Ground Water Quality - Semantic Scholar

2 downloads 0 Views 291KB Size Report
Taeyoon Lee and Craig H. Benson* ... tives, such as wood flour, cob flour, and oat or rice hulls, ... addressed in a leaching study by Lee and Benson (2002).
Ground Water Quality

Reproduced from Journal of Environmental Quality. Published by ASA, CSSA, and SSSA. All copyrights reserved.

Sorption and Degradation of Alachlor and Metolachlor in Ground Water Using Green Sands Taeyoon Lee and Craig H. Benson* ABSTRACT

jaru et al. (2003) have shown that ZVI can degrade metolachlor in aqueous solutions via hydrogenolysis. Eykholt and Davenport (1998) and Koppensteiner (1998) showed that alachlor can be treated similarly. Comfort et al. (2001) also conducted a field test where soil contaminated from a metolachlor spill was treated ex situ in windrows using mixtures of ZVI, acetic acid, and aluminum sulfate. Removal rates as high as 99% were reported by Comfort et al. (2001). A drawback of treating ground water with RBs containing ZVI is the high cost of construction. One approach to reduce costs is to use a low-cost reactive medium, such as an industrial by-product, which can be obtained for little or no cost (Kim et al., 1997; Bailey et al., 1999; Czurda and Haus, 2002). One potential medium is waste green sand from gray-iron foundries. Green sand is used by gray-iron foundries to make molds for castings, and consists of a mixture of fine sand, a small fraction of clay binder, an organic carbon source, and a small fraction of other additives. The organic carbon source is added to improve the finish of castings by providing a reducing atmosphere. Common organic carbon sources include powdered bituminous coal (“seacoal”) and Gilsonite, a naturally occurring asphaltic material. Cellulose additives, such as wood flour, cob flour, and oat or rice hulls, may also be added to absorb moisture and improve flowability. The clay binder usually is sodium bentonite, calcium bentonite, kaolinite, or a combination thereof. Waste green sands also contain between 1 and 11% residual iron particles (by weight) as a by-product of casting operations (Lee and Benson, 2002). The presence of iron particles, organic carbon, and clay renders green sand from gray-iron foundries as a reactive and sorbing medium. Moreover, green sands can be obtained for little or no cost because they are landfilled in large quantities by the U.S. foundry industry. For example, approximately 800 000 Mg of green sand is landfilled annually in Wisconsin, USA, with an annual cost of $18 million to the foundry industry. Because waste green sands generally are considered as a solid waste, regulations often require that the environmental risks associated with their re-use be considered. This issue is not addressed in this paper, but is addressed in a leaching study by Lee and Benson (2002). They conducted batch and column leaching tests on the green sands described here as well as natural soils and a commercially available granular iron used for RBs. Leaching tests on the green sands showed that U.S. ground

Reactive barriers are used for in situ treatment of contaminated ground water. Waste green sand, a by-product of gray-iron foundries that contains iron particles and organic carbon, was evaluated in this study as a low-cost reactive material for treating ground water contaminated with the herbicides alachlor [2-chloro-2ⴕ,6ⴕ-diethyl-N-(methoxymethyl)acetanilide] and metolachlor [2-chloro-6ⴕ-ethyl-N-(2-methoxy1-methylethyl)-o-acetoluidide]. Batch and column tests were conducted with 11 green sands to determine transport parameters and reaction rate constants for the herbicides. Similar Fe-normalized rate constants (KSA) were obtained from the batch and column tests. The KSA values obtained for green sand iron were also found to be comparable with or slightly higher than KSA values for Peerless iron, a common reactive medium used in reactive barriers. Partition coefficients ranging between 3.6 and 50.2 L/kg were obtained for alachlor and between 1.0 and 54.8 L/kg for metolachlor, indicating that the organic carbon and clay in green sands can significantly retard the movement of the herbicides. Partition coefficients obtained from the batch and column tests were similar (⫾25%), but the batch tests typically yielded higher partition coefficients for green sands exhibiting greater sorption. Calculations made using transport parameters from the column tests indicate that a 1-m-thick reactive barrier will result in a 10-fold reduction in concentration of alachlor and metolachlor for seepage velocities less than 0.1 m/d provided the green sand contains at least 2% iron. This level of reduction generally is sufficient to reduce alachlor and metolachlor concentrations below maximum contaminant levels in the United States.

A

lachlor and metolachlor are chlorinated herbicides that have been used widely in agriculture and are ubiquitous ground water contaminants in agricultural regions of the United States (Roux et al., 1991; Kolpin et al., 1996; Barbash et al., 1999; Spalding et al., 2003). Alachlor and metolachlor are mobile in ground water, and have been characterized as probable (alachlor) and possible (metolachlor) human carcinogens (Nowell and Resek, 1994). One potential method to remediate ground water contaminated with alachlor and metolachlor is to treat the ground water in a reactive barrier (RB) using zerovalent iron (ZVI). Laboratory studies by Eykholt and Davenport (1998), Koppensteiner (1998), Comfort et al. (2001), Gaber et al. (2002), and SatapanaT. Lee, Construction Environment Department, Korea Institute of Construction Technology, 2311, Daehwa-Dong, Iilsan-Gu, Goyang-Si, Gyeonggi-Do, 411-712, Republic of Korea. C.H. Benson, Department of Civil and Environmental Engineering, University of Wisconsin, 2214 Engineering Hall, 1415 Engineering Drive, Madison, WI 53706-1691. Received 20 Aug. 2003. *Corresponding author ([email protected]). Published in J. Environ. Qual. 33:1682–1693 (2004).  ASA, CSSA, SSSA 677 S. Segoe Rd., Madison, WI 53711 USA

Abbreviations: RB, reactive barrier; SSA, specific surface area; TOC, total organic carbon; ZVI, zerovalent iron.

1682

Reproduced from Journal of Environmental Quality. Published by ASA, CSSA, and SSSA. All copyrights reserved.

LEE & BENSON: TREATING CONTAMINATED GROUND WATER WITH GREEN SANDS

water quality standards were only exceeded for Fe and that these exceedances were modest (⬍10%). Moreover, they found that Fe, Cr, and Pb leached at lower concentrations from green sand than from the natural soils and the granular iron that were tested. This study assessed the feasibility of using waste green sand as a low-cost reactive medium for treating ground water contaminated with alachlor and metolachlor. The intent of the study was to determine if green sands may be useful as medium in reactive barriers (RBs); however, the material may be applicable to other ground water treatment approaches as well. MATERIALS AND METHODS Green Sands Twelve green sands from foundries in Wisconsin, Illinois, and Ohio, USA, were used in the study. All of the green sands included a small fraction of bentonite clay (blended calcium and sodium bentonite) as a binder and seacoal as the organic additive. Other additives to the green sands were not divulged by the foundries supplying the sands. The green sands were obtained from segregated waste streams at the foundries. Core sands, core butts, scraps, litter, and other debris commonly found in foundry wastes were excluded from the green sands. Each sand was placed in a sealed 200-L drum and shipped to the University of Wisconsin-Madison. Before testing, each sand was spread and air-dried for one day at 24⬚C. After drying, any clumps in the sand were crushed using a small hammer. Index properties of the green sands are summarized in Table 1. Each green sand consists primarily of uniformly graded fine sand. The percent fines (fraction passing the U.S. no. 200 sieve, 75 ␮m) ranges from 4.3 to 16.0%, the clay fraction (finer than 2 ␮m) ranges from 2.9 to 13.2%, and the specific gravity ranges from 2.51 to 2.73. Total organic carbon (TOC) content of each green sand was measured using a Lab 2100 TOC analyzer (Zellweger Analytics, Lincolnshire, IL), and was found to range between 0.5 and 4.0%. Total iron content was determined by elemental analysis on a digestion and ranged between 1.2 and 11.9%. Surface area of the iron in the green sands was determined by Brunauer–Emmett–Teller (BET) analysis (Brunauer et al., 1938). The surface area was 2.45 m2/g, on average, and varied little between sands (Lee and Benson, 2002).

