Microbial Fuel Cells Generating Electricity from Rhizodeposits of ...

Environ. Sci. Technol. XXXX, xxx, 000–000
Microbial Fuel Cells Generating
Electricity from Rhizodeposits of
Rice Plants
LIESJE DE SCHAMPHELAIRE, †
LEEN VAN DEN BOSSCHE, †
HAI SON DANG, † MONICA HÖFTE, ‡
NICO BOON, † KORNEEL RABAEY, § AND
WILLY VERSTRAETE* ,†
Laboratory of Microbial Ecology and Technology (LabMET)
and Laboratory of Phytopathology, Ghent University, Coupure
Links 653, B-9000 Ghent, Belgium, and The Advanced Water
Management Centre, University of Queensland,
St Lucia QLD 4072, Australia
Received August 03, 2007. Revised manuscript received
January 14, 2008. Accepted January 22, 2008.
Living plants transport substantial amounts of organic material
into the soil. This process, called rhizodeposition, provides
the substrate for the rhizospheric microbial community. In this
study, a laboratory-scale sediment microbial fuel cell, of
which the anode is positioned in the rhizosphere of the rice
plants, is used to microbially oxidize the plant-derived organics.
An electrical current was generated through the in situ
oxidation of rhizodeposits from living rice plants. The electrical
power output of a sediment microbial fuel cell was found to
be a factor 7 higher in the presence of actively growing plants.
This process offers the potential of light-driven power
generation from living plants in a nondestructive way.
Sustainable power productions up to 330 W ha -1 could be
attributed to the oxidation of the plant-derived compounds.
Introduction
Plants continuously provide an input of organic matter to
the soil throughout their plant life. As plants decay, dead
roots and shoot residues remain in the soil. During the
growing season, organic carbon enters the soil as rhizodeposits
(1). The latter comprise several groups of organics
such as (i) water-soluble, low molecular weight, and passively
lost exudates, (ii) secretions which are of high molecular
weight and actively lost, (iii) lysates from sloughed-off cells
and decaying roots, (iv) gases, and (v) mucilage covering
roots (2). Moreover, rhizodeposition also comprises the C
flux that is directed out of roots through symbiotic mycorrhizal
fungi, which are associated with over 80% of the plants
(3). Since it is difficult to experimentally distinguish exudates
from other rhizodeposits, root exudates are often defined as
all organic substances released by healthy and intact roots
into the environment. They comprise a gathering of carbohydrates,
amino acids, amides, aliphatic acids, aromatic acids,
fatty acids, sterols, enzymes, hormones, vitamins, and others
(2). The release of exudates plays an important role in nutrient
acquisition (4).
* Corresponding author phone: 0032 9 264 59 76; fax: 0032 9 264
62 48; e-mail: Willy.Verstraete@UGent.be.
† LabMET, Ghent University.
‡ Laboratory of Phytopathology, Ghent University.
§ University of Queensland.
In rice paddies, rhizodeposition counts for 200 kg organic
C ha -1 crop cycle -1 (5). In a flooded rice system, this
substantial input of organic material is transformed into
methane to the extent that rice agriculture worldwide
contributes 7-20% of the total methane emissions (6).
Rhizodeposition was shown to be the main origin of methane
evolution in rice paddies, with a share of 25% from exudates
and 75% from decomposing root residues (5). Next to being
a source of greenhouse gases, these rhizospheric processes
also represent a significant loss of energy from the rice system:
rice plants lose substantial amounts of trapped solar energy
as rhizodeposition, while the gaseous end product of the
anaerobic composition thereof, methane, has a high energetic
value. It would most certainly be interesting to recuperate
this flow of energy from living plants, as it represents a true
source of green energy. In this research, the latter was
attempted through the installation of a so-called sediment
microbial fuel cell (SMFC) in the rhizosphere of the rice plants.
