COD removal characteristics in air-cathode microbial fuel cells
Bioresource Technology 176 (2015) 23–31
Contents lists available at ScienceDirect
Bioresource Technology journal homepage: www.elsevier.com/locate/biortech
COD removal characteristics in air-cathode microbial fuel cells Xiaoyuan Zhang a,b, Weihua He c, Lijiao Ren b, Jennifer Stager b, Patrick J. Evans d, Bruce E. Logan b,⇑ a
State Key Joint Laboratory of Environment Simulation and Pollution Control, School of Environment, Tsinghua University, Beijing 100084, PR China Department of Civil & Environmental Engineering, Penn State University, 231Q Sackett Building, University Park, PA 16802, USA c State Key Laboratory of Urban Water Resource and Environment, Harbin Institute of Technology, No. 73 Huanghe Road, Nangang District, Harbin 150090, PR China d CDM Smith, 14432 S.E. Eastgate Way, Suite 100, Bellevue, WA 98007, USA b
h i g h l i g h t s COD removal in air-cathode MFCs fit
first-order kinetics. Current generation accelerated COD removal for both acetate and wastewater. COD removal rate constants for wastewater with current were higher than acetate. Current generation stopped before substrates were fully depleted. For acetate, coulombic efficiencies were inversely related to COD concentration.
a r t i c l e
i n f o
Article history: Received 15 September 2014 Received in revised form 29 October 2014 Accepted 1 November 2014 Available online 7 November 2014 Keywords: Microbial fuel cell COD removal rate First-order reaction Coulombic efficiency Domestic wastewater
g r a p h i c a l a b s t r a c t Open Circuit Open Circuit 100 Ω 1.1
Exoelectrogens Anode
Exoelectrogens Heterotrophs
C a t h o d e
100 Ω 0.6
4.0
Average COD removal rates (kg/m3/d) COD removal: 1st Order Reaction
a b s t r a c t Exoelectrogenic microorganisms in microbial fuel cells (MFCs) compete with other microorganisms for substrate. In order to understand how this affects removal rates, current generation, and coulombic efficiencies (CEs), substrate removal rates were compared in MFCs fed a single, readily biodegradable compound (acetate) or domestic wastewater (WW). Removal rates based on initial test conditions fit first-order kinetics, but rate constants varied with circuit resistance. With filtered WW (100 O), the rate constant was 0.18 h1, which was higher than acetate or filtered WW with an open circuit (0.10 h1), but CEs were much lower (15–24%) than acetate. With raw WW (100 O), COD removal proceeded in two stages: a fast removal stage with high current production, followed by a slower removal with little current. While using MFCs increased COD removal rate due to current generation, secondary processes will be needed to reduce COD to levels suitable for discharge. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction A microbial fuel cell (MFC) is an innovative method to generate electricity from organic matter using exoelectrogenic bacteria (Li et al., 2014; Logan, 2008; Lovley, 2008; Rabaey and Verstraete, 2005). MFCs have drawn increasing attention during the past ⇑ Corresponding author at: Department of Civil and Environmental Engineering, Penn State University, University Park, PA, USA. Tel.: +1 814 863 7908; fax: +1 814 863 7304 (B.E. Logan). E-mail addresses: [email protected] (X. Zhang), [email protected] (B.E. Logan). http://dx.doi.org/10.1016/j.biortech.2014.11.001 0960-8524/Ó 2014 Elsevier Ltd. All rights reserved.
Heterotrophs
decade as a possible new technology for wastewater treatment, following an early demonstration that electricity could be produced using domestic wastewater (Liu et al., 2004). Since then, MFCs have been modified for many other purposes, including desalination, chemical production, and biosensor applications (Wang and Ren, 2013). However, if the primary use of the MFC is for wastewater treatment, better information is needed on treatment rates relative to wastewater characteristics, such as the concentration of organic matter. Most critically, the final concentration of organic matter, measured in terms of chemical (COD) or biochemical oxygen demand (BOD), must be low enough to meet discharge limitations.
24
X. Zhang et al. / Bioresource Technology 176 (2015) 23–31
In most MFC studies, the emphasis has been placed on maximum power densities, rather than the extent of COD removal. Several types of substrates have been examined, including: defined substrates that allow reproducible experimental conditions, such as acetate (Liu et al., 2005; Torres et al., 2007), butyrate (Liu et al., 2005), and glucose (Liu and Logan, 2004) as well as actual wastewaters that have a composition and concentration that can vary over time, such as domestic (Min and Logan, 2004) and brewery wastewaters (Feng et al., 2008). In general, power densities using these substrates have been found to increase with the concentration of the COD up to a maximum, when batch tests are initiated at specific substrate concentrations, and then they remain constant with further increases in COD, resulting in a Monod-like relationship between power and substrate concentrations (Feng et al., 2008; Liu et al., 2005; Min and Logan, 2004). COD removal rates and the extent of COD removal depend on the reactor operation conditions. In continuous flow tests, removal rates are calculated based on reactor influent and effluent COD concentrations as a function of hydraulic retention time (HRT) (Fan et al., 2012; Liu et al., 2008; Moon et al., 2005). COD removal rates generally increase with shorter HRTs (Zhang et al., 2013a). While high power densities are achieved at these shorter HRTs, longer HRTs are needed to reduce COD levels suitable for wastewater discharge. At very long HRTs, current densities can also be quite low, making it impractical to produce any useful electrical power at these low effluent COD concentrations (Ahn and Logan, 2012). In fed-batch reactors, COD removal rates are usually only reported for the whole cycle, based on initial and final COD concentrations (Hays et al., 2011). Removal rates are rarely examined over time, due in part to the difficulty of replacing solution without affecting current generation, for example, as a result of introduction of air into the reactor. Sometimes different initial substrate concentrations are used, but usually a single substrate concentration is tested (Ge et al., 2013; Pant et al., 2010). The rate of COD removal in an MFC is affected by microbial growth with current generation, aerobic growth due to oxygen leaking in through the cathode, and anaerobic growth using other terminal electron acceptors in the wastewater, including carbon dioxide. The high surface areas of the electrodes, particularly the anode, can greatly affect the rate of COD removal even in the absence of current generation. Microbial communities can change with different operational conditions, which can affect COD removal rates and current generation (Zhang et al., 2011). COD removal under open circuit conditions is usually not reported for continuous flow experiments, and only infrequently examined in fed-batch tests. For example, COD removal was examined over time in a baffled single-chambered air-cathode MFC using glucose as substrate in phosphate buffer solution with current generation, but not under open circuit conditions (Li et al., 2008). Reporting the coulombic efficiency (CE) is sufficient to determine the amount of COD that was captured as electrical current by the end of a fedbatch cycle, but without COD measurements over the course of the fed batch cycle, the COD removal due to current production cannot be separated from that due to other anaerobic or aerobic processes. Substantial COD removal can occur due to other anaerobic processes, particularly in reactors containing a high surface area anode, and aerobic COD removal is supported by diffusion of oxygen through the cathode. For example, substrate removal was 94% in a single chamber, air-cathode MFC fed acetate under open circuit conditions, compared to 98% with current generation (Shehab et al., 2013). Sealing the air-cathode with a solid plate reduced substrate removal to only 7%, suggesting most acetate removal was due to oxygen leaking through the cathode. In order to effectively design MFCs for wastewater treatment, the relationship between current production and COD removal relative to current generation versus other aerobic and anaerobic
