methanol. Seven different stock solutions containing a constant concentration of 1,2-DCB and various concentrations of a competitor up to 15∼20% of its solubility were prepared and spiked into each reactor simultaneously. All reactors were treated following the same procedures as the single-solute isotherm experiments. After 30 d of contact time, both 1,2-DCB and competitor concentrations in the aqueous phase were measured after hexane extraction using the analytical methods described above.
Results and Discussion Single-Solute Isotherms. 1,2-DCB sorption isotherm data in four natural geosorbents with a 30-d contact time were modeled using the Freundlich (eq 4) and the P-M (eq 5) models. The Freundlich isotherm model results are summarized in Table 3. All four sorption isotherms deviated significantly (p < 0.01) from linearity, showing Freundlich n values from 0.64 to 0.80. The sorption isotherms in the peat and Yolo soil are relatively closer to linearity (n ) 1) than those for Forbes soil and Ohio shale. Ohio shale showed more than 2 orders of magnitude greater organic carbon VOL. 38, NO. 22, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5865
TABLE 3. Single-Component Freundlich Isotherm Parameters for 1,2-DCB
soil
no. of data
log KF (µmol/kg OC) (µmol/L)-n ( 95% confidence level
n ( 95% confidence level
R2
average error (%)
Yolo Forbes Ohio shale Pahokee peat
69 58 58 53
2.914 ( 0.010 3.273 ( 0.011 4.562 ( 0.010 2.990 ( 0.011
0.796 ( 0.012 0.745 ( 0.012 0.639 ( 0.012 0.797 ( 0.012
0.996 0.996 0.995 0.997
6.6 7.1 6.2 6.9
FIGURE 1. Sorption isotherms for 1,2-DCB in the sorbents. Data were fitted using the partition-adsorption model based on PolanyiManes theory (eq 5). normalized Freundlich capacity factors and the greatest degree of isotherm nonlinearity consistent with previous results for other solutes on this material (2, 20). The sorption data were also well-described by the P-M model (Figure 1). The abscissa in Figure 1 is the volume of sorbate in the solid phase at equilibrium per mass of organic carbon. The solute concentration in the solid phase covers almost 3 orders of magnitude difference in Yolo, Forbes, and the peat and covers 2 orders of magnitude in the shale. Where the volume of sorbed solute is 1.0 cm3/kg OC, organic carbon normalized distribution coefficient values (KOC ) qe/(Ce foc); foc is the fraction of organic carbon in sorbent) for 1,2-DCB are approximately 70 000 (shale), 1000 (Forbes), 500 (peat), and 400 (Yolo); the more than 2 orders of magnitude greater 1,2-DCB KOC in the shale suggests that its NOM is structurally quite different from the soil-derived ones. Excluding the shale, it is notable that Forbes soil showed much stronger sorption affinity as compared with another soil and the peat. Its affinity was approximately 2 or 3 times greater depending on solute concentration in the solid phase, even after being normalized by OC content. Binary Solute Isotherms. The most common way to graphically display competitive sorption behavior has been to plot the isotherm of the primary solute in the presence and in the absence of a fixed initial (typically high) concentration of the competing solute (2, 5, 6). Although our experiments were not designed for this purpose for reasons discussed below, we first analyzed our data by examining the changes to the isotherms of the other chlorobenzenes introduced by the presence of 0.1 mg/L initial concentration of 1,2-DCB (Figure S-1 in the Supporting Information). Although the 1,2-DCB concentration was low, all 16 of the chlorobenzene isotherms became more linear (larger Freundlich n values), and 12 of the 16 displayed increases in the Freundlich capacity factor (KF). These changes are generally subtle and not statistically significant (p > 0.05), but their consistency of direction suggests they are real. For comparison McGinley et al. (2) observed competition when 5866
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 22, 2004
