Environ. Sci. Technol. 2008, 42, 1268–1275
Direct Measurement of VOC Diffusivities in Tree Tissues: Impacts on Tree-Based Phytoremediation and Plant Contamination KRISHNA K. BADURU, STEFAN TRAPP, AND JOEL G. BURKEN* Department of Civil, Architectural and Environmental Engineering, Room 224 Butler Carlton Hall, 1870 Miner Circle, Missouri University of Science and Technology, Rolla, Missouri 65409, and Institut of Environment & Resources, Technical University of Denmark, Bygningstorvet 115, DK-2800 Kongens Lyngby, Denmark
Received June 25, 2007. Revised manuscript received November 7, 2007. Accepted November 19, 2007.
Recent discoveries in the phytoremediation of volatile organic compounds (VOCs) show that vapor-phase transport into roots leads to VOC removal from the vadose zone and diffusion and volatilization out of plants is an important fate following uptake. Volatilization to the atmosphere constitutes one fundamental terminal fate processes for VOCs that have been translocated from contaminated soil or groundwater, and diffusion constitutes the mass transfer mechanism to the plant-atmosphere interface. Therefore, VOC diffusion through woody plant tissues, that is, xylem, has a direct impact on contaminant fate in numerous vegetation-VOC interactions, including the phytoremediation of soil vapors and dissolved aqueous-phase contaminants. The diffusion of VOCs through freshly excised tree tissue was directly measured for common groundwater contaminants, chlorinated compounds such as trichloroethylene, perchloroethene, and tetrachloroethane and aromatic hydrocarbons such as benzene, toluene, and methyl tert-butyl ether. All compounds tested are currently being treated at full scale with tree-based phytoremediation. Diffusivities were determined by modeling the diffusive transport data with a one-dimensional diffusive flux model, developed to mimic the experimental arrangement. Wood-water partition coefficients were also determined as needed for the model application. Diffusivities in xylem tissues were found to be inversely related to molecular weight, and values determined herein were compared to previous modeling on the basis of a tortuous diffusion path in woody tissues. The comparison validates the predictive model for the first time and allows prediction for other compounds on the basis of chemical molecular weight and specific plant properties such as water, lignin, and gas contents. This research provides new insight into phytoremediation efforts and into potential fruit contamination for fruit-bearing trees, specifically establishing diffusion rates from the transpiration stream and modeling volatilization along the transpiration path, including the trunk and branches. This work also has
* Corresponding author e-mail:
[email protected]. 1268
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 4, 2008
importance in other plant-VOC interactions, such as potential uptake from the atmosphere for hydrophobic compounds and also uptake from vapor-phase soil contaminants.
Introduction Phytoremediation for groundwater and soil that are contaminated with volatile organic compounds (VOCs) includes vapor-phase uptake processes recently discovered and investigated. Lingering concerns related to uncertain contaminant fate and potential risks have been detrimental to regulatory approval and caused public concern that has led to specific research efforts on entry to the food chain (1, 2). Concerns are well-founded as phytoremediation inherently brings known contaminants into contact with vegetation and thus the building blocks of the food chain. Phytoremediation is a long-term process and includes ecorestoration aspects, that is, improving ecological health, which makes phytoremediation totally unique in remediation approaches; therefore, a comprehensive understanding of contaminants’ fate is necessary. Recent discoveries have been made in the phytoremediation of problematic VOCs such as chlorinated solvents and aromatic hydrocarbons that persist in the environment. Vapor-phase pathways of VOCs in phytoremediation have been shown, including, as recently noted, uptake (3) and volatilization, which has been termed “phytovolatilization” from the earliest studies on VOC phytoremediation (4, 5). These transport processes occur in addition to other “loss” or attenuation processes, such as sorption to plant tissues, degradation in the tree biomass due to phytodegradation or endophytes, and degradation in the soil and root zone. These mechanisms can all be important, particularly rhizodegradation, which can prohibit uptake and has been hypothesized as the main mechanism in hydrocarbon phytoremediation, as the rhizosphere is aerobic and can lead to high degradation rates. Among the various fates, contaminant uptake and subsequent volatilization has been of great interest and noted as an area lacking in knowledge (6). Vroblesky and colleagues first hypothesized diffusion and volatilization from the xylem tissues to be the contaminant-loss mechanism following uptake (7). Analysis of tree trunk samples at a contaminated site showed the presence of trichloroethylene (TCE) and cisdichlorothene and a reduction in concentration with height up the trunk. Similar contaminant profiles were confirmed in field and laboratory studies for chlorinated VOCs, including capture of the VOCs in diffusion traps placed upon the trees (8–10). Similar decreasing contaminant profiles with height and volatilization have been observed for methyl tert-butyl ether (MTBE) in alfalfa plants (11) and poplar trees (12, 13). While diffusion and volatilization have been observed, the processes are not adequately understood at a level to allow for the prediction of chemical fate. Diffusion is the key, rate-limiting process in phytovolatilization, limiting radial contaminant transport to the plant-atmosphere interface where volatilization occurs. The diffusivity in wood tissues predicates the loss from the transpiration pathway, that is, xylem flow, and thereby the relative concentrations that may reach leaves and even fruit. Studies evaluating sorption into or desorption from plant tissues estimated diffusivity on the basis of field and laboratory study results as well as molecular-based theory, but none were based upon measurements through tissues. Some initial TCE diffusivities were estimated from volatilization rates from plant tissues that had been grown in TCE10.1021/es071552l CCC: $40.75
