The Effect of Different Pretreatment of Chicken Manure for Electricity Generation in Membrane-Less Microbial Fuel Cell

Abstract

The need for energy resources is growing all the time, which means that more fossil fuels are needed to provide them. People prefer to consume chicken as a source of protein, and this creates an abundance of waste. Thus, microbial fuel cells represent a new technological approach with the potential to generate electricity through the action of electrogenic bacteria toward chicken manure, while reducing the abundance of chicken manure. This study investigated the effect of different pretreatment (thermal, alkaline, and sonication pretreatment) of chicken manure to improve the performance of a membrane-less microbial fuel cell (ML-MFC). Statistical response surface methodology (RSM) through a central composite design (CCD) under a quadratic model was conducted for optimization of the ML-MFC performance focusing on the COD removal efficiency (R² = 0.8917), biomass (R² = 0.9101), and power density response (R² = 0.8794). The study demonstrated that the highest COD removal (80.68%), biomass (7.8539 mg/L), and power density (220 mW/m²) were obtained when the pretreatment conditions were 140 °C, 20 kHz, and pH 10. The polarization curve of the best condition of ML-MFC was plotted to classify the behavior of the ML-MFC. The kinetic growth of Bacillus subtillis (BS) showed that, in treated chicken manure, the specific growth rate µ = 0.20 h⁻¹ and doubling time Td = 3.43 h, whereas, in untreated chicken manure, µ = 0.11 h⁻¹ and Td = 6.08.

1. Introduction

Electricity is a basic energy need for human beings in daily life, and its consumption has grown over the years. However, more than one million people are still without energy throughout the world. According to a report from the Global Energy Review (GER) in the year 2019, there were around 770 million people in the globe who do not have access to electricity [1]. This situation has brought about a high demand for basic energy (electricity) consumption. The world now commonly depends on energy; thus, energy access is currently considered as a basic human need that facilitates education, communication, poverty reduction, and overall socioeconomic growth. Mostly, the world relies on nonrenewable energy sources (coal, oil, and natural gas) to gain energy. There are several drawbacks of using nonrenewable energy, which have contributed to the global energy crisis. The utilization of fossil fuels triggers environment problems such as the emission of greenhouse gases (GHG), which later lead to the global warming. Furthermore, nonrenewable energy sources will eventually run out and cannot be replenished by the nature. Thus, new alternative energy sources are urgently needed to solve these problems. The most suitable alternative sources are renewable energy (biomass, solar power, wind energy, and hydropower energy) [1,2,3].

On the other hand, the Sustainable Development Goal on Energy (SDG7) focuses on a global effort to ensure that everyone has access to affordable, reliable, sustainable, and modern energy. In order to achieve universal access to modern energy services, a considerable improvement in the enabling environments is needed for relevant projects and programs. In the government’s diversification goal and power sector growth strategy, renewable energy sources and other low CO₂-emitting sources, such as nuclear, have been highlighted [4].

The world population has increased drastically over the years, thus contributing to the high demand for food. Most people consume protein sources in their diet. Malaysia consumes chicken meat every day because it is a cheaper protein source than other poultry and meat. Thus, a high demand of chicken is recorded every day in the market. Therefore, to cope with this market demand, a huge number of chicken farms can be found in Malaysia, thus resulting in a huge amount chicken manure. Chicken manure has a negative effect on the health of the chickens; thus, it must be removed from farm, and its disposal contributes to environmental problems. However, chicken manure has high unrecovered energy that can be translated into electricity. To achieve the proposed energy targets, new and alternative methods of electricity production from renewable resources without environmental problems are greatly desired to save our environment [5,6,7]. Previous research conducted several studies on the pretreatment of animal manure for improvised biogas production, with biological pretreatment showing the most prominent result with a ~74% increase in methane yield [8]. In microbial fuel cell (MFC) technology, various pretreatments have been suggested to provide better electricity generation [9]. Hassan et al. [10] showed that raw rice straw in a dual-chambered MFC generated a power density of up to 145 mW/m² using cellulose-degrading bacteria (CDB). Compared to sonication pretreatment and alkaline pretreatment, Tao et al. achieved a maximum power density of 10.19 W/m³ using vegetable and grass wastes [11]. Ong et al. [12] highlighted that pretreated rice straw used in MFC was 35 times higher in power density compared to the untreated substrate. Thus, MFC technology can be enhanced by pretreatment of the substrate to yield alternative renewable energy technology for electricity generation.

An MFC is a bio-electrochemical device capable of converting chemical energy to electrical energy via the actions of microbes (biocatalyst) on the biomass waste acting as a substrate. MFCs can utilize a variety of substrates such as acetate [13,14], glucose [15,16], cellulose [17], synthetic wastewater [18], food wastes [19,20], sugarcane waste [21,22], lignocellulosic waste [23,24], and activated sludge [25].

