Bioelectrochemistry 78 (2010) 8–24

Contents lists available at ScienceDirect

Bioelectrochemistry
j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / b i o e l e c h e m

Model based evaluation of the effect of pH and electrode geometry on microbial fuel
cell performance
Cristian Picioreanu a,⁎, Mark C.M. van Loosdrecht a, Thomas P. Curtis b, Keith Scott c
a
Department of Biotechnology, Faculty of Applied Sciences, Delft University of Technology, Julianalaan 67, 2628 BC Delft, The Netherlands
b
School of Civil Engineering and Geosciences, Devonshire Building, University of Newcastle Upon Tyne, Newcastle Upon Tyne, Tyne & Wear NE1 7RU, UK
c
School of Chemical Engineering and Advanced Materials, Merz Court, University of Newcastle Upon Tyne, Newcastle Upon Tyne, Tyne & Wear NE1 7RU, UK

a r t i c l e i n f o a b s t r a c t

Article history: A mathematical model for microbial fuel cells (MFC) which integrates macro-scale time-dependent mass
Received 21 January 2009 balances for solutes and biomass in the anodic liquid with a micro-scale individual-based two-dimensional
Received in revised form 20 April 2009 biofilm model is developed. Computational fluid dynamics and Nernst–Plank mass and charge balances with
Accepted 30 April 2009
diffusion, electromigration, convection and electroneutrality in the biofilm are combined to calculate spatial pH
Available online 8 May 2009
distribution and solutes speciation. Soluble redox mediators are the electron shuttle between microbial cells
Keywords:
and the electrode. The model describes the generally observed variations of pH, solute concentrations and
Microbial fuel cell electrical current produced over time from electroactive biofilms. Numerical simulations also show the effect
Model of bicarbonate buffer and mass transfer through the proton exchange membrane on the microbial population
Biofilm within a mixed anaerobic digestion sludge consortium of methanogenic and electrogenic microorganisms.
Porous electrode In addition, the new modeling approach opens the way to study the influence of fluid flow and any two- or
Redox mediator three-dimensional biofilm and electrode geometry on the MFC output parameters. Hydrodynamic calculations
show that porous bio-electrodes with greater specific surface area do not necessarily produce more current, as
long as convection through the pores is absent. An innovative model solution strategy combines in a very
efficient and flexible way MATLAB, COMSOL finite element and Java codes.
© 2009 Elsevier B.V. All rights reserved.

1. Introduction bio-catalyzed H2 production [9,10], or biosensors [11,12]. We believe

Microbial fuel cells (MFC) are a type of fuel cell that converts the
chemical energy contained in organic matter to electricity using
microorganisms as a biocatalyst. In MFC, bacteria do not transfer their
electrons directly to their characteristic terminal electron acceptor,
but rather to a solid electrode [1,2]. Although MFCs are not a new
discovery [3], the interest in their study and development has been
increasing extraordinarily in recent years. This is mainly because it is
believed that MFCs offer the possibility of directly harvesting
electricity from organic waste and renewable biomass [4]. Potentially,
microbial fuel cells add the diversity of microbial catalytic abilities to
the high-efficiency design of fuel cells, allowing the energy of mixed
organic compounds to be converted into electricity [5]. Driving factors
are not only harvesting small amounts of electricity while polishing
wastewater. Other applications seem even more interesting such as
powering remote sensors [4] or bio-robots [6,7]. More general bio-
electrochemical systems related with the MFCs include applications
such as: conversion of residual glycerol from biodiesel into ethanol [8],
⁎ Corresponding author. Tel.: +31 15 2781166; fax: +31 15 2782355.
E-mail addresses: C.Picioreanu@tudelft.nl (C. Picioreanu),
M.C.M.vanLoosdrecht@tudelft.nl (M.C.M. van Loosdrecht), Tom.Curtis@ncl.ac.uk
(T.P. Curtis), K.Scott@ncl.ac.uk (K. Scott).
that the modeling framework presented here can be extended to
describe also these related bio-electrochemical systems.
A few options exist currently for design of MFC, and these are
outlined in a number of reviews dedicated both to the understanding of
the microbiology of MFC processes [4,13,14] and to advances in the
technological aspects [1,5,15,16]. Although MFCs promise sustainable
energy generation in the future, many major bottlenecks still exist
[17]. Identification of the rate-limiting steps allows the development of
strategies to enhance the MFC output. The performance of a MFC can be
influenced by several physical, chemical and biological factors such as
the rates of [18]: (1) mass transfer of substrates to and into biofilm,
(2) microbial oxidation of substrate, (3) electron transfer from the
microbial cells to the electrode, (4) electric load in the circuit, (5) proton
transfer to the cathode compartment, and (6) oxygen supply and
reduction at the cathode. The mechanism of electron transfer in step
(3) is one of the most disputed areas in current MFC research. First,
direct electron transfer involves either cytochromes (contact conduc-
tion) or proposed conducting structures termed “nano-wires” [19,20].
The second mechanism involves diffusible electron transfer relays
(mediators) that can shuttle electrons between the microbial biocata-
lytic system and the electrode [21]. The biocatalytic process performed
by the microorganisms differs from the natural situation because the
electron flow goes (partly) to the anode instead of to a natural electron

1567-5394/$ – see front matter © 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.bioelechem.2009.04.009
 C. Picioreanu et al. / Bioelectrochemistry 78 (2010) 8–24
acceptor. Fuel cells with externally supplied mediators tend to be
inefficient and expensive [12]. However, it has been demonstrated that
electrochemically active bacteria (EAB) from a biofilm enriched in a MFC
can produce mediator compounds in-situ [22,23]. Therefore, in spite of a
present tendency of the scientific community to focus on the direct
electron transfer, we believe that the study of MFCs with soluble redox
mediators is also very significant. In addition, besides MFCs, other bio-
electrochemical systems using mediators have been reported for which
the present model could be adapted [8].
A suitable method to integrate information gathered from several
disciplines ranging from biology to engineering is by mathematical
modeling. Essentially, the rigor of modeling provides a framework for
testing hypotheses. Most recent MFC studies are experimental, and
focus either on a detailed understanding of the microbiology of
bacteria involved in MFC or on engineering aspects with reports of
incremental increases in MFC performance from MFCs with increas-
ingly advanced designs. A breakthrough that will propel MFC into a
large-scale feasible technology can come only from a multidisciplinary
approach integrating both microbiology and engineering. This is
precisely the bridge computational models can make and it is our aim
to promote this approach through the present study. Only a handful of
modeling studies have been yet dedicated to microbial fuel cells. More
than a decade ago preliminary work attempted to simulate the
currents produced by MFC with suspended cells and an added
mediator [24]. This line of research has not been pursued further until
in 2007 when we introduced a model for the biofilm-based microbial
fuel cells with mediated electron transfer [25]. The model realistically
described many experimental observations, including the influence of
operational parameters on the current/power-voltage characteristics
and the current development in time. A further extension of this
model included methanogenic processes, with preliminary results
presented in [26]. Based on a different hypothesis, Kato-Markus et al.
interestingly explained the functioning of biofilm-based MFCs based
on a hypothetical biofilm electrical conduction property instead of
diffusible mediators [27]. This approach combined in the microbial
growth kinetics limitations by the electron donor substrate and by
electrical potential, introducing a rate equation dubbed “Monod–
Nernst”.
This study's main aim is to extend the MFC modeling framework
presented in [25] in order to include pH calculations. The acidification
of the anodic electrolyte is a primordial limitation hampering the use
of MFCs. Because the electrons are collected by the anode, an equi-
valent amount of protons is released into the solution and must
be eventually transferred away from the anode surface to reach the
cathode, where H+ can be consumed. Important limiting steps in
the proton transport from anode to cathode are the transport in the
biofilm [28] and through the proton exchange membrane (PEM, [29]).
There are at least two effects of anolyte acidification. First, the
reversible electrochemical oxidation of the soluble electron mediator
is negatively influenced by the protons. Second, the activity of
neutrophilic biofilm microorganisms would be decreased if the pH
drops too suddenly. However, on the long term, the second effect may
not be the main problem because acidophilic organisms with similar
activities will be naturally selected. The mathematical model of the
anodic compartment of a MFC introduced in this study is more
general, versatile and rigorous because: (i) in order to calculate pH,
Nernst–Planck fluxes of ions (electromigration and diffusion) together
with an ionic charge balance are introduced instead of molecular
diffusion only; (ii) mass transport by convection and liquid flow over
the irregular biofilm and electrode surface are considered; (iii) any
two- or three-dimensional geometry of the electrode/biofilm can be
accommodated in the model, not only the planar electrode system.
Three case studies are presented here to exemplify the use of the
model for pH calculations, for mixed species electroactive biofilms
with methanogenic and fermentative microorganisms and for flow/
mass/charge transport with biofilm growth in geometrically complex
9
electrodes. Including other electron transfer mechanisms such as by
direct contact or connection to the anode via proposed conducting
structures termed “nano-wires” (e.g., [19]) in the model framework is
possible, and shall be presented in further studies. In this model we
aim to describe in detail only the dynamic behavior of the anodic
compartment, having in mind that a description of the cathode
chamber can follow the same approach. The mediator was assumed in
this study to diffuse freely in the biofilm and bulk liquid. Retention
mechanisms of a produced electrochemical mediator in the biofilm
must be further identified and introduced in an improved model.
2. Model description
The model accounts for two spatial scales: a macroscale for the
bulk liquid anolyte and a microscale for the biofilm on the electrode
and its close environment. These two scales are linked via the fluxes
integrated over the open boundaries of the microscale domain. It is
implicitly assumed that the microscale domain chosen is representa-
tive for the whole biofilm developed on the electrode. A schematic
representation of the model domains, links and fluxes is presented in
Fig. 1.
2.1. Macroscale mass balances in the anodic bulk liquid domain
The bulk liquid in the anodic compartment of a MFC is assumed
completely mixed so that the concentrations CB of all soluble chemical
and microbial species are uniform in the whole volume vB. The system
(1) of nS ordinary differential equations representing mass balances of
each model component i (chemical or microbial) in the bulk liquid (B)
will be solved:
dCi;B Φ
 A A
= Ci;in − Ci;B + ri;B + E Ni;F + M Ni;M ð1Þ
dt vB vB vB
with a set of specified initial conditions Ci,B(t = 0) = Ci,0. Molar units
are used for each component. Batch operation is assumed in the first
two cases of this paper, therefore the rate of exchange with the
exterior (first term in the right hand side of Eq. (1)) is zero. ri,B is
the net rate of reaction in the bulk for a component i, and it is made up
of contributions from rates of all individual reactions multiplied
by the corresponding stoichiometric coefficients. The third term in
Eq. (1) is the rate of mass transfer with the biofilm on the electrode
with area AE. Finally, the last term is the rate of mass transfer through
the membrane (area AM) that separates the anodic and cathodic
compartments. Fluxes Ni,F and Ni,M are given by Eqs. (12) and (13),
respectively. Although Eq. (1) is general and can be applied to solutes
and biomass as well, we choose for the illustration cases here not to
account for the biomass in the bulk liquid. The effect of the biomass
concentration in the bulk liquid on the MFC performance was
analyzed in [25], and found that suspended microorganisms con-
tribute much less to the current production than the biofilm cells. The
solution of Eq. (1) is a set of bulk liquid concentrations Ci,B needed at
each moment in time as the boundary condition for solute mass
balances in the biofilm domain (see Fig. 1).
2.2. Microscale mass, charge and momentum balances in the biofilm
domain
The biofilm domain is characterized by concentration gradients
both for solutes and for biomass components (Fig. 1). We present here
for illustration only two-dimensional model simulations because of
the good balance between calculation time and model realism, but
one-dimensional reduction or three-dimensional extensions are
perfectly possible (as shown in [25,26] for MFC diffusion/reaction
systems only). The concentrations Ci,F (t, x, y) are also time dependent
as the biofilm structure continuously changes.
10 C. Picioreanu et al. / Bioelectrochemistry 78 (2010) 8–24

