Metabolism-Induced CaCO3 Biomineralization During Reactive

Article pubs.acs.org/est

Metabolism-Induced CaCO3 Biomineralization During Reactive Transport in a Micromodel: Implications for Porosity Alteration Rajveer Singh,*,†,‡ Hongkyu Yoon,§ Robert A. Sanford,‡,∥ Lynn Katz,⊥ Bruce W. Fouke,‡,∥ and Charles J. Werth⊥ †

Civil and Environmental Engineering, ‡Institute for Genomic Biology, and ∥Department of Geology, University of Illinois at Urbana−Champaign, Urbana, Illinois 61801, United States § Geoscience Research and Applications, Sandia National Laboratories, Albuquerque, New Mexico 87185, United States ⊥ Civil, Architectural and Environmental Engineering, University of Texas at Austin, Austin, Texas 78712, United States S Supporting Information *

ABSTRACT: The ability of Pseudomonas stutzeri strain DCP-Ps1 to drive CaCO3 biomineralization has been investigated in a microfluidic flowcell (i.e., micromodel) that simulates subsurface porous media. Results indicate that CaCO3 precipitation occurs during NO3− reduction with a maximum saturation index (SIcalcite) of ∼1.56, but not when NO3− was removed, inactive biomass remained, and pH and alkalinity were adjusted to SIcalcite ∼ 1.56. CaCO3 precipitation was promoted by metabolically active cultures of strain DCP-Ps1, which at similar values of SIcalcite, have a more negative surface charge than inactive strain DCP-Ps1. A two-stage NO3− reduction (NO3− → NO2− → N2) pore-scale reactive transport model was used to evaluate denitrification kinetics, which was observed in the micromodel as upper (NO3− reduction) and lower (NO2− reduction) horizontal zones of biomass growth with CaCO3 precipitation exclusively in the lower zone. Model results are consistent with two biomass growth regions and indicate that precipitation occurred in the lower zone because the largest increase in pH and alkalinity is associated with NO2− reduction. CaCO3 precipitates typically occupied the entire vertical depth of pores and impacted porosity, permeability, and flow. This study provides a framework for incorporating microbial activity in biogeochemistry models, which often base biomineralization only on SI (caused by biotic or abiotic reactions) and, thereby, underpredict the extent of this complex process. These results have wide-ranging implications for understanding reactive transport in relevance to groundwater remediation, CO2 sequestration, and enhanced oil recovery.

■

INTRODUCTION Biomineralization refers to the process of microbial-mediated mineral formation. A large number of studies have shown that it can significantly alter the porosity and permeability of porous media, and it has been implicated as an important process in many natural and engineered applications, including geological CO2 storage,1−4 enhanced oil recovery,2,5,6 reinforcement of soil in geotechnical engineering,7−9 and immobilization of heavy metals and radionuclides.10−14 Prior work has demonstrated that microorganisms can promote the precipitation of carbonate crystals (termed biomineralization) by catalyzing reactions that increase solution pH and alkalinity,1,2,15−17 and/ or providing mineral nucleation sites.18−21 There is also evidence that active microorganisms enhance mineralization through molecular and genetic controls.21−24 Biomineralization within subsurface porous media is further complicated by complex mineralogical grain assemblages, hydrodynamics, and diffusive mixing of microbial substrates in pore spaces, whose geometries evolve over time with geochemical reactions.25−28 A better understanding of the complex mechanisms that control the extent and location of biomineralization in the subsurface is © 2015 American Chemical Society

needed to develop more accurate predictive models for the aforementioned applications. Over the past decade, biologically induced carbonate mineralization has been extensively studied with a focus on metabolic activities that influence solution chemistry, such as pH and alkalinity.1,2,15−17,21,29,30 A change in solution chemistry can directly affect the saturation index (SI) with respect to CaCO3. For example, bacterial ureolysis has been used to promote biomineralization, where metabolic pathways produce NH4+ and OH− ions. This increases pH and alkalinity, which provides favorable conditions for CaCO3 precipitation.2,16,30 Denitrification has also been widely studied and causes the consumption of H+ and/or the production of bicarbonate, which increases SI values for CaCO3 precipitation.15,31 Both urea hydrolysis and denitrification processes can alter the saturation index favorably, which in the presence of Ca2+ can Received: Revised: Accepted: Published: 12094

