THE MODELLING OF A MICROBIAL FUEL CELL

(1)

THE MODELLING OF A MICROBIAL FUEL CELL

Bachelor Thesis

(2)

Samenvatting

Men wordt steeds bewuster dat de mensheid een sterke bijdrage levert aan klimaatveranderingen door hun broeikasgasemissies. Door deze trend zal de mondiale drang voor een transitie naar een duurzame, hernieuwbare energiebronnen toenemen. Er wordt een scala aan veelbelovende technologieën ontwikkeld die potentieel kunnen bijdragen in deze energietransitie. Een van deze veelbelovende technologieën zijn microbiële brandstofcellen ofwel MFCs. Dit zijn apparaten die chemische energie dat vrijkomt bij de microbiële afbraak van organisch materiaal om zet in elektrische energie. Gedurende de laatste decennia hebben MFCs meer aandacht gekregen en wordt er uitgebreid onderzoek over gedaan. Desalniettemin is het nog steeds een technologie die in de kinderschoenen staat en nog meer onderzoek vereist. Microbiële brandstofcellen zijn redelijk complexe systemen die meerdere disciplines als microbiologie en elektrochemie bevatten. Onderzoek met microbiële brandstofcellen is relatief tijdrovend en de materialen om de brandstofcel te bouwen hebben hoge kosten. Modelleren en simuleren kan gebruikt worden als een effectief methode om onderzoek te doen met MFCs en kost daarnaast minder tijd en geld in vergelijking met een experimentele opstelling. In dit onderzoek is een computer model gebouwd en vergeleken met data van een experimentele opstelling van een microbiële brandstofcel. Het model bevat de basisprocessen die voorkomen in een MFC. De gemodelleerde processen bevatten modellen met massatransport (diffusiemodellen), een elektrochemisch model (Nernst vergelijking) en een model dat de snelheid van glucose decompositie beschrijft (gebruikmakende van een Monod model). Het gedrag van het voltage in vooraf-ingestelde modellen hebben een enkele overeenkomsten in vergelijking met de experimentele data. Echter is het model nog niet in staat om een precieze en adequate voorspelling te geven van het werkelijke voltage. Om dit wel te bereiken zullen complexere relaties moeten worden toegevoegd aan het uitbreiden van het model.

(4)

Abstract

The realisation that humans are contributing to climate change by greenhouse gas emissions is growing larger. With this trend, a global urge for the transition towards renewable energy sources has been and is still growing as well. Many promising technologies are being developed that could potentially contribute to a sustainable future. One of these technologies are microbial fuel cells – or MFCs. These are devices that convert chemical energy released by de microbial decomposition of organic matter, to produce electricity. Last decades MFCs are drawing a lot of attention and is research extensively. Nevertheless, it is still a technology in a developing state and requires more research. Microbial fuel cells are complex systems, that involve multiple disciplines such as microbiology and electrochemistry. In addition to this, microbial fuel cells are dependent on lots of external and internal variables. Research on microbial fuel cells is time consuming and the materials to design and produce microbial fuel cells are expensive. Modelling and simulation can be an effective tool to gain understanding of microbial fuel cells and it also saves time and money compared to experimental setups. In this research, a computer model is built and compared with data from an experimental setup of a microbial fuel cell. The model contains the principle processes for modelling a fuel cell and consisting of mass transport models (diffusion models), electrochemical model (Nernst equation) and a model representing the rate glucose decomposition (using a Monod model). The behavior of the voltage output in preset situations of the model show rough similarities compared to the experimental setup. However, the model is not yet able to give an adequate and precise prediction of the voltage output. Therefore, more complex relations are required implement in extension of the model.

(5)

Introduction

At the end of 2015 during the COP21 in Paris, the world acknowledged the threat that global climate change poses to our future. The key result of this conference was an arrangement to keep the global increase in temperature below two degrees Celsius compared to pre-industrial levels. To reach this goal, nations worldwide will need to reduce their greenhouse gas emissions to zero between 2050 and 2100 (UNFCCC, 2015). Part of achieving this requires a worldwide transition from fossil fuel sources towards more sustainable and renewable energy sources. In this transition the common renewables like solar and wind energy are most likely to take the leading roles in this transition. Nonetheless, newly emerging technologies must also be accounted in order to fulfil this transition. One of these technologies are microbial fuel cells (hereinafter named MFCs).

Although the first sight of bacteria generating electrical energy was observed by Potter in 1911, the phenomenal has not received much attention until the early 1900s (Allen and Bennett, 1993). In the last decades the interest has grown and MFCs have been studied with increasing intensity as a potential green renewable energy source. This technology uses the natural process microbial decomposition of organic material to produce electrical energy. The process of decomposition is a reduction-oxidation (redox) reaction, which means that there is a transfer of electron(s) occurring during the reaction. The transfer of electrons is also the key characteristic of a fuel cell, due to the fact that charged particles that are flowing through a wire generates electricity. By controlling and manipulating the process of microbial decomposition of organic matter, a fuel cell can be designed where it becomes possible that the electrons that are produced are able to flow through a wire and generate electricity. There are three specific types of MFCs that are most commonly found in literature. These are plant MFCs, sediment MFCs and wastewater MFCs. All these types make use of the same principle of microbial decomposition to produce electricity. However, they differ in for instance fuel source and prospect of where and how it can be implemented.

