Environ. Sci. Technol. 2001, 35, 1231-1239
Phytoremediation of MTBE from a Groundwater Plume MARJORIE S. HONG, WALTER F. FARMAYAN, IRA J. DORTCH, AND CHEN Y. CHIANG Equilon Enterprises, LLC, Westhollow Technology Center, Houston, Texas 77082 SARA K. MCMILLAN AND JERALD L. SCHNOOR* Department Civil & Environmental Engineering, The University of Iowa, Iowa City, Iowa 52242
The feasibility of phytoremediation to both remediate and hydraulically contain a methyl tert-butyl ether (MTBE)contaminated groundwater plume was investigated in a threephase study that included the following elements: (i) a laboratory bioreactor study that examined the fate and transport of 14C-radiolabeled MTBE in hybrid poplar trees, (ii) a novel approach for a mathematical modeling study that investigated the influence of deep-rooted trees on unsaturated and saturated groundwater flow, and (iii) a field study at a Houston site with MTBE-contaminated groundwater where hybrid poplar trees were planted. In the laboratory study, the predominant fate pathway was uptake and evapotranspiration of [14C]-MTBE from leaves and stems of poplar cuttings rooted in hydroponic solution. The modeling study demonstrates that phytohydraulic containment of MTBE in groundwater by deep-rooted trees can be achieved. The field study demonstrated significant groundwater uptake of groundwater by deep-rooted trees via direct measurement in the first three seasons. The use of vegetation may provide a cost-effective insitu alternative for containment and remediation of MTBEcontaminated groundwater plumes.
Introduction The production and use of methyl tert-butyl ether (MTBE) has increased in response to the 1990 Clean Air Act; it is used widely as a gasoline oxygenate to decrease urban smog and to enhance octane performance. Annual domestic production increased from 1.38 billion lb in 1984 to 24.1 billion lb in 1993 (1). While less than 3% of MTBE levels detected in drinking water wells exceeded the lower limit of the health advisory, MTBE is the second most frequently detected chemical in urban shallow groundwater samples (2). MTBE is volatile (vapor pressure 3.27-3.35 × 104 Pa; Henry’s constant 59-305 Pa m3 mol-1), highly soluble in water (water solubility 23.2-54.4 g L-1; log Kow ) 0.94-1.16), and therefore has great potential to migrate in environmental media. MTBE can contaminate groundwater supplies by preferentially leaching from gasoline plumes (2, 3). Constituents in gasoline, such as BTEX (benzene, toluene, ethyl benzene, and xylenes), are preferentially sorbed and degraded while MTBE may be transported beyond the other contaminants. * Corresponding author phone: (319)335-5649; fax: (319)335-5777; e-mail:
[email protected]. 10.1021/es001911b CCC: $20.00 Published on Web 02/10/2001
2001 American Chemical Society
The problems associated with the increased mobility of MTBE are compounded by its tertiary carbon structure and ether linkage rendering this compound recalcitrant to biological attack. Hence, MTBE poses a challenge for remediation of contaminated groundwater. In this paper, we report on a full-scale phytoremediation effort for treatment of a groundwater plume contaminated with MTBE at a site in Houston, TX. (The plume is currently being captured by a preexisting pump-and-treat system; groundwater extraction systems are widely used in environmental remediation to hydraulically contain plumes.) In this research, the trees are intended to serve as natural “pumps”, not only to extract water and create a cone of depression in the saturated zone but also to potentially remove and remediate MTBE from groundwater. If phytoremediation is judged successful, the groundwater pumpand-treat system would no longer be needed. A groundwater flow modeling study was performed to assess the potential of deep-rooted hybrid poplar trees to hydraulically contain the MTBE plume at the Houston site. This study evaluates the impact of deep-rooted hybrid poplar trees on the groundwater hydraulics at the site, in particular, the ability of poplars to locally depress the water table and create a capture zone for plume containment. The field design was guided by detailed mathematical modeling of water movement in both the unsaturated and the saturated zones, accounting for evapotranspiration and site-specific hydrogeologic conditions. An integrated, parallel effort was also undertaken in the laboratory using poplar cuttings exposed to 14C-radiolabeled MTBE-contaminated media. These experiments allow the issues of uptake and fate to be addressed in a tightly controlled, physical environment where constituent mass balances can be achieved. These experiments in the lab and field are meant to demonstrate the potential for phytoremediation of MTBE plumes.