processes must be better understood. The amount of substrate lost to processes that do not generate electrical current varies, depending on reactor operation, even for reactors operated over the same period of time. For example, 67% of acetate was removed without current generation (open circuit conditions) in an air cathode MFC in 9 h (Ren et al., 2014b). With current generation over the same period of time, this loss was reduced to 35%, with 53% of the substrate converted to electricity by exoelectrogens (88% COD removal). Although maximum current densities are often reported at the beginning of a fed batch cycle, current generation is not constant over time. Thus, the rate of substrate removal changes during the fed batch cycle. The impact of substrate concentration on substrate removal through current generation versus loss to aerobic processes, has not been well examined. The effect of COD concentration on current generation is particularly important at the end of a fed-batch cycle, when COD concentrations are low, as no current may be generated. Therefore, without a comparison of COD removals in MFCs generating current over time, to open circuit controls, the impact of COD concentration on the extent of treatment cannot be adequately understood. In this study, COD removal rates were examined for domestic wastewater (soluble COD = 220 mg/L), and compared to those obtained using identical reactors, but with acetate as the fuel with a low COD concentration (0.3 g/L sodium acetate) that is more representative of a domestic wastewater. In addition, a higher concentration (1 g/L) of acetate was used that is more representative of laboratory MFC tests. Current and COD concentrations were examined under both open and closed circuit conditions (two different external resistances) to better understand the competition by microorganisms for biodegradable COD in air-cathode MFCs, under these operational different conditions. 2. Methods 2.1. MFC construction and operation Air-cathode MFCs were cube-shaped reactors with a single cylindrical chamber (4 cm length, 26 mL net liquid volume), constructed as previously described (Logan et al., 2007). The anode was a graphite fiber brush made from a core of two twisted titanium wires that functioned as current collectors (Logan et al., 2007). The brush was heat treated in a muffle furnace at 450 °C for 30 min, and then placed in the middle of the MFC chamber. The air-cathode was made of 30% wet-proofed carbon cloth (Type B, BASF Fuel Cell, Inc., USA) with a platinum (0.5 mg/cm2) catalyst layer on the water side, and four polytetrafluoroethylene (PTFE) diffusion layers on the air side to prevent water loss (7 cm2 projected area) (Cheng et al., 2006). Anodes for acetate COD tests were inoculated with effluent from MFCs originally inoculated with domestic wastewater that were acclimated to sodium acetate (1 g/L, medium described below) for over one year. Anodes for domestic wastewater COD tests were inoculated and acclimated only using domestic wastewater from the primary clarifier overflow. All MFCs were operated under fed batch conditions at 30 ± 1 °C. 2.2. COD removal tests Domestic wastewater was collected from the primary clarifier overflow of the Pennsylvania State University Wastewater Treatment Plant, with a total COD (tCOD) of 439 ± 55 mg/L. Except as noted, wastewater was filtered (0.4 lm diameter, polycarbonate membrane, Poretics Corporation, USA), producing a final soluble COD (sCOD) of 223 ± 6 mg/L. The use of filtered wastewater provided a substrate that was more similar to dissolved acetate due
X. Zhang et al. / Bioresource Technology 176 (2015) 23–31
to removal of particles, and it avoided production of additional soluble substrate from particulate COD. In some additional tests, raw (unfiltered) domestic wastewater was also used to separately examine the effect of the particulate COD on removal rates. Acetate fed MFCs were supplied with a medium containing acetate in a 50 mM phosphate buffer solution (PBS; 4.57 g/L Na2HPO4, 2.45 g/L NaH2PO4H2O, 0.31 g/L NH4Cl and 0.13 g/L KCl) amended with 12.5 mL/L mineral and 5 mL/L vitamin solutions (Logan et al., 2007). Two different concentrations of acetate were used: 0.3 g/L sodium acetate (260 mg/L COD) to produce a COD similar to that of the sCOD of the local domestic wastewater; and 1 g/L sodium acetate (840 mg/L COD), which is a concentration typically used in MFC laboratory studies. Tests were first conducted using 1 g/L, followed by tests at 0.3 g/L. Three circuit conditions were examined with these solutions: an external resistance of 1000 O, which is commonly used for MFC start-up and operation (Zhang et al., 2013b); an external resistance of 100 O, as this produces a high current and typically the maximum power density (1000 mW/m2) in these MFCs when fed acetate; and an open circuit control, to examine the effect of substrate removal in the absence of current generation. In order to obtain COD concentrations of acetate in the reactor over time, and avoid leaving air in the reactor due to the removal of the solution, different feeding strategies were developed for acetate and wastewater samples. These procedures enabled a reasonable number of samples to be taken over time, while minimizing the need for a large number of reactors (Fig. S1). Since tests were conducted at the same time under three different conditions (open circuit, 1000 O and 100 O), a total of 12 reactors were used. For acetate, 4 MFCs were started at the same initial condition, with two of these four reactors sampled after 2, 6, 10, 23 and 48 h, and the other pair of duplicate reactors sampled at 4, 8, 12, 23, 48 h (i.e., two pairs of duplicate reactors sampled at different times). After a sample (2 mL) was taken, the COD was immediately determined, and solution was replaced with fresh media at the same COD based on the COD data of the other pair of reactors (Fig. S1). For example, at 4 h, after a sample was taken, the solution removed was replenished with new solution at the same COD of the sample at 2 h (Fig. S1). The same procedure was conducted from 2 to 12 h. At 23 h, a sample was taken and at 25 h the solution was replenished with the same COD based on the COD measurement at 23 h. At 48 h, the end of the test, a sample was taken and all the solution in the reactor was replaced by fresh influent for the next cycle. Using this approach, there was at least a 4 h interval between sampling from the same reactor. This procedure did introduce minor changes in COD added back into the reactors as the two concentrations were not identical. However, as the volume of sample withdrawn was small (2 mL) compared to the total solution volume (26 mL), it was assumed that this volume difference and only a small difference in COD would result in a relatively negligible error in the COD concentration in the reactors, compared to changes that would occur with leaving air in the reactor. A shorter interval between sampling of 2 h was also examined, but results were inconsistent (data not shown). For domestic wastewater COD tests, a different sampling strategy was developed due to need for a larger volume of sample for COD analysis. Total COD and soluble COD tests for raw wastewater required 5 mL per sample, which was not negligible relative to the total volume of solution. Therefore, wastewater-fed MFCs were sampled only once, and the reactor was not sampled again. Using 4 reactors, samples were taken in duplicate after 2, 4, 8, 10 and 24 h, and tests were repeated several times. COD removal can occur due to biofilms on both the anode brush and the cathode. Additional tests were therefore conducted to separately examine COD removal by the anode and cathode biofilms. Cathodes containing thick biofilms were removed from the MFCs