2008 American Chemical Society
Published on Web 01/16/2008
contaminated solutions (14). Mackay and Gschwend (15) investigated the sorption of monoaromatics to wood in subsurface fill and calculated effective wood diffusivities of monoaromatic compounds using partition coefficients and aqueous diffusivities. Wood diffusivities were hypothesized to be influenced by aqueous diffusivities and a degree of physical hindrance due to microscale partitioning between wood and chemicals, along the diffusion path. Effective diffusivity values were determined from a regression of kinetic profiles of sorption data. The model included diffusion in the gas phase and approximating the resistance by a tortuosity factor for the diffusion path in the wood tissue (15). Whole-plant mathematical models have also been used to estimate diffusivities; however, these models include many assumptions with respect to multiple, concurrent processes as well as physical size, geometry, and homogeneity of plants and must be calibrated with laboratory data collected under controlled conditions at a minute scale when considering trees. One whole-plant model was developed to evaluate experimentally observed MTBE volatilization from alfalfa tissues (11). This model assumed a cylindrical-shaped plant stem and uniform transpiration in the stem. MTBE diffusion was clearly measured as efflux from excised tissues, and the model successfully tracked the efflux, that is, phytovolatilization. A unique output of the modeling was a radial profile of contaminant concentrations. In the model, alfalfa stems were assumed to be conducting uniformly throughout entire the stem, and alfalfa plants do not have vascular woody xylem like many trees; thereby, the model assumptions would be violated in application to woody trees often used in phytoremediation for VOCs. In trees, water transport is notably nonuniform in xylem tissues. Perennial trees that are easily established and grow rapidly are desired in many phytoremediation applications. Having the desired traits, poplars (Populus sp.) and willows (Salix sp.) are commonly utilized. In both families, water is prominently conducted in outer annual rings, since Salix and Populus species are diffusive porous (16). On the basis of this concept, a different whole-plant model was developed considering the tree trunk as a ring model and assuming an average diffusion path length (17). The development of the model was mainly based on the advective upward transport and diffusive transport radially outward. Effective diffusivities were estimated by calibrating the model solution to the obtained data in hydroponic laboratory studies conducted with hybrid poplar trees dosed with TCE. The effective diffusivities in this model were hypothesized to be influenced by advection and dispersion along the transpiration path. Due to approximations and inherent complexity and uncertainty associated with the whole-plant systems evaluated, the model was unable to accurately approximate VOC flux from the plant tissues, that is, volatilization. Both models predict an exponential decrease in VOC concentrations with height and in the radial direction of the stem or trunk. The radial decreasing concentration has also been confirmed in trees at phytoremediation sites (7–9). The above models focused on explaining the transport mechanism of VOCs in the plant as a whole rather than on specific determination of the diffusivities. These efforts require numerous assumptions relating to multiple transport processes in actively transpiring, whole-plant experiments, such as advection and dispersion. These assumptions and assuming homogeneous water conduction and tissues are necessary to solve the models but are innately inaccurate and may in part lead to the wide range of values noted. Other modeling efforts focused upon sorption and diffusion of the VOCs into tissues (wood), investigating the removal of VOCs from solution (15, 18) or from the atmosphere (19) where wood diffusivity calculations were based on aqueous diffusivities and the degree of retardation
or physical hindrance encountered moving in the wood matrix. These models are not subject to the whole-plant assumptions above but must, in turn, assume that the chemical loss from solution is (1) kinetically controlled by diffusion and (2) only due to sorption. In these studies focusing on removal from a contaminated environment, the end point was the key experimental objective. The study presented here focuses on the direct determination of diffusion coefficients from the measurement of contaminant transport in isolated xylem tissues of hybrid poplar trees. Compounds investigated herein have all been targeted with full-scale phytoremediation systems, making the understanding of contaminant transport at the tissue level of considerable importance. No model assumptions are necessary with respect to advective flow, plant geometry or homogeneity, or metabolism in tissues. The directly measured component is the diffusive flux of contaminants through tree tissues.
