Accepted Manuscript Impact of Electrode Configurations on Retention Time and Domestic Wastewater Treatment Efficiency Using Microbial Fuel Cells Kyoung-Yeol Kim, Wulin Yang, Bruce E. Logan PII:
S0043-1354(15)00300-0
DOI:
10.1016/j.watres.2015.05.021
Reference:
WR 11296
To appear in:
Water Research
Received Date: 20 March 2015 Revised Date:
5 May 2015
Accepted Date: 7 May 2015
Please cite this article as: Kim, K.-Y., Yang, W., Logan, B.E., Impact of Electrode Configurations on Retention Time and Domestic Wastewater Treatment Efficiency Using Microbial Fuel Cells, Water Research (2015), doi: 10.1016/j.watres.2015.05.021. 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.
RI PT
ACCEPTED MANUSCRIPT
SC
1
AC C
EP
TE D
M AN U
2
1
ACCEPTED MANUSCRIPT
1 2 3
Date: March 17, 2015 Submitted to: Water Research
Impact of Electrode Configurations on Retention Time and Domestic
5
Wastewater Treatment Efficiency Using Microbial Fuel Cells
RI PT
4
6
Kyoung-Yeol Kim, Wulin Yang, and Bruce E. Logan*
8
Department of Civil and Environmental Engineering, The Pennsylvania State University, 231Q
9
Sackett Building, University Park, PA 16802, USA
10
*Corresponding author: e-mail:
[email protected]; phone: +1-814-863-7908; fax: +1-814-863-7304
M AN U
SC
7
11 Abstract
13
Efficient treatment of domestic wastewater under continuous flow conditions using microbial
14
fuel cells (MFCs) requires hydraulic retention times (HRTs) that are similar to or less than those
15
of conventional methods such as activated sludge. Two MFCs in series were compared at
16
theoretical HRTs of 8.8, 4.4 and 2.2 h using two different brush-electrode MFC configurations: a
17
full brush evenly spaced between two cathodes (S2C); and trimmed brush anodes near a single
18
cathode (N1C). The MFCs with two cathodes produced more power than the MFCs with a single
19
cathode, with 1.72 mW for the S2C, compared to and 1.12 mW for the N1C at a set HRT = 4.4 h.
20
The single cathode MFCs with less cathode area removed slightly more COD (54.2 ± 2.3%, N1C)
21
than the two-cathode MFCs (48.3 ± 1.0%, S2C). However, the higher COD removal was due to
22
the longer HRTs measured for the MFCs with the N1C configuration (10.7, 5.3 and 3.1 h) than
23
with the S2C configuration (7.2, 3.7 and 2.2 h), despite the same theoretical HRT. The longer
AC C
EP
TE D
12
1
ACCEPTED MANUSCRIPT
HRTs of the N1C MFCs also resulted in slightly higher coulombic efficiencies (≤ 37%) than
25
those of the S2C MFCs (≤ 29%). While the S2C MFC configuration would be more
26
advantageous based on electrical power production, the N1C MFC might be more useful on the
27
basis of capital costs relative to COD removal efficiency due to the use of less cathode surface
28
area per volume of reactor.
RI PT
24
29
Keywords: Microbial fuel cell; Continuous flow mode; Brush anode electrode; Flow pattern;
31
Hydraulic retention time
SC
30
33
M AN U
32 1. Introduction
Microbial fuel cells (MFCs) have been studied as an energy-neutral process which can
35
produce electrical energy from organic compounds while accomplishing wastewater treatment
36
(Li et al. 2014, Rabaey and Verstraete 2005, Zhang et al. 2013). However, recent studies have
37
shown that both high power densities and low effluent concentrations of chemical oxygen
38
demand (COD) cannot be simultaneously achieved in MFCs under continuous flow conditions
39
(Akman et al. 2013, Nam et al. 2010, Zhuang et al. 2012). When the COD concentration is
40
reduced to a range of ~100 mg/L, there is minimal current generation relative to higher COD
41
concentrations (Ren et al. 2014a, Zhang et al. 2015). Thus, a secondary treatment process such as
42
an anaerobic fluidized bed membrane bioreactor (AFMBR) is needed to reduce the COD to
43
levels suitable for discharge (Ren et al. 2014b). The goal of MFC treatment is therefore not COD
44
removal to a level suitable for discharge, but COD reduction to levels that maintain good power
45
production. The low COD levels needed for discharge can be achieved using AFMBRs or other
AC C
EP
TE D
34
2
ACCEPTED MANUSCRIPT
46
types of processes such as biofilters or solids contact systems (Chernicharo 2006, Kim et al.