competitor concentrations were 5-15 mg/L, and Xia and Ball observed competition at competitor concentrations of 0.005-25 mg/L (7). In some cases competitor concentrations have exceeded 100 mg/L (5, 6). The generally observed nonlinearity of isotherms for soils and sediments in which n < 1 (38) should result in reduced KOC values with increasing solid-phase solute concentration as observed in Figure 1. Showing the reduction in primary solute KOC with increasing competitor concentration, as is typically done, implies a “displacement” of primary solute when the addition of primary solute instead of competitor might produce a KOC depression of similar magnitude. Competitive sorption in soils should therefore be assessed by comparing the KOC lowering potential of a particular quantity of a competitor to that caused by the addition of an equal amount of the primary solute to the solid phase. This demands that the equilibrium concentration of both the primary and secondary solutes be measured, which has not typically been done in competitive sorption studies with soils and sediments. The effect of secondary solutes at various initial aqueous loadings on the sorption of a primary solute, 1,2-DCB, with fixed initial concentration by four different sorbents was investigated (Figure 2). We used the volume of sorbate (primary solute plus competitor) in the solid phase (cm3/kg OC) at equilibrium as a measure of the total solute uptake so that single-solute (1,2-DCB) and bisolute (1,2-DCB and competitor) data could be displayed on the same graph. On the basis of the potential theory, the same volume of adsorbates is expected to sorb for all liquid HOCs where the adsorption potential densities are the same (i.e., at the same fraction of their aqueous solubilities; 34). As expected, KOC reductions were small at very low cosolute concentration but as the volumes of sorbed competitor increased in the solid phase, more substantial impacts were observed. 1,2-DCB KOC values were extensively lowered in some cases (up to 89% in the shale with the presence of 1,2,4-TCB at its highest aqueous loading), but the extent of KOC reduction significantly varied among the sorbents. Influence of Sorbate Molecule Structure on the Degree of Competition. In almost every case, the largest reduction in KOC for a given volume increment of sorbed chemical occurred for the “1,2-DCB-only” treatments. KOC reductions of almost 2 orders of magnitude occur as additional 1,2-DCB is added to the reactors, and this curve asymptotically approaches the organic carbon normalized linear partition coefficient (KP) from Table 4 (indicated by a dashed line in Figure 2). This “self-competitive” behavior is presumably caused by the occupation of successively less energetically favorable sorption sites by each additional increment of sorbed chemical, thereby reducing the average distribution coefficient (20). After a sufficient amount of solute has sorbed, the linear partitioning portion of the solute’s uptake will come to dominate the overall sorption process. We expect that if the structural properties of competitors such as size, polarity, and hydrophobicity are close to those of the primary solute, then the composite profile of 1,2-DCB and individual competitors will be nearly coincident with the data profile for 1,2-DCB alone.
FIGURE 2. Effect of the presence of varying amounts of competing solute on 1,2-DCB KOC in four different sorbents.
TABLE 4. Fitting Parameters for the Polanyi-Manes Model for 1,2-DCB Single-Solute Isotherms Where β Is Fixed at 2.0 soil
KP (L/kg OC)
VO (cm3/kg OC)
r
error (%)
Yolo Forbes Ohio shale Pahokee peat
283 615 5898 432
0.50 0.80 20.3 0.32
-1.8E-03 -1.5E-03 -1.5E-03 -1.5E-03
5.9 7.1 5.2 7.9
All five of the competitors reduced the KOC value for 1,2DCB significantly in all sorbents including the peat, but in many cases they did not reduce KOC as much as an equal volume of 1,2-DCB (Figure 2). Significant differences were not generally observed among the chlorinated benzenes in their competitive efficiencies. In a few cases the solutes that are liquids at the experimental temperature (CB and 1,2,4TCB) were more effective competitors than the solid solutes (1,4-DCB, 1,2,3,4-TeCB). The absence of clear and systematic differences between the competitive efficiency of liquid and solid chlorinated benzenes contrasts with the results of Xia and Ball, who attributed the differences to the less efficient packing of solids within the sorbent pore structure (7). The fact that 1,2-DCB KOC data with phenanthrene as a competitor are far above the 1,2-DCB only line suggests that the sorption mechanism(s) of phenanthrene to NOM matrixes associated with the sorbents may differ from those of 1,2-DCB or other chlorinated benzenes. We expected 1,4-DCB to show relatively stronger competing ability than other cosolutes, because its molecular size and hydrophobicity are almost the same as that of the primary solute (1,2-DCB). Structurally similar molecules are reported to compete more strongly with each other (27, 39).
