International Journal of

Molecular Sciences Article

Development and Long-Term Stability of a Novel Microbial Fuel Cell BOD Sensor with MnO2 Catalyst Shailesh Kharkwal 1, *, Yi Chao Tan 1 , Min Lu 2 and How Yong Ng 1, * 1 2

*

Centre for Water Research, Department of Civil & Environmental Engineering, National University of Singapore, Engineering Drive 3, Singapore 117580, Singapore; [email protected] Key Laboratory of Flexible Electronics (KLOFE), Institute of Advanced Materials (IAM), Jiangsu National Synergetic Innovation Center for Advanced Materials (SICAM), Nanjing Tech University (NanjingTech), Nanjing 211816, China; [email protected] Correspondence: [email protected] (S.K.); [email protected] (H.Y.N.); Tel.: +65-8432-6713 (S.K.); +65-6516-4777 (H.Y.N.)

Academic Editor: Deepak Pant Received: 31 October 2016; Accepted: 25 January 2017; Published: 28 January 2017

Abstract: A novel microbial fuel cell (MFC)-based biosensor was designed for continuous monitoring of biochemical oxygen demand (BOD) in real wastewater. To lower the material cost, manganese dioxide (MnO2 ) was tested as an innovative cathode catalyst for oxygen reduction in a single chamber air-cathode MFC, and two different crystalline structures obtained during synthesis of MnO2 (namely β- and γ-MnO2 ) were compared. The BOD sensor was studied in a comprehensive way, using both sodium acetate solution and real domestic wastewater (DWW). The optimal performance of the sensor was obtained with a β-MnO2 catalyst, with R2 values of 0.99 and 0.98 using sodium acetate solution and DWW, respectively. The BOD values predicted by the β-MnO2 biosensor for DWW were in agreement with the BOD5 values, determined according to standard methods, with slight variations in the range from 3% to 12%. Finally, the long-term stability of the BOD biosensor was evaluated over 1.5 years. To the best of our knowledge, this is the first report of an MFC BOD sensor using an MnO2 catalyst at the cathode; the feasibility of using a low-cost catalyst in an MFC for online measurement of BOD in real wastewater broadens the scope of applications for such devices. Keywords: microbial fuel cell; manganese dioxide; cathode; biochemical oxygen demand; biosensor; wastewater

1. Introduction To ensure efficient operation of wastewater treatment processes, the continuous monitoring of the biochemical oxygen demand (BOD) of the treated effluent is essential. However, the conventional 5-day BOD (BOD5 ) test is laborious, sometimes poorly reproducible and inaccurate, particularly at low concentrations of organic compounds [1]. Moreover, membrane fouling of the dissolved oxygen probe can become a serious problem with wastewater, and frequent maintenance is therefore needed to maintain high sensitivity [2]. These issues supported the development of microbial biosensors for BOD measurement, based on respirometry [3,4], bioluminescence [5,6], bioreactor/chemostat technology [7,8], and structure-based biodegradation estimation [9]. Biosensors provide the additional benefit of automation; however, they usually make use of single microbial strains, which may not be able to oxidize a broad range of substrates. Consequently, dissolved oxygen consumption in such sensors may not be directly proportional to the total concentration of biodegradable organics present in solution [10]. In addition, the intrinsic limitation of oxygen solubility in aqueous solutions, the short-term stability of probes, and the lysis of immobilized microbial strains challenge the viability of such sensors [11,12]. Int. J. Mol. Sci. 2017, 18, 276; doi:10.3390/ijms18020276

www.mdpi.com/journal/ijms

Int. J. Mol. Sci. 2017, 18, 276

2 of 10

In recent years, microbial fuel cells (MFCs) have been examined as an alternative BOD sensing device [1,2,13–17]. MFC BOD sensors are quick and portable, and contain a mixed population of microorganisms (by using environmental inoculum from a wastewater reclamation plant in Singapore) able to oxidize a wide range of substrates [1,18]. Yet the design and operation conditions as reported in the literature are suboptimal, consisting of two-chambered MFC systems [13,14] functioning in batch mode [15–17]. For the operator of a wastewater treatment plant to continuously monitor the evolution of BOD in effluent samples, a continuous “online” system is needed. Furthermore, the use of single-chambered air cathode MFCs should be preferred, as these are easier and cheaper to operate than their two-chambered counterparts, mostly because the cathode can be directly exposed to ambient air as a free and sustainable source of oxygen. Another area of improvement for MFC sensors is the choice of an affordable and sustainable cathode catalyst. So far, most sensors reported in the literature have made use of expensive platinum [19–21]. However, considering the high cost and low durability of platinum due to corrosion and degradation of catalytic activity over time [22], alternative metal catalysts such as indium/tin [23], cobalt [24], titanium [25], tungsten [26], or manganese [27] should be sought. In our previous work, we explored the efficacy of manganese oxide (MnO2 ) as a catalyst for the oxygen reduction reaction [27]. Specifically, three types of MnO2 catalysts (i.e., α-, β-, and γ-MnO2 ), based on different preparation methods and crystalline structure, were incorporated individually into three air-cathode MFCs. The performance of the MFCs were characterized and investigated by cyclic voltammetry in neutral medium. The performance of the MFCs based on their electrical performance showed the superior catalytic activity of β- and γ-MnO2 over α-MnO2 ; yet, the efficacy and long-term stability of such catalyst in a sensor device remains to be assessed. In this study, a compact single-chambered MFC, with an air cathode and MnO2 as a cathode catalyst, was tested as a BOD sensor in a continuous mode. With the aim of implementing this MFC BOD sensor in wastewater treatment plants, the performance stability of the sensor must be evaluated over a long time period in order, for example, to address the repeatability of results and the need to re-calibrate the sensor over time. Therefore, unlike previous studies, the performance of the BOD sensor was evaluated over 1.5 years in this study using both a synthetic substrate and real domestic wastewater (DWW). The performance of the MFC BOD sensor was assessed holistically in terms of the linear relationship between signal output and organic concentration, sensor stability, and the compliance of predicted BOD values using the MFC biosensor with BOD5 , obtained according to standard methods [28]. 2. Results 2.1. Optimization of Hydraulic Retention Time (HRT) A good sensor should measure “true” BOD5 concentrations, provide a strong signal (in this case, voltage output), and have a quick response time. For this purpose, the HRT of the MFC-based BOD sensor was optimized. A sodium acetate solution at a fixed BOD5 concentration of 386 ± 10 mg/L was used to avoid the variability of DWW and study the reproducibility of the MFC-based BOD sensor. Figure 1 shows that the BOD5 removal rate decreased with decreasing HRT from 30 to 2 min. It was observed that an HRT of 30 min removed 85% of BOD5 in the influent feed, while an HRT of 2 min decreased the BOD5 removal efficiency down to 6% for both β- and γ-MnO2 -type BOD sensors. Only 6% of BOD5 removal efficiency at the HRT of 2 min ensured that the MFC BOD sensor would measure BOD values substantially similar to the BOD5 concentration present in the influent feed. The response time of the BOD sensors was calculated at the HRT of 2 min by alternating the phases of feeding and starvation periods and monitoring the time taken for the voltage output to reach a maximum stable voltage value. The shortest response time of the MFC BOD sensors was measured in the range of 2–3 h at a 2 min HRT for both β- and γ-MnO2 -type BOD biosensors.