25
and placed in 4-cm length cubic reactors that lacked an anode. The end of the reactor that still contained the original anode was either sealed with an endplate to substantially reduce oxygen transfer into the reactor, or a new cathode was placed in the reactor with a glass fiber separator allowing some oxygen transfer into the reactor to better simulate conditions where oxygen might leak into the anode chamber when the cathode contained a biofilm (Zhang et al., 2010, 2009). COD removal rates were then examined in these reactors under open circuit conditions using a high concentration of sodium acetate (1 g/L). 2.3. Analysis of MFCs Voltages (U) were recorded across an external resistance (R, 1000 O or 100 O) at 20 min intervals using a data acquisition system (2700, Keithley Instrument, USA) connected to a personal computer. Current densities (J) were normalized by air-cathode projected area (A = 7 cm2) using J = U/RA (Logan et al., 2006). All CODs were measured using standard methods and a spectrophotometer (DR/2010, Hach Co., Loveland, CO, USA). COD removal data were analyzed using the Solver and Data Analysis Tools in Microsoft Excel (Professional Plus 2010). COD removal rates were fitted assuming a first-order reaction with respect to concentration, and calculated as follows
LnðCODt =COD0 Þ ¼ kt
ð1Þ
where COD0 is the influent COD, CODt the effluent COD, t is time, and k is the first order removal rate constant. While kinetics are assumed to be first-order with respect to concentration, they can vary with external resistance (Ren et al., 2014b). Coulombic efficiency (CE), defined as the fractional recovery of electrons from the substrate, was calculated (Logan et al., 2006) as:
Rt M 0 Idt CE ¼ nv FðCOD0 CODt Þ
ð2Þ
where M is the molecular weight of oxygen, I is the current, F is Faraday’s constant, n = 4 is the number of electrons exchanged per mole of oxygen, and v is the anolyte volume. 3. Results and discussion 3.1. COD removal rates with acetate COD removal rates with acetate at the low initial COD concentration (COD0 = 260 mg/L) showed good agreement with an assumption of first-order degradation kinetics, although rates were different depending on the circuit resistance (Fig. 1 and Fig. S2). For the low acetate concentration, the largest rate constant of 0.14 ± 0.03 h1 was obtained with the highest maximum current density produced with the lower external resistance of 100 O. At the lower current density produced using a higher resistance (1000 O), the rate constant (0.071 ± 0.008 h1) was about half that at the higher current density, and it decreased to 0.034 ± 0.002 h1 under open circuit (zero current) conditions (Fig. 1 and Fig. S2). The COD after 8 h of operation was 58 ± 11 mg/L with an external resistance of 100 O, compared to 108 ± 12 mg/L (1000 O) and 167 ± 10 mg/L (open circuit). After 8 h with the 100 O resistor, COD removal became very low resulting in a nearly constant COD of 58 ± 10 mg/L from 8 to 48 h. The final CODs after 48 h of operation were 57 ± 13 mg/L (78% removal, 100 O), 24 ± 15 mg/L (91% removal, 1000 O), and 47 ± 11 mg/L (82% removal, open circuit) (Fig. 1). These variations in the final CODs were not clearly related to the initial COD or circuit conditions, and they are all higher than reported in some previous studies. For example, Liu et al. (2005) reported >99% acetate removal (initial acetate concentration of
26
X. Zhang et al. / Bioresource Technology 176 (2015) 23–31
0
300
COD (mg/L)
250
Ln(COD/COD0) (Arbitrary)
Ac Low OCV Ac Low 1000 Ω Ac Low 100 Ω
200 150 100
A
50
0
10
20
30 Time (h)
40
50
-1 -1.5
B
-2
0
60
900
10
20
30 Time (h)
40
50
60
0
750
Ln(COD/COD0) (Arbitrary)
Ac High OCV
COD (mg/L)
-0.5
-2.5
0
Ac High 1000 Ω Ac High 100 Ω
600 450 300
C
150
-1 -2
Ac High OCV Ac High 1000 Ω Ac High 100 Ω N-Ac High 100 Ω Fit Ac High OCV Fit Ac High 1000 Ω Fit Ac High 100 Ω
-3 -4
0 0
10
20
30 Time (h)
40
50
0
60
Ln(sCOD/sCOD0) (Arbitrary)
WW OCV
250
WW 1000 Ω
200
WW 100 Ω
150 100
E
50 0 0
5
10
15 Time (h)
10
20
D 30
40
50
60
Time (h)
300
sCOD (mg/L)
Ac Low OCV Ac Low 1000 Ω Ac Low 100 Ω N-Ac Low 1000 Ω N-Ac Low 100 Ω Fit Ac Low OCV Fit Ac Low 1000 Ω Fit Ac Low 100 Ω
20
25
30
0 -0.2 -0.4 -0.6 -0.8 -1 -1.2 -1.4 -1.6 -1.8
WW OCV WW 1000 Ω WW 100 Ω N-WW OCV N-WW 1000 Ω N-WW 100 Ω Fit WW OCV Fit WW 1000Ω Fit WW 100 Ω
F 0
5
10
15 Time (h)
20
25
30
Fig. 1. COD of low (A) and high (C) concentrations of acetate (Ac) in MFCs changes with time; soluble COD (sCOD) of filtered domestic wastewater changes with time in the MFCs initially acclimated to wastewater (WW) (E). (B), (D), and (F): first order reaction fitting data; only the data with filled symbols were used to fit the first order relationship, and the data with open symbols were not (N), as the COD removal had essentially ceased after current generation stopped with 1000 O and 100 O (shown in Fig. 3). The MFCs were operated under three conditions including open circuit (OCV), 1000 O and 100 O external resistors.
800 mg/L) using an air-cathode MFC, while Freguia et al. (2007) reported 100% acetate removal using a ferricyanide catholyte. In both of these studies (Freguia et al., 2007; Liu et al., 2005), acetate concentrations were measured rather than COD, and thus some of the remaining COD may not be acetate. COD was used here so that the same methods were used for acetate and wastewater tests. Also, the different reactor conditions, such as flat, widely spaced electrodes (Liu et al., 2005) compared to brush electrodes placed close to the cathodes, and much lower current densities than those achieved here, could have affected the development of microbial communities and their responses to low substrate conditions and oxygen transfer from the cathode into solution. For the high acetate concentration (COD0 = 840 mg/L), the rate constants again showed agreement with first-order kinetics, with rate constants that increased with current. The highest rate constant was 0.086 ± 0.006 h1 (100 O, high current), followed by 0.065 ± 0.002 h1 (1000 O), and 0.030 ± 0.002 h1 (open circuit) (Fig. 1 and Fig. S2). After 24 h of operation, the CODs in the MFCs were 99 ± 51 mg/L with the 100 O circuit, compared to 188 ± 79 mg/L (1000 O) and 323 ± 112 mg/L (open circuit). After 48 h of operation, the final COD under open circuit was 184 ± 81 mg/L (78% removal), which was much higher than that obtained with current generation of 59 ± 18 mg/L (93% removal, 100 O) and 33 ± 5 mg/L (96% removal, 1000 O) (Fig. 1). In these
tests at the higher COD concentration, it was shown that current production accelerated COD removal compared to that with no current (open circuit), and that the final COD was appreciably lower with current generation than the open circuit control. While the COD removal rates can be seen to obey first-order kinetics for any specific initial COD concentration, the overall COD removal rate depended on operating conditions. In some cases, the rate constant was unchanged based on the initial COD. For example, under open circuit conditions (no current), the rate constants were the same (within a standard deviation) at the low initial COD of 260 mg/L (0.034 ± 0.002 h1) and high COD of 840 mg/L (0.030 ± 0.002 h1). Similarly, with the high resistance of 1000 O (current data discussed below), the rate constants were also similar at both initial COD concentrations (0.071 ± 0.008 h1, 260 mg/L; and 0.065 ± 0.002 h1, 840 mg/L). However, when the lower resistance of 100 O was used, rate constants were significantly increased. At the initial COD of 840 mg/L, the rate constant was 0.086 ± 0.006 h1, which was 39% lower than that obtained at an initial COD concentration of 260 mg/L (0.14 ± 0.03 h1). If the overall rate remained truly first order under all conditions, the rate concentration would be independent of the initial concentration, as shown in Eq. (1). Thus, COD removal rates were appreciably affected by initial COD concentrations at the lowest resistance (higher initial current densities).