Fig. 1. Model scales and computational domains. In the bulk liquid domain (top scheme), time-dependent mass balances are solved for each soluble component i with concentration
Ci,B present in the anolyte volume vB. Fluxes exchanged with the exterior (Φ), with the biofilm (Ni,F) on the electrode with area AE, with the cathode via the membrane (Ni,M) with
area AM, and reaction rates in the bulk liquid (ri,B) are considered. In the biofilm domain, mass, charge and liquid flow balance equations are solved both in the biofilm solid matrix
and in the mass/hydrodynamic boundary layer. Two-dimensional distributions of concentrations Ci,F, potential V and liquid velocity u are calculated both for the biofilm on a planar
electrode (left scheme) and on the porous electrode (right scheme). Specific boundary conditions are applied on the biofilm domain boundaries at inlet ΓI, outlet ΓO, electrode ΓE, bulk
ΓB, and biofilm surface ΓF. At the smallest scale (bottom scheme), a set of biological and acid–base volume-based reactions (ri,F) are coupled to a set of electrochemical surface-based
reactions (ri,E).

2.2.1. Solute components
For any soluble component i, a mass balance is set up by assuming
transport by diffusion Ni,D = −Di∇Ci,F, convection Ni,C = uCi,F and
electromigration Ni,E = − zium,iFCi,F∇V fluxes (the gradient operator is
∇ = i∂/∂x + j∂/∂y, with i and j the unit vectors in the x and y
directions). Soluble components can be produced or consumed in
several biotic or abiotic transformation processes with net rate ri,F.
Therefore, the nS mass balances written for each soluble component i
will form the following system of partial differential equations written
in molar units:
biofilm matrix subdomain) and partially with liquid (the mass
transfer boundary layer sub-domain, see Fig. 1). Particular forms of
Eq. (2) will apply for different species and sub-domains. For example,
neutral species (charge zi = 0) will not have electromigration, no
convection will occur in the biofilm matrix (liquid velocity u = 0), and,
while acid–base equilibria apply everywhere, no microbial reactions
will occur in the liquid sub-domain.
The field of liquid velocity u results from the Navier–Stokes
equations of momentum balance and liquid continuity for incom-
pressible liquid in laminar flow regime:

ACi;F

2
= Di j Ci;F + zi um;i Fj  Ci;F jV − u  jCi;F + ri;F ð2Þ Au 1
At 2
= −ðu  jÞu − jp + mj u
At ρ ð4Þ
by assuming constant diffusion coefficients Di and ion mobilities ju = 0
um,i = Di/RT, and incompressible fluid (see [30] for details on the
derivation, and [31] for an application of a similar approach to model
microbially influenced corrosion). The present model does not account For a complete model setup, the system of Eqs. (2)–(4) needs
for potential gradients between anode and cathode, which could lead to boundary conditions. At biofilm domain/bulk liquid interface ΓB
an electrokinetic flow of ions. However, the ions will move in the po- (e.g., y = LY for planar electrode) bulk concentrations of all soluble
tential gradient ∇V developed by imposing an electroneutrality con- components are assumed, while the potential V is set to zero, as the
dition everywhere in the biofilm domain: constant reference value in the bulk liquid. Symmetry boundary is
assumed for the flow on ΓB (Eq. (5)):
X
zi Ci;F = 0 ð3Þ
i
8
< Ci;F = Ci;B ðt Þ
Eqs. (2) and (3) are applied on the microscale rectangular com- V=0 on CB ð5Þ
:
putational domain partially filled with attached microbial cells (the uY = 0; AuX = Ay = 0
 C. Picioreanu et al. / Bioelectrochemistry 78 (2010) 8–24 11

The electrode surface on which the biofilm develops ΓE (at y = 0 The liquid inlet has a fully developed laminar velocity profile with
for the planar electrode, but on boundaries with other orientations an average velocity uin. The inlet concentrations are those from the
for the porous electrode, see Fig. 1) is electrochemically active for bulk liquid and electrical insulation applies:
certain chemical species, e.g. for mediators and protons. The surface- 8
based rate or reaction ri,E equals the total flux normal on the surface. > Ci;F = Ci;B ðt Þ
>
>
For the electrochemically inactive chemical species, zero-flux con-
>
>
> X  AC AV

<F zi Di i + zi um;i FCi =0 on CI
dition applies (insulation because ri,E = 0). A current inflow condi- Ax Ax ð7Þ
>
>
i
 
tion applies for the charge balance, with the inward current density >
>
>
> 3 y y
ii given by a Butler–Volmer expression (see Table 1). The electrode : uY = 0; uX = uin 2−
2 LY LY
is a no-slip wall for the flow:

In the liquid outlet there is a zero pressure (arbitrary reference value),

8
 convective flux only for the chemical species and electrical insulation:
>
> n  Di jCi + zi um;i FCi jV = ri;E
>
< 8
X

AC AV
ni = nF zi Di jCi + zi um;i FCi jV = ii ð6Þ >
>
> on CE > Di i + zi um;i FCi
> =0
>
> >
> Ax Ax
: <
X  AC 
i
uX = 0; uY = 0 i AV ð8Þ
> F z D + z u FC = 0 on CO
>
>
i i
Ax i m;i i
Ax
>
> i
>
:
p=0
In the simpler case of a planar electrode surface, the flux and
X  conditions become: D i∂Ci / ∂y + zium,i FC i∂V / ∂y = ri,E and
current The boundary layer/biofilm matrix interface ΓF is an internal
F zi Di ACi = Ay + zi um;i FCi AV = Ay = ii . boundary with flux and current continuity for the mass and charge
i

Table 1
Molar stoichiometry and reaction rates for the chemical, microbial and electrochemical conversions included in model. All reactions are used in Case 2, whereas only reactions a1–a5,
b1 and e1–e3 are used in Case 1 and Case 3.

Reaction name Reaction equationa Reaction rate

Acid–base equilibria

a1. Water dissociation H2 O±OH− + Hþ ra;HOH = kHOH 1 − CKHHOH
COH


a2. Carbonic acid dissociation CO2 + H2 O±HCO−
3 + H
þ
ra;CO2 = kCO2 CCO2 − CHKCCO2HCO3


a3. Acetic acid dissociation AcH±Ac− + Hþ ra;AcH = kAcH CAc − CHKAcH
CAcH


a4. Reduced thionine first dissociation MHþ
3 ±MH2 + H
þ
ra;MH3 = kMH3 CMH3 − CHKMH3 CMH2


þ þ
a5. Reduced thionine second dissociation MH2+
4 ±MH3 + H ra;MH4 = kMH4 CMH4 − CHKMH4 CMH3


a6. Propionic acid dissociation PrH±Pr− + Hþ ra;PrH = kPrH CPr − CHKPrH
CPrH


a7. Butyric acid dissociation BuH±Bu− + Hþ ra;BuH = kBuH CBu − CHKBuH
CBuH

Biological conversions
b1. Acetate with oxidized mediator 2:163 Ac − + 6:553 MHþ + 0:2 NHþ
4 + 6:153 H2 O rb;AcM = kAcM CXAc CAc;t
CAc;t CMH
+ KAc;AcM CMH + KMH;AcM
−
→ 1 XAc + 6:553 MHþ
3 + 3:326 HCO3 + 1:363 H
þ

b2. Glucose fermentation 1:755 Gl + 0:2 NHþ

4 + 3:212 H2 O rb;GlFer = kGlFer CXGl CGl CGl
+ KGl;GlFer
→ 1 XGl + 0:252 Bu − + 0:728 Pr− + 1:932 Ac−
+ 3:612 H2 + 2:473 HCO3− + 5:586 Hþ
−
b3. Butyrate fermentation 3:51 Bu− + 0:2 NHþ
4 + 0:158 HCO3 + 6:199 H2 O rb;BuFer ¼kBuFer CXBu CBu;t
CBu;t KH2;BuFer
+ KBu;BuFer CH2 + KH2;BuFer
→ 1 XBu + 6:599 Ac − + 6:599 H2 + 3:131 Hþ

b4. Propionate fermentation 7:522 Pr − + 0:2 NHþ

4 + 21:336 H2 O rb;PrFer ¼kPrFer CXPr CE;t
CPr;t KH2;PrFer
+ KPr;PrFer CH2 + KH2;PrFer
→ 1 XPr + 7:203 Ac − + 21:737 H2 + 7:16 HCO−
3 + 7:041 H
þ

b5. Acetotrophic methanogenesis 10:531 Ac− + 0:273 Hþ + 0:2 NHþ

4 + 9:604 H2 O rb;AcFer = kAcFer CXAc CAc;t
CAc;t
+ KAc;AcFer
→ 1 XAc + 10:004 CH4 + 10:058 HCO3−

b6. Hydrogenotrophic methanogenesis 35:104 H2 + 9:249 HCO3− + 0:2 NHþ

4 + 9:049 H
þ
rb;H2Met ¼kH2Met CXH2 CH2 CH2
+ KH2;H2Met
→ 1 XH2 + 8:249 CH4 + 27:254 H2 O
Electrochemical conversionsb