1683

Saturated hydraulic conductivity of each green sand was measured following the methods described in ASTM D 5886 (American Society for Testing and Materials, 2000). The constant head method was used and tap water was the permeant liquid. The green sands were tamped into the permeameter with a rod in three layers of equal thickness using 15 tamps per layer. The intent was to create a moderately compacted material that would simulate in situ placement of the sand by loose dumping followed by vibratory densification. Saturated hydraulic conductivities of the sands are summarized in Table 1. They range from 0.00079 to 23.3 m/d, but most are between 0.2 and 2.0 m/d. The saturated hydraulic conductivity is controlled primarily by the amount of clay in the green sand (i.e., green sands with a larger clay fraction typically have lower hydraulic conductivity). For example, Sand 11 has the lowest saturated hydraulic conductivity and the highest clay content of the sands that were tested, whereas Sand 9 has the smallest clay fraction and the highest hydraulic conductivity. This sensitivity to clay content is typical of mixtures of sand and bentonite, as described in Abichou et al. (2002). The hydraulic conductivities of the green sands are lower than the hydraulic conductivity of granular ZVI, which typically is on the order of 10 m/d (Elder et al., 2002). Thus, green sands most likely would be used in permeable reactive barriers (PRBs) placed in less permeable aquifer materials (fine sands, silty sands) or in low-permeable reactive barriers (LPRBs) used in conjunction with ground water cutoff walls. Alternatively, the hydraulic conductivity of green sands can be increased to more conventional levels by mixing with coarse materials such as gravel, coarse sand, crushed glass, or tire chips (Lee and Benson, 2002).

Zerovalent Iron Commercially available ZVI particles obtained from Peerless Metal Powders and Abrasives Co. (Detroit, MI) were used for comparative tests. Purity of the ZVI ranged from 92 to 95% (by weight), the total carbon content was 3.5% (by weight), the mean particle size was 0.7 mm, and the specific surface area was 0.87 m2/g (Koppensteiner, 1998). The Peerless ZVI particles are larger and have lower specific surface area than the iron in green sands, as will be discussed subsequently.

Chemicals Alachlor (94% purity) and its dechlorination by-product acetyl alachlor (98% purity) were obtained from Monsanto

Table 1. Properties of the green sands. Particle size characteristics

Green sand

d50†

Fines‡

1 2 3 4 5 6 7 8 9 10 11 12

mm 0.19 0.19 0.20 0.19 0.19 0.28 0.19 0.19 0.21 0.19 0.20 0.20

10.7 14.3 11.3 13.2 12.4 10.2 10.9 11.1 4.3 10.0 16.0 10.0

2-␮m Clay content

USCS classification§

Specific gravity of solids

TOC¶

SP–SM SM SW–SM SC–SM SC–SM SP–SM SC–SM SP SP SP–SM SM–SC SP

2.62 2.53 2.52 2.63 2.54 2.61 2.72 2.68 2.64 2.73 2.51 2.73

1.5 2.6 2.5 0.5 1.8 1.1 2.2 2.5 0.8 2.5 4.0 2.4

%

Total iron content

Saturated hydraulic conductivity#

2.8 1.5 11.9 2.9 1.4 1.6 6.5 7.8 1.2 1.4 1.8 11.3

m/d 1.35 (1.31) 1.99 (1.34) 0.52 (1.35) 0.00081 (1.26) 0.24 (1.34) 0.35 (1.36) 0.34 (1.38) 0.0033 (1.38) 23.3 (1.51) 0.47 (1.34) 0.00079 (1.26) 1.64 (1.32)

% 6.7 9.2 7.7 9.3 8.0 5.2 4.5 6.2 2.9 3.5 13.2 3.5

† Median particle size. ‡ Percent finer than 75 ␮m. § Unified Soil Classification System. SP, poorly graded sand with little or no fines; SM, silty sand; SW, well-graded sand with little or no fines; SC, clayey sand. ¶ Total organic carbon. # Dry density (Mg/m3) of specimen tested for hydraulic conductivity is shown in parentheses.

1684

J. ENVIRON. QUAL., VOL. 33, SEPTEMBER–OCTOBER 2004

Reproduced from Journal of Environmental Quality. Published by ASA, CSSA, and SSSA. All copyrights reserved.

Corporation (St. Louis, MO). Novartis Crop Protection (Greensboro, NC) provided the metolachlor (97.3% purity) and its dechlorination by-product (99% purity), which is known as “metolachlor by-product” or MBP. Deuterium oxide, which was used as a tracer, was obtained from Aldrich Chemical Co. (Milwaukee, WI).

Analytical Methods Herbicides and By-Products A Varian (Palo Alto, CA) 3400 gas chromatograph with a flame ionization detector (FID) and DB-5 column (30-m length ⫻ 0.25-mm i.d.) was used to analyze for alachlor, metolachlor, and their dechlorination by-products. A 1-␮L sample was drawn using a 10-␮L gas-tight syringe and manually injected into the gas chromatograph. The detector temperature was set at 300⬚C and the injector temperature was set at 210⬚C. Air was used as the carrier gas at a flow rate of 300 mL/min. Helium was used as the make-up gas using a flow rate of 30 mL/min. The column temperature was maintained at 110⬚C for 6 min and raised to 170⬚C at 20⬚C/min until 240⬚C was reached, at which the temperature was held for 2 min to remove any residuals. The total run time was 25 min. The detection limits for alachlor, acetyl alachlor, metolachlor, and MBP were 0.9, 1.1, 1.0, and 0.8 mg/L, respectively. Samples were prepared for analysis using the method described in Koppensteiner (1998). A 2-mL sample of solution was extracted into 2 mL of methylene chloride and then mixed for 1 min with a vortex mixer. A gas-tight syringe was used to transfer 1 mL of the methylene chloride solution into an autosampler vial with an open-top closure and a Teflon-lined silicone septa. All samples were stored in a refrigerator at 4⬚C before analysis. Recovery factors determined by extraction from stock solutions of alachlor and metolachlor having known concentration (50 mg/L) were 104 and 102%, with standard deviations of 6.3 and 8.1%, respectively. Deuterium Oxide Tracer The D2O concentrations were determined via refractive index (RI) using a Beckman 156 refractive index detector (Beckman Coulter, Fullerton, CA). Before analysis, samples were filtered using a 0.45-␮L Teflon syringe filter to remove colloidal materials and transferred to an autosampler vial with a closure having a Teflon-lined silicone septum. The samples were loaded into a Varian autosampler 8000/8100 and an 80-␮L sample was injected into the RI detector at a flow rate of 1 mL/min. Calibration of the RI detector was conducted using Type I deionized water containing 1, 3, 5, and 10% D2O (by volume). The detection limit for D2O was 0.4% (by volume).

