Environ. Sci. Technol. 2001, 35, 4457-4462
Gaseous Transport of Volatile Organic Chemicals in Unsaturated Porous Media: Effect of Water-Partitioning and Air-Water Interfacial Adsorption HEONKI KIM,† M I C H A E L D . A N N A B L E , * ,‡ A N D P. SURESH C. RAO§ Department of Environmental Science, Hallym University, South Korea, Department of Environmental Engineering Sciences, University of Florida, and School of Civil Engineering, Purdue University
Laboratory experiments were conducted employing gas chromatographic techniques to evaluate the gaseous transport of volatile organic chemicals (VOCs) in water-unsaturated soil columns as influenced by interfacial (air-water) adsorption and water partitioning. VOCs [methylene chloride, tetrachloroethene (PCE), 1,1,1-trichloroethane (TCA), ethylbenzene, p-xylene, chlorobenzene] with different waterpartitioning and interfacial adsorption coefficients (airwater) were used to evaluate the theoretical basis of using these coefficients to predict the retardation factors (Rt) observed during gaseous transport. A loamy sand from Dover Air Force Base, DE, and a commercial sand were used as the column packing material to assess the effect of grain size on the air-water interfacial area (ai) and retardation at different water saturations (Sw). The ai were measured using n-alkanes. At low Sw, interfacial adsorption contributed most to the retardation for all VOCs during gaseous transport in the Dover soil which has little sorption capacity for the VOCs. As Sw increased, the fraction of Rt attributed to interfacial adsorption decreased, while that due to water partitioning increased for all of the VOCs used for this study. For the sand, with a more uniform grain-size distribution than the Dover soil, the contribution of airwater interfacial adsorption to the Rt of a VOC (p-xylene) was not as significant as that for the Dover soil due to small ai. The fractions of Rt attributed to interfacial adsorption and water partitioning were quantified. The observed Rt for the VOCs agreed well with those predicted based on the sorption coefficients and the quantities of sorption domains (Sw, ai).
Introduction Understanding the sorption mechanisms of organic contaminants in all physical domains in unsaturated soils is of great importance for the estimation and prediction of their transport and fate in the subsurface environment. Since, at many contaminated sites, hazardous organic compounds * Corresponding author phone: (352) 392-3294; fax: (352) 3923076; e-mail:
[email protected]. Present address: Department of Environmental Engineering Sciences, A.P. Black Hall, University of Florida, Gainesville, Florida 32611-2013. † Hallym University. ‡ University of Florida. § Purdue University. 10.1021/es001965l CCC: $20.00 Published on Web 10/13/2001
2001 American Chemical Society
are trapped in the vadose zone providing a source for the gaseous plume, volatile organic chemical (VOC) transport in the gaseous phase is of interest not only for the assessment of the time-dependent spatial distribution of those contaminants but also for optimizing the design of vapor-phase remediation techniques (e.g., soil vapor extraction, bioventing). Four mechanisms have been identified as causing retardation of VOCs during gaseous transport in unsaturated porous media (1-4). Sorption by the soil mineral and organic components is known to be responsible for the retardation of VOCs during gaseous transport. The magnitude of VOC sorption from the gaseous phase on the soil (solid) depends on the solubility (or Henry’s law constant) of the chemical in aqueous phase, the degree of water saturation (Sw), and the aqueous phase sorption coefficient (Kd) of the soil. Soil water is the second physical domain capable of retaining VOCs during gaseous transport. VOCs with relatively high aqueous solubilities partition into water and as a result are retained longer by the soil water. This is manifested as an increase in the observed total retardation during gaseous transport. At higher vapor pressures (non-Henry region), the third process, capillary condensation within intraparticle microporosity, may contribute to the apparent adsorption of VOCs (4). Finally, adsorption at the air-water interface can be a significant factor contributing to retardation, the specific air-water interfacial area (ai), and the interfacial adsorption coefficient (Ki) of a chemical control this retardation of the VOC. Although it has been known that a substantial mass of VOCs can accumulate at the air-water interface in unsaturated soils (5, 6), a systematic investigation of this effect on the gaseous transport of VOCs has not been performed quantitatively, primarily because of the difficulties encountered in measuring ai in water-unsaturated porous media. A gaseous interfacial tracer technique has been reported by Brusseau et al. (7) and Kim et al. (8) for the experimental measurement of ai that is exposed to a continuous gaseous phase. An aqueous interfacial tracer technique was also reported by Saripalli et al. (9, 10) and Kim et al. (11) to measure air-water [or nonaqueous phase liquid (NAPL)-water] interfacial areas using anionic surfactants as tracers, which detect the air-water interface in contact the continuous aqueous phase. The effect of interfacial adsorption on the aqueous-phase transport of surface-active organic chemicals was also reported (12, 13). In the present study, we examine the effect of air-water interfacial adsorption and water partitioning on the retardation of VOCs during gaseous transport in water-unsaturated columns packed with a soil and a sand. We evaluated, at different degrees of water saturation (Sw), the partial contributions of interfacial adsorption and other mechanisms to the observed retardation factors for several VOCs. The effect of grain size on the interfacial adsorption, hence on the retardation factor, was also demonstrated using two soils.
