Micellar Partitioning and Its Effects on Henry's Law ... - ACS Publications

The overall objective of this study was to determine the effects of both anionic and nonionic surfactants on the Henry's constant and hence the aboveg...
0 downloads 0 Views 180KB Size
Download PDF
Environ. Sci. Technol. 2006, 40, 208-214

Micellar Partitioning and Its Effects on Henry’s Law Constants of Chlorinated Solvents in Anionic and Nonionic Surfactant Solutions CHUNLONG ZHANG,* GANG ZHENG, AND COURTNEY M. NICHOLS Department of Environmental Sciences, University of HoustonsClear Lake, 2700 Bay Area Blvd., Houston, Texas 77058

Micellar partitioning of volatile chlorinated hydrocarbons in surfactant solutions and its effects on vapor-liquid equilibrium is fundamental to the overall design and implementation of surfactant-enhanced aquifer remediation. Surfactant micelles greatly enhance contaminant recovery from the subsurface; however, the reduced volatility of organic compounds compromises the aboveground treatment of surfactant-laden wastewaters using air-stripping process. Batch equilibrium tests were performed to acquire micellar partition coefficients (Km) and apparent Henry’s law constants (H*) of three prominent groundwater contaminants (tetrachloroethylene, trichloroethylene, cisdichlorethylene) in the presence of two anionic surfactants (sodium dodecyl sulfate, SDS; sodium dodecylbenzene sulfonate, SDBS) and two nonionic surfactants (Triton X-100 and Tween 80). The H* values were significantly reduced in the presence of all four surfactants over their critical micelle concentrations (cmc’s). On a cmc basis, the anionic surfactant SDS had the greatest effect on H*, followed by SDBS, Triton X-100, and Tween 80. Anionic surfactants decreased H* to an order of magnitude lower than nonionic surfactants, although nonionic surfactants decreased the H* at concentrations significantly lower than the anionic surfactants due to their lower cmc’s. Nonionic surfactants present higher Km and molar solubilization ratio than anionic surfactants. Tetrachloroethylene has the highest Km values among three chlorinated solvents, which agrees well with the hydrophobicity (Kow) of these chemicals. An empirical correlation between log Km and log Kow is developed on the basis of data from this study and the Km values reported for a number of chlorinated and nonchlorinated hydrocarbons. Equilibrium data were also tested against three sets of models that describe the partitioning of volatile compounds in vapor-water-micelle phases. Applications of these models in experimentally determining Km from batch vapor-water equilibrium data are discussed.

Introduction The use of surfactant to enhance subsurface aquifer remediation has been the subject of substantial research during the last 2 decades. Of particular promise for remediation are * Corresponding author phone: +1-281-283-3746; fax: +1-281283-3709; e-mail: [email protected]. 208

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 1, 2006

the chlorinated aliphatic hydrocarbons (CAHs) that are shown to be the most prevalent and the most difficult contaminants to remediate in soils and groundwaters in the United States (1, 2). A number of successful field tests to date have been achieved to recover CAHs from contaminated subsurface using surfactants (3-5). Aquifers contaminated with these compounds have not been restored to regulation-acceptable levels by conventional techniques such as pump-and-treat and bioremediation. Incorporating surfactants into current remediation strategies such as in situ soil flushing can potentially reduce significant time and cost by increasing the aqueous solubility and mobility of hydrophobic CAHs. Whereas solubilization by surfactant micelles enhances contaminant recovery from the subsurface, this benefit is compromised due to the needed aboveground treatment of surfactant-laden waste streams. The decontamination of volatile CAHs in these liquid wastes is particularly challenging because of the reduced volatility (apparent Henry’s law constant) in the presence of surfactant (6, 7). When surfactant is present at a high concentration, direct carbon adsoption is not efficient for CAHs removal due to the high affinity of most surfactants to organic carbon surfaces (8). Above the critical micelle concentrations (cmc’s), micellized contaminants are also less amenable to the removal by aboveground treatment processes such as air stripping, vacuum striping, and steam stripping. Consequently, the overall successful implementation of surfactant-enhanced aquifer remediation (SEAR) is still limited. The present work has been largely motivated to address this issue with an emphasis on micellar partitioning and its effects on the apparent Henry’s constants of three prominent volatile groundwater contaminants (tetrachloroethylene, PCE; trichloroethylene, TCE; cis-dichlorethylene, DCE). Micellar partitioning in surfactant solutions has been extensively reported for various hydrophobic contaminants of low volatilities such as polycyclic aromatic hydrocarbons (PAHs) and polychlorinated biphenyls (PCBs) (9-13). Associated with these efforts, parameters such as the micellar partitioning coefficient (Km) and molar solubilization ratio (MSR) are readily available using various experimental equilibrium and kinetic approaches (14, 15). Theoretical models have also been developed to estimate Km using surfactant and solute specific properties, including surface tension, molecular surface area, solute’s structural formula, and Henry’s constant (16, 17). The effect of surfactants on the vapor-liquid partitioning of volatile contaminants such as PCE, TCE, and DCE and the acquisition of those parameters have not been thoroughly investigated. Only a few batch studies addressed the partitioning of volatile chlorinated solvents in air-water-micelle systems (18-22). These studies often used chlorinated solvents at concentrations lower than those prevalent in dense nonaqueous phase liquid (DNAPLs) source zones and high concentration plumes. In a related study on a pilotscale diffused aeration system, Chern and Chou (23) noted significantly reduced emission of volatile organic compounds when surfactants were present at a sub-cmc concentration. A reduced apparent Henry’s constant to lower than 0.01 (unitless) renders air stripping ineffective in removing volatile contaminants from liquid medium (24). Kibbey et al. (7) revealed the importance of solublization-induced volatility reduction when a sieve-tray air stripper was used to remove PCE from a wastewater containing a nonionic surfactant. Their results concluded that neglecting micellar effect can produce a several-orders-of-magnitude overprediction of air stripper performance at application-relevant surfactant 10.1021/es051387e CCC: $33.50