A SMFC is a microbial fuel cell, with an anode buried in
a reduced matrix and a cathode floating in the overlying,
oxidized water layer (7). The submerged matrix can serve as
a support for plant growth; it can be composed of various
types of materials. At the anode, a microbially catalyzed
oxidation of reduced compounds is responsible for a delivery
of electrons to the anodic electrode. The electrons pass
through an electrical circuit, containing a power user. Arriving
at the cathodic electrode, they react with the available electron
acceptor, such as oxygen. A microbial fuel cell in general is
thus able to extract electrical power from the oxidation of
bioconvertible substrates (8). Thus far, research on SMFCs
primarily involved marine SMFCs and typically resulted in
sustained power productions of 9-16 mW m -2 total anodic
surface (7, 9, 10).
The rice rhizodeposits potentially form substrates for an
SMFC. This research aims to exploit this flow of energy by
oxidizing the substrates derived from living plants directly
at an anode. The purpose is to demonstrate that growing
plants can serve as ongoing sources of organic substrates for
power generation in a SMFC configuration. To achieve this
goal, several SMFC reactors were set up, comprising reactors
with and without plants, operated in open and closed
electrical circuit and with several types of substratum, namely,
soil, vermiculite, and graphite granules.
Materials and Methods
Growth of Rice Plants. Rice plants used in the experiments
belonged to Oryza sativa ssp. indica, cultivar C101PKT,
originating from the International Rice Research Institute.
After 1 week humid incubation of the rice seeds at 28 °C and
3 weeks growth in soil or vermiculite, the plants were
transplanted into containers equipped with electrodes and
further referred to as sediment microbial fuel cell (SMFC)
reactors. Nutrients were dosed at a rate of 5 g of (NH 4 ) 2 SO 4
m -2 week -1 and 10 g of FeSO 4 · 7H 2 Om -2 week -1 with soil as
support layer, while plants growing in vermiculite were fed
with Hoagland’s hydroponics solution (11) every 2 or 3 days.
During operation of the SMFC reactors in soil and vermiculite,
a regular administering of nutrients was only applied during
the first 3 weeks of reactor runs and during 3 weeks around
day 150 of the reactor runs. No nutrients were given to the
plants in reactors with graphite granules. The omission of
nutrients was intended to stimulate root exudation (4).
Trials during the Summer of 2006: Construction. For
the first two series of SMFC reactors (see Figure 1), plastic
containers with an upper area or plant growth area (PGA) of
272.25 cm 2 and a total volume of 4.6 L were filled with either
10.1021/es071938w CCC: $40.75 © XXXX American Chemical Society VOL. xxx, NO. xx, XXXX / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 A
Published on Web 03/18/2008

FIGURE 1. Schematic representation of the reactor setup. Each
of the four separated compartments holds a so-called sediment
microbial fuel cell (SMFC): a support matrix in which plants
can root and which holds one or more anodes is submerged by
water in which a cathode is placed. The SMFC can be planted
or not. The support matrix can be, e.g., regular soil, vermiculite,
etc. In the case of soil as matrix, the reactors presented above
are respectively called soil 0 (control reactor in closed circuit,
without plants), soil 1 (reactor in closed circuit, with plants),
soil 2 (repetition of soil 1 ), and soil OC (reactor in open circuit,
with plants).
soil (Structural Professional type 1, M. Snebbout N.V.,
Kaprijke, Belgium) or vermiculite (exfoliated vermiculite, Sibli
SA Vermiculite et Perlite, Andenne, Belgium) as supports for
plant growth. These containers were placed in subdivided
aquaria filled with tap water to a height of 25 or 8 cm above
the support surface, thus leading to a total volume of 13.3
L/(microbial fuel cell). Before placement, two anodic graphite
mats (3.18 mm thickness, Alfa Aesar) of 9 cm by 12 cm were
placed in the support layers, one at 6 cm and one at 14 cm
below the support surface, resulting in a total anodic
geometric area (GA) of 216 cm 2 . Only one anode of 6 cm by
9 cm was placed at a depth of 6 cm below the surface in
reactors meant for open circuit (no current harvesting was
required). All mats were interwoven with a graphite rod (5
mm diameter, Morgan), which was attached to the electrical
circuitry through an insulated connection. The third series
of reactors consisted of glass cylinders with a diameter of 14
cm, filled with 11.5 cm of graphite granules (5 mm diameter,
Le Carbon), a sand layer of 1.5 cm, and a standing water
layer of 5 cm. The anodic structure as a whole had a geometric
area, equal to the plant growth area, of 154 cm 2 . A graphite
rod of 13 cm and insulated wire connected the anodic granule
matrix to the outside. Four plants grown in soil were planted
in the containers filled with soil, while respectively 6 and 2
plants grown in vermiculite were planted in the containers
filled with vermiculite and cylinders with granules. Control
reactors were unplanted. Bacterial inocula were added to all
anodic compartments, by injecting all reactors with 10 mL
of the effluent of an acetate oxidizing MFC reactor (12). Soil
and vermiculite matrices were furthermore initially mixed
with 20 mL of a methanogenic culture (presettling tank of
a constructed wetland, Wontergem, Belgium).
