- Email: [email protected]
A biofilm model for assessing perchlorate reduction in a methane-based membrane biofilm reactor
Accepted Manuscript A Biofilm Model for Assessing Perchlorate Reduction in a Methane-Based Membrane Biofilm Reactor Jing Sun, Xiaohu Dai, Lai Peng, Yiwen Liu, Qilin Wang, Bing-Jie Ni PII: DOI: Reference:
S1385-8947(17)31087-2 http://dx.doi.org/10.1016/j.cej.2017.06.136 CEJ 17217
To appear in:
Chemical Engineering Journal
Received Date: Revised Date: Accepted Date:
30 April 2017 23 June 2017 24 June 2017
Please cite this article as: J. Sun, X. Dai, L. Peng, Y. Liu, Q. Wang, B-J. Ni, A Biofilm Model for Assessing Perchlorate Reduction in a Methane-Based Membrane Biofilm Reactor, Chemical Engineering Journal (2017), doi: http://dx.doi.org/10.1016/j.cej.2017.06.136
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
A Biofilm Model for Assessing Perchlorate Reduction in a Methane-Based Membrane Biofilm Reactor
Jing Sun1, Xiaohu Dai1, Lai Peng2, Yiwen Liu3, Qilin Wang4, Bing-Jie Ni1,*
1
State Key Laboratory of Pollution Control and Resources Reuse, College of
Environmental Science and Engineering, Tongji University, Shanghai 200092, PR China 2
Laboratory of Microbial Ecology and Technology (LabMET), Ghent University, Coupure Links 653, 9000 Ghent, Belgium
3
Centre for Technology in Water and Wastewater, School of Civil and Environmental Engineering, University of Technology Sydney, Sydney, NSW 2007, Australia 4
Griffith School of Engineering, Griffith University, Nathan Campus, QLD 4111, Australia
*Corresponding author: Bing-Jie Ni, P: +86 21 65986849; F: +86 21 65983602; E-mail: [email protected]
Abstract Perchlorate (ClO4-) is recognized as an important contaminant in surface water and groundwater, which would pose health risks at very low concentrations. A methane-based
membrane
biofilm
reactor
(MBfR)
has
been
successfully
demonstrated for perchlorate reduction, which provided an alternative solution for perchlorate remediation with low cost. In this work, a multispecies biofilm model was developed to evaluate perchlorate reduction in the methane-based MBfR under 1
different operational conditions. The model was calibrated and validated using the experimental data from the long-term operation of the MBfR at seven distinct stages. The results suggested that the developed model could satisfactorily describe perchlorate reduction and denitrification performances in the MBfR (R2>0.9). The modeling results provided insight into the microbial community distribution in the biofilm, with aerobic methanotrophs and perchlorate reduction bacteria being mainly located at the membrane side (~ 60%) and heterotrophic bacteria being situated near the liquid side (~ 50%). The model simulations indicated that over 80% of perchlorate removal efficiency could be achieved through controlling the optimal combinations of methane pressure (PCH4) and perchlorate loading (LClO4) (e.g., applying a PCH4 of 30kPa at a LClO4 of 0.08 gCl/m2/d). In addition, the perchlorate reduction would be inhibited by the presence of nitrate and nitrite in the MBfR, which should be appropriately controlled during the future practical application of the promising process.
Keywords: membrane biofilm reactor; perchlorate removal; methane; biofilm model; practical application
1. Introduction Perchlorate salts, including ammonium, potassium and sodium perchlorate have been widely used in rocket fuels, pyrotechnics, matches, munitions and many other industries [1, 2]. These perchlorate salts usually have high solubilities and would be readily dissociated in water producing perchlorate anion, which is relatively nonreactive and stable, but extremely mobile in aqueous systems [3]. Perchlorate (ClO4-) is recognized as an important contaminant in surface water and groundwater, 2
as it would pose health risks at very low concentrations and is difficult to remediate from the water. It can interfere with the body’s iodine intake, consequently inhibiting thyroid hormone production, which is thought essential for growth and development [4]. Therefore, the U.S. Environmental Protection Agency is in the process of evaluating a maximum contaminant level (MCL) for perchlorate in the range of 1−18 μg/L [5].
Many previous studies have been carried out to investigate the remediation of perchlorate in water through microbiological reduction [6-8]. A recent study, for instance, reported a successful example of microbial perchlorate removal in a methane-based membrane biofilm reactor (MBfR) [9]. In this study, methane was supplied into lumens of hollow-fiber membranes and further permeated through the non-porous walls into biofilms attached on outer surfaces of membranes. Meanwhile, perchlorate in wastewater was diffused into biofilms, where perchlorate reduction bacteria (PRB) developed and the microbial perchlorate reduction took place. The result showed that the biofilms could reduce up to 5 mg/L ClO4- to a non-detectable level when methane delivery was not limiting. While, under the methane limiting condition, the removal of perchlorate could be interfered by the presence of nitrate and nitrite in water.
As methane is an inexpensive and widely available carbon source, its utilization for perchlorate remediation could largely reduce the cost, compared with other methods such as chemical reduction and iron exchange [2, 10]. Also, it proposed a promising solution for energy recovery from the wastewater system, because methane is a key product of the anaerobic digestion of biosolids [11, 12]. Although, the utilization of 3
methane has potential risks due to its relatively low explosion limit [13], the application of the “bubble-less” membranes could enhance the operational safety by preventing methane losses to the atmosphere. In addition, as methane is a potent greenhouse gas, using methane for perchlorate reduction could potentially alleviate the greenhouse gas emission from the wastewater treatment system [14, 15]. Therefore, the application of methane-based MBfR for perchlorate remediation has a good prospect of practical application. In this regard, a mathematical model for describing perchlorate reduction in the methane-based MBfR is highly desirable to facilitate its full-scale application. The comprehensive investigation into this promising system, especially into the effect of different operational parameters on perchlorate removal, is also required to potentially improve the system performance.
The experimental investigation implied that the microbial perchlorate removal in the methane-based MBfR system could be affected by three key operational parameters, i.e. the methane pressures, perchlorate loadings and nitrate or nitrite loadings [9]. These parameters governed the perchlorate removal not only by directly affecting the perchlorate reduction rate, but also by influencing the development of PRB in the biofilm. In the biofilm, the possible competition and promotion mechanisms mainly include: i) The competition between PRB and heterotrophic bacteria (HB) for the same electron donor [16, 17]; ii) The competition between different microorganisms and associated products for space in the biofilm [18]; iii) The competition within PRB between nitrate/nitrite and perchlorate for the same resources such as electrons and possibly reductase enzymes[19]; and iv) the promotion of the growth of PRB through their utilization of nitrate or nitrite [20]. Because of multiple mechanisms, microbial species, and substrates in the methane-based MBfR system, the links between these 4
operating parameters and perchlorate removal are not straightforward. In this case, multispecies biofilm modeling is advantageous for quantitatively integrating the microbiological and physical phenomena that control the perchlorate removal in the methane-based MBfR.
This study aimed to develop a multispecies biofilm model to predict the perchlorate removal in the methane-based MBfR system under different operational conditions. The developed model comprehensively considered the interaction between different microbial processes in the biofilm as well as the gas delivery characteristics through the membrane substratum in the reactor. The model was calibrated and validated using the experimental data from a long-term operated methane-based MBfR at different operational stages. The effect of key operation parameters on the perchlorate removal was then investigated using the validated model. It is expected that the established model could provide supports for the further development of the methane-based MBfR for efficient perchlorate remediation.
2. Material and methods 2.1. Development of the biological processes model The key biological processes occurred in the biofilm of the MBfR was considered based on the experimental observations [21], which were summarized in Figure 1. In the biofilm, the methane could be aerobically oxidized by methanotrophs (AMO) with oxygen in the influent. The oxidation product, methanol, was then severed as the electron donor for perchlorate reduction and heterotrophic denitrification with the respiration of PRB and HB. On the other hand, methane could also be anaerobically oxidized by denitrifying anaerobic methane oxidation (DAMO) bacteria or archaea 5
using nitrate and nitrite as electron acceptor in the biofilm where oxygen could not be penetrated to. The anaerobic methane oxidation coupled to perchlorate reduction was not included in this work, as so far there still no direct evidence to prove the occurrence of such process in the system. Through aforementioned oxidation or reduction processes, the microorganisms gained energy to synthesis new biomass, and meanwhile
to
produce
extracellular
polymeric
substances
(EPS)
and
substrate-utilization-associated products (UAP) [22]. The formed EPS could be hydrolyzed into biomass-associated products (BAP). The BAP and UAP could be both used by HB as organic electron donors with nitrate or nitrite as their electron acceptors. All microorganisms involved in the system are subject to endogenous respiration, with the production of inert biomass (IB) and the reduction of the electron acceptors to gain energy for cell maintenance.
According to the biological processes described above, the developed biological model includes seven particulate species, i.e. AMO (XAMO), HB (XHB), PRB (XPRB), DAMO bacteria (XDB), DAMO archaea (XDA), EPS (XEPS) and inert biomass (XI), and eight soluble species, i.e. methane (SCH4), oxygen (SO2), perchlorate (SClO4), nitrate (SNO3), nitrite (SNO2), methanol (SCH3OH), UAP (SUAP), and BAP (SBAP), as listed in Table S1. The unit was gN/m3 for all nitrogen species, gCl/m3 for perchlorate and gO/m3 for oxygen, while concentrations of other compounds were quantified based their chemical oxygen demand (COD) equivalent, i.e. gCOD/m3. The stoichiometric matrix of the model was summarized in Table S2 (SI). The rates of the growth and endogenous respiration of microorganisms were modelled using Monod-type kinetics while hydrolysis of the EPS was simulated through first-order kinetics [22, 23]. Competition for the same resources in PRB is expressed by 6
competitive-inhibition coefficients in the acceptor part of dual-substrate Monod kinetics [24]. The kinetic expressions of the reactions included in the model were listed in Table S3 (SI). The definitions of parameters in the kinetic expressions were shown in Table S4 (SI).
2.2. Methane-based membrane biofilm reactor model A multispecies one-dimensional biofilm model was constructed to model the methane-based membrane biofilm reactor through employing the software AQUASIM 2.1d [25]. The one-dimensional conservation law is formulated as an overall balance equation of between the mass conserved and utilized in this model [18]. The methane-based MBfR system was modeled with two different compartments in AQUASIM: the membrane lumen was modelled with a completely mixed gas compartment and the reactor bulk liquid and the biofilm was modelled through a biofilm compartment. The gas compartment was connected to the base of the biofilm via a diffusive link. The flux of methane (LCH4) from the gas to the biofilm matrix compartment through the membrane was modeled using the following equation (Equation 1) according to Sun et al. [26] (1) where
and
are concentrations of oxygen in the gas and biofilm matrix
compartments (gCOD/m3), respectively; of oxygen (m/day); and
is the overall mass transfer coefficient
is the Henry’s coefficient for oxygen (mole m-3
gas/mole m-3 liquid).
In the biofilm compartment, the biofilm structure was represented as a continuum without considering diffusive mass transport of biomass. The steady-state biofilm 7
thickness was established by controlling the detachment using a global detachment velocity, i.e. ude, (μm/d), in model simulations [26]. The composition of solids detached from the biofilm conformed to their composition at the biofilm surface. The detached particulates were assumed to be washed out of the system with the effluent, and no reattachment of detached particulates was considered in the model. The water fraction of the biofilm matrix was kept constant at 0.6, which was in the range of most probable water fractions in the wastewater biofilm [27]. The biomass density was kept at 100000 gCOD/m3 [28].