Theoretical Background The total retardation factor (Rt) for VOC transport via the gaseous phase can be divided to four terms, each representing a physical domain contributing to VOC retention (2, 3):
Rt ) βg + βw + βi + βd
(1)
[t ) total; g ) gas; w ) water; i ) interface; d ) solid] βg ) 1, βw )
θw aiKi FKd , βi ) , βd ) K H θg θg KHθg
VOL. 35, NO. 22, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
(2) 9
4457
where θw and θg are the volumetric water and gas contents (cm3/cm3), respectively; KH is the Henry’s law constant (dimensionless); Ki is the adsorption coefficient at the airwater interface (cm); ai is the specific air-water interfacial area (cm2/cm3); D is the bulk density of the medium (g/cm3); Kd is the sorption coefficient for the medium (cm3/g); and β represents the partial retardation factor (dimensionless). Note that linear sorption isotherms for water partitioning, air-water interfacial adsorption, and soil-sorption are assumed, as is equilibrium conditions for VOC distribution among the phases. Over a limited range of VOC concentrations, and especially for small VOC pulses introduced, the linearity assumption is acceptable. Ki values represent the linear relationship between the gaseous phase concentration (Cg, mol/cm3) and the surface excess of a chemical at the air-water interface (Γ, mol/cm2). Ki values are available in the literature (14) or can be estimated from a surface tension-Cg relationship. The Henry’s law constants (KH) for VOCs are also found in the literature (15, 16). The Henry’s law constants used in eq 2 are dimensionless
Cg ) KHCw
(3)
where Cg and Cw are the gaseous and aqueous concentrations of a VOC, respectively (mol/cm3). Since the sorption capacity of a soil for a VOC varies depending the chemical composition of the soil, it is necessary to measure the Kd value for each soil-VOC pair. Soil sorption coefficients can be estimated given published partition coefficients with respect to the organic fraction (Koc) and the soil-specific organic carbon content (foc), as follows (17): Kd ) focKoc. To investigate the relative magnitudes of βg, βw, βi, and βd as Sw varies, or for different porous media, it is useful to normalize these terms with respect to the total retardation factor, Rt.
fg )
βg , Rt
fw )
βw , Rt
fi )
βi Rt
,
fd )
βd ; Rt
∑f ) 1 j
(4)
j
Note that βg is always 1 (eq 2), while the normalized term fg (fractional retardation factor) varies depending on the magnitude of Rt. For a breakthrough curve (BTC) measured with a small pulse input of a VOC, Rt is estimated based on the mean residence time (th) for a nonreactive tracer (n) and the VOC of interest (c).
Rt )
µ′c ht c ) µ′n ht n
(5)
The mean residence time is equal to the pulse-corrected, normalized, first temporal moment µ′ calculated as (18)
∫ C(t)t dt 1 µ′ ) - t ∫ C(t) dt 2 ∞
0
∞
p
(6)
0
where C(t) is the gaseous concentration of a chemical in the effluent (mol/cm3); t is time (min); and tp is the input pulse duration (min). When the input duration is sufficiently small (i.e., a Dirac input), compared to the normalized, first temporal moment, tp may be assumed to be zero. This is indeed the case for all of the gas chromatography (GC) experiments reported in this paper.
