Environ. Sci. Technol. 2010, 44, 3085–3092
Effects of Pore-Scale Heterogeneity and Transverse Mixing on Bacterial Growth in Porous Media CHANGYONG ZHANG,† QINJUN KANG,‡ XING WANG,§ JULIE L. ZILLES,§ ¨ LLER,| AND ROLAND H. MU C H A R L E S J . W E R T H * ,§ Chemical and Materials Sciences Division, Fundamental and Computational Sciences Directorate, Pacific Northwest National Laboratory, Richland, Washington, Computational Earth Science Group (EES-16), Earth and Environmental Sciences Division, Los Alamos National Laboratory, Los Alamos, New Mexico, Department of Civil and Environmental Engineering, University of Illinois at UrbanasChampaign, 205 North Mathews Avenue, Urbana, Illinois 61801, and Department of Environmental Microbiology, Helmholtz Center for Environmental Research UFZ, D-04318 Leipzig, Germany
Received November 8, 2009. Revised manuscript received January 25, 2010. Accepted February 8, 2010.
Microbial degradation of contaminants in the subsurface requires the availability of nutrients; this is impacted by porous media heterogeneity and the degree of transverse mixing. Two types of microfluidic pore structures etched into silicon wafers (i.e., micromodels), (i) a homogeneous distribution of cylindrical posts and (ii) aggregates of large and small cylindricalposts,wereusedtoevaluatetheimpactofheterogeneity on growth of a pure culture (Delftia acidovorans) that degrades (R)-2-(2,4-dichlorophenoxy)propionate (R-2,4-DP). Following inoculation, dissolved O2 and R-2,4-DP were introduced as two parallel streams that mixed transverse to the direction of flow. In the homogeneous micromodel, biomass growth was uniform in pore bodies along the center mixing line, while in the aggregate micromodel, preferential growth occurred between aggregates and slower less dense growth occurred throughout aggregates along the center mixing line. The homogeneous micromodel had more rapid growth overall (2 times) and more R-2,4-DP degradation (9.5%) than the aggregate pore structure (5.7%). Simulation results from a pore-scale reactive transport model indicate mass transfer limitations within aggregates along the center mixing line decreased overall reaction; hence, slower biomass growth rates relative to the homogeneous micromodel are expected. Results from this study contribute to a better understanding of the coupling between mass transfer, reaction rates, and biomass growth in complex porous media and suggest successful implementation and analysis of bioremediation systems requires knowledge of subsurface heterogeneity.
* Corresponding author phone: 217-333-3822; e-mail: werth@ illinois.edu. † Pacific Northwest National Laboratory. ‡ Los Alamos National Laboratory. § University of Illinois at UrbanasChampaign. | Helmholtz Center for Environmental Research UFZ. 10.1021/es903396h
2010 American Chemical Society
Published on Web 03/01/2010
Introduction Bioremediation is an established technology that relies on the microbial degradation of subsurface contaminants to clean groundwater (1). Several studies indicate that substrates and/or nutrients injected into groundwater during bioremediation are often consumed in the source zone and that contaminant biodegradation is controlled by transverse mixing along the plume margins (2–4). Transverse mixing can also limit natural attenuation when substrates and/or nutrients from the surrounding aquifer are depleted in the source zone (5). A number of studies have considered the impact of pore-scale heterogeneity on transverse mixing rates and plume attenuation, and several have evaluated biomass morphology in uniform pore structures and the associated reaction rates. However, none have examined the impact of pore-scale heterogeneity on biomass growth and reaction rates, the focus of this work. Several studies have evaluated the impacts of pore-scale heterogeneity on transverse mixing and transverse mixinglimited chemical reactions in both periodic arrays and random distributions of cylinders or ellipses in the absence of biological growth (6–9). Greater contact time between fluids with reacting solutes and flow focusing of streamlines, either in pore throat constrictions or in high-porosity regions surrounded by low-porosity regions, were both found to enhance transverse mixing and transverse mixing-limited reactions. Similar results were found at the continuum scale (2, 10, 11). The impact of high- and low-permeability inclusions embedded in a contrasting permeability matrix were evaluated, and flow focusing in high-permeability inclusions was found to enhance transverse mixing and transverse mixing-limited reactions. Analytical solutions were developed to quantify this effect as a function of relative flow rates, solute concentrations, and diffusion coefficients (11). Other studies have evaluated biomass growth and biodegradation at the pore scale using micromodels (12–19) and