2.3. Experimental data used for testing the developed model The experiment data from a methane-based MBfR reported by Luo et al.[9] was used to calibrate and validated the developed model. Briefly, the MBfR had a total volume of 65ml and comprised hollow fiber gas-transfer membranes (Model MHF-200TL, Mitsubishi, Ltd., Japan), which resulted in a total membrane surface of 7.0 cm2. The fibers were glued into a gas-supply manifold at the bottom of the MBfR, with the top of each fiber being sealed. The simplified graphical scheme of the experimental methane-based membrane biofilm reactor was presented in Figure S1 (SI). Methane was delivered into lumens of the membranes and the bulk liquid in reactor was completely mixed by recirculation with a peristaltic pump at 100 mL/min. The MBfR was inoculated with 10 mL of ANMO-D culture and was fed with 2 mg N/L of NO2for 40 days to accumulate enough biomass. Then the MBfR was operated under 7 stages to investigate ClO4- reduction in the presence of NO3- and NO2-. The operation conditions of all stages are summarized in Table S5 (SI). The influent flow rate was 0.5 mL/min and the CH4 pressure was 10 psi (69 kPa) for stages 1−3 and 15 psi (103 kPa) for stages 4-7. The dissolved oxygen concentration was ~0.2 mg/L for the 8
influent and ≤0.1 mg/L for the effluent. The pH was controlled at 7.0 ± 0.2 with hydrochloric acid. For each stage, the MBfR was operated to reach the steady state, which was determined as effluent concentrations stable (<10% variation) for a minimum of three hydraulic retention time (HRTs). The effluent concentrations of ClO4- NO3- and NO2- were measured using ion chromatography. More details could be found in Luo et al.[9]. The removal flux of ClO4- , NO3- and NO2- was calculated according to the following equation: (2) where S° and S are the influent and effluent ClO4- , NO3- or NO2- concentration (g/L), Q is the influent flow rate to the MBfR system (L/d), and A is the membrane surface area (m2).
2.4. Model calibration and validation The developed biofilm model includes 27 species-specific biochemical processes and 58 kinetic and stoichiometric parameters (Tables S2 and S3 (SI)). The values of most of these parameters were well established in previous studies and thus were adapted from literature, as presented in Table S4 (SI). However, limited information was available in literature for the parameters related to XAMO and XPRB. Therefore, parameter estimation based on experimental measurements was conducted for three parameters of XSOB and three parameters of XPRB, which are the maximum growth rate of XAMO (μXAMO), the half-saturated concentration of oxygen for XAMO (KXAMO,O2) , the inhibition constant of NO2- on XAMO (KIXAMO,NO2), the maximum growth rate of XPRB (μXPRB), the inhibition constant of NO2- on XPRB (KIXPRB,NO2) and perchlorate reduction factor on the maximum growth of XPRB (ηClO4). The experimental data obtained in the operational stage of 1-4 were used for model 9
calibration and parameters estimation, which was conducted by minimizing the sum of squares of the deviations between the measured data and the model predictions. The objective function to be minimized in the parameter estimation is as follows [25]: (3) where
is the measured data at time ti (i from 1 to n).
is the calculated
value by the model at time ti (i from 1 to n). p is the parameters of μXAMO, KXAMO,O2, KIXAMO,NO2, μXPRB, KIXPRB,NO2 and ηClO4.
is the standard deviation of the
measurements. With the built-in simplex and secant algorithms, at each iteration, all parameter arrays were replaced by new values until
are close enough to fulfill
the convergence criterion.
Model validation was then conducted with calibrated parameters using the experimental data during the stage 5-7 that were operated under completely different experimental conditions compared to the calibration experiments (e.g., different methane pressures and substrates loadings). The software AQUASIM 2.1d was employed for both calibrating the model parameters and simulating the system [25].
2.5. Model-based analysis of microbial community in the biofilm The validated MBfR model was then used to investigate the microbial community distribution in the methane-based perchlorate-reduction biofilm to provide insight into the key microbial interactions in the MBfR under different conditions. The model was simulated with different operational conditions as described in Section 2.2 until reaching the steady states. The fractions of different microorganisms along the depth of the biofilm under different operational conditions were calculated and compared. 10
Then the microbial distribution pattern in the biofilm could be revealed together with its variation with the different operational conditions. In addition, the relative abundances of different microorganisms in the biofilm were also calculated to provide information on the overall microbial community in the biofilm.
2.6. Impact of key operational parameters on perchlorate removal efficiency The simulation study was conducted with the validated model to investigate the effect of key operational parameters, including the methane pressures, perchlorate loadings and nitrate or nitrite loadings on perchlorate removal in the methane-based MBfR. The simulated MBfR system had a total water volume of 5L and a membrane surface area of 0.5 m2, which resulting in the area to volume (A/V) ratio of 100 m-1, a typical A/V ratio in the MBfR system. To evaluate the effect of methane pressures (PCH4) and perchlorate loading (LClO4), the comprehensive model simulations were conducted with different combinations of PCH4 and LClO4, covering a range of 20-120 kPa for PCH4 and 0.08-0.8 gCl/m2/d for L ClO4. The settings of other operational parameters were listed in Table S6 (SI), Scenario I. The perchlorate removal efficiency was calculated and compared under different conditions (SI Table S6), i.e., the different combinations of PCH4 (20-120 kPa) and LClO4 (0.08-0.8 gCl/m2/d). Similarly, the perchlorate removal efficiency was also calculated and compared to evaluate the effect of nitrate or nitrite loadings (LNO2 or LNO3) on perchlorate removal. The model simulation was conducted with different LNO2 or LNO3, covering a range of 0-0.8 gN/m2/d. Also, the LClO4 was varying from 0.08-0.8 gCl/m2/d. The settings of other operational parameters were listed in Table S6 (SI), Scenario II and Scenario III.
3. Results and discussion 11
3.1. Model calibration The effluent concentrations and the removal fluxes of perchlorate, nitrate and nitrite at the steady state during the operational stage 1-4 were used for the model calibration (SI Table S7). Three parameters of XAMO (μXAMO, KXAMO,O2, KIXAMO,NO2) and three parameters of XPRB (μXPRB, KIXPRB,NO2, ηClO4) were estimated by fitting simulation results to monitored data. The calibrated parameter values giving the optimum model fittings were listed in Table S4 (SI) and the model predictions and experiment measurements of the reactor performance are compared in Figure 2 (A-D).
Figure 2A and 2C presents the effluent concentrations of perchlorate, nitrate and nitrite obtained from both experimental study and model prediction. At Stage 1, the perchlorate removal efficiency was 23% with the presence of nitrite in the influent. It recovered to over 90% at Stage 2 when there was no nitrate or nitrite was in the influent. The high perchlorate removal efficiency (98%) was also observed during Stage 3, while the perchlorate reduction was totally inhibited when the influent nitrate concentration reached 11.3 mgN/L at Stage 4. The comparison in these figures suggested that the developed model could well capture the perchlorate reduction described above. Also, the model can satisfactorily described nitrite or nitrate removal in the system as shown in Figure 2C. Both the experimental results and model predictions suggested that most of the nitrite and nitrate could be removed at Stage 1 and 3, while the nitrate removal decreased to 69% at Stage 4.
The removal fluxes of perchlorate, nitrate and nitrite obtained from experimental study and model simulation was compared in Figure 2B and 2D. The maximum removal flux of perchlorate was reached at Stage 3 (0.07 g/m2/d), while the minimum 12
flux was only 0.0002 at Stage 4.The maximum removal flux of nitrate was reached at 0.56 gN/m2/d at Stage 4 and the removal flux of nitrite was 0.12 gN/m2/d at Stage 1. Figure 2B and 2D indicated that the model simulation also well described the removal fluxes showed in the experimental study. Overall, the well match between the model predictions and the experimental results (R2>0.9) as suggested in Figure 2 (A-D) supported the ability of the developed model for predicting perchlorate reduction in the methane-based MBfR system.
3.2. Model validation Model and parameters validation were conducted based on the comparison between the model predictions and the experimental measurements at Stage 5-7 (SI Table S7). The operational conditions at Stage 5-7 were significantly different from that at Stage 1-4, of which the data was used for model calibration. The methane pressure at Stage 5-7 was 1.5 times higher than that at Stage 1-3 and the influent perchlorate concentration at Stage 6-7 was over 5 times higher than that at Stage 1-4. The influent nitrite concentration at Stage 7 was over 3 times higher than that at the stage 1, while the influent nitrate concentration at Stage 5 was half of that at Stage 4.
The model and its parameters were firstly validated by the effluent perchlorate concentration and its corresponding removal flux at Stage 5-7. The comparison of the model predictions and experimental measurements as shown in Figure 3A and 3B suggested that the developed model could well describe the perchlorate removal as determined by the experimental study. The effluent perchlorate concentration at Stage 5 and 6 were negligible as indicated by both experimental and modelling results. The corresponding removal flux were 1.0 g/m2/d and 5.9 g/m2/d, respectively While at the 13
Stage 7, about 50% of perchlorate in the influent was removed, resulting in the effluent concentration of 2.5mg/L. The corresponding removal flux was 2.9 g/m2/d.
The effluent nitrite and nitrate concentrations and their corresponding removal flux determined by the experimental measurements were further used to validate the developed model. The model predictions were compared with experimental results in Figure 3C and 3D. Both the experimental and modelling results show that most nitrite and nitrate were removed at Stage 5 and Stage 7. The corresponding removal flux were 0.3 gN/m2/d for nitrite at Stage 5 and 0.4 gN/m2/d for nitrate at Stage 7, respectively. To assess the the accuracy of the developed model, we have conducted root mean square error (RMSE) analysis for Figures 2-3 and compare each RMSE to the corresponding maximum measurement. The results indicated that all of the RMSE are within 20% of maximum measurement value, indicating that the model prediction results generally match the experimental data. In sum, the good agreement between model simulations and the measured results (R2>0.9) confirmed the validity of the proposed model to describe the reactor performance related to perchlorate reduction and denitrification in the methane-based MBfR system.
3.3. Microbial community distribution in the biofilm Both model simulations and experimental results suggested that changes of substrate concentrations in the influent posed significant impact on perchlorate removal in the methane-based MBfR. This could be related to the evolution of microbial communities in the biofilm under different substrate conditions [29]. Therefore, simulation study was carried out to investigate the distribution of the microbial community in the biofilm in MBfR and its variation according to substrate 14
concentrations using the validated model. The distribution of solid components and their relative abundance in the biofilm on two operational stages, i.e. Stage 5 and Stage 6, were simulated and compared in Figure 4 (A-D).
As shown in Figure 4A and 4B, AMO was mainly located in the inner layer of the biofilm closed to the membrane, as methane was supplied through the membrane and consequently methane concentration was higher in the membrane side than the liquid side [30]. On the other hand, HB was mainly situated near the liquid side of the biofilm. This could be due to that the substrates for HB, such as oxygen, nitrite or nitrate was more available in the liquid side as they were diffused from the liquid to the biofilm. PRB showed a higher abundance near the membrane side in the biofilm, though it also required electron acceptors such as perchlorate, nitrite and nitrate penetrating from the liquid side. Thus, the relative low abundance of PRB at the liquid side possibly resulted from the competition between HB and PRB for the same substrates [31]. Since HB had a higher growth rate than the PRB (Table S4, SI), it is likely to outcompete PRB at the liquid side of the biofilm. Thus, most of PRB was developed in the inner side of the biofilm, where the penetration of substrate for HB was limited. As for the DAMO groups, DAMO bacteria accounted for about 5% of the solid components in the inner layer of the biofilm, while DAMO archaea was negligible. The low abundance of DAMO groups in the biofilm could be due to their relatively slow growth rate, compared with other microorganisms [32]. The inert biofilm presented a higher proportion in the inner biofilm while the EPS had a higher fraction near the biofilm surface. The EPS near the biofilm surface could enhance the stability of the biofilm by protecting microorganisms from erosion caused by hydrodynamic and mechanical shears [33]. 15
The changing of operational conditions from Stage 5 to Stage 6 would not change the microbial distribution pattern in the biofilm as shown in Figure 4A and 4B. However, the relative abundance of certain microorganisms could be varied according to substrates concentration. As presented in Figure 4C and 4D, the relative abundance of PRB increased from 6% at the Stage 5 to 18% at Stage 6. This could be the result of the increase of perchlorate concentration in the influent. On the contrary, the relative abundance of HB at Stage 6 was lower than that at Stage 5, as the influent concentration of nitrate decreased. The variation of microbial community in the biofilm is in good agreement with the change of the biofilm activities as suggested in the experimental study [9].
3.4. Effect of methane pressure and perchlorate loading on perchlorate removal The experimental study and modelling results both revealed that methane pressure (PCH4) and perchlorate loading (LClO4) were two key factors affecting the perchlorate removal in the methane-based MBfR. Therefore, simulation study using the validated model was carried out to investigate the combining effect of methane pressure and perchlorate loading on perchlorate removal efficiency. The tested PCH4 and LClO4 covered a wide range in order to potentially offer different options for real application [34]. Figure 5A and 5B illustrated the perchlorate removal efficiency under the extensive simulation conditions.