Recently, several types of MFCs have been proposed, with the double-chamber MFC most commonly used by researchers. This MFC contains mediators, a membrane, and an aeration system, which are expensive. Moreover, the chemicals used as mediators, such as natural red and thionine, are toxic, and the proton exchange membrane (PEM) can suffer from membrane fouling, thus disturbing the power generation and leading to a lower performance of the MFC. Hence, an alternative solution is required to overcome this problem. In this study, MFCs were based on a single chamber with no mediator or membrane, using a graphite felt electrode with an air cathode. Chicken manure was supplemented as the substrate, and Bacillus subtilis acted as the biocatalyst. Response surface methodology (RSM) was conducted for optimization of the pretreatment of chicken manure so as to improve the performance of the membrane-less MFC as a function of the biomass response, chemical oxygen demand (COD) removal, and power density generation.

2. Results and Discussion

2.1. Proximate Analysis of Substrates

The chicken manure composition was identified to determine its usefulness as a substrate for the growth of electrogenic bacteria (Bacillus subtilis (BS)), thus achieving power generation in the ML-MFC.

A proximate analysis was performed to determine the characteristics of chicken manure and to verify its usefulness as a substrate in the membrane-less microbial fuel cell (ML-MFC). The moisture content of dry chicken manure was 16.79% with pH 7.83, slightly differing from the characteristics observed in [26]. The proximate analysis involved the analysis of macronutrient, micronutrient, and trace elements. AAS, elemental analysis, and ICP-OES were used to analyze the elements in the chicken manure sample. As shown in

Table 1. Compound analysis in chicken manure.

Types	Element	Unit	Untreated Chicken Manure	Treated Chicken Manure
Macronutrient	Carbon (C)	%	34.20	35.12
	Hydrogen (H)	%	2.86	3.18
	Nitrogen (N)	%	4.05	4.16
Micronutrient	Magnesium (Mg)	mg/L	71.61	78.03
	Iron (Fe)	mg/L	9.7	10.5
Trace element	Zinc (Zn)	mg/L	4.6	3.9
	Manganese (Mn)	mg/L	5.4	6.3
	Cadmium (Cd)	mg/L	1.7	2.2
Organic carbon		mg/L	517	610

, AAS revealed that the chicken manure contained a high amount of magnesium (Mg), 71.60 mg/L, followed by iron (Fe), manganese (Mn), zinc (Zn), and cadmium (Cd) with 9.76 mg/L, 4.63 mg/L, 5.41, and 1.7 mg/L, respectively. The high concentration of elements such as Fe, Mn, Zn, and heavy metals can have a harmful effect on certain microbial species via metabolite reactions linked to redox reactions. However, small quantities of these elements are still needed for building compounds and proton generation in bacterial cells. For example, Mg is required by bacteria for the enzymatic replication of DNA [27].

Additionally, elemental analysis using an elemental analyzer was carried out to examine the macronutrient (C, H, and N) composition in treated and untreated chicken manure. The results in

Table 2. The results of COD removal efficiency, biomass, and power density generation from CCD of RSM.

Run	Variables a			Response 1 COD Removal Efficiency (%)	Response 2 Biomass (mg/L)	Response 3 Power Density (mW/m2)
	A	B	C
	°C	kHz	pH
1	−1	0	−1	30.4	6.521	77.25
2	+1	0	−1	43.855	6.723	93.241
3	−1	+1	−1	35.12	6.608	89.168
4	+1	+1	−1	48.21	6.689	95.32
5	−1	0	+1	79.548	7.402	168.66
6	+1	0	+1	81.025	6.780	160.41
7	−1	+1	+1	76.124	6.833	173.48
8	+1	+1	+1	74.891	6.608	186.34
9	−1	0	0	64.194	6.724	120.6
10	+1	0	0	98.305	7.698	205.228
11	0	−1	0	86.546	7.690	118.195
12	0	+1	0	89.618	6.807	62.842
13	0	0	−1	81.320	6.834	85.566
14	0	0	+1	87.213	7.03	200.14
15	0	0	0	72.412	7.256	211.02
16	0	0	0	84.745	7.02	180.3
17	0	0	0	79.731	7.381	194.3
18	0	0	0	83.792	7.156	195.7
19	0	0	0	80.685	7.853	220.089
20	0	0	0	80.658	7.831	220.089

All variables are in uncoded units. A: thermal pretreatment; B: sonication pretreatment; C: alkaline pretreatment. ^a The code that being used in the Design of Expert for RSM optimization.

reveal that untreated chicken manure contained 34.20%, 2.86%, and 4.05% of C, H, and N, respectively. Treated chicken manure contained a higher percentage of C, H, and N than untreated chicken manure with 35.12%, 3.18%, and 4.16%, respectively. The treated chicken manure was prepared by performing three types of pretreatment: thermal, sonication, and alkaline pretreatment. The COD values of untreated and treated chicken manure were 610 mg/L and 571 mg/L, respectively. Nutrient consumption by EB relies on this COD value. The pretreatment process degrades the lignocellulosic structure of chicken manure via chemical and physical pathways [8], making it more accessible to microbes as a substrate for metabolic reactions and, thus, increasing the performance of the ML-MFC by generating a higher voltage. These macronutrients are the most important elements for the growth of bacterial cells [28]. Therefore, chicken manure has high potential as a feedstock for the growth of electrogenic bacteria in an ML-MFC.

2.2. Optimization of Pretreatment for Electricity Generation Using Response Surface Methodology (RSM) with a CCD

The response of three independent variables (thermal pretreatment, A; sonication pretreatment, B; alkaline pretreatment, C) to the COD removal efficiency (Y1), biomass (Y2), and power density (Y3) was further investigated using response surface methodology (RSM) with a central composite design (CCD).

