Environ. Sci. Technol. 2001, 35, 2995-3001
Solvent Extraction for Separating Micellar-Solubilized Contaminants and Anionic Surfactants HEFA CHENG,† DAVID A. SABATINI,* AND TOHREN C. G. KIBBEY School of Civil Engineering and Environmental Science, University of Oklahoma, Norman, Oklahoma 73019-0631
Decontamination of contaminant-laden surfactant solutions is critical to successful implementation of surfactantenhanced aquifer remediation (SEAR). Solvent extraction was studied for removing micellar-solubilized contaminants having low equivalent alkyl carbon numbers (EACNs) from surfactant solutions. Factors influencing the solvent extraction of micellar-solubilized contaminant were studied, including surfactant concentration, solution salinity, solvent solubilization, and solvent/solution volumetric ratio. A model was developed to quantify the impacts of these factors on contaminant removal. The good agreement between experimental results and model predictions corroborates assumptions made in the model development. From these results, it is concluded that extracting solvents must have much higher EACNs than that of the contaminant to reduce the micellar solubilization of the solvents, which can significantly reduce contaminant removal efficiency. However, the highest EACN solvent is not necessarily the best one for contaminant removal due to other constraints (e.g., molecular weight and viscosity). Increasing the total surfactant concentration or salinity of an anionic surfactant solution increases its contaminant solubilization capacity but reduces the contaminant removal efficiency by solvent extraction. Continuous column extraction operated at a low column surface loading rate allowed contaminant partitioning to approach equilibrium conditions, and multistage column extraction was able to improve the contaminant removal efficiency while minimizing solvent requirement.
Introduction Surfactant-enhanced aquifer remediation (SEAR) is a wellestablished technology for cleaning up subsurface nonaqueous phase liquid (NAPL) contamination. Anionic surfactants are often considered for SEAR to minimize surfactant sorption losses (1). Because of the costs of surfactant and the hazardous nature of contaminants, removal of contaminants from surfactant solutions is necessary from both economic and environmental perspectives. Air-stripping, which can effectively remove volatile organic compounds from surfactant solutions (2-4), is not eco* Corresponding author phone: (405) 325-4273; fax: (405) 3254217; e-mail:
[email protected]. † Present address: Rm. 181 EWRE Bldg., 1351 Beal Ave., Department of Civil and Environmental Engineering, The University of Michigan, Ann Arbor, MI 48109-2125. 10.1021/es002057r CCC: $20.00 Published on Web 06/02/2001
2001 American Chemical Society
nomically viable for the removal of low-volatility organic compounds from surfactant solutions. Solvent extraction has been tested as an alternative to remove semi- and nonvolatile organic compounds from surfactant solutions (5-7). However, more detailed studies are needed to identify the interactions among the contaminant, extracting solvent, and surfactant in a solvent extraction system and to quantify the impacts of factors influencing the contaminant removal. The objectives of this research were to identify and quantify the factors influencing solvent extraction of micellarsolubilized contaminant and to optimize the continuous extraction process. The fundamental hypothesis directing this research was that the competitive partitioning of a contaminant among the micellar pseudo-phase, solvent phase, and water in the extraction system controls the contaminant removal that can be achieved by solvent extraction.
