|Home | About | Journals | Submit | Contact Us | Français|
Laboratory experiments and mathematical modeling were conducted to examine the influence of a hydroxypropyl-beta-cyclodextrin (HPCD) solution on the dissolution of single- and three-component organic liquids. The results of batch experiments showed that HPCD-enhanced solubilization of the organic-liquid mixtures was ideal (describable using Raoult’s Law), and that solubilization-enhancement factors were independent of mixture composition. Addition of the HPCD solution to columns containing residual saturations of the organic liquid enhanced the dissolution and removal of all three compounds in the mixture. The results of the column experiments and multicomponent rate-limited dissolution modeling suggest that solubilization was ideal for both water and cyclodextrin flushing. Concomitantly, the mass-flux reduction versus mass removal behavior was ideal for all experiments. Mass transfer was increased for HPCD solubilization relative to the water flushing due to solubility and concentration-gradient enhancement. Organic-liquid composition did not significantly impact mass transfer coefficients, and fractional mass removal behavior during HPCD solubilization was nearly identical for each compound whether present as a single component or in a mixture. Additionally, mass transfer coefficients for aqueous and HPCD solubilization for single and multicomponent mixtures were not statistically different upon normalizing by the solubility enhancement factor.
The use of enhanced solubilization agent (ESA) solutions, including cosolvents, surfactants, and cyclodextrins, to increase mass removal is one option for remediation of source zones contaminated by organic liquids. Many organic liquids are comprised of multiple compounds, and solubilization of multicomponent organic liquids may be complicated by nonideal behavior (e.g., Rostad et al., 1985; Borden and Kao, 1992; Chen et al., 1994; Lesage and Brown, 1994; Whelan et al., 1994; Adeel et al., 1996; Mukherji et al., 1997; McCray and Dugan, 2002, Burris et al., 2006). Previous research evaluating ESA applications has focused primarily on factors of mobilization and solubilization, with few investigations of organic liquid compositional impacts (Khachikian and Harmon, 2000; Oostrom et al., 2006). Additionally, very little effort has been devoted to specifically examining mass-flux-reduction/mass-removal behavior (e.g., Stroo et al., 2003; Brooks et al., 2004; NRC, 2004; Parker and Park, 2004; Phelan et al., 2004; Soga et al., 2004; Falta et al., 2005; Jawitz et al., 2005; Fure et al., 2006; Brusseau et al., 2007; Brusseau et al., 2008; DiFilippo and Brusseau, 2008) for multicomponent organic liquids (e.g., D’Affonseca et al., 2008).
Hydroxypropyl-beta-cyclodextrin (HPCD), the solubilization agent evaluated in this paper, is a glucose-based molecule comprising a hydrophobic interior cavity into which compounds can partition, thereby increasing their apparent solubilities (Wang and Brusseau, 1993; Ko et al., 1999; McCray et al., 2000; Badr et al., 2004; Hanna et al., 2004; He and Yalkowsky, 2004; Viglianti et al., 2006; Yang et al., 2006; Caro et al., 2007). Previous investigations involving cyclodextrin have primarily considered single-component organic liquids (Wang and Brusseau, 1993; Boving et al., 1999; Boving and Brusseau, 2000, Boving et al., 2000; McCray et al., 2000; Tick et al., 2003; Cai et al., 2006). However, there is interest in the impact of cyclodextrin on organic-liquid mixtures (McCray and Brusseau, 1998; 1999; Viglianti et al., 2006; Yang et al., 2006). The purpose of this paper is to investigate the dissolution and cyclodextrin-enhanced solubilization of a three-component organic liquid. Batch and column aqueous dissolution and HPCD-enhanced solubilization experiments were conducted, and the measured concentrations were compared to concentrations predicted using Raoult’s Law to evaluate the ideality of solubilization. Model simulations and mass-flux reduction versus mass removal analyses were used to evaluate multicomponent organic-liquid mass-transfer behavior.