Batch and Column Test Methods Solution Preparation Aqueous solutions of the herbicides were prepared by injecting a designated amount of alachlor or metolachlor into a 2-L volumetric flask filled with Type I deionized water. The Type I deionized water was sparged with nitrogen gas for 1 h beforehand to remove dissolved oxygen, and then mixed with sodium azide (0.1% by weight) to prevent biological activity. The solution was mixed with a magnetic stirring bar for 24 h. To evaluate losses during preparation, aliquots of alachlor and metolachlor (1 ␮L) were analyzed. The variations were ⫾5% of the expected alachlor and metolachlor concentrations.

Batch Sorption Tests Batch sorption tests were conducted on green sands without iron so that reactivity and sorptivity could be evaluated separately. Sands 1 through 5 and 7 through 12 were tested. Sand 6 was not tested because insufficient quantities of Sand 6 remained after the hydraulic conductivity tests. A strong magnet was passed over the surface of a thin layer of green sand to remove the iron particles. The layer was mixed several times during removal to ensure that the iron particles could be removed from the magnet. The batch tests were conducted by varying the mass (2–7 g) of green sand and holding the concentration of herbicide constant. Aqueous solutions and green sand were placed into 40-mL Teflon bottles sealed with Teflon caps. Aqueous solutions with no green sand were used as controls to estimate losses. Final concentrations of these controls were assumed to be the initial concentration of the mixtures to account for losses during the test (Zytner, 1992). All samples and controls were treated identically to maintain procedural similarities (Zytner, 1992). The vials were tumbled at 30 rpm. A series of preliminary tests were conducted to determine the time required to achieve equilibrium. Results of these tests showed that the concentration ceased changing or was changing a very small amount after 24 h of tumbling. Thus, for practical purposes, a tumbling time of 24 h was used for the batch tests. After tumbling, the vials were centrifuged at 8000 rpm (8200 ⫻ g ) for 20 min. Samples were obtained from the vials using a 5-mL gas-tight syringe, and then prepared for analysis as described previously. Analyses were conducted for both the herbicides and their dechlorination by-products (acetyl alachlor and MBP). No dechlorination by-products were detected, which suggests that the iron removal process was successful and that dechlorination and degradation during the batch sorption tests were negligible. Batch Degradation Tests Batch tests were conducted to evaluate the rate of degradation of alachlor and metolachlor in the presence of green sand iron. Tests were also conducted with Peerless iron for comparison. Iron particles for the batch tests were extracted from the green sands using a magnet, washed with methanol, and then dried for 5 min with an external heater set at 40⬚C for 5 min. No brown oxides (i.e., iron oxyhydroxides) were visible on the washed particles. Iron particles extracted from the green sands were placed in 50-mL glass vials filled completely with solution (no headspace). Controls without iron were prepared in an identical manner. The vials were loaded on a tumbler and rotated at 30 rpm. Vials were removed from the tumbler intermittently for analysis. The vial was allowed to sit for 5 min so the iron particles would settle, and then a 5-mL gas-tight syringe was used to collect a 2.5-mL sample. The sample was filtered through a 0.45-␮m Teflon syringe filter and prepared for analysis as described previously. A first-order decay model with instantaneous sorption was used to determine the bulk reaction rate constant (Kobs) and the iron-phase partition coefficient (Kpi) for both herbicides. The model is (Koppensteiner, 1998):

Caq(t) ⫽



⫺Kobst C0 exp KpiSr ⫹ 1 KpiSr ⫹ 1



[1]

where Caq is the concentration of herbicide in the batch reactor at time t (mg/L), C0 is the initial herbicide concentration (mg/L), Kobs is the bulk first-order decay rate constant (h⫺1), and Sr is

1685

Reproduced from Journal of Environmental Quality. Published by ASA, CSSA, and SSSA. All copyrights reserved.

LEE & BENSON: TREATING CONTAMINATED GROUND WATER WITH GREEN SANDS

Fig. 1. Concentrations of alachlor and metolachlor from batch degradation tests conducted with green sand iron from Green Sand 11.

solid to solution ratio (kg/L). Equation [1] was fit to the data using a least-squares algorithm to obtain Kobs and Kpi. Typical fits of Eq. [1] are shown in Fig. 1 for batch tests conducted with iron particles from Sand 11. Analyses were conducted to determine concentrations of the herbicides as well as their dechlorination by-products to confirm that dechlorination was occurring and to conduct mass balances. Dechlorination by-products were observed in all tests and mass balances of 94% (alachlor) and 96% (metolachlor) were obtained. Moreover, the rate of depletion of both herbicides agreed closely with the rate at which the dechlorination by-products accumulated (Lee and Benson, 2002).

Column Tests Column tests were conducted to determine partition coefficients and rate constants for flow-through conditions. A schematic of the column test setup is shown in Fig. 2. A glass column (length ⫽ 450 mm, diameter ⫽ 25 mm) was used with Teflon adapters and porous ceramic disks attached to both ends. A glass fiber filter was placed between the porous disk and the reactive medium to prevent clogging. At least two pore volumes of Madison tap water were passed through the column initially. The influent solution was then switched to a herbicide solution for the column test. After finishing a test with herbicide solution, D2O solution (10% by weight) was introduced as a conservative tracer to determine the effective porosity. Solution was introduced into the columns at a constant flow rate (2.2 ⫻ 10⫺4 L/min) using a peristaltic pump. All tests were conducted with upward flow. The influent was contained in a 2-L Teflon bag. All contact parts in the pump were made of Teflon, except for the tubing, which was Viton. Concentrations of alachlor or metolachlor in the influent ranging between 34 and 55 mg/L were used. A loss of approximately 3% occurred when the herbicides passed through the pump, but a constant influent concentration (⫾2.6% for alachlor, ⫾3.5% for metolachlor) was maintained during testing (Lee and Benson, 2002). Samples of the influent and effluent were collected from the sampling ports (Fig. 2) at regular intervals. Approximately

Fig. 2. Schematic of column test apparatus.

2 mL of solution was transferred to a 4-mL glass vial in less than 1 min, and was immediately prepared for analysis as described previously. Samples were stored in a refrigerator at 4⬚C before analysis and were manually injected into the gas chromatograph using a gas-tight syringe. The bulk reaction rate constant (Kobs), partition coefficient (Kp), dispersivity (␣L), and effective porosity (ne) were determined for each column test by fitting the relative concentration (Ce/C0) data to an analytical solution of the advection–dispersion–reaction equation (ADRE) with instantaneous sorption. Typical fits are shown in Fig. 3a. The solution to the ADRE that was used is (van Genuchten, 1981):

冤 冤

冥 冤 冥 冤

冥 冥

Ce 1 (v ⫺ u)L RL ⫺ ut ⫽ exp erfc ⫹ C0 2 2D 2√DRt 1 (v ⫹ u)L RL ⫹ ut exp erfc 2 2D 2√DRt

[2]

where Ce is the concentration at the effluent end of the column at time t, C0 is the influent concentration, L is the column length, v is the seepage velocity (discharge velocity/effective porosity), D is the effective dispersion coefficient, R is the retardation factor, exp is the exponential function, and erfc is the complementary error function. This analytical solution is derived for the first-type initial and boundary conditions (i.e., background concentration is assumed to be zero, influent concentration is constant, and the concentration gradient is zero at great distance from the influent boundary). The variable u in Eq. [2] is:



u⫽v1⫹



4KobsD 1/2 v2

[3]

Use of Eq. [3] implicitly assumes that degradation of ad-

Reproduced from Journal of Environmental Quality. Published by ASA, CSSA, and SSSA. All copyrights reserved.