X. Zhang et al. / Bioresource Technology 176 (2015) 23–31
In all these tests, there was no indication of saturation kinetics relative to COD removal. However, the current did not always appreciably decrease following acetate addition, and thus current production did not vary in proportion to COD removal. Thus, the rate of COD removal in a fed batch reactor for these substrate concentrations could be described by first order kinetics, with the recognition that rates were dependent on the organic loading range and external resistance (circuit load). 3.2. COD removal rates with domestic wastewater COD removal rates with filtered domestic wastewater in the MFCs inoculated with domestic wastewater were also well described by first order kinetics, with rate constants increased by current production (Fig. 1 and Fig. S2). The rate constants with current generation using filtered wastewater were 0.18 ± 0.03 h1 (100 O) and 0.17 ± 0.03 h1 (1000 O). These rate constants are 80% (100 O) and 70% (1000 O) higher than those obtained under open circuit conditions (0.10 ± 0.01 h1). The final CODs were similar with and without current generation, but the time needed for treatment varied due to differences in rates. The CODs with current generation decreased from 214 ± 2 mg/L to 51 ± 4 mg/L (100 O) and 52 ± 5 mg/L (1000 O) after 8 h, and did not further decrease from 8 to 24 h. The final COD of the open circuit controls was similar at 56 ± 2 mg/L when the experiment was stopped, but this extent of treatment required 24 h. After 8 h, under open circuit conditions, the COD was 84 ± 2 mg/L. These final COD concentrations were comparable to those previously obtained using MFCs fed domestic wastewater (60 to 80 mg/L total COD and soluble COD) (Hays et al., 2011). In those studies, the final BOD5 was estimated to be 30 to 40 mg/L. Previous research has shown that reactors acclimated to wastewater can be effectively used with other wastewaters (Ren et al., 2013). However, the converse was not found to be true here: long-term acclimation of the reactor to acetate showed very low COD removal rates when tested with domestic wastewater. Reactors fed acetate for over one year, exhibited similar and very low COD removal rates with domestic wastewater, with all rate constants [0.02 h1 (Fig. S2). These low rate constants showed that long-term acclimation to acetate hindered the ability of the anode to degrade the more complex substrates in filtered wastewater. 3.3. Comparison of removal rates in acetate and wastewater A comparison of the first-order rate constants shows that COD removal rate constants with domestic wastewater were actually higher than those of acetate at each comparable condition (100 O, 1000 O, or open circuit) (Fig. 2). This higher rate constant for the wastewater was surprising given that acetate is considered to be readily biodegradable compared to the more complex organic matter in domestic wastewater, and acetate produces higher power densities than wastewater. The rate constants obtained here for acetate are lower than those previously reported for smaller but quite similar MFCs, likely as a result of different electrode specific surface areas (electrode area per anolyte volume). Ren et al. obtained a 1.4 higher rate constant (0.33 ± 0.02 h1) with current (100 O), and a 3.1 higher rate constant under open circuit conditions (0.14 ± 0.01 h1) using the same sized cathodes as those used here, but tests were conducted with a smaller reactor volume (14 mL) (Ren et al., 2014b). Therefore, the cathode specific surface area in that study was 64 m2/m3 (based on liquid volume) or 2.4 that used here (27 m2/m3). This higher electrode surface area per volume enabled much more rapid depletion of substrate, particularly under open circuit conditions by aerobic heterotrophs using oxygen transfer
27
into the reactor through the cathode. These different rate constants based on the electrode surface area per volume highlight the importance of the specific electrode area when evaluating COD removal rates. 3.4. Coulombic efficiency For both acetate and domestic wastewater, the CEs over time were on average higher in tests with lower resistances (higher current densities) (Fig. 2). For acetate, the CEs were also higher for the lower initial COD, under both 1000 O and 100 O. The acetate-fed MFCs with COD0 = 260 mg/L and 100 O achieved the highest CE of 90% (Fig. 2), indicating that exoelectrogenic microorganisms outcompeted other microorganisms for substrates as 90% of the available electrons in the removed substrate were used for current production. The rest of the removed substrate was degraded by bacteria using oxygen as an electron acceptor, as further discussed below. During the fed-batch cycle with acetate, the CEs also increased over time under all conditions (Fig. 2). The CEs for the filtered domestic wastewater, were 22% (100 O) and 14% (1000 O), which are both lower than those obtained with acetate (Fig. 2). It is commonly observed that CEs are lower for domestic wastewater compared to acetate (Ahn and Logan, 2010; Liu et al., 2004). The CEs for the filtered wastewaters were essentially constant or they slightly decreased over time, which was different from that observed using acetate. 3.5. Rapid decreases in current production at low CODs A comparison of current densities and COD concentrations in the MFCs over time (during a fed-batch cycle) shows that there was a larger impact of COD concentration on the current densities produced, than the rates of COD removal (Fig. 3). The effect of the COD concentration on current can more directly be seen by plotting current as a function of COD (rather than each variable versus time), as shown in Fig. 4. When the COD concentration decreased to less than 100 mg/L for acetate or filtered domestic wastewater, current production rapidly decreased (Fig. 4). This significant loss in current production could be due to a combination of factors, such as mass transfer limitations or increased dissolved oxygen concentrations due to reduced dissolved oxygen removal at lower substrate concentrations. The final CODs in these tests (50 mg/L for acetate and filtered domestic wastewater) were greater than CODs typically needed for wastewater discharge to the environment (Fig. 4). At 100 O, the current densities decreased rapidly when the COD was below only 200 mg/L, and the current was very low even there was still 100 mg/L COD for both acetate and filtered domestic wastewater (Fig. 4). For both acetate and wastewater, with either resistance, current generation was minimal well before the substrate in the reactor was fully depleted. This indicates that it will be difficult to maintain high power production in an MFC if the goal is to reduce the effluent COD levels to values suitable for wastewater discharge (
Contents lists available at ScienceDirect
Bioresource Technology journal homepage: www.elsevier.com/locate/biortech
COD removal characteristics in air-cathode microbial fuel cells Xiaoyuan Zhang a,b, Weihua He c, Lijiao Ren b, Jennifer Stager b, Patrick J. Evans d, Bruce E. Logan b,⇑ a
State Key Joint Laboratory of Environment Simulation and Pollution Control, School of Environment, Tsinghua University, Beijing 100084, PR China Department of Civil & Environmental Engineering, Penn State University, 231Q Sackett Building, University Park, PA 16802, USA c State Key Laboratory of Urban Water Resource and Environment, Harbin Institute of Technology, No. 73 Huanghe Road, Nangang District, Harbin 150090, PR China d CDM Smith, 14432 S.E. Eastgate Way, Suite 100, Bellevue, WA 98007, USA b
h i g h l i g h t s COD removal in air-cathode MFCs fit
first-order kinetics. Current generation accelerated COD removal for both acetate and wastewater. COD removal rate constants for wastewater with current were higher than acetate. Current generation stopped before substrates were fully depleted. For acetate, coulombic efficiencies were inversely related to COD concentration.