Overall, the perchlorate removal efficiency increased with the increase of PCH4 while decreased with the increase of LClO4. This is consistent with the findings reported in Long et al. [35], in which the Cr(VI) reduction increased by 40.3% and the 16
nitrate reduction increased by 30.2% when the methane pressure increased from 0.02 to 0.03 MPa in a methane-based membrane biofilm reactor. A biofilm model was developed to evaluate the key mechanisms including microbially-mediated perchlorate and nitrate reduction in the hydrogen-based membrane biofilm reactor and similarly the hydrogen loading was revealed to be properly managed at certain critical level to maximize the perchlorate and nitrate removal [17]. The perchlorate removal efficiency of up to 80% could be achieved when the combination of PCH4 and LClO4 was located in the region above the dash-dot line as indicated in Figure 5B. Specifically, when the applied LClO4 was 0.08 gCl/m2/d, a PCH4 of 30kPa was high enough to remove 80% of perchlorate under the simulated condition, whereas when the applied LClO4 was 0.64 gCl/m2/d, a PCH4 of 120 kPa was required to achieve the similar perchlorate removal efficiency. When LClO4 was above 0.64 gCl/m2/d, the increase of PCH4 from 100 kPa to 120 kPa did not resulted in significant improvement in the perchlorate removal and the perchlorate removal efficiency remained lower than 80%. This indicated that, apart from methane, other factors might also limit the perchlorate reduction in the methane-based MBfR, especially when LClO4 was relatively high. One possible factor could be the oxygen concentration in the reactor. In the MBfR system, methane could be aerobically oxidized by AMO, with its oxidation product, i.e. methanol, being used as electron donor for perchlorate reduction [10, 21]. If the oxygen in the system was depleted, the production of methanol would be inhibited. Consequently, the perchlorate reduction would be restricted due to the lack of electron donors.
3.5. Effect of nitrate and nitrite loading on perchlorate removal Nitrate and nitrite are two contaminants usually co-existed with the perchlorate in the 17
natural water bodies such as groundwater [9, 36]. The experimental investigation suggested that both nitrate and nitrite could affect the perchlorate removal in the methane-based MBfR system. Thus, the effect of nitrate loading (LNO3) and nitrite loading (LNO2) on perchlorate removal efficiency were assessed based on the simulation study using the validated model. Figure 6A and 6B presented the variation of perchlorate removal efficiency according to nitrate and nitrite loading rate under different perchlorate loading conditions.
As shown in Figure 6A, the increase of the LNO3 could result in a decrease of the perchlorate removal efficiency for all tested LClO4. Also, for a certain LNO3, the perchlorate removal efficiency declined with increasing LClO4. Specifically, when LClO4 was 0.08 gCl/m2/d, the perchlorate removal efficiency decreased from 80% to 20% when LNO3 increased from 0.5 gN/m2/d to 0.72 gN/m2/d. For the LClO4 of 0.08 gCl/m2/d, the same decrease of perchlorate removal efficiency was corresponding to the increase of LNO3 from 0.12 gN/m2/d to 0.60 gN/m2/d. The inhibition effect of nitrate loading on perchlorate removal has also been observed by previous studies. Nerenberg et al. [8] reported that an increase in nitrate loading from 0 to 1.2 g N/m2-d decreased perchlorate removal from 57% to 30% in a hydrogen-based membrane-biofilm reactor. Tang et al. [37] used biofilm modeling to investigate the impact of nitrate loading on perchlorate reduction with hydrogen as the electron donor and found nitrate loading would decrease perchlorate reduction by competing for the hydrogen. Choi and Silverstein also found that the reduction rate of perchlorate decreased by up to 70% in the presence of 28 mg/L of nitrate in a fixed biofilm reactor when the electron donor was limiting [38]. One possible reason for the inhibition on perchlorate reduction was the competition within PRB between nitrate 18
and perchlorate for the same resources including electrons and reductase enzymes [19]. PRB may prefer to utilize nitrate first as it has a higher anoxic reduction factor than perchlorate reduction factor (Table S4, SI). In addition, the existence of nitrate could promote the growth of HB, which had a faster growth rate than PRB [24]. As a result, the space in the biofilm for PRB could decrease and consequently limited perchlorate reduction activities.
Nitrite had a similar effect on perchlorate removal as nitrate. As show in Figure 6B, the perchlorate removal efficiency decreased with the increase of LNO2 and LClO4. The 80% of perchlorate removal efficiency could be achieved with LNO2 below 0.7 gN/m2/d when LClO4 was 0.08 gCl/m2/d. However, when LClO4 was 0.8 gCl/m2/d, even a LNO2 of 0.02 gN/m2/d could resulted in the perchlorate removal efficiency decreasing below 80%. The inhibition effect of LNO2 on perchlorate removal efficiency could also be caused by the competition within PRB between nitrite and perchlorate for the same resources as well as the competition between PRB and HB for spaces in the biofilm. In addition, the toxicity of nitrite could inhibit the activity of AMO [39], which might decrease the amount of electron donors for both nitrite and perchlorate reduction, and consequently reduced the perchlorate removal efficiency.
3.6. Practical implication Perchlorate in natural water bodies could pose potential biological hazards and its remediation was important. The feasibility of perchlorate reduction using the methane-based MBfR has been demonstrated by previous experimental investigation and is further proven by the developed model in this study. It is expected that the established model could provide supports for the further development of the 19
methane-based MBfR for efficient perchlorate remediation. However, this model is subjected to two key assumptions, i.e., aerobic methane oxidation would produce methanol, and the anaerobic methane oxidation coupled to perchlorate reduction would not occur. As a result, the model may not be able to describe the experimental observations of other products (e.g., acetate) formation during aerobic methane oxidation as well as anaerobic methane oxidation coupled to perchlorate reduction. These simplifications can be relaxed in the future as more information about methane oxidation mechanisms becomes available. While this model may not yet serve as a precise and quantitative predictor of methane oxidation in various systems with different cultures, it can nevertheless serve as tool to explore the effect of operational conditions on the studied methane-based MBfR for perchlorate reduction. The modelling results also clearly indicated that the perchlorate removal efficiency would be affected by several key operational parameters which need to be well designed/controlled during the practical operation in future.
Methane pressure and perchlorate loading are two key factors affecting the perchlorate removal efficiency in the methane-based MBfR. Higher methane pressure was preferable for perchlorate reduction with increasing perchlorate loadings. However, in practice, the exceeding supply of methane not only would be a waste of energy, but also could lead to the greenhouse gas emission from the system. Therefore, the proper selection of the methane pressure according to the perchlorate loading was crucial. The dash-dotted line in Figure 5B indicated that a high perchlorate removal efficiency (>80%) could be achieved with corresponding combinations of the methane pressure and the perchlorate loading rate, which could be served as a reference for controlling methane pressure and perchlorate loading during practical operation. In 20
addition, nitrate and nitrite in the water could pose inhibition effect on the perchlorate removal in the methane-based MBfR (Figure 6), as PRB preferred to reduce nitrate and nitrite firstly rather than perchlorate. Therefore, when the nitrate or nitrite loading was high, more electron donor would be required to maintain high perchlorate removal efficiency. During the practical application, the increase of electron donor could be achieved by enhancing methane pressure or through adding other organic substances such as methanol or ethanol [40]. Also, the presence of nitrate or nitrite could favor the growth of HB, which might occupy more space in the biofilm and resulted in less space for PRB to develop. As indicated in Figure 4A and 4B, HB was mainly situated in the outer layer of the biofilm while PRB were growth in the inner layer. Thus, a relatively thicker biofilm were preferred with the high nitrite or nitrate loadings, in which case more space could be left inside the biofilm for the development of PRB. In practice, the thickness of the biofilm could be increased by decreasing the hydrodynamic shears through slowing down the mixing speed or the water recirculation rate in the MBfR system. Furthermore, simulation study revealed that other factors, oxygen for instance, might also limit the perchlorate reduction in the methane-based MBfR. The effect of these factors on perchlorate removal efficiency is worthwhile to be investigated in further study.
4. Conclusion In this work, a multi-species biofilm model describing perchlorate reduction in a methane-based membrane biofilm reactor was developed for the first time. The model has a satisfied applicability and predictive abilities to reproduce the experimental data from the long-term operation perchlorate reduction MBfR at different operational stages (R2>0.9). The modeling results indicated that the aerobic methanotrophs and 21
perchlorate reduction bacteria were mainly located in the biofilm at the membrane side (~ 60%) while heterotrophic bacteria were mainly situated near the liquid side (~ 50%). Methane pressure (PCH4) and perchlorate loading (LClO4) were two key factors affecting perchlorate removal efficiency in the methane-based MBfR. The perchlorate removal efficiency of over 80% could be achieved by controlling the proper combination of PCH4 and LClO4 settings (e.g., applying a PCH4 of 30kPa at a LClO4 of 0.08 gCl/m2/d). In addition, nitrate and nitrite loading could both have inhibitory effect on perchlorate reduction, which need to be carefully considered and designed during the practical application.
Acknowledgements This work was supported by the National Natural Science Foundation of China (No. 51608374 and No. 51578391), the Recruitment Program of Global Experts, the Program for Young Excellent Talents in Tongji University and the Foundation of State Key Laboratory of Pollution Control and Resource Reuse (Tongji University), China (No. PCRRY15011).
Reference [1] P.K. Dasgupta, P.K. Martinelango, W.A. Jackson, T.A. Anderson, K. Tian, R.W. Tock, S. Rajagopalan, The Origin of Naturally Occurring Perchlorate: The Role of Atmospheric Processes, Environ. Sci. Technol. 39 (2005) 1569-1575. [2] L. Ye, H. You, J. Yao, H. Su, Water treatment technologies for perchlorate: A review, Desalination 298 (2012) 1-12. [3] E.T. Urbansky, Perchlorate Chemistry: Implications for Analysis and Remediation, Bioremediation Journal 2 (1998) 81-95. 22
[4] J.D. Coates, L.A. Achenbach, Microbial perchlorate reduction: rocket-fuelled metabolism, Nat Rev Micro 2 (2004) 569-580. [5] R. Srinivasan, G.A. Sorial, Treatment of perchlorate in drinking water: A critical review, Sep. Purif. Technol. 69 (2009) 7-21. [6] H.-P. Zhao, S. Van Ginkel, Y. Tang, D.-W. Kang, B. Rittmann, R. Krajmalnik-Brown, Interactions between Perchlorate and Nitrate Reductions in the Biofilm of a Hydrogen-Based Membrane Biofilm Reactor, Environ. Sci. Technol. 45 (2011) 10155-10162. [7] M. Gao, S. Wang, Y. Ren, C. Jin, Z. She, Y. Zhao, S. Yang, L. Guo, J. Zhang, Z. Li, Simultaneous removal of perchlorate and nitrate in a combined reactor of sulfur autotrophy and electrochemical hydrogen autotrophy, Chem. Eng. J. 284 (2016) 1008-1016. [8] R. Nerenberg, B.E. Rittmann, I. Najm, Perchlorate reduction in a hydrogen-based membrane-biofilm reactor, Journal (American Water Works Association) 94 (2002) 103-114. [9] Y.-H. Luo, R. Chen, L.-L. Wen, F. Meng, Y. Zhang, C.-Y. Lai, B.E. Rittmann, H.-P. Zhao, P. Zheng, Complete Perchlorate Reduction Using Methane as the Sole Electron Donor and Carbon Source, Environ. Sci. Technol. 49 (2015) 2341-2349. [10] P.J. Strong, S. Xie, W.P. Clarke, Methane as a Resource: Can the Methanotrophs Add Value?, Environ. Sci. Technol. 49 (2015) 4001-4018. [11] Y. Liu, Y. Zhang, B.-J. Ni, Zero valent iron simultaneously enhances methane production and sulfate reduction in anaerobic granular sludge reactors, Water Res. 75 (2015) 292-300. 23
[12] W. Wei, X. Zhou, D. Wang, J. Sun, Q. Wang, Free ammonia pre-treatment of secondary sludge significantly increases anaerobic methane production, Water Res. 118 (2017) 12-19. [13] J. Sun, S. Hu, K.R. Sharma, B.-J. Ni, Z. Yuan, Stratified Microbial Structure and Activity in Sulfide- and Methane-Producing Anaerobic Sewer Biofilms, Appl. Environ. Microb. 80 (2014) 7042-7052. [14] J. Sun, I. Pikaar, K.R. Sharma, J. Keller, Z. Yuan, Feasibility of sulfide control in sewers by reuse of iron rich drinking water treatment sludge, Water Res. 71 (2015) 150-159. [15] A. Silva-Teira, A. Sánchez, D. Buntner, L. Rodríguez-Hernández, J.M. Garrido, Removal of dissolved methane and nitrogen from anaerobically treated effluents at low temperature by MBR post-treatment, Chem. Eng. J. 326 (2017) 970-979. [16] H.-P. Zhao, A. Ontiveros-Valencia, Y. Tang, B.-O. Kim, S. VanGinkel, D. Friese, R. Overstreet, J. Smith, P. Evans, R. Krajmalnik-Brown, B. Rittmann, Removal of multiple electron acceptors by pilot-scale, two-stage membrane biofilm reactors, Water Res. 54 (2014) 115-122. [17] X. Chen, Y. Liu, L. Peng, B.-J. Ni, Perchlorate, nitrate, and sulfate reduction in hydrogen-based membrane biofilm reactor: Model-based evaluation, Chem. Eng. J. 316 (2017) 82-90. [18] O. Wanner, P. Reichert, Mathematical modeling of mixed-culture biofilms, Biotechnol. Bioeng. 49 (1996) 172-184. 24
[19] S.W.M. Kengen, G.B. Rikken, W.R. Hagen, C.G. van Ginkel, A.J.M. Stams, Purification
and
Characterization
of
(Per)Chlorate
Reductase
from
the
Chlorate-Respiring Strain GR-1, J. Bacteriol. 181 (1999) 6706-6711. [20]
R.