Hydroxypropyl-beta-cyclodextrin (90% purity technical grade with an average molecular weight of 1,363 g mol−1 from Cerastar USA, Inc., Hammond, Indiana, Lot #F8028) and distilled-deionized NANOpure (Series 550, Barnstead Thermolyne Corp., Dubuque, Indiana) water were used for all experiments. Toluene, ethylbenzene, and butylbenzene, common constituents of gasoline, were selected as representative aromatic constituents, and were obtained from Aldrich Chemical Company (greater than 99% purity; Milwaukee, WI). The two- and three-component organic-liquid mixtures were prepared with equal volume fractions by syringe injection into a headspace-free glass vial. The mass of each compound added was determined gravimetrically, and masses and molecular weights for individual compounds were used to calculate mole fractions. Stainless-steel chromatography columns (7 cm long stainless-steel tubing with 3 cm2 cross-sectional area) were used for the dissolution experiments (Alltech). Each column was packed with a well-sorted natural sand (Accusand, North Kato Supply, MN). This sand has a median grain diameter of 0.323 mm (40/50 mesh) and uniformity coefficient of 1.17.
Batch experiments were conducted to measure apparent solubilities of the compounds in HPCD solutions. Experiments were conducted with single-component organic liquids as well as two- and three-component organic-liquid mixtures. The organic liquids were prepared (5 ml) and injected (Gastight Hamilton Company syringe, Reno, Nevada) into glass vials (VWR Trace-Clean 25 ml clear vials with open top caps and Teflon lined septa, Brisbane, CA) with water or solutions (20 ml) with various concentrations of HPCD (10, 15, and 20% by weight). The experiments were conducted in triplicate without headspace. The vials were placed on a Orbit Model shaker table (Lab-Line Instruments, Inc., Melrose Park, Illinois) set to 200 RPM for 72 hours to achieve equilibration (as previously determined), and then centrifuged (Beckman GP Centrifuge, Palo Alto, CA) at 1500 RPM for 30 minutes to effect phase separation. The aqueous or HPCD solutions were sampled with a gastight syringe through the Teflon septa of the vials, and sealed in headspace vials for immediate chemical analysis. Statistical confidence intervals (95% C.I.) were calculated from the triplicates.
Water and cyclodextrin-enhanced dissolution of the single- and three-component organic liquids were examined by conducting a series of column experiments. Before each experiment, all materials and solutions were autoclaved, and the columns were packed in a sterile hood (HEPA filtered forced air) to avoid biodegradation during experiments. The columns were packed incrementally with air-dried porous media to establish a consistent porosity and bulk density, and saturated with autoclaved NANOpure water for approximately 80 pore volumes (until constant mass was achieved). Upon completion of column packing and saturation, a non-reactive tracer test and a three-component reactive tracer test were conducted to evaluate hydrodynamic and chemical transport processes (sorption), respectively.
The nonreactive tracer test was initiated by injecting a pulse of tracer solution containing pentaflourobenzoic acid (Aldrich Chemical Co., Inc., Milwaukee, WI) into the column for approximately 11 pore volumes, after which tracer-free water was injected to elute the tracer solution from the column. The pentaflourobenzoic acid concentrations in the column effluent were analyzed with a flow-through UV-VIS Spectrophotometer (Shimadzu, Japan) at 254 nanometer wavelength. A similar experiment was conducted with a tracer solution containing toluene, ethylbenzene, and butylbenzene. The three-component mixture solution was prepared by equilibration with the equal volume organic-liquid mixture and NANOpure water. The solution was injected for 22 pore volumes, and then eluted by switching the injection solution to tracer-free NANOpure water. Samples were collected from the column effluent and the injection solution during the experiment with a gastight syringe and placed in headspace vials for immediate chemical analysis. The column was cleaned and repacked prior to each experiment (sorption and organic-liquid dissolution). All miscible-displacement experiments were conducted at 25(±2)°C.