Materials and Methods Diffusion Experiment. Tree cores (5 mm diameter × 6-8 mm in length) were isolated from a 15-cm-diameter poplar stem (P. deltoides x nigra, clone DN34) using an increment borer. Two Erlenmeyer flasks of 1.1 L volume were modified to have four threaded ports. The freshly extracted tree cores were sleeved into a Teflon tube of 5-mm-diameter and 7-cmlength, such that the cores snugly fit into the Teflon tubes. The other end of the Teflon tube was fitted with a threaded mininert valve, and the remaining space in the tube was filled with deionized (DI) water. Each tube assembly was attached to one of the four outlets of the Erlenmeyer flask, Figure 1. The junction between the Teflon tube assembly and the flask was sealed water-tight with Teflon tape (not shown). A mininert valve was fitted to the fourth outlet of the flask. For each compound tested, six Teflon tubes with cores were constructed to make duplicate arrangements, Figure 1. The flasks were filled headspace-free with a solution of the compound to be analyzed, with a known concentration, C1. The VOC solution (DI water with VOC added) in the flask was kept mixed throughout the experiment using a magnetic stirrer bar. Sodium azide, 1 g/L, was added as a biocide to the solution in the flask and to the DI water in the Teflon tube. The concentration of VOCs in the Erlenmeyer flask was maintained at an approximately constant level (targeting 16 mg/L). This was achieved by means of measuring the concentration in the flask daily and dosing with a stock solution to target 16 mg/L. Experimental concentrations were selected for analytical ease, considering analytic linearity and detection limits at minor fractions of the initial concentration. The concentration loss in the flask due to sorption and diffusion was usually minimal due to the volume of the flask (V1 ) 1100 mL) compared to that of Teflon tubes (V2 ≈ 1.5 mL). Each day, sampling was done by collecting 2 µL from each Teflon tube and from the flask. The samples were analyzed using gas chromatography, an HP5890 apparatus with electron capture and flame ionization detectors (ECD/ FID). ECD was used for analyzing the chlorinated solvents (TCE; perchloroethane, PCE; and tetrachloroethane, TeCA), and FID was used to analyze the hydrocarbons (MTBE, benzene, and toluene). Peak areas of the samples were calibrated with prepared known standards to obtain the concentration. Sampling was continued for 7-15 days. Partitioning Experiment. The compounds investigated in partitioning experiments were benzene, toluene, and MTBE, as partitioning for the chlorinated solvents had been previously conducted (20). Tree cores 5 mm in diameter and 5 cm in length were obtained from a 15-cm-diameter poplar tree stem. Each core was weighed and placed in a 21 mL vial with DI water. Vials were then dosed with contaminants from VOL. 42, NO. 4, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1269
FIGURE 1. Schematic of diffusion experiment setup. The reactor was maintained at constant concentration (C1), stirred continuously, and completely filled with a solution of VOC, that is, no headspace. a prepared stock solution and were sealed headspace-free with septa and a crimp top. Five different concentrations were prepared (0.6-9.0 mg/L) with three replicates for each concentration. Sampling was done by collecting aqueous samples of 3 µL and analyzing them as noted above. Samples were analyzed each day until a constant concentration was achieved (less than 5% fluctuation in concentration). The steady concentration indicates that other loss mechanisms (leaks or degradation) were most likely not contributing to decreased aqueous concentrations. Final aqueous concentrations were recorded, CW, and the sorbed mass was calculated by contaminant loss from the solution. Earlier studies performed in this manner were also tested for desorption, verifying that sorption was the mechanism leading to the decreased aqueous concentration over time. To determine core dry weight, cores were dried at 105 ( 2 °C and weighed. The sorbed tissue concentration was calculated, Ct. Detailed calculations and procedures of the experiments have been previously described (20).
Results Partitioning Experiment. Hydrocarbon compounds were sorbed to tree tissues from the aqueous phase marked by a reduction of the aqueous concentration over time. Tissue sorption isotherms were obtained for Ct and Cw, and while a slight nonlinear trend was noted on the isotherms, linear isotherms were accepted over the range of concentrations, Figure 2. At higher solute concentrations, nonlinear sorption (i.e., Freundlich) models would most likely be applicable; however, such concentrations are rare in environmental settings. The biomass-water partitioning coefficient (Kwood) is the ratio between the concentration of a compound in the tree core (mg/g) and that of water (mg/L) at equilibrium. With the linear sorption isotherms, the Kwood values, listed in Table 1, were accepted for use in the model analysis below. Measured Kwood coefficients were within the range of values obtained by other researchers, Table 1. A linear relationship was observed between log Kwood and log Kow, Figure 3. Partitioning coefficients obtained from this study are comparable to previous studies and calculated values. Mackey and Gschwend (15), Trapp et al. (19), and Boving and Zhang (18) were able to establish a linear relationship between log Kow, and this is illustrated in Figure 3. Mackay and Gschwend were also able to establish a relation between log Kow and partitioning to the lignin content in the wood (15). This research indicated that wood-water partitioning coefficients can be estimated from log Kow, and the percentage of lignin content in the wood. In earlier work, another linear relation1270
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 4, 2008