 2006 American Chemical Society Published on Web 11/16/2005

concentrations. The overall objective of this study was to determine the effects of both anionic and nonionic surfactants on the Henry’s constant and hence the aboveground air-stripping process for the removal of chlorinated solvents. Specific aims of this study were (i) to investigate the effects of micellar partitioning on the volatility of chlorinated solvents at concentrations relevant to DNAPLs source zone and (ii) to develop a new methodology to obtain Km directly from batch vapor-water partitioning data in place of the apparent solubility model commonly used for semivolatile or nonvolatile organic compounds. An improved understanding of micellar partitioning and its effect on the apparent Henry’s constant is essential for the design and performance evaluation of aboveground treatment techniques.

Materials and Methods Test Surfactants and Chlorinated Solvents. The physicochemical properties of two anionic and two nonionic surfactants used in this study are given in Table SI-1 of the Supporting Information. Sodium dodecyl sulfate (SDS) and sodium dodecyl benzene sulfonate (SDBS) were purchased from Sigma-Aldrich. Polyoxyethylene (20) sorbitan monooleate (Tween 80) and tert-octylphenoxypoly-ethoxyethanol (Triton X-100) were purchased from Fisher. These surfactants were used without further purification. The selection of the four surfactants was based on their potential in remediation uses (8, 25-27). Three volatile chlorinated solvents of various hydrophobicities were HPLC-grade PCE and reagent-grade TCE and cis-DCE (Aldrich). The physicochemical properties of these test contaminants are listed in Table SI-2 of the Supporting Information. Batch Studies on Micellar Solubilization and VaporLiquid Partitioning. A series of glass vials with an average volume of 71.59 ( 0.48 mL (measured from 10 randomly selected vials) were used for the equilibrium partitioning tests. Each set of duplicate vials was filled with 20 mL of aqueous surfactant solution at concentrations equivalent to and above the cmc (up to 500 × cmc) that are typically used in surfactant-based remediation (8). A duplicate surfactantfree control was also included. The vials were then sealed airtight with Teflon-lined septa and crimp aluminum caps. After the vials were sealed, a known amount of PCE, TCE, or DCE dissolved in methanol was injected. The amount of PCE, TCE, or DCE was predetermined from trial runs to confirm the absence of emulsification and free DNAPLs while saturating the solutions. The concentrations of PCE, TCE, and DCE in the samples were examined in a wider range than previously reported (20). The amount of methanol in each vial was less than 0.5% (v/v). At such low concentrations, methanol does not affect vapor-liquid partitioning (28). The vials were then placed in a shaker to equilibrate for 24 h at room temperature (21.1 ( 0.8 °C). After equilibration, the vials were removed from the shaker and further stabilized for several hours. To minimize potential volatilization losses, the vials were incubated in an inverted position, with the liquid phase in contact with septa. After stabilization, a 0.2mL gas sample was withdrawn from the headspace using a gastight syringe. The samples were then injected into the gas chromatograph equipped with a flame ionization detector (GC-FID) to analyze PCE, TCE, and DCE concentrations in the headspace of the sample vials. The total concentrations in the aqueous phase (dissolved plus micellized) were calculated from the difference between the total mass of contaminant added initially and the mass remaining in the vapor phase. Details on the headspace analysis of volatile chlorinated solvents are given in the Supporting Information.

Results Micellar Solubilization of Chlorinated Solvents. Micellar solubilization of chlorinated solvents by four surfactants were

evaluated using the apparent solubility data at various surfactant concentrations. The apparent solubility (CA,total) is the concentration of contaminant dissolved in aqueous phase (CA) plus the concentration of micellized contaminant, and the latter is linearly proportional to micellized surfactant concentration (S - cmc) according to eq 1:

CA,total ) CA + KmCA(S - cmc)

(1)

where S is the total surfactant concentration and cmc is the critical micelle concentration (mg L-1). The proportionality constant, Km, is the micellar partition coefficient. Km has a unit of L mg-1 when contaminant concentrations in the micelle pseudophase and in the aqueous phase are expressed in mg mg-1 (mg of micellized CAHs per mg of micellized surfactant) and mg L-1 (mg of aqueous CAHs per L water), respectively. Equation 1 is valid by assuming a constant cmc, no changes in micelle shape and size during solubilization, a constant Km value independent of S, and the absence of monomer solubilization. With these assumptions, eq 1 can be used to determine Km and CA by the least-squares fit of the linear plot between CA,total and S - cmc. The slope of this plot can be used to estimate the molar solubilization ratio (MSR)