1686

J. ENVIRON. QUAL., VOL. 33, SEPTEMBER–OCTOBER 2004

Fig. 3. (a) Typical breakthrough curves for column tests with green sands and (b) breakthrough curves for control tests conducted with Portage sand along with fits of the advection–dispersion–reaction equation (ADRE; Eq. [2]).

sorbed herbicides is negligible. The retardation factor is related to the bulk partition coefficient (Kp) for the linear isotherm via:

R⫽1⫹

␳d Kp n

[4]

where ␳d is the dry density (mass of solid per total volume) and n is the porosity. For the steady-state condition, the effluent concentration (Css) is:

Css ⫽ C(L,∞) ⫽ C0exp

冤(v ⫺2Du)L冥

[5]

The dispersion coefficient (D ) obtained by fitting Eq. [2] to the column test data is the sum of the mechanical dispersion coefficient (Dm) and the molecular diffusion coefficient (D*). The dispersivity (␣L) is related to the mechanical dispersion coefficient Dm via (Freeze and Cherry, 1979):

␣L ⫽ Dm /v

[6]

A least squares algorithm was used for fitting Eq. [2] to the data (Lee and Benson, 2002). Equation [2] was fit to the relative concentrations of D2O with R ⫽ 1 and Kobs ⫽ 0 to determine v and D. The specific discharge (known from the pumping rate and the cross-sectional area of the test specimen) was divided by v to obtain ne and ␣L was computed from D using Eq. [6]. The molecular diffusion coefficient (D*) in Eq. [6] was obtained from Yaws (1995) and was reduced by a factor of 0.4 to account for tortuosity. However, the solution was not sensitive to D* because D was dominated by Dm at the flow rates that were used. Equation [2] was fit to the relative concentrations of alachlor and metolachlor to obtain Kobs and Kp. During these fits, u and R were fitting parameters, v was set at the value obtained from the tracer test, and Dm was computed with Eq. [6] using ␣L and v from the tracer test. Equation [3] was used to compute Kobs from u and Eq. [4] was used to compute Kp from R. The bulk reaction rate (Kobs) was also computed independently from the steady-state effluent concentration using Eq. [3] and [5]. The Fe-normalized rate constant (KSA) was computed as Kobs/SSA, where SSA is the specific surface area of iron per volume of solution in the column. Two control tests were conducted with Portage sand to assess the potential for sorption on the tubing and glass column. One test was conducted with alachlor and the other with metolachlor. Portage sand is a poorly graded silica sand from a sand deposit in Portage, WI. The sand has a median particle size of 1.8 mm, a uniformity coefficient of 2.0, a coefficient of curvature of 1.39, and a classification of SP in the Unified Soil Classification System. The sand was washed with deionized water until any visible impurities were removed, air-dried, and packed in the glass column. The control test was conducted in the same manner as all other column tests. Breakthrough curves for alachlor and metolachlor from the control tests are shown in Fig. 3b. The relative concentration reaches steady state at unity, indicating that no continuous losses were occurring. Fitting Eq. [2] to the data yielded partition coefficients of 1.56 L/kg for alachlor and 0.23 L/kg for metolachlor. This modest partition coefficient is believed to be due in part to organic carbon in the sand. To assess this possibility, the partition coefficients for alachlor and metolachlor were estimated using the empirical equation from Briggs (1981):

Kp ⫽ foc10AlogKow⫹B

[7]

where foc is weight fraction of organic carbon in the Portage sand, Kow is the octanol–water partition coefficient, and A and B are empirical constants. The empirical constants A and B were set at 0.52 and 0.88, respectively, as reported by Briggs (1981). The foc for the Portage sand was determined using the same procedure used for the green sands and was found to be 0.2%. The Kp values for alachlor and metolachlor estimated using this approach are 0.35 and 0.94 L/kg, respectively, which suggests the modest amount of partitioning observed in the control tests may have been due to the presence of organic carbon. Regardless, the Kp obtained from the control tests are small relative to the Kp obtained for most of the green sands. Thus, losses to the column materials are believed to have been negligible.

RESULTS AND DISCUSSION Batch Tests Batch Sorption Tests Sorption isotherms for alachlor and metolachlor obtained from the batch tests on green sands without iron

Reproduced from Journal of Environmental Quality. Published by ASA, CSSA, and SSSA. All copyrights reserved.

LEE & BENSON: TREATING CONTAMINATED GROUND WATER WITH GREEN SANDS

1687

Fig. 4. Batch sorption isotherms for alachlor: (a) Sands 1 through 5 and (b) Sands 7 through 12.

Fig. 5. Batch sorption isotherms for metolachlor: (a) Sands 1 through 5 and (b) Sands 7 through 12.

are shown in Fig. 4 and 5 along with fits of the linear isotherm. Between 5 and 12 individual batch tests were conducted on each sand to develop an isotherm. No replicate tests were conducted, but each isotherm follows a well-defined trend without excessive scatter. Thus, the isotherms are believed to be reliable despite the lack of replication. All of the isotherms are approximately linear within the range of concentrations that was tested. Partition coefficients for alachlor and metolachlor were obtained by fitting data from the batch tests to linear and Freundlich isotherm models using a least-squares algorithm. The parameters of the linear and Freundlich models are summarized in Table 2. Good fits were obtained with both isotherm models for the range of concentrations that were used. Partition coefficients obtained from the linear model range from 3.6 L/kg (Sand 9) to 50.2 L/kg (Sand 11) for alachlor and 1.0 L/kg (Sand 9) to 54.8 L/kg (Sand 11) for metolachlor and generally are within the range typically reported for soils with TOC similar to the green sands (Peter and Weber, 1985; Chesters et al., 1989; Johnson

and Sims, 1993; Gaston and Locke, 1994; Seybold and Mersie, 1996; Clay et al., 1997; Wang et al., 1999; Patakioutas and Albanis, 2002). The fits with the Freundlich model confirm that the isotherms are essentially linear. Except for Sands 1, 3, and 4 (alachlor) and Sands 5 and 8 (metolachlor), the Freundlich parameter 1/n is approximately 1. Similar values of 1/n have been reported by Seybold and Mersie (1996), Wang et al. (1999), and Patakioutas and Albanis (2002) for organic soils, but slightly smaller values have been reported Clay et al. (1997). The Kp values for alachlor and metolachlor are shown as a function of TOC and clay content in Fig. 6. The open circles are data points and the numbers in italics adjacent to the open circles are the measured Kp. The contours were prepared with the minimum curvature algorithm in Surfer (Version 8.04; Golden Software, 2002). For both alachlor and metolachlor, Kp generally increases with TOC and clay content. For example, Sand 9 has the lowest Kp (3.6 L/kg for alachlor and 1.0 L/kg for metolachlor) as well as the lowest TOC and clay content of all sands. Similarly, Sand 11 has the highest Kp (50.2 L/kg for alachlor and 54.8 L/kg for metolachlor)

1688

J. ENVIRON. QUAL., VOL. 33, SEPTEMBER–OCTOBER 2004

Table 2. Isotherm parameters from batch sorption tests.

Reproduced from Journal of Environmental Quality. Published by ASA, CSSA, and SSSA. All copyrights reserved.