a r t i c l e
i n f o
Article history: Received 15 September 2014 Received in revised form 29 October 2014 Accepted 1 November 2014 Available online 7 November 2014 Keywords: Microbial fuel cell COD removal rate First-order reaction Coulombic efficiency Domestic wastewater
g r a p h i c a l a b s t r a c t Open Circuit Open Circuit 100 Ω 1.1
Exoelectrogens Anode
Exoelectrogens Heterotrophs
C a t h o d e
100 Ω 0.6
4.0
Average COD removal rates (kg/m3/d) COD removal: 1st Order Reaction
a b s t r a c t Exoelectrogenic microorganisms in microbial fuel cells (MFCs) compete with other microorganisms for substrate. In order to understand how this affects removal rates, current generation, and coulombic efficiencies (CEs), substrate removal rates were compared in MFCs fed a single, readily biodegradable compound (acetate) or domestic wastewater (WW). Removal rates based on initial test conditions fit first-order kinetics, but rate constants varied with circuit resistance. With filtered WW (100 O), the rate constant was 0.18 h1, which was higher than acetate or filtered WW with an open circuit (0.10 h1), but CEs were much lower (15–24%) than acetate. With raw WW (100 O), COD removal proceeded in two stages: a fast removal stage with high current production, followed by a slower removal with little current. While using MFCs increased COD removal rate due to current generation, secondary processes will be needed to reduce COD to levels suitable for discharge. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction A microbial fuel cell (MFC) is an innovative method to generate electricity from organic matter using exoelectrogenic bacteria (Li et al., 2014; Logan, 2008; Lovley, 2008; Rabaey and Verstraete, 2005). MFCs have drawn increasing attention during the past ⇑ Corresponding author at: Department of Civil and Environmental Engineering, Penn State University, University Park, PA, USA. Tel.: +1 814 863 7908; fax: +1 814 863 7304 (B.E. Logan). E-mail addresses: [email protected] (X. Zhang), [email protected] (B.E. Logan). http://dx.doi.org/10.1016/j.biortech.2014.11.001 0960-8524/Ó 2014 Elsevier Ltd. All rights reserved.
Heterotrophs
decade as a possible new technology for wastewater treatment, following an early demonstration that electricity could be produced using domestic wastewater (Liu et al., 2004). Since then, MFCs have been modified for many other purposes, including desalination, chemical production, and biosensor applications (Wang and Ren, 2013). However, if the primary use of the MFC is for wastewater treatment, better information is needed on treatment rates relative to wastewater characteristics, such as the concentration of organic matter. Most critically, the final concentration of organic matter, measured in terms of chemical (COD) or biochemical oxygen demand (BOD), must be low enough to meet discharge limitations.
24
X. Zhang et al. / Bioresource Technology 176 (2015) 23–31
In most MFC studies, the emphasis has been placed on maximum power densities, rather than the extent of COD removal. Several types of substrates have been examined, including: defined substrates that allow reproducible experimental conditions, such as acetate (Liu et al., 2005; Torres et al., 2007), butyrate (Liu et al., 2005), and glucose (Liu and Logan, 2004) as well as actual wastewaters that have a composition and concentration that can vary over time, such as domestic (Min and Logan, 2004) and brewery wastewaters (Feng et al., 2008). In general, power densities using these substrates have been found to increase with the concentration of the COD up to a maximum, when batch tests are initiated at specific substrate concentrations, and then they remain constant with further increases in COD, resulting in a Monod-like relationship between power and substrate concentrations (Feng et al., 2008; Liu et al., 2005; Min and Logan, 2004). COD removal rates and the extent of COD removal depend on the reactor operation conditions. In continuous flow tests, removal rates are calculated based on reactor influent and effluent COD concentrations as a function of hydraulic retention time (HRT) (Fan et al., 2012; Liu et al., 2008; Moon et al., 2005). COD removal rates generally increase with shorter HRTs (Zhang et al., 2013a). While high power densities are achieved at these shorter HRTs, longer HRTs are needed to reduce COD levels suitable for wastewater discharge. At very long HRTs, current densities can also be quite low, making it impractical to produce any useful electrical power at these low effluent COD concentrations (Ahn and Logan, 2012). In fed-batch reactors, COD removal rates are usually only reported for the whole cycle, based on initial and final COD concentrations (Hays et al., 2011). Removal rates are rarely examined over time, due in part to the difficulty of replacing solution without affecting current generation, for example, as a result of introduction of air into the reactor. Sometimes different initial substrate concentrations are used, but usually a single substrate concentration is tested (Ge et al., 2013; Pant et al., 2010). The rate of COD removal in an MFC is affected by microbial growth with current generation, aerobic growth due to oxygen leaking in through the cathode, and anaerobic growth using other terminal electron acceptors in the wastewater, including carbon dioxide. The high surface areas of the electrodes, particularly the anode, can greatly affect the rate of COD removal even in the absence of current generation. Microbial communities can change with different operational conditions, which can affect COD removal rates and current generation (Zhang et al., 2011). COD removal under open circuit conditions is usually not reported for continuous flow experiments, and only infrequently examined in fed-batch tests. For example, COD removal was examined over time in a baffled single-chambered air-cathode MFC using glucose as substrate in phosphate buffer solution with current generation, but not under open circuit conditions (Li et al., 2008). Reporting the coulombic efficiency (CE) is sufficient to determine the amount of COD that was captured as electrical current by the end of a fedbatch cycle, but without COD measurements over the course of the fed batch cycle, the COD removal due to current production cannot be separated from that due to other anaerobic or aerobic processes. Substantial COD removal can occur due to other anaerobic processes, particularly in reactors containing a high surface area anode, and aerobic COD removal is supported by diffusion of oxygen through the cathode. For example, substrate removal was 94% in a single chamber, air-cathode MFC fed acetate under open circuit conditions, compared to 98% with current generation (Shehab et al., 2013). Sealing the air-cathode with a solid plate reduced substrate removal to only 7%, suggesting most acetate removal was due to oxygen leaking through the cathode. In order to effectively design MFCs for wastewater treatment, the relationship between current production and COD removal relative to current generation versus other aerobic and anaerobic