2.2.1. Statistical Analysis and Regression Model

Table 2 lists the results of the RSM studies using Design Expert 13.0. As can be seen, COD removal efficiency ranged from 30.4% to 98.30%, while biomass ranged from 6.521 mg/L to 7.8539 mg/L, and power density ranged from 77.25 mW/m² to 220.089 mW/m². Four types of models were used to fit the experimental data: linear, two-factor interaction, quadratic, and cubic polynomial. A probability value (“prob > F”) less than 0.05 indicates significance according to the sequential model sum of squares (

Table 3. Sequential model sum of squares of COD removal efficiency, biomass, and power density generation.

Responds	Source	Sum of Squares	df	Mean Square	F-Value	p-Value
COD removal efficiency (%)	Linear	1461.69	3	487.23	0.6569	0.5904
	2FI	2355.60	3	785.20	1.07	0.3945
	Quadratic	3831.79	3	1277.26	2.25	0.0325
	Cubic	1905.01	4	476.25	0.7569	0.5890
Biomass (mg/L)	Linear	0.1228	3	0.0409	1.13	0.3682
	2FI	0.0335	3	0.0112	0.265	0.8493
	Quadratic	0.3179	3	0.1060	4.60	0.0285
	Cubic	0.0959	4	0.0240	1.07	0.4464
Power density (mW/m2)	Linear	40,798.46	3	13,599.49	3.01	0.0612
	2FI	9376.89	3	3125.63	0.645	0.5997
	Quadratic	15,037.17	3	5012.39	1.05	0.0381
	Cubic	20,691.97	4	5172.99	1.14	0.4215

). Because the p-values for COD removal efficiency, biomass, and power density were all less than 0.05, the quadratic model was chosen as the best model.

A lack-of-fit test was conducted to identify whether the model was adequate. Lack-of-fit p-values for the quadratic model for COD removal efficiency, biomass, and power density were 0.2681, 0.3435, and 0.5928, respectively (

Table 4. Lack-of-fit tests for COD removal efficiency, biomass, and power density generation.

	Source	Sum of Squares	df	Mean Square	F-Value	p-Value
COD removal efficiency (%)	Linear	11,438.60	11	1039.87	12.11	0.0015
	2FI	9083.00	8	1135.37	13.23	0.0009
	Quadratic	5251.21	5	1050.24	12.23	0.2681
	Cubic	3346.20	1	3346.20	38.98	0.0002
	Pure error	429.24	5	85.85
Biomass (mg/L)	Linear	0.4881	11	0.0444	2.37	0.1755
	2FI	0.4546	8	0.0568	3.04	0.1181
	Quadratic	0.1367	5	0.0273	1.46	0.3435
	Cubic	0.0408	1	0.0408	2.18	0.1999
	Pure error	0.0935	5	0.0187
Power density (mW/m2)	Linear	45,749.87	11	4159.08	0.7815	0.6606
	2FI	36,372.98	8	4546.62	0.8543	0.5995
	Quadratic	21,335.81	5	4267.16	0.8018	0.5928
	Cubic	643.84	1	643.84	0.1210	0.7421
	Pure error	26,609.24	5	5321.85

).

All of the results were then analyzed using analysis of variance (ANOVA), as shown in

Table 5. Analysis of variance (ANOVA) for the quadratic model of COD removal efficiency.

Respond	Source	Sum of Square	df	Mean Square	F-Value	p-Value
COD removal efficiency	Model	7649.08	9	849.90	1.50	0.0171
	A—thermal	804.16	1	804.16	1.42	0.0186
	B—sonication	139.64	1	139.64	0.2458	0.0201
	C—alkaline	1063.64	1	1063.64	1.87	0.0194
	AB	221.96	1	221.96	0.3908	0.0215
	AC	1273.80	1	1273.80	2.24	0.0369
	BC	859.83	1	859.83	1.51	0.0281
	A²	304.07	1	304.07	0.5353	0.4812
	B²	40.40	1	40.40	0.0711	0.7951
	C²	3473.47	1	3473.47	6.11	0.0329
	Lack of fit	5251.21	5	1050.24	12.23	0.5271
	R2	0.8917

,

Table 6. Analysis of variance (ANOVA) for the quadratic model of biomass.

Respond	Source	Sum of Square	df	Mean Square	F-Value	p-Value
Biomass (mg/L)	Model	0.4742	9	0.0527	2.29	0.0265
	A—thermal	0.0587	1	0.0587	2.55	0.0175
	B—sonication	0.0642	1	0.0642	2.79	0.0249
	C—alkaline	0.0010	1	0.0010	0.044	0.0146
	AB	0.0022	1	0.0022	0.097	0.0134
	AC	0.0101	1	0.0101	0.440	0.0352
	BC	0.0212	1	0.0212	0.919	0.0371
	A²	0.1587	1	0.1587	6.90	0.0253
	B²	0.0539	1	0.0539	2.34	0.1571
	C²	0.0768	1	0.0768	3.34	0.0978
	Lack of fit	0.1367	5	0.0273	1.46	0.3435
	R2	0.9101

A p-value less than 0.05 indicates that the model term is significant.

and

Table 7. Analysis of variance (ANOVA) for the quadratic model of power density.