Materials and Methods Laboratory Study. Uptake of 14C-radiolabeled MTBE by hybrid poplar cuttings was studied in hydroponic solution conducted in a manner similar to methods reported by other investigators (4, 5). Hydroponic studies used one-quarter strength Hoagland’s inorganic nutrient solutions, fed to the root zone of batch reactors containing hybrid poplar cuttings (Populus deltoides x nigra DN-34, Imperial Carolina). Eightinch long cuttings (0.75-1.0 in diameter) were utilized. The cuttings were affixed with predrilled screw caps and predrilled Teflon-lined septa. Teflon tape was wrapped around the stem, and acrylic caulk was used to seal the caps and septa to the cutting. Poplar cuttings were rooted first in hydroponic solution until a vascular root system appeared and then placed in 1-L bioreactors containing 400 mL of nutrient solution. The bioreactors consisted of 1-L Erlenmeyer flasks modified by attaching a sampling port to the bottom and top of each flask. Controls were included in the experiment in triplicate to isolate the uptake of MTBE from trees by the quantification of losses from the system. First, a capped control with only one-quarter strength Hoagland’s solution and [14C]-MTBE was included to determine the contribution of biodegradation possibly occurring inside the hydroponic solution and sorption to the glass. To determine loss of compound through leaks in the system, controls were used with solid glass rods in the place of the cutting. Finally, two sets of planted cuttings were cut off just above the cap to create excised tree controls, which were necessary to observe the effect of live roots and VOL. 35, NO. 6, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1231
FIGURE 1. Conceptual model for the three-dimensional unsaturated/saturated flow modeling study performed to assess the impact of trees for obtaining phytohydraulic capture of an MTBE plume at a site in Houston, TX.
FIGURE 2. MTBE concentrations (August 1998) and areas planted with hybrid poplar according to the modeling study at a site in Houston, TX. an excised tree on [14C]-MTBE removal. Two sizes of trees were used; the first was approximately the same diameter as the glass rod, and the second was similar to the full tree reactors. 1232
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 6, 2001
Bioreactors were treated with a mixture of 4.32 mg of cold MTBE (Fisher Chemical, Inc.; chemical purity of 99.9%) and 7.07 µCi of [14C]-MTBE (NEN Life Science Products; specific activity of 1.48 Ci/mmol and a radiochemical purity of 97%).
FIGURE 3. Mass of MTBE remaining in aqueous solution as determined by LSC analysis. Impulse input of 4.32 mg of MTBE at t ) 0 for an initial concentration of 10.2 mg/L in solution. Error bars on the full trees (n ) 5), excised trees (n ) 3), and capped controls (n ) 3) indicate one standard deviation. Recovered radiolabel was calculated on a mass basis as [14C]-MTBE.