January 9, 2015 September 1, 2015 September 8, 2015 September 8, 2015 DOI: 10.1021/acs.est.5b00152 Environ. Sci. Technol. 2015, 49, 12094−12104

Article

Environmental Science & Technology

by which this happens, and investigate the influence of transverse mixing limitations on the extent and location of CaCO3 precipitation. Results are directly compared to predictions from an existing pore scale reactive transport model51,57 that was modified to include denitrification. This comparison was then used to identify the mechanisms that control the extent and location of biomineralization in the micromodel. We note that transverse mixing-limited conditions in the micromodel experiment allow precise control of the spatial distribution of electron donor and acceptor concentrations in the pore network, which results in a detailed mechanistic interpretation of the results. Implications of CaCO3 biomineralization are then considered for a variety of important applications in the subsurface.

help induce biomineralization, and subsequently reduce porosity5,32 and strengthen soils.8,31 Bacterial surfaces (or cells) can also play a significant role in providing heterogeneous nucleation sites for precipitation and crystal growth by lowering the activation energy (i.e., supersaturation) required for biomineralization.29,33 Heterogeneous nucleation can occur on bacterial surfaces, where positively charged ions (e.g., Ca2+, Mg2+) bind to negatively charged surface groups.19,21,34 Other bacterial components that can promote precipitation are extracellular polymeric substances (EPS), which refer to a large variety of organic polymers secreted by microorganisms.35−38 There is also evidence of more direct microbial involvement in promoting biomineralization. Previous authors have attributed this to both active Ca2+ metabolism39,40 and bacterial genetic control.22,24 For the former, Ca2+ is actively transported out of cells by an ATP-dependent pump to prevent accumulation within cells by passive diffusion. This is coupled to H+ uptake, which results in high pH and Ca2+ concentrations immediately adjacent to the cell,39 conditions that favor precipitation. For the latter (i.e., bacterial genetic control), specific microbial genes are associated with regulation of the proton motive force in bacteria, which affects the biomineralization process via control of H+ excretion.24 A number of studies have investigated the role of fluid hydrodynamics and diffusive mixing of reactive substrates in pore spaces on (bio)mineralization and feedback associated with pore blockage.26,27,31,41−45 Other studies have considered these effects on biomass growth without mineralization46−50 and on abiotic carbonate mineralization.51,52 Active biomineralization-stimulated near column inlet zones resulted in significant permeability alteration in lab scale experiments.31,41,43 Purely mixing-induced CaCO3 precipitation under reactive transport conditions has been demonstrated in quasi-two-dimensional laboratory flowcells,26,27 where two solutes (e.g., Ca2+ and CO32−) react via transverse mixing. In these studies, a narrow CaCO3 precipitation zone was observed that caused pore clogging in the transverse mixing zone and subsequently hindered transverse diffusion of reactants. Similar observations were made in abiotic CaCO3 precipitation experiments in micromodels51,52 In this effort, the role of transverse mixing-limited nitrate reduction on biologically induced carbonate precipitation is evaluated. This has not been examined in the laboratory, but its importance has been implicated in a number of applications. For example, nitrate is the most common groundwater contaminant in the world;53,54 bioremediation can involve injecting an aqueous phase electron donor through a well into a polluted aquifer to stimulate denitrification.54 Similarly, an aqueous phase electron donor could be injected near the cap rock of a nitrate-rich saline aquifer as a preconditioning step for geological CO2 storage.41 Also, (sea)water injected into deep oil and gas reservoirs is amended with nitrate to mitigate biosouring (i.e., sulfate reduction to sulfide).55 These scenarios all lead to the formation of plumes created by injected fluids that can mix with the surrounding pore water to promote denitrification through transverse mixing of electron donor and acceptor,56 and denitrification can promote biomineralization. In this study, we investigate CaCO3 biomineralization driven by the denitrifying bacterium Pseudomonas stutzeri strain DCPPs1 under transverse mixing conditions in micromodels. The objectives were to determine if microbial metabolic activity promotes CaCO3 biomineralization, identify the mechanisms