47
2011). The materials used in larger-scale MFCs for wastewater treatment need to be inexpensive,
49
effective for treatment of actual wastewaters that contain particulate organic matter, and produce
50
stable power over time. Flat anodes made of carbon cloth or mesh can allow close electrode
51
spacing and high power densities with acetate solutions (Liu et al. 2005), but they have been
52
found to result in unstable power production when used in MFCs treating domestic wastewater
53
(Hays et al. 2011) and small electrode spacing could result in reactor clogging. Graphite fiber
54
brush anodes can be used in MFCs to achieve good rates of COD removal, they have high
55
surface areas and porosities for growth of exoelectrogenic microorganisms, and they produce
56
stable power over time using actual domestic wastewaters (Ahn and Logan 2013, Logan et al.
57
2007). MFCs have been designed and tested with multiple brush anodes configured primarily in
58
two ways: with space between the anodes and the cathodes in the presence and absence of a
59
separator; or with brushes placed directly on top of the separator with little space between the
60
electrodes (Ahn et al. 2014, Zhang et al. 2014a). The two configurations have different
61
advantages relative to power production, but separators are needed for both configurations to
62
reduce cathode biofouling (Ren et al. 2014b). Cathode materials and performance are also
63
important factors in MFC designs. Activated carbon cathodes have been shown to produce power
64
densities similar to, or even larger than, cathodes made with a Pt catalyst (Cheng and Wu 2013,
65
Zhang et al. 2009, Zhang et al. 2014b). The performance of activated carbon is maintained much
66
better over time compared to Pt, and power production can be nearly restored to levels
67
comparable to new cathodes using an acid treatment. Cathode performance typically limits
AC C
EP
TE D
M AN U
SC
RI PT
48
3
ACCEPTED MANUSCRIPT
68
power generation in MFCs at higher COD concentrations. Doubling the cathode surface area, for
69
example, increased maximum power densities from domestic wastewater by 62% in fed-batch
70
tests (Cheng and Logan 2011). In order for large-scale wastewater treatment using MFCs to be feasible using these low-cost
72
brush anodes and activated carbon cathodes, the hydraulic retention times (HRTs) in the reactors
73
has to be reduced to be more similar to that used for conventional activated sludge processes.
74
However, there have been relatively few studies focused on achieving wastewater treatment with
75
HRTs in the range of ~10 h or less using domestic wastewater (Min and Logan 2004, Puig et al.
76
2011). Short HRTs can be accomplished by improving current densities, and therefore achieving
77
higher rates of COD removal. Two different graphite fiber brush anode reactor configurations
78
were examined here for wastewater treatment at HRTs of less than 10 h: anodes (trimmed flat on
79
one side) placed directly on top of a separator covering a single cathode so that they are very near,
80
in order to reduce solution ohmic resistance (N1C), resulting in no flow between the electrodes;
81
and anodes (full brush) positioned in the middle of the chamber with space for flow between the
82
brush and the two cathodes each covered by a separator (S2C). The two types MFCs were
83
operated under continuous flow conditions at three different theoretical HRTs (8.8, 4.4, and 2.2 h)
84
using two MFCs in series to minimize the change in COD in a single reactor. Power generation
85
and COD removal were monitored over time, with maximum power densities evaluated using
86
polarization data. The different anode configurations can affect flow patterns in these reactors,
87
and thus the actual HRTs were measured using a salt tracer.
89
SC
M AN U
TE D
EP
AC C
88
RI PT
71
2. Materials and methods 4
ACCEPTED MANUSCRIPT
90
2.1. MFC construction
91
Anodes were graphite fiber brushes (Mill-Rose, Mentor, OH) with a titanium core (25 mm
92
diameter by 35 mm length) heat treated at 450 °C for 30 min before use. Cathodes (40 cm2,
93
projected
94
poly(vinylidenefluoride, PVDF) binder and a mixture of activated carbon (AC, 8.8 mg/cm2) and
95
carbon black (CB, Vulcan XC-72, Cabot Corporation, USA) as previously described (Yang et al.
96
2014).
were
made
by
a
phase
inversion
technique
using
RI PT
area)
a
SC
surface
Four single-chamber, air-cathode MFCs were constructed using three graphite fiber brush
98
anodes per reactor. Two of these reactors had the anodes placed directly next to a single cathode
99
(N1C) similar to previous designs (Ahn et al. 2014) (Fig. 1). The other two MFCs had the anodes
100
placed in the middle of the anode chamber with a space between the edge of the anodes and two
101
cathodes on either side of the brushes (S2C). The spaced electrode reactor design is different
102
from previous spaced electrode designs as two cathodes each containing separators were used
103
per anode array, compared to MFCs tested with a single cathode without a separator (Zhang et al.