Nerenberg,
hydrogen-oxidizing,
Y.
Kawagoshi,
perchlorate-reducing
B.E.
Rittmann,
bacterium,
Water
Kinetics Res.
40
of
a
(2006)
3290-3296. [21] R. Chen, Y.-H. Luo, J.-X. Chen, Y. Zhang, L.-L. Wen, L.-D. Shi, Y. Tang, B.E. Rittmann, P. Zheng, H.-P. Zhao, Evolution of the microbial community of the biofilm in a methane-based membrane biofilm reactor reducing multiple electron acceptors, Environ. Sci. Pollut. R 23 (2016) 9540-9548. [22] C.S. Laspidou, B.E. Rittmann, Non-steady state modeling of extracellular polymeric substances, soluble microbial products, and active and inert biomass, Water Res. 36 (2002) 1983-1992. [23] Y. Liu, Y. Zhang, Z. Zhao, H.H. Ngo, W. Guo, J. Zhou, L. Peng, B.-J. Ni, A modeling approach to direct interspecies electron transfer process in anaerobic transformation of ethanol to methane, Environ. Sci. Pollut. R 24 (2017) 855-863. [24] Y. Tang, H. Zhao, A.K. Marcus, R. Krajmalnik-Brown, B.E. Rittmann, A Steady-State Biofilm Model for Simultaneous Reduction of Nitrate and Perchlorate, Part 1: Model Development and Numerical Solution, Environ. Sci. Technol. 46 (2012) 1598-1607. [25] O. Wanner, E. Morgenroth, Biofilm modeling with AQUASIM, Water Science & Technology 49 (2004) 137-144. 25
[26] J. Sun, X. Dai, Y. Liu, L. Peng, B.-J. Ni, Sulfide removal and sulfur production in a membrane aerated biofilm reactor: Model evaluation, Chem. Eng. J. 309 (2017) 454-462. [27] Z. Lewandowski, H. Beyenal, J. Myers, D. Stookey, The effect of detachment on biofilm structure and activity: the oscillating pattern of biofilm accumulation, Water Sci. Technol. 55 (2007) 429-436. [28] G. Koch, K. Egli, J. Van der Meer, H. Siegrist, Mathematical modeling of autotrophic denitrification in a nitrifying biofilm of a rotating biological contactor, Water Sci. Technol. 41 (2000) 191-198. [29] T. Ito, S. Okabe, H. Satoh, Y. Watanabe, Successional Development of Sulfate-Reducing Bacterial Populations and Their Activities in a Wastewater Biofilm Growing under Microaerophilic Conditions, Appl. Environ. Microbiol. 68 (2002) 1392-1402. [30] Y. Shi, S. Hu, J. Lou, P. Lu, J. Keller, Z. Yuan, Nitrogen Removal from Wastewater by Coupling Anammox and Methane-Dependent Denitrification in a Membrane Biofilm Reactor, Environ. Sci. Technol. 47 (2013) 11577-11583. [31] Z. Wang, S. Xia, X. Xu, C. Wang, Utilizing a one-dimensional multispecies model to simulate the nutrient reduction and biomass structure in two types of H2-based membrane-aeration biofilm reactors (H2-MBfR): model development and parametric analysis, Environ. Sci. Pollut. R 23 (2016) 3436-3449. [32] X. Chen, J. Guo, Y. Shi, S. Hu, Z. Yuan, B.-J. Ni, Modeling of Simultaneous Anaerobic Methane and Ammonium Oxidation in a Membrane Biofilm Reactor, 26
Environ. Sci. Technol. 48 (2014) 9540-9547. [33] G.-P. Sheng, H.-Q. Yu, X.-Y. Li, Extracellular polymeric substances (EPS) of microbial aggregates in biological wastewater treatment systems: A review, Biotechnol. Adv. 28 (2010) 882-894. [34] E. Sahinkaya, H. Hasar, A.H. Kaksonen, B.E. Rittmann, Performance of a Sulfide-Oxidizing,
Sulfur-Producing
Membrane
Biofilm
Reactor
Treating
Sulfide-Containing Bioreactor Effluent, Environ. Sci. Technol. 45 (2011) 4080-4087. [35] M. Long, C. Zhou, S. Xia, A. Guadiea, Concomitant Cr(VI) reduction and Cr(III) precipitation with nitrate in a methane/oxygen-based membrane biofilm reactor, Chem. Eng. J. 315 (2017) 58-66. [36] S. Xia, C. Wang, X. Xu, Y. Tang, Z. Wang, Z. Gu, Y. Zhou, Bioreduction of nitrate in a hydrogen-based membrane biofilm reactor using CO2 for pH control and as carbon source, Chem. Eng. J. 276 (2015) 59-64. [37] Y. Tang, H. Zhao, A.K. Marcus, R. Krajmalnik-Brown, B.E. Rittmann, A steady-state biofilm model for simultaneous reduction of nitrate and perchlorate, part 2: parameter optimization and results and discussion, Environ. Sci. Technol. 46 (2012) 1608-1615. [38] H. Choi, J. Silverstein, Inhibition of perchlorate reduction by nitrate in a fixed biofilm reactor, J. Hazard. Mater. 159 (2008) 440-445. [39] P. Dunfield, R. Knowles, Kinetics of inhibition of methane oxidation by nitrate, nitrite, and ammonium in a humisol, Appl. Environ. Microb. 61 (1995) 3129-3135. 27
[40] J. Sun, X. Dai, Q. Wang, Y. Pan, B.-J. Ni, Modelling Methane Production and Sulfate Reduction in Anaerobic Granular Sludge Reactor with Ethanol as Electron Donor, Scientific Reports 6 (2016) 35312.
28
Legends of Figures
Figure 1. Schematic representation of microbial processes occurred in the methane-based MBfR. Lines in the figure connect substances, including CH4, O2, ClO4-, NO2-, NO3-, extracellular polymeric substances (EPS), substrate-utilization -associated products (UAP), biomass-associated products (BAP), inert biomass (IB), which could be either consumed or produced by microorganisms involved in the system, i.e. aerobic methanotrophs (AMO), heterotrophic bacteria (HB), perchlorate reduction bacteria (PRB) and denitrifying anaerobic methane oxidation bacteria or archaea (DAMOb or DAMOa).
Figure 2. Comparison of effluent perchlorate concentration (A), perchlorate removal flux (B), effluent nitrate and nitrite concentration (C) and nitrate and nitrite removal flux (D) at Stage 1-4 from the experiment and from the model with optimized parameters.
Figure 3. Comparison of effluent perchlorate concentration (A), perchlorate removal flux (B), effluent nitrate and nitrite concentration (C) and nitrate and nitrite removal flux (D) at Stage 5-7 from the experimental measurements and from the model prediction
Figure 4.
The simulated distribution of solid components in the biofilm at Stage 5
(A) and Stage 6 (B). The simulated relative abundance of solid components in the biofilm at Stage 5 (C) and Stage 6 (D)
29
Figure 5. Model simulated perchlorate removal efficiency under different combinations of methane pressure and perchlorate loading in 3D (A) and 2D (B) plots. The efficiency (%) was represented by the color scale in all 2D plots. The dash-dotted line in Figure represented for perchlorate removal efficiency being up to 80%.
Figure 6. Model simulated perchlorate removal efficiency under different nitrate (A) and nitrite loading (B) in 2D plots. The efficiency (%) was represented by the color scale. The dotted line in the figure represented for perchlorate removal efficiency at 20%, 40%, 60% and 80%, respectively.
30
CH4
DAMOb NO2NO3-
AMO
EPS
O2
BAP
DAMOa
CH3OH UAP IB HB AMO related processes
NO2-
ClO4NO3
HB related processes PRB related processes
-
DAMOb related processes
PRB
DAMOa related processes Hydrolysis of EPS
Figure 1. Schematic representation of microbial processes occurred in the methane-based MBfR. Lines in the figure connect substances, including CH4, O2, ClO4-, NO2-, NO3-, extracellular polymeric substances (EPS), substrate-utilization -associated products (UAP), biomass-associated products (BAP), inert biomass (IB), which could be either consumed or produced by microorganisms involved in the system, i.e. aerobic methanotrophs (AMO), heterotrophic bacteria (HB), perchlorate reduction bacteria (PRB) and denitrifying anaerobic methane oxidation bacteria or archaea (DAMOb or DAMOa).
31
Figure 2. Comparison of effluent perchlorate concentration (A), perchlorate removal flux (B), effluent nitrate and nitrite concentration (C) and nitrate and nitrite removal flux (D) at Stage 1-4 from the experiment and from the model with optimized parameters.
32
Figure 3. Comparison of effluent perchlorate concentration (A), perchlorate removal flux (B), effluent nitrate and nitrite concentration (C) and nitrate and nitrite removal flux (D) at Stage 5-7 from the experimental measurements and from the model prediction
33
Figure 4.
The simulated distribution of solid components in the biofilm at Stage 5
(A) and Stage 6 (B). The simulated relative abundance of solid components in the biofilm at Stage 5 (C) and Stage 6 (D)
34
Figure 5. Model simulated perchlorate removal efficiency under different combinations of methane pressure and perchlorate loading in 3D (A) and 2D (B) plots. The efficiency (%) was represented by the color scale in all 2D plots. The dash-dotted line in Figure represented for perchlorate removal efficiency being up to 80%.
35
Figure 6. Model simulated perchlorate removal efficiency under different nitrate (A) and nitrite loading (B) in 2D plots. The efficiency (%) was represented by the color scale. The dotted line in the figure represented for perchlorate removal efficiency at 20%, 40%, 60% and 80%, respectively.
36
Graphical Abstract
B
A
DAMOb NO2
AMO
-
NO3EPS
O2
BAP
80 AMO
HB
DA
XI
PRB
DB
EPS
60
40
20
0
DAMOa
Solid component fraction (%)
CH4
Solid component fraction (%)
80
20
40
20
40
60
80
100
0
C
D
AMO related processes
NO3
HB related processes PRB related processes DAMOb related processes
PRB
DAMOa related processes Hydrolysis of EPS
37
40
Biofilm
HB
-
20
Biofilm thickness ( m)
UAP
ClO4-
X
60
CH3OH
NO2-
H
0 0
IB
AMO DA
AMO HB PRB DB XI EPS
Highlights
A model describing perchlorate reduction in a methane-based MBfR was developed
The model was verified by experimental data under different operational conditions
The stratified microbial distribution in the biofilm was reveal by model analysis
Over 80% perchlorate removal efficiency can be achieved with proper PCH4 and LClO4.