Two types of cathodes were used. The first type consisted
of a tube of cation exchange membrane (Ultrex), closed at
the bottom (diameter, 3.4 cm; length, 20 cm). These tubes
were filled with graphite granules, a regularly replenished
100 mM phosphate buffered solution of 100 mM K 3 Fe(CN) 6 ,
and a graphite rod for electrical contact. This type of cathode
was used for the reactors with graphite granules at the anode
(except for gran OC , which did not contain a cathode) and
during the first 100 and 125 days respectively of the test runs
of reactors with soil and vermiculite. During the remaining
time of the latter two series of reactors, a biologically catalyzed
oxygen reducing cathode was used. This graphite mat cathode
(5 cm by 12.5 cm) was connected with a graphite rod, floating
in the top water layer of the reactors and slightly aerated
through an air pump. Before placement of the cathodes, the
mats were inoculated with a performant cathodic culture
and the oxygen reduction reaction was initiated in separate
SMFCs.
Trials during the Summer of 2006: Operation. The entire
reactor setup consisted of 13 SMFC reactors, 4 with soil (soil),
4 with vermiculite (verm), and 5 with graphite granules (gran)
as support material. To denote the different types of reactors,
a subscript is added to the support abbreviation: 1 and 2 for
the repetitions of the reactors with plants in closed circuit,
allowing current generation (soil 1 , soil 2 , verm 1 , verm 2 , gran 1 ,
and gran 2 ), 0 and 0′ for the control reactors without plants
in closed circuit (soil 0 , verm 0 , gran 0 , and gran 0′ ) and OC for
the reactors with plants in open circuit (soil OC , verm OC , and
gran OC ), where no current is possible. Table S1 of the
Supporting Information gives an overview of the different
types of reactors and their names.
The electrical circuits were closed through the use of a
variable external resistance, which at any time had one value
(between 75 and 500 Ω) per series of reactors. The reactors
were positioned in a greenhouse with a temperature thermostat
set at 28 °C and under special horticulture lamps
(HQI, 400 W) with a 16-h-day-8-h-night cycle. On the basis
of temperature and light conditions in- and outside the
greenhouse, the reactor runs can be subdivided in three
succeeding test periods, starting from the beginning of April
2006. Average outside shortwave radiation and outside
temperatures during test period 1 were 157 W m -2 (second
half of this test period) and 13.2 ( 3.5 °C (inside around 30
°C); during test period 2, 231 W m -2 and 20.0 ( 4.0 °C (inside
around 42 °C) and during test period 3, 105 W m -2 and 16.1
( 2.5 °C (with intermediate inside temperatures). Radiation
inside the greenhouse was determined to be around 18 MJ
m -2 day -1 during the second test period (LICOR, LI-190S,
(13)). Test periods 1-3 respectively started on days 1, 52,
and 120 of the soil reactor runs, on days 1, 60, and 128 of the
verm reactor runs, and days 1, 2, and day 70 of the gran
reactor runs. The total reactor runs from the soil and verm
reactors respectively comprised 204 and 188 days, thus
covering the entire cycle of the rice plant life from transplanting
onward. The reactor runs from the gran reactors
lasted 142 days.