This was attributed to the greater overlap in the affinity of the two sorbates for the adsorption site where the competitive sorption takes place. Strong competition between atrazine and its analogues and between trichloroethylene (TCE) and halogenated hydrocarbons, but weak competition between atrazine and TCE has been reported (5). However, 1,4-DCB does not look to be a relatively strong competitor against 1,2-DCB in the sorbents tested in this study, possibly because of the reduced packing efficiency of solids within the sorbent pore structure noted above. Influence of NOM Nature on the Degree of Competition. The influence of NOM properties on the competitive sorption behaviors of 1,2-DCB was investigated by employing four natural geosorbents with NOM matrixes expected to have significant differences in structure and composition. Competitive sorption has been attributed to the condensed NOM domain in geosorbents (1, 3, 5). For example, the removal of a large fraction of hard and aromatic moieties from sorbents suppressed the competitive sorption of phenanthrene in the presence of pyrene (40). In this study, the condensed nature of the sorbents’ NOM was evaluated based on (i) the fraction of hard carbon domain and (ii) atomic H/O ratios. As summarized in Table 2, the fractional distribution among the operationally defined soft, hard, and soot carbon domains for NOM in each sorbent varied widely. Ohio shale has a distinctly different distribution pattern, with most (91%) of its OC comprised by either hard (89.0%) or soot carbon (1.7%), the highest fractions for any of the sorbents. Forbes soil contains a greater fraction of hard carbon (66%) than the other modern NOM matrixes, Yolo soil, and Pahokee peat. The atomic H/O ratio has been used to assess the sorbent’s oxidation state (16, 41-44) and thus its age because the VOL. 38, NO. 22, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5867
FIGURE 3. Reduction in KOC for 1,2-DCB in the presence of varying amounts of chlorobenzene. (a) Comparison of competitor effects in four natural materials, and (b) data from (a) normalized by P-M pore volume, VO. relative amount of oxygen containing functional groups tends to gradually decrease over very long diagenetic time scales since most long-term carbon storage occurs in reducing environments. Lower values of H/O generally indicate a greater degree of NOM oxidation. The H/O ratios increase in the order Pahokee peat < Forbes soil < Yolo soil < Ohio shale. The analysis of elemental composition of the sorbents qualitatively suggests that the shale is in the most reduced state, the peat is the most oxidized, and Yolo and Forbes soils are at intermediate redox states, consistent with the expected ages of NOM in shale, soil, and peat samples. Combining the results from the elemental analysis and the hard carbon content measurement, we infer that the NOM structure of the shale is the most condensed probably with well-developed internal micropores, the peat contains the most amorphous NOM structure, while Yolo and Forbes soils have intermediate degrees of condensed NOM character. The differences in competitive effects among the sorbents when CB was used as a competitor are shown in Figure 3; similar figures for the other competitors are included in the Supporting Information (Figure S-2). The effects were depicted based on the reduction in 1,2-DCB KOC values in cosolute systems relative to those values without CB. Each point is an average of triplicate experimental data, and the error bars represent standard deviations. The first data point at the lowest solute concentration on the abscissa of Figure 2, measured in the absence of competitor, was set to a standard point (100%) to normalize competitive effects. Competitive effects decrease exponentially as the concentrations of cosolutes in the solid phase increase (5). The marginal decrease of 1,2-DCB KOC values per unit volume of 5868
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 22, 2004
competitor sorbed dropped noticeably when the sorbed amount of cosolute exceeded a certain range (approximately 1 for Yolo and the peat, 2 for Forbes and 40 for the shale in the unit of cm3 solute sorbed/kg OC). This finding has important implications in interpreting competitive sorption mechanisms and designing multicomponent sorption experiments for HOCs in geosorbents. Reviewing previous papers that reported no competition in HOC sorption in multiple-solute systems, one of the common traits is that the initial concentrations of competitors were usually set at extremely high levels (11, 12, 40). Choosing very high levels of initial concentration for competitors (and primary solutes) may result in finding no competitive sorption. In Figure 3a, the shale shows markedly different competition features as compared with the other sorbents. The reduction in 1,2-DCB KOC for the shale occurs at much higher sorbed species volumes, indicating a much smaller degree of competition at a particular adsorbate volume (potential). In the shale, no competitive effect was observed until sorbed solute volume exceeded approximately 2.0 cm3/kg OC, while 1,2-DCB KOC values for other sorbents decreased by 50-60% at that point. The competitive effect is in the order of Yolo ≈ peat > Forbes . Ohio shale for a given sorbate volume. The fact that the degree of competition appeared to be the weakest in Ohio shale may be explained by its much greater maximum adsorption capacity (VO), which exceeds those for the other sorbents by at least an order of