−1
−3
þ
e1. Oxidation double protonated mediator 1 2+
2 MH4 ⇄ 2 MH
1
+ 3
2 Hþ + e − re;1 = i1 = F; i1 = i0;r CMH4
CMH4;r
CMH
CMH;r
CH
CH;r f ðEM ; IÞ


−1
−2
þ þ
e2. Oxidation single protonated mediator 2 MH3 ⇄ 2 MH
1 1
+ Hþ + e − re;2 = i2 = F; i2 = i0;r CMH3
CMH3;r
CMH
CMH;r
CH
CH;r f ðEM ; I Þ


−1
−1
þ 1 þ
e3. Oxidation neutral mediator 2 MH2 ⇄ 2 MH
1 1
+ 2H + e− re;3 = i3 = F; i3 = i0;r CMH2
CMH2;r
CMH
CMH;r
CH
CH;r f ð EM ; I Þ

a − − − − − −
Abbreviations used for chemical and microbial species names: Ac ≡ CH3COO , AcH ≡ CH3COOH, Pr ≡ C2H5COO , PrH ≡ C2H5COOH, Bu ≡ C3H7COO , BuH ≡ C3H7COOH,
Gl ≡ C6H12O6, MH2+ + +
4 , MH3 , MH2 ≡ protonated and neutral forms of the reduced thionine mediator, MH ≡ protonated form of the oxidized thionine mediator. XAc ≡ acetate consumer
both electrochemically and by methanogenic fermentation, XGl ≡ glucose consumer, XBu ≡
C4(butyrate) consumer, XPr ≡
C3(propionate) consumer,  XH2 ≡ H2 consumer.
b
The potential and current dependency of the electrochemical rates is: f ðEM ; IÞ = exp 2:303 ðVC − Re I − EM Þ − exp − 2:303 ðVC − Re I − EM Þ with the equilibrium potential EM


 bM
 b M
0 C CH2 0 C CH KMH3 0 CMH CH3
calculated as: EM = EM + 0:0295 lg CMH
MH3
= EM + 0:0295 lg MHCMH2 = EM + 0:0295 lg CMH4 KMH4 .
12 C. Picioreanu et al. / Bioelectrochemistry 78 (2010) 8–24
balances, but with zero velocity (no-slip wall) for the flow equations.
In the Case 2 with porous electrode (see Fig. 1), on the external
boundaries ΓIE, ΓOE and ΓWE no-flux (insulation) conditions are applied
for the mass, charge and momentum transfer, when the flow is only
over the electrode. When the flow is through the electrode, ΓIE is inlet
like ΓI and ΓOE is outlet like ΓO.
For the initial state at t = 0, a uniform distribution of concentra-
tions Ci,F = Ci,0 and zero potential V throughout the whole domain
are assumed. The initial concentrations for the dissociated species
are calculated from total concentrations and dissociation (acidity)
constants (see footnotes of Table 2).
2.2.2. Biomass balances
The model for biomass production/consumption and transport
within the biofilm matrix closely follows the individual-based
modeling approach (IbM) described and applied in [32,33]. Multiple
microbial species (metabolic groups) can be easily considered with
this approach. The relevant parameters are: nP0, initial number of
biomass particles, ρX, density of biomass particles, m0, initial mass
of biomass particles and mX,max, critical biomass for biomass particle
division. The biomass particles grow with the rates expressed in
Table 1, as a function of solute concentrations at the position of the
spherical particle center. Then, the biomass particles divide and a
hard collisions shoving algorithm are applied to avoid particle
overlapping [34]. The position of the moving boundary ΓF is the
result of this algorithm.
2.3. Model components and reactions
A choice of the set of chemical and biological model components
(“species”) together with the set of chemical, biological and elec-
trochemical reactions (conversions) must be made. The microbial
metabolism needs a reduced substrate (electron donor) to be
oxidized with an electron donor in order to gain the energy needed
by the cells for growth and maintenance. A series of metabolic
products can be formed during the biological conversions per-
formed by the cells. The present model is based on the mediated
electron transfer between microbial cells and electrode. In this
simple model it is therefore assumed that the organic substrate is
converted by the oxidized mediator into more oxidized products,
while the mediator is reduced. The reduced mediator diffuses to the
anode where it is electrochemically oxidized and can be re-used by
the cells (Fig. 1).
2.3.1. Electron mediator
Given this simple mechanism for electron transfer, a diffusible
mediator species must be chosen. There are exogenous mediators
naturally present or just added in the medium and there are also
endogenous mediators produced by the microbes. Among the many
choices that can be made, we opted here (like in [25]) for thionine as
an exogenous mediator because this is a well known electron shuttle,
studied and used before in MFCs, and for which kinetic and
thermodynamic data exist. Most of the rate parameters listed in
Table 2 were obtained by Bennetto's group on MFCs with added
thionine and suspended Proteus cells ([21,35–37]).
Thionine exists in two forms: an oxidized purple dye and reduced
colorless one, called also leuco-thionine. The reduced leuco-thionine
can have several protonated states present around the neutral
and slightly acidic pH, which must be considered in this study that
focuses on pH effects on the MFC performance. These states are
denoted: MH2, MH+ 2+
3 and MH4 . The two acid–base equilibria with
pKa 4.4 and 5.3 ([38,39]) are shown in Table 1 (reactions a4 and a5).
The oxidized thionine exists in a single protonated form MH+ bet-
ween pH 2 and 10 (pKa ≈ 11, from [39]). It is assumed that at the
anode surface all forms of the reduced mediator can be electro-
chemically oxidized (reactions e1–e3 in Table 1), with production of
protons. The reaction rates are of Butler–Volmer reversible type,
dependent on the concentrations of participant species at the elec-
trode surface and on the anodic activation overpotential ηa (see
footnotes of Table 1). The activation overpotential depends on the
ohmic losses of potential (due to internal and external electrical
resistance Re and current I), on the cathodic potential VC and equi-
librium redox potential of the mediator oxidation reaction, EM. The
derivation of ηa was presented in [25].
2.3.2. Substrates and products
In this study we investigated two reaction systems. In the first, one
simple substrate — acetate — is converted by a single microbial species
(XAc) into CO2. In this conversion the oxidized mediator is the electron
acceptor. The reaction stoichiometry (b1 in Table 1) was derived in
[25] based on elemental and charge balances and thermodynamic
considerations [40].
The second case study presents a more complex system with a
mixed microbial community of electroactive (XAc) organisms compet-
ing with fermentative organisms (see Table 1, reactions b2–b6). The
scheme of reactions and the stoichiometric and rate coefficients
follow closely the IWA Anaerobic Digestion Model described in [41]. In
conjunction with biofilm modeling, the kinetic model from [41] was
already applied to methanogenic granular sludge in [42,43] and in a
preliminary MFC study [26]. The initial substrate is glucose (Gl). This is
converted into several low-chain fatty acids (butyrate, Bu, propionate,
Pr, and acetate Ac) together with hydrogen and CO2 production by the
glucose fermentative microorganisms XGl (reaction b2). This acid-
ification step is followed by acetogenesis (reactions b3 and b4)
performed by microorganisms XBu and XPr, which produce acetate and
H2 from the C4 and C3 organic acids. These reactions are thermo-
dynamically inhibited by the H2. In the last step, methanogenesis,
acetate is transformed into methane and CO2 (XAc in reaction b5),
while H2 and CO2 are converted into methane by hydrogenotrophs
(XH2 in reaction b6).
The pH changes due to all these biological conversions. For pH
calculations, together with the water dissociation (reaction a1),
several acid–base equilibria must be taken into account for the
carbonic, acetic, propionic and butyric acids (reactions a2, a3, a6 and
a7 in Table 1). To maintain electroneutrality and to directly change
the ionic conductivity of the solution, inert cations (Na+) and anions
(Cl−) can also be added.
2.3.3. Net reaction rates and fluxes
Having the system of reactions defined as in Table 1, the net
rates for each chemical or microbial component at each model
scale, in each sub-domain and on each reactive boundary can be
calculated by the algebraic sum of individual reaction rates rj
multiplied by the stoichiometric coefficient ni,j of species i in
reaction j. In the bulk liquid there are only the acid–base equilibria,
needed in Eq. (1):
X
ri;B = ra;j mi;j ð9Þ
j
In the biofilm domain there are in addition the biological
conversions, needed in Eq. (2), as well as for the biomass growth in
the biofilm biomass balances:
X X
ri;F = ra;j mi;j + rb;j mi;j ð10Þ
j j
On the electrode surface ΓE the net rates used in Eq. (6) are:
X
ri;E = re;j mi;j ð11Þ
j
 C. Picioreanu et al. / Bioelectrochemistry 78 (2010) 8–24 13

Table 2
Parameters used in the microbial fuel cell model.