Several organic-liquid dissolution experiments were conducted, and experimental conditions are listed in Table 1. Single-component and equal-volume, three-component organic-liquid samples were prepared and injected into columns using the following procedure. First, organic-liquid mixtures were created by adding equal volumes of each component to headspace-free glass vials, and allowing the compounds to mix for 24 hours. Second, 10 ml of the organic liquid were injected (0.02 cm3 min−1) with a syringe pump (Sage Instruments Model 335) into the top of the vertically-oriented column to maintain stable displacement. Third, an aqueous solution of NANOpure water equilibrated with the organic-liquid mixture was introduced into the column from the bottom to displace the mobile organic liquid. The water was pumped at the equivalent pore-water velocity of 8.5 cm hr−1 for 1 hour and 45 minutes (approximately 2 pore volumes), followed by a pore-water velocity of 85 cm hr−1 for another 45 minutes (approximately 10 pore volumes). The capillary number for this displacement process was approximately 10−6, consistent with prior research. The residual saturation of organic liquid was determined by measuring the volume of organic liquid injected and displaced during this procedure. After the establishment of residual saturation, the column was flushed (Shimadzu LC-10AS HPLC piston pump, Japan) in an upward direction with either NANOpure water or cyclodextrin solution (10% by weight) at a pore-water velocity of 10 cm hr−1 (Table 1). For each experiment, effluent samples were collected through time with a gastight syringe for immediate chemical analysis.
The chemical analysis of toluene, ethylbenzene, and butylbenzene was performed using gas chromatography (Hewlett-Packard 5890) with flame-ionization detection (Shimadzu GC-17A, Japan) and a headspace autosampler (Tekmar 7000). Aqueous and HPCD samples were prepared by adding 5 ml of the samples into empty headspace vials (Kimble) sealed with open caps and Teflon-lined septa. Standards prepared in NANOpure water and HPCD solutions depending on the experiment were used for quantification, and the quantifiable method detection limit was approximately 0.08, 0.1, and 0.05 mg L−1 for toluene, ethylbenzene, and butylbenzene, respectively.
Quantification of equilibrium solubilization of chemical mixtures has been developed from thermodynamic analysis (Burris and MacIntyre, 1985). Organic-liquid mixtures comprised of compounds with similar size and functional groups are considered ideal, and ideal equilibrium dissolution is described by Raoult’s Law (Banerjee, 1984; Burris and MacIntyre, 1985):
where the subscript A refers to aqueous phase and N refers to organic liquid; superscript o signifies the pure phase (single-component); SA is the aqueous concentration of the component in equilibrium with the organic-liquid mixture; S°A is the pure-compound aqueous solubility; and XN is the mole fraction of the component in the organic liquid.
The ESA solution enhancement of single-component organic-liquid solubilities has been defined as (e.g., Ji and Brusseau, 1998):
where E is the enhancement factor and S°E is the pure compound enhanced (apparent) solubility. For cyclodextrin solutions, the enhanced equilibrium solubility is a function of the cyclodextrin concentration:
where CCD is the aqueous concentration of HPCD, and Kcw is the partition coefficient for each compound between the HPCD and water. Thus, the enhancement factor (E) can be expressed as a function of the HPCD concentration and the compound specific partition coefficient:
if the ESA and organic-liquid mixture behave ideally (McCray and Brusseau, 1999).
A mathematical model (Ji and Brusseau, 1998) was previously developed to simulate organic-liquid dissolution and dissolved-phase transport of organic contaminants in porous media. This model considers the enhanced solubilization of single-component organic liquid with various ESAs. The model was modified for this investigation to incorporate ideal multicomponent organic-liquid solubilization, wherein the organic-liquid components dissolving into the flushing solution have apparent solubilities that follow Raoult’s Law.