magnitude. A high VO value suggests that a greater volume of micropores or number of adsorption sites exists in the NOM of the shale. In Figure 3b, the values on the abscissa in graph a are divided by the maximum adsorption capacities (VO) of the corresponding sorbents from Table 4. Differences in the amount of competing solute sorbed before the 1,2-DCB KOC reaches a plateau appear to be adequately normalized by VO. Sorption of 1,2-DCB is not apparently affected in a competitive way by the presence of additional CB beyond approximately 150-200% of the sorbent’s VO. This finding is consistent with the idea that sorption at high concentrations takes place primarily by partitioning, because almost all adsorption sites in the condensed domain of NOM are saturated beyond VO. Additional sorbate must then sorb in the amorphous domain. When the differences in competitive effects by sorbent are compared after normalization by VO, much stronger competition was observed in the shale than in the other sorbents. The order of the extent of competition is Ohio shale > Forbes > Yolo > Pahokee peat. The 1,2-DCB KOC decreases to 20% of its original value at the highest concentrations studied in the shale, showing an 80% competitive effect. A strong positive correlation (0.92 < R 2 < 0.99) between the observed fractional KOC reduction and the fraction of hard plus soot carbon of the sorbents is consistent with previous hypotheses that the degree of competition is related to the amount of the condensed organic domain (Figure 4). This finding also supports the notion that the extent of diagenetic alteration of NOM plays a central role in determining competitive sorption characteristics for hydrophobic contaminants by giving NOM a structure similar to that of glassy polymers. Competitive Sorption Mechanisms. None of the competitors tested here, chlorinated benzenes and phenanthrene, have strongly polar or ionizable functional groups, although some of the molecules have substantial permanent dipole moments (CB, 1,2-DCB). These sorbates are not likely to interact specifically with the functional groups of NOM components present in the sorbents (39). Therefore, the mechanism of competition is likely to be a nonspecific one. The following two mechanisms may contribute, either individually or collectively, to competition between 1,2-DCB and the competitors investigated in this study: (a) overlap-
FIGURE 4. Relationship between the competitive effect and the hard plus soot carbon fraction of corresponding sorbents. The competitive effect of each sorbent was expressed as the fractional 1,2-DCB KOC reduction when the sorbed volume of 1,2-DCB and competitor was twice the average V0 value for the sorbent. ping preference for sorption locations and/or (b) lowered sorption affinity of NOM matrixes as the consequence of sorbate-sorbent interaction. The generality of nonlinear sorption isotherms for organic chemicals in soils (38) suggests that sorption locations are energetically heterogeneous, and there is a fixed population of locations with any particular sorption energy (20). Several lines of evidence suggest that the most energetically favorable sorption locations may be located within relatively hydrophobic pore structure of condensed NOM phases. Evidence includes the success of the P-M model in fitting the sorption data (Figure 1), the convergence of the KOC reduction curves for chlorinated benzenes when plotted on a total sorbate volume basis (Figure 2), the success of the fitted value of VO in normalizing competitive effects (Figure 3), and the correlations among VO, condensed carbon fractions, and the strength of competition (Figure 4). In this view competition results when the competitor occupies a portion of the most favorable sorption locations displacing the primary solute to either less favorable sorption locations or into solution. Solute pairs with the greatest structural similarity (e.g., those that can form an ideal solution) are expected to exhibit the greatest competition because their interchangeability within the pore structure exhibits the greatest overlap (27). This view presumes that the pore structure within the condensed NOM matrixes, where competitive sorption is believed to occur, remains largely intact during sorption and desorption processes and that solute-solute interactions drive the process. A second possibility is that the competitive effect in soils and sediments might be the consequence of decreased sorption affinity for the primary solute caused by the change of macromolecular structure of the NOM matrixes following uptake of the competitor. Several recent research papers have invoked such “solute-induced plasticization” to explain a variety of sorption phenomena in humic substances, kerogens, synthetic glassy polymers, and soils and sediments (26, 44-50). Evidence for the plasticization hypothesis is largely circumstantial, but the hypothesis is consistent with a variety of sorption related observations. The sorption isotherms of