Respond	Source	Sum of Square	df	Mean Square	F-Value	p-Value
Power density (mW/m2)	Model	65,212.51	9	7245.83	1.51	0.0421
	A—thermal	25,964.81	1	25,964.81	5.42	0.0310
	B—sonication	1826.47	1	1826.47	0.381	0.0408
	C—alkaline	3265.08	1	3265.08	0.681	0.0344
	AB	6091.25	1	6091.25	1.27	0.2860
	AC	3188.00	1	3188.00	0.664	0.4338
	BC	97.64	1	97.64	0.020	0.8894
	A²	618.93	1	618.93	0.129	0.7268
	B²	5176.80	1	5176.80	1.08	0.3232
	C²	11,755.23	1	11,755.23	2.45	0.1485
	Lack of fit	21,335.81	5	4267.16	0.801	0.5928
	R2	0.8794

A p-value less than 0.05 indicates that the model term is significant.

for COD removal efficiency, biomass, and power density, respectively. By analyzing the correlation coefficient, ANOVA was also used to measure the precision of the given model (R²). The R² values for COD removal efficiency, biomass, and power density in this investigation were 0.8917, 0.9101, and 0.8794, respectively.

The relationship of the three independent factors for COD removal efficiency, biomass, and power density exhibited a good fit using the quadratic model. The regression coefficients of the three factors are defined below (Equations (1)–(6)).

COD removal efficiency (coded factor)

+75.8544 + 9.12475A − 3.37402B − 10.5015 C − 4.42925AB − 12.6185 AC + 8.71933BC − 4.677A² + 0.986037 B² − 15.8396 C².

(1)

COD removal efficiency (actual factor)

−6248.89 + 70.3692(Thermal) + 13.832(sonication) + 233.167 (alkaline) − 0.17717 (Thermal × sonication) − 1.26185 (thermal × alkaline) + 0.871933 (sonication × alkaline) − 0.18708 (thermal²) + 0.0394415 (sonication²) − 3.95989 (alkaline²).

(2)

Biomass (coded factor)

+2.66537 − 0.0779299 A − 0.0723381B + 0.0103275 C − 0.0140568 AB − 0.0355999AC − 0.0432544BC + 0.106857A² + 0.0360019B² − 0.0744689C².

(3)

Biomass (actual factor)

80.1517 − 1.16554 (thermal) + 0.0499017 (sonication) + 0.962416 (alkaline) − 0.000562271 (thermal × sonication) − 0.00355999 (thermal × alkaline) − 0.00432544 (sonication × alkaline) + 0.00427427 (thermal²) + 0.00144007 (sonication²) − 0.0186172 (alkaline²).

(4)

Power density (coded factor)

107.166 + 38.2588 A − 28.5984B + 16.6766C.

(5)

Power density (actual factor)

−933.07 + 7.65176 (thermal) − 5.71968 (sonication) + 8.33828 (alkaline).

(6)

The regression model-predicted results were compared to the experimental data. The predicted values, as seen in Figure 1, generally followed a straight line as compared to the experimental data. The ANOVA test indicated that the predicted values were extremely comparable to the observed value, justifying the quadratic model. Figure 2 shows the same result, with the normal percentage probability plot of residuals following a straight line, suggesting that the errors were normally distributed and the model was appropriate.

• COD removal efficiency

The highest COD removal in ML-MFC was 80.685% for a temperature of 140 °C, sonication at 20 kHz, and alkaline conditions at pH 10 after 7 days (168 h) of incubation according to Design Expert 13.0.

To visualize the significant interaction terms of the predicted model (Equations (1) and (2)) on COD removal efficiency, the results of the ANOVA (Table 6) are shown in the form of contour plots and response surface plots. The response is highlighted on these plots for the interaction between sonication and thermal (Figure 3), between alkaline and thermal (Figure 4), and between alkaline and sonication (Figure 5) pretreatment. These interactions formed a parabolic cylinder. All interactions in the experimental domain had a maximum ridge, according to the pattern. The thermal, sonication, and alkaline treatment methods yielded the maximum COD removal effectiveness at 148.41 °C, 20 kHz, and pH 10, respectively.

• Biomass

The highest biomass in ML-MFC was 7.8539 mg/L when the process conditions were a temperature of 140 °C, sonication at 20 kHz, and alkaline conditions at pH 10 after 7 days (168 h) of incubation according to Design Expert 13.0.

To visualize the projected model’s important interaction terms (Equations (3) and (4)) on biomass, the results of the ANOVA (Table 7) are displayed as response contour plots and surface plots. These plots highlight the interaction between sonication and thermal (Figure 6), between alkaline and thermal (Figure 7), and between alkaline and sonication (Figure 8) pretreatment. These interactions formed a parabolic cylinder. All interactions in the experimental domain had a maximum ridge. As the thermal, sonication, and alkaline pretreatment parameters approached 140 °C, 20 kHz, and pH 10, respectively, the maximum voltage generation was achieved.

• Power density

The highest power density in ML-MFC was 220.089 mW/m² when the process conditions were a temperature of 140 °C, sonication at 20 kHz, and alkaline conditions at pH 10 after 7 days (168 h) of incubation according to Design Expert 13.0.