The plant microbial fuel cell (PMFC) is a type of MFC that gets a direct fuel supply from a plant that is rooted in the soil. According to a study where C14 _{is used as tracer compound concluded}

that on average 17% of the total carbon that is fixed by plants is wasted through the roots and 12% is respired - broken down by microorganisms (Nguyen, 2003). Another source states that in most vascular plants a substantial amount (20%-60%) of the organic material that is fixed by the plant by photosynthesis is excreted through the roots. From this, a fraction in the range of 15%-60% is used for respiration in the soil (Neumann, 2011). The precise fraction that is excreted and used for respiration is dependent on a variety of factors such as the species of plants, size of microbial population and chemical composition of the soil. Also environmental

(6)

Sediment microbial fuel cells (SMFC) are MFCs that make use of the sediment from the bottom of water bodies (i.e. lakes, marshes, rivers, etc.) that preferably contain high amounts of organic matter. Like the PMFC the fuel is organic matter from the plant, however in a SMFC is organic matter is from dead plant parts instead of exudates from a living plant. Due to this difference, SMFC tend to have a less continuous source of fuel than PMFCs which could be a disadvantage. What offer a great opportunity is combining a PMFC and a SMFC. An aquatic environment offers advantages like higher conductivity and faster diffusion than a non-aquatic environment. Saline environments could even increase the conductivity and boost its potential. By using plant species that grown in aquatic environments, for example rice or species native to swamps, the continuous fuel flow of a PMFC can be combined into an environment typical to SMFCs to produce a fuel cell with a high potential (Kothapalli, 2013). The third technology are microbial fuel cells that use organic compounds like acetate, lactate and glucose that are present in wastewater. These compounds are broken down by microbes in anaerobic conditions to generate electricity. Because wastewater is the carrier of the fuel source, this technology focusses on implementation in wastewater treatment plants with the aim to making such a plant more self-sufficient by producing its own energy (Oh et al., 2010; Du et al., 2007).

As indicated earlier, MFCs are extensively researched. However, this research is often time consuming and expensive. Nowadays the costs to produce a, for current standards, high performance PMFC is approximately €600 per square meter. Of this the major part accounts the electrodes and membrane (Plant-E, 2015). Additionally, there is a large variety of (external) factors that can influence the electrical output of the fuel cell. This ranges from the type of microbe that decomposes the organic material, the material used to construct the MFC or environmental factors such as salinity, nutrient availability or acidity. If one desires to examine these factors in different experimental setups to review their influence on the process, it will become expensive and time consuming. An alternative for these experimental setups are computer models. A model that is validated can be used as a tool to conduct some of the same research more costs and time effectively (Oliveira et al., 2013a).

Conventional fuel cell such as direct methanol fuel cells or hydrogen fuel cells are more commonly researched. Although there are significant differences between a conventional fuel cell and a microbial fuel cell that have to be taken into account, there are also strong resemblances in some basic principles. There are already models produced for conventional fuel cell. Despite the fact that a MFC is not simply an extension of a conventional cell, the model basics of conventional fuel cells can be used to produce a model of a MFC.

(7)

Aim and objectives

In this research a simplified model will be designed and produced that is be able to estimate the behavior of the electrical output under certain circumstances. The model describes the main processes that occur in a MFC using basic mathematical relations. To analyse whether the model resembles a realistic output, the model is verified using an experimental setup of a MFC.

In order to achieve this objective, several sub questions can assist to achieve the aim of this research. These are:

 How do the main processes in a microbial fuel cell work?

 How can the basic processes in a microbial fuel cell be describes in mathematical relation?

 What processes are most common used in modeling fuel cells - in particular microbial fuel cells?

 Which assumptions for the model are appropriate to make in order to obtain a simplified, but adequate representation of a microbial fuel cell?

 What are more complex processes that could be included to add complexity in the model?

Theoretical background

As mentioned previously, MFCs are showing strong resembles with a plain fuel cell. In both cases electricity is produced by using chemical reactions known as redox reactions. These are reactions where an electron is transferred from one atom or molecule to another, or in other words there is a transfer of a charged particle. Also the design is similar: both consist of an anode chamber with an anode, a semipermeable membrane and a cathode-chamber with a cathode. Despite the similarities, there are also differences. For example, a conventional fuel cell is based on spontaneous occurring chemical reactions and do not require a catalyser or other assisting processes. A MFC needs microbes to catalyse the process of the decomposition of organic material. The performance of the MFC is therefore dependent on the presence and health status of a microbial population on the anode. Another difference is that the fuel is less predictable in a MFC. A conventional fuel cell like a direct methanol fuel cell is characterised by one reaction with one fuel source (methanol) that occur continuously. However, the microbes in a MFC are not restricted to one fuel type, but a variety of organic compounds that are present in the mixture of organic material that is decomposed in different reactions. The different reactions that may occur can differ in the electrical potential which makes predicting of output more challenging.