CB and 1,2,4-TCB in two organic-rich soils were found to shift upward with greater nonlinearity after brief exposure of the sorbents to high concentrations of dichloromethane or benzene (48). Lu and Pignatello (48) postulated that this conditioning effect occurs due to the expansion of nanovoids in the condensed domain of NOM matrixes by a conditioning agent. LeBoeuf and Weber revealed that NOM manifests transitions from glassy states to rubbery states at elevated temperatures as observed for synthetic organic polymers; the existence of a glass transition temperature implies the existence of a glass transition concentration (37, 47). Huang and Weber (45) postulated that faster rates of uptake occur at higher solute concentrations because of reconfiguration of the condensed NOM matrixes upon solute sorption. Weber and Young (26) showed that NOM swelling provided a consistent explanation for changing sorption equilibria in supercritical fluid systems. Xia and Pignatello (33) showed that Freundlich n values are highly dependent on the solute concentration range over which isotherm data is collected and suggested that maxima in the Freundlich n value (minimum nonlinearity) indicated the onset of plasticization, which in Pahokee peat occurred at a Ce/Sw value of around 0.002. On the basis of their examination of several solute classes, chlorinated benzenes were categorized among the solutes having the greatest plasticizing ability. The combined evidence supporting the plasticization hypothesis suggests a conceptual model of NOM in which the “voids” or micropores are more dynamic, flexible, and expandable than in the case of activated carbon or charcoal. Plasticization increases the amorphous fraction of NOM at the expense of the condensed domain. Because the sorption affinity of HOCs is lower in the amorphous domain of NOM than in the condensed domain, this change will lower the distribution coefficient of a primary solute in the presence of a cosolute, producing an apparent competitive effect. The plasticization hypothesis implies that the degree of competition will depend on the relative plasticizing ability of competing species, the ease of plasticization of a given sorbent, and the sorbent’s initial condensed carbon fraction. Competitive effects will be largest for a competitor with good plasticizing ability and a sorbent with a significant fraction of condensed NOM that has a low resistance to plasticization (i.e., a low glass transition concentration). There is presently no clear method for assessing the ease of plasticizing particular soils and sediments beyond direct measurements with a range of plasticizing agents and primary solutes; such measurements do not clearly link the behavior to NOM structure. Further investigation of structural features of soils and sediments that promote or resist swelling and more direct observation of the phenomenon are required before the relative importance of the two causes of competitive sorption behavior can be elucidated.
Acknowledgments This publication was made possible by Grant 5 P42 ES04699 from the National Institute of Environmental Health Sciences, NIH. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the NIEHS, NIH. Funding was also provided as part of National Science Foundation Grant BES 9733621. This study was partly supported by the Ecotoxicology Lead Campus Program of the University of California Toxic Substances Research & Teaching Program and the Jastro Shields Summer Fellowship. Peter G. Green provided valuable assistance with analytical method development, and Anna Hepler assisted with the sorption experiments.
Supporting Information Available Two figures. This material is available free of charge via the Internet at http://pubs.acs.org. VOL. 38, NO. 22, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5869
Literature Cited (1) Li, J.; Werth, C. J. Environ. Sci. Technol. 2001, 35, 568-574. (2) McGinley, P. M.; Katz, L. E.; Weber, W. J. Environ. Sci. Technol. 1993, 27, 1524-1531. (3) McGinley, P.; Katz, L.; Weber, W. J. Water Resour. Res. 1996, 32, 3571-3577. (4) White, J.; Pignatello, J. Environ. Sci. Technol. 1999, 33, 42924298. (5) Xing, B.; Pignatello, J. J.; Gigliotti, B. Environ. Sci. Technol. 1996, 30, 2432-2440. (6) Xing, B.; Pignatello, J. J. Environ. Sci. Technol. 1998, 32, 614619. (7) Xia, G.; Ball, W. Environ. Sci. Technol. 2000, 34, 1246-1253. (8) White, J. C.; Hunter, M.; Pignatello, J. J.; Alexander, M. Environ. Toxicol. Chem. 1999, 18, 1728-1732. (9) Stuart, B. J.; Bowlen, G. F.; Kosson, D. S. Environ. Prog. 1991, 10, 104-109. (10) White, J. C.; Alexander, M.; Pignatello, J. Environ. Toxicol. Chem. 1999, 18, 182-187. (11) Chiou, C. T.; Porter, P. E.; Schmedding, D. W. Environ. Sci. Technol. 1983, 17, 227-231. (12) Chiou, C. T.; Shoup, T. D.; Porter, P. E. Org. Geochem. 1985, 8, 9-14. (13) Schaefer, C.; Schuth, C.; Werth, C. J.; Reinhard, M. Environ. Sci. Technol. 2000, 34, 4341-4347. (14) Schwarzenbach, R. P.; Gschwend, P. M.; Imboden, D. M. Environmental Organic Chemistry; Wiley: New York, 1993. (15) Schultz, L. F.; Young, T. M.; Higashi, R. M. Environ. Toxicol. Chem. 1999, 18, 1710-1719. (16) Young, T. M. Ph.D Dissertation Thesis, The University of Michigan, 1996. (17) Gustafsson, O.; Haghseta, F.; Chan, C.; Macfarlane, J.; Gschwend, P. M. Environ. Sci. Technol. 1997, 31, 203-209. (18) Martin, F.; Sainz-Jimenez, C.; Gonzalez-Vila, F. J. Soil Sci. 1981, 132, (19) Weber, W. J.; McGinley, P. M.; Katz, L. E. Environ. Sci. Technol. 1992, 26, 1955-1962. (20) Young, T. M.; Weber, W. J. Environ. Sci. Technol. 1995, 29, 9297. (21) Cuypers, C.; Grotenhuis, T.; Nierop, K. G. J.; Franco, E. M.; Jager, A. D.; Rulkens, W. Chemosphere 2002, 48, 919-931. (22) Gelinas, Y.; Prentice, K. M.; Baldock, J. A.; Hedges, J. I. Environ. Sci. Technol. 2001, 35, 3519-3525. (23) Schmidt, M. W. I.; Skjemstad, J. O.; Czimezik, C. I.; Glaser, B.; Prentice, K. M.; Gelinas, Y.; Kuhlbushc, T. A. J. Global Biogeochem. Cycles 2001, 15, 163-167. (24) Nguyen, T. H.; Brown, R. A.; Ball, W. P. Org. Geochem. 2004, 35, 217-234.