To illustrate the projected major interaction terms of the model (Equations (5) and (6)) on biomass, the ANOVA results (

Table 8. Optimum conditions as suggested by RSM using Design-Expert.

Parameters	Unit	Value
Sonication	kHz	20.8916
Thermal	°C	140.632
Alkaline	pH	11.0583
Response 1—biomass	mg/L	7.34048
Response 2—COD removal	%	86.192
Response 3—power density	mW/m2	207.108
Desirability		1.000

) are displayed as response contour plots and surface plots. The response surface plots and contour plots highlight the interactions between sonication and thermal (Figure 9), between alkaline and thermal (Figure 10), and between alkaline and sonication (Figure 11) treatment. These three interaction terms formed a parabolic cylinder. The interaction in Figure 9 presents an inappropriate interaction with maximum power density. It can be seen that the highest power density was achieved when the thermal, sonication, and alkaline pretreatment parameters approached 140 °C, 20 kHz, and pH 10, respectively, declining beyond these values.

2.2.2. Process and Validation of Models

A vital step of the experimental study was to determine the optimum conditions for operation of the ML-MFC. Table 8 presents the suggested conditions for generating the highest COD removal efficiency, biomass, and power density.

The results suggested that, for the ML-MFC, the COD removal efficiency, biomass, and power density generated under optimum conditions were 86.192%, 7.34048 mg/L, and 207.108 mW/m², respectively. The value of desirability was 1.00. A response over the permissible limit would yield a desirability value of ‘0’, whereas a desirability value of ‘1’ indicates an ideal response. Five further experimental runs were performed to test the accuracy and justify the indicated optimum.

Table 9. Experimental results and predicted data for COD removal, biomass, and power density.

	COD Removal			Biomass			Power Density
No.	Exp.	Pred.	ε	Exp.	Pred.	ε	Exp.	Pred.	ε
1	82.002	86.19	−5.19	7.102	7.340	−0.23	198.400	207.10	−8.70
2	84.874	86.19	−1.31	7.119	7.340	−0.22	202.100	207.10	−5.00
3	87.300	86.19	1.108	7.404	7.340	0.063	209.970	207.10	2.862
4	89.351	86.19	3.159	7.588	7.340	0.248	211.471	207.10	4.363
5	90.062	86.19	3.870	7.542	7.340	0.202	213.620	207.10	6.512
Mean	86.7178	86.192	0.3258	7.35142	7.34048	0.01094	207.1122	207.108	0.0042
SD	3.320			0.230			6.529
95%cl	81.8824–90.5016			6.973456–7.707504			196.7526–217.4634

ε: error, SD: standard deviation, 95%: range within confidence level.

shows the arrangement of the results for the validation runs.

The mean and standard deviations were calculated for the experimental runs. The experimental and predicted data were compared according to their ranges within the 95% confidence level. The calculation revealed that the mean COD removal efficiency, biomass, and power density were 86.7178%, 7.35142 mg/L, and 207.1122 mW/m², respectively, with standard deviations of 3.32019, 0.23006, and 6.52959, respectively. The 95% confidence intervals for COD removal efficiency, biomass, and power density were 81.8824% to 90.5016%, 6.973456 to 7.707504 mg/L, and 196.7526 to 217.4634 mW/m², respectively. It can be concluded that the experimental values and the projected values were identical within a 95% confidence interval, and that the ideal conditions identified by RSM were correct. Thus, RSM using CCD was found to be an effective strategy for optimizing the process conditions of ML-MFC for COD removal efficiency, biomass, and power density.

2.3. ML-MFC Performance

2.3.1. Voltage Generation and Power Density

As shown in Figure 12, the voltage increased with time; the highest voltage was generated at 48 h (262 mV) with a power density of 220 mW/m². In general, bacteria grow in the ML-MFC by consuming the substrate and generating electrons, thus creating a redox potential between the anode and cathode, which generates voltage. After 48 h, the bacterial growth started to slow down and fluctuate, as also observed in the trend of power density.

2.3.2. Power Density, COD Removal Efficiency, and Biomass

Figure 13 illustrates the trends of power density and COD removal efficiency in the ML-MFC for the best pretreatment conditions (140 °C, 20 kHz, and pH 10). As can be seen, the results for power density and COD removal efficiency fluctuated slightly at the early stage. After 4 h, power density dropped to 29.766 mW/m², while COD removal efficiency decreased to 6.443% at t = 6 h. After 6 h, the power density increased rapidly to reach maximum growth, along with optimum power density (220 mW/m²) and COD removal efficiency (80.68%), at t = 48 h. Then, the trend fluctuated and decreased for both power density and COD removal efficiency. This is due to the activation losses occurring during the transfer of electrons from or to the compounds reacting at the electrode surface. In general, an increase in power density resulted in increased COD removal efficiency. This phenomenon indicated that the studied electrogenic bacterium, Bacillus subtilis (BS), was active in degrading the chicken manure substrate to generate electricity with the presence of sufficient supporting enzymes such as β-glucosidase, alkaline serine protease, β-mannosidase, and thioglycosides, secreted by BS, which facilitated the degradation of chicken manure substrate in the exponential phase.