(8)

The organic material for a MFC can originate from different sources. In a PMFC it is ideal if exudates that were excreted through the root system is used as source. However, for a SMFC, dead plant parts are used as source for organic material. For this research, glucose is selected as the representative substance. This is due to the fact that glucose is one of the more common molecules in organic matter and it is a simple sugar molecule that can be decomposed by most microbes.

The microbial decomposition of the monosaccharide glucose proceeds as follows: C6H12O6 + 6O2  6CO2 + 6H2O

This is a reduction-oxidation reaction that can be subdivided into two reactions. In the first part of the reaction electrons are produced and in the second part these electrons are used to complete the reaction. This reaction can occur in two situations either in the presence of oxygen or the presence of water in the anode chamber. The two half reactions are different, but result into the same total reaction. For electricity production the anode reaction with water is preferred, because it produces twice the amount of electrons and thus a higher energy output potential.

For situations where oxygen is present on the anode, the following reaction will occur: C 6H12O6 + 3O2  6CO2 + 12H+ + 12e- (anode reaction)

3 O2 + 12H+ + 12e-  6H2O (cathode reaction)

For situations where oxygen is absent, or less abundance, on the anode, the following reaction will occur:

C 6H12O6 + 6H2O  6CO2 + 24H+ + 24e- (anode reaction)

(9)

In a soil the anode reaction is caused by the microbe. This means that the electrons are in fact a waste product. The reaction products are transported out of the cell into the surrounding environment. In a normal, non-controlled, situation the produced electrons react almost instantaneously, because loose hydrogen ions and electrons are reactive. The hydrogen ions, electrons will react with oxygen molecules that are present resulting into water molecules, and will thereby complete the redox reaction. The transportation of especially the electrons could pose a challenge for a microbe since the cell membrane consist of non-conductive materials phospholipids and polysaccharides (Kothapalli, 2013). Many microbes have evolved to cope with this challenge by producing so-called nanowires. These are microscopically small conductive wires that a microbe makes to transport the electrons that are produced in the reaction from

inside the cell to outside the cell to a place where they can discard their electrons (Logan, 2006; Gorby et al., 2006). This is useful for a MFC, since the microbes can dispose their electrons at an anode through the conductive wires. Besides this, it poses a beneficial situation for microbes, resulting in a biofilm of these microbes on the anode.

FIGURE 1:MICROBES THAT FORMED NANOWIRES ON THE ANODE OF A MFC.

(10)

Methods & Materials

This research consists out of two parts; producing a simplified computer model and an experimental setup to validate the model. The design and explanation of the model is the first part of this chapter and in the second part the experimental setup will be discussed.

Model

The first aspect that is addressed is the decision regarding the desired complexity of this model. The discussion of complex versus simple models is key in addressing this. Where on the one hand too simple can lead to an inadequate representation of the reality, but on the other hand over-complexity uses more processing time than necessary (and more effort). The processes in a MFC are, like many other natural processes, imbedded in a complex web of processes and influencing factors. Building a model of such processes requires simplifications. The model of the MFC will consist out of roughly three parts: electrochemical part, diffusion model and a model the predicts the rate of glucose decomposition based on a Monod equation (which is often used for models of microbial population size). The electrochemical part will represent the part that calculates the performance. This can be achieved by calculating the difference in electrical potential on the anode and the cathode.

Ecell = Ecathode – Eanode

The potential of the reaction of glucose decomposition with water is equal to the standard electron potential of the first half reaction minus the second half reaction. The standard electron potential (EO) can be calculated by the following definition formula:

𝐸0 = ∆𝐺 −𝑛𝐹

Where ∆𝐺 is the Gibbs free energy, 𝑛 the number of transferred electrons and F the Faradays constant. In the table the standard potentials are calculated as well as the total cell potential.

TABLE 1:STANDARD ELECTRON POTENTIALS OF THE REACTIONS.SOURCE:VANCOUVER ISLAND UNIVERSITY(VIU),2008 Gibbs free energy Standard electron

(11)

However, the electric potential on the cathode and anode are dependent on the concentration of the reacting substances, and is by this means subjected to Nernst law. This equation uses the fraction coefficient of the concentration of the reacting substances to calculate the electrical potential under dynamic concentrations. Nernst law is described as: In the reaction,

𝑝𝐴 + 𝑞𝐵 → 𝑟𝐶 + 𝑠𝐷

The electrical potential can be calculated by, 𝐸_{𝑐𝑒𝑙𝑙} = 𝐸0− (

𝑅𝑇

𝑛𝐹∗ 𝑙𝑛 𝑄) , where 𝑄 =

[𝐶]𝑟[𝐷]𝑠 [𝐴]𝑝_[𝐵]𝑞

This equation can be applied on both the anode and cathode side respectively. Where on the anode side the following reaction happens:

C 6H12O6 + 6H2O 6CO2 + 24H+ + 2e- 𝐸_{𝐴𝑛𝑜𝑑𝑒} = 𝐸_{0,𝐴𝑛𝑜𝑑𝑒} − (𝑅𝑇 𝑛𝐹∗ ln 𝑄) , where 𝑄 = [𝐶𝑂2]6[𝐻+]24 [𝐶6𝐻12𝑂6] [𝐻2𝑂]6 = [𝐶𝑂2]6[𝐻+]24 [𝐶6𝐻12𝑂6] .