FIGURE 2. Isotherms for partitioning of aromatic VOCs from water to tree cores at equilibrium. Aqueous concentrations were measured, and sorbed tissue concentrations were calculated from aqueous concentration changes. Error bars indicate 95% CI for measured water concentrations. ship between log Kwood and log Kow was established (19), Figure 3. The variability observed shows a clear difference in values for different tree species and also a slight difference in experimental approaches. In summary, the relationships cited above appear to be consistent in general, order-of-magnitude estimates, but when specific partitioning values for contaminant-tissue (type, age, etc.) are desired, Kwood values should be determined directly. Diffusion Studies. Tested contaminants diffused through xylem tissue in the absence of any advection or dispersion process as shown by increasing concentration in the Teflon tubes, C2. Wood effective diffusivity was assumed to be independent of concentration and one-dimensional across the grain of the tissue. C2 remained less than 10% of the flask concentration C1 over the short period of experimentation. The experimental duration was limited (7-15 days) to ensure that the tissues did not degrade and were essentially unchanged from in vivo xylem tissues. The concentration increase of contaminants varied for each core of different lengths and volumes in the sampling compartments (Teflon tubes). Concentrations were plotted as the -ln[(C1 - C2)/C1] with time, as the slope of the curve has been used in the model to determine effective diffusivities, using the model developed, Figure 3. Diffusivity tests were run perpendicular to the conducting tissues, to mimic the radial transport, that is, horizontal. This work makes no efforts to evaluate vertical diffusion as transport is assumed to be dominated by advection. However, in earlier studies, diffusion of nonsorbing
TABLE 1. Wood-Water Partitioning Coefficients from (A) Current Study and Reported Values and for (B) Compounds Previously Investigated (A) Wood-Water Partitioning Coefficients from Current Study and Reported Values benzene (mL/g) toluene (mL/g) MTBE (mL/g) a
poplar tree cores poplar treesb Ponderosa pinec Douglas fird poplar cuttingse
8.6 3.95–8.95 6.3–6.9 11.4–12.6
13.9 7.9–23.3 12.0–14.0 16.0–18.0
4.8 1.03–1.04 3.8–5.8
(B) Wood-Water Partitioning Coefficients for Compounds Previously Investigated TCE (mL/g) poplar tree cores poplar trees
PCE (mL/g)
38.5 ( 6.8 4.8–41.4b
f
g
51.0
TeCA (mL/g) 20.7 ( 3.0f 4.7–91.8b
a
Measured values for poplar cores from current study. Calculated values using Mackay and Gschwend (15) equation for 20% lignin content. c Measured values for Ponderosa pine by Mackay and Gschwend (15). d Measured values for Douglas fir by Mackay and Gschwend (15). e Measured values for poplar cuttings Ma et al. (12). f Measured values for poplar tree cores Ma and Burken (20). g Measured values for poplar tree cores Struckhoff, et al. (3).
FIGURE 4. Plot showing -ln[(C1 - C2)/C1] versus time (h) for compounds investigated. Plot shows concentration in six individual cores with slightly different lengths, volumes, and masses. Equating mass rates from the core and into the sampling port (Teflon tube), equation 3a is obtained:
b
-DA
∂Cw ∂C2 )V ∂x ∂t
(3a)
Equation 3a can be written as -D
∂Cw ∂Cw V ) × ∂X KwoodA ∂t
(3b)
where Cw is the concentration in wood, C1 is the constant concentration of VOC in the flask, C2 is the concentration of VOC in the Teflon tube (varies as a function of time), Kwood is the equilibrium partition coefficient between water and wood, X refers to the distance along the core, V is the volume of solution in the Teflon tube, and A is the cross-sectional area of the core. The solution of eq 1 with the above given boundary conditions (eqs 2a and 3a) was given by Carslaw and Jaeger (22) in their work Conduction of Heat in Solids. Suitable changes were made to modify the solution for a heat equation solution to an analogous diffusion equation, shown by ∞
FIGURE 3. Relation between log Kwood and log Kow for the hydrocarbons investigated and comparison with previous research (15, 18, 19). Error bars indicate 95% CI for Trapp’s data (19). salts in wood tissues was shown to be lower than aqueous diffusivities, and transverse diffusivities in wood were significantly lower, Dwood/Daq ) 0.02:0.032, showing much greater radial versus axial resistance (21). Model Development to Determine Experimental Effective Diffusivity. To determine effective diffusivity, a mathematical model was used which was developed on the basis of Fick’s second law of diffusion with suitable initial and boundary conditions. Instantaneous equilibrium is assumed to take place between aqueous solution and the exposed face of the tree core. Diffusion resistance at the interface between the core and aqueous solution is assumed to be much lower than diffusion within the core itself. Other losses were assumed to be minimal. Fick’s one-dimensional diffusion equation with constant diffusion coefficient (D) is ∂2Cw ∂Cw )D 2 ∂t ∂X
(1)
at X ) 0, Cw ) KwoodC1
(2a)
at X ) L, Cw ) KwoodC2
(2b)
{
2(Rn2 + h2) sin(RnX) e-DRnt C2 )1C1 Rn[L(Rn2 + h2) + h] n)1
∑
2
}
(4a)
where Rn are the roots of the transcendental equation, R tan R L ) h, and h ) (AKwood)/V. Equation 4a expands in infinite series fashion for an infinite number of R values (roots). The series converges rapidly, especially at larger times, making the first term of eq 4a dominant and the rest of the terms approach zero. So, for mathematical convenience, the first term of eq 4a and the first root of the transcendental equation were assumed to be dominant. So, for the first root R1, and at X ) L, eq 4a can be written as
{
}
2(R21 + h2) sin(R1L) C1 - C2 2 ) e-DR1t C1 R1[L(R21 + h2) + h]
(4b)
A plot of -ln(C1 - C2)/C1 against t gives a straight line slope, -DR12 (first root), where R1 depends on the geometry and liquid-wood partition coefficient, Figure 4. Partition coefficients determined herein and from earlier research (20) were used in the model, Table 1. Effective diffusivities were calculated from the slope and are listed in Table 2.