104
2014a). Brushes used in the N1C reactor (100 mL empty bed volume) were trimmed along their
105
length to form a half cylinder, with the flat side placed on the separator that was positioned
106
against the air-cathode (0.5 cm distance from the cathode). Full brushes were used in the S2C
107
reactors (140 mL empty bed volume) with an edge-cathode distance of 0.8 cm (Fig. 1). These
108
different configurations resulted in a higher cathode surface area (57 m2/m3) for the S2C reactor
109
than the N1C reactors (40 m2/m3). Each reactor had an entrance and exit chamber (no electrodes)
110
of 10 mL each to help distribute flow evenly into the reactors (Fig. S1). Two layers of a textile
111
separator (46% cellulose, 54% polyester; 0.3 mm thick; Amplitude Prozorb, Contec Inc.) were
AC C
EP
TE D
M AN U
97
5
ACCEPTED MANUSCRIPT
used as separators on each cathode surface (liquid side) to reduce cathode biofouling and oxygen
113
intrusion into the anode chamber, and in the case of the N1C reactors, to prevent electrode short-
114
circuiting. A reference electrode (Ag/AgCl; + 200 mV vs. standard hydrogen electrode, SHE)
115
was installed in the upper and middle part of the MFCs to record electrode potentials (all
116
potentials reported versus SHE).
RI PT
112
117 2.2. Operational methods
SC
118
The four MFC reactors (2 N1Cs, 2 S2Cs) were arranged with two flow paths, with the two
120
similar types of MFCs hydraulically connected in series (Fig. 1). The first MFC was the labeled
121
the upstream (U) reactor, and the second one the downstream (D) reactor. The MFCs were
122
inoculated and fed domestic wastewater from the primary clarifier of the Pennsylvania State
123
University Wastewater Treatment Plant. Wastewater was collected once a week and stored in a
124
refrigerator (4 °C). When used, the wastewater was placed next to the reactor in a container filled
125
with ice, which preserved the wastewater composition and allowed the solution to warm as it was
126
pumped into the upstream MFCs using peristaltic pumps (model no.7523-90, Masterflex, Vernon
127
Hills, IL).
EP
TE D
M AN U
119
The theoretical HRT was calculated from the set flow rate (Q, mL/h) and reactor volume (V,
129
mL) as HRT = V/Q. Flow rates were chosen to obtain the same theoretical HRTs for the two
130
reactors of: 8.8 h (31.8 mL/h, S2C; 22.8 mL/h, N1C); 4.4 h (63.6 mL/h, S2C; 45.6 mL/h, N1C);
131
2.2 h (127.2 mL/h, S2C; 91.2 mL/h, N1C). A salt tracer test was conducted to measure the actual
132
residence time of solutes in the two pairs of MFCs during continuous mode operation. A NaCl
133
solution (1 M) was injected into the flow line of the upstream flow line of the MFC for 36 min
AC C
128
6
ACCEPTED MANUSCRIPT
(8.8 h HRT), 18 min (4.4 h HRT), or 9 min (2.2 h HRT), with the effluent conductivity monitored
135
using a meter (SB90M5, VWR International, USA). Measured HRTs were defined as the time to
136
the peak of the tracer input as previously described (Min et al. 2004, Zhang et al. 2006). The
137
HRT measured using this approach includes the entrance and exit volumes (not working area) of
138
the reactors. Thus, the measured HRTs were adjusted by multiplying a ratio of working volume
139
to total reactor volume. A dispersion number (d) was calculated from the normalized variance of
140
the salt tracer curve. A Peclet number (Pe) was calculated as Pe = 1/d, where Pe = uL/E is a ratio
141
of mass transfer by advection to that of dispersion, where u is the average cross-sectional
142
velocity, L is the distance of flow, and E is the dispersion coefficient.
143 144
2.3. Analysis and calculations
M AN U
SC
RI PT
134
Cell voltages (external resistor of 200 Ω) of each MFC were monitored at every 10 min using
146
a multimeter (Model 2700, Keithley Instruments, USA) and used to calculate the current based
147
on the voltage drop across the resistor. Polarization and power curves were obtained by varying
148
the external resist from open circuit to 50 Ω. Reactors were first operated under open circuit
149
conditions for at least 4 h prior to each polarization test (Kim et al. 2014).
EP
150
TE D
145
Coulombic efficiency (CE) was calculated as CE (%) = Ct/Cth
100, where Ct was the total
coulombs calculated by integrating the current over time (Ct = ∑I ∆t, where ∆t is the time
152
interval of one HRT), and Cth was the theoretical amount of coulombs available based on the
153
COD removed in the MFC over the same amount of time, calculated as Cth = [Fb (CODin –
154
CODout) Q ∆t]/M, where F is Faraday’s constant, b = 4 is the number of electrons exchanged per
155
mole of oxygen, CODin and CODout are the influent and effluent COD, ∆t is the time interval
AC C
151
7
ACCEPTED MANUSCRIPT
156
(HRT), and M = 32 is the molecular weight of oxygen (Logan et al. 2006). Total COD was
157
measured following standard methods (method 5220, HACH company, Loveland, CO).