All responses of EB biomass, power density generation, and COD removal were correlated. Generally, as the ML-MFC systems were applied, the Bacillus subtilis present in the chicken manure substrate was degraded. The degradation of substrate in ML-MFC system generated the voltage, which could be attributed to power density [29]. As shown in Figure 13, the graph trend for COD removal and power density was similarly aligned with the growth profile of EB biomass.

The kinetic growth profile, doubling time Td, and its specific growth µ are presented in

Table 10. Kinetic growth profile of BS growth using untreated and treated chicken manure in the ML-MFC.

Condition	Doubling Time, Td	Specific Growth Rate µ
Untreated chicken manure	6.08 h	0.11 h−1
Treated chicken manure	3.43 h	0.20 h−1

.

The doubling time is the time required for the cells to double in size, which results in the cell mass and cell number density increasing exponentially with time. The specific growth rate is the rate at which a cell population’s biomass increases per unit of biomass concentration. The specific growth rate and doubling time of BS were 0.11 h⁻¹ and 6.08 h, respectively (Table 10). In order to achieve the maximum biomass in a shorter time, a higher specific growth rate and an earlier doubling time are required. The calculated results for µ and Td in the ML-MFC were 0.202107 h⁻¹ and 3.4296 h, respectively. There were slight differences in the growth of BS in the shake flask and in the ML-MFC. This is because their growth originated from the same soluble COD.

As shown in Table 10, µ and Td for treated chicken manure in the ML-MFC were 0.20 h⁻¹ and 3.43 h, respectively, while these values for untreated chicken manure were µ = 0.11 h⁻¹ and 6.08 h. The higher µ and earlier Td indicate that BS had the ability to create maximum biomass productivity in a shorter period of time. The thermal, alkaline, and sonication pretreatment helped in the degradation of chicken manure. Pretreatment simplified the biomass, thus facilitating the consumption of nutrients by BS. The degradation by BS generated electrons, thus improving the power generation in the ML-MFC. Even though there were slight differences in the values of µ and Td between these two conditions, the growth profile showed the same trend.

2.3.3. Performance of ML-MFC

The polarization was elaborated to test the performance of the ML-MFC. The data were obtained for the best conditions (140 °C, 20 kHz, and pH 10) of the ML-MFC according to Design expert 3.0. Five types of resistors were included (47, 100, 470, and 1000 Ω) in the polarization curve. The power density and current were calculated using Ohm’s law (V = IR, P = IV). Figure 14 shows the trend of power density and voltage generated. The highest power density of 220 mW/m² was achieved at 47 Ω, as identified from the peak of the polarization curve. When current increased from the start point to 0.12 mA, a rapid increase in voltage was observed (from 30 to 120 mV). Then, when the current increased from 0.12 mA to 5.57 mA until the highest power density of 220 mW/m² was attained, the voltage increased from 120 to 262 mV.

The polarization curve was characterized by three losses (activation losses, ohmic losses, and mass transfer losses) [30], which need to be minimized to improve the performance of the MFC. Activation losses are caused by the slowness of the reaction taking place on the electrode surface, resulting in a low power density and a drop in cell potential. This phenomenon can be attributed to the energy lost (as heat) during the oxidation and reduction reactions, as well as the energy lost in transferring electrons from the cell terminal protein or enzyme of the electrogenic bacteria to the anode surface. This loss can be minimizing by increase the surface area and optimizing the operating temperature for better electron transfer.

Ohmic losses include the resistance flow of electrons through the electrode and the resistance flow of ions. They can be limited by minimizing the electrode distance to reduce the loss of energy during the transmission of protons and by ensuring good contact between the circuit and electrode. Mass transfer losses occur at high current densities due to the limited mass transfer of chemicals via diffusion to the electrode. Thus, insufficient mass transport causes reactant depletion or product accumulation. These losses can be overcome by using a proton exchange membrane (PEM) with a higher mass transfer area. The highest power density of 220 mW/m² was achieved when the ML-MFC was operated under closed-circuit voltage (CCD) with a 47 Ωresistor (Figure 15).

3. Methodology

3.1. Sample Characterization

Chicken manure was obtained from a broiler farm at Permatang Berangan TasekGelugor, Penang, Malaysia. The collected sample was kept at 4 °C and thawed to room temperature (27 °C). The sample was then autoclaved to remove pathogens and sterilize the chicken manure sample prior to use in the study.

3.2. Analytical Methods

3.2.1. Elemental Analysis

An elemental analyzer (PerkinElmer 2400 series II,), Germany) was used to evaluate the content of hydrogen (H), nitrogen (N), carbon (C), and sulfur (S) in the chicken manure at 20, 15, and 60 psi. The apparatus was purged using purified hydrogen, oxygen, and compressed air. The combustion and reduction furnaces were set to 9758 °C and 5008 °C, respectively.