Int. J. Mol. Sci. 2017, 18, 276 Int. J. Mol. Sci. 2017, 18, 276

3 of 10 3 of 9

Int. J. Mol. Sci. 2017, 18, 276

3 of 9

Figure 1. Optimization of hydraulic retention time (HRT) for the β-and γ-MnO2-type microbial fuel

Figure 1. Optimization of hydraulic retention time (HRT) for the β-and γ-MnO2 -type microbial fuel cell (MFC) biochemical oxygen demand (BOD) sensors by measuring the BOD5 removal efficiency at cell (MFC) biochemical oxygen demand (BOD) sensors by measuring the BOD5 removal efficiency at different HRTs. different HRTs. Figure 1. Optimization of hydraulic retention time (HRT) for the β-and γ-MnO2-type microbial fuel

2.2. Correlation BODoxygen 5 and Voltage Output cell (MFC)between biochemical demand (BOD) sensors by measuring the BOD5 removal efficiency at

2.2. Correlation between different HRTs. BOD5 and Voltage Output 2.2.1. Calibration with Sodium Acetate Solution

2.2.1. Calibration with Sodium Solution 2.2. Correlation between BOD5Acetate and Voltage Output

After three months of operation, the sensors were calibrated with the sodium acetate solution at

different BOD 5 concentrations ranging 44 towere 483 mg/L (Figurewith 2A,B). linear relationship was After months ofSodium operation, thefrom sensors calibrated theAsodium acetate solution at 2.2.1. three Calibration with Acetate Solution obtained the BOD 5 and the maximum steady-state voltage as long as the BOD 5relationship remained in different BODbetween concentrations ranging from 44 to 483 mg/L (Figure 2A,B). A linear was 5 Afterofthree months operation, with sodium acetate solution at the range 44–343 mg/L.of The R2 valuethe wassensors high atwere 0.99 calibrated and 0.97 for the the β- and γ-MnO 2, respectively. obtained between the BOD 5 and the maximum steady-state voltage as long as the BOD5 remained in different BOD5 concentrations ranging from 44found to 483to mg/L (Figure 2A,B). A linear relationship was The reproducibility of the sensor response was be better with β-MnO 2 (standard deviation 2 the range of 44–343 mg/L. The R value was high at 0.99 and 0.97 for the β- and γ-MnO2 , respectively. obtained between BOD25 and the maximum voltage as in long as the BOD 5 remained of 0.05%) than withthe γ-MnO (standard deviationsteady-state of 0.22%). As shown Table 1, the F-values werein The reproducibility of the sensor Rresponse was found toand be better deviation 2 (standard 2 value was theand range offor 44–343 mg/L. high at 0.99 0.97 forwith the β-β-MnO and γ-MnO 2, respectively. 511 236 the βand The γ-MnO 2 type MFC BOD biosensor, respectively, implying that the linear of 0.05%) than with γ-MnO (standard deviation of 0.22%). As shown in Table 1, the F-values 2 The reproducibility of the sensor response was found to be better with β-MnO deviation model was significant. Additionally, the value of “Prob > F” (probability that2 (standard null hypothesis iswere 511 and 236 for the βand γ-MnO type MFC BOD biosensor, respectively, implying that the of 0.05%) than with γ-MnO 2 (standard deviation 0.22%). As shown in Table 1, at thea F-values were 2 correct) remained below 0.05, indicating that theofmodel terms were significant 95% level oflinear model was significant. Additionally, value of BOD “Probbiosensor, >was F” obtained (probability thatimplying nullvoltage hypothesis isand correct) 511 and 236Infor thetypes β- and γ-MnOthe MFC respectively, that the linear confidence. both of sensor, a2 type good relationship between the output model was significant. Additionally, value of “Prob > F” (probability hypothesis is remained below 0.05, indicating that thethe model terms were significant at athat 95%null level of confidence. the BOD 5, indicating reliable performance and calibration. correct) remained 0.05, indicating that obtained the model terms the were significant at aafter 95% level of 5 , The performance the sensors was determined againbetween under same conditions one year In both types of sensor,below a of good relationship was the voltage output and the BOD confidence. In both typesBOD of sensor, a good relationship was aobtained between the voltage and of operation of performance the MFC sensor (Figure 2C,D). Again, good linear relationship wasoutput obtained indicating reliable and calibration. 2 value 2 value of 0.96 theaBOD 5, range indicating reliable performance and calibration. for BOD of 44–343 mg/L with an R of 0.97 (F-value of 173) and an R The performance of the sensors was determined again under the same conditions after one year Theofperformance sensors was determined again under the same after onewere year (F-value 145) for theofβ-the and γ-MnO 2-type BOD sensors, respectively. Allconditions the experiments of operation of the MFC BOD sensor (Figure 2C,D). Again, a good linear relationship was obtained for of operation the MFC and BODthe sensor (Figure 2C,D). Again, a good linear relationship was obtained conducted in of duplicate, standard deviation equaled 0.075% and 0.16%, for the βand a BODfor range of 44–343 mg/L with anwith R2 value 0.97of (F-value of 173)ofand R2 an value of 0.96 a BOD range 44–343 mg/L an R2of value 0.97 (F-value 173)an and R2 value of (F-value 0.96 γ-MnO 2-type MFCof BOD sensors, respectively. of 145)(F-value for the of β-145) and for γ-MnO -type BOD sensors, respectively. All the experiments were conducted 2 and γ-MnO2-type BOD sensors, respectively. All the experiments were in the βduplicate, and the equaleddeviation 0.075% and 0.16%, for the and γ-MnO conducted in standard duplicate, deviation and the standard equaled 0.075% andβ-0.16%, for the2 -type β- andMFC BOD sensors, respectively. γ-MnO2-type MFC BOD sensors, respectively.