159
3. Results and discussion
160
3.1. Power generation as a function of theoretical HRT
RI PT
158
When the MFCs were started up at the longest HRT of 8.8 h (200 Ω external resistance),
162
there was inconsistent power produced for the similar reactor configurations (Fig. 2). For
163
example, the S2C upstream MFC produced the most power, but the downstream S2C produced
164
the least power. When operation was switched to shorter HRTs of 4.4 and 2.2 h, there was more
165
consistent power production, with the S2C MFCs consistently producing more power (1.36 mW
166
at 4.4 h, and 1.30 mW at 2.2 h) than the N1C MFCs (1.01 mW at 4.4 h, and 1.22 mW at 2.2 h).
167
The upstream MFCs generally produced greater power than the downstream MFCs, likely due to
168
the higher COD concentrations in the upstream MFCs. After these tests at the two shortest HRTs,
169
the reactors were switched back to the longer 8.8 h HRT, and a fresh wastewater sample was
170
used. This time at this longer HRT, the MFCs produced more consistent power densities based on
171
reactor configuration, with power by the two-cathode MFCs (1.14 mW, S2C) larger than that
172
produced by the single-cathode MFCs (0.85 mW, N1C).
EP
TE D
M AN U
SC
161
Power and polarization curves (Fig. 3) also demonstrated that higher power could be
174
produced by the S2C MFCs than the single cathode N1C MFCs at all HRTs. For example, at the
175
shortest HRT of 2.2 h, the maximum power of the two S2C MFCs was 1.67 mW, compared to
176
1.20 mW for the N1C MFCs. In most cases, similar power densities were produced by both the
177
upstream and downstream MFCs. At the shortest theoretical HRT, however, the maximum power
AC C
173
8
ACCEPTED MANUSCRIPT
produced by the downstream MFCs based on these polarization data was equal to or slightly
179
higher than that of the upstream MFCs. The largest difference was obtained at an HRT of 2.2 h
180
where the downstream S2C produced 0.89 ± 0.03 mW compared to 0.78 ± 0.04 mW for the
181
upstream reactor (Fig. 3c). For the N1C reactors, the downstream reactor produced 0.65 ± 0.03
182
mW, compared to 0.55 ± 0.02 mW for the upstream MFC. Together, these averaged 209 mW/m2
183
for the S2C and 300 mW/m2 for the N1C. There were smaller differences between the upstream
184
and downstream MFCs at the longer HRTs, with average power produced of 1.72 mW (S2C) and
185
1.12 mW (N1C) at a 4.4 h HRT, and 1.61 mW (S2C) and 1.06 mW (N1C) at an 8.8 h HRT. There
186
are several possible reasons for the slightly higher maximum power densities in the downstream
187
MFCs. Polarization data represent the capability of the MFCs for power production based on
188
short-term tests, but with longer acclimation the power densities may have been more similar.
189
For example, in Figure 2 it was shown that under steady flow conditions, and at the same
190
external resistance, the upstream reactor produced higher voltages than the downstream reactors.
191
Power overshoot, where the power density curve doubles back to lower current densities
192
oftentimes before a higher power density that is possible with further reactor acclimation, could
193
also be a factor as overshoot was observed in several tests (Fig. 3). Differences in the
194
composition and quality of the wastewater in the upstream and downstream reactors could also
195
be important. Domestic wastewater contains both soluble and particulate COD, and it thus there
196
may be more fermentation and particle hydrolysis occurring in the upstream reactor, with more
197
labile and biodegradable organics in the downstream reactor.
AC C
EP
TE D
M AN U
SC
RI PT
178
198
The higher power production of the S2C MFCs with two cathodes is consistent with previous
199
results showing that increasing cathode area can result in higher power densities (Cheng and 9
ACCEPTED MANUSCRIPT
Logan 2011). The S2C reactors had 43% more surface area (57 m2/m3) than the N1C reactors (40
201
m2/m3), and produced 39-53% more power on average at the three different flow rates based on
202
polarization data. In contrast, only a difference of 16% was observed previously when the three
203
anodes were placed closer to a single cathode (Ahn et al. 2014). Thus, the closer placement of
204
the anode to the cathode in the N1C reactors, which would reduce ohmic resistance, was less
205
important for higher power production than the use of more cathode surface area.
RI PT
200
207
SC
206 3.2. COD removals as a function of HRTs
COD removals with the two-cathode S2C MFCs decreased in proportion to HRTs, with
209
removals of 64.8 ± 1.7% at an HRT of 8.8 h, decreasing to 48.3 ± 1.0% at 4.4 h, and 32.8 ± 1.9%
210
at 2.2 h (Fig. 4). It was expected that the S2C MFC would have greater COD removal than the
211
N1C MFC due to higher current densities (Zhang et al. 2015) and greater cathode surface area
212
per volume of reactor (which would allow more leakage of oxygen into the reactor) at the same
213
set HRT. However, the COD removals with the single-cathode N1C MFC were higher than those
214
of the S2C MFCs, with removals of 69.0 ± 0.4% at 8.8 h, 54.2 ± 2.3% at 4.4 h, and 40.7 ± 2.7%
215
at 2.2 h. This is 6.4 to 24.4% greater COD removal for the single cathode MFCs than the two-
216
cathode MFCs at each of these set HRTs.