On reaction on the cathode side is:

O2 + 4H+ + 4e-  2H2O |6×| 𝐸_{𝑐𝑎𝑡ℎ}= 𝐸_{0,𝑐𝑎𝑡ℎ}− (𝑅𝑇 𝑛𝐹∗ ln 𝑄) , where 𝑄 = [𝐻2𝑂]12 [𝑂2]6[𝐻+]24 = 1 𝑝𝑂2,𝑐𝑎𝑡ℎ𝑜𝑑𝑒[𝐻+]24 .

In a reaction coefficient the water concentration is not accounted for (otherwise it would use the concentration of water in a solution of water). Also the concentration of oxygen ( [𝑂2] ) is

equal to the partial oxygen pressure (pO2). On the cathode the pO2 is similar to the

atmospheric pressure, e.g. 1 atm.

In order to use the Nernst equation in a dynamic model, it is key to know the concentration of the substances at a certain point of time and a certain location in the fuel cell. To obtain this data, models of the mass transportation have to be included in the model. These are diffusion models that are based on Fick’s’ second law of diffusion:

(12)

The diffusion is used to model the dispersion of glucose on the anode side (1) and for the diffusion of hydrogen ions through the membrane (2). The CO2 concentration is not calculated

with diffusion, but it is assumed to be proportional to the glucose concentration that is decomposed on the anode at a ratio of 6:1 (per mole glucose, six mole carbon dioxide is produced). In the model it is assumed that the CO2 concentration does not accumulate in the

anode chamber. The input for the hydrogen ion concentration is also proportional to the glucose concentration that is decomposed, but now on a ratio of 24:1.

Important for the development of the concentration over time is the rate of decomposition of glucose by the microbial population. The size of the microbial population is dependent on the availability of nutrient. However, if the microbial population grows, due to an abundance of nutrients, the decay rate of these nutrients will decrease and this will influence the growth of the microbial population. Modelling the microbial population growth with the presence of a limiting substance can be done by implementing the Monod equation, describes as:

𝜇 = 𝜇𝑚𝑎𝑥 ∗

𝑆 𝐾 + 𝑆

Here the maximum growth rate is influence by a ratio between the reacting substance and the reacting substance plus the concentration of that substance where 𝜇 = 𝜇𝑚𝑎𝑥

2 (= 𝐾). The

limiting factor in the model is the presence of glucose.

When assuming that the size of the microbial population is proportional to rate of glucose decomposition, the formula can be rewritten as,

𝑘_𝑑𝑒𝑐 = 𝑘_{𝑑𝑒𝑐,𝑚𝑎𝑥} ∗ 𝑆

𝐾+𝑆

The last addition to the model is an efficiency factor, or loss factor. Since the processes are simplified and not all processes are taken into account, it is likely that the model will give an overestimation of the reality. This is caused by the fact that the model is treated as an ideal situation. However, there are also processes that cause the fuel cell to be less efficient then an ideal situation. For example, the model assumes that all the microbial decomposition happens in a biofilm on the anode. More realistically would be that it is more likely that there is an increased amount of microbial decomposition happens at the anode, but also in the rest of the fuel cell. Another example is that the model assumes that the two reaction are perfectly

(13)

These examples, as well as others, result in an overall voltage loss. Or in other words, a lower potential that can be reached due these inefficiencies. Note that this is not an efficiency factors over Eanode or Ecathode, but an overall loss. This factor is added in the equation as follows,

Ecell = Ecathode – Eanode – ɳloss

The value of ɳloss is determined a posteriori (after calibration for an experiment with the

setup). A typical value ranges from 0.5-0.9. Although this helps to fit the model during verification, the fact that the factor is determined afterwards makes it arbitrary.