Figure 2. Dynamic response of the MFC BOD sensors against organic concentration (BOD5) in sodium acetate solution at different time intervals. (A,B) show a linear relationship between maximum steadystate voltage generated and BOD5 of the β- and γ-MnO2-type MFC BOD biosensors, respectively, after 3Figure months; (C,D)response demonstrate thethe same type of linear relationship afterconcentration 1concentration year. Standard deviations in 2. Dynamic response of MFC BOD sensors against organic (BOD 5) in Figure 2. Dynamic of the MFC BOD sensors against organic (BOD in sodium 5 )sodium voltages are indicated in red color. acetate solution at different time intervals. show a linear relationship between maximum acetate solution at different time intervals.(A,B) (A,B) show a linear relationship between steadymaximum state voltage generated andand BODBOD 5 of the β- and γ-MnO2-type MFC BOD biosensors, respectively, after steady-state voltage generated 5 of the β- and γ-MnO2 -type MFC BOD biosensors, respectively, months; (C,D) (C,D) demonstrate same type of linear relationship afterafter 1 year. Standard deviations in after 33months; demonstratethe the same type of linear relationship 1 year. Standard deviations voltages are indicated in red color.

in voltages are indicated in red color.

Int. J. Mol. Sci. 2017, 18, 276

4 of 10

Table 1. Overall performance shown by the β- and γ-MnO2 -type microbial fuel cell (MFC) biochemical oxygen demand (BOD) sensors. Type of Sensor

Time Period

Type of Wastewater

β-MnO2 β-MnO2 β-MnO2 β-MnO2 γ-MnO2 γ-MnO2 γ-MnO2 γ-MnO2

3 months 6 months 1 year 1.5 years 3 months 6 months 1 year 1.5 years

AWW DWW AWW DWW AWW DWW AWW DWW

Variables for Polynomial Regression Voltage vs. Voltage vs. Voltage vs. Voltage vs. Voltage vs. Voltage vs. Voltage vs. Voltage vs.

Range of Organic Concentration (BOD5 ) Measured (ppm)

R Square

F-Value

Prob > F

44–343 36–178 44–343 33–160 44–343 36–178 44–343 33–160

0.990 0.980 0.971 0.934 0.979 0.967 0.966 0.906

511 360 173 114 236 209 145 77

3.14 × 10−6 2.281 × 10−7 4.55 × 10−5 5.18 × 10−6 2.12 × 10−5 1.79 × 10−6 6.93 × 10−5 2.20 × 10−5