Extraction Model Development Surfactants are amphiphilic, surface-active molecules, consisting of both hydrophilic and hydrophobic moieties. Above a solution concentration known as the critical micelle concentration (cmc), surfactant molecules aggregate together to form aggregates known as micelles. Surfactant molecules in micelles are oriented such that their hydrophilic moieties shield the hydrophobic moieties from the water environment. The resulting hydrophobic micellar core allows surfactant solutions to increase the solubility of hydrophobic organic contaminants, by causing the contaminants to partition into the micellar core. Solvent extraction involves reversal of this process, by creation of a driving force (the solvent) to compete with the micellar core for the contaminant. Figure 1 shows the distribution of surfactant and contaminant molecules in a solvent extraction system. Surfactant molecules exist as monomers and micelles in the aqueous phase and as partitioned monomers in the solvent phase, whereas contaminant molecules exist in extramicellar (watersolubilized) and micellar-solubilized forms in the aqueous phase, as well as in solvent-solubilized form in the solvent phase. The distribution of species in different phases are related by corresponding partitioning processes. The aqueous monomer concentration is constant at the cmc when the total surfactant concentration exceeds it. Due to the anionic nature of the surfactants studied, surfactant partitioning into the solvent phase is assumed to be negligible. Partitioning equilibrium exists between extramicellar and micellarsolubilized contaminant in surfactant solution and also between extramicellar and solvent-solubilized contaminant when an external solvent phase is present. The solvent can be solubilized into the micellar surfactant solution, introducing extramicellar and micellar-solubilized solvent and further complicating the above distribution equilibria. In the following model development, concentrations of surfactant, contaminant, and solvent (oil) species are represented by the superscripts s, c, and o, respectively, and the pure water phase, micellar pseudo-phase, aqueous phase (including water phase and micellar pseudo-phase), and solvent phase are represented by the subscripts w, m, a, and o, respectively. The substantial solubilization enhancement of organic compounds in micellar surfactant solutions results from the partitioning of the compounds in micelle cores (micellar solubilization). The molar solubilization ratio (MSR) describes the hydrocarbon solubilization capacity of a micellar surfVOL. 35, NO. 14, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2995
FIGURE 1. Interactions among surfactant, contaminant, and solvent and corresponding species distributions in solvent extraction system: (a) surfactant and solvent interactions; (b) surfactant, contaminant, and solvent interactions. actant solution (8-11) and is defined as ms MSR ) (Cmaxc - Cwc a w )/Ca
(1)
where Cmaxc is the maximum contaminant solubilization a capacity of a given surfactant solution, Cwc w is the contaminant water solubility, and Cms is the micellar surfactant a concentration in solution, that is, the total surfactant concentration minus the monomer concentration (cmc). The partition equilibrium between extramicellar and micellarsolubilized contaminant is described by a contaminant micelle-water partition coefficient (Kcm) as
Kcm
)
wc Xmc m /Xw
(2)
where Xmc m is the mole fraction of micellar-solubilized contaminant in the micellar pseudo-phase and Xwc w is the mole fraction of extramicellar contaminant in the water phase. Kcm has been observed to be independent of total aqueous contaminant concentration and can be estimated from MSR as (8, 12)
Kcm )
(
)
MSR 55.556 1 + MSR Cwc w
(3)
The extramicellar contaminant readily partitions into the solvent phase upon contacting an extracting solvent, and the contaminant solvent-water partition coefficient (Ko) is wc Ko ) Xoc o /Xw
(4)
where Xoc o is the mole fraction of contaminant in the solvent phase. The micellar-solubilized contaminant molecules are shielded from the solvent phase by the surfactant micelles, and they must partition out of the micelle into the water phase before partitioning into the solvent phase (6, 13). The activities of contaminant molecules in water, micellar pseudophase, and solvent phase are equal at equilibrium, and contaminant removal is controlled by the competitive partitioning of the contaminant among the pseudo-ternary phases (solvent, micellar, and water). 2996
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 14, 2001
The extracting solvent can also be solubilized into the surfactant micelles during extraction, which affects the contaminant partitioning capacity of the micelles as its nonpolar organic content is increased (14). The increased mass fraction of nonpolar organic groups interacting with contaminant molecules in the micelles is expected to lead to decreased partitioning of contaminant into the external solvent phase. By taking this into consideration and assuming that the contaminant micelle-water partition coefficient (Kcm) is independent of solvent solubilization, Kcm can be reformulated as
Kcm )
Xmc m Xwc w
)
Xmc a ms mc mo Xwc w (Xa + Xa + Xa )
(5)
where Xms a is the mole fraction of surfactant in micellar form in the aqueous phase and is treated as constant because of the negligible anionic surfactant partition loss and the fact that the solubilization of solvent will produce only minimal increase of aqueous mass and Xmc and Xmo are the mole a a fractions of micellar-solubilized contaminant and solvent in the aqueous phase. Similarly, the solvent micelle-water partition coefficient (Kom) is expressed as
Kom )
Xmo m Xwo w
)
Xmo a ms mc mo Xwo w (Xa + Xa + Xa )
(6)
where Xmo m is the mole fraction of extracting solvent in the micellar pseudo-phase and Xwo w is the mole fraction of extramicellar solvent in the water phase. The solubilization of solvent in micelles eventually increases the contaminant solubilization potential of the micellar pseudo-phase, and the mole fraction of micellar-solubilized contaminant remaining in the aqueous phase at equilibrium can be derived from eqs 5 and 6 as
Xmc a )
ms KcmXwc w Xa c wc (1 - KomXwo w - KmXw )
(7)
Because the extramicellar contaminant concentration cannot
TABLE 1. Surfactants, Contaminants, and Extracting Solvents Used in the Extraction Study and Corresponding EACNs surfactant
contaminanta and EACN
extracting solventa and EACN
SDBS (sodium dodecyl benzenesulfonate) Aerosol MA (sodium dihexyl sulfosuccinate)
PCE (tetrachloroethylene), 2.9b hexane, 6 octane, 8
hexane, 6 octane, 8 decane, 10 dodecane, 12 tetradecane, 14 hexadecane, 16 octadecene, 18 squalane (2,6,10,15,19,23hexamethyltetracosane), 30
a
Purchased from Aldrich Chemical Co. (Miwaukee, WI).
b
From ref 17.