mathematical modeling (20–22). In micromodels, a variety of biomass growth patterns have been observed, including biomass webs (12), clumps of growth in pores and attached to pore walls (13), thin films on grains (14), and fingers of growth that extend into pore bodies in low-flow regions (17). In most cases, nutrients were coinjected and biomass growth was distributed throughout the pore network. In one such case, Vayenas et al. (23) found that increased substrate loading rates and flow velocities reduced the efficiency of biodegradation. Biofilm thickness also decreased with increased distance from the inlet due to a reduction in nutrient levels. However, in two studies the electron donor and acceptor were injected separately (3, 17), such that biomass growth was restricted to transverse mixing zones and transverse mixing controlled the growth and reaction rates. Nambi et al. (17) had a higher level of resolution in imaging biomass and observed that the biomass grew as fingers in pore bodies, where flow was relatively slow, but not in pore throats. The transverse width of this growth was restricted to a few pore bodies. Knutson et al. (20) developed a pore-scale model to simulate this biomass growth based on the solute degradation rate (which controls solute concentrations) and shear stress. We are not aware of any prior studies examining the effect of pore-scale heterogeneity on biomass growth and biodegradation. The objective of this study is to investigate the impacts of pore-scale heterogeneity and transverse mixing on the growth, distribution, and activity of a herbicide-degrading VOL. 44, NO. 8, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3085
bacterium. Two different model pore structures were created in etched silicon substrates (i.e., micromodels): a homogeneous pore structure consisting of a uniform array of cylindrical posts and an aggregate pore structure with clusters of large and small cylindrical posts separated by larger interaggregate pore spaces. The aggregate pore structure is used to represent heterogeneous soil structures reported to occur in the field (24, 25). Each micromodel was first inoculated with Delftia acidovorans strain R39bR and then injected with the herbicide (R)-2-(2,4-dichlorophenoxy)propionate (R-2,4-DP) from one inlet and dissolved oxygen from another inlet. The two reactants flowed parallel to each other in the micromodel pore structures, and strain R39bR grew along the transverse mixing line between R-2,4-DP and oxygen. A pore-scale flow and transport model was used to quantitatively evaluate water flow, transport of aqueous species, and the kinetic reaction rate in identical pore structures. Biomass growth rates, distribution, and activity were interpreted from images of biomass in the pore structures, effluent R-2,4-DP concentrations, and pore-scale model results of reaction rate distribution.
Materials and Methods Bacterial Strains and Growth Conditions. Delftia acidovorans (formerly Comamonas (26)) strain MC1 is a Gramnegative, motile aerobic bacterium capable of degrading phenoxypropionate herbicides (27) in the presence of oxygen. A derivative of MC1, strain R39bR, that is unable to utilize (S)-2-(2,4-dichlorophenoxy) propionate but maintains the ability to utilize (R)-2-(2,4-dichlorophenoxy) propionate (R2,4-DP) was used in this work (Mu ¨ ller, R. H. Unpublished results). Active cultures used to inoculate the micromodels were prepared as follows: several colonies were picked from an agar plate to inoculate 10 mL of LB media (10 g/L tryptone, 10 g/L NaCl, and 5 g/L yeast extract, pH 7.0) amended with 100 µM R-2,4-DP (99.9%, Sigma) in 100 mM of NaOH, followed by incubation at 30 °C with aeration for 12-16 h. Next, the biomass was centrifuged and resuspended in 25 mL of mineral salts solution (MSS) (28) amended with 100 µM R-2,4-DP as the sole carbon source and 2 mM cycloheximide (99.9%, Aldrich). Cycloheximide is a eukaryotic inhibitor that was added to prevent fungal contamination; control experiments showed that it had no effect on degradation of R-2,4-DP by R39bR (data not shown). After the initial R-2,4-DP was completely degraded (3-5 h), higher concentrations of R-2,4-DP (sequentially increased from 0.5 to 1.2 mM) were added to batch cultures every ∼12 h over a 3 day period to obtain more biomass. Each round of feeding and degradation was confirmed by chemical analysis, and immediately prior to inoculation the culture was again centrifuged and resuspended in MSS to remove the R-2,4DP and cycloheximide. All media and amendments were sterilized prior to use, either by autoclaving or by filter sterilization. Chemical Analysis. The R-2,4-DP concentration was analyzed by HPLC (Shimadzu LC-20AT) using a silica-based reversed-phase C18 column (3 µm, 120 Å, 4.6 × 100 mm) with 50% methanol and 50% 0.2 mM sodium phosphate buffer (pH ≈ 2.5) as the mobile phase and a flow rate of 1 mL/min; the sample injection volume was 20 µL. The R-2,4DP peak was detected at the UV wavelength of 205 nm and had a retention time of 6.4 min. Micromodels. Two types of micromodel pore network structures were used in this study: a homogeneous micromodel (HC) containing a uniform distribution of cylindrical posts 300 µm in diameter (Figure 1a) and an aggregate micromodel (AG) containing aggregates of large (300 µm in diameter) and small (100 µm in diameter) cylindrical posts (Figure 1b) separated by larger interstitial pore spaces. For 3086