BOD BOD BOD BOD BOD BOD BOD BOD

J. Mol. Sci. 2017, 18, 276 2.2.2.Int. Calibration with Domestic Wastewater (DWW)

4 of 9

For applications, theWastewater MFC BOD(DWW) sensors must be able to work with DWW; therefore, 2.2.2.practical Calibration with Domestic this was tested both with β- and γ-MnO months In with thoseDWW; conditions, a linear 2 after For practical applications, the MFC BOD six sensors mustofbeoperation. able to work therefore, relationship between the voltage output and the BOD concentration was again obtained, confirming 5 this was tested both with β- and γ-MnO2 after six months of operation. In those conditions, a linear the results obtained with substrate. 3A,B show awas linear relationship between the relationship between thesynthetic voltage output and theFigure BOD5 concentration again obtained, confirming 2 valuesthe results obtained with Figure 3A,B show a linear relationship BOD5the (ranging from 36 to 178synthetic mg/L) substrate. and the maximum steady-state voltage with Rbetween of 0.98 2 values of 0.98 BOD5and (ranging from of 36 to mg/L) theβmaximum steady-state voltage with R and 0.96, F-values 360178 and 209,and with and γ-MnO , respectively. Additionally, the very 2 and 0.96, and F-values of 360 and 209, with βand γ-MnO 2, respectively. Additionally, the very low low value of model parameter, “Prob > F” indicated that the model was statistically significant. value ofa model parameter,examination “Prob > F” of indicated that thestability model was statistically significant. To complete comprehensive the long-term of the MFC BOD sensors, they To complete a comprehensive examination of the long-term stability of the MFC BOD sensors, they were finally tested again with DWW after 1.5 years (Figure 3C,D). When measuring their performance were finally tested again with DWW after 1.5 years (Figure 3C,D). When measuring their performance with respect to BOD5 , which ranged from 33 to 160 mg/L, the regression coefficient, R2 , was 0.93 with respect to BOD5, which ranged from 33 to 160 mg/L, the regression coefficient, R2, was 0.93 and and 0.90, was114 114and and77, 77,for forthe theβ-βand γ-MnO MFC sensors, respectively. 2 -type 0.90, and and the the F-value F-value was and γ-MnO 2-type MFC BODBOD sensors, respectively. TableTable 1 summarizes the the overall performance typesofofMFC MFCBOD BOD sensors shows 1 summarizes overall performanceof of both both types sensors andand shows that that the the performances in terms of regression F-valuewere wereslightly slightly decreased 1.5 years. performances in terms of regressioncoefficient coefficient and and F-value decreased afterafter 1.5 years. Nonetheless, MFC BOD sensorswere werestill stillable able to measure of of water samples withwith goodgood Nonetheless, the the MFC BOD sensors measurethe theBOD BOD water samples accuracy, confirming the longevity of MFC BOD sensors. accuracy, confirming the longevity of MFC BOD sensors.

Figure 3. Dynamic response ofofMFC againstorganic organicconcentration concentration (BOD ) in DWW. Figure 3. Dynamic response MFCBOD BOD sensors sensors against (BOD 5) in 5DWW. a relationship linear relationship between the independent variable (maximum voltage) and (A,B)(A,B) show show a linear between the independent variable (maximum voltage) and dependent dependent (BOD of theγ-MnO β- and γ-MnO 2-type MFC BOD sensors, respectively, after 6 months; variable (BOD5variable ) of the β-5)and -type MFC BOD sensors, respectively, after 6 months; 2 demonstrate sametype typeof of relationship relationship after 1.51.5 years. Standard deviations in voltages are (C,D)(C,D) demonstrate thethe same after years. Standard deviations in voltages indicatedin in red red color. color. are indicated Table 1. Overall performance shown by the β- and γ-MnO2-type microbial fuel cell (MFC) biochemical oxygen demand (BOD) sensors. Type of Sensor

Time Period

Type of Wastewater

β-MnO2 β-MnO2 β-MnO2 β-MnO2 γ-MnO2

3 months 6 months 1 year 1.5 years 3 months

AWW DWW AWW DWW AWW

Variables for Polynomial Regression Voltage vs. BOD Voltage vs. BOD Voltage vs. BOD Voltage vs. BOD Voltage vs. BOD

Range of Organic Concentration (BOD5) Measured (ppm) 44–343 36–178 44–343 33–160 44–343

R Square

F-Value

Prob > F

0.990 0.980 0.971 0.934 0.979

511 360 173 114 236

3.14 × 10−6 2.281 × 10−7 4.55 × 10−5 5.18 × 10−6 2.12 × 10−5

Int. J. Mol. Sci. 2017, 18, 276 Int. J. Mol. Sci. 2017, 18, 276

5 of 10 5 of 9

2.3. Compliance of Predicted BOD Values (BODpp) with BOD55

The compliance of the BOD BOD values values as as predicted predicted by by the the MFC MFC sensors sensors (BOD (BODpp) with with BOD BOD55 was operation, using the calibration calibration determined using five random samples of DWW after 1.5 years of operation, curves obtained at that time. Figure 4 plots the BOD p values measured with βand γ-MnO 2 MFC curves obtained at that time. Figure 4 plots the BODp values measured with β- and γ-MnO 2 BOD sensors against the BOD 5 for all five samples . The BOD 5 of the five samples ranged from 56 to MFC BOD sensors against the BOD5 for all five samples. The BOD5 of the five samples ranged 155 mg/L, a standard comprised between 14%–27%. For both types of catalysts, a good from 56 to with 155 mg/L, withdeviation a standard deviation comprised between 14%–27%. For both types of correlation was obtained between BOD 5 and BOD p ; yet the standard deviation obtained with β-MnO catalysts, a good correlation was obtained between BOD5 and BODp ; yet the standard deviation2 (3%–12%)with was lower than that of γ-MnO (8–21%). Nevertheless, these standard deviations remained obtained β-MnO was 2lower than that of γ-MnO Nevertheless, these 2 (3%–12%) 2 (8–21%). lower than that of remained BOD5, therefore showing the improved reliability of the sensors over the standard deviations lower than that of BOD , therefore showing the improved reliability of 5 conventional BOD 5 method, especially when β-MnO 2 was used as the catalyst. the sensors over the conventional BOD5 method, especially when β-MnO2 was used as the catalyst.