MSR ) KmCAMWSurfactant/MWA

(2)

where MW denotes the molecular weight. MSR is defined as the number of moles of solubilizate incorporated into micelles per mole surfactant micelles at equilibrium or the average number of solubilizate molecules per micelle divided by the aggregation number (9). Figure 1 shows the linear plots in which surfactant concentrations are normalized to the number of cmc for a comparison purpose. As can be seen, the apparent solubility of chlorinated solvents increased linearly as surfactant concentrations increased (p < 0.05). The slopes in Figure 1 further indicate that two anionic surfactants, on a cmc basis, present higher solubility enhancement than two nonionic surfactants. On a mass concentration basis (not shown), however, the opposite is true. This is expected since SDS and SDBS have much higher cmc values than two nonionic surfactants (Table SI-1 of the Supporting Information). The solubilization potential of three chlorinated solvents can be further evaluated from the parameters summarized in Table 1. Nonionic surfactants present higher MSR than anionic surfactants. The MSR value is the highest for TCE with the four surfactants tested, excluding SDS. The highest solubilization enhancement observed was for TCE by nonionic Tween 80, the MSR value being 930 mmol of TCE/mole of surfactant micelle. For both PCE and TCE, the MSR values are approximately 6 times higher for Tween 80 than Triton X-100. This discrepancy cannot be simply explained by their difference in aggregation number (Triton X-100 ) 100-155; Tween 80 ) 58) (12) and molar volume of surfactant molecule (Triton X-100 ) 0.58 L/mol and Tween 80 ) 1.21 L/mol). It appears that surfactant structure and its interaction with solubilizate may attribute to such a discrepancy as well. The concentration of contaminant dissolved in aqueous phase (CA) in Table 1 appears to be independent of surfactant type and characteristic of the solute of interest. An additional observation is that the estimated CA values are approximately 2 times the aqueous solubility for the more hydrophobic PCE and TCE (Table SI-2 of the Supporting Information). This increase may be presumably due to solubility enhancement by surfactant monomers, a phenomenon observed for highly hydrophobic PCBs and DDT (10, 29). In the case of nonionic surfactants, this enhancement is related to successive micellization of heterogeneous surfactant homologues without a well-defined cmc (31). VOL. 40, NO. 1, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

209

TABLE 1. Parameters Describing Micellar Solubilization of Chlorinated Solvents by Surfactants anionic parameters

chlorinated solvents

SDS

cis-DCE TCE PCE cis-DCE TCE PCE

1858 2493 311 8.5 9.9 4.7

CAa (mg L-1) MSR (mmol mol-1)b

nonionic

SDBS

Triton X-100

Tween 80

1659 2334 352 24 31 5.2

1599 2407 324 129 145 27

1714 2216 266 184 930 193

a Reported solubilities (C ) are given in Table SI-2 in the Supporting w Information. b mmol mol-1 ) millimoles micellized chlorinated solvent per mole surfactant.

PCE, 66 mg of TCE, and 48 mg of DCE), and VV and VA are the total volume (L) of the vapor and aqueous phases, respectively. Assuming that only the dissolved phase contaminant (CA) is available for gas exchange between the aqueous and vapor phases, CA is related to CV by Henry’s constant (i.e., CA ) CV/H). By substituting CA into eq 3, it can be rearranged to become

(

)

VA VAKm VV 1 ) + + (S - cmc) CV MT MTH MT H

FIGURE 1. Effects of surfactant concentrations on the apparent solubility of (a) PCE, (b) TCE, and (c) DCE. Surfactants were SDS (b), SDBS (1), Triton X-100 (4), and Tween 80 (O). Regression lines denote solubilization model of eq 1. Effect of Micellar Partitioning on Vapor Phase Concentrations. A three-phase model, similar to Anderson’s (18), is used herein to derive a simple model describing the effect of surfactant concentration (S) on the equilibrium vapor phase concentration (CV) of chlorinated solvents. As will be described later, this modified equation allows for a direct estimation of Henry’s constant (H) and Km using batch equilibrium data. In a three-phase system, volatile organics are partitioned into vapor, aqueous, and micellar phases as follows

MT ) VVCV + VACA + KmCAVA(S - cmc)

(3)

where MT is the total mass (mg) of contaminant in the closed system, which was kept constant during the study (9 mg of 210

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 1, 2006

(4)