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 8, 2010
the homogeneous pore network, each pore space is 180 µm and each pore throat is 40 µm; for the aggregate pore network, the pore space between two aggregate units is 240 µm and each pore throat is 82 µm. The average porosities are 0.39 (homogeneous) and 0.34 (aggregates). The micromodels have two inlet channels (A, B) and a 2 cm (length) × 1 cm (width) pore network; they were etched to a depth of 15-22 µm. A third inlet (C) is located after the pore network for biocide injection, followed by the outlet (D). Detailed dimensions of both micromodels are summarized in Table S1 in the Supporting Information. Standard photolithography techniques were used to fabricate micromodels; details of the fabrication process can be found in Chomsurin and Werth (29) and Willingham et al. (9). The finished micromodels were each mounted in an aluminum manifold through which polymer tubing (PEEK, which was selected because of low gas permeability) was connected through threaded fittings. Three-way PTFE valves were installed before each micromodel inlet and connected to syringes containing substrates. Flow through the micromodel was maintained using a dual-syringe pump. Micromodel Experiments. Prior to inoculation, all assembled micromodel systems were flushed with 10 mL of H2O2 solution (10%) for sterilization and rinsed with 10 mL of autoclaved and degassed nanopure water. Inoculation was carried out by injecting an active R39bR culture through inlet A and a MSS solution containing ∼85 µM R-2,4-DP and dissolved oxygen (DO) through inlet B. Solutions with DO were prepared by bubbling air through a 0.2 µm filter at 35 °C; maintaining DO below saturation during the experiments at 25 °C prevented the formation of air bubbles. The resulting influent DO was approximately 5 mg/L. Inoculation lasted 5-8 days, during which the inoculum syringe was refilled once with freshly prepared liquid culture. Very limited biomass in the form of fingers and clumps were randomly distributed in the micromodel pore space during inoculation, accounting for at most 4-9% of the final biomass. After inoculation, the electron donor and acceptor were supplied through separate inlets. Specifically, R-2,4-DP (∼85 µM) dissolved in degassed MSS was supplied through inlet A, and MSS with DO was supplied through inlet B. Both syringes were operated at a volumetric flow rate of 50-60 µL/h (actual flow rate depends on depth of micromodel), which corresponds to a Darcy flux of 1.02 cm/min in both micromodels. These flow rates were found to be optimal for supplying sufficient nutrient mass flux into the micromodels without seriously impacting biomass attachment due to hydrodynamic shear. The corresponding Darcy flux is also in the range of those found in sandy groundwater aquifers during active bioremediation (30). After inoculation, inline filters with an average pore size of 2 µm were installed before each micromodel inlet to reduce the chemotaxis or movement of the bacteria back toward the substrates in the syringes. A schematic of the micromodel flow set up is included in Supporting Information Figure S1. The outlet tubing was also submerged in a liquid solution containing 5 mM sodium azide (NaN3) to avoid contamination of the micromodel through the outlet tubing. To allow analysis of the degradation occurring in the micromodel, biomass growth in the outlet and outlet tubing was inhibited via injection of 10 mM NaN3 solution through inlet C for 3 days prior to collection of effluent samples for HPLC analysis. The flow rate at inlet C was 50 µL/h, one-half that of inlets A and B combined; backward flow of NaN3 was minimal and did not reach the pore network, as confirmed by a separate test using a fluorescent tracer (data not shown). Effluent samples were collected beginning when growth was observed (20-25 days for the homogeneous micromodel and 30-35 days for the aggregate micromodel). Two replicate experiments were performed with both micromodels.
FIGURE 1. Micromodel pore structures: (a) homogeneous, (b) aggregates, and (c) flow set up.
FIGURE 2. Biomass distribution in (a) homogeneous (34 day) and (b) aggregate (70 day) micromodels. Flow is from left to right, with R-2,4-DP supplied from the top inlet and DO from the bottom inlet.