Trials during the Summer of 2007: Construction and
Operation. One year after the operation of the three series
of reactors described above, a fourth series of reactors was
constructed analogous to the soil reactors. This series is
referred to as b-soil reactors. The construction of the b-soil
reactors differed from that of the soil reactors by the use of
plastic containers with an upper PGA of 231.04 cm 2 , height
of 19.6 cm, and total volume of 3.3 L. Three anodic graphite
mats (Sigratherm, KFA, 2.5 mm thickness), each of 6 cm by
11 cm, were placed horizontally at respectively 5, 11 and 17
cm below the support surface and resulted in a total anodic
geometric area of 198 cm 2 . The cathode consisted in each
case of an aerated graphite mat (Alfa Aesar, 3.18 mm
thickness) of 5 cm by 12.5 cm. Three containers were planted
with 5 4-week-old rice plants each (b-soil 1 , b-soil 2 , and b-soil 3 ),
two containers remained unplanted (b-soil 0 and b-soil 0′ )
(Table S1). An extra horticulture lamp (HQI, 400 W) was placed
above the reactors to account for lower light influx. The
reactor runs started in the beginning of June 2007 and could
be divided in two test periods. The average shortwave
radiation and outside temperature were respectively 166 W
m -2 and 18.4 ( 3.6 °C in the first period and 78 W m -2 and
13.6 ( 3.8 °C in the second test period, which started on day
88 of the reactor runs. An external resistance of either 497
or 100 Ω was used. Data collected until day 144 of the reactor
run are included in this paper.
B 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. xxx, NO. xx, XXXX

Electrochemical Data. Continuous potential measurements
were recorded every 5 or 7.5 min. Other electrochemical
measurements and the analysis thereof were
performed according to Logan et al. (14). Negative signs were
assigned to power outputs corresponding with reverse
(negative) currents. The expression of current production as
chemical oxygen demand (COD) oxidation was based on the
release of 4 mol of electrons/(mol of COD oxidized or mol
of O 2 ).
Chemical Analysis. Chemical oxygen demand analysis
was chosen as a general measurement of the amounts of
oxidizable material in the solution of the anodic compartment.
Mixed samples of the solution were collected with a
spinal needle, stored at 4 °C, filtered (Whatman, 597 1 / 2 ) and
analyzed according to the dichromate method (15). The dry
weights (DW) of the rice shoots as well as the panicles were
determined by drying the rice biomass overnight at 105 °C.
The pooled above-ground biomass weights per reactor were
divided by the initial number of plants to normalize for the
different numbers of plants in reactors with vermiculite and
soil.
Statistical Analysis. Statistical analysis was performed
on the data from Figure 4, excluding the data points from
verm OC to prevent biases, resulting in 14 data pairs. On the
basis of the shape of the scatterplot of the data, linear, logistic,
and nonlinear regressions were applied and the Pearson
correlation coefficient (two-tailed) between the transformed
COD concentrations and the current productions was
determined. A Michaelis–Menten type of model with lateral
shift [current ∼ C max (COD - X)/(K + (COD - X))] and
parameters C max (maximum current) ) 2.4, K (Michaelis
constant) ) 141, and X (lateral shift) ) 74 resulted in the
highest correlation coefficient, being 0.76 (p ) 0.002). Normal
distribution of the data sets was checked through P-P plots
and the Kolmogorov–Smirnov test. Variability is expressed
as standard deviation throughout the paper.
Results
Overall Electrochemical Performance. In the summer of
2006, four SMFCssof which three were planted with
riceswere set up with natural soil as support and were held
in open circuit (no current was allowed) until a cell potential
higher than 600 mV was reached, indicating anaerobic
conditions around the anode. The stable open circuit cell
potential of soil OC , the control reactor with plants that was
kept in open circuit throughout the experimental period,
slowly increased during the entire reactor run of 204 days
from 695 to 800 mV with an anodic redox potential decreasing
from -240 mV vs SHE to -340 mV vs SHE. The sharpest
increase in open circuit cell potential occurred concomitantly
with the start of the second test period, with a stronger sun
regime. Notable currents of about 3 mA m -2 anodic geometric
area (GA) were harvested from the moment the electrical
circuits of reactors soil 1 and soil 2 (with plants) and soil 0
(without plants) were closed. A steady relation between the
subsequent, higher level performances of these reactors was
only observed from the start of the second test period onward.