Equation 4 can be used to determine H and Km by plotting 1/CV versus S - cmc, where S is the known concentration of surfactant added and CV can be determined directly from headspace analysis by GC. The linear curve has a slope of (VAKm/MTH) and an intercept of (VV/MT) + (VA/MTH). Figure 2 shows that experimental data agree well with the model (eq 4). The linear regressions between 1/CV and S cmc are significant (p < 0.05) for all combinations of surfactants and chlorinated solvents. In contrast to the micellar solubilization shown in Figure 1, where apparent solubility linearly increased with increasing surfactant concentration at the supra-cmc’s, the vapor-phase solvent concentration decreased linearly with increasing concentration of micellized surfactant. Also noted in Figure 2, anionic surfactant concentrations were an order of magnitude higher than nonionic surfactants due to the higher cmc’s of anionic surfactants employed in this study. Note also that H estimated from Figure 2, as defined previously, is the ratio of CV to CA when the contaminant in vapor and surfactant solution is in equilibrium. Depending on the extent of solubility enhancement by surfactant monomers, the CA values may differ from the aqueous phase equilibrium concentration in surfactant-free solutions. Consequently, these estimated H values (0.52 ( 0.28 for PCE, 0.20 ( 0.03 for TCE, and 0.19 ( 0.02 for DCE) may deviate from the Henry’s constant commonly reported in the literature (Table SI-2 of the Supporting Information), particularly the underestimation of the H value for TCE. The H values estimated are, as expected, independent of surfactant types (data not shown). Effect on Apparent Henry’s Constants. The experimentally determined apparent solubility (CA,total) and vapor-phase concentration (CV), as described in the preceding sections, can be combined in the form of an apparent Henry’s constant (H*) to determine the vapor-liquid partitioning of chlorinated solvents in surfactant solutions. Unlike H, which is characteristic of partition chemicals of interest, the apparent Henry’s constant (H*) is also a function of surfactant concentration and is defined as

H* )

CV CA,total

(5)

FIGURE 2. Effects of surfactant concentrations on the vapor-phase concentrations of (a) PCE, (b) TCE, and (c) DCE. Surfactants were SDS (b), SDBS (1), Triton X-100 (4), and Tween 80 (O). The linear plots between the inverse vapor-phase concentration (Cv) and micellized surfactant concentration (S - cmc) denote the model of eq 4. evaluated individually from apparent solubility or vaporphase concentration because H* is the ratio of these two. Interestingly, the reduction in H* increased with the increasing values of log Kow, implying that hydrophobicity may be the determining factor in vapor-liquid partitioning of chlorinated solvents in micellized surfactant solutions. Effect on Extramicellar Fraction. The extramicellar fraction (fex) is defined as the ratio of CA to the total contaminant concentration (i.e., apparent solubility, CA,total) in solution:

fex )

CA

(6)

CA,total

The above equation cannot be used to estimate fex, since the presence of a micellar phase precludes the direct determination of CA with known MT and CV commonly measured in batch vapor-liquid partition study. The fex, however, can be determined from the ratio of Henry’s constant measured with surfactant (i.e., H*) to that measured without (21):

fex )

H* H

(7)

FIGURE 3. Effect of surfactant concentrations on apparent Henry’s constants (H*): (a) SDS, (b) SDBS, (c) Triton X-100, and (d) Tween 80. Compounds were PCE (b), TCE (O), and DCE (2).

The fex value is a function of Km and micellized surfactant concentration (S - cmc), which can be derived by rearranging eq 1 and subsequent substitution into eq 6:

Figure 3 depicts the effects of four surfactants on the dimensionless H* in describing the vapor-liquid partitioning of the volatile contaminants. For comparison purposes, surfactant concentrations are expressed on a mass basis as well as a cmc basis. Results in Figure 3 clearly indicate that H* values were significantly reduced in the presence of all four surfactants at supra-cmc’s. On a cmc basis, the anionic SDS had the greatest effect on H*, followed by SDBS, Triton X-100, and Tween 80. This order follows the same sequence as the cmc’s of the surfactants (Table SI-1 of the Supporting Information). At the same cmc level, anionic surfactants have the higher surfactant mass in the solution. With the highest concentration tested, anionic surfactants decreased H* to an order of magnitude lower than nonionic surfactants. However, nonionic surfactants decreased the H* at mass concentrations significantly lower than the anionic surfactants. This is because supra-cmc concentrations are achievable using a lower mass of nonionic surfactants. In comparing H* among three test contaminants, Figure 3 indicates that PCE had the greatest percent reduction in H*, followed by TCE and DCE for a given surfactant. For instance, at 100 × cmc, the reduction in H* of PCE was 91%, 57%, and 4% for SDBS, Triton X-100, and Tween 80, respectively (For SDS, 100 × cmc was not tested and the reduction was 87% at 25 × cmc). The reduced H* cannot be

CA,total 1 ) ) 1 + Km(S - cmc) fex CA

(8)

Eq 8 clearly shows that fex is inversely related to Km and the micellized surfactant concentration (S - cms). Since fex can be estimated from experimentally measured H* and H, eq 8 can be used to determine Km values using a least-squares fit. Alternatively, by rearranging eq 8, the following equation can be used to calculate Km at various surfactant concentrations and Km values are averaged on the assumption that Km is theoretically independent of surfactant concentrations:

Km )

(

)

1 1 -1 fex S - cmc

(9)

The experimental data showing the inverse relationship between fex and surfactant concentration are demonstrated in Figure 4, where both mass- and cmc-based surfactant concentrations are plotted for a comparison. The linear plot (1/fex vs S - cmc) given in Figure 5 shows that the model (eq 8) fits the experimental data well. At the surfactant concentration tested, anionic surfactants resulted in the lower fex than two nonionic surfactants. It is also evident that the surfactant-induced reduction in fex of PCE was substantially VOL. 40, NO. 1, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