218
TE D
EP
AC C
217
M AN U
208
3.3. Effect of reactor configuration on HRTs in continuous flow mode MFCs
219
The unexpected results showing better COD removals for single cathode MFCs compared to
220
the two-cathode suggested that configuration of the MFCs may have produced HRTs different
221
than the theoretical ones (calculated from reactor empty bed volume and flow rate). Therefore, 10
ACCEPTED MANUSCRIPT
salt tracer tests were conducted to measure the actual residence time of solutes through the two
223
types (S2C and N1C) of reactors. The results showed that the actual HRTs of S2C reactors were
224
less than or equal to the theoretical HRTs, with actual:theoretical HRTs of 7.4:8.8 h, 3.8:4.4 h,
225
and 2.2:2.2 h (Table 1). A slight reduction in HRT due to short circuiting in these MFCs was
226
likely a result of the reactor volume occupied by the brushes as there may not have been
227
advective flow through the brushes. However, the HRTs of the two N1C reactors were longer
228
than those measured for the S2C reactors by 39% (7.4:10.3 h for S2C:N1C), 37% (3.8:5.2 h), and
229
32% (2.2:2.9 h). Thus, the greater removal by the N1C reactors was likely due to the much
230
longer HRTs produced using this configuration compared to that obtained with the S2C
231
configuration.
M AN U
SC
RI PT
222
A comparison of the COD removals on the basis of the measured HRTs shows that there was
233
generally good agreement with the observed COD removals and measured HRTs despite
234
differences in the two reactor configurations (Fig. 5). The COD removals for both types of
235
reactors show good agreement with measured HRTs based on a non-linear regression. However,
236
there was little agreement between the two configurations in terms of COD removal on the basis
237
of the theoretical HRTs, with slightly better COD removals indicated to occur for the N1C MFC
238
than the S2C MFC. These results suggest that the COD removal was more a function of the
239
HRTs, and thus how the HRT was altered by the use of the two different brush configurations,
240
rather other factors such as cathode surface area. The main impact of the reactor geometry was
241
on power production, with more power generated when using two cathodes than one cathode
242
with closely-spaced electrodes.
243
AC C
EP
TE D
232
The low dispersion numbers and high Peclet numbers indicate that flow through the reactors 11
ACCEPTED MANUSCRIPT
was consistent with plug flow reactors with dispersion. For both reactor configurations there was
245
little dispersion, with the dispersion increasing slightly with flow rate (Table 1). Given the high
246
Peclet numbers at the longest theoretical HRT, the reactors could be modeled as a series of
247
completely mixed, constant flow reactors (CSTRs). For example, the S2C MFCs (Pe = 36) could
248
be modeled as 19 CSTRs in series, with a rate constant of –0.0066 h–1 (Logan 1999).
RI PT
244
249 3.4. CEs as a function of the HRTs
SC
250
The highest CE obtained was 36 ± 2% at theoretical HRT of 8.8 h for the two N1C MFCs in
252
series (Fig. 6). The CEs ranged from 18 ± 2 to 36 ± 2% for the N1C MFCs, compared to 18 ± 5
253
to 29 ± 3% for the S2C MFCs. While the CEs were generally higher for N1C MFCs for each
254
pairwise comparison of the two reactor configurations on the basis of theoretical HRT, as noted
255
above the measured HRTs were longer for the N1C MFCs than the S2C MFCs. The downstream
256
MFCs consistently had higher CEs than the upstream MFCs (Fig. 6) although there was a much
257
greater percentage of COD removal in the upstream than downstream MFCs (Fig. 4). This
258
indicated that more of the COD removed in the upstream MFC was not used for electricity
259
generation, and was lost to alternate electron acceptors, such as in dissolved oxygen. The CEs are
260
known to be much lower in MFCs with domestic wastewater than with defined media such as
261
acetate in a phosphate buffer solution.
263
TE D
EP
AC C
262
M AN U
251
4. Conclusions
264
Brush anode and activated carbon cathodes can be used as effective and inexpensive
265
electrode materials in MFCs. The use of closely spaced electrodes and high cathode specific 12
ACCEPTED MANUSCRIPT
surface areas have been shown to improve power generation in MFCs, but little attention has
267
been paid to treatment performance in these systems. Higher power production was achieved
268
using reactors with greater cathode specific surface area than reactors with a configuration based
269
on close electrode spacing. However, there was little difference in COD removals in these two
270
reactor designs on the basis of measured HRTs. The MFC with the closely spaced electrodes had
271
longer measured HRTs, and thus slightly greater COD removals than MFCs with spaced
272
electrodes despite the same set HRT. This different results in terms of power and COD removal
273
suggest that there is a tradeoff in MFC design. The use of more cathode surface area per volume
274
of reactor will increase power production, but that design would have a higher capital cost due to
275
the use of more cathodes. It might be that the use of less cathode area per reactor would be more
276
cost effective than the recovery of more electrical power. Thus, the benefits of power production
277
will need to be considered relative capital costs and advantages of the different reactor designs.