This numerical model was used to simulate the effluent data from the organic-liquid dissolution column experiments, and organic-liquid dissolution mass transfer coefficients were quantified. The commonly used, first-order approach was used to represent interphase mass transfer of compounds between an immobile organic-liquid source and a mobile water phase (e.g., Miller et al., 1990; Powers et al., 1992; 1994; Ji and Brusseau, 1998):
where Mn is the mass concentration of each organic-liquid component, kf is the mass-transfer coefficient, ao is the specific organic-liquid/water interfacial area for mass transfer, SA is the equilibrium aqueous phase concentration at the interface between the organic liquid and water, and CA is the bulk aqueous phase concentration. A lumped dissolution rate coefficient kL, the product of ao and kf, is commonly used because the specific interfacial area is not usually known.
The model was also developed to quantify the effect of an ESA injection on organic-liquid dissolution by including the enhancement factor (E) in the driving force equation (Ji and Brusseau, 1998):
where CE is the bulk aqueous phase ESA-enhanced concentration for that organic-liquid component. The reduction in interfacial area as organic liquid is removed over time is accounted for by modifying the dissolution rate coefficient from the initial value based on changes in the organic liquid volumetric content (θN) (e.g., Powers et al., 1994). The transient behavior of the dissolution rate coefficient is expressed as:
where β4 is an optimized parameter. The model updates the transient dissolution rate coefficient for each component based on the change in each component’s organic-liquid volume [θN(t)], which accounts for compound-specific changes in organic liquid interfacial area with time during the dissolution process. Each single blob of multicomponent organic liquid during dissolution is conceptualized to have a shrinking core for each compound within the organic-liquid matrix, with a compound specific θN(t) decreasing over time as that compound becomes depleted.
The modeling of organic-liquid dissolution with an ESA solution requires the coupled solution of the transport equations for both the ESA compound and each of the compounds contained in the organic liquid, with a source term that describes the dissolution. The transport of cyclodextrin (subscript CD) in solution can be described with the nonreactive form of the advection-dispersion equation (Ji and Brusseau, 1998):
where D is the dispersion coefficient, and v is the interstitial velocity. The transport of each organic-liquid component dissolved in water (subscript A) is given by:
where CA is the concentration of each organic-liquid compound in aqueous solution (CE replaces CA as the component concentration for ESA solutions), and S1 and S2 are the instantaneous and rate-limited sorbed concentrations of the compounds (Ji and Brusseau, 1998).
Application of this model requires knowledge of several parameters, which were obtained independently except for the dissolution rate coefficients. The nonreactive tracer (PFBA) test was used to determine the dispersivity, the three-component tracer sorption experiment was used to determine compound specific retardation coefficients and sorption rate coefficients, and batch experiments were used to measure single-component and organic-liquid mixture solubilities in water and the HPCD solution. Temporal moment analysis was used to calculate the mass balance and the retardation factor for all tracer experiments. The mean travel times were calculated using the normalized 1st moments, and the ratio of the reactive and non-reactive tracer travel times was used to calculate the retardation factor (data not shown). Travel times for both toluene and ethylbenzene were not significantly different from that of the nonreactive tracer, thus exhibiting minimal retardation. Conversely, butylbenzene did exhibit minor retardation (R = 1.6). The sorption properties were quantified and considered in the organic-liquid dissolution modeling.
The organic-liquid aqueous and HPCD enhanced dissolution rate coefficients were compared to previous empirical mass-transfer correlations for the Sherwood number (Sh) and Reynolds number (Re) (Powers et al, 1994). The Reynolds number is defined as:
where ρW is the density of water, d50 is the median grain size, and μW is the viscosity of water, and the Sherwood numbers were calculated from the initial dissolution rate coefficients:
where DL is the bulk water diffusion coefficient for each organic-liquid compound.
The relationship between mass removal and mass-flux reduction was examined for the dissolution experiments. Fractional mass removal was defined as the mass removed relative to the initial contaminant mass, and it was plotted as a function of fractional removal time for each of the experiments. The fractional removal time was defined as the experimental time relative to the total mass removal time (a cutoff of 1 mg L−1 was used for consistency). Fractional mass-flux reduction (MFR) was calculated using:
where J is the mass flux, Q is the volumetric flow rate, C is concentration (represents CE or CA as the component concentration above), and the subscripts i and f indicate initial and final times, respectively.