TABLE 2. Fundamental Surfactant Properties surfactant
av MW
designed molecular formula
cmc, mg/L
1000a SDBS 348 C12H25C6H4(SO3Na) Aerosol MA 376 C6H13OOCH2CH5200c (SO3Na)COOC6H13 c
active % supplier
b d
100 80
a Measured in this study. b Aldrich Chemical Co. (Milwaukee, WI). From ref 23. d CYTEC Industries, Inc. (Willow Island, WV).
be easily measured whereas the total contaminant concentration is of concern in decontamination, a contaminant solvent-solution partition coefficient (Kp), expressed in terms c of Xms a , Ko, and Km, can be represented as
Kp )
Xoc o Xtc a
≈
KoXwc w mc (Xwc w + Xa )
{
) KcmXms a
Ko 1 -
wc o wo [1 + Kcm (Xms a - Xw ) - KmXw ]
}
(8)
In the above equation, the mole fraction of extramicellar contaminant in the aqueous phase is approximated by the mole fraction of extramicellar contaminant in the water phase (Xwc w ) because the surfactant and contaminant molecules are negligible portions of the total solution phase. The lumped parameter, Kp, can be easily applied to find the final aqueous contaminant mole fraction (Xf) after a solvent extraction
Kp )
[
]
Msltn (Xi - Xf)
Mo + Msltn (Xi - Xf) Xf
(9)
where Xi is the initial contaminant mole fraction in the surfactant solution and Msltn and Moil are the total initial moles of the aqueous phase and solvent phase. Total moles of aqueous phase (Msltn) are assumed to be constant in the above solution becausesince the moles of contaminant transferred from the aqueous phase to the solvent phase are insignificant compared to the amount of water presented.
Materials and Methods Table 1 lists the equivalent alkyl carbon numbers (EACNs) (that is, the number of alkyl carbons in an organic molecule when it is treated equivalently to an alkane) of contaminants and extracting solvents used in this study, and Table 2 lists the fundamental properties of the surfactants studied. All chemicals were used as received, and all experiments were conducted at a room temperature of 23 ( 1 °C. Sodium dodecyl benzenesulfonate (SDBS) partitioning experiments were conducted by introducing 5 mL of solutions with SDBS concentrations below and above its cmc (see Supporting Information) and 10 mL of alkanes into 20 mL vials. The
impact of salinity on surfactant partitioning and contaminant solubilization was studied by mixing 5 mL of surfactant solutions at different salinities (see Supporting Information) with 10 mL of alkanes in 20 mL vials. The vials were shaken horizontally on a table shaker for 3 weeks and then allowed to stand for another week. Contaminant solvent-solution partitioning studies were conducted by adding 36 mL of contaminant-saturated surfactant solution and 3.6 mL of alkane solvent into I-Chem 40 mL vials (Fisher Scientific). When the effect of the solvent/solution volumetric ratio was studied, the solution volume was held at 36 mL while the added solvent volume varied from 0.6 to 6 mL (see Supporting Information). The vials were immediately sealed by polypropylene closures with Teflon fluorocarbon resin/silicone septa. The headspaces in vials were generally 14. This can be explained by the collective contribution of decreased solvent solubilization in the surfactant solution and decreased 2998
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 14, 2001
contaminant solvent-water partitioning as solvent EACN increases. Table 3 summarizes the solubilized solvent concentrations in 50000 mg/L SDBS solutions with NaCl concentrations ranging from 0 to 10000 mg/L. Solvent with a higher EACN has a lower solubilization in the SDBS solution than the one with a lower EACN, and the solubilization of solvent increases with salinity of the SDBS solution. Increased solvent solubilization reduces partitioning of contaminant out of the micelles into the extracting solvent. Although the PCE solvent-water partition coefficients (Ko) are not expected to vary much among these alkanes, they decrease slightly as the solvent EACN increases because of the increasing molecular dissimilarity between PCE and these solvents (15). Differences in KPCE in Figure 2a suggest that the solubilizap tion of these solvents in micelles significantly affects the contaminant partitioning. Figure 2b shows similar trends for hexane and octane partitioning between various solvents and 40000 mg/L Aerosol MA/50000 mg/L NaCl solution. The alkanes evaluated are more similar to octane than to hexane, whereas their solubilizations in the surfactant solution are the same. This explains why the octane solventsolution partition coefficient (KC8 p ) in a given solvent is always higher than the corresponding hexane solventsolution partition coefficient (KC6 p ). The Salager equation suggests that a solvent with a higher EACN has less potential of