Parameter Description Value Units Source

Case 1, Case 3 Case 2
Dissolved components
CAc,t,0 Initial concentration total acetate 1.56 0.01 mmol L− 1 [36]
CC,t,0 Initial concentration total carbonate 2, 20 and 100 2 and 20 mmol L− 1 Chosen
CMH2,t,0 Initial concentration total reduced mediator 0.001 0.001 mmol L− 1 Chosen
CBu,t,0 Initial concentration total butyrate – 0.01 mmol L− 1 Chosen
CPr,t,0 Initial concentration total propionate – 0.01 mmol L− 1 Chosen
CMH,0 Initial concentration oxidized mediator 1 1 mmol L− 1 [36]
CH,0 Initial concentration protons 10− 7 10− 7 mol L− 1 pH 7
CGl,0 Initial concentration glucose – 0.8 mmol L− 1 Chosen
CH2,0 Initial concentration H2 – 10− 6 mmol L− 1 Chosen
CCH4,0 Initial concentration CH4 – 0.01 mmol L− 1 Chosen
CCl,0 Initial concentration Cl− 10− 5 10− 5 mmol L− 1 Chosen
Ci,0 Initial concentration of dissociated/protonated Equilibrium valuesa mmol L− 1 Calculateda
species (i = Ac, AcH, Pr, PrH, Bu, BuH, OH,
MH2, MH3, MH4)
DMH, DMH2, DMH3, DMH4 Diffusion coefficient mediator, all forms 2 × 10− 10 m2 s− 1 [25]
DAc Diffusion coefficient acetate 1.1 × 10− 9 m2 s− 1 [30,44]
DAcH Diffusion coefficient acetic acid 1.3 × 10− 9 m2 s− 1 [30,44]
DCO2 Diffusion coefficient CO2 1.9 × 10− 9 m2 s− 1 [30,44]
DHCO3 Diffusion coefficient HCO3− 1.2 × 10− 9 m2 s− 1 [30,44]
DH Diffusion coefficient H+ 9.3 × 10− 9 m2 s− 1 [30,44]
DOH Diffusion coefficient OH− 5.3 × 10− 9 m2 s− 1 [30,44]
DNa Diffusion coefficient Na+Na+ 1.3 × 10− 9 m2 s− 1 [30,44]
DCl Diffusion coefficient Cl− 2 × 10− 9 m2 s− 1 [30,44]
DGl Diffusion coefficient glucose – 0.5 × 10− 9 m2 s− 1 [44]
DBu, DBuH Diffusion coefficients butyrate and butyric acid – 0.7 × 10− 9 m2 s− 1 [44]
DPr, DPrH Diffusion coefficients propionate and propionic acid – 0.9 × 10− 9 m2 s− 1 [44]
DH2 Diffusion coefficient H2 – 5 × 10− 9 m2 s− 1 [44]
DCH4 Diffusion coefficient CH4 – 1.5 × 10− 9 m2 s− 1 [44]
R Universal gas constant 8.314 J mol− 1 K− 1
T Temperature 300 K
F Faraday's constant 96,480 C mol− 1 e−
km,H Mass transfer coefficient for protons across 10− 2 and 10− 6 m s− 1 Chosen
the PEM
KHOH Water dissociation equilibrium constant 10− 14 mol2 L− 2 [41]
KAcH Acetic acid dissociation equilibrium constant 10− 4.756 mol L− 1 [41]
KMH3 Equilibrium constant for dissociation of 10− 5.3 mol L− 1 [38,39]
thionine MH3+
KMH4 Equilibrium constant for dissociation of 10− 4.376 mol L− 1 [38,39]
thionine MH42+
− 6.6 −1
KCO2 Carbonic acid dissociation equilibrium constant 10 mol L [41]
KPrH Propionic acid dissociation equilibrium constant – 10− 4.874 mol L− 1 [41]
KBuH Butyric acid dissociation equilibrium constant – 10− 4.819 mol L− 1 [41]
ki Rate constants for the dissociation equilibria 107 s− 1 Large values
(i = HOH, CO2, AcH, PrH, BuH, MH3, MH4) 107 (kHOH) mol m− 3 s− 1

Biomass components
nP0 Initial number of biomass particles 2, 25 15 each type Particles Chosen
ρX Density of biomass particlesb 8900 Cmol m− 3 [25]
m0 Initial mass of biomass particles (4.5 ± 1.2) × 10− 14 Cmol [25]
mX,max Critical biomass for biomass particle division 6 × 10− 14 Cmol [25]

System dimensions
vB Bulk liquid volume 2.5 × 10− 4 m3 [36]
AF Anode surface area 10− 3 m2 [36]
LX × LY Size of biofilm computational domain (1) 300 × 100, (3) 300 × 400 (2) 300 × 100 μm × μm Chosen
and 2300 × 100

Electrical parameters
VC Cathode potential 0.68 V (SHE) [36] Ferricyanide
Re Total MFC resistance 100 Ω Chosen

Electrochemical rate parameters

i0,r Exchange current density for mediator oxidation 2 × 10− 4 A m− 2 [25] based on [36]
in reference conditions (CMH4,r = CMH3,r = CMH2,r =
CMH,r = 1mM, CH,r = 10− 7 M)
EM0 Standard reduction potential for the mediator 0.477 V (SHE) [21] at pH 0
couple (vs. SHE)
bM Tafel coefficient for mediator oxidation 0.120 V For 2 e−

Microbial rate parameters

kAcM Rate constant for XAc growth on acetate and 2.8 × 10− 5 s− 1 [25] from [21]
mediator
−3
KAc,AcM Monod half-saturation coefficient for acetate in 1.56 mol Ac,t m [25]
growth with mediator

(continued on next page)

14 C. Picioreanu et al. / Bioelectrochemistry 78 (2010) 8–24

Table 2 (continued)
Parameter Description Value Units Source
Case 1, Case 3 Case 2
Microbial rate parameters
KMH,AcM Monod half-saturation coefficient for 0.1 mol m− 3 [25]
oxidized mediator
−6 −1
kAcFer Rate constant for XAc growth on acetate – 5.8 × 10 s [41] at 300K
fermentation
KAc,AcFer Monod half-saturation coefficient for acetate – 1.56 mol Ac,t m− 3 [41]
in XAc growth
with Ac fermentation
kH2Met Rate constant for XH2 growth on H2 – 1.4 × 10− 5 s− 1 [41]
KH2,H2Met Monod half-saturation coefficient for H2 – 8 × 10− 4 mol H2 m− 3 [41]
in XH2 growth
kPrFer Rate constant for XPr growth on propionate – 6 × 10− 6 s− 1 [41]
fermentation
KPr,PrFer Monod half-saturation coefficient for propionate – 0.9 mol Pr,t m− 3 [41]
in XPr growth
KH2,PrFer Inhibition coefficient for H2 in XPr growth – 2.2 × 10− 4 mol H2 m− 3 [41]
kBuFer Rate constant for XBu growth on butyrate – 1.4 × 10− 5 s− 1 [41]
fermentation
KBu,BuFer Monod half-saturation coefficient for butyrate – 1.25 mol Bu,t m− 3 [41]
in XBu growth
KH2,BuFer Inhibition coefficient for H2 in XBu growth – 6.2 × 10− 4 mol H2 m− 3 [41]
kGlFer Rate constant for XGl growth on glucose – 3.4 × 10− 5 s− 1 [41]
fermentation
KGl,GlFer Monod half-saturation coefficient for glucose – 2.6 mol Gl m− 3 [41]
in XGl growth

Hydrodynamic parameters
ρ Water density 1000 kg m− 3
ν Water kinematic viscosity 10− 6 m2 s− 1
uin Water mean inlet velocity 0.001 m s− 1 Chosen
a
The initial equilibrium concentrations for the dissociated and protonated chemical species were calculated as: COH,0 = KHOH / CH,0; CHCO3,0 = CC,t,0KCO2 / (KCO2 + CH,0);
CCO2,0 = CC,t,0CH,0 / (KCO2 + CH,0); CC,t,0 = CCO2,0 + CHCO3,0; CAcH,0 = CAc,t,0CH,0 / (KAcH + CH,0); CAc,0 = CAc,t,0KAcH / (KAcH + CH,0); CAc,t,0 = CAcH,0 + CAc,0; CPrH,0 = CPr,t,0CH,0 / (KPrH +CH,0);
CPr,0 =CPr,t,0KPrH / (KPrH +CH,0); CPr,t,0 =CPrH,0 +CPr,0; CBuH,0 =CBu,t,0CH,0 /(KBuH +CH,0); CBu,0 =CBu,t,0KBuH / (KBuH + CH,0); CBu,t,0 =CBuH,0 +CBu,0; CMH2,0 =CMH2,t,0KMH3KMH4 / (KMH3KMH4 +
CH,0KMH4 +CH,02); CMH3,0 =CMH2,t,0CH,0KMH4 / (KMH3KMH4 +CH,0KMH4 +CH,02); CMH4,0 =CMH2,t,0CH,02 / (KMH3KMH4 + CH,0KMH4+CH,02).
b
One C-mol biomass with the elemental formula CH1.8O0.5N0.2 corresponds to 24.6 g dry weight biomass or 33.7 g COD biomass.