The perchlorate reduction could be affect by nitrate and nitrite concentrations
38
S1385-8947(17)31087-2 http://dx.doi.org/10.1016/j.cej.2017.06.136 CEJ 17217
To appear in:
Chemical Engineering Journal
Received Date: Revised Date: Accepted Date:
30 April 2017 23 June 2017 24 June 2017
Please cite this article as: J. Sun, X. Dai, L. Peng, Y. Liu, Q. Wang, B-J. Ni, A Biofilm Model for Assessing Perchlorate Reduction in a Methane-Based Membrane Biofilm Reactor, Chemical Engineering Journal (2017), doi: http://dx.doi.org/10.1016/j.cej.2017.06.136
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
A Biofilm Model for Assessing Perchlorate Reduction in a Methane-Based Membrane Biofilm Reactor
Jing Sun1, Xiaohu Dai1, Lai Peng2, Yiwen Liu3, Qilin Wang4, Bing-Jie Ni1,*
1
State Key Laboratory of Pollution Control and Resources Reuse, College of
Environmental Science and Engineering, Tongji University, Shanghai 200092, PR China 2
Laboratory of Microbial Ecology and Technology (LabMET), Ghent University, Coupure Links 653, 9000 Ghent, Belgium
3
Centre for Technology in Water and Wastewater, School of Civil and Environmental Engineering, University of Technology Sydney, Sydney, NSW 2007, Australia 4
Griffith School of Engineering, Griffith University, Nathan Campus, QLD 4111, Australia
*Corresponding author: Bing-Jie Ni, P: +86 21 65986849; F: +86 21 65983602; E-mail: [email protected]
Abstract Perchlorate (ClO4-) is recognized as an important contaminant in surface water and groundwater, which would pose health risks at very low concentrations. A methane-based
membrane
biofilm
reactor
(MBfR)
has
been
successfully
demonstrated for perchlorate reduction, which provided an alternative solution for perchlorate remediation with low cost. In this work, a multispecies biofilm model was developed to evaluate perchlorate reduction in the methane-based MBfR under 1
different operational conditions. The model was calibrated and validated using the experimental data from the long-term operation of the MBfR at seven distinct stages. The results suggested that the developed model could satisfactorily describe perchlorate reduction and denitrification performances in the MBfR (R2>0.9). The modeling results provided insight into the microbial community distribution in the biofilm, with aerobic methanotrophs and perchlorate reduction bacteria being mainly located at the membrane side (~ 60%) and heterotrophic bacteria being situated near the liquid side (~ 50%). The model simulations indicated that over 80% of perchlorate removal efficiency could be achieved through controlling the optimal combinations of methane pressure (PCH4) and perchlorate loading (LClO4) (e.g., applying a PCH4 of 30kPa at a LClO4 of 0.08 gCl/m2/d). In addition, the perchlorate reduction would be inhibited by the presence of nitrate and nitrite in the MBfR, which should be appropriately controlled during the future practical application of the promising process.
Keywords: membrane biofilm reactor; perchlorate removal; methane; biofilm model; practical application
1. Introduction Perchlorate salts, including ammonium, potassium and sodium perchlorate have been widely used in rocket fuels, pyrotechnics, matches, munitions and many other industries [1, 2]. These perchlorate salts usually have high solubilities and would be readily dissociated in water producing perchlorate anion, which is relatively nonreactive and stable, but extremely mobile in aqueous systems [3]. Perchlorate (ClO4-) is recognized as an important contaminant in surface water and groundwater, 2
as it would pose health risks at very low concentrations and is difficult to remediate from the water. It can interfere with the body’s iodine intake, consequently inhibiting thyroid hormone production, which is thought essential for growth and development [4]. Therefore, the U.S. Environmental Protection Agency is in the process of evaluating a maximum contaminant level (MCL) for perchlorate in the range of 1−18 μg/L [5].
Many previous studies have been carried out to investigate the remediation of perchlorate in water through microbiological reduction [6-8]. A recent study, for instance, reported a successful example of microbial perchlorate removal in a methane-based membrane biofilm reactor (MBfR) [9]. In this study, methane was supplied into lumens of hollow-fiber membranes and further permeated through the non-porous walls into biofilms attached on outer surfaces of membranes. Meanwhile, perchlorate in wastewater was diffused into biofilms, where perchlorate reduction bacteria (PRB) developed and the microbial perchlorate reduction took place. The result showed that the biofilms could reduce up to 5 mg/L ClO4- to a non-detectable level when methane delivery was not limiting. While, under the methane limiting condition, the removal of perchlorate could be interfered by the presence of nitrate and nitrite in water.
As methane is an inexpensive and widely available carbon source, its utilization for perchlorate remediation could largely reduce the cost, compared with other methods such as chemical reduction and iron exchange [2, 10]. Also, it proposed a promising solution for energy recovery from the wastewater system, because methane is a key product of the anaerobic digestion of biosolids [11, 12]. Although, the utilization of 3
methane has potential risks due to its relatively low explosion limit [13], the application of the “bubble-less” membranes could enhance the operational safety by preventing methane losses to the atmosphere. In addition, as methane is a potent greenhouse gas, using methane for perchlorate reduction could potentially alleviate the greenhouse gas emission from the wastewater treatment system [14, 15]. Therefore, the application of methane-based MBfR for perchlorate remediation has a good prospect of practical application. In this regard, a mathematical model for describing perchlorate reduction in the methane-based MBfR is highly desirable to facilitate its full-scale application. The comprehensive investigation into this promising system, especially into the effect of different operational parameters on perchlorate removal, is also required to potentially improve the system performance.
The experimental investigation implied that the microbial perchlorate removal in the methane-based MBfR system could be affected by three key operational parameters, i.e. the methane pressures, perchlorate loadings and nitrate or nitrite loadings [9]. These parameters governed the perchlorate removal not only by directly affecting the perchlorate reduction rate, but also by influencing the development of PRB in the biofilm. In the biofilm, the possible competition and promotion mechanisms mainly include: i) The competition between PRB and heterotrophic bacteria (HB) for the same electron donor [16, 17]; ii) The competition between different microorganisms and associated products for space in the biofilm [18]; iii) The competition within PRB between nitrate/nitrite and perchlorate for the same resources such as electrons and possibly reductase enzymes[19]; and iv) the promotion of the growth of PRB through their utilization of nitrate or nitrite [20]. Because of multiple mechanisms, microbial species, and substrates in the methane-based MBfR system, the links between these 4
operating parameters and perchlorate removal are not straightforward. In this case, multispecies biofilm modeling is advantageous for quantitatively integrating the microbiological and physical phenomena that control the perchlorate removal in the methane-based MBfR.
This study aimed to develop a multispecies biofilm model to predict the perchlorate removal in the methane-based MBfR system under different operational conditions. The developed model comprehensively considered the interaction between different microbial processes in the biofilm as well as the gas delivery characteristics through the membrane substratum in the reactor. The model was calibrated and validated using the experimental data from a long-term operated methane-based MBfR at different operational stages. The effect of key operation parameters on the perchlorate removal was then investigated using the validated model. It is expected that the established model could provide supports for the further development of the methane-based MBfR for efficient perchlorate remediation.
2. Material and methods 2.1. Development of the biological processes model The key biological processes occurred in the biofilm of the MBfR was considered based on the experimental observations [21], which were summarized in Figure 1. In the biofilm, the methane could be aerobically oxidized by methanotrophs (AMO) with oxygen in the influent. The oxidation product, methanol, was then severed as the electron donor for perchlorate reduction and heterotrophic denitrification with the respiration of PRB and HB. On the other hand, methane could also be anaerobically oxidized by denitrifying anaerobic methane oxidation (DAMO) bacteria or archaea 5
using nitrate and nitrite as electron acceptor in the biofilm where oxygen could not be penetrated to. The anaerobic methane oxidation coupled to perchlorate reduction was not included in this work, as so far there still no direct evidence to prove the occurrence of such process in the system. Through aforementioned oxidation or reduction processes, the microorganisms gained energy to synthesis new biomass, and meanwhile
to
produce
extracellular
polymeric
substances
(EPS)
and
substrate-utilization-associated products (UAP) [22]. The formed EPS could be hydrolyzed into biomass-associated products (BAP). The BAP and UAP could be both used by HB as organic electron donors with nitrate or nitrite as their electron acceptors. All microorganisms involved in the system are subject to endogenous respiration, with the production of inert biomass (IB) and the reduction of the electron acceptors to gain energy for cell maintenance.
According to the biological processes described above, the developed biological model includes seven particulate species, i.e. AMO (XAMO), HB (XHB), PRB (XPRB), DAMO bacteria (XDB), DAMO archaea (XDA), EPS (XEPS) and inert biomass (XI), and eight soluble species, i.e. methane (SCH4), oxygen (SO2), perchlorate (SClO4), nitrate (SNO3), nitrite (SNO2), methanol (SCH3OH), UAP (SUAP), and BAP (SBAP), as listed in Table S1. The unit was gN/m3 for all nitrogen species, gCl/m3 for perchlorate and gO/m3 for oxygen, while concentrations of other compounds were quantified based their chemical oxygen demand (COD) equivalent, i.e. gCOD/m3. The stoichiometric matrix of the model was summarized in Table S2 (SI). The rates of the growth and endogenous respiration of microorganisms were modelled using Monod-type kinetics while hydrolysis of the EPS was simulated through first-order kinetics [22, 23]. Competition for the same resources in PRB is expressed by 6
competitive-inhibition coefficients in the acceptor part of dual-substrate Monod kinetics [24]. The kinetic expressions of the reactions included in the model were listed in Table S3 (SI). The definitions of parameters in the kinetic expressions were shown in Table S4 (SI).
2.2. Methane-based membrane biofilm reactor model A multispecies one-dimensional biofilm model was constructed to model the methane-based membrane biofilm reactor through employing the software AQUASIM 2.1d [25]. The one-dimensional conservation law is formulated as an overall balance equation of between the mass conserved and utilized in this model [18]. The methane-based MBfR system was modeled with two different compartments in AQUASIM: the membrane lumen was modelled with a completely mixed gas compartment and the reactor bulk liquid and the biofilm was modelled through a biofilm compartment. The gas compartment was connected to the base of the biofilm via a diffusive link. The flux of methane (LCH4) from the gas to the biofilm matrix compartment through the membrane was modeled using the following equation (Equation 1) according to Sun et al. [26] (1) where
and
are concentrations of oxygen in the gas and biofilm matrix
compartments (gCOD/m3), respectively; of oxygen (m/day); and
is the overall mass transfer coefficient
is the Henry’s coefficient for oxygen (mole m-3
gas/mole m-3 liquid).
In the biofilm compartment, the biofilm structure was represented as a continuum without considering diffusive mass transport of biomass. The steady-state biofilm 7
thickness was established by controlling the detachment using a global detachment velocity, i.e. ude, (μm/d), in model simulations [26]. The composition of solids detached from the biofilm conformed to their composition at the biofilm surface. The detached particulates were assumed to be washed out of the system with the effluent, and no reattachment of detached particulates was considered in the model. The water fraction of the biofilm matrix was kept constant at 0.6, which was in the range of most probable water fractions in the wastewater biofilm [27]. The biomass density was kept at 100000 gCOD/m3 [28].