Alachlor Linear model

Metolachlor Freundlich model

Linear model

Freundlich model

Green sand

Kp †

R2

K f‡

1/n

R2

Kp

R2

Kf

1/n

R2

1 2 3 4 5 7 8 9 10 11 12

L/kg 5.8 18.8 7.1 20.5 15.6 19.3 11.2 3.6 17.2 50.2 17.8

0.93 0.96 0.97 0.99 0.98 0.94 0.96 0.86 0.97 0.96 0.98

L/kg 13.5 18.3 23.1 67.1 17.8 19.0 5.6 5.0 21.5 33.8 19.4

1.3 1.0 1.5 1.6 1.1 1.0 0.8 1.1 1.1 0.9 1.0

0.95 0.96 0.99 0.97 0.94 0.93 0.98 0.89 0.96 0.97 0.98

L/kg 14.9 16.5 8.5 18.9 15.3 10.5 16.3 1.0 23.6 54.8 15.4

0.95 0.95 0.99 0.98 0.98 0.99 0.97 0.96 0.96 0.96 0.96

L/kg 8.2 15.8 11.5 11.7 30.4 17.1 38.5 2.1 20.3 32.4 10.5

0.9 1.0 1.1 0.9 1.3 1.1 1.4 1.2 0.9 0.9 0.9

0.97 0.93 0.98 0.99 0.97 0.98 0.96 0.98 0.89 0.97 0.97

† Partition coefficient. ‡ Freundlich coefficient.

and the highest TOC and clay content. Both TOC and clay content are important because the herbicides have polar functional groups that permit hydrogen bonding to mineral surfaces (Chesters et al., 1989; Grundl and Small, 1993; Wang et al., 1999). For example, Sand 4 has relatively high Kp (18.9 L/kg) because the sand has high clay content (10.5%) even though the TOC is low (0.5%). A direct assessment of the relative contributions of the clay fraction and organic carbon in the green sands was not possible because the proportions of TOC, clay, and nonclay particles could not be determined with the data that were available. However, the relative contributions of mineral solids and organic carbon to the partition coefficient were estimated using the linear mixing model described in Grundl and Small (1993):

Kp ⫽ (1 ⫺ foc)Km ⫹ focKoc

[8]

where foc is the fraction of organic carbon (foc ⫽ TOC/ 100), Km is the partition coefficient for the mineral solid, and Koc is the partition coefficient for the organic carbon. Least-squares regression was used to determine Km and Koc from the partition coefficients for the linear isotherms in Table 2. The regression yielded Km ⫽ 0.65 L/kg and Koc ⫽ 809 L/kg for alachlor and Km ⫽ 0.50 L/kg and Koc ⫽ 854 L/kg for metolachlor. The Koc and Km values obtained from the regression analysis are similar to those reported by Grundl and Small (1993) for surficial organic soils having similar TOC as the green sands, but are larger than Koc reported for soils with low organic carbon content (⬍1.3%) by John-

Fig. 6. Contours of partition coefficient (L/kg) as a function of total organic carbon (TOC) and 2-␮m clay content for (a) alachlor and (b) metolachlor. Open circles represent data points. Numbers in italics next to data points are measured partition coefficients. Numbers in bold are contour intervals for the partition coefficient.

1689

Reproduced from Journal of Environmental Quality. Published by ASA, CSSA, and SSSA. All copyrights reserved.

LEE & BENSON: TREATING CONTAMINATED GROUND WATER WITH GREEN SANDS

son and Sims (1993) and Seybold and Mersie (1996). The large Koc demonstrates the affinity of the organic carbon fraction for alachlor and metolachlor. However, sorption on the mineral fraction is also important. Even though Km is much smaller than Koc, the mass of mineral solid in green sands is much greater than the mass of organic carbon, making sorption on the mineral solid appreciable. The clay fraction likely is responsible for most of the sorption on mineral solids, as suggested by Peter and Weber (1985) and Wang et al. (1999). This inference is also consistent with the larger Kp associated with green sands having higher clay content (Table 2). Batch Degradation Tests Batch degradation tests were conducted with iron from the green sands as well as Peerless ZVI for comparison. The parameters obtained from the model fits are summarized in Table 3. Rate constants and iron partition coefficients from tests conducted by Koppensteiner (1998) and Eykholt and Davenport (1998) using Peerless granular ZVI are also shown in Table 3. The rate constants in Table 3 (KSA) are normalized by SSA (i.e., KSA ⫽ Kobs/SSA). The KSA obtained for the green sands iron is comparable with, but slightly larger (approximately 10%), than the KSA obtained for Peerless iron for both alachlor and metolachlor. The green sand Kpi values are similar to the Kpi values for Peerless iron. The KSA measured in this study also fall within the range reported by Eykholt and Davenport (1998) and Koppensteiner (1998) for Peerless ZVI (KSA ⫽ 3.9 ⫻ 10⫺4 to 1.5 ⫻ 10⫺3 L/m2-h for alachlor and 3.3 ⫻ 10⫺4 to 9.9 ⫻ 10⫺4 L/m2-h for metolachlor). However, the bulk reactivity of the green sand iron is higher than that of Peerless iron because the green sand has higher surface area than Peerless iron (2.45 vs. 0.87 m2/g). Although particle size analyses were not conducted on the green sand iron, visual comparison of green sand iron and Peerless iron showed that green sand iron consists of particles that are appreciably finer and more uniform than Peerless iron. The smaller particle size results in more surface area and greater reactivity per mass of iron. Green sands contain smaller iron particles because the particles are often derived from small pieces

of flashing as well as residuals from grinding (the latter forming very fine particles), whereas commercial iron particles often are from recycled iron from machining operations.

Column Tests The column tests differed from the batch tests in several important ways. Column tests simulate flow-through conditions where solid–liquid contact may be limited. Rate limitations may also exist or be exacerbated due to mass transfer restrictions between solutes in different pore spaces (e.g., pores active in flow and less accessible pores). Additionally, the batch tests were conducted to evaluate sorption and dechlorination separately, whereas both processes occur simultaneously during column tests. A key issue is whether transport parameters obtained from relatively simple batch tests (i.e., partition coefficients and rate constants) are comparable with those operative during flow-through conditions, which occur in column tests and the field. Transport parameters obtained from the column tests are summarized in Table 4. Five green sands (1, 2, 4, 5, and 12) having a broad range of TOC, clay content, and iron content were used for the tests with alachlor. Tests with alachlor were also conducted with Sands 1 and 5 using admixtures of Peerless ZVI (10 and 20% by weight, respectively) to evaluate how elevated iron content affects reactivity of the green sand. Three green sands (1, 5, and 12) were used for the tests conducted with metolachlor. Two of the tests with Sands 1 and 5 contained an admixture of 20% Peerless iron (by weight). The tests with ZVI admixtures are reported in Table 4 with a suffix of A in the sand-type designation. Two additional comparative tests were conducted with alachlor and metolachlor using mixtures of Peerless iron and Portage sand (10 and 20% iron by weight, respectively). Effective Porosity and Dispersivity The effective and total porosities are summarized in Table 4, along with the ratio of effective to total porosity. The ratio of effective to total porosity ranges from 0.82 to 1.0, which is comparable with ranges reported by others for compacted fine-textured soils (Kim et al., 1997; Bin-Shafique et al., 2002). The dispersivities are

Table 3. Results of batch degradation tests for alachlor and metolachlor using foundry iron and Peerless granular zerovalent iron (ZVI). Compound Alachlor Alachlor Alachlor# Alachlor†† Metolachlor Metolachlor Metolachlor# Metolachlor††

Iron source green sand Peerless Peerless Peerless green sand Peerless Peerless Peerless

Initial concentration mg/L 49 49 11 9 84 38 14 9

SSA† 2

m /L 186 40 308 28 186 22 308 28

† Specific surface area of iron per unit volume of solution. ‡ Bulk first-order decay rate constant. § Normalized first-order rate constant. ¶ Iron-phase partition coefficient for herbicide. # Parameters obtained by fitting Eq. [1] to data from Eykholt and Davenport (1998). †† Parameters obtained by fitting Eq. [1] to data from Koppensteiner (1998).