Discussion In the model solution, the first term is assumed to dominate, and at greater times, the negative exponentials VOL. 42, NO. 4, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1271
TABLE 2. Chemical Properties and Diffusivities of VOCs in Wood and Experimental and Reported Values from This Study diffusivities, D × 10-7 (cm2/s) compound
current studya
reported values
molecular weight (g/mol)
log Kow
solubility (mg/L)
TCE PCE TeCA MTBE benzene toluene
1.24 ( 0.54 0.32 ( 0.05 0.68 ( 0.29 1.78 ( 0.87 2.98 ( 1.11 1.92 ( 0.60
4.00–25.00,b 0.10–1.00c
131.40 165.80 167.90 88.15 78.10 92.10
2.42 3.40 2.39 1.06 2.13 2.69
1366 150 2900 51,000 1789 518
4.14–8.00d 0.80e 0.50e
a
Measured values for hybrid poplar tree cores with propagated uncertainties. b Measured values by Ma and Burken (20) for hybrid poplar tree. c Data cited from Hu et al. (32) for poplar stems. d Measured values by Zhang et al. (11) for alfalfa plants. e Measured values by Mackay and Gschwend (15) for Douglas fir.
TABLE 3. Calculated Half-Distances z1/2 (m) for Loss by Volatilization from a Stem (dia = 20 cm) and a Branch (dia = 2 cm) compound
z1/2 stem (m)
z1/2 branch (m)
KAWa
Db (10-7 cm2/s)
TCE PCE TeCA MTBE benzene toluene
8.09 3.69 3.12 88.1 9.2 10
0.054 0.025 0.021 0.588 0.061 0.067
0.34 0.84 1 0.024 0.23 0.23
23.5 38.9 113 17.3 92.4 52.7
summation of terms to estimate Dwood that was used to estimate DWood for 10 compounds. Dwood ) TfwDw + TfGDG
where T represents tortuosity, fW and fG are the fractions of chemicals in water and gas, respectively, and DW and DG are the diffusivities in water and gas. The chemical partitioning to fractions in the respective compartments are calculated by eqs 6 and 7.
of equation 4a vanish, making the first term significant. The influence of lesser terms is negligible in calculating effective diffusivities. A similar assumption was adopted to calculate the gaseous diffusivities in porous soil media (23). To validate the assumption, the sum of the terms in equation 4a was performed for selected cores. The summation was carried until the series converged, and a constant value of the sum was achieved. The summation value was compared to the value of the first term in equation 4a for various times. The difference was also calculated and found to be less than 6–8% at lesser times (0-3 days) for most of the compounds (8–12% for PCE), and less than 4% at greater times (6-8 days) for all the compounds investigated. This suggests that the technique used is adequate and explicit assumptions are valid. The uncertainties associated with the linear model fit, measured quantities (length of the core, area, volume, etc.), and partition coefficients (Kwood) were also estimated. Errors from individual parameters (data, measured quantities, and Kwood) were propagated toward effective diffusivity for each tree core sample, reported in Table 2. While this is the first study to determine effective diffusivities of problematic VOCs in live, excised xylem tissue, earlier models were based on physicochemical properties of the compounds and physiology of the plant species to estimate diffusivities of organic chemicals in wood tissues. Models were based upon measured aqueous diffusivities of empirical molecular weight-aqueous diffusivity relationships. Models also incorporated some degree of physical hindrance and microscale partitioning between chemicals in wood along the diffusion path (15). Vapor-phase diffusion was included in a similar approach that included diffusion through both the fraction of pores filled with gas and that of pores filled with water (19). Effective diffusivities in each phase were estimated, and the fractions of contaminants in each phase were calculated using partitioning coefficients and a resistance term based on the tortuosity of the path. Equation 5 shows a 1272
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 4, 2008
PW,Wood KWood
(6)
PA,Wood × KAW KWood
(7)
fW )
a
Data taken from Rippen (33), except TeCA, which is estimated. b Calculated according to Trapp et al. (19) with 10% gas pores.
(5)
fG )
where PW,Wood and PA,Wood are the portions of water and air in the wood, respectively, and KAW is the air–water partition coefficient (i.e., dimensionless Henry’s law constant). A similar version of this model was used herein to calculate effective diffusivities for comparison to the experimentally determined values. The predictive model calculated DG and DW on the basis of the molecular weight of the selected contaminants, eqs 8 and 9. Values for KWood were also calculated using empirical relationships presented previously (eqs 1 and 2 in Trapp et al. (19)) DW ) DO2 ×
√32 √M
(8)
DG ) DH2O ×
√18 √M
(9)
where DO2 is the diffusion coefficient of O2 (M ) 32 g mol-1) in water ) 1.728 × 10-4 m2/d; M is the molecular weight of the contaminant of interest, and DH2O ) 2.22 m2 d-1. In using the predictive model to calculate the effective diffusivities of the selected contaminants in the experimental arrangement, the portion of air in the xylem tissues was set to 0, as the gas phase quickly displaces under saturated conditions. Calculating the effective diffusivities on the basis of the tissue properties of the poplar cores leads to calculated values quite similar to the estimated D values, Figure 5. Diffusion coefficients have been shown to be influenced by physical and chemical properties of the compound, such as molecular weight or molecular volume. Research by Bodalal et al. (24) and Little et al. (25) on the diffusion of VOCs through building materials (floor tiles, plywood, etc.) showed that diffusivity decreased as a function of partition coefficients and the molecular weight of the compound. A similar trend was noticed between effective diffusivities and the molecular weight (Figure 6A), as heavier, larger molecules pass more slowly through the wood matrix. However, a general decreasing trend was noted for the relationship for
tions above to include 10% gas phase in the tissues increases the calculated diffusivity to 21 × 10-7 cm2/s, which is essentially the same value that was estimated in the live, whole-plant studies. Live xylem contains varying amounts of the gas phase. For oak, 20% has been determined for live tissues (19). Calculation of Loss from Stem and Branches. The loss from the stem can be calculated, using the equations recently developed for a fruit tree model (26). The basic equation is CStem(z) ) CStem(0) × e-kz/uc
FIGURE 5. Ratios of calculated and measured Kwood and D for values determined experimentally herein and in an earlier work (20) and predicted Kwood and D from the previously published model (19).