2.3. Experimental data used for testing the developed model The experiment data from a methane-based MBfR reported by Luo et al.[9] was used to calibrate and validated the developed model. Briefly, the MBfR had a total volume of 65ml and comprised hollow fiber gas-transfer membranes (Model MHF-200TL, Mitsubishi, Ltd., Japan), which resulted in a total membrane surface of 7.0 cm2. The fibers were glued into a gas-supply manifold at the bottom of the MBfR, with the top of each fiber being sealed. The simplified graphical scheme of the experimental methane-based membrane biofilm reactor was presented in Figure S1 (SI). Methane was delivered into lumens of the membranes and the bulk liquid in reactor was completely mixed by recirculation with a peristaltic pump at 100 mL/min. The MBfR was inoculated with 10 mL of ANMO-D culture and was fed with 2 mg N/L of NO2for 40 days to accumulate enough biomass. Then the MBfR was operated under 7 stages to investigate ClO4- reduction in the presence of NO3- and NO2-. The operation conditions of all stages are summarized in Table S5 (SI). The influent flow rate was 0.5 mL/min and the CH4 pressure was 10 psi (69 kPa) for stages 1−3 and 15 psi (103 kPa) for stages 4-7. The dissolved oxygen concentration was ~0.2 mg/L for the 8
influent and ≤0.1 mg/L for the effluent. The pH was controlled at 7.0 ± 0.2 with hydrochloric acid. For each stage, the MBfR was operated to reach the steady state, which was determined as effluent concentrations stable (<10% variation) for a minimum of three hydraulic retention time (HRTs). The effluent concentrations of ClO4- NO3- and NO2- were measured using ion chromatography. More details could be found in Luo et al.[9]. The removal flux of ClO4- , NO3- and NO2- was calculated according to the following equation: (2) where S° and S are the influent and effluent ClO4- , NO3- or NO2- concentration (g/L), Q is the influent flow rate to the MBfR system (L/d), and A is the membrane surface area (m2).
2.4. Model calibration and validation The developed biofilm model includes 27 species-specific biochemical processes and 58 kinetic and stoichiometric parameters (Tables S2 and S3 (SI)). The values of most of these parameters were well established in previous studies and thus were adapted from literature, as presented in Table S4 (SI). However, limited information was available in literature for the parameters related to XAMO and XPRB. Therefore, parameter estimation based on experimental measurements was conducted for three parameters of XSOB and three parameters of XPRB, which are the maximum growth rate of XAMO (μXAMO), the half-saturated concentration of oxygen for XAMO (KXAMO,O2) , the inhibition constant of NO2- on XAMO (KIXAMO,NO2), the maximum growth rate of XPRB (μXPRB), the inhibition constant of NO2- on XPRB (KIXPRB,NO2) and perchlorate reduction factor on the maximum growth of XPRB (ηClO4). The experimental data obtained in the operational stage of 1-4 were used for model 9
calibration and parameters estimation, which was conducted by minimizing the sum of squares of the deviations between the measured data and the model predictions. The objective function to be minimized in the parameter estimation is as follows [25]: (3) where
is the measured data at time ti (i from 1 to n).
is the calculated
value by the model at time ti (i from 1 to n). p is the parameters of μXAMO, KXAMO,O2, KIXAMO,NO2, μXPRB, KIXPRB,NO2 and ηClO4.
is the standard deviation of the
measurements. With the built-in simplex and secant algorithms, at each iteration, all parameter arrays were replaced by new values until
are close enough to fulfill
the convergence criterion.
Model validation was then conducted with calibrated parameters using the experimental data during the stage 5-7 that were operated under completely different experimental conditions compared to the calibration experiments (e.g., different methane pressures and substrates loadings). The software AQUASIM 2.1d was employed for both calibrating the model parameters and simulating the system [25].
2.5. Model-based analysis of microbial community in the biofilm The validated MBfR model was then used to investigate the microbial community distribution in the methane-based perchlorate-reduction biofilm to provide insight into the key microbial interactions in the MBfR under different conditions. The model was simulated with different operational conditions as described in Section 2.2 until reaching the steady states. The fractions of different microorganisms along the depth of the biofilm under different operational conditions were calculated and compared. 10
Then the microbial distribution pattern in the biofilm could be revealed together with its variation with the different operational conditions. In addition, the relative abundances of different microorganisms in the biofilm were also calculated to provide information on the overall microbial community in the biofilm.
2.6. Impact of key operational parameters on perchlorate removal efficiency The simulation study was conducted with the validated model to investigate the effect of key operational parameters, including the methane pressures, perchlorate loadings and nitrate or nitrite loadings on perchlorate removal in the methane-based MBfR. The simulated MBfR system had a total water volume of 5L and a membrane surface area of 0.5 m2, which resulting in the area to volume (A/V) ratio of 100 m-1, a typical A/V ratio in the MBfR system. To evaluate the effect of methane pressures (PCH4) and perchlorate loading (LClO4), the comprehensive model simulations were conducted with different combinations of PCH4 and LClO4, covering a range of 20-120 kPa for PCH4 and 0.08-0.8 gCl/m2/d for L ClO4. The settings of other operational parameters were listed in Table S6 (SI), Scenario I. The perchlorate removal efficiency was calculated and compared under different conditions (SI Table S6), i.e., the different combinations of PCH4 (20-120 kPa) and LClO4 (0.08-0.8 gCl/m2/d). Similarly, the perchlorate removal efficiency was also calculated and compared to evaluate the effect of nitrate or nitrite loadings (LNO2 or LNO3) on perchlorate removal. The model simulation was conducted with different LNO2 or LNO3, covering a range of 0-0.8 gN/m2/d. Also, the LClO4 was varying from 0.08-0.8 gCl/m2/d. The settings of other operational parameters were listed in Table S6 (SI), Scenario II and Scenario III.
3. Results and discussion 11
3.1. Model calibration The effluent concentrations and the removal fluxes of perchlorate, nitrate and nitrite at the steady state during the operational stage 1-4 were used for the model calibration (SI Table S7). Three parameters of XAMO (μXAMO, KXAMO,O2, KIXAMO,NO2) and three parameters of XPRB (μXPRB, KIXPRB,NO2, ηClO4) were estimated by fitting simulation results to monitored data. The calibrated parameter values giving the optimum model fittings were listed in Table S4 (SI) and the model predictions and experiment measurements of the reactor performance are compared in Figure 2 (A-D).
Figure 2A and 2C presents the effluent concentrations of perchlorate, nitrate and nitrite obtained from both experimental study and model prediction. At Stage 1, the perchlorate removal efficiency was 23% with the presence of nitrite in the influent. It recovered to over 90% at Stage 2 when there was no nitrate or nitrite was in the influent. The high perchlorate removal efficiency (98%) was also observed during Stage 3, while the perchlorate reduction was totally inhibited when the influent nitrate concentration reached 11.3 mgN/L at Stage 4. The comparison in these figures suggested that the developed model could well capture the perchlorate reduction described above. Also, the model can satisfactorily described nitrite or nitrate removal in the system as shown in Figure 2C. Both the experimental results and model predictions suggested that most of the nitrite and nitrate could be removed at Stage 1 and 3, while the nitrate removal decreased to 69% at Stage 4.
The removal fluxes of perchlorate, nitrate and nitrite obtained from experimental study and model simulation was compared in Figure 2B and 2D. The maximum removal flux of perchlorate was reached at Stage 3 (0.07 g/m2/d), while the minimum 12
flux was only 0.0002 at Stage 4.The maximum removal flux of nitrate was reached at 0.56 gN/m2/d at Stage 4 and the removal flux of nitrite was 0.12 gN/m2/d at Stage 1. Figure 2B and 2D indicated that the model simulation also well described the removal fluxes showed in the experimental study. Overall, the well match between the model predictions and the experimental results (R2>0.9) as suggested in Figure 2 (A-D) supported the ability of the developed model for predicting perchlorate reduction in the methane-based MBfR system.
3.2. Model validation Model and parameters validation were conducted based on the comparison between the model predictions and the experimental measurements at Stage 5-7 (SI Table S7). The operational conditions at Stage 5-7 were significantly different from that at Stage 1-4, of which the data was used for model calibration. The methane pressure at Stage 5-7 was 1.5 times higher than that at Stage 1-3 and the influent perchlorate concentration at Stage 6-7 was over 5 times higher than that at Stage 1-4. The influent nitrite concentration at Stage 7 was over 3 times higher than that at the stage 1, while the influent nitrate concentration at Stage 5 was half of that at Stage 4.
The model and its parameters were firstly validated by the effluent perchlorate concentration and its corresponding removal flux at Stage 5-7. The comparison of the model predictions and experimental measurements as shown in Figure 3A and 3B suggested that the developed model could well describe the perchlorate removal as determined by the experimental study. The effluent perchlorate concentration at Stage 5 and 6 were negligible as indicated by both experimental and modelling results. The corresponding removal flux were 1.0 g/m2/d and 5.9 g/m2/d, respectively While at the 13
Stage 7, about 50% of perchlorate in the influent was removed, resulting in the effluent concentration of 2.5mg/L. The corresponding removal flux was 2.9 g/m2/d.
The effluent nitrite and nitrate concentrations and their corresponding removal flux determined by the experimental measurements were further used to validate the developed model. The model predictions were compared with experimental results in Figure 3C and 3D. Both the experimental and modelling results show that most nitrite and nitrate were removed at Stage 5 and Stage 7. The corresponding removal flux were 0.3 gN/m2/d for nitrite at Stage 5 and 0.4 gN/m2/d for nitrate at Stage 7, respectively. To assess the the accuracy of the developed model, we have conducted root mean square error (RMSE) analysis for Figures 2-3 and compare each RMSE to the corresponding maximum measurement. The results indicated that all of the RMSE are within 20% of maximum measurement value, indicating that the model prediction results generally match the experimental data. In sum, the good agreement between model simulations and the measured results (R2>0.9) confirmed the validity of the proposed model to describe the reactor performance related to perchlorate reduction and denitrification in the methane-based MBfR system.
3.3. Microbial community distribution in the biofilm Both model simulations and experimental results suggested that changes of substrate concentrations in the influent posed significant impact on perchlorate removal in the methane-based MBfR. This could be related to the evolution of microbial communities in the biofilm under different substrate conditions [29]. Therefore, simulation study was carried out to investigate the distribution of the microbial community in the biofilm in MBfR and its variation according to substrate 14
concentrations using the validated model. The distribution of solid components and their relative abundance in the biofilm on two operational stages, i.e. Stage 5 and Stage 6, were simulated and compared in Figure 4 (A-D).
As shown in Figure 4A and 4B, AMO was mainly located in the inner layer of the biofilm closed to the membrane, as methane was supplied through the membrane and consequently methane concentration was higher in the membrane side than the liquid side [30]. On the other hand, HB was mainly situated near the liquid side of the biofilm. This could be due to that the substrates for HB, such as oxygen, nitrite or nitrate was more available in the liquid side as they were diffused from the liquid to the biofilm. PRB showed a higher abundance near the membrane side in the biofilm, though it also required electron acceptors such as perchlorate, nitrite and nitrate penetrating from the liquid side. Thus, the relative low abundance of PRB at the liquid side possibly resulted from the competition between HB and PRB for the same substrates [31]. Since HB had a higher growth rate than the PRB (Table S4, SI), it is likely to outcompete PRB at the liquid side of the biofilm. Thus, most of PRB was developed in the inner side of the biofilm, where the penetration of substrate for HB was limited. As for the DAMO groups, DAMO bacteria accounted for about 5% of the solid components in the inner layer of the biofilm, while DAMO archaea was negligible. The low abundance of DAMO groups in the biofilm could be due to their relatively slow growth rate, compared with other microorganisms [32]. The inert biofilm presented a higher proportion in the inner biofilm while the EPS had a higher fraction near the biofilm surface. The EPS near the biofilm surface could enhance the stability of the biofilm by protecting microorganisms from erosion caused by hydrodynamic and mechanical shears [33]. 15
The changing of operational conditions from Stage 5 to Stage 6 would not change the microbial distribution pattern in the biofilm as shown in Figure 4A and 4B. However, the relative abundance of certain microorganisms could be varied according to substrates concentration. As presented in Figure 4C and 4D, the relative abundance of PRB increased from 6% at the Stage 5 to 18% at Stage 6. This could be the result of the increase of perchlorate concentration in the influent. On the contrary, the relative abundance of HB at Stage 6 was lower than that at Stage 5, as the influent concentration of nitrate decreased. The variation of microbial community in the biofilm is in good agreement with the change of the biofilm activities as suggested in the experimental study [9].
3.4. Effect of methane pressure and perchlorate loading on perchlorate removal The experimental study and modelling results both revealed that methane pressure (PCH4) and perchlorate loading (LClO4) were two key factors affecting the perchlorate removal in the methane-based MBfR. Therefore, simulation study using the validated model was carried out to investigate the combining effect of methane pressure and perchlorate loading on perchlorate removal efficiency. The tested PCH4 and LClO4 covered a wide range in order to potentially offer different options for real application [34]. Figure 5A and 5B illustrated the perchlorate removal efficiency under the extensive simulation conditions.