■

MATERIALS AND METHODS Bacterial Strain and Culture Media. The denitrifying bacterial pure culture Pseudomonas stutzeri strain DCP-Ps1 (strain DCP-Ps1) was used in this study (provided from R. Sanford’s lab at the University of Illinois Urbana−Champaign). Strain DCP Ps-1 reduces NO3− via denitrification with the production of N2 gas under anaerobic conditions. P. stutzeri has been shown to accumulate nitrite in the presence of nitrate during denitrification.58−60 It was proposed that this is because nitrate reductase has a competitive advantage over nitrite reductase for the electron acceptor, which causes transient accumulation of extra-cellular nitrite during denitrification.60 The medium for culture growth and mineralization experiments was prepared using a modified Bold’s basal medium,14 whose composition is given in the Supporting Information. The solution was sparged with a mixture of He and CO2 gases to remove both oxygen and adjust the pH to 6.70 ± 0.03. A growth solution of strain DCP-Ps1 was prepared by adding it from a previously enriched pure culture isolate (∼5% v/v) to culture-tubes containing 20 mL of media, 4 mM CH3COONa (electron donor) (Sigma-Aldrich, ≥ 99%), and 1 mM NaNO3 (electron acceptor) (Sigma-Aldrich, ≥ 99%). The NO3− was depleted in ∼24 h, and repeated additions of this limiting substrate were performed over 3−4 days until the optical density (OD600) reached ≥0.1 (∼107 cells/ml). The bacterial growth was conducted at 30(±1) °C under anaerobic conditions. Micromodel. Micromodel experiments were performed using a homogeneous pore network 2 cm long by 1 cm wide, consisting of a uniform distribution of cylindrical posts (300 μm in diameter) representing soil grains, and separated by 180 μm pore bodies and 40 μm pore throats (Figure 1). The pore network depth was ∼18 μm, and the calculated porosity based on geometric design was ∼0.39. The permeability of the micromodel, calculated based on the Kozeny−Carman porosity-permeability relationship,61,62 was 9.56 × 10−7 cm2. The micromodel has two up-gradient inlets (A and B) for injecting reactants, one down-gradient inlet (C) for a quenching reagent to prevent reactions downstream of the pore network, and one outlet (D) (Figure 1). As described previously,63,64 standard photolithography methods and inductively coupled plasma-deep reactive ion etching (ICPDRIE) were used to etch pore networks in silicon wafers, which were treated by thermal oxidation to create a thin hydrophilic silicon dioxide layer (∼100 nm); etched wafers were sealed with a glass coverslip by anodic bonding to form closed flow channels. 12095

DOI: 10.1021/acs.est.5b00152 Environ. Sci. Technol. 2015, 49, 12094−12104

Article

Environmental Science & Technology NO3− + 5/8CH3COO− + H+ = 1/2N2 + 5/8CO2 + 5/8HCO3− + 9/8H 2O

NO3−

(1)

+

One mole of leads to one mole of H consumption and, coupled with bicarbonate generation, results in an increase of pH and alkalinity. This will promote environmental conditions favorable for biologically induced CaCO3 precipitation via an increase in the saturation index (SI) with respect to CaCO3 precipitation, which is defined as the logarithm of the ratio of the product of ion activity of calcium and carbonate ions (= aCa2+ aCO32−) to the saturation constant (Ksp) of CaCO3.65 Volumetric flow rates of 35 and 20 μL/h through inlet A and B and inlet C, respectively, were maintained during all three steps of the experiment. Step I represents the biofilm growth period. Here, syringes for inoculation were replaced with syringes containing 5 mM nitrate (inlet A) and 7.5 mM acetate (inlet B) in growth media, both amended with 5 mM CaCl2 (Aldrich ≥99.9%). CaCl2 was added because cations have been shown to promote biomass formation.66 The pH of these solutions was 6.70 ± 0.03 with an estimated alkalinity of ∼10 mM (Table 1). Biomass growth was established during step I and was maintained for 8−10 days until no further biomass growth was apparent. Step II represents our effort to evaluate CaCO3 precipitation in the absence of metabolically active cells generating alkalinity and consuming protons via denitrification. Here, inlets A and B were connected to syringes containing 0 mM nitrate and 15 mM acetate, respectively, in growth media amended with 30 mM Ca2+. The pH of these solutions was 7.2 with an estimated alkalinity of ∼18 mM (Table 1), which corresponds to the maximum expected alkalinity generated when denitrification is occurring in the micromodel during step III. Step II was maintained for ∼6 days to monitor CaCO3 precipitation. Step III conditions tested the extent of biomineralization with denitrification; inlets A and B were connected to syringes containing 10 mM nitrate and 15 mM acetate, respectively, in growth media amended with 30 mM Ca2+. The pH of these solutions was 6.70 ± 0.03 with alkalinity similar to that in step I. During denitrification, pH and alkalinity values can increase in the micromodel to maximum values of 7.2 and 18 mM (eq 1). Step III was maintained for at least 8 days to monitor CaCO3 precipitation. Steps I−III were replicated in two separate micromodels using separate inocula. An additional control experiment was performed with only steps I and III to evaluate whether the no-nitrate feed condition (step II) had an influence on biologically promoted CaCO3