TABLE 2: MODEL PARAMETERS NAMES, VALUES AND UNITS

Name Value Dimension Source

Ecell - Volts (V) Calculated in model

EAnode - Volts (V) Calculated in model

E0,Anode - 0.42 Volts (V) Vancouver Island University(VIU), 2008

Ecath - Volts (V) Calculated in model

E0,cath + 0.82 Volts (V) Vancouver Island University(VIU), 2008

R 8,3144621 J /(K.mole) By definition

T 288 K By definition (at 25°C)

n Cathode:24/Anode:4 #electrons/reaction Half reactions

F 96485,3329 C/mole Faradays constant, definition

Q - - Calculated in model

pO2, cathode 1 Atm. By definition in standard situation

C - mole/cm3 _{Calculated in model}

t - hours Set in model

x - cm Defined in model

∂C/∂t - mole/(cm3_.h) _{Calculated in model}

∂C/∂x - mole/cm2 _{Calculated in model}

Dglucose 0.10-0.25* cm2/hour Jones, et al., 1980

DH+, PEM 0.10 cm2/hour Assumed

Kdec - mole/hour Calculated in model

Kdec, max 1.6e-5 mole/hour Assumed

K 4e-5 mole/cm3 _Assumed

S - mole/cm3 _{Calculated in model}

(14)

Experimental Setup

In order to confirm whether the model is a realistic representation of the actual process, a fuel cell is built to analyse the electrical output in an environment where the basic variables are known. Important for validating the model is that both are comparable with each other in setup and structure.

The build MFC is a single chamber reaction that contains, apart from the building materials, a young humic topsoil with has a high organic matter fraction. Figure 3 gives a schematic illustration of how the MFC is designed. As can be seen, there is an anode chamber

(a container) where the organic material, which is in this research glucose, will be added to the soil in which a healthy microbial population is present. To form a fuel cell, the microbes have to be able to connect their nanowires to the anode. This means that a biofilm of microbes has to grow on the anode. At the beginning, the microbes will not have colonised on the anode yet, but are randomly present in the soil. Time is needed for the microbes to grow on the anode and produce their nanowires. According to a model from Picioreanu et al. (2008) the biofilm on the anode can develop itself in less than a week. By this means the output produced in the beginning is not representative for potential of the cell. In the figure can also be seen that the anode, membrane and cathode are joined together. This is beneficial for the direct diffusion of hydrogen ions through the membrane towards

the cathode. The cathode reaction requires oxygen. Therefore, the cathode is directly exposed to the air where oxygen supply is not a limiting factor on the cathode side (in some literature known as an air-cathode).

Design of the cell

Figure 4 illustrates how the fuel cell will be built. The fuel cell will consist out of a Plexiglas (3mm thick) container that will have the dimensions 15×15×15cm. The top side will be perforated with holes for gas exchange. The material for the anode and cathode will be carbon cloth (20% platinum, 0,03mg/cm2 with gas diffusion layer) and the membrane will be a proton exchange membrane (PEM), so only the hydrogen ions will pass through the membrane. After putting the anode, PEM and the cathode on the open side of the container, a sealing frame will hold the components in place. This sealing frame is only a frame, because the cathode has to be exposed to air.

(15)

Experiment

As said earlier, the anode chamber will be filled with young humic topsoil. This was collected from the Utrechtse Heuvelrug and consist mainly of humus from litter of a Douglas fir forest. A glucose solution of 12.500 mg/L (0.07 M glucose) will be added at different times and in different amounts. The exact time and amount is viewed in the table along with some additional information.

TABLE 3: SETUP OF THE EXPERIMENT AND TREATMENT OF THE MFC When (hour) Amount (mL) Amount (gr Glucose) Amount (moles) Where Extra

Week 1 1 25 0,3125 1,66*10e-3 Close to Anode -

60 25 0,3125 1,66*10e-3 Close to Anode -

120 25 0,3125 1,66*10e-3 Close to Anode -

Week 2 160 100 1,25 6,64*10e-3 Close to Anode Rest of soil

moistened

245 100 1,25 6,64*10e-3 Close to Anode Rest of soil

moistened The voltage output of the microbial fuel cell is measured by a model CR10 data logger form Campbell Scientific. The output is set on a range of 0-250 millivolts with a corresponding resolution of 33 microvolts (0.033 millivolts). The current of the output is not measured/calculated due to the absence of a resistance in the electrical circuit.

(16)

Results & Interpretation

In this chapter, the results of the model and experiment will be shown and discussed. Firstly, the data from the model, which is divided in two runs that are set to give an approximation of experiment. Secondly, the data of the experiment are shown. Finally, and comparison between the data of the model and the experiment.

Model

In order to verify the model results, the influencing parameters were adjusted to fit the experimental result. Therefore, the time when glucose was added, in which quantities, the diffusion coefficient (adding more glucose in solution, results in a higher potential diffusion rate) and the ɳloss were set. The other parameters are desired to be equal in both models for

an adequate verification.

First week

In the first week, glucose is added in low amounts. This amount was added at three different times during the first week (with ±60 hour intervals). The parameter’s values can be seen in the table (table 4) and the results in the graph (figure 5).

TABLE 4:SET PARAMETERS FOR FIRST WEEKS RUN

Time of addition t = 0, 60, 120 hours

Amount of addition 1.8e-03 mole glucose spread over 45cm2*

Dglucose 0.15 cm2/hour

ɳloss 0.55 V

* = 45cm2_{is the area where glucose is added; over the whole with of the fuel cell (15cm) and approximately till 3cm near the} anode. This results that the amount is distributed (assumed equally) over an area of 45cm2 _{near the anode.}

(17)

Week 2

In the second week, four times the amount of first week of glucose was added. Different than during the first run; the interval of addition was increased to 80 hour intervals. This was done because the microbes in the MFC had more fuel to digest. Due to the fact that the glucose solution was added in higher quantities, the diffusion coefficient is set higher (more H2O

added, results in a wetter soils and subsequently a higher diffusion rate). The chosen parameter’s values are shown in table 5 and the results in graph 6.