The total flux exchanged between the small-scale biofilm domain element methods for solving the PDEs, ODEs and coupling integral
and the bulk liquid in Eq. (1) is calculated by integrating total fluxes condition, and own Java code for the individual-based biofilm model.
over the inlet, outlet and bulk boundaries: The main program code is a MATLAB script, which in a first section
defines the:
Z

1
Ni;F = Ni;D + Ni;E + Ni;C dC ð12Þ (a) model input parameters — from Table 2;
LX
CI [CO [CB (b) model geometry (the rectangular biofilm domain and the anode
geometry);
The flux of protons through the membrane between anode and (c) initial biomass distribution on the anode surface;
cathode solutions is expressed as function of the mass transfer (d) two-dimensional COMSOL application modes (1: momentum
coefficient km,i and assuming a constant pH in the cathodic solution transfer — incompressible Navier–Stokes Eq. (4); 2: mass
(a set-point concentration Ci,C): transfer — Nernst–Planck with diffusion, convection, electro-
migration and electroneutrality condition Eqs. (2) and (3)),

 with corresponding boundary conditions Eqs. (5)–(8) and
Ni;M = km;i Ci;B − Ci;C ð13Þ equilibrium initial conditions;
(e) integral-algebraic Eq. (14) for the total current;
Because the anodic activation overpotential is a function of the (f) ODE system of mass balances in bulk liquid Eq. (1);
total current I (see Table 1), an implicit integral equation must be (g) reaction rates in the bulk liquid (9), in the biofilm domain (10)
solved simultaneously with the Eqs. (2) and (3) (see also the model and on the electrode surface (11), with expressions from Table 1;
description and solution algorithm from [25]): (h) integral coupling condition (12) for the biofilm/bulk sub-
models; and
Z X
 (i) 2d mesh for finite element solution of the hydrodynamics and
I= ij Ej ; I ds ð14Þ mass balances.
CE j

The second section of the script solves the model equations in a

time loop (time step Δt = 1 hour). The finite element mesh size was
2.4. Model solution 2 μm. At any time t there are successively solved:

The model was implemented in a combination of MATLAB code (j) biofilm domain hydrodynamics at steady state to get u and p
(MATLAB 2007b, MathWorks, Natick, MA) as the main algorithm script, (with COMSOL finite element methods), for a given geometry
COMSOL Multiphysics (COMSOL 3.5, Comsol Inc., Burlington, MA) finite of the biofilm matrix and electrode;
 C. Picioreanu et al. / Bioelectrochemistry 78 (2010) 8–24 15

(k) biofilm domain mass balances at steady state to get Ci,F (with
COMSOL finite element methods) by using the calculated
velocity field u, given 2d biomass distribution CX,F and given
bulk liquid concentrations Ci,B. All the model state variables at
this time t, i.e., u(t,x,y), p(t,x,y), Ci,F(t,x,y), CX,F(t,x,y), Ci,B(t) are
saved after this step.
(l) time evolution of bulk liquid concentrations Ci,B from t to t + Dt
using the coupling fluxes with biofilm domain;
(m) biomass balances in the biofilm to get CX,F from the pre-
viously calculated Ci,F, using own Java algorithms as described
in [32–34].
Finally, the biofilm matrix distribution CX,F(t,x,y) is updated from
the particle-based biomass model. With the new biofilm geometry
and biomass distribution so obtained a new time step starts with the
hydrodynamic calculations (j).
3. Model results and discussion
Three case studies are presented here to exemplify the use of the
model: (1) for pH calculations, (2) for multiple species electroactive
biofilms mixed with methanogenic and fermentative microorganisms
and (3) for mass/charge transport in electrodes with complex

Fig. 2. Time progress of current production and of pH, acetate and mediator concentrations in the bulk and biofilm biomass calculated in Case 1. (A) pH; (B) current; (C) mediator
species; (G) acetate and biofilm biomass development in time for very slow proton transfer rates through PEM (km,H = 10− 6 m/s) and at different carbonate buffer concentrations. 1
C-mol biomass = 24.6 g biomass. (D) pH; (E) current; (F) mediator species concentration in time for faster proton transfer rates through PEM (km,H = 10− 2 m/s) and at different
carbonate buffer concentrations (2, 20 and 100 mM).
16 C. Picioreanu et al. / Bioelectrochemistry 78 (2010) 8–24

geometry (e.g., porous electrodes). Table 2 lists all parameters used in cathode proton transfer rate on the functioning of MFCs. Moreover,
the simulations. the model also points to the crucial role of this transfer process in
the MFC.
3.1. Case 1 — mono-species electroactive biofilm on planar electrode

This is a relatively simplified model system with one microbial

species (XAc) that consumes acetate and oxidized mediator as
substrates for growth. We studied the effect of anolyte buffering
with CO2/HCO3− in different concentrations (CC,t = CCO2 + CHCO3) and
of different mass transfer rates of protons through the PEM (km,H) on
the anolyte pH and on the current obtained with the MFC. Spatially
two-dimensional mass balances in the biofilm domain and time-
dependent balances in the ideally mixed bulk anolyte liquid were
solved for the following chemical species: H+, HO−, MH+, MH2, MH3+,
MH42+, AcH, Ac−, CO2, HCO− + −
3 , Na and Cl (see footnote (a) of Table 1
for abbreviations).
When the rate of proton mass transfer through the PEM is
negligible (km,H = 10− 6 m/s), the anode compartment is practically a
closed system, in which H+ will accumulate. The consequence is that
pH decreases during the microbial conversion of acetate and will
then remain constant when the substrate is depleted (Fig. 2A). The
more CO2/HCO3− in the solution (from 2 mM to 100 mM) the closer
the pH remains to the initial value of 7. How much bicarbonate in the
solution is needed to keep the pH constant depends on the initial
concentration of acetate. In an ideal electrochemical conversion of
acetate into CO2, 1 mol acetate releases 8 mol H+ into the solution.
With the microbial/electrochemical system, where acetate is also
consumed for microbial growth, about 7 mol H+ will be released
(summed eq. b1 and e2 in Table 1). We used 1.56 mM acetate (which
means 100 g COD/m3, with COD = chemical oxygen demand, a
measure of reduction equivalents contained in a substrate frequently
used in environmental technology), for which it is clear that 2 mM
bicarbonate buffer is insufficient. Consequently, in that case the
medium acidifies in such extent (pH b 4.5 after 7 days) that the
electrochemical oxidation of the mediator — assumed here as a
reversible process which produces H+ — is practically blocked. The
result of poor buffering is lower current production (Fig. 2B) and
incomplete acetate conversion (Fig. 2G). Obviously, both these
effects are negative for the use of MFC technology in wastewater
treatment. At pH ~ 5 the reduced thionine exists in three forms, as
shown in Fig. 2C. Although this justifies the need for considering all
thionine dissociation equilibria, we did not account for the
possibility that only some of these species may be electrochemically
active (all three species react at the electrode cf. eq. e1–e3 in Table 1).
Another known effect not investigated here is the dependency of
microbial metabolism on pH.
In the presence of significant proton transfer to the cathode
(km,H = 10− 2 m/s), the pH remains above 6 during the acetate
conversion even for low bicarbonate content of the feed solution
(Fig. 2D). Also, when the rate of acetate consumption and proton
production slow down, the pH increases again towards the initial
values around 7. The current production is first limited by the
biomass attached on the anode (the anodic biofilm). The current
increases together with the biofilm biomass (Fig. 2G) as long as
there is substrate to sustain microbial activity and growth. After a
peak of 0.7–0.8 A/m2 around day 4, the current decreases again due
to the substrate exhaustion (Fig. 2B and E). Low buffering and low
proton transfer lead to lower and later (day 5) current peaks. With
sufficient buffer in the anode (here, 100 mM), the rate of H+ transfer
matters less for the anodic process and assuming that the cathodic Fig. 3. Spatial distributions of reduced mediator and ionic currents developed from the
process is not influenced, the same current-time curves are obtained biofilm on the planar anode calculated in Case 1 (km,H = 10− 6 m/s, 2 mM carbonate).
(Fig. 2B and E). For a rather constant pH around 7, the mediator (A) Surface plots of total reduced mediator concentration (CMH2,t) and arrow plots of
speciation can be neglected. The concentration of reduced mediator total ionic current vectors. The length of the current arrows is scaled relative to the
maximum current over the 15 days simulated MFC operation, thus the same scale
in time follows closely the current production and only one chemical applies for days 2 to 6. Microbial cells are shown as small black circles attached to
species dominates (Fig. 2F). The above simulations clearly indicate the electrode surface. (B) Distribution of produced current density over the planar
that the current model is capable of simulating the effect of anode to electrode at days 1 to 6.
 C. Picioreanu et al. / Bioelectrochemistry 78 (2010) 8–24 17

The production of electrons at the anode is accompanied by an

ionic current through the solution, defined as:

X
2 
i= −F Fzi um;i Ci jΦ + zi Di jCi :
i

The intensity and 2d spatial distribution of the ionic current at

different moments in time can be seen in Fig. 3A. The ionic current —
shown with arrows scaled relative to the maximum current obtained
during the whole operation of 15 days — is maximum at day 4 (locally
can reach 1 A/m2). Although the general direction of the current is
from the electrode towards the bulk liquid (constrained by the
insulation condition imposed on the lateral boundaries), it can be
clearly seen that the current mainly originates on the anode places
covered by the microbial colonies (Fig. 3B).
The 2d distributions of reduced mediator concentration (MH2 as
the total of the three leuco-thionine species) are also shown in Fig. 3.
The highest concentration of reduced thionine is in the middle of
the microbial colonies, as the result of microbial metabolism. The
concentration decreases at the anode surface due to MH2 consump-
tion in the electrochemical reactions and towards the bulk liquid due
to diffusive transport.
The concentration of protons is slightly higher (lower pH) in the
biofilm-covered areas (Fig. 4). However, there is no maximum
concentration in the middle of the colonies, but rather near the
electrode, where the protons are mainly produced. Protons will be
transported out of the biofilm by diffusion and electromigration.
Diffusion is driven by concentration gradients, and dominates near the
biofilm surface (Fig. 4B). The electromigration is driven only by the
potential gradient formed as a result of electroneutrality condition.
As a result of the electrode polarization, there is a higher potential
near the electrode, relative to the biofilm. This potential gradient
leads to more electromigration near the electrode (Fig. 4A), with
values comparable with those of molecular diffusion fluxes. In Fig. 4,
the maximum diffusion flux is NH+,D = 4 ∙ 10− 9 mol/m2s, and the
maximum electromigration flux is NH+,M = 3 ∙ 10− 9 mol/m2s. These
transport mechanisms are however negligible in the bulk liquid (or
anyway, far from the biofilm surface), where the convection clearly
dominates (maximum in Fig. 4 is NH+,C = 5 ∙ 10− 7 mol/m2s). We did

Fig. 5. Time progress of current production, pH and solutes concentrations in the bulk
and of biofilm biomass calculated in Case 2 (electroactive vs. fermentative bacteria). (A)
Concentrations of solutes in the bulk over time: glucose CGl,B, total butyrate CBu,t,B, total
propionate CPr,t,B, total acetate CAc,t,B, hydrogen CH2,B and methane CCH4,B; (B) Biomass
on the electrode over time for the different microbial species; (C) Current density and
pH calculated with km,H = 10− 2 m/s and 2 mM total carbonate (1 — thick lines); and
km,H = 10− 6 m/s and 20 mM total carbonate (2 — dashed lines).

not consider in this model electrophoretic fluxes created due to the

potential difference between cathode and anode. Supposedly, the
electrophoretic fluxes may also be important either for the proton
transfer through the biofilm [28], or for retaining negatively-charged
mediator species in the anodic biofilm (such as the phenazines, [2]).
Following model extensions will investigate these electrophoretic
fluxes too.