Micromodel Visualization. All micromodel images were acquired with a Nikon Epiphot 200 epi-fluorescent microscope using a 5× inverted objective and reflected differential interference contrast (DIC) microscopy. The microscope is equipped with an automated stage (Prior Scientific Instrument) and a monochrome digital charge-coupled device (CCD) camera (RT Spot, Diagnostic Inc.), both connected to a computer and controlled by Metamorph imaging software (Molecular Devices). A total of 70 (10 × 7) separate images were taken at selected time points for each experiment; an
additional image from outside the pore network area was also acquired, and image correction was systematically applied to correct for nonuniform illumination prior to montaging the 70 separate images to form a single image which captured the entire 2 cm × 1 cm pore network. Image Analysis. Montaged images of micromodels were analyzed to quantitatively evaluate biomass growth. The image intensity of the biomass is lower than that for silicon posts and pore spaces; a threshold value was determined for each image to distinguish biomass, and the area (i.e., number VOL. 44, NO. 8, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3087
FIGURE 3. Pore-scale visualization of biomass distribution in homogeneous micromodel after 21 (a) and 34 days (b), and aggregates micromodel after 36 (c) and 70 days (d).
of pixels) occupied by biomass was determined. The optical image is two dimensional; hence, the biomass below the threshold value was assumed to occupy a consistent depth of the micromodel channel, and the total amount of biomass is reported as number of pixels. This approach results in an approximate indicator of the amount of biomass, as the biomass density and biomass growth with depth is probably not uniform. Model Simulations. A lattice-Boltzmann model (LBM) was used to simulate water flow, mass transport of substrates, and the kinetic reaction between these substrates in the pore networks. The LBM for incompressible flow was used to solve for interstitial water velocities (31). The LBM for multicomponent reactive transport was used (after modification to account for kinetic reactions) to solve for reactant and product concentrations (32, 33). Details of the mathematical model and its LBM implementation for the transport of aqueous species with kinetic reactions in the bulk solution are described in Kang et al. (34). A numerical grid spacing of 10 µm was used, and two-dimensional model pore structures of 0.384 cm × 0.192 cm and 0.416 cm × 0.208 cm were used for homogeneous and aggregate systems, respectively. Aqueous diffusion coefficients of O2 and R-2,4-DP were determined using the Wilke-Chang equation (35). For strain R39bR, the stoichiometric ratio between oxygen and R-2,4DP is approximately 6:1 (36, 37); the following simplified irreversible reaction was therefore considered 6A + B ) C The reaction between O2 and R-2,4-DP is complicated and involves a series of enzymatic reactions. We assumed for simplicity second-order nonlinear reaction kinetics to calculate the reaction rate at each numerical grid r ) kCACB where CA and CB represent the concentrations of O2 and R-2,4DP, respectively, and k is the reaction rate constant; all other trace nutrients are assumed to be in excess. Since k is constant for each simulation and biomass growth is not considered, an implicit assumption is that biomass does not affect the velocity field or the transport/reaction of substrates. Hence, 3088
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 8, 2010
the model is used to evaluate the effects of mass transport (i.e., advection and diffusion) and reaction rate constant (k) on the reaction rate distribution (i.e., kCACB), and this is used to qualitatively interpret biomass growth observed in the experiments. There are two major dimensionless parameters that characterize the advection-diffusion-reaction process. One is the Peclet number, defined as Pe ) UL/D, which indicates the relative strength between advection and diffusion; the other is Damkohler number, defined as Da ) kCL2/D, which denotes the ratio between reaction and diffusion. In the above definitions, U, L, C, and D are characteristic velocity, length, concentration, and diffusion coefficient, respectively. Here the Darcy velocity, average pore space of the aggregate model, inlet concentration of O2, and diffusion coefficient of O2 are chosen for these characteristic quantities. On the basis of the experimental condition, Pe is calculated as 15.5, which is used to determine parameter values in the LBM simulation. The value of the reaction rate constant k is varied to investigate the effect of Da number.