processes must be better understood. The amount of substrate lost to processes that do not generate electrical current varies, depending on reactor operation, even for reactors operated over the same period of time. For example, 67% of acetate was removed without current generation (open circuit conditions) in an air cathode MFC in 9 h (Ren et al., 2014b). With current generation over the same period of time, this loss was reduced to 35%, with 53% of the substrate converted to electricity by exoelectrogens (88% COD removal). Although maximum current densities are often reported at the beginning of a fed batch cycle, current generation is not constant over time. Thus, the rate of substrate removal changes during the fed batch cycle. The impact of substrate concentration on substrate removal through current generation versus loss to aerobic processes, has not been well examined. The effect of COD concentration on current generation is particularly important at the end of a fed-batch cycle, when COD concentrations are low, as no current may be generated. Therefore, without a comparison of COD removals in MFCs generating current over time, to open circuit controls, the impact of COD concentration on the extent of treatment cannot be adequately understood. In this study, COD removal rates were examined for domestic wastewater (soluble COD = 220 mg/L), and compared to those obtained using identical reactors, but with acetate as the fuel with a low COD concentration (0.3 g/L sodium acetate) that is more representative of a domestic wastewater. In addition, a higher concentration (1 g/L) of acetate was used that is more representative of laboratory MFC tests. Current and COD concentrations were examined under both open and closed circuit conditions (two different external resistances) to better understand the competition by microorganisms for biodegradable COD in air-cathode MFCs, under these operational different conditions. 2. Methods 2.1. MFC construction and operation Air-cathode MFCs were cube-shaped reactors with a single cylindrical chamber (4 cm length, 26 mL net liquid volume), constructed as previously described (Logan et al., 2007). The anode was a graphite fiber brush made from a core of two twisted titanium wires that functioned as current collectors (Logan et al., 2007). The brush was heat treated in a muffle furnace at 450 °C for 30 min, and then placed in the middle of the MFC chamber. The air-cathode was made of 30% wet-proofed carbon cloth (Type B, BASF Fuel Cell, Inc., USA) with a platinum (0.5 mg/cm2) catalyst layer on the water side, and four polytetrafluoroethylene (PTFE) diffusion layers on the air side to prevent water loss (7 cm2 projected area) (Cheng et al., 2006). Anodes for acetate COD tests were inoculated with effluent from MFCs originally inoculated with domestic wastewater that were acclimated to sodium acetate (1 g/L, medium described below) for over one year. Anodes for domestic wastewater COD tests were inoculated and acclimated only using domestic wastewater from the primary clarifier overflow. All MFCs were operated under fed batch conditions at 30 ± 1 °C. 2.2. COD removal tests Domestic wastewater was collected from the primary clarifier overflow of the Pennsylvania State University Wastewater Treatment Plant, with a total COD (tCOD) of 439 ± 55 mg/L. Except as noted, wastewater was filtered (0.4 lm diameter, polycarbonate membrane, Poretics Corporation, USA), producing a final soluble COD (sCOD) of 223 ± 6 mg/L. The use of filtered wastewater provided a substrate that was more similar to dissolved acetate due
X. Zhang et al. / Bioresource Technology 176 (2015) 23–31
to removal of particles, and it avoided production of additional soluble substrate from particulate COD. In some additional tests, raw (unfiltered) domestic wastewater was also used to separately examine the effect of the particulate COD on removal rates. Acetate fed MFCs were supplied with a medium containing acetate in a 50 mM phosphate buffer solution (PBS; 4.57 g/L Na2HPO4, 2.45 g/L NaH2PO4H2O, 0.31 g/L NH4Cl and 0.13 g/L KCl) amended with 12.5 mL/L mineral and 5 mL/L vitamin solutions (Logan et al., 2007). Two different concentrations of acetate were used: 0.3 g/L sodium acetate (260 mg/L COD) to produce a COD similar to that of the sCOD of the local domestic wastewater; and 1 g/L sodium acetate (840 mg/L COD), which is a concentration typically used in MFC laboratory studies. Tests were first conducted using 1 g/L, followed by tests at 0.3 g/L. Three circuit conditions were examined with these solutions: an external resistance of 1000 O, which is commonly used for MFC start-up and operation (Zhang et al., 2013b); an external resistance of 100 O, as this produces a high current and typically the maximum power density (1000 mW/m2) in these MFCs when fed acetate; and an open circuit control, to examine the effect of substrate removal in the absence of current generation. In order to obtain COD concentrations of acetate in the reactor over time, and avoid leaving air in the reactor due to the removal of the solution, different feeding strategies were developed for acetate and wastewater samples. These procedures enabled a reasonable number of samples to be taken over time, while minimizing the need for a large number of reactors (Fig. S1). Since tests were conducted at the same time under three different conditions (open circuit, 1000 O and 100 O), a total of 12 reactors were used. For acetate, 4 MFCs were started at the same initial condition, with two of these four reactors sampled after 2, 6, 10, 23 and 48 h, and the other pair of duplicate reactors sampled at 4, 8, 12, 23, 48 h (i.e., two pairs of duplicate reactors sampled at different times). After a sample (2 mL) was taken, the COD was immediately determined, and solution was replaced with fresh media at the same COD based on the COD data of the other pair of reactors (Fig. S1). For example, at 4 h, after a sample was taken, the solution removed was replenished with new solution at the same COD of the sample at 2 h (Fig. S1). The same procedure was conducted from 2 to 12 h. At 23 h, a sample was taken and at 25 h the solution was replenished with the same COD based on the COD measurement at 23 h. At 48 h, the end of the test, a sample was taken and all the solution in the reactor was replaced by fresh influent for the next cycle. Using this approach, there was at least a 4 h interval between sampling from the same reactor. This procedure did introduce minor changes in COD added back into the reactors as the two concentrations were not identical. However, as the volume of sample withdrawn was small (2 mL) compared to the total solution volume (26 mL), it was assumed that this volume difference and only a small difference in COD would result in a relatively negligible error in the COD concentration in the reactors, compared to changes that would occur with leaving air in the reactor. A shorter interval between sampling of 2 h was also examined, but results were inconsistent (data not shown). For domestic wastewater COD tests, a different sampling strategy was developed due to need for a larger volume of sample for COD analysis. Total COD and soluble COD tests for raw wastewater required 5 mL per sample, which was not negligible relative to the total volume of solution. Therefore, wastewater-fed MFCs were sampled only once, and the reactor was not sampled again. Using 4 reactors, samples were taken in duplicate after 2, 4, 8, 10 and 24 h, and tests were repeated several times. COD removal can occur due to biofilms on both the anode brush and the cathode. Additional tests were therefore conducted to separately examine COD removal by the anode and cathode biofilms. Cathodes containing thick biofilms were removed from the MFCs
25