M AN U
SC
RI PT
266
TE D
278 Acknowledgments
280
This research was supported by the grant from the Strategic Environmental Research and
281
Development Program (SERDP).
282
EP
279
References
284
Ahn, Y. and Logan, B. (2013) Domestic wastewater treatment using multi-electrode continuous
285
flow MFCs with a separator electrode assembly design. Appl. Microbiol. Biotechnol. 97, 409-
286
416.
287
Ahn, Y., Hatzell, M.C., Zhang, F. and Logan, B.E. (2014) Different electrode configurations to
AC C
283
13
ACCEPTED MANUSCRIPT
optimize performance of multi-electrode microbial fuel cells for generating power or treating
289
domestic wastewater. J. Power Sources 249, 440-445.
290
Akman, D., Cirik, K., Ozdemir, S., Ozkaya, B. and Cinar, O. (2013) Bioelectricity generation in
291
continuously-fed microbial fuel cell: Effects of anode electrode material and hydraulic retention
292
time. Bioresour. Technol. 149, 459-464.
293
Cheng, S. and Logan, B.E. (2011) Increasing power generation for scaling up single-chamber air
294
cathode microbial fuel cells. Bioresour. Technol. 102(6), 4468-4473.
295
Cheng, S. and Wu, J. (2013) Air-cathode preparation with activated carbon as catalyst, PTFE as
296
binder and nickel foam as current collector for microbial fuel cells. Bioelectrochemistry 92, 22-
297
26.
298
Chernicharo, C.D. (2006) Post-treatment options for the anaerobic treatment of domestic
299
wastewater. Rev. Environ. Sci. Bio/Technol. 5(1), 73-92.
300
Hays, S., Zhang, F. and Logan, B.E. (2011) Performance of two different types of anodes in
301
membrane electrode assembly microbial fuel cells for power generation from domestic
302
wastewater. J. Power Sources 196(20), 8293-8300.
303
Kim, J., Kim, K., Ye, H., Lee, E., Shin, C., McCarty, P.L. and Bae, J. (2011) Anaerobic fluidized
304
bed membrane bioreactor for wastewater treatment. Environ. Sci. Technol. 45(2), 576-581.
305
Kim, K.-Y., Yang, E., Lee, M.-Y., Chae, K.-J., Kim, C.-M. and Kim, I.S. (2014) Polydopamine
306
coating effects on ultrafiltration membrane to enhance power density and mitigate biofouling of
307
ultrafiltration microbial fuel cells (UF-MFCs). Water Res. 54, 62-68.
308
Li, W.-W., Yu, H.-Q. and He, Z. (2014) Towards sustainable wastewater treatment by using
309
microbial fuel cells-centered technologies. Energy Environ. Sci. 7(3), 911-924.
AC C
EP
TE D
M AN U
SC
RI PT
288
14
ACCEPTED MANUSCRIPT
Liu, H., Cheng, S. and Logan, B.E. (2005) Power generation in fed-batch microbial fuel cells as
311
a function of ionic strength, temperature, and reactor configuration. Environ. Sci. Technol.
312
39(14), 5488-5493.
313
Logan, B.E. (1999) Environmental Transport Processes, Wiley, New York.
314
Logan, B.E., Hamelers, B., Rozendal, R., Schröder, U., Keller, J., Freguia, S., Aelterman, P.,
315
Verstraete, W. and Rabaey, K. (2006) Microbial fuel cells: methodology and technology. Environ.
316
Sci. Technol. 40(17), 5181-5192.
317
Logan, B.E., Cheng, S., Watson, V. and Estadt, G. (2007) Graphite fiber brush anodes for
318
increased power production in air-cathode microbial fuel cells. Environ. Sci. Technol. 41(9),
319
3341-3346.
320
Min, B., Evans, P.J., Chu, A. and Logan, B.E. (2004) Perchlorate removal in sand and plastic
321
media bioreactors. Water Res. 38(1), 47-60.
322
Min, B. and Logan, B.E. (2004) Continuous electricity generation from domestic wastewater and
323
organic substrates in a flat plate microbial fuel cell. Environ. Sci. Technol. 38(21), 5809-5814.
324
Nam, J.-Y., Kim, H.-W., Lim, K.-H. and Shin, H.-S. (2010) Effects of organic loading rates on
325
the continuous electricity generation from fermented wastewater using a single-chamber
326
microbial fuel cell. Bioresour. Technol. 101(1), S33-S37.