211

TABLE 2. Micellar Partitioning Coefficients (log Km) of Chlorinated Solvents in the presence of anionic and nonionic surfactants log Km (L/kg) anionic surfactant chlorinated solvents

model numbera

SDS

SDBS

Triton X-100

Tween 80

cis-DCE

1 2 3 1 2 3 1 2 3

0.19 1.40 1.50 0.26 1.85 1.64 0.94 2.55 2.16

0.61 1.33 1.41 0.70 1.54 1.47 0.85 3.20 2.33

1.11 1.65 1.69 1.11 1.80 1.83 1.36 2.17 2.11

0.90 1.58 1.54 1.63 2.77 2.49 1.97 3.17 2.68

TCE PCE

a

FIGURE 4. Effect of surfactant concentrations on extramicellar fraction (fex): (a) SDS, (b) SDBS, (c) Triton X-100, and (d) Tween 80. Compounds were PCE (b), TCE (O), and DCE (2).

FIGURE 5. Linear plots of eq 8 between the inverse extramicellar fraction (fex) and micellized surfactant concentration (S - cmc). Compounds were PCE (b), TCE (O), and DCE (2). greater than that of TCE and DCE when either anionic surfactant was present (Figure 4a,b). As seen in Figure 4c, the fex values of the three test chemicals exhibited similar decreases in the presence of Triton X-100. For example, fex was 0.43 for PCE, 0.56 for TCE, and 0.62 for DCE. For Tween 80 (Figure 4d), little to no effect on fex was noted when the concentrations of Tween 80 were less than 100 × cmc, which corresponded to a lower mass concentration compared to other test surfactants. Effect on Micellar Partition Coefficient (Km). The preceding sections have shown that three equations (eqs 1, 4, 8) can be used to estimate Km from batch vapor-liquid equilibrium data when headspace analysis is performed. The best estimates of Km can be obtained using the least-squares fit through each equation. 212

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 1, 2006

nonionic surfactant

Km values estimated by (1) eq 1, (2) eq 4, and (3) eq 8.

Table 2 summarizes the log Km values estimated by differing models described above. Two models, based on experimentally determined vapor phase concentration (eq 4) and derived extramicellar fraction (eq 8), are in general agreement in estimating Km values. Regardless of the large deviation from eq 1, all models correlate with a minimum R of 0.74 and predict the same sequence with regard to the effect of the types of surfactant and chlorinated solvents used. Using model 2 (eq 4) and model 3 (eq 8) data (Table 2), a two-way analysis of variance (ANOVA) indicates the significant effect of four surfactants and three CAHs on Km (both are significant at p < 0.01). When data from models 2 and 3 are further grouped into anionic vs nonionic surfactants, a two-way ANOVA shows the significant effect of three CAHs at p < 0.01, and the difference between anionic surfactants (average Km ) 1.87) and nonionic surfactants (average Km ) 2.12) is significant only at p ()0.085) < 0.1. Among three chlorinated solvents, the Km values (Models 2 and 3 data only) are in an increasing order of DCE (1.51) < TCE (1.92) < PCE (2.55), consistent with the hydrophobicity of these compounds. This is also in good agreement with the earlier analysis on the effects on Henry’s constants (19, 21). The variations in the Km values acquired by different experimental procedures and model assumptions were noted by several studies (16, 18, 30). The discrepancies between various model results (Table 2) and the reported Km may be partially related to the much higher concentrations of surfactants and chlorinated solvents used in this study than the previous ones (20, 21). When the concentration of chlorinated solvents was higher than the aqueous solubility, it was critical to ensure that the system was free of additional DNAPLs. The presence of even a microscopic DNAPL droplet can result in a significant overestimation of the Km value. In some cases (i.e., TCE in SDS and PCE in SDS and Triton X-100), the calculated Km values from the current study are lower than those reported (21). Ko et al. (31) revealed that Km values, which should be independent of the concentrations of hydrophobic organic compounds (HOCs), were in fact decreased with increasing HOCs concentration. The Km values reported by Vane and Giroux (21) were for solutions of 0.2-0.9 mg of PCE and TCE in 3-16 mL of water. Therefore, even the highest test concentration (300 mg L-1) was an order of magnitude lower than the concentrations used in the present study. Thus, the magnitude of difference in HOC test concentrations may justify the discrepancy of the Km values. Relevant to the variations was also the observations that solubilization caused micelle to swell, which in turn promoted micelle growth and associated transformation of micelle shapes (32).

Discussion Relationship between Octanol- and Micelle-Water Partition Coefficients for Chlorinated Hydrocarbons. The Km value depends on the properties of both solubilizate and surfactant and their interactions (31, 33, 34). Its dependence on the solubilizate’s hydrophobicity (Kow) has been reported for various volatile and semivolatile hydrocarbons (9, 14, 20, 29, 35). Several studies (15, 34, 36) have examined the correlation using previously published data, all indicating a significant linear relationship between log Km and log Kow. Nonlinearity was observed for extremely hydrophobic PCBs (log Kow in the range of 5.0-8.5) in the presence of SDS micelles (10). The nonlinearity was explained by the hydration of the polar headgroup of the SDS micelles and their polar palisade layer, which may screen off the hydrophobic core for more hydrophobic molecules. The larger and more hydrophobic molecules also require more energy to form a cavity in more structured micellar pseudophases, resulting in lower Km and a loss of linearity. Figure SI-1 of the Supporting Information includes a total of 13 surfactants and 21 organic contaminants with a log Kow range of 1.3-6.36. The combined data result in the following semiempirical correlation that relates experimentally determined Km to Kow