Figure 1. (a) Micromodel experimental setup with two inlets (A and B) on the upstream side of the porous medium (20 mm × 10 mm) for injecting reactants, one down stream inlet (C) for injecting reaction quenching reagent, and one downstream outlet (D). (b) The porous medium consists of a homogeneous pore network with cylindrical posts ∼0.018 mm deep.

Micromodel Inoculation. The inoculum was prepared by adding 1 mL of the fresh growth solution to 20 mL of growth medium (pH 6.70 ± 0.03) containing 4 mM of acetate (and no NO3−) in a sterilized gastight glass syringe (Hamilton). Fresh medium containing 1 mM NO3− and no bacteria was put into a separate glass syringe. These two syringes were connected to inlets A and B, respectively. A volumetric flow rate of 35 μL/h was continuously injected into each inlet using a dual syringe pump (Cole-Parmer), resulting in an average Darcy flux of 0.47 cm/min. A solution of H2O2 (5% v/v) was injected through inlet C at 20 μL/h to inhibit biomass growth in the outlet. The inoculation process was performed until the first immobile biomass was observed along the mixing zone between acetate and NO3− solutions (∼4−5 days). This very small amount of biomass (relative to the amount that grew later) served as seeds for the subsequent biofilm growth period described in the next section. CaCO3 Biomineralization Experiment. After micromodel inoculation, a three-step experiment was performed to test the impacts of solution pH and alkalinity, nucleation, and metabolic activity on CaCO3 precipitation (Table 1). Metabolic activity was driven by denitrification, and the overall reaction is as follows

Table 1. Summary of Biotic CaCO3 Precipitation Experimental Conditions maximum SI based on influent conditionsb

influent conditionsa experiments step I: biofilm growth step II: CaCO3 precipitation test without denitrification step III: microbially induced CaCO3 precipitation test

inlet A

inlet B

7.5 mM Acet,c 5 mM Ca2+ 5 mM NO3−, 5 mM Ca2+ d pH 6.7 (6.95), TALK = 10 mM (14) 0 mM NO3−, 30 mM Ca2+ 15 mM Acet, 30 mM Ca2+ pH 6.7 (7.2), TALK = 10 mM (18) 10 mM NO3−, 30 mM Ca2+ 15 mM Acet, 30 mM Ca2+ pH 6.7 (7.2), TALK = 10 mM (18)

calcite

vaterite

0.23 (0.62)

−0.34 (0.05)

1.56 (1.56)

0.99 (0.99)

0.81 (1.56)

0.24 (0.99)

a

pH and TALK values in parentheses are the calculated maximum values based on eq 1 due to mixing in the pore network. The maximum values ideally occur at the central mixing point in the first pore body from the inlet. bSI calculation is based on ideal mixing along the centerline, and the values in parentheses are based on eq 1. pKsp values for calcite and vaterite are 8.42 and 7.73, respectively (Stumm, 199265). cAcet stands for acetate. d Total alkalinity (TALK) = 2[CO32−] + [HCO3−] + [OH−] − [H+]. 12096

DOI: 10.1021/acs.est.5b00152 Environ. Sci. Technol. 2015, 49, 12094−12104

Article

Environmental Science & Technology

Figure 2. Bacterial-driven CaCO3 precipitation. (a) Step I: biofilm formation along transverse mixing line of micromodel with acetate (7.5 mM) and nitrate (5 mM) as the electron donor and acceptor, respectively, at an inlet pH 6.7. (b) Step II: when 30 mM Ca2+ was introduced in the micromodel at elevated pH (7.2) and alkalinity (18 mM) (maximum expected pH and alkalinity values due to 5 mM nitrate reduction). No nitrate was present, which eliminated microbial activity, and no precipitation was observed. (c) Step III: CaCO3 polymorphs were precipitated when 5 mM nitrate was supplied with 30 mM Ca2+ at an inlet pH 6.7. Flow is from left to right.