Figure S1a of the Supporting Information contains a simplified
performance profile of the soil reactors through time,
representing the cumulative electron transfer.
Averaged over the entire period of closed circuit, soil 1 ,
soil 2 , and soil 0 respectively delivered 68 ( 45, 54 ( 38, and
40 ( 20 mA m -2 GA. When averaging over concurrent, stable,
and representative periods (not limited by the cathode
reaction, plant age, or climate), comprising at least 27 days
per reactor (between day 63 and day 129), the reactors with
plants in soil produced a current of 120 ( 19 mA m -2 GA,
which is a factor 2.7 higher than the current of 44 ( 8mA
m -2 GA, produced by the reactor with soil. During that period,
the reactors with plants produced a factor 7 more power
FIGURE 2. Electrical power output per m 2 geometric anode area
(GA) and standard deviation. (a) First three series of reactors,
operated in the summer of 2006. At the left side of the figure,
power output is given for periods with a ferricyanide cathode
(cross-hatched bars) and with an oxygen reducing cathode
(black bars). The right side only refers to periods with
ferricyanide cathodes. A subscript 0 points to the absence of
plants. Each bar depicts a representative performance period
per reactor, comprising a period of respectively 16, 15, and 5
days or more in case of reactors with soil (soil), vermiculite
(verm), and graphite granules (gran). The cross-hatched bars
from soil and vermiculite reactors originated from test period 2;
all other bars, from test period 3. (b) Repetition series with soil
(b-soil), operated during the summer of 2007. The bars depict
representative performance periods of at least 18 days. At any
instance, an oxygen reducing cathode was used. A subscript 0
points to the absence of plants.
than the control reactor without plants, being 26 ( 7mW
m -2 GA versus 3.7 ( 1.8 mW m -2 GA (Figure 2a; Figure S2
and Table S2 of the Supporting Information).
A similar reactor setup was used with vermiculite as
support. These SMFCs were held in open circuit for 29 days,
during which the obtained open circuit cell potentials
remained fluctuating with daily peaks between -50 and +130
mV. At the start of test period 2 (day 60), the cell potentials
of the reactors with plants started to increase steeply to reach
higher and more stable values: the open circuit cell potential
of verm OC , the control reactor with plants that was kept in
open circuit throughout the experimental period, reached
consistently positive values up to 760 mV, verm 1 increased
its cell potential with a factor 15 and verm 2 changed its
regularly negative cell potential into a stable positive cell
potential. It required 104 dayssstill in test period 2sfor verm 2
to steeply increase its cell potential to become as high as that
of verm 1 . During periods with high cell potentials, a fluctuation
in the latter could also be observed, this time without
negative cell potentials (Figure 3). These fluctuations were
determined by the anodic potentials; high cell potentials and
low anodic potentials were reached when light conditions
prevailed and low cell potentials and high anodic potentials
were reached in the absence of light. The closed circuit reactor
without plants (verm 0) demonstrated low cell potentials (with
a negative average) during the entire reactor run. The
VOL. xxx, NO. xx, XXXX / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 C

FIGURE 3. Fluctuation in voltage produced in reactors
containing vermiculite and plants. Vertical bars show the
artificial light conditions inside the greenhouse (grey bars ) no
light). The external resistance for the reactor in closed circuit
was 495 Ω, and a ferricyanide type of cathode was used. Dark
line, verm OC ; light line, verm 1 .
FIGURE 4. Anodic organic substrate of the reactors with
vermiculite as support layer. Current production, expressed as
rate of COD oxidation per day and per liter total anode
compartment (TAC), is plotted versus COD concentration
measured in the anodic solution. Reactors with plants in closed
circuit (b, verm 1 ; O, verm 2 ) and with plants in open circuit
(2, verm OC ) are presented.
cumulative electron transfer of the three reactors with
vermiculite support in closed circuit is given in Figure S1b.