4. Variation between BODpp by the β- and and γ-MnO γ-MnO22-type MFC BOD sensors and and the the BOD BOD55 Figure 4. 5-day BOD BOD method. method. measured with the conventional 5-day

3. Discussion 3. Discussion 3.1. Impact of Operating Parameters For achieving thethe “true” concentration of BOD be achieving reliable reliableperformance performanceofofany anyBOD BODsensor, sensor, “true” concentration of BOD 5 must 5 must be measured; the concentration of5 BOD 5 is almost the samethe in influent both theand influent effluent measured; the concentration of BOD is almost the same in both effluentand wastewater wastewater feed. In other words, a lower BODrate 5 removal rate the indicates the measurement of the “true” feed. In other words, a lower BOD indicates measurement of the “true” influent 5 removal influent BOD5 concentration, which is BOD the 5desired BOD5 measurement. of a5 BOD which is the desired measurement. The observationThe of a observation decreasing BOD 5 concentration, decreasing BOD 5 removal rate with decreasing HRT illustrates a lower HRT may help an to removal rate with a decreasing HRT aillustrates how a lower HRT how may help to measure nearly measure nearly an equal organic concentration, which is present in the influent feed. Moreover, equal organic concentration, which is present in the influent feed. Moreover, a lower HRT increasesa lower HRT increases the electrochemical of the mass biofilm by increasing transfer rate and the electrochemical activity of the biofilm activity by increasing transfer rate andmass subsequently helps in subsequently in decreasing the response timeThese of theresults MFC are BOD sensor. These results are in decreasing the helps response time of the MFC BOD sensor. in accordance with the findings accordance with the[19], findings of Di Lorenzo et al. [19], who reported that the dynamic response of Di Lorenzo et al. who reported that the dynamic response of MFC sensors became faster of at MFC sensors became faster at lower HRTsat(experiments were conducted at HRTs lower HRTs (experiments were conducted HRTs ranging from 89.3 to 417.0 min).ranging from 89.3 to 417.0 min). 3.2. Impact of Catalysts

3.2. Impact Catalystsstudy, β-MnO was reported as yielding the best performance, as assessed by In the of previous 2 cyclicInvoltammetry and confirmed by actual MFC experiments Nevertheless, the performance of the previous study, β-MnO2 was reported as yielding [27]. the best performance, as assessed by γ-MnO was good enough to be assessed for its efficacy as an MFC BOD sensor. The present study 2 cyclic voltammetry and confirmed by actual MFC experiments [27]. Nevertheless, the performance confirms superiority of β-MnO over γ-MnO sensors terms The of sensitivity and 2 MFC sensors of γ-MnOthe 2 was good enough to be assessed for its efficacy as an 2MFC BODinsensor. present study confirms the superiority of β-MnO2 MFC sensors over γ-MnO2 sensors in terms of sensitivity and stability. This can be explained by different crystallographic structures of β- and γ-MnO2 catalysts.

Int. J. Mol. Sci. 2017, 18, 276

6 of 10

stability. This can be explained by different crystallographic structures of β- and γ-MnO2 catalysts. Indeed, β-MnO2 presents a homogeneous nano-rod structure, while γ-MnO2 forms fine needles, dispersed in clusters around carbon nanotubes [27]. The improved structure of the former is likely to allow for better adsorption of O2 on its surface and achieve overall higher catalytic activity. In our MFC BOD sensor, this translated into a stronger and more stable signal. This is in accordance with other studies that have correlated the higher catalytic activity of β-MnO2 with its higher average oxidation state, as compared to γ-MnO2 [29]. 3.3. Impact of the Substrate One of the main goals of this study is to provide the operators of wastewater treatment plants with an efficient and sustainable BOD sensor; for this purpose, it is essential to assess the sensor performance not only with synthetic feed but also with real DWW. It is a well-known fact that simple substrates like acetate are more readily assimilable at the anode of an MFC than complex effluents, leading to higher coulombic recoveries [30]. Additionally, a complex substrate such as DWW helps more in the development of electrochemically active biofilms; meanwhile, as an easily consumable feed, a simple substrate improves voltage generation [31]. Indeed, in an MFC BOD sensor, lower assimilable rates of complex organics in DWW lead to slightly reduced sensitivity; nevertheless, a good linear relationship was established with DWW too, at a BOD5 ranging between 33 and 172 mg/L, even after 1.5 years, making the MFC BOD sensor applicable for practical use. 3.4. Long-Term Stability Over a long time period of 1.5 years, the voltage generated by both the β- and γ-MnO2 MFC BOD sensors eventually decreased, probably due to changes in biofilm composition, bio-fouling on the proton exchange membrane [32], and decreased catalyst activity [33]. Kim et al. [18] reported that their MFC BOD sensor remained stable for five years, but provided little evidence to support their claim. In particular, re-calibration was not addressed. In this study, we noticed that re-calibration is one of the critical factors that affected the performance of the MFC BOD sensor and that it was definitely needed after a year; however, the exact frequency of re-calibration should be evaluated in future works in order to maintain a good proportionality between BOD5 and the recorded voltage at all times. 3.5. BODp vs. BOD5 The β-MnO2 -type sensor showed a close compliance of BODp with BOD5 , while both MFC BOD sensors exhibited low standard deviation in the repetition of measuring BODp . Results indicated that, for higher BOD values (>150 mg/L), the MFC BOD sensors had a tendency to slightly overestimate the BOD. Yet, the compliance was extremely good in a lower BOD range (between 50 and 100 mg/L), which corresponds to the typical BOD of treated effluents from wastewater treatment plants. This, along with good accuracy, result reproducibility, a fast response time, and an ability to measure BOD, continuously emphasizes the suitability of the MFC BOD sensor developed in this study for application in wastewater treatment plants. 4. Materials and Methods 4.1. MFC Construction and Operation The MFCs used in this study were designed and constructed as acrylic single-chambered air-cathode cylindrical reactors, with a diameter of 6 cm and a width of 1 cm (i.e., a total volume of 28.3 mL) (Figure 5). MnO2 nanoparticles were mixed with carbon nanotubes as the conductive material and polyvinylidene fluoride as the binder. Both the anode and cathode consisted of carbon cloth (E-Tek, Woburn, MA, USA), with the cathode coated on one side with MnO2 ink (either β or γ) at a catalyst loading of 3 mg·cm−2 following the procedure of Lu et al. [27]. The cathode was then hot-pressed together with Nafion, a proton exchange membrane (PEM). All MFCs were duplicated.