DCP-Ps1 inoculum grown with acetate and nitrate. One tube, representing the metabolically inactive bacteria in step II, was further amended with OH− and HCO3− to obtain pH/Alk ∼7.2/18 mM. A second tube was amended with 5 mM NO3−, and it represented the metabolically active bacteria in step III of the micromodel experiment. A third tube was amended with 5 mM of NO2− because nitrite was expected to be an initial product of nitrate reduction in the system. A fourth tube (control) was given no additional amendments. All four samples (in duplicate) were incubated at 30 °C for 30 h before zeta potential measurements. Pore-Scale Reactive Transport Model. An existing51,57 reactive transport model was modified to account for Monod− Haldane denitrification kinetics67 under the experimental conditions in this study. An open source lattice-Boltzmann solver68 and a finite volume method (FVM) were used to simulate two-dimensional steady, laminar fluid flow, and reactive transport in pore spaces. A model domain of 6 mm × 3.36 mm with an inlet region length of 1.05 mm (i.e., no porous network) was used. A 5 μm grid spacing was used, as in our previous studies,57,64,69,70 and this results in negligible numerical dispersion. Details of denitrification kinetics considered in the model are summarized in the Results and Discussion. Inlet conditions are defined as total concentrations of primary species (e.g., nitrate, acetate, calcium, H+, HCO3−) based upon both measured and calculated pH values of inlet solutions. Molins et al.71 demonstrated that a single diffusion coefficient can be used for all aqueous species involved in calcite dissolution without significant errors at pH ≥ 5. Therefore, for simplicity, the diffusion coefficients for all aqueous species are assumed to be equal (i.e., 1.0 × 10−5 cm2/s) to ensure local charge neutrality44,72 and avoid complicated charge-induced interactions between species. Key transport and reaction parameters used in simulations are reported in Table S1, and aqueous reactions and stability constants are listed in Table S2.

precipitation in step III. Abiotic control batch experiments were also performed in which microbe-free growth medium was amended to obtain a pH, alkalinity, and Ca2+ concentration of 7.2, 18 mM, and 30 mM, respectively, similar to influent solutions in step II. Triplicate samples were incubated for more than 6 days. All of the experiments were maintained at 30 ± 1 °C. Raman Spectroscopy, Fluorescence Tracer Experiments, and Energy Dispersive Spectroscopy. After completion of step III, carbonate precipitates were analyzed in situ using Raman spectroscopy with a LabRAM HR 3D (Horiba Scientific) to identify CaCO3 polymorphs at a spatial resolution of 2 μm. Next, a nonreactive fluorescence tracer test was used to evaluate the impact of biominerals on pore blockage and transverse mixing. Last, micromodels were cut with a dicing saw (Disco DAD-6TM Wafer Dicing Saw) into ∼1−2 mm strips transverse to the flow direction. A high resolution (2 nm) environmental scanning electron microscope (ESEM) (Philips/FEI XL30 ESEM-FEG) was used to image carbonate crystals in individual pores, and a coupled energydispersive spectroscopy (EDS) system was used to identify the elemental composition of minerals. Further details of these analyses are in the Supporting Information. Data Acquisition and Image Analysis. All micromodel images were obtained using a Nikon Epiphot 200 inverted microscope with reflected differential interference contrast (DIC) and fluorescent microscopy. Selected images were analyzed via thresholding to determine areas corresponding to individual CaCO3 biominerals. At each time step, care was taken to analyze a similar fraction (i.e., ∼15−23%) of the total number of biomineral aggregates. Details of imaging and thresholding methods are in Supporting Information. Bacterial Surface Charge Measurement. Bacterial surface charge was evaluated using a Zetasizer Nano ZS90 instrument (Malvern Instruments, Southborough, MA) under conditions representative of those during step II (inactive biomass) and step III (active biomass) in micromodel experiments. Four samples, each containing 20 mL of growth media (pH 6.7, Alk = 10 mM) in culture tubes, were prepared by adding 7.5 mM acetate, 5 mM Ca2+, and 1 mL of strain

■

RESULTS AND DISCUSSION Biofilm Growth and Distribution in the Micromodel. After inoculation, growth occurred during the first 4−5 days in 12097