Although there was variability in the length of the start-up
period, the reactors with plants clearly behaved similar to
each other and differed from the reactor without plants, which
produced low negative currents.
The total time-averaged current production of verm 1 ,
verm 2 , and verm 0 was respectively 36 ( 17, 14 ( 16, and -0.7
( 1.0 mA m -2 GA. The mean current averaged over total
representative periods of at least 38 days per reactor (between
day 54 and day 188) gives 44 ( 9mAm -2 GA for the reactors
with rice (verm 1 and verm 2 ) and -0.56 ( 0.69 mA m -2 GA
for the reactor without rice plants (verm 0 ). The mean power
for verm 1 and verm 2 was 21 ( 8mWm -2 GA; that of verm 0
was -0.008 ( 0.01 mW m -2 GA (Figure 2a and Table S2).
Verm 1 could sustain a power production of 33 mW m -2 GA
for at least 6 days.
A third series of reactors contained graphite granules as
plant support. These SMFCs generally yielded low cell
potentials, as was the case with the vermiculite reactors. Near
the end of the reactor runs, current production from the
reactors with plants (gran 1 and gran 2 ) started to increase
steeply, whereas that of the reactors without plants (gran 0
and gran 0′ ) did not. By subtracting the cumulative electron
transfer of the last type of reactor from the first one, the net
cumulative electron transfer could be obtained (Figure S1c).
Averaging power production over a high-performance period
of a minimum of 5 days per reactor (between day 119 and
day 141) yielded 15.8 ( 2.1 mW m -2 GA for the granule
reactors with plants and 0.14 ( 0.03 mW m -2 GA for those
without plants (Figure 2a).
To confirm the test results and show that repetition in
time was possible, an additional series of reactors was set up
during the summer 1 year later, with soil as support layer
(b-soil reactors). The cumulative electron transfer for these
reactors is shown in Figure S1d. Averaged over the entire
reactor run the reactors b-soil 1 , b-soil 2 , b-soil 3 , b-soil 0 , and
b-soil 0′ respectively produced 43 ( 30, 35 ( 27, 39 ( 23, 22
( 17, and 7 ( 12 mA m -2 GA. The mean current production
during a representative period of at least 18 days (between
day 62 and day 85) was a factor 3.4 higher for the reactors
with plants (b-soil 1 , b-soil 2 , and b-soil 3 ) than for those without
plants (b-soil 0 and b-soil 0′ ), being 68 ( 20 versus 20 ( 12 mA
m -2 GA. The mean power output for the reactors with plants
was 9.9 ( 6.0 mW m -2 GA, which was a factor 9 higher than
the power output from the reactors without plants (1.1 ( 1.1
mW m -2 GA) (Figure 2b; Figure S2 and Table S2). This
confirmed the observations of the previous summer.
Energy Considerations. The energy obtained through the
planted SMFC systems can be compared with the energy
cycle of irradiant solar energy, photosynthesis, and exudation.
An extended version of these calculations can be found in
the Supporting Information. On the basis of the actual rice
biomass production in these experiments and a release of
3% as exudates, an exudate release of 0.41 g of C m -2 PGA
day -1 was obtained. Considering a general composition of
[CH 2 O] for the released material, this can be converted in a
daily release of electrons and energy. The averaged current
and power production from the SMFCs, which could be
attributed to the plants, was 47.5 and 17 mW m -2 PGA,
corresponding with 0.043 mol of electrons and 1.47 kJ m -2
PGA day -1 . On the basis of the biomass production and the
oxidation of the readily biodegradable theoretical exudate
release, the SMFCs with rice plants thus obtain a coulometric
efficiency (recovery of electrons) of 31% and an energetic
efficiency of 9%. Through a combination of the sunlight
interception efficiency (89%; based on LAI), the photosynthetic
efficiency for C 3 plants (4.9%), the exudates release
(3%), and the fuel cell efficiency (9%), an overall efficiency
for the conversion of light energy into electrical energy is
obtained, being 0.01%. This can be confirmed by comparing
the harvested 1.47 kJ m -2 PGA day -1 with the radiation energy
in the greenhouse (about 18 MJ m -2 day -1 , the same order
of magnitude as the solar radiation of 15 MJ m -2 day -1 in rice
producing countries).