Experimental Section Materials. Compressed methane gas with 99% purity was purchased from Aldrich Chemical Co. Tetrachloroethene (PCE), 1,1,1-trichloroethane (TCA), and straight-chain hy4458
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 22, 2001
TABLE 1. Chemical Properties of the VOCs Used in This Study (at 25 °C)
chemical methane n-hexane n-heptane n-octane n-nonane n-decane methylene chloride PCE TCA chlorobenzene ethylbenzene p-xylene
molecular KHa weight (amu) (cm3gas/cm3aq)
K ib (µm)
vapor solubility pressurea in watera (kPa) (g/m3)
16.04 86.17 100.21 114.23 128.26 142.28 84.9
27 70 90 120 200 300 0.11
0.109 0.233 0.541 1.23c 2.84c 0.186
27260 20.2 6.11 1.88 0.571 0.175 58.4
24.1 9.5 2.93 0.66 0.122 0.052 19400
165.83 133.4 112.56 106.2 106.2
0.93 1.13 0.14 0.32 0.29
0.327 0.292 1.23 2.35 2.41d
2.48 16.53 1.581 1.27 1.17
140 720 472 152 185
a Estimated from ref 15. b Data from ref 14. c Estimated based on the Ki values of n-hexane, n-heptane, and n-octane and the measured airwater interfacial areas. d Data from ref 5.
drocarbons (n-hexane, n-heptane, n-octane, n-nonane, and n-decane) were purchased from Aldrich Chemical Co. Aromatic compounds (chlorobenzene, ethylbenzene, and p-xylene) were purchased from Fisher Scientific Co. All of the chemicals used in this study were reagent grade and were used as received without further treatment. High-purity nitrogen was used as the carrier gas for the gaseous transport experiments. Chemical properties relevant to this study are listed in Table 1. Two porous media were used as column packing: (1) a loamy sand collected at 10.4 m below ground surface at Dover Air Force Base (Delaware) that consists of mostly sand with small fractions of silt and clay (called Dover soil hereafter) and (2) the clean commercial sand used in previous studies (11, 12). The porosities and particle size distributions of these two soils are presented in Table A in the Supporting Information. The Dover soil was washed with HPLC-grade water (W-4, Fisher Scientific Co.) several times to remove any debris other than soil and dried in an oven at 105 °C for 24 h before packing. Before packing, the sand was washed with HPLC-grade water and baked in an muffle furnace at 500 °C for 24 h to remove any possible organic carbon residue. Miscible Displacement Experiments. The experimental procedures used in this study were identical to those of a previous study (8) and were only slightly different from those used by others (7) where gaseous-phase transport experiments were conducted using GC systems. A stainless steel column [30.0 cm (length) × 1.0 cm (inner diameter)] was packed with the soil or sand and installed in a GC (Shimadzu, GC-14A). A bubble humidification flask was installed between the carrier gas supply and the GC injection port to prevent water loss from the column during the experiment. The column was maintained at room temperature (22 ( 0.5 °C) for all the experiments. The temperature at the GC injection port was set at 120 °C. A flame ionization detector (FID) was used for monitoring the VOC concentrations in the gaseffluent from the column. The FID temperature was set at 240 °C. The carrier gas flow rate was controlled using a needle valve to generate a constant pore-gas velocity at each water saturation about 10 cm/min throughout the experiment. The columns were packed with dry soil or sand. To achieve an intended water saturation (Sw ) θw/φ; φ-porosity) for each experiment, an equal amount of water (100-200 µL) was injected into each end of the column using a syringe. The column was then placed in an oven and heated at 105 °C for about 24 h to allow the water to distribute uniformly along the column. The column was then cooled to room temper-
ature and installed in the GC. After each experiment, another small increment of water was injected, followed by heat treatment to redistribute the water. Homogeneous water saturation along the column was verified by measuring the water contents in segments of soil samples from the column after the experiments at the highest Sw. Water saturation changes before and after a set of experiments were carefully monitored. No measurable Sw change during an experiment was found, based on the weights of the column before and after a series of experiments with the VOCs at a specified water saturation. About 10 mL of each VOC, except methane, was put into 40 mL vials, and vapor samples from the headspace were taken using a gastight syringe for GC injection. The injection volume varied between 0.5 and 10 µL depending on the vapor pressure of the VOCs at room temperature. Compressed methane gas was depressurized to the ambient atmospheric pressure prior to the injection on the soil column. The retention time and BTC for each chemical were monitored using a GC integrator (Shimadzu, CR601) and a chart recorder (Fisher Scientific Co., Recodall Series 5000) simultaneously. The first temporal moment for each measured BTC was calculated using numerical integration of eq 6. All 12 VOCs were used in the experiments with the Dover soil column to