The solubility enhancement factors for toluene, ethylbenzene, and butylbenzene were obtained for HPCD solutions (0 to 20% by weight) in equilibrium with the one-, two-, and three-component organic-liquid samples. Solubility enhancement factors as a function of HPCD concentrations are presented in Figure 1. As expected, the solubility enhancement increased linearly with increasing HPCD concentration, and was inversely proportional to the aqueous solubility for each compound (Wang and Brusseau, 1993). The 95% confidence interval error bars overlap for the single- and three-component enhancement factor results, which suggests that partition coefficients (KCW) for the distribution of the compounds between HPCD solution and water were independent of the organic-liquid mixture composition for all three compounds. For example, the KCW for toluene obtained for both the single-component and three-component organic liquids was 0.06 L g−1, and the enhancement factor for 10% HPCD (the column experiment concentration) was 7.7 and 8.0 for the single- and three-component organic liquids, respectively (Figure 1 and Table 1). These results suggest that HPCD-induced solubility enhancement was ideal for each compound.
The elution curves for aqueous flushing of the single-component and three-component organic-liquid column systems are presented in Figure 2 (modeling results are discussed in Section 4.3). The majority of the toluene was removed after approximately 300 pore volumes for the single-component organic liquid. The initial aqueous concentration was not statistically different from the single-component solubility, which suggests that the initial phase of dissolution was at equilibrium for this system. For the three-component organic-liquid experiments, the initial aqueous concentrations for each of the three components were approximately equal to the concentrations predicted by Raoult’s Law. This indicates initial dissolution was ideal. As expected, toluene was removed from the organic-liquid mixture first due to its higher solubility, with a majority of the toluene removed after approximately 500 pore volumes. As toluene was depleted, the concentrations, and removal rate, of ethylbenzene and butylbenzene increased due to their increasing mole fractions.
The effluent concentration curves obtained from flushing the single-component and three-component organic-liquid column systems with HPCD solution are presented in Figures 3 and and4,4, respectively. The effluent concentrations at the beginning of the two sets of experiments matched the aqueous single- and three-component solubilities, respectively. After one pore volume, the concentrations increased to levels consistent with the enhanced apparent solubilities predicted with Equation 2 for the single-component organic liquid and Equation 5 for three-component organic liquid, incorporating the batch-determined enhancement factors. This similarity suggests that the initial phase of dissolution was ideal for these systems.
The majority of the toluene was removed through solubilization in the HPCD solution after approximately 150 pore volumes for the single-component organic liquid and less then 80 pore volumes for the three-component organic liquid, which is a significant reduction compared to aqueous dissolution. After toluene was depleted (Figure 4), the concentrations, and removal rate, of ethylbenzene and butylbenzene increased due to their increasing mole fractions, as was observed for the aqueous multicomponent dissolution experiment. Concentration decreases were observed in each experiment over time as organic liquid was removed through solubilization.
Fractional mass removal as a function of fractional removal time (the experimental time relative to the total mass removal time with a cutoff of 1 mg L−1 for consistency) was compared for each of the experiments (Figure 5a). The results indicate that mass removal of single and multiple-component organic liquid followed the same trend for each of the HPCD solubilization experiments. However, the HPCD solubilization experiments had significantly more rapid mass removal compared to the aqueous dissolution experiments, due to the solubility enhancement. Thus, the impact of HPCD solubility enhancement on mass removal was significantly more dominant than the organic-liquid mixture compositional changes. The fractional mass-flux reduction (MFR) versus fractional mass removal (MR) behavior (Figure 5b) was generally ideal for both HPCD solubilization and aqueous dissolution for both single and multiple-component organic-liquid experiments. This indicates that the enhanced solubilization had no significant impact on the MFR-MR relationship for this ideal system, and the overall fluid flow and dissolution behavior was identical for the experiments.