Rice Plants and the Anodic Organic Substrate. Measurements
of soluble COD in the interstitial anodic solution
of reactors with vermiculite as support were taken as a
representative for the amount of reduced organics available
as microbial substrate at the anode, as this forms the source
of electrical energy in the systems. Higher COD concentrations
could be measured in verm 1 than verm 2 . The first reactor
generally performed better from an electrical point of view
than the latter. There was more COD in solution in reactors
producing current than in the reactor left in open circuit
(see Figure S3 in the Supporting Information for CODcentrations
through time). Figure 4 demonstrates a positive
Michaelis–Menten-like correlation between the current
production in the vermiculite reactors and the COD concentration
in the anodic compartment. This type of correlation
might imply that the attainable current output is
limited for the used setup.
At the end of the reactor runs, the aboveground biomass
mainly consisted of senescent leaves, suggesting that decaying
plant material could be an important anodic substrate at
this point. This was especially the case with the reactors in
D 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. xxx, NO. xx, XXXX

graphite granules, where the electrical current production
(Figure S1c) notably increased after the plants had almost
died off. The reason for this plant mortality presumably was
a visible leakage of traces of ferricyanide from the cathode
into the anode compartment, since the plants in gran OC (open
circuit without a ferricyanide cathode) remained healthy
(results not shown).
The total aboveground production of plant biomass was
determined at the end of the soil and verm reactor runs.
With an overall production of 27 ( 7 g of DW/(plant in closed
circuit) and 21 ( 1 g of DW/(plant in open circuit), there
appears to be no effect of the electrical circuit on the plant
biomass production.
Discussion
In this study it was demonstrated that the presence of plants
increased the power production from sediment microbial
fuel cells (SMFCs) (see also Figure 2, Figure S2, and Table S2)
and that it is possible to oxidize plant-derived material in
situ. A reactor with soil as support, holding organic oxidizable
substrate, enabled a freshwater SMFC to produce electrical
power in a sustained way. The presence of plants increased
the current output from this type of reactor with a factor 2.7
and the power output with a factor of 7. The series of 2007,
only using soil as support, showed that our data were
reproducible during a subsequent summer period. Differences
in absolute values and trends between the two soil
series are probably related to differences in plant growth
and age, time of reactor run, and weather which indeed
slightly varied between the two experimental years. If the
reactors were filled with a support matrix, which did not
contain organic material, being vermiculite (hydroponic plant
growth substratum) or graphite granules (anodic matrix for
reactor type MFCs), no apparent electrical current could be
produced in the absence of plants. The presence of plants,
more specifically the presence of living plant roots around
the anode, allowed substantial power generation. Roots of
living plants transport organic material such as exudates into
the rhizosphere. As exudates have a major stimulatory effect
on microbial growth and activity because of their rapid
assimilation (2), they indeed qualify as a readily oxidizable
anodic substrate. The highest sustained electrical output,
based on plant-derived material, was an output of 33 mW
m -2 GA or 330 W ha -1 GA.
On the basis of the oxidation of theoretical exudate release,
the SMFCs with rice plants obtained a coulometric efficiency
of 31% and an energetic efficiency of 9%. Further engineering
improvements on the system are necessary to enable an
increase in fuel cell efficiency. The results from Figure 2a
demonstrate that the electrical output from a system with
ferricyanide at the cathode (used before in reactor type MFCs
for a reliable output (12)) was at equivalent levels as that
from a system with a sustainable, biologically catalyzed
cathode, using dissolved oxygen as cathodic reagent.