Results and Discussion Micromodel Experiments of Biomass Growth and Distribution. Images of established biomass in the entire pore networks of both homogeneous and aggregate micromodels are shown in Figure 2, and representative images at the pore scale are shown in Figure 3. In the homogeneous micromodel, a continuous line of biomass (appearing as dark regions in Figure 2a) is observed in the center of the transverse mixing zone along the entire length of the pore structure after 34 days. In individual pores (Figure 3a,b), biomass growth spans entire pore bodies along the transverse mixing zone. Water velocity and shear are lowest in pore bodies (not pore throats), where biomass growth is primarily located; observations of biomass growth in the same locations were made using a different microorganism by Nambi et al. (17). In the aggregate micromodel, biomass growth appears to be discontinuous along the transverse mixing zone (Figure 2b). Biomass is darker, indicating it is denser, in the interaggregate pore spaces than in intra-aggregate pore spaces. Inside aggregates, biomass is distributed over a wider
FIGURE 5. Fraction of R-2,4-DP degradation in the effluent (error bars represent standard deviation among 3-5 effluent samples collected).
FIGURE 4. Biomass growth rate in (a) the entire micromodel and (b) selected pore spaces. Two replicates are shown for each pore structure in a. region in the transverse direction but with lower density (Figure 3c,d). Biomass Growth Kinetics in Micromodels. The total number of pixels containing biomass is normalized by the value at steady state in both micromodels and plotted vs time in Figure 4; this is a surrogate for the biomass growth rate. Two phases of biomass growth are observed in both micromodel pore structures: an initial slow growth phase (i.e., the first 15 days for the homogeneous micromodel and the first 20 days for the aggregate micromodel) followed by a second, faster growth phase (between 15 and 35 and 20 and 60 days for homogeneous and aggregate micromodels, respectively); this indicates the existence of an initial lag phase in these hydrodynamic porous media systems. During the second growth phase, the growth rate in the homogeneous micromodel is approximately 2 times greater than in the aggregate micromodel. On the basis of growth patterns, we attribute this to the limited growth that occurs in intraaggregate pore spaces compared to the more uniform and dense growth that occurs in all pores along the mixing zone in the homogeneous micromodel. Biomass growth rates in selected pore spaces in both micromodels are compared in Figure 4b. Growth rates in individual pore spaces in the homogeneous micromodel matched the overall growth rate, while in the aggregate micromodel, the initial growth was dominated by biomass in the interaggregate pore spaces while growth in intra-aggregate pore space was slower. Effluent Concentration. Approximately 9.5% of the R-2,4DP was degraded in the homogeneous micromodel and approximately 5.7% in the aggregate micromodel (Figure 5). Without NaN3 injection in inlet C, greater than 97% of R-2,4DP was degraded in the effluent samples, confirming that substantial activity occurs in the flow channel after the pore network in the absence of biocide injection. Willingham et al. (9) observed that the amount of product formed from two
reactants that mix transverse to flow in the homogeneous and aggregate pore structures used in this study is not significantly different when instantaneous reaction (i.e., high Da) and identical flow rates were considered. This suggests that the greater amount of R-2,4-DP degraded in the homogeneous micromodel is due to the greater amount of biomass in this pore network compared to the aggregate micromodel at the time of effluent measurement. Flow Velocity Simulation. LBM results for water flow in the homogeneous pore structure (Figure S2 in the Supporting Information) indicate that the interstitial velocity through pore throat regions in the absence of biomass is approximately 2 times greater than in pore bodies. They also indicate that the interaggregate pore water velocity is approximately 40% higher than in the homogeneous pore structure and approximately 2 orders of magnitude higher than in intra-aggregate pore bodies, where the fluid is almost stagnant near the center (i.e., diffusion controlled). The higher velocities in pore throats of the homogeneous micromodel and in interaggregate pore spaces of the aggregate micromodel result in relatively high shear stresses. The impacts on biomass growth are discussed in the next section. Reactive Transport Simulation. The simulated reaction rate distributions (i.e., kCACB) in the homogeneous and aggregate pore structures for three different Da values are shown in Figure 6 (O2 concentration distributions are included in the Supporting Information, Figure S3). It is clear that for both pore structures, the reaction rate is primarily distributed in a narrow transverse zone one to several pore bodies in width along the center of the domain. The larger the Da, the narrower the reaction zone in the transverse direction. In the homogeneous pore structure at low Da (0.139), the highest reaction rate is along the centerline of pore bodies in the transverse mixing zone, with smaller reaction rates extending one to two pore bodies above and below the centerline. These results are consistent with those in the homogeneous micromodel experiment, in which most biomass is present in the middle of pore bodies along the centerline and less biomass is present in pore bodies one to two pore widths above and below the centerline. However, the simulated reaction rate is similar in pore throats and pore bodies, while biomass growth is primarily in pore bodies, not pore throats. This suggests that biomass grows preferentially in pore bodies where the flow rate is slower and there is less shear force. These results are consistent with those from Knutson et al. (20), who determined that in a homogeneous pore structure biomass grows preferentially in pore bodies but not pore throats due to lower shear in the former. In the aggregate pore structure at low Da (0.139), the reaction is primarily confined to a single-pore body width between aggregates (i.e., interaggregate) but is distributed VOL. 44, NO. 8, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3089