investigate the effect of chemical properties on the retardation behavior of VOCs. However, only methane, p-xylene, and n-decane were used for the clean sand column: methane as nonreactive tracer; n-decane as an interfacial tracer for ai measurement; and p-xylene as a reference compound for assessing the effect of different particle size and distribution of a medium on the transport of gaseous VOCs. A series of aqueous-phase miscible displacement experiments was conducted to measure the adsorption coefficients (Kd in eq 2) of the VOCs. Straight-chain hydrocarbons and methane were not attempted due to their low water solubilities. The experimental setup was the same as that used for a previous study (13). A stainless steel column [6.0 cm (length) × 1.0 cm (inner diameter)] was packed with the soils under water-saturated conditions. A pulse input of about 30% of the pore volume was used for this experiment. The Rt (aq) was calculated using eqs 5 and 6, and Kd was estimated from the following relationship, Rt (aq) ) 1 + (FKd)/θw,where F is the bulk density (g/cm3). Since all the experiments for Kd estimation were carried out under water-saturated conditions, interfacial adsorption at the air-water interface was assumed to be zero. At very low water saturations it is possible that the gaseous VOCs might interact with soil surface exposed to the gaseous phase or that interfacial sorption at the surface of a thin water film might be affected by the soil surface. We conducted additional gaseous transport experiments to confirm that the predicted retardation factors were due to the mechanisms described in eqs 1 and 2, with no interference by the soil surface. First, the sorption coefficients for three selected VOCs (n-hexane, methylene chloride, and TCA) for dry Dover soil were measured using an experimental setup similar to that used for the gaseous VOC transport experiments. A new column (25 cm × 0.46 cm) was packed with glass wool and a small amount of dry Dover soil was sandwiched between glass wool at the center of the column; this method is often used to determine very large sorption coefficients. Approximately 1 g of the soil was used. Using the same column, the sorption coefficient of glass wool was measured and corrected for estimating the soil sorption coefficients of the VOCs. After the dry soil column experiments were completed, the column was then equilibrated with humidified carrier gas (N2) for 24 h using the same device used for the VOC transport experiments. The final Sw was 0.029 after humidification. Transport experiments were conducted for the three
FIGURE 1. Breakthrough curves of n-nonane and chlorobenzene at different water saturations; Dover soil, pore volume - displaced volume of carrier gas, normalized with respect to the volume of gaseous phase in the soil-packed column. VOCs (n-hexane, methylene chloride and TCA) to estimate the difference between sorption on absolutely dry soil and soil with minimal water saturation.
Results and Discussion Retardation of VOCs. All of the BTCs for the VOCs were reasonably symmetric, implying both water-partitioning and interfacial adsorption processes are not significantly ratelimited under the conditions employed for the gaseous-phase transport with the soils used in this study. This suggests that the mean residence time was long enough to ensure equilibrium phase distribution. The size of the injection pulse was found to have no impact on n-heptane mean travel time, suggesting linear partitioning between air and the air-water interface (Supporting Information, Figure A). The first temporal moments, representing mean residence times, of the n-alkanes estimated from experiments with Dover soil decreased with increasing water saturation (Sw). BTCs for only n-nonane are presented in Figure 1; the other three n-alkanes were found to have the same trend as shown in Figure 1. The retardation of other VOCs decreased initially as Sw increased, reaching a minimum point, and then increased with further increase in Sw (Figure 1). The gaseous transport of other VOCs, except n-alkanes, used for Dover soil experiments, was found to have the same characteristics as chlorobenzene (Figure 1). Retardation factors of all VOCs for Dover soil are listed in Table B, Supporting Information. From the aqueous miscible displacement experiments, the Kd values for all of the VOCs were found to be insignificant for both soils used in this study. Aqueous-phase retardation factors [Rt (aq)] were no larger than 1.02 for all VOCs used in this study [e.g., Rt (aq) ) 1.02 for p-xylene, Dover soil], which corresponds to a Kd less than 0.005 cm3/g. Thus, most of the observed retardation factors of VOCs during gaseous transport resulted from either adsorption at the air-water VOL. 35, NO. 22, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4459
FIGURE 2. Observed and predicted total retardation factors of VOCs in Dover soil. The predicted partial retardation factors (β) of chlorobenzene are shown with the observed total retardation factor, Rt; solid line in all graphs - predicted Rt.