Overall, the perchlorate removal efficiency increased with the increase of PCH4 while decreased with the increase of LClO4. This is consistent with the findings reported in Long et al. [35], in which the Cr(VI) reduction increased by 40.3% and the 16
nitrate reduction increased by 30.2% when the methane pressure increased from 0.02 to 0.03 MPa in a methane-based membrane biofilm reactor. A biofilm model was developed to evaluate the key mechanisms including microbially-mediated perchlorate and nitrate reduction in the hydrogen-based membrane biofilm reactor and similarly the hydrogen loading was revealed to be properly managed at certain critical level to maximize the perchlorate and nitrate removal [17]. The perchlorate removal efficiency of up to 80% could be achieved when the combination of PCH4 and LClO4 was located in the region above the dash-dot line as indicated in Figure 5B. Specifically, when the applied LClO4 was 0.08 gCl/m2/d, a PCH4 of 30kPa was high enough to remove 80% of perchlorate under the simulated condition, whereas when the applied LClO4 was 0.64 gCl/m2/d, a PCH4 of 120 kPa was required to achieve the similar perchlorate removal efficiency. When LClO4 was above 0.64 gCl/m2/d, the increase of PCH4 from 100 kPa to 120 kPa did not resulted in significant improvement in the perchlorate removal and the perchlorate removal efficiency remained lower than 80%. This indicated that, apart from methane, other factors might also limit the perchlorate reduction in the methane-based MBfR, especially when LClO4 was relatively high. One possible factor could be the oxygen concentration in the reactor. In the MBfR system, methane could be aerobically oxidized by AMO, with its oxidation product, i.e. methanol, being used as electron donor for perchlorate reduction [10, 21]. If the oxygen in the system was depleted, the production of methanol would be inhibited. Consequently, the perchlorate reduction would be restricted due to the lack of electron donors.
3.5. Effect of nitrate and nitrite loading on perchlorate removal Nitrate and nitrite are two contaminants usually co-existed with the perchlorate in the 17
natural water bodies such as groundwater [9, 36]. The experimental investigation suggested that both nitrate and nitrite could affect the perchlorate removal in the methane-based MBfR system. Thus, the effect of nitrate loading (LNO3) and nitrite loading (LNO2) on perchlorate removal efficiency were assessed based on the simulation study using the validated model. Figure 6A and 6B presented the variation of perchlorate removal efficiency according to nitrate and nitrite loading rate under different perchlorate loading conditions.
As shown in Figure 6A, the increase of the LNO3 could result in a decrease of the perchlorate removal efficiency for all tested LClO4. Also, for a certain LNO3, the perchlorate removal efficiency declined with increasing LClO4. Specifically, when LClO4 was 0.08 gCl/m2/d, the perchlorate removal efficiency decreased from 80% to 20% when LNO3 increased from 0.5 gN/m2/d to 0.72 gN/m2/d. For the LClO4 of 0.08 gCl/m2/d, the same decrease of perchlorate removal efficiency was corresponding to the increase of LNO3 from 0.12 gN/m2/d to 0.60 gN/m2/d. The inhibition effect of nitrate loading on perchlorate removal has also been observed by previous studies. Nerenberg et al. [8] reported that an increase in nitrate loading from 0 to 1.2 g N/m2-d decreased perchlorate removal from 57% to 30% in a hydrogen-based membrane-biofilm reactor. Tang et al. [37] used biofilm modeling to investigate the impact of nitrate loading on perchlorate reduction with hydrogen as the electron donor and found nitrate loading would decrease perchlorate reduction by competing for the hydrogen. Choi and Silverstein also found that the reduction rate of perchlorate decreased by up to 70% in the presence of 28 mg/L of nitrate in a fixed biofilm reactor when the electron donor was limiting [38]. One possible reason for the inhibition on perchlorate reduction was the competition within PRB between nitrate 18
and perchlorate for the same resources including electrons and reductase enzymes [19]. PRB may prefer to utilize nitrate first as it has a higher anoxic reduction factor than perchlorate reduction factor (Table S4, SI). In addition, the existence of nitrate could promote the growth of HB, which had a faster growth rate than PRB [24]. As a result, the space in the biofilm for PRB could decrease and consequently limited perchlorate reduction activities.
Nitrite had a similar effect on perchlorate removal as nitrate. As show in Figure 6B, the perchlorate removal efficiency decreased with the increase of LNO2 and LClO4. The 80% of perchlorate removal efficiency could be achieved with LNO2 below 0.7 gN/m2/d when LClO4 was 0.08 gCl/m2/d. However, when LClO4 was 0.8 gCl/m2/d, even a LNO2 of 0.02 gN/m2/d could resulted in the perchlorate removal efficiency decreasing below 80%. The inhibition effect of LNO2 on perchlorate removal efficiency could also be caused by the competition within PRB between nitrite and perchlorate for the same resources as well as the competition between PRB and HB for spaces in the biofilm. In addition, the toxicity of nitrite could inhibit the activity of AMO [39], which might decrease the amount of electron donors for both nitrite and perchlorate reduction, and consequently reduced the perchlorate removal efficiency.
3.6. Practical implication Perchlorate in natural water bodies could pose potential biological hazards and its remediation was important. The feasibility of perchlorate reduction using the methane-based MBfR has been demonstrated by previous experimental investigation and is further proven by the developed model in this study. It is expected that the established model could provide supports for the further development of the 19
methane-based MBfR for efficient perchlorate remediation. However, this model is subjected to two key assumptions, i.e., aerobic methane oxidation would produce methanol, and the anaerobic methane oxidation coupled to perchlorate reduction would not occur. As a result, the model may not be able to describe the experimental observations of other products (e.g., acetate) formation during aerobic methane oxidation as well as anaerobic methane oxidation coupled to perchlorate reduction. These simplifications can be relaxed in the future as more information about methane oxidation mechanisms becomes available. While this model may not yet serve as a precise and quantitative predictor of methane oxidation in various systems with different cultures, it can nevertheless serve as tool to explore the effect of operational conditions on the studied methane-based MBfR for perchlorate reduction. The modelling results also clearly indicated that the perchlorate removal efficiency would be affected by several key operational parameters which need to be well designed/controlled during the practical operation in future.
Methane pressure and perchlorate loading are two key factors affecting the perchlorate removal efficiency in the methane-based MBfR. Higher methane pressure was preferable for perchlorate reduction with increasing perchlorate loadings. However, in practice, the exceeding supply of methane not only would be a waste of energy, but also could lead to the greenhouse gas emission from the system. Therefore, the proper selection of the methane pressure according to the perchlorate loading was crucial. The dash-dotted line in Figure 5B indicated that a high perchlorate removal efficiency (>80%) could be achieved with corresponding combinations of the methane pressure and the perchlorate loading rate, which could be served as a reference for controlling methane pressure and perchlorate loading during practical operation. In 20
addition, nitrate and nitrite in the water could pose inhibition effect on the perchlorate removal in the methane-based MBfR (Figure 6), as PRB preferred to reduce nitrate and nitrite firstly rather than perchlorate. Therefore, when the nitrate or nitrite loading was high, more electron donor would be required to maintain high perchlorate removal efficiency. During the practical application, the increase of electron donor could be achieved by enhancing methane pressure or through adding other organic substances such as methanol or ethanol [40]. Also, the presence of nitrate or nitrite could favor the growth of HB, which might occupy more space in the biofilm and resulted in less space for PRB to develop. As indicated in Figure 4A and 4B, HB was mainly situated in the outer layer of the biofilm while PRB were growth in the inner layer. Thus, a relatively thicker biofilm were preferred with the high nitrite or nitrate loadings, in which case more space could be left inside the biofilm for the development of PRB. In practice, the thickness of the biofilm could be increased by decreasing the hydrodynamic shears through slowing down the mixing speed or the water recirculation rate in the MBfR system. Furthermore, simulation study revealed that other factors, oxygen for instance, might also limit the perchlorate reduction in the methane-based MBfR. The effect of these factors on perchlorate removal efficiency is worthwhile to be investigated in further study.
4. Conclusion In this work, a multi-species biofilm model describing perchlorate reduction in a methane-based membrane biofilm reactor was developed for the first time. The model has a satisfied applicability and predictive abilities to reproduce the experimental data from the long-term operation perchlorate reduction MBfR at different operational stages (R2>0.9). The modeling results indicated that the aerobic methanotrophs and 21
perchlorate reduction bacteria were mainly located in the biofilm at the membrane side (~ 60%) while heterotrophic bacteria were mainly situated near the liquid side (~ 50%). Methane pressure (PCH4) and perchlorate loading (LClO4) were two key factors affecting perchlorate removal efficiency in the methane-based MBfR. The perchlorate removal efficiency of over 80% could be achieved by controlling the proper combination of PCH4 and LClO4 settings (e.g., applying a PCH4 of 30kPa at a LClO4 of 0.08 gCl/m2/d). In addition, nitrate and nitrite loading could both have inhibitory effect on perchlorate reduction, which need to be carefully considered and designed during the practical application.
Acknowledgements This work was supported by the National Natural Science Foundation of China (No. 51608374 and No. 51578391), the Recruitment Program of Global Experts, the Program for Young Excellent Talents in Tongji University and the Foundation of State Key Laboratory of Pollution Control and Resource Reuse (Tongji University), China (No. PCRRY15011).
Reference [1] P.K. Dasgupta, P.K. Martinelango, W.A. Jackson, T.A. Anderson, K. Tian, R.W. Tock, S. Rajagopalan, The Origin of Naturally Occurring Perchlorate: The Role of Atmospheric Processes, Environ. Sci. Technol. 39 (2005) 1569-1575. [2] L. Ye, H. You, J. Yao, H. Su, Water treatment technologies for perchlorate: A review, Desalination 298 (2012) 1-12. [3] E.T. Urbansky, Perchlorate Chemistry: Implications for Analysis and Remediation, Bioremediation Journal 2 (1998) 81-95. 22
[4] J.D. Coates, L.A. Achenbach, Microbial perchlorate reduction: rocket-fuelled metabolism, Nat Rev Micro 2 (2004) 569-580. [5] R. Srinivasan, G.A. Sorial, Treatment of perchlorate in drinking water: A critical review, Sep. Purif. Technol. 69 (2009) 7-21. [6] H.-P. Zhao, S. Van Ginkel, Y. Tang, D.-W. Kang, B. Rittmann, R. Krajmalnik-Brown, Interactions between Perchlorate and Nitrate Reductions in the Biofilm of a Hydrogen-Based Membrane Biofilm Reactor, Environ. Sci. Technol. 45 (2011) 10155-10162. [7] M. Gao, S. Wang, Y. Ren, C. Jin, Z. She, Y. Zhao, S. Yang, L. Guo, J. Zhang, Z. Li, Simultaneous removal of perchlorate and nitrate in a combined reactor of sulfur autotrophy and electrochemical hydrogen autotrophy, Chem. Eng. J. 284 (2016) 1008-1016. [8] R. Nerenberg, B.E. Rittmann, I. Najm, Perchlorate reduction in a hydrogen-based membrane-biofilm reactor, Journal (American Water Works Association) 94 (2002) 103-114. [9] Y.-H. Luo, R. Chen, L.-L. Wen, F. Meng, Y. Zhang, C.-Y. Lai, B.E. Rittmann, H.-P. Zhao, P. Zheng, Complete Perchlorate Reduction Using Methane as the Sole Electron Donor and Carbon Source, Environ. Sci. Technol. 49 (2015) 2341-2349. [10] P.J. Strong, S. Xie, W.P. Clarke, Methane as a Resource: Can the Methanotrophs Add Value?, Environ. Sci. Technol. 49 (2015) 4001-4018. [11] Y. Liu, Y. Zhang, B.-J. Ni, Zero valent iron simultaneously enhances methane production and sulfate reduction in anaerobic granular sludge reactors, Water Res. 75 (2015) 292-300. 23
[12] W. Wei, X. Zhou, D. Wang, J. Sun, Q. Wang, Free ammonia pre-treatment of secondary sludge significantly increases anaerobic methane production, Water Res. 118 (2017) 12-19. [13] J. Sun, S. Hu, K.R. Sharma, B.-J. Ni, Z. Yuan, Stratified Microbial Structure and Activity in Sulfide- and Methane-Producing Anaerobic Sewer Biofilms, Appl. Environ. Microb. 80 (2014) 7042-7052. [14] J. Sun, I. Pikaar, K.R. Sharma, J. Keller, Z. Yuan, Feasibility of sulfide control in sewers by reuse of iron rich drinking water treatment sludge, Water Res. 71 (2015) 150-159. [15] A. Silva-Teira, A. Sánchez, D. Buntner, L. Rodríguez-Hernández, J.M. Garrido, Removal of dissolved methane and nitrogen from anaerobically treated effluents at low temperature by MBR post-treatment, Chem. Eng. J. 326 (2017) 970-979. [16] H.-P. Zhao, A. Ontiveros-Valencia, Y. Tang, B.-O. Kim, S. VanGinkel, D. Friese, R. Overstreet, J. Smith, P. Evans, R. Krajmalnik-Brown, B. Rittmann, Removal of multiple electron acceptors by pilot-scale, two-stage membrane biofilm reactors, Water Res. 54 (2014) 115-122. [17] X. Chen, Y. Liu, L. Peng, B.-J. Ni, Perchlorate, nitrate, and sulfate reduction in hydrogen-based membrane biofilm reactor: Model-based evaluation, Chem. Eng. J. 316 (2017) 82-90. [18] O. Wanner, P. Reichert, Mathematical modeling of mixed-culture biofilms, Biotechnol. Bioeng. 49 (1996) 172-184. 24
[19] S.W.M. Kengen, G.B. Rikken, W.R. Hagen, C.G. van Ginkel, A.J.M. Stams, Purification
and
Characterization
of
(Per)Chlorate
Reductase
from
the
Chlorate-Respiring Strain GR-1, J. Bacteriol. 181 (1999) 6706-6711. [20]
R.