327
Puig, S., Serra, M., Coma, M., Balaguer, M. and Colprim, J. (2011) Simultaneous domestic
328
wastewater treatment and renewable energy production using microbial fuel cells (MFCs). Water
329
Sci. Technol. 64(4), 904-909.
330
Rabaey, K. and Verstraete, W. (2005) Microbial fuel cells: novel biotechnology for energy
331
generation. Trends Biotechnol. 23(6), 291-298.
AC C
EP
TE D
M AN U
SC
RI PT
310
15
ACCEPTED MANUSCRIPT
Ren, L., Ahn, Y., Hou, H., Zhang, F. and Logan, B.E. (2014a) Electrochemical study of multi-
333
electrode microbial fuel cells under fed-batch and continuous flow conditions. J. Power Sources
334
257, 454-460.
335
Ren, L., Ahn, Y. and Logan, B.E. (2014b) A two-stage microbial fuel cell and anaerobic fluidized
336
bed membrane bioreactor (MFC-AFMBR) system for effective domestic wastewater treatment.
337
Environ. Sci. Technol. 48(7), 4199-4206.
338
Yang, W., He, W., Zhang, F., Hickner, M.A. and Logan, B.E. (2014) Single-step fabrication using
339
a phase inversion method of poly (vinylidene fluoride)(PVDF) activated carbon air cathodes for
340
microbial fuel cells. Environ. Sci. Technol. Lett. 1(10), 416-420.
341
Zhang, F., Cheng, S., Pant, D., Bogaert, G.V. and Logan, B.E. (2009) Power generation using an
342
activated carbon and metal mesh cathode in a microbial fuel cell. Electrochem. Commun. 11(11),
343
2177-2179.
344
Zhang, F., Ge, Z., Grimaud, J., Hurst, J. and He, Z. (2013) Long-term performance of liter-scale
345
microbial fuel cells treating primary effluent installed in a municipal wastewater treatment
346
facility. Environ. Sci. Technol. 47(9), 4941-4948.
347
Zhang, F., Ahn, Y. and Logan, B.E. (2014a) Treating refinery wastewaters in microbial fuel cells
348
using separator electrode assembly or spaced electrode configurations. Bioresour. Technol. 152,
349
46-52.
350
Zhang, H., Bruns, M.A. and Logan, B.E. (2006) Biological hydrogen production by Clostridium
351
acetobutylicum in an unsaturated flow reactor. Water Res. 40(4), 728-734.
352
Zhang, X., Xia, X., Ivanov, I., Huang, X. and Logan, B.E. (2014b) Enhanced activated carbon
353
cathode performance for microbial fuel cell by blending carbon black. Environ. Sci. Technol.
AC C
EP
TE D
M AN U
SC
RI PT
332
16
ACCEPTED MANUSCRIPT
48(3), 2075-2081.
355
Zhang, X., He, W., Ren, L., Stager, J., Evans, P.J. and Logan, B.E. (2015) COD removal
356
characteristics in air-cathode microbial fuel cells. Bioresour. Technol. 176, 23-31.
357
Zhuang, L., Zheng, Y., Zhou, S., Yuan, Y., Yuan, H. and Chen, Y. (2012) Scalable microbial fuel
358
cell (MFC) stack for continuous real wastewater treatment. Bioresour. Technol. 106, 82-88.
RI PT
354
AC C
EP
TE D
M AN U
SC
359
17
ACCEPTED MANUSCRIPT
1
Figure captions
2 Figure 1. (A) Full brush anodes spaced evenly between two cathodes (S2C) and trimmed brush
4
anodes near a single cathode (N1C), not drawn to scale. (B) Experimental set-up used in this
5
study. Pairs of S2Cs and N1Cs were arranged in a series, and domestic wastewater was
6
continuously provided through the inlet channel to each of the paired reactors.
7
Figure 2. Power generation (external resistor of 200 Ω for each MFC) produced by S2Cs and
8
N1Cs (U = upstream, D = downstream) during continuous operation under different theoretical
9
HRTs (8.8, 4.4 and 2.2 h).
M AN U
SC
RI PT
3
Figure 3. Polarization and power curves of 4 MFCs with different theoretical HRTs. (a)(d): 8.8 h;
11
(b)(e): 4.4 h; (c)(f): 2.2 h.
12
Figure 4. COD concentrations in the influent and effluent of the S2Cs and N1Cs (U = upstream,
13
D = downstream) and percentage of COD removal at theoretical HRTs of 8.8, 4.4 and 2.2 h.
14
Figure 5. COD removals (%) by S2C and N1C MFCs based on the theoretical and measured
15
HRTs. The regression line (black dashed) was drawn to fit the COD removals based on the
16
measured HRTs of both MFC configurations.
17
Figure 6. CEs of S2Cs and N1Cs (U = upstream, D = downstream) based on theoretical HRTs
18
(8.8, 4.4 and 2.2 h).