FIGURE 3. Estimated specific air-water interfacial areas of columns packed with Dover soil and clean sand in the water saturation range of 0.05-0.55. interface or partitioning into the aqueous phase or both. Solid-phase sorption from the vapor phase was determined not to be significant but is discussed in detail in a separate section. Based on the shape of the Rt(measured)-Sw relationship for VOCs shown in Figure 2, it was obvious that only air-water interfacial adsorption was the cause of the n-alkane retardation, while both air-water interfacial adsorption and water partitioning were responsible for the retardation the other VOCs. The combined effect of water-partitioning and interfacial adsorption on the VOC retardation is consistent with earlier reports (1). The total retardation factors for n-decane in the clean sand column were smaller than those from the experiments with Dover soil but followed the same trend with changing water saturation followed the same pattern. Air-Water Interfacial Areas. Since the n-alkanes have extremely low aqueous solubilities (Table 1), the Rt measured was assumed to be primarily due to the adsorption at the air-water interface. Note that low aqueous solubilities of n-alkanes limit direct interaction with the soil surface, provided that the soil surface is completely water-wet. Thus, neglecting the βw and βd terms in eqs 1 and 2, the ai values could be estimated with known Ki, θg, and measured Rt values for n-alkanes. The estimated ai values for both media decreased with increasing Sw (Figure 3). In these calculations, Ki values from Hoff et al. (14) were used for n-hexane, n-heptane, and n-octane. The ai values estimated using the Ki values of n-nonane and n-decane from the literature (14) were systematically off from the ai values estimated from the rest of the n-alkanes. Since the ai values from n-hexane, n4460
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 22, 2001
heptane, and n-octane were very consistent, we used the average ai from these chemicals as a reference for calculating new Ki values for n-nonane and n-decane. We justify this based on the symmetry and consistency of the BTCs, indicating no evidence of nonlinear or nonequilibrium sorption behavior. Literature values for n-nonane and n-decane may have been determined at much higher concentrations than applied in our experiments. Although the total retardation factors of n-alkanes varied considerably depending on their carbon-chain lengths, the estimated ai values were very consistent; cases where Rt e 1.1 were not used due to concerns of reliability when retardation is too small. Though the two media had very different mean particle sizes and size distributions (Supporting Information, Table A), the estimated ai showed roughly the same functional dependency on Sw. However, the ai values of Dover soil were about 1-2 orders of magnitude larger than those of the clean sand. The largest ai value for Dover soil, which was measured at the lowest Sw achieved in this study, was comparable with the ai values reported by Conklin et al. (2) and Brusseau et al. (7). Retardation Components. Prediction of the Rt values as well as the partial retardation factor terms (β, in eqs 1 and 2) for the VOCs used in this study was attempted using the estimated ai values based on the first moments of the n-alkanes, the Henry’s law constants, and the water/gas contents measured gravimetrically (Figure 2). Since the Kd values of all the VOCs were found to be insignificant for both media used in this study, based on the aqueous phase column experiments, the soil sorption term (βd) was not included for Rt prediction. In general, the predicted Rt values were in good agreement with those measured for all of the VOCs. As expected, the Rt values for n-alkanes decreased consistently with increasing Sw. However, the retardation behavior of other VOCs varied as Sw changed, depending on the relative magnitude of Ki and KH. In experiments with Dover soil, the Rt for methylene chloride, which has relatively small Ki and KH values, increased over most of the Sw range except at a very low Sw (Figure 2). This implies that the water-partitioning effect was dominant at most Sw except when ai was large enough to contribute significantly to Rt. When the rate of decreasing partial retardation factor from interfacial adsorption with respect to the change of Sw ()dβi/dSw) was smaller than the increase rate of combined partial retardation factor due to water partitioning and soil sorption [)d(βw+βd)/dSw], Rt increased regardless of the absolute magnitudes of βi, βw, and βd. The Rt of PCE, TCA, ethylbenzene, and p-xylene decreased rather rapidly at low Sw range, reached the minimal point, and then increased slowly from that point compared to Rt for methylene chloride. This trend is due to the larger Ki and KH values for the four VOCs compared to those for methylene chloride. The contribution of different sorption mechanisms to the observed Rt for chlorobenzene is illustrated in Figure 2. It was obvious that at low Sw, the major contributor to the total retardation was that from air-water interfacial adsorption (represented by βi), while the water-partitioning effect became more significant as Sw increased with negligible soil sorption. The normalized, fractional retardation factors (f) for p-xylene in Dover soil and clean sand columns are shown in Figure 4. Note that the advantage of normalized fractional retardation factors is that the relative composition of each partial retardation factor with respect to the total retardation factor can be easily identified and assessed regardless of the magnitude of the total retardation factor. It is clear that there was a distinct difference in the relative composition of Rt depending on the porous media. This difference was mainly due to the absolute amount of ai generated in the media. For the clean sand, within the Sw range achieved in this study,