FIGURE 6. Reaction rate distribution (color bar indicates reaction rate in mol/m3/s) in homogeneous (left) and aggregate (right) pore networks at different reaction rate constants (Da) simulated using LBM. Da ) 0.139, 1.39, and 13.9 from top to bottom. more widely within aggregates (i.e., intra-aggregate). This is illustrated more clearly by overlaying transverse reaction rate profiles between aggregates and within an aggregate in Figure 7. This result is consistent with biomass growth in the aggregate micromodel experiment, where biomass is narrower transverse to the direction of flow between aggregates than within aggregates, and with slower growth rates within aggregates than between aggregates. However, simulation results also indicate that the reaction rate is higher along the centerline than off center within aggregates. This is not consistent with biomass growth results that show growth is relatively uniform within aggregates; this may be due to biomass growth that has not achieved steady state. Simulation results from both homogeneous and aggregate pore networks indicate that the reaction rate is lower than the rate of transverse mixing (i.e., Da ) 0.139) and that this results in greater transport of reactants across the centerline and wider reaction and biomass growth zones transverse to the direction of flow. In the aggregate pore network, the biomass growth zone is more narrowly distributed between than within aggregates because between aggregates advection dominated transport results in steeper transverse concentration gradients and higher overall reaction rates (kCACB), and within aggregates diffusive transport results in more diffuse transverse concentration gradients, lower overall reaction rates, and more transverse mixing of reactants. Implications. In this study, the impacts of pore-scale heterogeneity on transverse mixing and biomass growth and distribution were quantitatively evaluated. Biomass grew uniformly in pore bodies along the center mixing zone in the 3090
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 8, 2010
homogeneous micromodel and primarily in the interaggregate pore space in the aggregate pore structure, where the reaction rate (kCACB) is higher. Due to reduced growth that occurred in the intra-aggregate pore spaces, the overall degradation rate in the aggregate micromodel was lower than in the homogeneous micromodel. While these results were all obtained with D. acidovorans, similar results with another microorganism in homogeneous micromodels (17) and good agreement with simulated substrate distributions suggest that they may be more generally applicable. Flow in threedimensional subsurface porous media systems is more complicated due to more complex physical heterogeneity, which may result in different patterns of high and low flow with respect to pore bodies and throats, of substrate distribution in transverse mixing zones, and hence of bacterial growth and distribution. During active bioremediation of contaminated aquifers, degradation of contaminants often requires injection of nutrient substrates through pumping wells and the effectiveness of such strategy can be limited by the subsurface heterogeneity. Characterization of subsurface heterogeneity and upscaled modeling approaches are essential to designing bioremediation systems and understanding their performance.
Acknowledgments We thank Robert Sanford from the University of Illinois and Peter Lichtner from Los Alamos National Laboratory for helpful discussions; we also thank Albert Valocchi from the University of Illinois and Thomas Willingham from Exxon Mobil Upstream Research Company for help with pore-scale
FIGURE 7. Transverse profile of reaction rate in the aggregate pore network for Da ) 0.139 (a), 1.39 (b), and 13.9 (c). modeling. This project was supported by the National Research Initiative Grant no. 2007-35107-17817 from the USDA National Institute of Food and Agriculture. Partial financial support for C.Y.Z was provided by the Environmental Molecular Sciences Laboratory (EMSL), a national scientific user facility sponsored by the DOE, Office of Biological and Environmental Research and located at PNNL.
Supporting Information Available Summary of micromodel pore network dimensions, schematic of micromodel flow set up, LBM simulation of longitudinal velocity profiles and dissolved O2 concentration distribution. This material is available free of charge via the Internet at http://pubs.acs.org.
Literature Cited (1) Loffler, F. E.; Edwards, E. A. Harnessing microbial activities for environmental cleanup. Curr. Opin. Biotechnol. 2006, 17 (3), 274–284. (2) Cirpka, O. A.; Frind, E. O.; Helmig, R. Numerical simulation of biodegradation controlled by transverse mixing. J. Contam. Hydrol. 1999, 40 (2), 159–182.