being emulsified or solubilized by a surfactant solution compared to one with an EACN close to that of the contaminant (16, 17). The solubilization data in Table 3 support such a relationship between solvent solubilization and its EACN. The greater the solvent EACN compared to that of the contaminant, the less the interference on contaminant partitioning from interaction between solvent and surfactant solution. Selection of an extracting solvent with an EACN much higher than that of the contaminant, especially when the surfactant solution has a high salinity, is desirable to minimize the solvent’s solubilization. Figure 2c shows PCE percent removal from 50000 mg/L SDBS solution with different solvents at a solvent/ solution volumetric ratio of 0.1. Although squalane has the highest nominal Kp among all of the solvents, it did not produce the highest PCE removal. Furthermore, the PCE removal efficiencies achieved by dodecane, tetradecane, hexadecane, octadecene, and squalane were not statistically different. This can be explained by the “mass action” of extracting solvent molecules. Although a high EACN solvent has a high nominal Kp, it has fewer molecules per unit volume to interact with those of the contaminant (disproportional increase in liquid density with molecular weight as EACN increases). Therefore, under a given solvent/solution volumetric ratio, the solvent with the highest EACN does not necessarily produce the greatest contaminant removal. The considerations of solvent solubilization in surfactant solution and its molecular weight (and density as well) should be integrated in solvent selection. Also, solvents with higher
FIGURE 3. Effect of surfactant concentration on KPCE between p tetradecane and SDBS solution. EACNs have higher melting points and viscosity, which limit their application in solvent extraction. The nominal partition coefficient, Kp, is independent of the contaminant concentration existing in the surfactant solution because the extramicellar contaminant and micellarsolubilized contaminant are related through the micellewater partitioning (Kcm), as indicated by eq 8. This was corroborated by experimental results, which showed a KPCE p value of 680 ( 38 between tetradecane and 50000 mg/L SDBS solution at various initial PCE concentrations. Effect of Surfactant Concentration. Although other studies have shown significant partitioning of nonionic surfactants into organic phases (9, 18), SDBS partitioning into the alkane solvents was undetectable under the conditions examined. This was probably due to the low cmc of SDBS and the low salt concentration range studied. Figure 3 shows the experimentally measured KPCE and the fitted p model prediction of KPCE between tetradecane and SDBS as p a function of aqueous SDBS concentration. The tetradecanewater partition coefficient of PCE (Ko) was found to be 1.55 × 105 by fitting eq 8 with experimental results. Increasing SDBS concentration was found to reduce KPCE significantly, p although the contaminant solubilization capacity of the solution increased. Because the product of Kcm and Xms a represents the overall contaminant affinity of the micellar pseudo-phase relative to water, whereas Ko represents the contaminant affinity of the solvent phase relative to water, Kp decreases with increasing micellar mass. In addition, more solvent is solubilized in micelles with increase in surfactant concentration, which will further increase the contaminant retained by the micellar pseudo-phase after solvent extraction. Higher surfactant concentration means a greater “sink” for contaminant to stay in rather than partition out and results in lower contaminant removal efficiency and greater solvent solubilization loss. Effect of Surfactant Solution Salinity. When an electrolyte is added to an ionic surfactant solution, the surfactant cmc decreases with increasing counterion concentration, as indicated by the Corrin-Harkins equation (19, 20). At the same time, the contaminant solubilization capacity of the solution increases as the micelles transition from spherical to rodlike or disklike shapes caused by closer packing of the hydrophilic moieties (21, 22). Adding electrolyte to increase the solubilization capacity of ionic surfactant solutions, which greatly reduces the amount of surfactant required to solubilize the target NAPLs, is a common practice in SEAR. Figure 4a shows that the PCE solubilization capacity of 50000 mg/L SDBS increased almost linearly with increases in NaCl concentration from 0 to 10000 mg/L, resulting in an almost 2.5-fold increase in solubilized PCE concentration. The