log Km ) 1.02 log Kow + 1.02

(R2 ) 0.970, n ) 74) (10)

where reported Km values are converted to mole fraction ratio for comparison purposes, and the near-unity slope (≈1.0) is significant (p < 0.01). Similar near-unity slopes were also found by others (15, 16), suggesting that the solvency of a micellar pseudophase is comparable with the bulk octanol phase. Km values lower than Kow were also reported, which was explained by the large Laplace pressure acting across the curved micelle-water interface (35). Figure SI-1 of the Supporting Information clearly demonstrates the linear dependence of log Km on solute hydrophobicity. The log Km values are in an increasing order of hydrophobic Cl substitution for both chlorinated ethanes (CH2Cl2 < CHCl3 < CCl3) and chlorinated ethenes (cis-DCE < tran-DCE < TCE < PCE). The same trend is also apparent for mono- and ploycyclic aromatic hydrocarbons and their corresponding methyl or halogen substitutes. The log Km values increase with the increase in benzene rings and the substitution of hydrophobic functional groups (e.g., benzene < toluene < p-xylene < naphthalene < 1-methylnaphthalene < 1-bromonaphthalene < anthracene ≈ phenanthrene < pyrene < perylene). The dependence of micellar partitioning on various surfactants can be seen from the vertical variations in Figure SI-1 of the Supporting Information. It is noted that 13 surfactants of all types (anionic, cationic, and nonionic) are included for the regression analysis. As expected, a difference of approximately one log unit is frequently seen when the same compound is present in micelles of various surfactants. The log-log correlation between Km and Kow is therefore better expressed to be surfactant-specific. However, such information is very limited. The insert of Figure SI-1 of the Supporting Information illustrates that Km values measured from this study are generally in agreement with what is predicted from the semiempirical correlation derived in this study (eq 10). The Km values estimated by eq 1 are excluded, since this method underestimated the Km values (see discussion below). The Km values calculated from eqs 4 and 8 are generally in good agreement with the literature values, and no apparent distinction can be made between these two due to variations among surfactants.

Headspace Analysis To Determine Km of Volatile Chemicals. Unlike other partition coefficients, Km has been the subject of some ambiguity and debate in the literature, for example, with regard to its definition (9, 29), experimental approach (14, 36, 37), and choice of units (reference state) (15, 38). An accurate quantification of Km value is further complicated by other issues, such as the dynamics of micellar growth (size and shape) during the course of solubilization, change in cmc’s, and its dependence on surfactant and solute concentration (12, 22). Many of these are still the subjects of ongoing research, and a detailed account is beyond the scope of the present study. It is, however, relevant to the following discussion when examining the three equations (eqs 1, 4, 8) used to determine the Km values of volatile chemicals. The common experimental approach in determining Km is to solubilize under saturation followed by the removal of an excess amount of solute (9, 14, 29, 36), and Km is calculated by the slope of the solubilization curve (eq 1). On the basis of the results from this study and the results from Almegren et al. (14) on the more volatile benzene series compounds, it is concluded that this traditional method is appropriate for more hydrophobic compounds but will likely underestimate the Km for volatile compounds. A more frequently used approach for volatile compounds is to use gas-phase equilibrium data when the solute is present below saturation (18-21, 38). In this case, eq 8 can be used to estimate the Km from headspace analysis (Figure 5). This equation is analogous to that previously reported (20, 39), except a unit conversion factor was present to convert between M-1 and mole fraction ratio and a micellar fraction (fm ) 1 - fex) was used. Alternatively, eq 4 derived in this paper can directly use vapor phase concentration (Cv) to estimate Km (Figure 2). An added advantage of this linearized equation is its estimation of H from the headspace data. Micellar partition coefficient (Km) is theoretically independent of solute concentration (CA, saturation or below saturation) if a linear solubilization model is assumed (eq 1). This was supported by the Km values of benzene determined by various methodologies over considerable ranges of micellar saturation in SDS (15). All the Km values reported by several investigators were within 25% of each other. This, however, was in significant contrast with pentanol solubilization by cationic surfactants (30). A factor of 3-5 reduction in the Km was observed as the mole fraction of pentanol in the micelles (X) increased from 0 to 0.9. Their results suggest a Langmuir-type adsorption model, and the observed decreasing Km with increasing X was attributed to the limited adsorptive capacity of the cationic surfactant micelles when pentanol -OH groups bind at the micelle surface. Morgan et al. (30) also noticed significant discrepancies of the solubilization results determined using various experimental techniques and model assumptions in data analysis and implied headspace chromatography was the preferred method. Solubilization data from this study imply that nonionic surfactants appear to be the more favorable economical choice for surfactant-enhanced remediation of chlorinated solvents due to the lower amount of mass required. Although the anionic surfactants had a greater impact on the solubilization of the test chemicals on a cmc basis, a substantially greater mass of surfactant is required to initiate solubilization. Vapor-liquid partition data further imply that the significantly lower H* will preclude the removal of volatile chemicals through stripping techniques as a part of the SEAR strategy, resulting in a higher remediation cost. The correction of Henry’s constants due to solubilization-induced volatility reduction is needed to better design and evaluate the costeffectiveness of air-strippers. VOL. 40, NO. 1, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

213

Acknowledgments This work was supported by Welch Foundation and the EIH. We thank Dean Muirhead of NASA-JSC for his insightful comments.