(3) Thullner, M.; Mauclaire, L.; Schroth, M. H.; Kinzelbach, W.; Zeyer, J. Interaction between water flow and spatial distribution of microbial growth in a two-dimensional flow field in saturated porous media. J. Contam. Hydrol. 2002, 58 (3-4), 169–189. (4) Bauer, R. D.; Rolle, M.; Bauer, S.; Eberhardt, C.; Grathwohl, P.; Kolditz, O.; Meckenstock, R. U.; Griebler, C. Enhanced biodegradation by hydraulic heterogeneities in petroleum hydrocarbon plumes. J. Contam. Hydrol. 2009, 105 (1-2), 56–68. (5) Thornton, S. F.; Quigley, S.; Spence, M. J.; Banwart, S. A.; Bottrell, S.; Lerner, D. N. Processes controlling the distribution and natural attenuation of dissolved phenolic compounds in a deep sandstone aquifer. J. Contam. Hydrol. 2001, 53 (3-4), 233–267. (6) Edwards, D. A.; Shapiro, M.; Brenner, H.; Shapira, M. Dispersion of Inert Solutes in Spatially Periodic, 2-Dimensional Model Porous-Media. Transp. Porous Med. 1991, 6 (4), 337–358. (7) Cao, J.; Kitanidis, P. K. Pore-scale dilution of conservative solutes: An example. Water Resour. Res. 1998, 34 (8), 1941–1949. (8) Acharya, R. C.; Valocchi, A. J.; Werth, C. J.; Willingham, T. W. Pore-scale simulation of dispersion and reaction along a transverse mixing zone in two-dimensional porous media. Water Resour. Res. 2007, 43 (10), W10435, doi: 10.1029/2007WR005969. (9) Willingham, T. W.; Werth, C. J.; Valocchi, A. J. Evaluation of the effects of porous media structure on mixing-controlled reactions using pore-scale modeling and micromodel experiments. Environ. Sci. Technol. 2008, 42 (9), 3185–3193. (10) Cirpka, O. A.; Kitanidis, P. K. Characterization of mixing and dilution in heterogeneous aquifers by means of local temporal moments. Water Resour. Res. 2000, 36 (5), 1221–1236. (11) Werth, C. J.; Cirpka, O. A.; Grathwohl, P. Enhanced mixing and reaction through flow focusing in heterogeneous porous media. Water Resour. Res. 2006, 42 (12), W12414, doi: 10.1029/ 2005WR004511. (12) Paulsen, J. E.; Oppen, E.; Bakke, R. Biofilm morphology in porous media, a study with microscopic and image techniques. Water Sci. Technol. 1997, 36 (1), 1–9. (13) Dupin, H. J.; McCarty, P. L. Mesoscale and microscale observations of biological growth in a silicon pore imaging element. Environ. Sci. Technol. 1999, 33 (8), 1230–1236. (14) Kim, D. S.; Fogler, H. S. Biomass evolution in porous media and its effects on permeability under starvation conditions. Biotechnol. Bioeng. 2000, 69 (1), 47–56. (15) Stewart, T. L.; Fogler, H. S. Biomass plug development and propagation in porous media. Biotechnol. Bioeng. 2001, 72 (3), 353–363. (16) Stewart, T. L.; Fogler, H. S. Pore-scale investigation of biomass plug development and propagation in porous media. Biotechnol. Bioeng. 2002, 77 (5), 577–588. (17) Nambi, I. M.; Werth, C. J.; Sanford, R. A.; Valocchi, A. J. Porescale analysis of anaerobic halorespiring bacterial growth along the transverse mixing zone of an etched silicon pore network. Environ. Sci. Technol. 2003, 37 (24), 5617–5624. (18) Lanning, L. M.; Ford, R. M.; Long, T. Bacterial chemotaxis transverse to axial flow in a microfluidic channel. Biotechnol. Bioeng. 2008, 100 (4), 653–663. (19) Long, T.; Ford, R. M. Enhanced Transverse Migration of Bacteria by Chemotaxis in a Porous T-Sensor. Environ. Sci. Technol. 2009, 43 (5), 1546–1552. (20) Knutson, C. E.; Werth, C. J.; Valocchi, A. J. Pore-scale simulation of biomass growth along the transverse mixing zone of a model two-dimensional porous medium. Water Resour. Res. 2005, 41 (7), W07007, doi: 10.1029/2004WR003459. (21) Knutson, C.; Valocchi, A.; Werth, C. Comparison of continuum and pore-scale models of nutrient biodegradation under transverse mixing conditions. Adv. Water Resour. 