5870
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 22, 2004
(25) Huang, W.; Yu, H.; Weber, W. J., Jr. J. Contam. Hydrol. 1998, 31, 129-148. (26) Weber, W. J.; Young, T. M. Environ. Sci. Technol. 1997, 31, 16861691. (27) Xing, B.; Pignatello, J. J. Environ. Sci. Technol. 1997, 31, 792799. (28) Schultz, L. F. Ph.D Dissertation, University of California, Davis, 1999. (29) Chiou, C. T.; Rutherfod, D. W. Environ. Sci. Technol. 1992, 26, 965-970. (30) Ball, W. P.; Roberts, P. V. Environ. Sci. Technol. 1991, 25, 12231236. (31) Manes, M.; Hofer, L. J. E. J. Phys. Chem. 1968, 73, 584-590. (32) Kleineidam, S.; Schuth, C.; Grathwohl, P. Environ. Sci. Technol. 2002, 36, 4689-4697. (33) Xia, G.; Pignatello, J. J. Environ. Sci. Technol. 2001, 35, 84-94. (34) Xia, G.; Ball, W. Environ. Sci. Technol. 1999, 33, 262-269. (35) Moon, H.; Tien, C. Chem. Eng. Sci. 1988, 43, 1269-1279. (36) Ju, D.; Young, T. M. Unpublished manuscript, 2004, (37) Weber, W. J., Jr.; LeBoeuf, E. J.; Young, T. M.; Huang, W. Water Res. 2001, 35, 853-868. (38) Huang, W. L.; Young, T. M.; Schlautman, M. A.; Yu, H.; Weber, W. J. Environ. Sci. Technol. 1997, 31, 1703-1710. (39) Pignatello, J. Adv. Colloid Interface Sci. 1998, 77, 445-467. (40) Gunasekara, A. S.; Simpson, M. J.; Xing, B. Environ. Sci. Technol. 2003, 37, 852-858. (41) Tissot, B. P.; Welte, D. H. Petroleum Formation and Occurrence; Springer-Verlag: Berlin, 1984. (42) Garbarini, D. R.; Lion, L. W. Environ. Sci. Technol. 1986, 20, 1263-1269. (43) Grathwohl, P. Environ. Sci. Technol. 1990, 24, 1687-1693. (44) Huang, W. L.; Weber, W. J. Environ. Sci. Technol. 1997, 31, 25622569. (45) Huang, W. L.; Weber, W. J. Environ. Sci. Technol. 1998, 32, 35493555. (46) Pignatello, J. J.; Xing, B. Environ. Sci. Technol. 1996, 30, 1-11. (47) Leboeuf, E. J.; Weber, W. J. Environ. Sci. Technol. 1997, 31, 16971702. (48) Lu, Y.; Pignatello, J. Environ. Sci. Technol. 2002, 36, 4553-4561. (49) Gunasekara, A. S.; Xing, B. J. Environ. Qual. 2003, 32, 240-246. (50) Braida, W. J.; Pignatello, J.; Lu, Y.; Ravikovitch, P. I.; Neimark, A. V.; Xing, B. Environ. Sci. Technol. 2003, 37, 409-417.
Received for review March 2, 2004. Revised manuscript received August 11, 2004. Accepted August 17, 2004. ES049668U