Supporting Information Available Additional details on the physicochemical properties of test surfactants and chlorinated solvents and the log-log correlation diagram between Km and Kow. This material is available free of charge via the Internet at http://pub.acs.org.

Literature Cited (1) Pankow, J. F.; Cherry, J. A. Dense Chlorinated Solvents and Other DNAPLs in Groundwater; Waterloo Press: Portland, OR, 1996. (2) Stroo, H. F.; Unger, M.; Ward, C. H.; Kavanaugh, M. C.; Vogel, C.; Leeson, A.; Marqusee, J. A.; Smith, B. P. Remediating chlorinated solvent source zones. Environ. Sci. Technol. 2003, 225-230A. (3) Fountain, J. C.; Starr, R. C.; Middleton, T.; Beikirch, M.; Taylor, C.; Hodge, D. S. A controlled field test of surfactant-enhanced aquifer remediation. Ground Water 1996, 34, 910-916. (4) Londergan, J. T.; Meinardus, H. W.; Mariner, P. E.; Jackson, R. E.; Brown, C. L.; Dwarakanath, V.; Pope, G. A.; Ginn, J. S.; Taffinder, S. DNAPL removal from a heterogeneous alluvial aquifer by surfactant-enhanced aquifer remediation. Ground Water Monit. Remediat. 2001, 21, 57-67. (5) Ramsburg, C. A.; Pennell, K. D.; Abriola, L. M.; Daniels, G.; Drummond, C. D.; Gamache, M.; Hsu, H.-L.; Petrovskis, E. A.; Rathfelder, K. M.; Ryder, J. L.; Yavaraski, T. P. Pilot-scale demonstration of surfactant-enhanced PCE solubilization at the Bachman Road Site. 2. System operation and evaluation. Environ. Sci. Technol. 2005, 39, 1791-1801. (6) Cheng, H.; Sabatini, D. A.; Kibbey, T. C. G. Solvent extraction for separating micellar-solubilized contaminants and anionic surfactants. Environ. Sci. Technol. 2001, 35, 2995-3001. (7) Kibbey, T. C. G.; Pennell, K. D.; Hayes, K. F. Application of sievetray air strippers to the treatment of surfactant-containing wastewaters. AIChE J. 2001, 47, 1461-1470. (8) Abriola, L. M.; Drummond, C. D.; Hahn, E. J.; Hayes, K. F.; Kibbey, T. C. G.; Lemke, L. D.; Pennell, K. D.; Petrovskis, E. A.; Ramsburg, C. A.; Rathfelder, K. M. Pilot-scale decomposition of surfactantenhanced PCE solubilization at the Bachman Road Site. 1. Site characterization and test design. Environ. Sci. Technol. 2005, 39, 1778-1790. (9) Edwards, D. A.; Luthy, R. G.; Liu, Z. Solubilization of polycyclic aromatic hydrocarbons in micellar nonionic surfactant solutions. Environ. Sci. Technol. 1991, 25, 127-133. (10) Dulfer, W. J.; Bakker, M. W. C.; Govers, H. A. J. Micellar solubility and micelle/water partitioning of polychlorinated biphenyls in solutions of sodium dodecyl sulfate. Environ. Sci. Technol. 1995, 29, 985-992. (11) Guha, S.; Jaffe´, P. R.; Peters, C. A. Solubilization of PAH mixtures by a nonionic surfactant. Environ. Sci. Technol. 1998, 32, 930935. (12) Hill, A. J.; Ghoshal, S. Micellar solubilization of naphthalene and phenanthrene from nonaqueous-phase liquids. Environ. Sci. Technol. 2002, 36, 3901-3907. (13) Yoem, I. T.; Ghosh, M. M.; Cox, C. D. Kinetic aspects of surfactant solubilization of soil-bound polycyclic aromatic hydrocarbons. Environ. Sci. Technol. 1996, 30, 1589-1595. (14) Almgren, M.; Grieser, F.; Thomas, J. K. Dynamic and static aspects of solubilization of neutral arenes in ionic micellar solutions. J. Am. Chem. Soc. 1979, 101, 279-291. (15) Jafvert, C. T. Sediment- and saturated-soil-associated reactions involving an anionic surfactant (dodecylsufate). 2. Partition of PAH compounds among phases. Environ. Sci. Technol. 1991, 25, 1039-1045. (16) Liu, G. G.; Roy, D.; Rosen, M. J. A simple method to estimate the surfactant micelle-water distribution coefficients of aromatic hydrocarbons. Langmuir 2000, 16, 3595-3605. (17) Smith, G. A.; Christian, S. D.; Tucker, E. E.; Scamehorn, J. F. Group contribution model for predicting the solubilization of organic solutes by surfactant micelles. Langmuir 1987, 3, 598599. (18) Anderson, M. A. Influence of surfactants on vapor-liquid partitioning. Environ. Sci. Technol. 1992, 26, 2186-2191. (19) Shimotori, T.; Arnold, W. A. Measurement and estimation of Henry’s law constants of chlorinated ethylenes in aqueous surfactant solutions. J. Chem. Eng. Data 2003, 48, 253-261. 214