Int. J. Mol. Sci. 2017, 18, 276

7 of 10

Int. J. Mol. Sci. 2017, 18, 276

7 of 9

Figure 5. Schematic diagram of theof MFC this study. Influent (B) Microbial Figure 5. Schematic diagram the BOD MFCsensor BOD used sensorinused in this(A) study. (A) tank; Influent tank; (B) fuel cell; (C) Computer and data acquisition system (DAS); (D) Details of MFC configuration—1: MFC Microbial fuel cell; (C) Computer and data acquisition system (DAS); (D) Details of MFC body; 2: anode carbonMFC clothbody; with 2: current 3: Nafion membrane; 4: catalyst layer; 5: carbon configuration—1: anode collector; carbon cloth with current collector; 3: Nafion membrane; 4: base catalyst layer; 6:layer; cathode carbon cloth with 7: cathode frame. 5: carbon base layer; 6: current cathode collector; carbon cloth with current collector; 7: cathode frame.

MFCs were inoculated and operated initiallywith withreal realDWW DWWcollected collected from from the The The MFCs were inoculated and operated initially the primary primary clarifier of a wastewater treatment plant in Singapore. DWW was used as-is, except for preliminary clarifier of a wastewater treatment plant in Singapore. DWW was used as-is, except for preliminary screening, 0.45 pore-sized mm pore-sized filter to cloth to remove big particles from accumulating screening, usedused withwith a 0.45a mm filter cloth remove big particles from accumulating inside inside the MFCs. DWW had a COD of 355.0 ± 75.0 mg/L, total suspended solids of 218.5 ± 105.0 mg/L, the MFCs. DWW had a COD of 355.0 ± 75.0 mg/L, total suspended solids of 218.5 ± 105.0 mg/L, total total dissolved solids of 110.2 ± 62.3 mg/L, a pH of 7. ± 0.4, and a conductivity of 0.855 ± 0.085 ms/cm. dissolved solids of 110.2 ± 62.3 mg/L, a pH of 7. ± 0.4, and a conductivity of 0.855 ± 0.085 ms/cm. In some instances, a sodium acetate solution was used, prepared according to methodology described In some instances, a sodium acetate solution was used, prepared according to methodology described by Lefebvre et al. [24]. Influent feeding was done using peristaltic pumps (Masterflex 07523-70, by Lefebvre et al. [24]. Influent feeding was done using peristaltic pumps (Masterflex 07523-70, Spectra-Teknik Pte Ltd., Singapore) in an upflow mode. A 10 Ω resistor was connected externally to Spectra-Teknik Pte Ltd., Singapore) in an upflow mode. A 10 Ω resistor was connected externally each MFC, following recommendations by Di Lorenzo et al. [19] that low external resistance can to each MFC, following by Di Lorenzo et al.was [19]recorded that low at external resistance shorten the responserecommendations time, and the voltage over the resistor fixed intervals of 1can h shorten thea response time, and connected the voltagetoover the resistor recorded at fixed intervals of 1 hArray using using digital multimeter a computer via was a data acquisition system (M3500A, a digital multimeter connected Electronic, Taipei, Taiwan). to a computer via a data acquisition system (M3500A, Array Electronic, Taipei, Taiwan). 4.2. Synthesis of MnO2 Catalysts 4.2. Synthesis of MnO2 Catalysts The β-MnO2 catalyst was prepared by dissolving MnSO4·H2O (1.6 mmol) and (NH4)2S2O8 The β-MnO wasatprepared by dissolving ·H2 Oa homogeneous (1.6 mmol) and (NH4)which 2 S2 O8 (1.6 mmol) in 220catalyst mL of water ambient temperature (~25MnSO °C) to4form solution, ◦ C) to form a homogeneous solution, which (1.6 mmol) in 20 mL of water at ambient temperature (~25 was then transferred to a Teflon-lined stainless steel autoclave, sealed, and maintained at 125 °C for ◦C was 14 then to a Teflon-lined autoclave, sealed, maintained at 125 h. transferred After the solution was cooled stainless down to steel ambient temperature, theand resulting dark-gray solid for 14 h. After the solution was cooled down to ambient temperature, the resulting solid product was filtered, washed, and dried. γ-MnO 2 was prepared by dissolving MnSO4dark-gray ·H2O (16 mmol) and was (NHfiltered, 4)2S2O8 (16 mmol) and in 200 mL ofγ-MnO water.2After that, the mixture was heated up4 ·to in mmol) an oil product washed, dried. was prepared by dissolving MnSO H29O°C(16 for in 24200 h with magnetic stirring. Thethe resulting was and bath (NH4and )2 S2maintained O8 (16 mmol) mL of water. After that, mixtureblack was precipitate heated up to 9 ◦filtered, C in an washed, and dried [27]. oil bath and maintained for 24 h with magnetic stirring. The resulting black precipitate was filtered, washed, and dried [27]. 4.3. Calibration Procedure 4.3. Calibration Procedure The BOD sensor was calibrated by plotting the steady-state voltage (the signal) against BOD5 in aThe linear relationship. a fixedby external load 10 Ω was used in all(the experiments, the relationship BOD sensor wasSince calibrated plotting theofsteady-state voltage signal) against BOD5 in a between current and voltage was proportional and followed Ohm’s law. For calibration with the linear relationship. Since a fixed external load of 10 Ω was used in all experiments, the relationship sodium acetate solution, the MFC BOD sensors were fed with a specific BOD 5 until the maximum between current and voltage was proportional and followed Ohm’s law. For calibration with the steady voltage was generated and recorded. Between twowith sets a ofspecific experiments, period of maximum starvation sodium acetate solution, the MFC BOD sensors were fed BOD5a until the allowed the voltage to drop to 1.5 ± 0.5 mV. The MFC BOD sensor was fed with acetate solutions with steady voltage was generated and recorded. Between two sets of experiments, a period of starvation different BOD5 concentrations, and each concentration was tested twice to ensure reproducibility. For allowed the voltage to drop to 1.5 ± 0.5 mV. The MFC BOD sensor was fed with acetate solutions calibration with DWW, a similar methodology was used with several dilutions of feed prepared using with different BOD5 concentrations, and each concentration was tested twice to ensure reproducibility. deionized water.