DOI: 10.1021/acs.est.5b00152 Environ. Sci. Technol. 2015, 49, 12094−12104

Article

Environmental Science & Technology

Figure 3. Dendritic (left) and lumped-type (right) CaCO3 crystal morphologies and their mineralogical composition (Raman spectra) formed during step III of the biotic CaCO3 precipitation experiment across the transverse mixing zone in the micromodel. The Raman spectra with peaks at 156, 283, 711, and 1085 cm−1 correspond to calcite and 266, 302, 750, 1073, and 1089 cm−1 correspond to vaterite (Edwards et al. 200585). The other peaks at ∼520 cm−1 and the shoulders at 920 cm−1 correspond to the silicon surface of the micromodel. Flow is from left to right in each image panel.

both influent syringes, and CaCO3 precipitation was evaluated without denitrification in step II, and then with denitrification in step III (Table 1). A microscopic image of biofilm on the last day of step II (t = 16 days from the start of experiment) is shown in Figure 2b. The biomass distribution is very similar to that at the end of step I, indicating no apparent biomass growth or decay occurred as expected due to the absence of nitrate (electron acceptor). It is also clear that no CaCO3 precipitation occurred. This means that the combination of the elevated saturation index (SIcalcite ∼ 1.56) and microbial nucleation sites alone did not promote CaCO3 precipitation over the six day period. Similarly, our abiotic control batch experiments with microbe-free media and similar SI showed no precipitation over 6 days. We note that in an abiotic study of CaCO3 precipitation in a similar micromodel without biomass, precipitation only occurred after the SIcalcite reached 2.8.52 In step III, the influent pH and alkalinity were reduced to levels used in step I, and nitrate feed was reestablished (Table 1). An image of the pore network after 6 days of step III (i.e., t = 22 days from start of the experiment) is shown in Figure 2c. Sequential images during step III are shown in Figure S2. New biomass grew in the first day, mainly in the primary growth zone, indicating that biomass remained viable during stage II. Precipitation of CaCO3 started to occur within 3 days of starting step III. In previous CaCO3 precipitation experiments performed in porous media, SI increased due to microbial activity and led to precipitation.76,77 The role of actively metabolizing microorganisms, however, was not previously identified as a controlling factor for precipitation independent of an elevated SI or the presence of biomass nucleation sites. Several studies have approximated the net cell surface charge by measuring the zeta potential of bacterial suspensions.78−81 CaCO3 precipitation has been attributed to nucleation and growth at Ca2+ binding sites on negatively charged surfaces of cyanobacteria,79,80 and metabolic activity has been shown to affect surface charge.81 In this study, the zeta potential of strain DCP-Ps1 was measured under metabolically inactive and active conditions. The sample containing metabolically active DCPPs1 (i.e., amended with NO3− or NO2−) showed larger negative zeta potential (i.e., a more negative surface charge) compared to metabolically inactive DCP-Ps1 (i.e., without NO3−), even when the latter was adjusted to the same pH and alkalinity as

the micromodel with inlet solutions of acetate and nitrate during step I. The primary zone of biomass at day 1 trended from near the centerline at the inlets to above the centerline downstream (i.e., near the outlet). A secondary zone of biomass growth appeared below the primary zone at nearly the same time, starting from just below the micromodel centerline at the inlet and trending even further lower in the transverse direction with increasing longitudinal distance from the inlet. The position of these two zones of biomass growth is apparent at 10 days (Figure 2a) when growth had leveled off (i.e., pseudo steady state). Biomass in the primary growth zone is visually much more dense (i.e., darker/lower image intensity) than biomass in the secondary growth zone. Also, biomass in the primary growth zone is wider near the inlet (∼3−4 pore bodies) and narrower downstream (∼2 pore bodies) (Figure 2a). The two zones of growth suggest an intermediate of nitrate, likely nitrite (which is shown to accumulate in anoxic systems for biological nutrient removal processes73,74), is being produced in the pore network and must diffuse away from nitrate transverse to flow in order to react. The denser primary growth line suggests the microbial conversion of nitrate to nitrite is more favorable with respect to biomass growth and/or attachment than that of nitrite. Enlarged images of biomass in single pores of the primary and secondary growth zones are shown in Figure 2. Biofilm growth in individual pore bodies took place on the downstream side of the silicon post in the primary growth zone, and thick biomass growth extended over the pore body and was slightly curled toward the micromodel centerline (Figure S1). In contrast, biofilm growth started on the upstream side of the silicon post in the secondary growth zone (Figure S1). Flow stagnation points are located on the upstream and downstream sides of silicon posts, and the growth patterns suggest biomass grows initially in the lowest flow zones where shear forces are smallest.75 The growth patterns are similar to those observed in our earlier works.48−50 Biomass detachment may also have occurred during the course of biofilm growth in step I, and therefore, the biofilm growth in Figure S1 represents apparent growth. Promotion of CaCO3 Precipitation in the Micromodel. After biomass was established over 10 days in step 1, the influent Ca2+ concentration was increased from 5 to 30 mM in 12098