(10)
where z is the height (m), k is the loss rate from the stem, and uc is the flow velocity of chemicals in the stem. Interestingly, this solution is identical to the solution developed independently in a previous work (17) except for the definition of flow velocity. In the fruit tree model (26), the flow velocity of the chemical (uC) is the flow velocity of water (uW) multiplied by the fraction of chemicals present in water, uC ) uW × fW. The loss rate via volatilization from the stem is kV )
A D × × F M ∆x
(11)
where A, M, and F are the surface area, mass, and density of the stem; or, an easier equation, because, for a cylinder of height h and radius r, volume V ) M/F ) π r2h and outside surface area A ) 2πrh, would be: kV )
FIGURE 6. Plots showing average diffusivities vs molecular weight (A) and partition coefficients Kwood (B) of compounds investigated. Error bars indicate 95% CI for diffusivities. diffusivity and the partition coefficient, but no clear relationships could be drawn (Figure 6B). Effective diffusivities for TCE compare favorably with the range of values (0.1-25 × 10-7 cm2/s) from earlier studies (19, 20). Earlier whole-plant research (17) on the same plant species (P. deltoides x nigra, clone DN34) showed higher values (4.0–25.0 × 10-7 cm2/s) than current research (1.24 ( 0.54 × 10-7 cm2/s), since, in the former case, effective diffusivity was estimated by calibrating the obtained concentration profile in a live tree system to a theoretical model. The effective diffusivities previously determined in live tissues certainly were influenced by the dispersion and upward advective flow of water in the tree system, as well as the differentiation of the phloem and xylem tissues, as both were included in the transport processes measured. However, the gas-phase transport in the tissues is likely the largest deviation from the measured values here. Adjusting the model equa-
2D r∆x
(12)
where D is the diffusion coefficient of the chemical in wood, and ∆x is the diffusion path length. The removal half-time t1/2 is ln(2)/kV. The removal half-distance z1/2 (m), that is, the height where 50% of the chemicals have been removed, is then t1/2 × uc. The removal half-distances were calculated using KWood (for fW) and diffusivities calculated from the earlier model (19), once for a stem with a diameter of 20 cm and a thickness of the sapwood ring of 2 cm plus 1 cm of bark (2 cm diffusion length), and once for a branch with a 2 cm diameter and a 4 mm diffusion length. The water flow velocity was taken to 24 m/d. Results are shown in Table 3. The removal halfdistances for the stem range from 3.12 m for TeCA to 88.1 m for MTBE. For the branch, the distances to remove 50% of the compound by volatilization are much lower; that is, the half-distances range from 2.1 to 2.5 cm for PCE and TeCA, respectively, and is 59 cm for MTBE. This calculation shows that volatilization is a rapid process for the removal of volatile compounds from trees, with a substantial loss from the stem or trunk for highly volatile compounds, but much greater loss rates from small branches. The calculation also illustrates that volatilization from tall, large-diameter trees is probably much slower than loss from smaller, thinner trees. Therefore, experience gained from laboratory experiments, usually with small-sized trees, needs to be transferred with caution to mature trees at field sites. This may in part explain why in some cases researchers have not observed a notable decrease in concentration in tree cores with height when sampling the trunk (1, 10, 27). In these cases, the concentration profile may be in part due to decreasing diameter at a rate similar to that for contaminant mass lost. Nonuniformity of the contaminant around the tree may also explain different findings noted in these references, as Vroblesky et al. and Schumacher et al. have demonstrated (10, 28). If volatile compounds do pass the stem, the major fate process will finally be volatilization. This does not exclude the possibility that compounds are degraded in the root zone or during xylem transport upward VOL. 42, NO. 4, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1273