Int. J. Mol. Sci. 2017, 18, 276

8 of 10

For calibration with DWW, a similar methodology was used with several dilutions of feed prepared using deionized water. 4.4. Analyzes BOD5 and COD tests were performed in accordance with standard methods [28], using the 5-day dilution method and the closed reflux titrimetric method, respectively. The samples were filtered through glass microfiber filters (0.45-µm pore-sized, Whatman, Irvine, UK) prior to analysis. The analysis of variance (ANOVA) was used as a statistical analysis tool to examine and interpret the linear model equations obtained during calibration. ANOVA is a simple regression analysis that uses the Fisher’s F-test, and its associated probability to assess the robustness of linear models and overall model significance. A “Prob > F” less than 0.05 and a large F-value indicate that the model terms are significant. 5. Conclusions This study is first to use an MnO2 catalyst (in place of an expensive Pt catalyst) for MFCs applied as online BOD sensors. A good linear relationship between the voltage and BOD5 in both a sodium acetate solution and DWW, a stable performance in the long term (more than 1.5 years of operation), and close compliance between BODp and BOD5 demonstrated the potential of MnO2 -MFC BOD sensors in substituting the conventional BOD5 methods, which takes a longer analytical time of five days and suffers from substantial variability. Acknowledgments: We acknowledge the support from National University of Singapore and Public Utility Board (PUB), Singapore’s national water agency. Author Contributions: Shailesh Kharkwal designed and performed the experiments, and wrote the manuscript; Yi Chao Tan analyzed the data; Min Lu performed some of the experiments; How Yong Ng conceived and supervised the experiments. Conflicts of Interest: The authors declare no conflict of interest.

References 1. 2. 3. 4. 5. 6. 7. 8.

9. 10.

Kim, B.H.; Chang, I.S.; Gil, G.C.; Park, H.S.; Kim, H.J. Novel BOD (biological oxygen demand) sensor using mediator-less microbial fuel cell. Biotechnol. Lett. 2003, 25, 541–545. [CrossRef] [PubMed] Gil, G.C.; Chang, I.S.; Kim, B.H.; Kim, M.; Jang, J.K.; Park, H.S.; Kim, H.J. Operational parameters affecting the performance of a mediator-less microbial fuel cell. Biosens. Bioelectron. 2003, 18, 327–334. [CrossRef] Hikuma, M.; Suzuki, H.; Yasuda, T. Amperometric estimation of BOD by using living immobilized yeasts. Eur. J. Appl. Microbiol. 1979, 8, 289–297. [CrossRef] Yang, Z.; Suzuki, H.; Sasaki, S.; McNiven, S.; Karube, I. Comparison of the dynamic transient- and steady-state measuring methods in a batch type BOD sensing system. Sens. Actuator B Chem. 1997, 45, 217–222. [CrossRef] Hyun, C.K.; Tamiya, E.; Takeuchi, T.; Karube, I. A novel BOD sensor based on bacterial luminescence. Biotechnol. Bioeng. 1993, 41, 1107–1111. [CrossRef] [PubMed] Carstea, E.; Bridgeman, J.; Baker, A.; Reynolds, D. Fluorescence spectroscopy for wastewater monitoring: A review. Water Res. 2016, 95, 205–219. [CrossRef] [PubMed] Bourgeois, W.; Burgess, J.E.; Stuetz, R.M. On-line monitoring of wastewater quality: A review. J. Chem. Technol. Biotechnol. 2001, 76, 337–348. [CrossRef] Jouanneau, S.; Recoules, L.; Durand, M.J.; Boukabache, A.; Picot, V.; Primault, Y.; Lakel, A.; Sengelin, M.; Barillon, B.; Thouand, G. Methods for assessing biochemical oxygen demand (BOD): A review. Water Res. 2014, 49, 62–82. [CrossRef] [PubMed] Raymond, J.W.; Rogers, T.N.; Shonnard, D.R.; Kline, A.A. A review of structure-based biodegradation estimation methods. J. Hazard. Mater. 2001, 84, 189–215. [CrossRef] Reynolds, D.M.; Ahmad, S.R. Rapid and direct determination of wastewater BOD values using a fluorescence technique. Water Res. 1997, 31, 2012–2018. [CrossRef]

Int. J. Mol. Sci. 2017, 18, 276

11. 12. 13. 14.

15.

16. 17.

18. 19. 20.

21.

22. 23. 24.

25.

26. 27.

28. 29. 30.

31.