3.2.2. Atomic Absorption Spectrometry (AAS)

The trace elements in chicken manure were determined using atomic adsorption spectrometry (AAS). The chicken manure underwent acid digestion (microwave method) before further analyzing its trace elements. For acid digestion, 1 g of dry chicken manure sample was added to MF/HF vessels in the microwave. Then, 10 mL of concentrated HNO₃ was added (in a fume hood), and the mixture was carefully swirled. Then, 5 mL of concentrated hydrochloric (HCl) acid was added, and the vessels were closed tightly with a lid. The vessel was inserted onto the rotor of the microwave. The microwave digester system was operated using the following profile: temperature, 100 °C; ramp, 10 min; hold, 10 min; cool, 30 min. After cooling, the solution was filtered through Whatman No. 42 filter paper into a 50 mL volumetric flask and then deionized water was added until the mark. The sample solution was used as the standard solution [31].

3.2.3. Inductively Coupled Plasma Optical Emission Spectrometry (ICP-OES)

Micronutrients were detected using inductively coupled plasma optical emission spectrometry (ICP-OES). Before analysis, the sample of chicken manure underwent microwave acid digestion as described in Section 3.2.2. After acid digestion, the sample was diluted to 50 mL with deionized water. Then, the sample was used for ICP-OES. Standards that were used for each element were 1000 ppm [32,33].

3.2.4. Determination of Biomass

A diluted (9 mL deionized in 1 g sample) sample solution was placed in a 30 mL falcon tube and centrifuged (4000 rpm, 15 min); then, around 1.5 mL of sample was added into a cuvette. The OD value was measured at a wavelength of 600 nm using a UV/Vis spectrometer (HITACHI U-1900) (Germany) with deionized water used as the blank. The OD results were recorded and then translated into biomass (mg/L) using the formula from the standard calibration curve of biomass (y = 0.1115x) (Appendix A, Figure A1).

3.2.5. Determination of Chemical Oxygen Demand (COD)

Chicken manure (1 g) was diluted in sterilized deionized water (9 mL), and then the solution was centrifuged at 4000 rpm for 15 min. Then, the sample was filtered using a 0.02 pore size syringe filter (Ks-Tek). The COD vial contained 3 mL of premixed chemicals (HgSO₄, K₂Cr₂O₇, AgNO₃, potassium hydrogen phthalate, H₂SO₄, Ag₂SO₄, and K₂Cr₂O₇), mixed with 2 mL of filtrate. The blank was prepared by adding 2 mL of deionized water into the vials. The COD vials were digested at 150 °C for 2 h using COD digester (Lovibond, Germany) The machine was heated for 15 min prior to digestion. After the vials were cooled to room temperature, the COD value was measured using a COD kit (Checkit Direct, Lovibond) [34]. The experimental COD removal was calculated as a function of the initial COD value and the final COD value after the reaction in MFC according to the equation below [35].

$$\mathrm{COD}\mathrm{removal} \%=\frac{{\mathrm{COD}}_{\mathrm{i}}{-\mathrm{COD}}_{\mathrm{f}}}{{\mathrm{COD}}_{\mathrm{i}}}\times 100\%.$$

(7)

3.3. Configuration of ML-MFC

The ML-MFC was set up using a cylindrical PVC vessel with a diameter and height of 6.5 cm and 10 cm, respectively. The anode (graphite felt) was located at the bottom of vessel, and 46 g of chicken manure was added to 3 cm height (electrode distance) to cover the anode. Later, the cathode (graphite felt) was placed at the top, and the upper surface of the cathode was exposed to air. The vessel was then covered with a lid and incubated at 35 °C for 7 days. The probe was connected to the anode and cathode to measure the power generation using a digital multimeter (UNI-TUT33D) device. Both electrodes had a 3.25 cm radius, with a thickness of 0.02 cm and surface area of 0.0033187 m². Chicken manure acted as a pseudo-membrane with 70% moisture content, and 10% inoculum (raw broth chicken manure) was added into the ML-MFC. Figure 16 illustrates a schematic diagram of the ML-MFC employed in this research.

3.4. Statistical Experimental Design

Optimization Using Response Surface Methodology with Central Composite Design (CCD)

The RSM method was used for optimization of the pretreatment. In RSM, the independent parameters (thermal, sonication, and alkaline pretreatment) were chosen with suitable ranges coded as −1, 0, and +1. This coded values indicated the lower, middle, and higher values, respectively. The conditions of each parameter utilized for ML-MFC optimization are summarized in

Table 11. Levels of variables chosen for process parameter optimization in RSM using CCD.

		Coded Variables Level
		Low	Centre	High
Factor	Variables	−1	0	+1
A	Thermal pretreatment (°C)	135	140	145
B	Sonication pretreatment (kHz)	15	20	25
C	Alkaline pretreatment (pH)	8	10	12

. RSM involved about 20 runs in random order to cover all combinations of the factor levels. The responses of COD removal efficiency, biomass, and power density were reported as averages.

3.5. Determination of Power Using Polarization Curve

A polarization curve was plotted as a function of the voltage measurement recorded by a multimeter (UNI-T). Five types of resistors (47, 100, 220, 470, and 1000 Ω) were used for the polarization curve. The polarization curve was plotted with the x-axis (current), y-axis (voltage), and z-axis (power density) derived from measurements [36]. The maximum power of the ML-MFC was identified at the peak of the power curve. The power and current generation were calculated according to Ohm’s law as expressed below.

$$\mathrm{V}=\mathrm{IR},$$

(8)

$$\mathrm{P}=\mathrm{VI},$$

(9)

where P is the power, I is the current, V is the voltage, and R is the resistance.