by endophytes. For the case of benzene, toluene, ethyl benzene, and xylene contaminants, endophytic degradation has been observed (29) and degradation in the rhizosphere is expected to be a major fate, if adequate oxygen levels can be maintained to sustain aerobic conditions needed for rapid biodegradation. In the absence of degradation processes, volatilization is a main fate-determining process for these VOCs. VOCs undoubtedly move through fresh wood tissues rapidly, and wood effective diffusivities were directly determined in xylem tissues, where the majority of contaminants reside in phytoremediation systems that utilize perennial tree species. Effective diffusivities were related to the molecular weight of the compound. Partitioning coefficients were linearly related to log Kow as expected. Current research can help to enhance understanding of the fate of VOCs in phytoremediation applications both in uptake and fate. As diffusion and partitioning form a fundamental fates of VOCs, knowledge of these parameters and in planta metabolic rates are necessary in estimating the contaminant residence and removal times by developing a full-scale tree model considering all possible fates. This is particularly of interest in efforts to evaluate the fate of contaminants in plants such as fruit trees that produce foods for human consumption. The diffusion rates in live trees that include considerable gas fractions will certainly be higher for VOCs and can explain the findings that VOCs are indeed not efficiently transported to fruits (1, 2, 30). The relationships shown can provide insight to the fate of other contaminants that were not evaluated herein, but caution should be exercised in making general assumptions of extrapolation to all compound-tissue interactions. This research also has considerable value in the use of trees as a biosensor in delineating contaminant plumes. As trees uptake contaminants from the subsurface, a signature of chemical composition in the subsurface is represented in the tissue and transpiration flow of the xylem tissues (3, 10, 31). Vegetation sampling can reduce the expensive site investigation costs and potentially the monitoring costs for phytoremediation sites. In order to gain acceptance of this approach, all fate and, particularly, loss rates from tissues must be better characterized. Residence times of compounds and the relative concentrations observed are functions of the affinity of the plant tissue sampled and the rate at which the compounds pass through and subsequently volatilize to the atmosphere. Through this research, a systematic approach can be taken in considering the diffusion flux to the plant-atmosphere interface where volatilization can occur for VOCs.
Acknowledgments We greatly acknowledge the support of the National Science Foundation (Grants No. 0320721 and 9984064), E.P.A. Midwest Hazardous Substance Research Center, and the Organization for Economic and Commercial Development (OECD), France. The authors would like to thank John Schumacher of the USGS, Drs. Glenn Morrison and David Grow of the Missouri University of Science and Technology; and Drs. Philipp Mayer and Ulrich Gosewinkel Karlson of the Danish National Environment Research Institute for their assistance and valued contributions to this work.
Literature Cited (1) Doucette, W. J.; Chard, J. K.; Fabrizius, H.; Crouch, C.; Petersen, M. R.; Carlsen, T. E.; Chard, B. K.; Gorder, K. Trichloroethylene Uptake into Fruits and Vegetables: Three-Year Field Monitoring Study. Environ. Sci. Technol. 2007, 41, 2505–2509. (2) Schnabel, W. E.; Dietz, A. C.; Burken, J. G.; Schnoor, J. L.; Alvarez, P. J. Uptake and Transformation of Trichloroethylene by Edible Garden Plants. Water Res. 1997, 31 (4), 816–824. 1274
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 4, 2008
(3) Struckhoff, G.; Burken, J. G.; Schumacher, J. Phytoremediation of Vadose Zone VOCs. Environ. Sci. Technol. 2005, 39 (6), 1563– 1568. (4) Burken, J. G.; Schnoor, J. L. Distribution and volatilization of organic contaminants following uptake by hybrid poplar trees. Int. J. Phytorem. 1999, 1 (2), 139–152. (5) Newman, L. A.; Strand, S. E.; Choe, N.; Duffy, J.; Ekuan, G.; Ruszaj, M.; Shurtleff, B. B.; Wilmoth, J.; Heilman, P. E.; Gordon, M. P. Uptake and Biotransformation of Trichloroethylene by Hybrid Poplars. Environ. Sci. Technol. 1997, 31 (4), 1062–1067. (6) Collins, C.; Fryer, M.; Grosso, A. Plant uptake of non-ionic organic chemicals. Environ. Sci. Technol. 2006, 40 (1), 45–52. (7) Vroblesky, D. A.; Neitch, C. T.; Morris, J. T. Chlorinated Ethenes from Groundwater in Tree Trunks. Environ. Sci. Technol. 1999, 33 (1), 510–515. (8) Burken, J. G.; Ma, X. Chlorinated Solvents Phytoremediation: Uptake and Diffusion. In Remediation of Chlorinated and Recalcitrant Compounds; Gavakar, A. R., Chen, A. S. C., Eds.; Battelle Press: Columbus, OH, 2002; pp 2B–24. (9) Ma, X.; Burken, J. G. Diffusion of TCE to the Atmosphere in Phytoremediation Applications. Environ. Sci. Technol. 