and placed in 4-cm length cubic reactors that lacked an anode. The end of the reactor that still contained the original anode was either sealed with an endplate to substantially reduce oxygen transfer into the reactor, or a new cathode was placed in the reactor with a glass fiber separator allowing some oxygen transfer into the reactor to better simulate conditions where oxygen might leak into the anode chamber when the cathode contained a biofilm (Zhang et al., 2010, 2009). COD removal rates were then examined in these reactors under open circuit conditions using a high concentration of sodium acetate (1 g/L). 2.3. Analysis of MFCs Voltages (U) were recorded across an external resistance (R, 1000 O or 100 O) at 20 min intervals using a data acquisition system (2700, Keithley Instrument, USA) connected to a personal computer. Current densities (J) were normalized by air-cathode projected area (A = 7 cm2) using J = U/RA (Logan et al., 2006). All CODs were measured using standard methods and a spectrophotometer (DR/2010, Hach Co., Loveland, CO, USA). COD removal data were analyzed using the Solver and Data Analysis Tools in Microsoft Excel (Professional Plus 2010). COD removal rates were fitted assuming a first-order reaction with respect to concentration, and calculated as follows
LnðCODt =COD0 Þ ¼ kt
ð1Þ
where COD0 is the influent COD, CODt the effluent COD, t is time, and k is the first order removal rate constant. While kinetics are assumed to be first-order with respect to concentration, they can vary with external resistance (Ren et al., 2014b). Coulombic efficiency (CE), defined as the fractional recovery of electrons from the substrate, was calculated (Logan et al., 2006) as:
Rt M 0 Idt CE ¼ nv FðCOD0 CODt Þ
ð2Þ
where M is the molecular weight of oxygen, I is the current, F is Faraday’s constant, n = 4 is the number of electrons exchanged per mole of oxygen, and v is the anolyte volume. 3. Results and discussion 3.1. COD removal rates with acetate COD removal rates with acetate at the low initial COD concentration (COD0 = 260 mg/L) showed good agreement with an assumption of first-order degradation kinetics, although rates were different depending on the circuit resistance (Fig. 1 and Fig. S2). For the low acetate concentration, the largest rate constant of 0.14 ± 0.03 h1 was obtained with the highest maximum current density produced with the lower external resistance of 100 O. At the lower current density produced using a higher resistance (1000 O), the rate constant (0.071 ± 0.008 h1) was about half that at the higher current density, and it decreased to 0.034 ± 0.002 h1 under open circuit (zero current) conditions (Fig. 1 and Fig. S2). The COD after 8 h of operation was 58 ± 11 mg/L with an external resistance of 100 O, compared to 108 ± 12 mg/L (1000 O) and 167 ± 10 mg/L (open circuit). After 8 h with the 100 O resistor, COD removal became very low resulting in a nearly constant COD of 58 ± 10 mg/L from 8 to 48 h. The final CODs after 48 h of operation were 57 ± 13 mg/L (78% removal, 100 O), 24 ± 15 mg/L (91% removal, 1000 O), and 47 ± 11 mg/L (82% removal, open circuit) (Fig. 1). These variations in the final CODs were not clearly related to the initial COD or circuit conditions, and they are all higher than reported in some previous studies. For example, Liu et al. (2005) reported >99% acetate removal (initial acetate concentration of
26
X. Zhang et al. / Bioresource Technology 176 (2015) 23–31
0
300
COD (mg/L)
250
Ln(COD/COD0) (Arbitrary)
Ac Low OCV Ac Low 1000 Ω Ac Low 100 Ω
200 150 100
A
50
0
10
20
30 Time (h)
40
50
-1 -1.5
B
-2
0
60
900
10
20
30 Time (h)
40
50
60
0
750
Ln(COD/COD0) (Arbitrary)
Ac High OCV
COD (mg/L)
-0.5
-2.5
0
Ac High 1000 Ω Ac High 100 Ω
600 450 300
C
150
-1 -2
Ac High OCV Ac High 1000 Ω Ac High 100 Ω N-Ac High 100 Ω Fit Ac High OCV Fit Ac High 1000 Ω Fit Ac High 100 Ω
-3 -4
0 0
10
20
30 Time (h)
40
50
0
60
Ln(sCOD/sCOD0) (Arbitrary)
WW OCV
250
WW 1000 Ω
200
WW 100 Ω
150 100
E
50 0 0
5
10
15 Time (h)
10
20
D 30
40
50
60
Time (h)
300
sCOD (mg/L)
Ac Low OCV Ac Low 1000 Ω Ac Low 100 Ω N-Ac Low 1000 Ω N-Ac Low 100 Ω Fit Ac Low OCV Fit Ac Low 1000 Ω Fit Ac Low 100 Ω
20
25
30
0 -0.2 -0.4 -0.6 -0.8 -1 -1.2 -1.4 -1.6 -1.8
WW OCV WW 1000 Ω WW 100 Ω N-WW OCV N-WW 1000 Ω N-WW 100 Ω Fit WW OCV Fit WW 1000Ω Fit WW 100 Ω
F 0
5
10
15 Time (h)
20
25
30
Fig. 1. COD of low (A) and high (C) concentrations of acetate (Ac) in MFCs changes with time; soluble COD (sCOD) of filtered domestic wastewater changes with time in the MFCs initially acclimated to wastewater (WW) (E). (B), (D), and (F): first order reaction fitting data; only the data with filled symbols were used to fit the first order relationship, and the data with open symbols were not (N), as the COD removal had essentially ceased after current generation stopped with 1000 O and 100 O (shown in Fig. 3). The MFCs were operated under three conditions including open circuit (OCV), 1000 O and 100 O external resistors.
800 mg/L) using an air-cathode MFC, while Freguia et al. (2007) reported 100% acetate removal using a ferricyanide catholyte. In both of these studies (Freguia et al., 2007; Liu et al., 2005), acetate concentrations were measured rather than COD, and thus some of the remaining COD may not be acetate. COD was used here so that the same methods were used for acetate and wastewater tests. Also, the different reactor conditions, such as flat, widely spaced electrodes (Liu et al., 2005) compared to brush electrodes placed close to the cathodes, and much lower current densities than those achieved here, could have affected the development of microbial communities and their responses to low substrate conditions and oxygen transfer from the cathode into solution. For the high acetate concentration (COD0 = 840 mg/L), the rate constants again showed agreement with first-order kinetics, with rate constants that increased with current. The highest rate constant was 0.086 ± 0.006 h1 (100 O, high current), followed by 0.065 ± 0.002 h1 (1000 O), and 0.030 ± 0.002 h1 (open circuit) (Fig. 1 and Fig. S2). After 24 h of operation, the CODs in the MFCs were 99 ± 51 mg/L with the 100 O circuit, compared to 188 ± 79 mg/L (1000 O) and 323 ± 112 mg/L (open circuit). After 48 h of operation, the final COD under open circuit was 184 ± 81 mg/L (78% removal), which was much higher than that obtained with current generation of 59 ± 18 mg/L (93% removal, 100 O) and 33 ± 5 mg/L (96% removal, 1000 O) (Fig. 1). In these
tests at the higher COD concentration, it was shown that current production accelerated COD removal compared to that with no current (open circuit), and that the final COD was appreciably lower with current generation than the open circuit control. While the COD removal rates can be seen to obey first-order kinetics for any specific initial COD concentration, the overall COD removal rate depended on operating conditions. In some cases, the rate constant was unchanged based on the initial COD. For example, under open circuit conditions (no current), the rate constants were the same (within a standard deviation) at the low initial COD of 260 mg/L (0.034 ± 0.002 h1) and high COD of 840 mg/L (0.030 ± 0.002 h1). Similarly, with the high resistance of 1000 O (current data discussed below), the rate constants were also similar at both initial COD concentrations (0.071 ± 0.008 h1, 260 mg/L; and 0.065 ± 0.002 h1, 840 mg/L). However, when the lower resistance of 100 O was used, rate constants were significantly increased. At the initial COD of 840 mg/L, the rate constant was 0.086 ± 0.006 h1, which was 39% lower than that obtained at an initial COD concentration of 260 mg/L (0.14 ± 0.03 h1). If the overall rate remained truly first order under all conditions, the rate concentration would be independent of the initial concentration, as shown in Eq. (1). Thus, COD removal rates were appreciably affected by initial COD concentrations at the lowest resistance (higher initial current densities).