Nerenberg,
hydrogen-oxidizing,
Y.
Kawagoshi,
perchlorate-reducing
B.E.
Rittmann,
bacterium,
Water
Kinetics Res.
40
of
a
(2006)
3290-3296. [21] R. Chen, Y.-H. Luo, J.-X. Chen, Y. Zhang, L.-L. Wen, L.-D. Shi, Y. Tang, B.E. Rittmann, P. Zheng, H.-P. Zhao, Evolution of the microbial community of the biofilm in a methane-based membrane biofilm reactor reducing multiple electron acceptors, Environ. Sci. Pollut. R 23 (2016) 9540-9548. [22] C.S. Laspidou, B.E. Rittmann, Non-steady state modeling of extracellular polymeric substances, soluble microbial products, and active and inert biomass, Water Res. 36 (2002) 1983-1992. [23] Y. Liu, Y. Zhang, Z. Zhao, H.H. Ngo, W. Guo, J. Zhou, L. Peng, B.-J. Ni, A modeling approach to direct interspecies electron transfer process in anaerobic transformation of ethanol to methane, Environ. Sci. Pollut. R 24 (2017) 855-863. [24] Y. Tang, H. Zhao, A.K. Marcus, R. Krajmalnik-Brown, B.E. Rittmann, A Steady-State Biofilm Model for Simultaneous Reduction of Nitrate and Perchlorate, Part 1: Model Development and Numerical Solution, Environ. Sci. Technol. 46 (2012) 1598-1607. [25] O. Wanner, E. Morgenroth, Biofilm modeling with AQUASIM, Water Science & Technology 49 (2004) 137-144. 25
[26] J. Sun, X. Dai, Y. Liu, L. Peng, B.-J. Ni, Sulfide removal and sulfur production in a membrane aerated biofilm reactor: Model evaluation, Chem. Eng. J. 309 (2017) 454-462. [27] Z. Lewandowski, H. Beyenal, J. Myers, D. Stookey, The effect of detachment on biofilm structure and activity: the oscillating pattern of biofilm accumulation, Water Sci. Technol. 55 (2007) 429-436. [28] G. Koch, K. Egli, J. Van der Meer, H. Siegrist, Mathematical modeling of autotrophic denitrification in a nitrifying biofilm of a rotating biological contactor, Water Sci. Technol. 41 (2000) 191-198. [29] T. Ito, S. Okabe, H. Satoh, Y. Watanabe, Successional Development of Sulfate-Reducing Bacterial Populations and Their Activities in a Wastewater Biofilm Growing under Microaerophilic Conditions, Appl. Environ. Microbiol. 68 (2002) 1392-1402. [30] Y. Shi, S. Hu, J. Lou, P. Lu, J. Keller, Z. Yuan, Nitrogen Removal from Wastewater by Coupling Anammox and Methane-Dependent Denitrification in a Membrane Biofilm Reactor, Environ. Sci. Technol. 47 (2013) 11577-11583. [31] Z. Wang, S. Xia, X. Xu, C. Wang, Utilizing a one-dimensional multispecies model to simulate the nutrient reduction and biomass structure in two types of H2-based membrane-aeration biofilm reactors (H2-MBfR): model development and parametric analysis, Environ. Sci. Pollut. R 23 (2016) 3436-3449. [32] X. Chen, J. Guo, Y. Shi, S. Hu, Z. Yuan, B.-J. Ni, Modeling of Simultaneous Anaerobic Methane and Ammonium Oxidation in a Membrane Biofilm Reactor, 26
Environ. Sci. Technol. 48 (2014) 9540-9547. [33] G.-P. Sheng, H.-Q. Yu, X.-Y. Li, Extracellular polymeric substances (EPS) of microbial aggregates in biological wastewater treatment systems: A review, Biotechnol. Adv. 28 (2010) 882-894. [34] E. Sahinkaya, H. Hasar, A.H. Kaksonen, B.E. Rittmann, Performance of a Sulfide-Oxidizing,
Sulfur-Producing
Membrane
Biofilm
Reactor
Treating
Sulfide-Containing Bioreactor Effluent, Environ. Sci. Technol. 45 (2011) 4080-4087. [35] M. Long, C. Zhou, S. Xia, A. Guadiea, Concomitant Cr(VI) reduction and Cr(III) precipitation with nitrate in a methane/oxygen-based membrane biofilm reactor, Chem. Eng. J. 315 (2017) 58-66. [36] S. Xia, C. Wang, X. Xu, Y. Tang, Z. Wang, Z. Gu, Y. Zhou, Bioreduction of nitrate in a hydrogen-based membrane biofilm reactor using CO2 for pH control and as carbon source, Chem. Eng. J. 276 (2015) 59-64. [37] Y. Tang, H. Zhao, A.K. Marcus, R. Krajmalnik-Brown, B.E. Rittmann, A steady-state biofilm model for simultaneous reduction of nitrate and perchlorate, part 2: parameter optimization and results and discussion, Environ. Sci. Technol. 46 (2012) 1608-1615. [38] H. Choi, J. Silverstein, Inhibition of perchlorate reduction by nitrate in a fixed biofilm reactor, J. Hazard. Mater. 159 (2008) 440-445. [39] P. Dunfield, R. Knowles, Kinetics of inhibition of methane oxidation by nitrate, nitrite, and ammonium in a humisol, Appl. Environ. Microb. 61 (1995) 3129-3135. 27
[40] J. Sun, X. Dai, Q. Wang, Y. Pan, B.-J. Ni, Modelling Methane Production and Sulfate Reduction in Anaerobic Granular Sludge Reactor with Ethanol as Electron Donor, Scientific Reports 6 (2016) 35312.
28
Legends of Figures
Figure 1. Schematic representation of microbial processes occurred in the methane-based MBfR. Lines in the figure connect substances, including CH4, O2, ClO4-, NO2-, NO3-, extracellular polymeric substances (EPS), substrate-utilization -associated products (UAP), biomass-associated products (BAP), inert biomass (IB), which could be either consumed or produced by microorganisms involved in the system, i.e. aerobic methanotrophs (AMO), heterotrophic bacteria (HB), perchlorate reduction bacteria (PRB) and denitrifying anaerobic methane oxidation bacteria or archaea (DAMOb or DAMOa).
Figure 2. Comparison of effluent perchlorate concentration (A), perchlorate removal flux (B), effluent nitrate and nitrite concentration (C) and nitrate and nitrite removal flux (D) at Stage 1-4 from the experiment and from the model with optimized parameters.
Figure 3. Comparison of effluent perchlorate concentration (A), perchlorate removal flux (B), effluent nitrate and nitrite concentration (C) and nitrate and nitrite removal flux (D) at Stage 5-7 from the experimental measurements and from the model prediction
Figure 4.
The simulated distribution of solid components in the biofilm at Stage 5
(A) and Stage 6 (B). The simulated relative abundance of solid components in the biofilm at Stage 5 (C) and Stage 6 (D)
29
Figure 5. Model simulated perchlorate removal efficiency under different combinations of methane pressure and perchlorate loading in 3D (A) and 2D (B) plots. The efficiency (%) was represented by the color scale in all 2D plots. The dash-dotted line in Figure represented for perchlorate removal efficiency being up to 80%.
Figure 6. Model simulated perchlorate removal efficiency under different nitrate (A) and nitrite loading (B) in 2D plots. The efficiency (%) was represented by the color scale. The dotted line in the figure represented for perchlorate removal efficiency at 20%, 40%, 60% and 80%, respectively.
30
CH4
DAMOb NO2NO3-
AMO
EPS
O2
BAP
DAMOa
CH3OH UAP IB HB AMO related processes
NO2-
ClO4NO3
HB related processes PRB related processes
-
DAMOb related processes
PRB
DAMOa related processes Hydrolysis of EPS
Figure 1. Schematic representation of microbial processes occurred in the methane-based MBfR. Lines in the figure connect substances, including CH4, O2, ClO4-, NO2-, NO3-, extracellular polymeric substances (EPS), substrate-utilization -associated products (UAP), biomass-associated products (BAP), inert biomass (IB), which could be either consumed or produced by microorganisms involved in the system, i.e. aerobic methanotrophs (AMO), heterotrophic bacteria (HB), perchlorate reduction bacteria (PRB) and denitrifying anaerobic methane oxidation bacteria or archaea (DAMOb or DAMOa).
31
Figure 2. Comparison of effluent perchlorate concentration (A), perchlorate removal flux (B), effluent nitrate and nitrite concentration (C) and nitrate and nitrite removal flux (D) at Stage 1-4 from the experiment and from the model with optimized parameters.
32
Figure 3. Comparison of effluent perchlorate concentration (A), perchlorate removal flux (B), effluent nitrate and nitrite concentration (C) and nitrate and nitrite removal flux (D) at Stage 5-7 from the experimental measurements and from the model prediction
33
Figure 4.
The simulated distribution of solid components in the biofilm at Stage 5
(A) and Stage 6 (B). The simulated relative abundance of solid components in the biofilm at Stage 5 (C) and Stage 6 (D)
34
Figure 5. Model simulated perchlorate removal efficiency under different combinations of methane pressure and perchlorate loading in 3D (A) and 2D (B) plots. The efficiency (%) was represented by the color scale in all 2D plots. The dash-dotted line in Figure represented for perchlorate removal efficiency being up to 80%.
35
Figure 6. Model simulated perchlorate removal efficiency under different nitrate (A) and nitrite loading (B) in 2D plots. The efficiency (%) was represented by the color scale. The dotted line in the figure represented for perchlorate removal efficiency at 20%, 40%, 60% and 80%, respectively.
36
Graphical Abstract
B
A
DAMOb NO2
AMO
-
NO3EPS
O2
BAP
80 AMO
HB
DA
XI
PRB
DB
EPS
60
40
20
0
DAMOa
Solid component fraction (%)
CH4
Solid component fraction (%)
80
20
40
20
40
60
80
100
0
C
D
AMO related processes
NO3
HB related processes PRB related processes DAMOb related processes
PRB
DAMOa related processes Hydrolysis of EPS
37
40
Biofilm
HB
-
20
Biofilm thickness ( m)
UAP
ClO4-
X
60
CH3OH
NO2-
H
0 0
IB
AMO DA
AMO HB PRB DB XI EPS
Highlights
A model describing perchlorate reduction in a methane-based MBfR was developed
The model was verified by experimental data under different operational conditions
The stratified microbial distribution in the biofilm was reveal by model analysis
Over 80% perchlorate removal efficiency can be achieved with proper PCH4 and LClO4.
The perchlorate reduction could be affect by nitrate and nitrite concentrations
38