FIGURE 4. Effect of NaCl concentration on maximum PCE solubilization capacity of 50000 mg/L SDBS solution, KPCE between p tetradecane and solution, and correlation between KPCE and p solution maximum PCE solubilization capacity: (a) PCE maximum solubilization capacity vs NaCl concentration; (b) measured and predicted KPCE vs NaCl concentration; (c) measured and predicted p KPCE (without considering tetradecane solubilization in SDBS p micelles) vs maximum PCE solubilization capacity. increased contaminant solubilization capacity of the surfactant solution results from an increase in MSR and Kcm at high salinity, which translates into an increased contaminant affinity of the micellar pseudo-phase (eqs 1 and 3). At the same time, the solubilization potential of the micelles toward the extracting solvent also increases with salinity (as shown in Table 3), which will further decrease the contaminant partitioning out of the micelles because of the increased nonpolar organic content of micelles. Therefore, the partitioning of contaminant into the solvent is expected to decrease with increasing solution electrolyte concentration. Figure 4b shows both observed and model-predicted KPCE values between tetradecane and 50000 mg/L SDBS p VOL. 35, NO. 14, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2999
TABLE 4. Measured and Predicted Tetradecane Solubilization in 40000 mg/L Aerosol MA Solution at Different NaCl Concentrations NaCl, mg/L
maximum solubility, mg/L model predicted, mg/L
20000
22000
24000
26000
28000
2726 ( 172 3097
3066 ( 107 3352
3334 ( 214 3499
3557 ( 177 3534
3774 ( 106 3573
FIGURE 5. Effect of NaCl concentration and solvent solubilization on KPCE between tetradecane and 40000 mg/L Aerosol MA solution. p solutions at different NaCl concentrations. Ko was assumed to be constant at 1.55 × 105 in this study, as changes of Ko with salinity would be expected to be negligible for most hydrophobic organic compounds. The cmc of SDBS was assumed to be constant at 1000 mg/L with NaCl addition, although the actual cmc of SDBS would decrease slightly with increasing electrolyte concentration, according to the Corrin-Harkins equation (20, 21). If the cmc is ignored, variations can be justified because of the relatively high total SDBS concentration and the narrow salinity range studied. Figure 4c shows the measured KPCE and model predictions p for 50000 mg/L SDBS solutions at different NaCl concentrations without considering the solubilization of tetradecane in them. The agreements between measured KPCE values and p model predictions are very encouraging even when micellarsolubilized tetradecane concentrations were ignored, which can be explained by the relatively low micellar solubilization of tetradecane in these solutions. However, it is also observed that the effect of tetradecane solubilization on PCE removal becomes more distinct as the solution salinity increased. between tetradecane and 50000 The overprediction of KPCE p mg/L SDBS solution increased from 5% at 1000 mg/L NaCl to 20% at 10000 mg/L NaCl when tetradecane micellar solubilization was neglected. The PCE solubilization capacity of 40000 mg/L Aerosol MA solutions was also found to increase almost linearly from 4500 to 62500 mg/L with the addition of NaCl from 0 to 28000 mg/L. At the same time, KPCE between tetradecane p and the solutions decreased from around 720 to 120. Figure 5 shows the effects of hypothetical tetradecane solubilization in the 40000 mg/L Aerosol MA solution, assuming the cmc of Aerosol MA is constant at 1000 mg/L over the salinity range. If the contribution of micellar-solubilized tetradecane is assumed to be constant, KPCE decreases dramatically with p increases in the maximum PCE solubilization capacity of the solution, resulting from the increased contaminant micellewater partition coefficient at higher electrolyte concentration. With increased tetradecane solubilization in the surfactant solution, KPCE is also substantially reduced. In solutions with p high PCE solubilization capacity (high salinity), neglecting 3000
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 14, 2001
FIGURE 6. Effect of solvent/solution volumetric ratio on PCE percent removal from 50000 mg/L SDBS solution by extraction with tetradecane. solvent solubilization can lead to significant overprediction of contaminant removal. Table 4 compares the predicted tetradecane concentrations in Aerosol MA solutions at NaCl concentrations from 20000 to 28000 mg/L to produce the reductions in KPCE p based on fitting eq 8 with experimental data and the measured tetradecane solubilization capacities of these solutions. The experimental and predicted results show reasonably good agreements. However, eq 8 predicted even higher solubilized tetradecane concentrations in 40000 mg/L aerosol MA solution with NaCl concentrations