Fig. 4. Spatial distributions of solution potential, pH, diffusion, electromigration and
convection fluxes for H+ calculated in Case 1 (km,H = 10− 6 m/s, 2 mM carbonate). (A)
Surface plots of solution potential (V, in mV) and arrow plots of H+ electromigration
flux vectors. (B) Surface plots of pH and arrow plots of H+ diffusion (black) and H+
convection (white) flux vectors. The length of the diffusion flux arrows is on the same
scale with that from electromigration fluxes in (A), but the convection fluxes are ~ 100
times scaled down in (B). Microbial cells are shown as small black circles attached to the
electrode surface.
3.2. Case 2 — multi-species electroactive and fermentative biofilm on
planar electrode
For wastewater treatment, a mixed microbial community consist-
ing mainly of anaerobic microorganisms will be active in the anode
space. Therefore, the electroactive microorganisms will compete for
substrate and space in the biofilm with fermentative bacteria. The
18 C. Picioreanu et al. / Bioelectrochemistry 78 (2010) 8–24

Fig. 6. Spatial distributions of microbial populations, ionic currents in solution and H2 concentration in the biofilm domain calculated in Case 2. (A) Two-dimensional distribution of
microbial populations in the biofilm at day 2, 5, 9, and 13. The circles show cells of: XGl (red), XBu (yellow), XPr (green), XAc (cyan) and XH2 (blue). Arrows show ionic currents and
lines are streamlines of the liquid flow over the biofilm. The length of the current arrows is scaled relative to the maximum current over the 15 days. (B) Two-dimensional distribution
of H2 concentration (μM) in the biofilm domain, at day 2, 5, 9, and 13. (C) Current density over the electrode at day 2, 5, 9, and 13. Simulations in this case are with km,H = 10− 6 m/s,
2 mM total carbonate buffer.

model includes for this case several microbial species: XGl, XBu, XPr,
XAc and XH2, all active in the anaerobic digestion process, but it is
assumed that only XAc is able to use the mediator to exchange
electrons with the anode. In addition to all species from Case 1, solute
mass balances are also solved for Gl, BuH, Bu−, PrH and Pr−. Table 1
presents the reaction rates and stoichiometry and Table 2 all the
parameters used.
In general, the calculated change in time of concentrations of
chemical species in the bulk liquid follows the normal course of the
anaerobic digestion process (Fig. 5A). Glucose is first converted in
acetic, propionic and butyric acids, with the concomitant production
of molecular H2. When glucose is depleted, the fast growing glucose
consumer microorganisms stop accumulating in the biofilm (XGl
in Fig. 5B). In this acidogenic period the pH decreases (Fig. 5C).
Methanogens (XH2) further convert the H2 formed by the fermenta-
tive organisms into CH4 and grow as long as H2 is present in sufficient
concentrations. In the methanogenic biofilm there are strong H2
gradients, clearly visible in Fig. 6B. Especially at day 5, corresponding

Fig. 7. Spatial distributions of biofilm cells, flow velocity, ionic and electronic currents, reduced mediator and pH in the porous electrode, calculated in Case 3 (flow over). (A1–A3)
Surface plot: acetate concentration (mM) in the porous electrode at days 3,4, and 5; White arrows: liquid velocity vectors; Black lines: liquid streamlines (although some computed
streamlines go through the biofilm, the velocity is nearly zero); Cells are shown in a gradient of color from white (high activity) to black (low activity). (B1–B3) Surface plot: total
reduced mediator concentration at days 3,4, and 5; Black contours: biofilm surface; Red arrows: electronic currents pointing into the electrode granules; (C1–C3) Surface plot: pH at
days 3,4, and 5; Black contours: biofilm surface; Black arrows: ionic currents in the solution. Animations of the complete time evolution can be downloaded from the on-line
Supplementary Material, files MFC_Fig_7A.avi and MFC_Fig_7C.avi.
C. Picioreanu et al. / Bioelectrochemistry 78 (2010) 8–24 19
20 C. Picioreanu et al. / Bioelectrochemistry 78 (2010) 8–24

Fig. 7 (continued).
 C. Picioreanu et al. / Bioelectrochemistry 78 (2010) 8–24
Fig. 8. Spatial distributions of the biofilm, reduced mediator and current produced over the planar electrode, calculated in Case 3. (A) Two-dimensional distribution of reduced
mediator concentration (mM) at day 5. The biofilm surface is shown with a black contour line. (B) Distribution of current density produced over the electrode at 3, 4 and 5 days.
to the peak in H2 concentration, high H2 concentrations (3 μM) are
in the XGl colonies and extremely low in the XH2 populated biofilm
areas.
Methane forms also during acetate fermentation. This process
competes with the acetate utilization for electricity production by XAc.
Current production begins only when sufficient XAc microorganisms
have accumulated in the biofilm (see the 2d microbial distribution in
Fig. 6A and biofilm biomass in Fig. 5B), and this can take place only
after acetate has been formed by fermentation. With the used model
parameters, current production starts only after day 5 and reaches a
peak around day 10 (Fig. 5C). After that, the current rapidly decreases
Fig. 9. Comparison between the current densities and biomass produced on the porous
and planar electrodes, calculated in Case 3. (A) Current density (thick lines) and
biomass (dashed lines) produced over time in the porous electrode. Black lines: flow
over; red lines: flow though the porous electrode; (B) Current density (thick lines) and
biomass (dashed lines) produced over time on the planar electrode with surface area
equivalent with that of the porous electrode.
21
because the acetate is consumed and it is not produced at a sufficient
rate from propionate and butyrate. However, the small acetate
formation from higher organic acids is still able to support a low
current production. The current is produced in localized areas on the
electrode (Fig. 6C), corresponding to places where the colonies of XAc
form (Fig. 6A). Compared with Case 1, less current (max. 0.3 A/m2) is
obtained in this mixed-species biofilm.
The pH evolution in time is shown in Fig. 5C along with the current
production. When a low bicarbonate concentration is present in the feed
solution (2 mM total CO2/HCO3−), the pH falls rapidly during
acidogenesis, and a visible accentuation of the pH decrease occurs
when current production takes place. Due to the proton transfer rate
through the PEM (km,H = 10− 2 m/s), the pH increases again after the
acetate conversion with current production. In the case of a buffered
feed solution (20 mM total CO2/HCO3−) without proton transfer, the pH
decreases slower but it does not recover after 15 days. The electrical
current is larger due to the more favorable conditions for the elec-
trochemical oxidation of the mediator (higher pH during the current
production phase).
This case study clearly shows that the proposed model reflects
the general variation over time of pH, solute concentrations and
current as experimentally obtained in MFCs operated with anaerobic
sludge in the anode ([26] and Katuri et al. 2009, submitted). The
main obstacle however would be the realistic identification of model
parameters.
3.3. Case 3 — mono-species electroactive biofilm on porous electrode
The model setup allows the definition of any arbitrary geometry of
the biofilm domain (not necessarily rectangular) and any electrode
shape. To illustrate the model use in such a case, we have chosen here
to compare the MFC performance with biofilm developing on a
porous anode vs. the planar electrode (Case 1). The porous electrode
consists of a granular highly conductive material with inter-granular
voids filled by the anodic solution and partially by the biofilm
developed on the granules (Fig. 1). In two dimensions the granules
are represented by irregular circular shapes, which must be seen
actually connected in the third space dimension so that they form a
conductive material of uniform electrical potential. The ohmic
resistance of the electrode is included in the total resistance (Re).
Real equivalents of the modeled porous electrode can be a bed of
graphite beads, a carbon sponge or any sintered material. The porous
electrode material in this simulation has a porosity εL = 0.6, a specific
area aE = 26000 m2/m3 and granule diameters in the range of 20–
70 μm. Concentrations of chemical species in the ideally mixed bulk
liquid (and thus in the inflow of the biofilm domain) were kept
constant in time in this case to simulate a continuous operation of
22 C. Picioreanu et al. / Bioelectrochemistry 78 (2010) 8–24

Fig. 10. Spatial distributions of biofilm cells, flow velocity, ionic and electronic currents and reduced mediator in the porous electrode, calculated in Case 3 (flow through). (A1–A2)
Surface plot: acetate concentration (mM) in the porous electrode at days 4 and 5; White arrows: liquid velocity vectors; cells are shown in a gradient of color from white (high activity)
to black (low activity). (B1–B2) Surface plot: total reduced mediator concentration at days 4 and 5; black contours: biofilm surface; red arrows: electronic currents pointing into the
electrode granules; black and white arrows: ionic currents in the solution. Length of the arrows is scaled and it is comparable with the arrows in Fig. 7. Animations of the complete time
evolution can be downloaded from the on-line Supplementary Material, files MFC_Fig_10A.avi and MFC_Fig_10B.avi.