2003, 37 (11), 2534–2539. (10) Vroblesky, D. A.; Clinton, B. D.; Vose, J. M.; Casey, C. C.; Harvey, G. J.; Bradley, P. M. Groundwater Chlorinated Ethenes In Tree Trunks: Case Studies, Influence Of Recharge, And Potential Degradation Mechanism. Ground Water Monit. Rem. 2004, 24 (3), 124–139. (11) Zhang, Q.; Davis, L.; Erickson, L. Transport of Methyl tert-Butyl Ether through Alfalfa Plants. Environ. Sci. Technol. 2001, 35, 725–731. (12) Ma, X.; Richter, A.; Burken, J. G.; Albers, S. Phytoremediation of MTBE with Hybrid Poplar Trees. Int. J. Phytorem. 2004, 6 (2), 157–167. (13) Winnike-McMillan, S. K.; Zhang, Q.; Davis, L. C.; Erickson, L. E.; Schnoor, J. L. Phytoremediation of Methyl Tertiary-Butyl Ether. In Phytoremediation: Transformation and Control of Contaminants; McCutcheon, S. C., Schnoor, J. L., Eds.; John Wiley and Sons, Inc.: New York, 2003; pp 805–828. (14) Davis, L. C.; Lupher, D.; Hu, J.; Erickson, L. E. In Transport of Trichloroethylene Through Living Plant Tissues, 14th Conference on Hazardous Waste Research, St. Louis, Mo, 1999. (15) Mackay, A. A.; Gschwend, P. M. Sorption of Monoaromatic Hydrocarbons to Wood. Environ. Sci. Technol. 2000, 34 (5), 839– 845. (16) Greenidge, K. N. H. Rates and patterns of moisture movement in trees. In The Physiology of Forest Trees; Thimann, K., Ed.; Ronald Press: New York, 1958. (17) Ma, X.; Burken, J. G. Modeling of TCE Diffusion to the Atmosphere and Distribution in Plant Stems. Environ. Sci. Technol. 2004, 38 (17), 4580–4586. (18) Boving, T. B.; Zhang, W. Removal of aqueous-phase polynuclear aromatic hydrocarbons using aspen wood fibers. Chemosphere 2004, 54 (7), 831–839. (19) Trapp, S.; Miglioranza, K.; Mosebaek, H. Sorption of Lipophilic Organic Compounds to Wood and Implications for Their Environmental Fate. Environ. Sci. Technol. 2001, 35, 1561–1566. (20) Ma, X.; Burken, J. G. VOCs Fate and Partitioning in Vegetation: Use of Tree Cores in Groundwater Analysis. Environ. Sci. Technol. 2002, 36 (21), 4663–4668. (21) Burr, H. K.; Stamm, A. J. Diffusion in wood. Phys. Colloid Chem. 1947, 51, 240–261. (22) Carslaw, H. S.; Jaeger, J. C. Conduction of Heat in Solids; Oxford University Press: Oxford, U.K., 1959. (23) Currie, J. A. Gaseous Diffusion in Porous Media Part1.-A Nonsteady State Method. Br. J. Appl. Phys. 1960, 11, 314–317. (24) Bodalal, A.; Zhang, J. S.; Plett, E. G. A method for measuring internal diffusion and equilibrium partition coefficients of VOCs for building materials. Build. Environ. 2000, 35, 101–110. (25) Little, J. C.; Hodgson, A. T.; Gadgil, J. A. Modeling Emissions of Volatile Organic Compounds from New Carpets. Atmos. Environ. 1994, 28 (2), 227–234. (26) Trapp, S. Fruit Tree model for uptake of organic compounds from soil and air. SAR QSAR Environ. Res. 2007, 18 (3–4), 367– 387. (27) Doucette, W. J.; Bugbee, B.; Hayhurst, S.; Pajak, C. J. S. G. Uptake, Metabolism, and Phytovolatilization of Trichloroethylene by Indigenous Vegetation: Impact of Precipitation. In Phytoremediation: Transformation and Control of Contaminants, McCutcheon, S. C.; Schnoor, J. L., Eds.; John Wiley and Sons, Inc.: New York, 2003; pp 561–588. (28) Schumacher, J. G.; Struckhoff, G. C.; Burken, J. G. Assessment of Subsurface Chloinated Solvent Contamination Using Tree
Cores at the Front Street Site and Former Dry Cleaning Facility at the Riverfront Superfund Site, New Haven MO, 1999–2003; Report 2004-5049; U.S. Geological Survey: Reston, VA, 2004; p 35. (29) Barac, T.; Taghavi, S.; Borremans, B.; Provoost, A.; Oeyen, L.; Colpaert, J.; Vangronsveld, J.; van der Lelie, D. Engineered endophytic bacteria improve phytoremediation of watersoluble, volatile, organic pollutants. Nat. Biotechnol. 2004, 22 (5), 583–588. (30) Chard, B. K.; Doucette, W. J.; Chard, J. K.; Bugbee, B.; Gorder, K. Trichloroethylene uptake by apple and peach trees and transfer to fruit. Environ. Sci. Technol. 2006, 40, 4788–4793.
(31) Gopalakrishnan, G.; Negri, M. C.; Minsker, B. S.; Werth, C. J. Monitoring subsurface contamination using tree branches. Ground Water Monit. Rem. 2007, 27 (1), 65–74. (32) Hu, J.; Davis, L. C.; Erickson, L. E. In Modeling contaminant transport in plants, 1998 Conference on Hazardous Waste Research, Snowbird, UT, 1998; Midwest HSRC: Snowbird, UT, 1998; pp 285–295. (33) Rippen, G. Handbuch Umweltchemikalien; Ecomed: Landsberg, Germany, 2007.
ES071552L
VOL. 42, NO. 4, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1275