FIGURE 4. The predicted, fractional retardation factors of p-xylene for the Dover soil and clean sand. The fractional retardation factors (f) were normalized with respect to the predicted total retardation factors.
factors for the three VOCs (n-hexane, Rt ) 277; methylene chloride, Rt ) 84; and TCA, Rt ) 524) were orders magnitude larger than those obtained at Sw ) 0.038, the lowest Sw for the gaseous VOC transport experiments (Table B, Supporting Information). By humidifying the dry Dover soil column without injecting bulk water into the column, the estimated retardation factors of the VOCs decreased dramatically to 2-3 times of those at Sw ) 0.038. This indicates that a very small amount of water (e.g., less than Sw ) 0.03) was sufficient for the soil surface to be covered with a water film with enough thickness to prevent the VOCs molecules adsorbed at the surface of the water film from interacting with the solid surface. Dorris and Gray (19) found that the presence of one to two monolayers of water adsorbed on glass was sufficient to considerably reduce the London force field of the glass. They also found that at water loading of 3.8 wt % for the medium they used (porous glass powder, surface area 18.9 m2/g), which was equivalent to Sw ≈ 0.015 (assuming the particle density ≈ 2.65 g/cm3), the effect of the solid on the adsorption of n-alkanes at the water surface appeared to vanish. Based on our observations with dry, humidified, and wet column experiments and the findings of Dorris and Gray (19), it is highly unlikely that VOC molecules interacted with the soil surface from the gaseous phase or that VOC molecules adsorbed at the air-water interface were affected by the soil surface for the Sw range (0.038-0.51) used in this study.
Transport Implications
FIGURE 5. The predicted fractional retardation factors (fi) due to interfacial (air-water) adsorption for VOCs used in this study. Rt for p-xylene increased consistently. Although the (airwater) interfacial adsorption was responsible for about 20% of the total retardation at Sw ) 0.05, the rate of βw increase with increasing Sw was enough to mask the decreasing effect of βi as described earlier. It is also apparent that the fractional retardation factor (fg) due to the retention of p-xylene in the gaseous phase during transport increased in the low Sw range for Dover soil, even though the partial retardation factor, βi, remained constant. This was because of a decreasing Rt, again, due to decreasing ai with increasing Sw. The magnitude of the fractional retardation factor due to air-water interfacial adsorption and its rate of change with respect to Sw were dependent not only on ai in the media but also on the chemical properties of the VOCs. With relatively large Ki and KH values, the fi of p-xylene was larger than that for other VOCs used in this study (Figure 5). The rate of change in fi values with respect to the water saturation change differed depending on the magnitude of Ki and KH values. Although the soil sorption term (βd in eqs 1 and 2) was not significant for the gaseous phase transport of VOCs for the two media used in this study, it was clear that it could be a major contributor for other soils. Note from eq 2 that βd will increase as θg decreases, provided that the Kd value is considerably large. Solid Vapor Sorption. Based on the gaseous phase sorption coefficients for dry Dover soil, the estimated retardation
The retardations observed with the two media used in this study provide some insight on the relevance of these processes for contaminant transport in the vadose zone. The fractional retardations presented in Figure 4 clearly indicate that the air-water interface becomes significant only at low water saturations. In the clean sand, retardation due to the air-water interface is always minor. This is simply due to low air-water interfacial areas generated even at low water saturations. The Dover soil, in contrast, shows quite significant retardation due to the air-water interface below water saturations of about 0.2. This does represent relatively dry conditions and may be most important for arid climates particularly with significant unsaturated zones. While neither soil has significant organic carbon content or solid-phase sorption, this process can dominate the retardation depending on the contaminant of interest. A contaminant with minimal aqueous solubility, such as n-decane, will be predominantly retarded by sorption at the air-water interface present.