FIGURE 4. Mass of MTBE remaining in aqueous solution as determined by LSC analysis. Impulse input of 4.32 mg of MTBE at t ) 0 for an initial concentration of 10.2 mg/L in solution. Error bars on the excised trees (n ) 3), glass rods (n ) 3), and capped controls (n ) 3) indicate one standard deviation. MTBE was uniformly labeled across all five carbons. A laboratory study indicated little or no toxicity to plants at MTBE concentrations less than 1000 mg/L. The trees were maintained in a laboratory growth chamber at 28 °C under artificial growth lights, which provided a photon flux at the
leaf surface of 100-160 µmol m-2 s-1 (photosynthetically active range). Samples were collected daily to monitor the disappearance of [14C]-MTBE from hydroponic solution. Bioreactors were weighed daily to gravimetrically monitor transpiration, and VOL. 35, NO. 6, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1233
FIGURE 5. Hydroponic mass balance: percent recoveries of 14C-radiolabel distributions for full tree reactors after 10 d as determined by LSC analysis at the termination of the experiment. Results are given as a percent of the total [14C]-MTBE initial spike (6.0 µCi) added to reactors at t ) 0. All recoveries were measured directly except transpiration from leaves and leaks, which were estimated by differences between the controls and the full tree experiments. nutrient solution was added as needed. Prior to adding nutrient solution, all headspace was removed through a series of traps to capture any MTBE that was present in the headspace above the hydroponic solution. A needle was inserted into the top Mininert and connected to tubing that pumped air out through an activated carbon trap and captured pure compound and any other organic metabolites (Orbo tube 32 large, Supelco), followed by a CO2 trap that allowed air to bubble through 1 M NaOH to trap the CO2. Less than 1% of 14CO2 was detected in the trap. To extract the trapped MTBE from the activated carbon, the carbon was poured into 4-mL glass vials and submerged in 2 mL of methanol (Fisher) for 24-48 h. Samples of 100 µL were taken and injected into 15 mL of Scintiverse for counting on the liquid scintillation counter (LSC). One-milliliter samples were taken from the NaOH in the CO2 traps and injected into Ultima Gold scintillation cocktail (Packard) for counting on the LSC. After 10 d, total 14C uptake was quantified in the poplar roots, lower stem, upper stem, leaves, and petioles via oxidation to 14CO2 in a RJ Harvey Bio-Oxidizer and 1234
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 6, 2001
subsequently by LSC. This oxidation was >96% efficient and was calibrated with pre- and post-analysis efficiency testing. Phytohydraulic Modeling Study. An MTBE plume at the Houston site extends several hundred feet and is present at a maximum concentration of 40 ppm downgradient of the original source area. The water table at the Houston site is approximately 10 ft below grade. Groundwater flows in an easterly direction with an average gradient of 0.003 ft/ft. The unsaturated zone comprises primarily a relatively impermeable clay (hydraulic conductivity of 10-7-10-6 cm/s), whereas the saturated zone, thought to be an old river channel, is composed of a more permeable silty sand (hydraulic conductivity of 10-4-10-3 cm/s). Rainfall in Houston is typically in the range of 100 cm/yr, and the potential evaporative demand as estimated by the Penman equation (8) is in the range of 150 cm/yr. The depth to groundwater at the site (9-10 ft below ground level) is considered near the practical limit with respect to poplar root penetration and water uptake from the capillary zone. It is still within achievable physiological limits. Rainfall
FIGURE 6. Pressure head (cm) contours for a horizontal slice (at elevation z ) 0 cm) in the soil column for a site in Houston, TX: (a) initial model condition without trees; (b) predicted model results with phytohydraulic containment. at the site is less than the potential evapotranspiration demand, but it may be larger than desirable considering the goal of phytoremediation to uptake water from the saturated zone instead of the shallow vadose zone. Hence, both of these factors make the field study a rigorous test of the feasibility of achieving phytohydraulic capture. The modeling, intended primarily as a preliminary assessment of feasibility of the phyto-based hydraulic capture concept, was performed using two public domain codes; a finite element-based three-dimensional unsaturated/saturated flow/transport code, SWMS_3D, developed by the U.S. Salinity Laboratory (6); and a finite difference-based onedimensional code, UNSAT-H, developed by the Pacific Northwest Laboratory (7). SWMS_3D has a number of capabilities that are especially suited for the current study. The code allows simulation of time-varying root water and contaminant uptake, surface evaporation, and infiltration.
The code also provides a means for estimating actual transpiration as a fraction of potential transpiration, based upon an experimentally determined “root-stress” curve provided by the user. This root-stress curve comprises an attenuation factor, applied to potential transpiration that varies depending upon the energy state (or head) of the water in the unsaturated zone (which can vary both spatially and with time in the domain). A root (or sink) zone of any desired shape or size within the domain can be assigned. Hence, roots concentrated near the ground surface or near the water table can be simulated. Spatially varying local water uptake within the root mass may also be taken into account by application of weighting factors. UNSAT-H is a one-dimensional code that has similar evapotranspiration simulation capabilities. It also has the capability of estimating daily potential evapotranspiration from site meteorological data using a form of a well-known VOL. 35, NO. 6, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1235
FIGURE 7. Pressure head (cm) contours for a vertical slice in the soil column for a site in Houston, TX: (a) initial model condition without trees; (b) predicted model results with phytohydraulic containment (0 cm head and greater represents saturated conditions, whereas