AC C
EP
TE D
10
1
ACCEPTED MANUSCRIPT
Table 1. Comparison of theoretical HRTs and measured HRTs with different anode
2
configurations (S2C and N1C). Dispersion and Peclet numbers at each HRT were obtained from
3
salt tracer tests. HRT (h)
Q (L/m2·h)
Configuration
N1C
0.028
36
8.8 (9.9)
7.4 (8.4)
18.2
4.4 (4.9)
3.8 (4.3)
36.3
0.035
28
2.2 (2.5)
2.2 (2.5)
72.7
0.077
13
8.8 (10.5)
10.3 (12.4)
18.2
0.020
49
4.4 (5.3)
5.2 (6.2)
36.5
0.015
68
2.2 (2.6)
2.9 (3.5)
73.0
0.043
23
SC
Measured*
TE D
*HRTs shown are corrected to include only the working reactor volume. Values in parentheses include the whole reactor volume (includes the entrance and exit volumes).
EP
7
Pe
Theoretical*
AC C
4 5 6
Dispersion number (d)
M AN U
S2C
RI PT
1
1
SC
RI PT
ACCEPTED MANUSCRIPT
M AN U
377
Figure 1. (A) Full brush anodes spaced evenly between two cathodes (S2C) and trimmed brush
379
anodes near a single cathode (N1C), not drawn to scale. (B) Experimental set-up used in this
380
study. Pairs of S2Cs and N1Cs were arranged in a series, and domestic wastewater was
381
continuously provided through the inlet channel to each of the paired reactors.
EP AC C
382
TE D
378
19
SC
RI PT
ACCEPTED MANUSCRIPT
M AN U
383 384
Figure 2. Power generation (external resistor of 200 Ω for each MFC) produced by S2Cs and
385
N1Cs (U = upstream, D = downstream) during continuous operation under different theoretical
386
HRTs (8.8, 4.4 and 2.2 h).
EP AC C
388
TE D
387
20
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
389
Figure 3. Polarization and power curves of 4 MFCs with different theoretical HRTs. (a)(d): 8.8 h;
391
(b)(e): 4.4 h; (c)(f): 2.2 h.
EP AC C
392
TE D
390
21
SC
RI PT
ACCEPTED MANUSCRIPT
M AN U
393 394
Figure 4. COD concentrations in the influent and effluent of the S2Cs and N1Cs (U = upstream,
395
D = downstream) and percentage of COD removal at theoretical HRTs of 8.8, 4.4 and 2.2 h.
AC C
EP
TE D
396
22
SC
RI PT
ACCEPTED MANUSCRIPT
M AN U
397 398
Figure 5. COD removals (%) by S2C and N1C MFCs based on the theoretical and measured
399
HRTs. The regression line (black dashed) was drawn to fit the COD removals based on the
400
measured HRTs of both MFC configurations.
AC C
EP
TE D
401
23
SC
RI PT
ACCEPTED MANUSCRIPT
M AN U
402 403
Figure 6. CEs of S2Cs and N1Cs (U = upstream, D = downstream) based on theoretical HRTs
404
(8.8, 4.4 and 2.2 h).
AC C
EP
TE D
405
24
ACCEPTED MANUSCRIPT
•
Compared performance of MFCs with different electrode configurations.
2
•
MFC with two cathodes and space between the electrodes produced highest power.
3
•
Measured HRTs were altered by different brush anode configurations.
4
•
COD removal was primarily a function of the measured HRTs and not configuration.
AC C
EP
TE D
M AN U
SC
RI PT
1
1
ACCEPTED MANUSCRIPT
Supporting Information Impact of Brush Electrode Configuration on Retention Time and Continuous Domestic Wastewater Treatment Efficiency Using Microbial Fuel Cells
RI PT
Kyoung-Yeol Kim, Wulin Yang, and Bruce E. Logan*
Department of Civil and Environmental Engineering, The Pennsylvania State University, 231Q Sackett Building, University Park, PA 16802, USA
EP
TE D
M AN U
SC
*Corresponding author: e-mail:
[email protected]; phone: +1-814-863-7908; fax: +1-814-863-7304
Figure S1. Diagrams of two MFC reactors demonstrating the total reactor volume; S2C (A),
AC C
N1C (B). Both reactors have an entrance and exit region (10 mL each) to evenly distribute the flow-in and -out.
S1
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
Figure S2. Salt tracer curve at a 8.8 h theoretical HRT. Arrows indicate the measured HRTs (8.4
AC C
EP
TE D
h for S2C, 12.4 h for N1C, based on the total reactor volume).
Figure S3. Salt tracer curve at a 4.4 h theoretical HRT. Arrows indicate the measured HRTs (4.3 h for S2C, 6.2 h for N1C, based on the total reactor volume).
S2
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
Figure S4. Salt tracer curve at a 2.2 h theoretical HRT. Arrows indicate the measured HRTs (2.5
AC C
EP
TE D
h for S2C, 3.5 h for N1C, based on the total reactor volume).
S3