Acknowledgments This research was supported by the Hallym Academy of Sciences, Hallym University, Korea, 2000-2001. This material is also based on work sponsored by the Air Force Office of Scientific Research, Air Force Material Command, U.S. Air Force, under grant F49620-95-1-0321.
Supporting Information Available Tabular data on media properties and measured retardation factors for VOC at different water saturations and breakthrough curves resulting from different injection volumes are also provided. This material is available free of charge via the Internet at http://pubs.acs.org.
Literature Cited (1) Okamura, J. P.; Sawyer, D. T. Anal. Chem. 1973, 45, 80-84. (2) Conklin, M. H.; Corley, T. L.; Roberts, P. A.; Davis, J. H.; van de Water, J. G. Water Resour. Res. 1995, 31, 1355-1365. (3) Popovicˇova´, J.; Brusseau, M. L. Water Resour. Res. 1998, 34, 83-92. (4) Corley, T. L.; Farrell, J.; Hong, B.; Conklin, M. H. Environ. Sci. Technol. 1996, 30, 2884-2891. VOL. 35, NO. 22, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4461
(5) Pennell, K. D.; Rhue, R. D.; Rao, P. S. C.; Johnston, C. T. Environ. Sci. Technol. 1992, 26, 756-763. (6) Valsaraj, K. T. Water Res. 1994, 28, 819-830. (7) Brusseau, M. L.; Popovicova, J.; Silva, J. A. K. Environ. Sci. Technol. 1997, 31, 1645-1649. (8) Kim, H.; Rao, P. S. C.; Annable, M. D. Soil Sci. Soc. Am. J. 2000, 63, 1554-1560. (9) Saripalli, K. P.; Kim, H.; Rao, P. S. C.; Annable, M. D. Environ. Sci. Technol. 1997, 31, 932-936. (10) Saripalli, K. P.; Rao, P. S. C.; Annable, M. D.; Hatfield, K. Measurement of Specific Interfacial Areas of Immiscible Fluids in Flow Systems; U.S. Patent No. 5,744,709. U.S. Patent Office, Washington, DC (April 28, 1998). (11) Kim, H.; Rao, P. S. C.; Annable, M. D. Water Resour. Res. 1997, 33, 2705-2711. (12) Kim, H.; Annable, M. D.; Rao, P. S. C. Environ. Sci. Technol. 1998, 32, 1253-1259.
4462
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 22, 2001
(13) Kim, H.; Rao, P. S. C.; Annable, M. D. J. Contam. Hydrol. 1999, 40, 79-94. (14) Hoff, J. T.; Mackay, D.; Gillham, R.; Shiu, W. Y. Environ. Sci. Technol. 1993, 27, 2174-2180. (15) Mackay, D.; Shiu, W. Y. J. Phys. Chem. Ref. Data 1981, 10, 11751199. (16) Yaws, C.; Yang, H.-C.; Pan, X. Chem. Eng. 1991, 98, 179-185. (17) Karickhoff, S. W.; Brown, D. S.; Scott, T. A. Water Res. 1979, 13, 241-248. (18) Valocchi, A. J. Water Resour. Res. 1985, 21, 808-820. (19) Dorris, G. M.; Gray, D. G. J. Phys. Chem. 1981, 85, 3628-3635.
Received for review December 12, 2000. Revised manuscript received June 13, 2001. Accepted September 4, 2001. ES001965L