TABLE 5: SET PARAMETERS FOR SECOND WEEKS RUN

Time of addition t = 0 (=160) and 80 (=240) hours

Amount of addition 7.4e-03 mole glucose spread over 45cm2*

Dglucose 0.20 cm2/hour

ɳloss 0.60

* = 45cm2_{is the area where glucose is added; over the whole with of the fuel cell (15cm) and approximately till 3cm near the} anode. This results that the amount is distributed (assumed equally) over an area of 45cm2 _{near the anode.}

FIGURE 6: MODEL OUTPUT OF THE SECOND WEEK

Similar to the first week, adding glucose shows a strong response in the electrical output of the model of the MFC. However, a higher amount of addition causes higher output that is in the range of 30mV till 100mV. In addition to this, the duration of the peaks were longer that in the first run.

(18)

Experimental setup

The experiment on the MFC ran for a total of 14 days. As said earlier, can be divided into two separate runs, with two different treatments on how much glucose is added. The first week, about 150 hours, the microbial fuel cell is supplied with a rather low amount of glucose. The second week, 180 hours, the MFC is supplied with four times as much as in the first run. Before the experiment started, the fuel cell was already set up and running for five days with glucose added. The purpose for this is to acclimate the soil and microbes to the situation. Besides this, as earlier indicated the first days the measurement may not be representative yet due to the absence of a biofilm on the anode. After these five days, the experiment and the measurements started.

The overall range of voltage output of the microbial fuel cell is 25mV till almost 90mV. Compared to more advanced setups in other research on microbial fuel cell, this is a reasonable output. For example, of a MFC in other research with a wastewater microbial fuel results in the range of 50 till 300mV (Lui and Logan, 2004).

First week:

In this graph, the voltage output of the MFC during the first week. At the arrows is de addition of glucose, which is at respectively 0, 60 and 120 hours into the run. It can be seen clearly that when the glucose is added, there is a response resulting in a peak of voltage output. The first peak (0h till ± 40h) has a duration of about 40 hours with an output range of 48-57mV. After this the added glucose was consumed and new glucose had to be added. The second peak (60h till ± 100h) has also a duration of 40 hours, but a lower output range of 43-47mV. The third peak (120h till ± 150h) is a shorter duration of about 30 hours. The voltage output is here even at a lower range 25-38mV.

(19)

Although the duration of the peaks are relative consistent, the range decreases in a negative trend, that can be observed clearly in the graph. There can be two causes for this. Firstly, all the glucose is decomposed which creates a shortage of fuel in the fuel cell and resulting in a lower voltage output. Secondly, the soil in de anode chamber could have dried up too much. This results in a stress on water availability that influences the microbes negatively since bacteria need a film of water to sustain their living (Manzoni et al., 2012). The illustration is a photograph taken from the MFC at the end of this run. The color difference in the upper part and the lower part, suggest that the dehydration of the soil is of influence on the voltage output. However, when observing that most of the decline appears to happen after the peaks, the fuel shortage is likely to affect the output as well.

Second week

In de second run, the behavior on a higher amount is of glucose is tested. Adding more glucose in solution, resolves both issues from the first run. Both the soil stays wetter and more fuel is added. As can be seen in the graphs, the trend is eliminated in this run.

The first peak (160h till 240h) gives a voltage output in the range of 35mV till 64mV and has an approximated duration of 80 hours. The second peak (245h till 330h) has a duration of 85 hours and a voltage range of 43mV till 71mV (highest single peak measured was 88mV).

FIGURE 8: THE SOIL IN THE ANODE CHAMBER AFTER THE FIRST WEEK WITH A CLEAR DIFFERENCE BETWEEN DRIED OUT SOIL AND WET SOIL

(20)

Comparison model vs experiment

In general, there are some resemblances between the model and the experiment. Most can be said at the level of behavior. Both the MFC in the experiment and the model shows the correct behavior that when glucose is added, the voltage output increases. In the side by side comparison in figure 10 of the model and experiment divided over the two weeks.

FIGURE 10:COMPARISONS OF FIRST WEEK VERSUS SECOND WEEK AND EXPERIMENT VERSUS MODEL. NOTE: THAT THE FIRST WEEK OUTPUT IS DETRENDED FOR AN EASIER VISUAL COMPARISON.

In the experimental setup as well as the model adding a higher amount of glucose results in a higher voltage output. Also the duration of the peaks are approximately equal in the model and the experimental setup (±50 and ±80 hours in the respectively first and second week). However, the models are fit arbitrarily by using the ‘loss factor’ (ɳloss). Although there are

resemblances and the output is range of the data from the experiment, the model cannot yet be used as an adequate and precise prediction model. Therefore, more complex processes have to be added. This is further discussed in the discussion.