3.6. Operation of ML-MFC

The ML-MFC was operated using chicken manure as a substrate following several pretreatment methods (thermal, sonication, and alkaline pretreatment) in order to determine the power density, biomass generation, and COD removal. The treated chicken manure underwent incubation at 35 °C. During the incubation period, samples were taken every 2 h until 6 h on the first day (to observe the lag phase phenomenon), and then continued at 12 h intervals for 7 days. The power density, biomass, and COD values were recorded at each interval. The power density was measured using a digital multimeter (UNI-T).

3.7. Pretreatment of Chicken Manure

3.7.1. Thermal Pretreatment

To study the thermal pretreatment, the dry chicken manure was heated at different temperatures in the range of 140 °C to 148 °C for RSM. The thermal pretreatment was performed in an oven (Memert, Germany) and took around 4 h. After this period, the sample was left to cool before continuing onto the next pretreatment.

3.7.2. Sonication Pretreatment

Sonication pretreatment was performed after thermal pretreatment. It involved the application of a probe sonicator device for 6 min in a range from 10 kHz to 30 kHz. The process involved adding deionized water at appropriate amounts to enable the sample to sonicate.

3.7.3. Alkaline Pretreatment

Alkaline pretreatment was performed by adding 1 M sodium hydroxide (NaOH) to the sample of chicken manure after sonication pretreatment to adjust the pH to a value from pH 8 to pH 10. Then, the sample was mixed and left for 30 min at room temperature before adding 10% of raw broth inoculum. Then, the mixture of chicken manure was added between the electrodes (3 cm) in the container and underwent fermentation in the incubator (37 °C) for 7 days. Voltage generation, COD removal, and biomass were monitored within this time interval using a multimeter, COD kit, and UV/Vis spectrometer, respectively.

3.8. Preparation of Standard Calibration Curve

One gram of sterilized chicken manure was diluted in 1000 mL of sterilized deionized water, and then mixed for 30 min at 80 °C. The solution was added into 15 mL falcon tubes with different amounts of sample in a concentration range from 0.1 to 1.0 mg/L to generate the stock solution. Then, the samples were measured to determine the optical density (OD) using a UV/Vis spectrometer (HITACHI U-1900, Düsseldorf, Germany) at a wavelength of 600 nm. The graph of concentration solution versus absorbance was plotted to obtain a linear calibration curve. The OD results for each sample were translated into biomass using the equation from the linear calibration curve (y = 0.1115x).

3.9. Growth Profile of Bacillus Subtillis

3.9.1. Preparation and Fermentation of Raw Broth from Chicken Manure

First, 100 g of sterilized chicken manure was added to 900 mL of deionized water and mixed for 30 min at 80 °C on a hot plate. The solution was left to settle for 15 min until two layers were formed. The solution was filtered using a vacuum pump, and the filtrate was centrifuged at 5500 rpm for 5 min. Next, supernatant was autoclaved for 15 min at 121 °C. The method proposed by Tranter (2016) [37] was slightly modified in order to identify the availability of bacteria.

The sterilized raw broth was added with a single colony of Bacillus subtilis under aseptic conditions and then fermented in an incubator shaker at 150 rpm and 35 °C for 48 h. Samples were taken every 2 h for the first 6 h and then at 6 h intervals. The biomass of a around 2 mL sample was recorded at every interval to check the OD value using a UV/Vis spectrometer.

3.9.2. Specific Growth Rate of Electrogenic Bacteria

The specific growth rate of Bacillus subtilis in raw broth was calculated using the equation below.

$$\frac{\ln X-\ln Xo}{t}=\mu ,$$

(10)

where X is the number or mass of cells, Xo is the initial number or mass of cells, t is the time, and $\mu$ is the specific growth rate.

3.9.3. Doubling Time of Electrogenic Bacteria

The doubling time of electrogenic bacteria [38] was calculated using the equation below.

$$Td=\frac{\ln 2}{\mu },$$

(11)

where Td is the doubling time, and µ is the specific growth rate.

4. Conclusions

The effect of pretreating chicken manure for electricity generation using an ML-MFC was determined. Chicken manure contains a high percentage of carbon, with values of 34.20% and 35.12% for untreated and treated chicken manure, respectively. The most common micronutrient in the substrate was Mg (71.61 mg/L). The COD values for treated chicken manure and untreated chicken manure were 610 mg/L and 517 mg/L, respectively. The nutrients consumed by EB rely on this COD value. The optimization of the pretreatment was achieved using respond surface methodology (RSM) with a CCD. The results showed that the best conditions for thermal, sonication, and alkaline pretreatment were 140 °C, 20 kHz, and pH 10, with the ML-MFC providing the highest COD removal efficiency, biomass, and power density at 80.68%, 7.8539 mg/L, and 220 mW/m², respectively. The optimal parameter conditions suggested by the RSM for the ML-MFC system were validated. The 95% confidence levels for COD removal efficiency, biomass, and power density were 81.8824% to 90.5016%, 6.973456 mg/L to 7.707504 mg/L, and 196.7526 to 217.4634 mW/m², respectively. Thus, RSM using a CCD was shown to be an effective tool for optimizing the process conditions of the ML-MFC in terms of COD removal efficiency, biomass, and power density.