Kobs‡ 1/h 0.148 0.0029 0.12 0.042 0.10 0.011 0.10 0.028

KSA§ 2

L/m -h 7.9 ⫻ 10⫺4 7.1 ⫻ 10⫺4 3.9 ⫻ 10⫺4 1.5 ⫻ 10⫺3 5.5 ⫻ 10⫺4 5.1 ⫻ 10⫺4 3.3 ⫻ 10⫺4 9.9 ⫻ 10⫺4

Kpi¶ L/kg 2.5 3.4 3.2 3.1 1.9 1.9 8.0 3.8

1690

J. ENVIRON. QUAL., VOL. 33, SEPTEMBER–OCTOBER 2004

Table 4. Test conditions and transport parameters for column tests conducted with reactive media.

Reproduced from Journal of Environmental Quality. Published by ASA, CSSA, and SSSA. All copyrights reserved.

Media† Sand 1 Sand 1-A Sand 2 Sand 4 Sand 5-A Sand 12 Portage sand–iron Sand 1 Sand 1-A Sand 5-A Sand 12 Portage sand–iron

Total porosity, n

Effective porosity, ne

Dispersivity

Dry density

Specific discharge

Initial concentration

Mg/m3 1.67 1.66 1.69 1.69 1.83 1.56 2.07

m/d 1.00 1.12 0.90 1.08 0.86 1.16 1.11

mg/L 34 35 35 35 48 42 20

0.36 0.47 0.33 0.36 0.49 0.43 0.34

0.36 0.40 0.33 0.36 0.40 0.36 0.29

1.00 0.85 1.00 1.00 0.82 0.84 0.85

1.68 1.89 1.89 1.64 2.04

0.99 0.60 0.79 1.42 0.76

50 49 49 49 38

0.36 0.48 0.48 0.47 0.44

0.36 0.40 0.40 0.40 0.37

1.00 0.83 0.83 0.85 0.84

ne/n

Measured

KSA#

0.1 L

Kp‡

Kobs§

SSA¶

0.017 0.074 0.039 0.031 0.012 NA†† 0.020

0.042 0.042 0.028 0.028 0.030 NA 0.043

L/kg 6.5 11.04 14.4 17.8 14.1 NA 0.49

1/h 0.036 0.082 0.009 0.032 0.453 0.570 0.267

m2/L 113 364 19 34 791 1053 529

L/m2-h ⫺ 3.21 ⫻ 10 4 2.56 2.95 ⫻ 10⫺4 3.07 4.69 ⫻ 10⫺4 4.51 9.36 ⫻ 10⫺4 1.11 5.67 ⫻ 10⫺4 5.89 NA 4.78 5.25 ⫻ 10⫺4 4.84

⫻ ⫻ ⫻ ⫻ ⫻ ⫻ ⫻

10⫺4 10⫺4 10⫺4 10⫺3 10⫺4 10⫺4 10⫺4

0.019 0.015 0.011 NA 0.047

0.030 0.030 0.030 NA 0.043

10.1 11.1 14.0 NA 1.6

0.023 0.448 0.433 0.301 0.209

114 945 828 1002 960

2.02 ⫻ 10⫺4 4.74 ⫻ 10⫺4 5.23 ⫻ 10⫺4 NA 2.18 ⫻ 10⫺4

⫻ ⫻ ⫻ ⫻ ⫻

10⫺4 10⫺4 10⫺4 10⫺4 10⫺4

m

Eq. [2]

Steady state

2.85 4.90 5.58 2.97 2.16

† The term “-A” indicates that Peerless zerovalent iron (ZVI) was added to the green sand. ‡ Partition coefficient. § Bulk first-order decay rate constant. ¶ Specific surface area. # Normalized first-order rate constant. †† Not available. Analytical difficulties prevented complete definition of the breakthrough curve for Sand 12. Thus, Kobs and KSA were determined only from the steady-state effluent concentration. The dispersivity and partition coefficient could not be determined.

also summarized in Table 4, along with estimates made assuming the dispersivity equals 10% of the column length. The measured and estimated dispersivities are typically within a factor of 2, which suggests that the fits of Eq. [2] are reasonable. Rate Constants and Partition Coefficients The normalized first-order rate constants (KSA) and partition coefficients are summarized in Table 4. Two KSA are reported for each test. One of the KSA values was obtained by fitting Eq. [2] to the entire breakthrough data set; the other was obtained by fitting Eq.

Fig. 7. Bulk first-order rate constants (Kobs) for alachlor as a function of specific surface area (SSA) of iron.

[5] to the steady-state concentrations in the latter part of each test. Both methods yield essentially the same KSA, which suggests that the fits to Eq. [2] are reasonable. The similarity of the KSA also suggests that assuming instantaneous sorption, which is implicit in Eq. [2], is reasonable for the relatively low flow rates that are characteristic of many ground water applications (and the column tests). Difficulties incurred when analyzing the effluent from Sand 12 prevented determination of the entire breakthrough curve. Thus, Kobs and KSA were only computed from the steady-state portion of the breakthrough curve for Sand 12 (hyphens are shown in Table 4 for parameters obtained from Eq. [2]). Nevertheless, the KSA for Sand 12 is comparable with the KSA obtained for the other sands. The bulk first-order rate constant (Kobs) for alachlor is shown as a function of SSA in Fig. 7. The slope of the trend line in Fig. 7, which was obtained by linear regression, equals the average KSA for the green sands (5.2 ⫻ 10⫺4 L/m2-h). This average KSA is similar to but slightly smaller than the KSA obtained from the batch degradation tests (7.9 ⫻ 10⫺4 L/m2-h for green sand iron, 7.1 ⫻ 10⫺4 L/m2-h for Peerless iron). Moreover, the relationship between Kobs and SSA obtained from the column tests is essentially the same for the green sand iron and Peerless iron, suggesting that both materials have similar reactivity. The KSA values for Sands 1 and 5 containing an admixture of Peerless iron also fall along the same trend line suggesting that the reactivities obtained from the batch and column tests for green sand and Peerless iron are comparable. A similar relationship was obtained between Kobs and SSA for metolachlor, but with more scatter (Lee and Benson, 2002). Regression of Kobs for metolachlor on SSA yielded KSA ⫽ 3.8 ⫻ 10⫺4 L/m2-h, which is comparable but slightly lower than the KSA from the batch degradation tests (5.5 ⫻ 10⫺4 L/m2-h for green sand iron, 5.1 ⫻ 10⫺4 L/m2-h for Peerless iron). Partition coefficients obtained from the fits to Eq. [2] are also summarized in Table 4. A partition coefficient

Reproduced from Journal of Environmental Quality. Published by ASA, CSSA, and SSSA. All copyrights reserved.

LEE & BENSON: TREATING CONTAMINATED GROUND WATER WITH GREEN SANDS

1691

Fig. 9. Normalized rate constants (KSA) for Sand 12 and iron–Portage sand mixture during long-term testing.