(21)

Discussion

There are some notes, points of consideration, regarding this research. These will be discussed in this chapter. The discussion is focused on the methodology of both model and experimental setup.

The model is built based on rough simplification of the reality. The processes that are included accounts for mass transportation, rate of decomposition and voltage output potential. The mathematical equation pairing these processes are amongst the most used in the modelling of microbial fuel cells (Ortiz-Martínez et al., 2015). Combining these relations give a resemblance of the voltage output of the fuel cell. However, the two important aspects are absent in this research; current output and voltage output. These could have been rather easy to implement in the model if the amount of moles of electrons that are produces per hour is known. It can be calculated by multiplying the amount of moles of electrons with the Avogadro’s constant – which is the number of electrons per mole. The output of this is the number of electrons produced per hour. Since every electron contains an electrical charge, also called elementary charge, the number of electrons can be multiplied with the charge per electron. This results in a certain coulomb per hour, which can be rewritten in coulomb per second, i.e. ampere. Subsequently the power (Watt) can be calculated by multiplying the voltage output with the ampere output. Power and current output are more interesting than voltage output when the performance of the fuel cell. The calculations were implemented in the model (line 237 – 256), but not given as output. The reason for this will be discussed later on where the experimental setup will be discussed.

Another difficulty is the estimation of parameters. Some parameters had to be assumed using knowledge in other research adjusted to environmental characteristics of the microbial fuel cell (the diffusion coefficients, decomposition rate). Others had to be assumed based on the model output and calibration (loss-factor, half saturation constant). The ‘loss-factor’, or ɳloss,

has the highest uncertainty due to its arbitrary nature and the value factor is determined afterwards. The factor could be further specified to extend the complexity of the model by analyzing the ohmic losses, internal resistance and over potential losses (Zhang, Halme, 1995;

Oliveira et al., 2013b). Beside these possible addition, there are other addition to the model that may result in a more precise predictive capacity.

The experimental setup encountered some challenges and shortcomings, as well. The design of the fuel cell was not the most efficient design. The dimensions of the fuel cell (15×15×15cm) could have been smaller. Maybe an even better option, was to divide the fuel cell into multiple smaller fuel cells. The surface area of the anode, cathode and membrane was 100cm2_{. This}

(22)

easily be calculated using Ohm’s law (I = U/R, where I is current in ampere; U voltage in volts; R resistance in ohm). The power can be calculated the same as explained earlier (P=U*I). Due to the absence of this in the electrical circuit of the experimental setup, this part is also excluded from the model.

Next to the imperfections of the method, the time used for this research is too short. In order to build a model with a higher predictive power, more time has to be invested. Also the validation may not be extensive enough, which means that more experimental data is needed to validate the model. On the contrary, this technology along with the research is still in an ‘young’ stage. More research must be conducted, including trail and error, to gain a higher output, efficiency and insight of the microbial fuel cells.

Conclusion

The aim of this research was to produce and verify a simplified model of a microbial fuel cell that gives an adequate prediction of the output. It can be concluded that the model produces a rough estimation of the output, but not very precise and by this means inadequate. This partly due to the inferiority of both the model and the microbial fuel cell itself.

The behavior of the model that shows similarities in the experimental setup includes:  The increase of voltage output when glucose is added;

 The arch shape in which the voltage output occurs after adding the glucose;  The increased voltage output when a higher amount of glucose is added;

 The longer duration of the voltage output peak when a higher amount of glucose is added.

Although, the rough behavior may be confirmed, the aim to produce a model with a higher predictive power is not truly completed. However, the model can be used as a basis to add more complexity by combining more complex relation in the model. This also means that more additional research in modelling as well as experimenting is necessary in order to gain more insight. Besides this, it is also important to increase the performance of the fuel cell to a point where implementation becomes more interesting.

(23)

Evaluation

The journey of this thesis started off with a research topic that might have been too ambitious. The uncertainties posed a higher risk of ‘failure’ and thereby a chance of disappointment. Especially with the latter, the opposite is true; by researching this topic, I was allowed to choose a subject that interests me and where I am eager for. Off course the results are not as rosy and perfect as I anticipated, but this meanly due to the optimism. In fact, I am very satisfied and pleased with the results I obtained and I enjoyed working on this subject for my Bachelor Thesis. Looking back, I would do a lot of things differently (especially implementing a resistance in the circuit). I have learned a lot and invested lots of time in the research, including things that eventually did not work out like I planned. Yet, I believe this is part of conducting research and it is part of a process in which I have learned a lot.

Finally, I want to express my gratitude to my two supervisors, Dr. Albert Tietema and Prof. dr. ir. Willem Bouten, for giving me the opportunity to follow my own topic and for being willing to have started this adventure. For me this has led to an end product for my bachelor from which I have learned a lot and, in my opinion, can be proud of.