DOI: 10.1021/acs.est.5b00152 Environ. Sci. Technol. 2015, 49, 12094−12104

Article

Environmental Science & Technology

Figure 4. (a) Spatial distribution of biofilm observed in experiments during step III. Simulated spatial distribution of (b) CH3COO−, (c) NO3−, (d) NO2− remaining, (e) NO2− removed, (f) HCO3−, (g) pH, and (h) saturation index (SI) for vaterite (log10 (activity product/Ksp)), assuming reactions only occur where biomass is present. The various species concentrations in the above panels are in M.

consistent with the control experiment (with only steps I and III). Distribution, Crystalline Morphology, and Mineralogy of CaCO3 Biominerals. Representative CaCO3 crystal morphologies that precipitated during step III are shown in Figure 3. In all cases, nucleation of CaCO3 within a pore consistently occurred at the same locations where microbial biofilms were present, as indicated for biominerals in Figure 3a. This correlation illustrates the fundamental role played by actively denitrifying microorganisms on CaCO3 precipitation and suggests that cell membranes may serve as nucleation sites via catalytic macromolecules on their surfaces.21 Additionally, microbial metabolism may also help to create localized high saturation index zones in the vicinity of the biofilm. Furthermore, CaCO3 nucleation first occurred downstream and then gradually populated upstream. This is illustrated in Figure S4, where total area associated with CaCO3 precipitate was calculated for the upstream half and downstream half of the micromodel. The representative CaCO3 crystal morphologies observed included: (1) euhedral to subhedral crystals of branched clusters of 5−25 μm length along the C-axis that form dendritic aggregates up to 180 μm in length (Figure 3b− d) and feather “shrub” aggregates up to 100 μm in length

the metabolically active strain DCP-Ps1 (Figure S3). The more negative surface charge of metabolically active bacteria may have been responsible for the observed CaCO3 precipitation in step III, possibly due to greater local pH and alkalinity, or enhanced adsorption of Ca2+ to a more negative surface and subsequent precipitation. It is also evident from Figure 2c that most CaCO 3 precipitation occurred in and slightly above the secondary growth zone (not in the primary growth zone). This was initially surprising because pH consumption and alkalinity, which contribute to CaCO3 precipitation, are associated with denitrification, and because in other studies segregation of biomass growth and precipitation was not observed. The predominance of CaCO3 precipitation near the secondary growth zone, however, suggests that more alkalinity production and proton consumption occur here. All alkalinity production associated with denitrification is affiliated with NO2− reduction to nitrous oxide (N2O) and not NO3− reduction to NO2−. This supports our assertion that NO3− reduction is occurring in at least two steps, and that produced NO2− is being transported away from NO3− transverse to flow before reacting. We note that replicate micromodel experiments were consistent and also 12099

DOI: 10.1021/acs.est.5b00152 Environ. Sci. Technol. 2015, 49, 12094−12104

Article

Environmental Science & Technology (Figure 3e−h), (2) spherule-shaped crystals up to 5−30 μm in diameter (Figure 3i−n), and (3) isolated euhedral rhombohedral crystals as large as 50 μm in diameter (Figure 3o). Both the dendritic and spherule shapes coupled with their surface roughness may be indicative of crystal formation under highly supersaturated conditions.82−84 Raman spectra peaks for biominerals indicate that two polymorphs of CaCO3 (vaterite and calcite85) precipitated in the micromodel during the biomineralization experiment (Figure 3). Among the ∼70 biomineral single crystals or aggregates analyzed in the entire micromodel pore network, the majority were calcite (∼70%), and among the ∼27 biomineral precipitates analyzed in the upstream 1/3rd of the pore network, 50% were vaterite. Irrespective of the crystal morphologies, 65% of the smaller and medium size crystals (