Fig. 8. Comparison of partition coefficients for alachlor and metolachlor from column tests and batch sorption tests.

was not computed for Sand 12 due to the difficulty mentioned previously. A comparison of the partition coefficients from the batch tests (linear isotherm) and column tests for alachlor and metolachlor is shown in Fig. 8. The partition coefficients from the batch and column tests generally are within ⫾25%, but the column tests typically yielded lower partition coefficients for green sands exhibiting greater sorption. One possible reason for this difference may be the different solid to solution ratios in the batch and column tests. Solid concentrations in batch tests ranged between 50 and 175 g/L, whereas the solid concentration in the column tests was approximately 4250 g/L (i.e., the solid concentration for the column tests is about 38 times larger than that of the batch tests). The modest variation in solid to solution ratio during the batch tests may also have contributed to the differences in the partition coefficients. O’Connor and Connolly (1980) report that partition coefficients for organic compounds can be reduced as much as 38% when the solid concentration is increased by an order of magnitude because closer contact between the soil solids prevents organic compounds from accessing the solid surface. Doust and Huang (1992) also report that Kp decreases as the solids concentration increases. Long-Term Tests The column tests with Sand 12 and the Portage sand– iron mixture were continued over a longer time period to assess persistence of the reactivity. Nearly 1500 pore volumes passed through the test with Sand 12, and 600 pore volumes passed through the test with the Portage sand–iron mixture. The KSA values were computed using

Eq. [5], and are shown in Fig. 9. A gradual reduction in KSA occurred between 300 and 400 pore volumes of flow for both tests, and the final KSA was two to three times lower than the initial KSA. This slight reduction in KSA may have been due to oxide formation on the surface of the iron, although an analysis of the iron surface was not conducted.

Practical Implications Calculations were made with Eq. [5] to estimate the required thickness of RBs constructed with green sands having different iron contents. The dispersion coefficient was computed using Eq. [6] using a dispersivity equal to one-tenth of the barrier thickness. The aqueous diffusion coefficient was assumed to be 9 ⫻ 10⫺6 cm2/s, which is a typical value for common organic ground water contaminants (Yaws, 1995). The effective molecular diffusion coefficient was estimated as the aqueous diffusion coefficient multiplied by a tortuosity of 0.4. Iron content of the green sand was varied from 0.1 to 12%. Typical values were assumed for dry density (1.5 Mg/m3) and specific gravity of solids (2.62) when calculating the SSA. The bulk first-order rate constant was calculated by multiplying by the SSA for a given iron content by the average KSA for the green sands obtained from the column tests. Sorption was ignored, although it may be important if degradation can occur while the herbicide is adsorbed. The seepage velocity was varied between 0.1 and 1.0 m/d, which is characteristic of field conditions (O’Hannesin and Gillham, 1998; McMahon et al., 1999; Puls et al., 1999; Kiilerich et al., 2000; Mayer et al., 2001; Yabusaki et al., 2001). At these seepage velocities, the molecular diffusion coefficient only contributes 0.1% to the dispersion coefficient. Thus, the value assumed for the molecular diffusion coefficient has no appreciable effect on the computed barrier thickness. The influent concentration (C0) was varied between 5 and 50 ␮g/L for alachlor

Reproduced from Journal of Environmental Quality. Published by ASA, CSSA, and SSSA. All copyrights reserved.

1692

J. ENVIRON. QUAL., VOL. 33, SEPTEMBER–OCTOBER 2004

may be feasible for RBs used to treat alachlor and metolachlor, there are other practical issues that need to be resolved before green sands can be used at full scale. For example, little is known about the long-term reactivity (i.e., over years or decades) of the iron in green sands. Other issues that need to be considered include in situ leaching characteristics, availability and fluctuations in the green sand supply, transportation costs to the project site, and variability in the properties of green sands (e.g., due to variations in foundry operations). Some of these issues can be inferred using existing data from full-scale RBs using ZVI. Others need to be evaluated by additional study or on a site-specific basis. Examples of site-specific issues include the potential for additional costs incurred by constructing a thicker barrier using green sands (relative to a barrier constructed with granular iron) and regulatory hurdles associated with using an industrial by-product in an application below the ground water table.

CONCLUSIONS

Fig. 10. Normalized concentration as a function of green sand iron content and barrier thickness: (a) alachlor and (b) metolachlor.

and 150 to 1500 ␮g/L for metolachlor, which bounds most conditions observed in the field (Spalding et al., 2003). Maximum contaminant levels (MCLs) for alachlor and metolachlor in the United States are 0.5 and 15 ␮g/L, respectively (Chesters et al., 1989). Accordingly, the normalized concentration required to meet the MCL (i.e., MCL/C0) ranges between 0.1 and 0.01 for the C0 that were used. Normalized concentrations for alachlor and metolachlor are shown in Fig. 10 for various zerovalent iron contents and barrier thicknesses. The horizontal lines on each graph correspond to the C/C0 needed to meet the MCL for the range of influent concentrations being considered. For seepage velocities less than 0.1 m/d, a 1-m-thick RB will reduce concentrations below the MCL for typical concentrations found in ground water if the iron content exceeds 1%. When the seepage velocity is 1.0 m/d and the barrier thickness is 1.0 m, iron contents on the order of 6 to 8% are required. Although these calculations indicate that green sands

The objective of this study was to evaluate the potential of using waste green sands from gray-iron foundries as inexpensive media for reactive barriers (RBs) used to treat ground water contaminated with alachlor or metolachlor. Batch and column tests were conducted to evaluate the reactivity and sorptive capacity of 11 green sands. Results of the tests showed that green sands have a high sorption capacity for alachlor and metolachlor, with partition coefficients ranging from 3.6 to 50.2 L/kg for alachlor and 1.0 to 54.8 L/kg for metolachlor. A linear sorption model can be used to describe the sorption isotherms. Partition coefficients obtained from the batch and column tests generally were within ⫾25%, although the partition coefficients from the column tests tended to be lower than those from the batch tests for more sorptive sands. Differences in the solid to solution ratios may be responsible for these differences in the partition coefficients. Normalized degradation rate constants obtained from the batch tests were similar for green-sand iron and Peerless granular ZVI, which suggests that the iron in the green sands has comparable reactivity as conventional granular ZVIs used in RBs. Column tests conducted with green sands also yielded similar normalized rate constants regardless of the sand source, and the rate constants obtained from the column tests were also similar to those from batch tests. Calculations showed that the required thickness of a RB constructed with green sand depends on the herbicide concentration in the influent and the iron content in the green sand. For influent concentrations less than 0.05 mg/L (alachlor) and 1.5 mg/L (metolachlor) and seepage velocities less than 0.1 m/d, the calculations suggest that RBs containing green sand will meet the maximum contaminant levels for herbicides if they are 1 m thick and the iron content is higher than 2%. For seepage velocities on the order of 1.0 m/d, a 1-m-thick RB with an iron content of at least 3% (alachlor) or 5% (metolachlor) will meet the MCL.

LEE & BENSON: TREATING CONTAMINATED GROUND WATER WITH GREEN SANDS

Reproduced from Journal of Environmental Quality. Published by ASA, CSSA, and SSSA. All copyrights reserved.

ACKNOWLEDGMENTS Financial support for this study was provided by the Wisconsin Department of Natural Resources (WDNR) and the State of Wisconsin’s Ground Water Research Advisory Council, which is administered through the Water Resources Institute (WRI) at the University of Wisconsin-Madison. This support is gratefully acknowledged. The findings in this report are solely those of the authors. Endorsement by WDNR or WRI is not implied and should not be assumed.