(24)

References

Allen, R.M. and H.P. Bennetto, (1993). Microbial fuel-cells: Electricity production from carbohydrates. Applied Biochem. Biotechnol., 39/40: 27-40.

Du, Z., Li, H., & Gu, T. (2007). A state of the art review on microbial fuel cells: a promising technology for wastewater treatment and bioenergy. Biotechnology advances, 25(5), 464-482.

Eaktasang, N., Kang, C. S., Ryu, S. J., Suma, Y., & Kim, H. S. (2013). Enhanced current production by electroactive biofilm of sulfate-reducing bacteria in the microbial fuel cell. Environmental Engineering Research, 18(4), 277-281.

Gorby, Y. A., Yanina, S., McLean, J. S., Rosso, K. M., Moyles, D., Dohnalkova, A., ... & Culley, D. E. (2006). Electrically conductive bacterial nanowires produced by Shewanella oneidensis strain MR-1 and other microorganisms. Proceedings of the National Academy of Sciences, 103(30), 11358-11363.

Jones J. P., Page J., Plant R. (1980). The Diffusivity of Glucose in Water at 25°C. Retrieved on June 26, 2016 from: http://d.umn.edu/~dlong/Sample%20Lab%20Report.pdf

Kothapalli, A. (2013). Sediment Microbial Fuel Cell as Sustainable Power Resource.

Logan, B. E., Hamelers, B., Rozendal, R., Schröder, U., Keller, J., Freguia, S., ... & Rabaey, K. (2006). Microbial fuel cells: methodology and technology. Environmental science & technology, 40(17), 5181-5192.

Logan, B. E., & Regan, J. M. (2006). Electricity-producing bacterial communities in microbial fuel cells. TRENDS in Microbiology, 14(12), 512-518.

Liu, H., & Logan, B. E. (2004). Electricity generation using an air-cathode single chamber microbial fuel cell in the presence and absence of a proton exchange membrane. Environmental science & technology, 38(14), 4040-4046.

Manzoni, S., Schimel, J. P., & Porporato, A. (2012). Responses of soil microbial communities to water stress: results from a meta‐analysis. Ecology, 93(4), 930-938.

Mink Justine E. (2013). Development of Micro-sized Microbial Fuel Cells as Ultra-Low Power Generators Using Nano-engineered Materials and Sustainable Designs

Nguyen, C. (2003). Rhizodeposition of organic C by plants: mechanisms and controls. Agronomie, 23(5-6), 375-396.

Neumann G. (2011). RHIZODEPOSITION – AN OVERVIEW. 1st international PlantPower Symposium. retrieved on 23 april 2016 from:

http://plantpower.eu/onewebmedia/Abstractboek%20Symposium%20Gent.pdf

Oh, S. T., Kim, J. R., Premier, G. C., Lee, T. H., Kim, C., & Sloan, W. T. (2010). Sustainable wastewaterq treatment: how might microbial fuel cells contribute. Biotechnology advances, 28(6), 871-881.

(25)

Potter, M. C. (1911). Electrical effects accompanying the decomposition of organic compounds. Proceedings of the Royal Society of London. Series B, Containing Papers of a Biological Character, 84(571), 260-276.

Roxby, D. N., Tran, N., & Nguyen, H. T. (2014, August). A simple microbial fuel cell model for improvement of biomedical device powering times. In Engineering in Medicine and Biology Society (EMBC), 2014 36th Annual International Conference of the IEEE (pp. 634-637). IEEE. Smagghe G., Boeckx P., Bossier P., Steurbaut W., van Damme Els JM, Verhoest N. (2011). 1st

international PlantPower Symposium. COMMUNICATIONS IN AGRICULTURAL AND APPLIED BIOLOGICAL SCIENCES, VOL 76/2 1-99

Steffensen J. F. (2012). Partial Pressure Of Oxygen In Water (pO2) ‐ at different temperatures and barometric pressures. retrieved on 10 may 2016 from: http://www.aquaresp.com/ Timmers, R. (2012). Electricity generation by living plants in a plant microbial fuel cell. Wageningen

Universiteit (Wageningen University).

Torres, C. I., Marcus, A. K., Parameswaran, P., & Rittmann, B. E. (2008). Kinetic experiments for evaluating the Nernst− Monod model for anode-respiring bacteria (ARB) in a biofilm anode. Environmental science & technology, 42(17), 6593-6597.

UNFCCC (2015). Adoption of the Paris Agreement. Conference of the Parties. Retrieved on April 21, 2016 from: http://unfccc.int/resource/docs/2015/cop21/eng/l09.pdf

Vancouver Island University (2008). Environmental Standard Reduction Potentials. Retrieved on June 26, 2016 from: https://web.viu.ca/krogh/chem331/Environmental%20Standard %20Reduction%20Potentials%202008.pdf

Zhang, X. C. and A. Halme (1995). "MODELING OF A MICROBIAL FUEL-CELL PROCESS." Biotechnology Letters 17(8): 809-814.

THE MODELLING OF A MICROBIAL FUEL CELL