the MFC that has reached steady-state in bulk concentrations.
This is achievable at high feed flow rates Φ or small bulk liquid
volumes vB corresponding to a low feed solution residence time
(see Eq. (1)). One more process was added as compared to Case 1:
biomass attachment on the electrode surface or on the biofilm
matrix surface occurred at a rate of one cell per 2 h per biofilm
domain area (i.e., 6.5 ∙ 10 − 6 C-mol biomass/m2 h = 4.45 ∙ 10 − 8
g biomass/m2s).
If the electrolyte flows parallel to the top of the porous electrode
(see fluid velocity vectors in Fig. 7A) then only a small liquid flow is
 C. Picioreanu et al. / Bioelectrochemistry 78 (2010) 8–24
entrained in the pores (streamlines in Fig. 7A). The contribution
of convection to the total mass transfer of substrate or products to or
from the inner biofilm is very small compared with the diffusion and
electromigration fluxes. Therefore, acetate diffusion gradients
develop along the porous electrode depth (y direction). The biofilm
cells on the deeper electrode granules will have a lower activity
relative to the top ones (Fig. 7A3) and a thinner biofilm will develop
on these granules in time. The electron currents collected on each
granule are shown in Fig. 7B by normal vectors pointing into
the granule. It can be seen that deeper granules (situated at small
coordinate y) collect initially as much current as the top ones
(Fig. 7B1), but later the currents are much higher on similar granules
near the electrode surface (Fig. 7B3). This is the combined effect of
mass transfer limitations and biofilm formation. In the first place,
acetate and oxidized/reduced mediator gradients form along y.
Second, these gradients lead to more biomass (thicker biofilm) near
the electrode surface, which means more biological conversion of
acetate with mediator takes place there. Third, acidity accumulates in
the inner electrode volume (Fig. 7C), which slows down the
reversible electrochemical mediator oxidation. Gradients in the
ionic current densities in electrolyte are also evident from Fig. 7C.
The ionic currents accumulate from contributions on each granule
and due to the insulated lateral boundaries can only escape
towards the bulk liquid on top of the electrode. As ions can travel
only through the liquid, bottlenecks for the ionic currents form in
the narrower inter-granular space (e.g., see ionic currents at
y = 200–300 μm).
Simulations with an electroactive biofilm developed on a planar
electrode surface with an equivalent surface were also performed.
Total concentrations of reduced mediator and the current density
obtained on the planar electrode are presented in Fig. 8. When
compared with the values from the porous electrode, the mean
current density achieved on the planar electrode after 6 days is
higher. Apparently, this is correlated with more biomass obtained on
the planar electrode (Fig. 9). Mass transfer limitations are easier
avoidable on the planar electrode, although per volume electrode
the porous electrode offers more surface for biofilm formation.
This is why in the first four days of operation, when no transport
limitations have yet occurred, the two types of electrode show
similar current values. Interestingly, another effect of the mass
transport limitations is that although the bulk liquid concentrations
are kept constant in time (like in a steady-state continuous MFC
operation), the current also reaches a plateau value (Fig. 9). In spite
of a still linear biomass accumulation, compared to the exponential
biofilm growth phase up to day 4, a saturation level has been
reached for the overall current production rate (0.9 A/m2 for porous
compared with 0.6 A/m2 for planar case). This can be both due to a
limitation in the biological rate (the deep biofilm layers do not
have substrate) or in the electrochemical rate (pH inhibition or
mediator limitation).
It is beyond the scope of this paper to analyze in detail all the
parameters (thickness, shape, porosity, different flow regimes, etc.)
that may influence the performance of porous electrodes. However,
for illustration of the model versatility, a simulation was also
performed with flow through the electrode pores. We changed the
no-flux conditions for flow, mass and charge on the boundaries ΓIE and
ΓOE, by applying inlet conditions on ΓIE and outlet conditions on ΓOE,
similar with those on ΓI and ΓO respectively. In order to make the
results comparable, the same liquid flow rate as on ΓI (mean velocity
1 mm/s) was passed through the whole inlet boundary, ΓI U ΓIE. As
shown in Fig. 10A, flow through the electrode prevents much of the
substrate (acetate) limitation that microorganisms growing in the
electrode “depth” experienced, as convection becomes the main
mechanism for mass transport. With more active biofilm cells
(compare visually the biomass at day 5 on Fig. 10 with Fig. 7, or
numerically 1.5 g/m2 and 1 g/m2 at day 5 in Fig. 9A), about 40% more
23
current is obtained in the “flow-through” system. The current will still
become limited when the liquid pores are blocked with biomass and
the liquid flows through preferential channels (Fig. 10A2). Never-
theless, the biomass and currents obtained are much closer to those
obtained with a planar electrode (Fig. 10B).
Not investigated here but nevertheless important is the effect of
hydrodynamics on the biofilm structure. High flow rates generate
large shear forces on the biofilm, which should affect the biofilm
structure by removing pieces of biofilm and promoting growth of a
more compact biofilm [45,46]. However, the experience with biofilm
growth in other porous media shows that clogging and preferential
flow channeling will still occur, even if after a longer period of time
[47,48].
In conclusion, Case 3 shows how to analyze the behavior of
porous “bio-electrodes” and in general of electrodes of any shape.
It shows also that more area for biomass attachment does not
necessarily mean more biomass and more current production, as
long as convection through the pores is absent. However, sending
the liquid flow through an electrode with the geometrical
characteristics chosen here would probably require a greater energy
input which must be balanced against any increased power from
the MFC. Only the initial calculated pressure drop was ~ 5600 Pa/cm
and this pressure exponentially increased in time, following the
pore clogging with biomass. Obviously, the flow regimes chosen
here are kind of extremes and reality will be in between the two
options: flow over and flow through. The model should however be
applied to explore other possibly more favorable configurations,
with packed bed anodes [49,50] of larger granules and thus larger
pore throat diameters and less pressure drop. The model will be
used to explore different geometric configurations (e.g. rectangular
vs. cylindrical), flow directions and positioning of current feeds,
electrodes of different thickness and material properties (e.g.
conductivity).
4. Conclusions
The proposed microbial fuel cell model successfully integrates
multi-scale time-dependent mass and charge balances for solutes
and biomass in the anodic liquid (at macro-scale) and in a two-
dimensional biofilm (at micro-scale). The proposed model reflects
the generally observed variations of pH, solute concentrations
and current over time in MFCs with mono-species electroactive
microorganisms and also in those operated with a mixed
methanogenic/electrogenic microbial consortium from anaerobic
sludge. In addition, the new modeling approach opens the way for
the study of the influence of any two-dimensional biofilm and
electrode geometry on the MFC output parameters. By including
hydrodynamic calculations it has been shown that porous “bio-
electrodes” with greater specific surface area do not necessarily
supply more current, as long as convection through the pores is
absent. Also an innovative model solution strategy is developed in
this study, using a very efficient and flexible combination of
MATLAB, COMSOL and Java codes. The modeling framework
presented here can be directly extended to describe three-
dimensional systems, direct electron transfer and also other related
bio-electrochemical systems. We trust that the progress made in
this study opens the door to integrated use of experimental and
modeling research, which will greatly enhance the speed of
developing the MFC technology.
Acknowledgements
This research was supported by the European Union through Marie
Curie Transfer of Knowledge fellowships to Cristian Picioreanu
(contract number MTKD-CT-2004-517215) and by the Netherlands
Organization for Scientific Research (NWO, VIDI grant 864.06.003).
24 C. Picioreanu et al. / Bioelectrochemistry 78 (2010) 8–24
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bioelechem.2009.04.009.
References
[1] K. Rabaey, W. Verstraete, Microbial fuel cells: novel biotechnology for energy
generation, Trends Biotechnol. 6 (2005) 291–298.
[2] K. Rabaey, J. Rodríguez, L.L. Blackall, J. Keller, P. Gross, D. Batstone, W. Verstraete, K.H.