Numerical modeling was used to simulate the dissolution and solute transport observed in each of the experiments. A comparison of the simulated and measured elution data for the aqueous dissolution experiments is presented in Figure 2. The model simulation captures both the steady state and transient dissolution behavior for both the single- and three-component experiments. Parameter values associated with the modeling, including calibrated initial dissolution rate coefficients are presented in Table 1. The dissolution rate coefficients (kL) are similar for all aqueous experiments, which suggests that organic-liquid composition did not have a significant effect on kL values. For example, the dissolution rate coefficients for toluene were 22±5 hr−1 for the single-component experiment and 21±4 hr−1 for the three-component experiment. There was a minor variation of kL values between components where the values increased with increasing solubility. The Sh numbers (Table 1) were comparable to values obtained with previously reported correlations (Miller et al., 1990; Powers et al., 1994). Additionally, values for the optimized parameter (β4) that modifies kL as a function of organic-liquid content during dissolution (Equation 8) were not statistically different for the aqueous experiments.
Simulated elution curves for HPCD-enhanced solubilization are presented in Figure 3 and Figure 4, and the parameter values are listed in Table 1. The simulations match the observed data relatively well for each of the three single-component organic-liquid experiments (Figure 3). The results for the three-component HPCD-enhanced solubilization simulation are compared to the measured data in Figure 4. The model simulations match the measured data relatively well, capturing the general behavior with respect to the initial aqueous concentrations, the increased concentrations due to the solubility enhancement, and the increased concentrations of ethylbenzene and butylbenzene with changing mixture mole fraction. However, the peak arrival times for the simulations deviate from the observed times for ethylbenzene and butylbenzene. A minor, although similar, deviation between simulated and observed peak arrival times may be observed in the aqueous three-component results (Figure 2). These deviations may have resulted from transient mass-transfer mechanisms, such as intra-organic-liquid diffusion, that have not been incorporated into the model.
The initial dissolution rate coefficients determined for HPCD flushing were significantly larger than the values determined for aqueous flushing (Table 1). However, the solubility-enhancement effect is not accounted for in such a comparison (Ji and Brusseau, 1998). The enhancement-factor normalized initial dissolution rate coefficients (kL/E) reported in Table 1 do account for the increase in concentration gradient, and it is observed that the kL/E values for the HPCD experiments are similar to the kL values obtained for the aqueous experiments. The values for the optimized parameter (β4) were statistically similar for the HPCD experiments, as well. These results suggest that the HPCD solution increased mass removal primarily through increases in solubilities and the concentration gradients, and had minimal impact on the mass-transfer process (e.g., interface configuration) as illustrated by the nearly identical temporal kL/E functions for the experiments (compare Figures 6a and 6b). This behavior is supported by a previous comparison of aqueous and cyclodextrin-enhanced dissolution rate coefficients for a single component organic liquid (Boving et al., 2000).
This investigation demonstrated significant solubility enhancement and increased mass-removal rates for a three-component organic liquid with the application of 10% HPCD solution. Multicomponent solubility enhancements were inversely proportional to the single-component aqueous solubilities, and followed Raoult’s Law. The mass-flux-reduction/mass-removal behavior was similar for experiments conducted with single and three-component organic liquids for both aqueous or HPCD flushing solutions. Dissolution rate coefficients determined from numerical modeling calibration for the aqueous-flushing experiments were similar to values obtained for the enhanced-solubilization experiments when the latter were normalized by the solubility enhancement factor. These results suggest that the HPCD solution had minimal impact on the interphase mass-transfer rate coefficient or interfacial area, and that the associated dissolution process was identical to that for aqueous flushing.
This research was supported in part by funding provided by the US EPA STAR Program and by the NIEHS Superfund Basic Research Program (P42ES04940). The authors gratefully acknowledge Asami Murau, Wei Ji, Guiyun Bai, and Zhuhui Zhang for their contributions, and assistance from the Brusseau Transport Group is also appreciated.