9 of 10

Riedel, K.; Lehmann, M.; Tag, K.; Renneberg, R.; Kunze, G. Arxula adeninivorans based sensor for the estimation of BOD. Anal. Lett. 1998, 31, 1–12. [CrossRef] Tan, T.C.; Wu, C. BOD sensors using multi-species living or thermally killed cells of a BODSEED microbial culture. Sens. Actuator B Chem. 1999, 54, 252–260. [CrossRef] Kumlanghan, A.; Liu, J.; Thavarungkul, P.; Kanatharana, P.; Mattiasson, B. Microbial fuel cell-based biosensor for fast analysis of biodegradable organic matter. Biosens. Bioelectron. 2007, 22, 2939–2944. [CrossRef] [PubMed] Moon, H.; Chang, I.S.; Kang, K.H.; Jang, J.K.; Kim, B.H. Improving the dynamic response of a mediator-less microbial fuel cell as a biochemical oxygen demand (BOD) sensor. Biotechnol. Lett. 2004, 26, 1717–1721. [CrossRef] [PubMed] Chang, I.S.; Jang, J.K.; Gil, G.C.; Kim, M.; Kim, H.J.; Cho, B.W.; Kim, B.H. Continuous determination of biochemical oxygen demand using microbial fuel cell type biosensor. Biosens. Bioelectron. 2004, 19, 607–613. [CrossRef] Chang, I.S.; Moon, H.; Jang, J.K.; Kim, B.H. Improvement of a microbial fuel cell performance as a BOD sensor using respiratory inhibitors. Biosens. Bioelectron. 2005, 20, 1856–1859. [CrossRef] [PubMed] Pandit, S.; Ghosh, S.; Ghangrekar, M.M.; Das, D. Performance of an anion exchange membrane in association with cathodic parameters in a dual chamber microbial fuel cell. Int. J. Hydrogen Energy 2012, 37, 9383–9392. [CrossRef] Kim, M.; Youn, S.M.; Shin, S.H.; Jang, J.G.; Han, S.H.; Hyun, M.S.; Gadd, G.M.; Kim, H.J. Practical field application of a novel BOD monitoring system. J. Environ. Monit. 2003, 5, 640–643. [CrossRef] [PubMed] Di Lorenzo, M.; Curtis, T.P.; Head, I.M.; Scott, K. A single-chamber microbial fuel cell as a biosensor for wastewaters. Water Res. 2009, 43, 3145–3154. [CrossRef] [PubMed] Kang, K.H.; Jang, J.K.; Pham, T.H.; Moon, H.; Chang, I.S.; Kim, B.H. A microbial fuel cell with improved cathode reaction as a low biochemical oxygen demand sensor. Biotechnol. Lett. 2003, 25, 1357–1361. [CrossRef] [PubMed] Kim, M.; Hyun, M.S.; Gadd, G.M.; Kim, G.T.; Lee, S.J.; Kim, H.J. Membrane-electrode assembly enhances performance of a microbial fuel cell type biological oxygen demand sensor. Environ. Technol. 2009, 30, 329–336. [CrossRef] [PubMed] Shao, Y.; Liu, J.; Wang, Y.; Lin, Y. Novel catalyst support materials for PEM fuel cells: Current status and future prospects. J. Mater. Chem. 2009, 19, 46–59. [CrossRef] Chhina, H.; Campbell, S.; Kesler, O. An oxidation-resistant indium tin oxide catalyst support for proton exchange membrane fuel cells. J. Power Sources. 2006, 161, 893–900. [CrossRef] Lefebvre, O.; Ooi, W.K.; Tang, Z.; Abdullah-Al-Mamun, M.; Chua, D.H.C.; Ng, H.Y. Optimization of a Pt-free cathode suitable for practical applications of microbial fuel cells. Bioresour. Technol. 2009, 100, 4907–4910. [CrossRef] [PubMed] Ekström, H.; Wickman, B.; Gustavsson, M.; Hanarp, P.; Eurenius, L.; Olsson, E.; Lindbergh, G. Nanometer-thick films of titanium oxide acting as electrolyte in the polymer electrolyte fuel cell. Electrochim. Acta 2007, 52, 4239–4245. [CrossRef] Maiyalagan, T.; Viswanathan, B. Catalytic activity of platinum/tungsten oxide nanorod electrodes towards electro-oxidation of methanol. J. Power Sources 2008, 175, 789–793. [CrossRef] Lu, M.; Kharkwal, S.; Ng, H.Y.; Li, S.F.Y. Carbon nanotube supported MnO2 catalysts for oxygen reduction reaction and their applications in microbial fuel cells. Biosens. Bioelectron. 2011, 26, 4728–4732. [CrossRef] [PubMed] APHA. Standard Methods for the Examination for Water and Wastewater, 21st ed.; American Public Health Association: Washington, DC, USA, 2005. Zhang, L.; Liu, C.; Zhuang, L.; Li, W.; Zhou, S.; Zhang, J. Manganese dioxide as an alternative cathodic catalyst to platinum in microbial fuel cells. Biosens. Bioelectron. 2009, 24, 2825–2829. [CrossRef] [PubMed] Lefebvre, O.; Shen, Y.; Tan, Z.; Uzabiaga, A.; Chang, I.S.; Ng, H.Y. Full-loop operation and cathodic acidification of a microbial fuel cell operated on domestic wastewater. Bioresour. Technol. 2011, 102, 5841–5848. [CrossRef] [PubMed] Pant, D.; van Bogaert, G.; Diels, L.; Vanbroekhoven, K. A review of the substrates used in microbial fuel cells (MFCs) for sustainable energy production. Bioresour. Technol. 2010, 101, 1533–1543. [CrossRef] [PubMed]

Int. J. Mol. Sci. 2017, 18, 276

32.

33.

10 of 10

Xu, J.; Sheng, G.-P.; Luo, H.-W.; Li, W.-W.; Wang, L.-F.; Yu, H.-Q. Fouling of proton exchange membrane (PEM) deteriorates the performance of microbial fuel cell. Water Res. 2012, 46, 1817–1824. [CrossRef] [PubMed] Rismani-Yazdi, H.; Carver, S.M.; Christy, A.D.; Tuovinen, O.H. Cathodic limitations in microbial fuel cells: An overview. J. Power Sources 2008, 180, 683–694. [CrossRef] © 2017 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).