Nealson, Microbial ecology meets electrochemistry: electricity-driven and driving
communities, ISME J. 1 (2007) 9–18.
[3] M.C. Potter, Electrical effects accompanying the decomposition of organic com-
pounds, R. Soc. (Formerly Proceedings of the Royal Society) B 84 (1911) 260–276.
[4] D.R. Lovley, Bug juice: harvesting electricity with microorganisms, Nat. Rev.,
Microbiol. 4 (2006) 497–508.
[5] E. Katz, A.N. Shipway, I. Wilner, Biochemical fuel cells, in: W. Vielstich, A. Lamm, H.A.
Gasteiger (Eds.), Handook of Fuel Cells-Fundamentals, Technology, and Application,
John Wiley & Sons, Ltd., Chichester, 2003, pp. 355–381.
[6] C. Melhuish, I. Ieropoulos, J. Greenman, I. Horsfield, Energetically autonomous
robots: food for thought, Auton. Robots 21 (2006) 187–198.
[7] I. Ieropoulos, J. Greenman, C. Melhuish, Microbial fuel cells based on carbon veil
electrodes: Stack configuration and scalability, Int. J. Energy Res. 32 (2008) 1228–1240.
[8] S. Sakai, T. Yagishita, Microbial production of hydrogen and ethanol from glycerol-
containing wastes discharged from a biodiesel fuel production plant in a
bioelectrochemical reactor with thionine, Biotechnol. Bioeng. 98 (2007) 340–348.
[9] R.A. Rozendal, H.V.M. Hamelers, G.J.W. Euverink, S.J. Metz, C.J.N. Buisman, Principle
and perspectives of hydrogen production through biocatalyzed electrolysis, Int. J.
Hydrogen Energy 31 (2006) 1632–1640.
[10] S. Cheng, B.E. Logan, Sustainable and efficient biohydrogen production via
electrohydrogenesis, Proc. Natl. Acad. Sci. 104 (2007) 18875-18873.
[11] B.H. Kim, I.S. Chang, G.C. Gil, H.S. Park, H.J. Kim, Novel BOD (biological oxygen
demand) sensor using mediator-less microbial fuel cell, Biotechnol. Lett. 25 (2003)
541–545.
[12] I.S. Chang, H. Moon, J.K. Jang, B.H. Kim, Improvement of a microbial fuel cell
performance as a BOD sensor using respiratory inhibitors, Biosens. Bioelectron. 20
(2005) 1856–1859.
[13] A.K. Shukla, P. Suresh, S. Berchmans, A. Rahjendran, Biological fuel cells and their
applications, Curr. Sci. 87 (2004) 455–468.
[14] I.S. Chang, H. Moon, O. Bretschger, J.K. Jang, H.I. Park, K.H. Nealson, B.H. Kim,
Electrochemically Active Bacteria (EAB) and Mediator-Less Microbial Fuel Cells,
J. Microbiol. Biotechnol. 16 (2006) 163–177.
[15] R.A. Bullen, T.C. Arnot, J.B. Lakemanc, F.C. Walsh, Biofuel cells and their
development, Biosens. Bioelectron. 21 (2005) 2015–2045.
[16] B.E. Logan, B. Hamelers, R. Rozendal, U. Schröder, J. Keller, S. Freguia, P. Aelterman,
W. Verstraete, K. Rabaey, Microbial fuel cells: methodology and technology,
Environ. Sci. Technol. 40 (2006) 5181–5192.
[17] R.A. Rozendal, H.V. Hamelers, K. Rabaey, J. Keller, C.J. Buisman, Towards practical
implementation of bioelectrochemical wastewater treatment, Trends Biotechnol.
26 (2008) 450–459.
[18] G.-C. Gil, I.-S. Chang, B.H. Kim, M. Kim, J.-K. Jang, H.S. Park, H.J. Kim, Operational
parameters affecting the performance of a mediator-less microbial fuel cell,
Biosens. Bioelectron. 18 (2003) 327–334.
[19] G. Reguera, K.D. McCarthy, T. Mehta, J.S. Nicoll, M.T. Tuominen, D.R. Lovley, Extracellular
electron transfer via microbial nanowires, Nature 435 (2005) 1098–1101.
[20] Y.A. Gorby, S. Yanina, J.S. McLean, K.M. Rosso, D. Moyles, A. Dohnalkova, T.J.
Beveridge, I.S. Chang, B.H. Kim, K.S. Kim, D.E. Culley, S.B. Reed, M.F. Romine, D.A.
Saffarini, E.A. Hill, L. Shi, D.A. Elias, D.W. Kennedy, G. Pinchuk, K. Watanabe, S. Ishii,
B. Logan, K.H. Nealson, J.K. Fredrickson, Electrically conductive bacterial nanowires
produced by Shewanella oneidensis strain MR-1 and other microorganisms, Proc.
Natl. Acad. Sci. 103 (2006) 11358–11363.
[21] S.D. Roller, H.P. Bennetto, G.M. Delaney, J.R. Mason, J.L. Stirling, C.F. Thurston,
Electron-transfer coupling in microbial fuel cells. 1. Comparison of redox-mediator
reduction rates and respiratory rates of bacteria, J. Chem. Technol. Biotechnol. B
Biotechnol. 34 (1984) 3–12.
[22] K. Rabaey, N. Boon, S.D. Siciliano, N. Verhaege, W. Verstraete, Biofuel cells select for
microbial consortia that selfmediate electron transfer, Appl. Environ. Microb. 70
(2004) 5373–5382.
[23] K. Rabaey, N. Boon, M. Höfte, W. Verstraete, Microbial phenazine production enhances
electron transfer in biofuel cells, Environ. Sci. Technol. 39 (2005) 3401–3408.
[24] X.-C. Zhang, A. Halme, Modeling of a microbial fuel cell process, Biotechnol. Lett. 17
(1995) 809–814.
[25] C. Picioreanu, I.M. Head, K.P. Katuri, M.C.M. Van Loosdrecht, K. Scott, A computational
model for biofilm-based microbial fuel cells, Water Res. 41 (2007) 2921–2940.
[26] C. Picioreanu, K.P. Katuri, I.M. Head, M.C.M. Van Loosdrecht, K. Scott, Mathematical
model for microbial fuel cells with anodic biofilms and anaerobic digestion, Water
Sci. Technol. 57 (2008) 965–971.
[27] A. Kato-Marcus, C.I. Torres, B.E. Rittmann, Conduction-based modeling of the
biofilm anode of a microbial fuel cell, Biotechnol. Bioeng. 98 (2007) 1171–1182.
[28] C.I. Torres, A. Kato-Markus, B.E. Rittmann, Proton transport inside the biofilm limits
electrical current generation by anode-respiring bacteria, Biotechnol. Bioeng. 100
(2008) 872–881.
[29] R.A. Rozendal, H.V.M. Hamelers, C.J.N. Buisman, Effects of membrane cation transport
on pH and microbial fuel cell performance, Environ. Sci. Technol. 40 (2006) 5206–5211.
[30] J.S. Newman, Electrochemical Systems, 2nd ed.Prentice Hall, Englewood Cliffs, NJ,
1991.
[31] C. Picioreanu, M.C.M. Van Loosdrecht, A mathematical model for initiation of
microbiologically influenced corrosion by differential aeration, J. Electrochem. Soc.
149 (2002) B211–B223.
[32] C. Picioreanu, J.U. Kreft, M.C.M. Van Loosdrecht, Particle-based multidimensional
multispecies model, Appl. Environ. Microbiol. 70 (2004) 3024–3040.
[33] J.B. Xavier, C. Picioreanu, M.C.M. Van Loosdrecht, A framework for multidimen-
sional modelling of activity and structure of multispecies biofilms, Environ.
Microbiol. 7 (2005) 1085–1103.
[34] J.U. Kreft, C. Picioreanu, J.W.T. Wimpenny, M.C.M. Van Loosdrecht, Individual-
based modelling of biofilms, Microbiol.-SGM 147 (2001) 2897–2912.
[35] H.P. Bennetto, M.E. Dew, J.L. Stirling, K. Tanaka, Rates of reduction of phenothiazine
‘redox’ dyes by E. coli. Chem. Ind. Nov. (1981) 776.
[36] G.M. Delaney, H.P. Bennetto, J.R. Mason, S.D. Roller, J.L. Stirling, C.F. Thurston,
Electron-transfer coupling in microbial fuel cell. 2. Performance of fuel cells
containing selected microorganisms-mediator-substrate combination, J. Chem.
Technol. Biotechnol. 34B (1984) 13–27.
[37] C.F. Thurston, H.P. Bennetto, G.M. Delaney, J.R. Mason, S.D. Roller, J.L. Stirling,
Glucose metabolism in a microbial fuel cell. Stoichiometry of product formation in
a thionine-mediated Proteus vulgaris fuel cell and its relation to coulombic yields,
J. Gen. Microbiol. 131 (1985) 1393–1401.
[38] W.M. Clark, B. Cohen, H.D. Gibbs, Methylene blue, Public Health Rep. 40 (1925)
1131–1201.
[39] B.P. Nikolsky, V.V. Palchevsky, State of oxidation–reduction indicators in solution,
Russ. Chem. Bull. 6 (1957) 547–554.
[40] J.J. Heijnen, Bioenergetics of microbial growth, in: M.C. Flickinger, S.W. Drew
(Eds.), Encyclopedia of Bioprocess Technology: Fermentation, Biocatalysis, and
Bioseparation, Wiley-Interscience, New York, 1999, pp. 267–291.
[41] D.J. Batstone, J. Keller, I. Angelidaki, S.V. Kalyuzhnyi, S.G. Pavlostathis, A. Rozzi, W.T.M.
Sanders, H. Siegrist, V.A. Vavilin, Anaerobic Digestion Model No. 1 (ADM1), IWA Task
Group for Mathematical Modelling of Anaerobic Digestion Processes, IWA Publishing,
London, 2002.
[42] C. Picioreanu, D.J. Batstone, M.C.M. Van Loosdrecht, Multidimensional modelling
of anaerobic granules, Water Sci. Technol. 52 (2005) 501–507.
[43] D.J. Batstone, C. Picioreanu, M.C.M. Van Loosdrecht, Multidimensional modelling
to investigate interspecies hydrogen transfer in anaerobic biofilms, Water Res. 40
(2006) 3099–3108.
[44] P. Vanysek, CRC handbook of chemistry and physics, 82nd ed., CRC Press LLC, Boca
Raton, 2001, pp. 5–95 and 6–194.
[45] C. Picioreanu, M.C.M. Van Loosdrecht, J.J. Heijnen, Two-dimensional model of
biofilm detachment caused by internal stress from liquid flow, Biotechnol. Bioeng.
72 (2001) 205–218.
[46] J.B. Xavier, C. Picioreanu, M.C.M. Van Loosdrecht, A general description of detachment
for multidimensional modelling of biofilms, Biotechnol. Bioeng. 91 (2005) 651–669.
[47] G.E. Kapellos, T.S. Alexiou, A.C. Payatakes, Hierarchical simulator of biofilm growth
and dynamics in granular porous materials, Adv. Water Res. 30 (2007) 1648–1667.
[48] D.A. Graf von der Schulenburg, T.R. Pintelon, C. Picioreanu, M.C.M. Van Loosdrecht,
M.L. Johns, Three-dimensional simulations of biofilm growth in porous media,
AIChE J. 55 (2009) 494–504.
[49] D. Sell, P. Kramer, G. Kreysa, Use of an oxygen gas diffusion cathode and a three-
dimensional packed bed anode in a bioelectrochemical fuel cell, Appl. Microbiol.
Biotechnol. 31 (1989) 211–213.
[50] K. Rabaey, P. Clauwaert, P. Aelterman, W. Verstraete, Tubular microbial fuel cell for
efficient energy generation, Environ. Sci. Technol. 39 (2005) 8077–8082.