REFERENCES Abichou, T., C. Benson, and T. Edil. 2002. Micro-structure and hydraulic conductivity of simulated sand-bentonite mixtures. Clays Clay Miner. 50:537–545. American Society for Testing and Materials. 2000. Standard test method for measurement of hydraulic conductivity of porous material using a rigid-wall, compaction-mold permeameter. ASTM D 5886. In Annual book of standards. ASTM Int., West Conshohocken, PA. Bailey, S., T. Olin, M. Bricka, and D. Adrian. 1999. A review of potentially low-cost sorbents for heavy metals. Water Res. 11:2469–2479. Barbash, J., G. Thelin, D. Kolpin, and R. Gilliom. 1999. Distribution of major herbicides in ground water of the United States. Water Resour. Investigations Rep. 98-4245. USGS, Reston, VA. Bin-Shafique, S., C. Benson, and T. Edil. 2002. Leaching of heavy metals from fly ash stabilized soils used in highway pavements. Geo Eng. Rep. 02-14. Dep. of Civil and Environ. Eng., Univ. of Wisconsin, Madison. Briggs, G. 1981. Theoretical and experimental relationships between soil adsorption, octanol-water partition coefficients, water solubilities, bioconcentration factors, and the parachor. J. Agric. Food Chem. 29:1050–1059. Brunauer, S., P. Emmett, and E. Teller. 1938. Adsorption of gases in multimolecular layers. J. Am. Chem. Soc. 60:309. Chesters, G., G. Sisiman, B. Levy, B. Alhajjar, R. Fathulla, and J. Harkin. 1989. Environmental fate of alachlor and metolachlor. Rev. Environ. Contam. Toxicol. 110:1–74. Clay, S., T. Moorman, D. Clay, and K. Scholes. 1997. Sorption and degradation of alachlor in soil and aquifer material. J. Environ. Qual. 26:1348–1353. Comfort, S., P. Shea, T. Machacek, H. Gaber, and B. Oh. 2001. Fieldscale remediation of a metolachlor-contaminated spill site using zerovalent iron. J. Environ. Qual. 30:1636–1643. Czurda, K., and R. Haus. 2002. Reactive barriers with fly ash zeolites for in situ groundwater remediation. Appl. Clay Sci. 21:13–20. Doust, H., and J. Huang. 1992. The fate and transport of hazardous chemicals in the subsurface environment. Water Sci. Technol. 25: 169–176. Elder, C., C. Benson, and G. Eykholt. 2002. Effects of heterogeneity on influent and effluent concentrations from horizontal permeable reactive barriers. Water Resour. Res. 38:27-1 to 27-2. Eykholt, G., and D. Davenport. 1998. Dechlorination of the chloroacetanilide herbicides alachlor and metolachlor by iron metal. Environ. Sci. Technol. 32:1482–1487. Freeze, R., and J. Cherry. 1979. Groundwater. Prentice Hall, Englewood Cliffs, NJ. Gaber, H., S. Comfort, P. Shea, and T. Machacek. 2002. Metolachlor dechlorination by zerovalent iron during unsaturated transport. J. Environ. Qual. 31:962–969. Gaston, L., and M. Locke. 1994. Predicting alachlor mobility using batch sorption kinetic data. Soil Sci. 158:345–353. Golden Software. 2002. Surfer. Version 8.04. Golden Software, Golden, CO. Grundl, T., and G. Small. 1993. Mineral contributions to atrazine and alachlor sorption in soil mixtures of variable organic carbon and clay content. J. Contam. Hydrol. 14:117–128. Johnson, R., and J. Sims. 1993. Influence of surface and subsoil proper-

1693

ties on herbicide sorption by Atlantic coastal plain soils. Soil Sci. 155:339–348. Kiilerich, O., J. Larsen, and C. Nielsen. 2000. Field results from the use of a permeable reactive wall. In 2nd International Conference on Remediation of Chlorinated and Recalcitrant Compounds, Monterey, CA. 22–25 May 2000. Battelle Press, Columbus, OH. Kim, J., J. Park, and T. Edil. 1997. Sorption of organic compounds in the aqueous phase onto tire rubber. J. Environ. Eng. 123:827–834. Kolpin, D., E. Thurman, and D. Goolsby. 1996. Occurrence of selected pesticides and their metabolites in near-surface aquifers of the midwestern United States. Environ. Sci. Technol. 30:335–340. Koppensteiner, B. 1998. The degradation of the herbicides alachlor and metolachlor by iron metal in water and soil systems. M.S. thesis. Dep. of Civil and Environ. Eng., Univ. of Wisconsin, Madison. Lee, T., and C. Benson. C. 2002. Using foundry sands as reactive media in permeable reactive barriers. Geo Eng. Rep. 02-01. Dep. of Civil and Environ. Eng., Univ. of Wisconsin, Madison. Mayer, K., D. Blowes, and E. Frind. 2001. Reactive transport modeling of an in situ reactive barrier for the treatment of hexavalent chromium and trichloroethylene in groundwater. Water Resour. Res. 37:3091–3103. McMahon, P., K. Dennehy, and M. Sandstrom. 1999. Hydraulic and geochemical performance of a permeable reactive barrier containing zero-valent iron, Denver Federal Center. Ground Water 37: 396–404. Nowell, L., and E. Resek. 1994. National standards and guidelines for pesticides in water, sediment, and aquatic organisms: Application to water-quality assessments. Rev. Environ. Contam. Toxicol. 140:1– 164. O’Connor, D., and J. Connolly. 1980. The effect of concentration of absorbing solids on the partition coefficient. Water Res. 14:1517– 1524. O’Hannesin, S., and R. Gillham. 1998. Long-term performance of an in situ ‘iron wall’ for remediation of VOCs. Ground Water 36:164–170. Patakioutas, G., and T. Albanis. 2002. Adsorption-desorption studies of alachlor, metolachlor, EPTC, chlorothalonil, and pirimiphosmethyl in contrasting soils. Pest Manage. Sci. 58:352–362. Peter, C., and J. Weber. 1985. Adsorption, mobility, and efficacy of alachlor and metolachlor as influenced by soil properties. Weed Sci. 33:868–873. Puls, R., D. Blowes, and R. Gillham. 1999. Long-term performance monitoring of a permeable reactive barrier at the U.S. Coast Guard Support Center, Elizabeth City, North Carolina. J. Hazard. Mater. 68:109–124. Roux, P., K. Balu, and R. Bennett. 1991. A large-scale retrospective ground water monitoring study for metolachlor. Ground Water Monit. Rev. 11:104–114. Satapanajaru, T., S. Comfort, and P. Shea. 2003. Enhancing metolachlor destruction rates with aluminum and iron salts during zerovalent ion treatment. J. Environ. Qual. 32:1726–1734. Seybold, C.A., and W. Mersie. 1996. Adsorption and desorption of atrazine, deethylatrazine, deisopropylatrazine, hydroxyatrazine, and metolachlor in two soils from Virginia. J. Environ. Qual. 25:1179– 1185. Spalding, R., M. Exner, D. Snow, D. Cassada, M. Burbach, and S. Monson. 2003. Herbicides in ground water beneath Nebraska’s management systems evaluation area. J. Environ. Qual. 32:92–99. Van Genuchten, M. 1981. Analytical solutions for chemical transport with simultaneous adsorption, zero-order production and firstorder decay. J. Hydrol. (Amsterdam) 49:213–233. Wang, Q., Y. Wang, and W. Liu. 1999. Adsorption of acetanilide herbicides on soils and its correlation with soil properties. Pestic. Sci. 55:1103–1108. Yabusaki, S., K. Cantrell, B. Sass, and C. Steefel. 2001. Multicomponent reactive transport in an in situ zero-valent iron cell. Environ. Sci. Technol. 35:1493–1503. Yaws, C. 1995. Handbook of transport property data. Gulf Publ., Houston, TX. Zytner, R. 1992. Adsorption-desorption of trichloroethylene in granular media. Water Air Soil Pollut. 65:245–255.