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 1, 2006

(20) Valsaraj, K. T.; Gupta, A.; Thibodeaux, L. J.; Harrison, D. P.; Partitioning of chloromethanes between aqueous and surfactant micellar phases. Water Res. 1988, 22, 1173-1183. (21) Vane, L. M.; Giroux, E. L. Henry’s law constants and micellar partitioning of volatile organic compounds in surfactant solutions. J. Chem. Eng. Data 2000, 45, 38-47. (22) Uchiyama, H.; Christian, S. D.; Tucker, E. E.; Scamehorn, J. F. Solubilization of trichloroethylene by polyelectrolyte/surfactant complexes. AIChE J. 1994, 40, 1969-1975. (23) Chern, J.-M.; Chou, S.-R. Effects of surfactant on volatile organic compound emission rates in a diffused aeration system. Ind. Eng. Chem. Res. 2002, 41, 5042-5048. (24) U. S. Environmental Protection Agency. Engineering Bulletin: Air Stripping of Aqueous Solutions, EPA/540/2-91/022, Center for Environmental Research Information: Cincinnati, OH, October 1991. (25) Zhang, C.; Valsaraj, K. T.; Constant, W. D.; Roy, D. Aerobic biodegradation kinetics of four anionic and nonionic surfactants at sub- and supra-critical micelle concentrations (CMCs). Water Res. 1999, 33, 115-124. (26) Zhang, C.; Valsaraj, K. T.; Constant, W. D.; Roy, D. Surfactant screening for soil washing: Comparison of foamability and biodegradability of a plant-based surfactant with commercial surfactants. J. Environ. Sci. Health 1998, 33, 1249-1273. (27) Zhou, M.; Rhue, R. D. Screening commercial surfactants suitable for remediating DNAPL source zones by solubilization. Environ. Sci. Technol. 2000, 34, 1985-1990. (28) Gossett, J. M. Measurement of Henry’s law constant for C1 and C2 chlorinated hydrocarbons. Environ. Sci. Technol. 1987, 21, 202-208. (29) Kile, D. E.; Chiou, C. T. Water solubility enhancements of DDT and trichlorobenzene by some surfactants below and above the critical micelle concentration. Environ. Sci. Technol. 1989, 23, 832-838. (30) Morgan, M. E.; Uchiyama, H.; Christian, S. D.; Tucker E. E.; Scamehorn, J. F. Solubilization of pentanol by micelles of cationic surfactants and binary mixtures of cationic surfactant in aqueous solution. Langmuir 1994, 10, 2170-2176. (31) Ko, S.-O.; Schlautman, M. A.; Carraway, E. R. Partitioning of hydrophobic organic compounds to sorbed surfactants. 1. Experimental studies. Environ. Sci. Technol. 1998, 32, 27692775. (32) Ranganathan, R.; Peric, M.; Medina, R.; Garcia, U.; Bales, B. L.; Almgren, M. Size, hydration, and shape of SDS/heptane micelles investigated by time-resolved fluorescence quenching and electron spin resonance. Langmuir 2001, 17, 6765-6770. (33) Diallo, M. S.; Abriola, L. M.; Weber, W. J. Jr. Solubilization of nonaqueous phase liquid hydrocarbons in micellar solutions of dodecyl alcohol ethoxylates. Environ. Sci. Technol. 1994, 28, 1829-1837. (34) Pennell, K. D.; Adinolfi, A. M.; Abriola, L. M.; Diallo, M. S.; Solubilization of dodecane, tetrachlobenzene, and 1,2-dichlorobenzene in micellar solutions of ethoxylated nonionic surfactants. Environ. Sci. Technol. 1997, 31, 1382-1389. (35) Valsaraj, K. T.; Thibodeaux, L. J. Relationships between micellewater and octanol-water partition constants for hydrophobic compounds of environmental interest. Water Res. 1989, 23, 183189. (36) Jafvert, C. T.; Van Hoof, P. L.; Heath, J. K. Solubilization of nonpolar compounds by nonionic surfactant micelles. Water Res. 1994, 28, 1009-1017. (37) De Lisi, R.; Genova, C.; Liveri, V. T. Solute distribution in micellar solutions by a calorimetric method. Thermodynamics of transfer of n-butanol and n-pentanol from water to sodium dodecyl sulfate micelles at 25 °C. J. Colloid Interface Sci. 1983, 95, 428434. (38) Simon, S. A.; McDaniel, R. V.; McIntosh, T. J. Interaction of benzene with micelles and bilayers. J. Phys. Chem. 1982, 86, 1449-1456. (39) Spink, C. H.; Colgan, S. Thermodynamics of alcohol binding to deoxycholate micelles and the hydrophobic effect. J. Phys. Chem. 1983, 87, 888-894.

Received for review July 17, 2005. Revised manuscript received October 10, 2005. Accepted October 25, 2005. ES051387E