2007, 30 (67), 1421–1431. (22) Olson, M. S.; Ford, R. M.; Smith, J. A.; Fernandez, E. J. Mathematical Modeling of Chemotactic Bacterial Transport through a Two-Dimensional Heterogeneous Porous Medium. Biorem. J. 2006, 10 (1), 13–23. (23) Vayenas, D. V.; Michalopoulou, E.; Constantinides, G. N.; Pavlou, S.; Payatakes, A. C. Visualization experiments of biodegradation in porous media and calculation of the biodegradation rate. Adv. Water Resour. 2002, 25 (2), 203–219. (24) Rao, P. S. C.; Rolston, D. E.; Jessup, R. E.; Davidson, J. M. Solute transport in aggregated porous-media - theoretical and experimental evaluation. Soil Sci. Soc. Am. J. 1980, 44 (6), 1139– 1146. (25) Ghezzehei, T. A.; Or, D. Pore-space dynamics in a soil aggregate bed under a static external load. Soil Sci. Soc. Am. J. 2003, 67 (1), 12–19. (26) Wen, A. M.; Fegan, M.; Hayward, C.; Chakraborty, S.; Sly, L. I. Phylogenetic relationships among members of the ComaVOL. 44, NO. 8, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3091
(27)
(28)
(29) (30)
(31)
3092
monadaceae, and description of Delftia acidovorans (den Dooren de Jong 1926 and Tamaoka et al. 1987) gen. nov., comb. nov. Int. J. Syst. Bacteriol. 1999, 49, 567–576. Muller, R. H.; Jorks, S.; Kleinsteuber, S.; Babel, W. Comamonas acidovorans strain MC1: a new isolate capable of degrading the chiral herbicides dichlorprop and mecoprop and the herbicides 2,4-D and MCPA. Microbiol. Res. 1999, 154 (3), 241–246. Muller, R. H.; Hoffmann, D. Uptake kinetics of 2,4-dichlorophenoxyacetate by Delftia acidovorans MC1 and derivative strains: Complex characteristics in response to pH and growth substrate. Biosci. Biotechnol. Biochem. 2006, 70 (7), 16421654. Chomsurin, C.; Werth, C. J. Analysis of pore-scale nonaqueous phase liquid dissolution in etched silicon pore networks. Water Resour. Res. 2003, 39 (9), 1265, doi: 10.1029/2002WR001643. Semprini, L.; Hopkins, G. D.; McCarty, P. L.; Roberts, P. V. Insitu Transformation of Carbon-Tetrachloride and Other Halogenated Compounds Resulting from Biostimulation under Anoxic Conditions. Environ. Sci. Technol. 1992, 26 (12), 2454–2461. Guo, Z. L.; Shi, B. C.; Wang, N. C. Lattice BGK model for incompressible Navier-Stokes equation. J. Comput. Phys. 2000, 165 (1), 288–306.
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 8, 2010
(32) Kang, Q. J.; Lichtner, P. C.; Zhang, D. X. Lattice Boltzmann porescale model for multicomponent reactive transport in porous media. J. Geophys. Res. 2006, 111 (B5), B05203, doi: 10.1029/ 2005JB003951. (33) Kang, Q. J.; Lichtner, P. C.; Zhang, D. X. An improved lattice Boltzmann model for multicomponent reactive transport in porous media at the pore scale. Water Resour. Res. 2007, 43 (12), W12S14, doi: 10.1029/2006WR005551. (34) Kang, Q. J.; Zhang, D. X.; Chen, S. Y.; He, X. Y. Lattice Boltzmann simulation of chemical dissolution in porous media. Phys. Rev. E 2002, 65 (3), 036318, doi: 10.1103/PhysRevE.65.036318. (35) Wilke, C. R.; Chang, P. Correlation of diffusion coefficients in dilute solutions. AIChE J. 1955, 1 (2), 264–270. (36) Muller, R. H.; Babel, W. Phenol and its derivatives as heterotrophic substrates for microbial-growth - an energetic comparison. Appl. Microbiol. Biotechnol. 1994, 42 (2-3), 446– 451. (37) Muller, R. H.; Babel, W. A theoretical study on the metabolic requirements resulting from alpha-ketoglutarate-dependent cleavage of phenoxyalkanoates. Appl. Environ. Microbiol. 2000, 66 (1), 339–344.
ES903396H