X. Zhang et al. / Bioresource Technology 176 (2015) 23–31
In all these tests, there was no indication of saturation kinetics relative to COD removal. However, the current did not always appreciably decrease following acetate addition, and thus current production did not vary in proportion to COD removal. Thus, the rate of COD removal in a fed batch reactor for these substrate concentrations could be described by first order kinetics, with the recognition that rates were dependent on the organic loading range and external resistance (circuit load). 3.2. COD removal rates with domestic wastewater COD removal rates with filtered domestic wastewater in the MFCs inoculated with domestic wastewater were also well described by first order kinetics, with rate constants increased by current production (Fig. 1 and Fig. S2). The rate constants with current generation using filtered wastewater were 0.18 ± 0.03 h1 (100 O) and 0.17 ± 0.03 h1 (1000 O). These rate constants are 80% (100 O) and 70% (1000 O) higher than those obtained under open circuit conditions (0.10 ± 0.01 h1). The final CODs were similar with and without current generation, but the time needed for treatment varied due to differences in rates. The CODs with current generation decreased from 214 ± 2 mg/L to 51 ± 4 mg/L (100 O) and 52 ± 5 mg/L (1000 O) after 8 h, and did not further decrease from 8 to 24 h. The final COD of the open circuit controls was similar at 56 ± 2 mg/L when the experiment was stopped, but this extent of treatment required 24 h. After 8 h, under open circuit conditions, the COD was 84 ± 2 mg/L. These final COD concentrations were comparable to those previously obtained using MFCs fed domestic wastewater (60 to 80 mg/L total COD and soluble COD) (Hays et al., 2011). In those studies, the final BOD5 was estimated to be 30 to 40 mg/L. Previous research has shown that reactors acclimated to wastewater can be effectively used with other wastewaters (Ren et al., 2013). However, the converse was not found to be true here: long-term acclimation of the reactor to acetate showed very low COD removal rates when tested with domestic wastewater. Reactors fed acetate for over one year, exhibited similar and very low COD removal rates with domestic wastewater, with all rate constants [0.02 h1 (Fig. S2). These low rate constants showed that long-term acclimation to acetate hindered the ability of the anode to degrade the more complex substrates in filtered wastewater. 3.3. Comparison of removal rates in acetate and wastewater A comparison of the first-order rate constants shows that COD removal rate constants with domestic wastewater were actually higher than those of acetate at each comparable condition (100 O, 1000 O, or open circuit) (Fig. 2). This higher rate constant for the wastewater was surprising given that acetate is considered to be readily biodegradable compared to the more complex organic matter in domestic wastewater, and acetate produces higher power densities than wastewater. The rate constants obtained here for acetate are lower than those previously reported for smaller but quite similar MFCs, likely as a result of different electrode specific surface areas (electrode area per anolyte volume). Ren et al. obtained a 1.4 higher rate constant (0.33 ± 0.02 h1) with current (100 O), and a 3.1 higher rate constant under open circuit conditions (0.14 ± 0.01 h1) using the same sized cathodes as those used here, but tests were conducted with a smaller reactor volume (14 mL) (Ren et al., 2014b). Therefore, the cathode specific surface area in that study was 64 m2/m3 (based on liquid volume) or 2.4 that used here (27 m2/m3). This higher electrode surface area per volume enabled much more rapid depletion of substrate, particularly under open circuit conditions by aerobic heterotrophs using oxygen transfer
27
into the reactor through the cathode. These different rate constants based on the electrode surface area per volume highlight the importance of the specific electrode area when evaluating COD removal rates. 3.4. Coulombic efficiency For both acetate and domestic wastewater, the CEs over time were on average higher in tests with lower resistances (higher current densities) (Fig. 2). For acetate, the CEs were also higher for the lower initial COD, under both 1000 O and 100 O. The acetate-fed MFCs with COD0 = 260 mg/L and 100 O achieved the highest CE of 90% (Fig. 2), indicating that exoelectrogenic microorganisms outcompeted other microorganisms for substrates as 90% of the available electrons in the removed substrate were used for current production. The rest of the removed substrate was degraded by bacteria using oxygen as an electron acceptor, as further discussed below. During the fed-batch cycle with acetate, the CEs also increased over time under all conditions (Fig. 2). The CEs for the filtered domestic wastewater, were 22% (100 O) and 14% (1000 O), which are both lower than those obtained with acetate (Fig. 2). It is commonly observed that CEs are lower for domestic wastewater compared to acetate (Ahn and Logan, 2010; Liu et al., 2004). The CEs for the filtered wastewaters were essentially constant or they slightly decreased over time, which was different from that observed using acetate. 3.5. Rapid decreases in current production at low CODs A comparison of current densities and COD concentrations in the MFCs over time (during a fed-batch cycle) shows that there was a larger impact of COD concentration on the current densities produced, than the rates of COD removal (Fig. 3). The effect of the COD concentration on current can more directly be seen by plotting current as a function of COD (rather than each variable versus time), as shown in Fig. 4. When the COD concentration decreased to less than 100 mg/L for acetate or filtered domestic wastewater, current production rapidly decreased (Fig. 4). This significant loss in current production could be due to a combination of factors, such as mass transfer limitations or increased dissolved oxygen concentrations due to reduced dissolved oxygen removal at lower substrate concentrations. The final CODs in these tests (50 mg/L for acetate and filtered domestic wastewater) were greater than CODs typically needed for wastewater discharge to the environment (Fig. 4). At 100 O, the current densities decreased rapidly when the COD was below only 200 mg/L, and the current was very low even there was still 100 mg/L COD for both acetate and filtered domestic wastewater (Fig. 4). For both acetate and wastewater, with either resistance, current generation was minimal well before the substrate in the reactor was fully depleted. This indicates that it will be difficult to maintain high power production in an MFC if the goal is to reduce the effluent COD levels to values suitable for wastewater discharge (