The electrical current and power produced in the SMFCs
in this study can be compared with the power obtained in
previous SMFC research, with a similar design but different
substrates. A long-term sustained current and power production
of 120 mA m -2 GA (geometric anode area) or 56 mA
m -2 AS (total anodic electrode surface) and 26 mW m -2 GA
or 12 mW m -2 AS was obtained in this work from a rice
freshwater ecosystem. Long-term sustained current and
power production from marine systems amounted to 34 mA
m -2 AS (7) and 16 mW m -2 AS (9). Reimers et al. (16) reached
a power output (sustainable for 24 h) of about a factor 2
higher at an ocean seep. Sustained electrical production from
a sediment freshwater system amounted to 9 mA m -2 AS
(17). Power production is more difficult to obtain from
freshwater systems than from salt water systems because of
a lower electric conductivity and decreased reactivity of the
electrodes. Hence, the long-term sustained results obtained
in this (freshwater) study are substantial.
During the trials in 2006, a start-up period of 50-100 days
was required in order for the reactors to start producing
electrical current derived from living plants (Figure S1). The
time delay could be due to many reasons, such as the life
cycle dependency of the exudate release, both qualitatively
and quantitatively (2, 18), the omission of nutrients (which
can induce exudation) (4), the release of oxygen, conducted
through the aerenchyma (19), scavenging the electrons
otherwise collected at the anode, and the lack of an adapted
anodic microbial consortium. Temperature and light conditions
are an important factor, as the onset of high voltages
of the plant reactors coincided for most reactors with the
start of test period 2, with higher temperatures and higher
photosynthetic radiation. Higher phototrophic production
allows a higher exudate release, while high temperatures have
been reported to decrease the oxygen release by plants (20)
as well as increase the exudate release (2). The vermiculite
reactor without plants was not affected by the periods of
high ambient temperature. Temperature and photosynthetic
radiation were higher at the onset of the trials in 2007 than
at the onset in 2006, which could explain the earlier onset
of stable trends in 2007. The fluctuations in cell potential of
reactors with plants, determined by the anodic potentials
(as the cathodic potentials were relatively stable, Figure 3),
were presumably due to the presence of reduced, oxidizable
compounds around the anode. The release of the latter is
higher during the day, as demonstrated by Leake et al. (3).
A relatively higher concentration of reduced compounds
would result in a lower anodic potential and hence a higher
cell potential during the day. Photosynthesis thus determines
the cell potential of the fuel cells and the power that can be
extracted from these reactors.
The positive relationship between COD and current
production (Figure 4) demonstrates that the COD derived
from the plants (representing rhizodeposition) acts as an
electron donor for electricity generation. The presence of an
oxidizing anode leads to higher concentrations of COD
(Figure 4). The latter infers that the withdrawal of rhizodeposits
appears to stimulate a further excretion of reduced
substrates by the roots. This is in accordance with the findings
of Barber and Lynch (21), who found an increased exudation
in the presence of microorganisms, possibly through the
utilization of exudates. Attempts to characterize the substances
comprising the measured COD are warranted. The
rice plants studied in our reactors produced current during
thousands of hours of their plant life. A large part of this was
during an active growth period, suggesting that excreted
compounds, mucilage, and sloughed-off cells, etc. made up
the anodic substrate, aside from the oxidation of dead plant
material. The latter would notably occur at the end of the
reactor runs, when the aboveground biomass mainly consisted
of senescent leaves. The fact that current can be
obtained from the oxidation of decomposing root residues
is in accordance with the observation that root residues are
the major contributor to methane production (5).
Photosynthetic activity has been applied before to drive
the electricity generation in reactor type MFCs, involving
whole-cell photosynthetic microorganisms or subcellular
photosynthetic components and requiring either electrocatalyzed
anodes or the addition of redox mediators (22, 23).
This paper demonstrated for the first time that living higher
plants can provide an ongoing supply for electrical current
production in a sediment type of microbial fuel cell, without
the need for chemical catalysts or added redox mediators.
The anodic oxidation of organic compounds set free by plant
roots into the rhizosphere offers several environmental
perspectives. A rice paddy field or any vegetated wetland
VOL. xxx, NO. xx, XXXX / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 E

system could thus for instance give rise to a renewable
resource for direct electricity production by oxidizing a
substrate, which is conventionally transformed into methane.
A potential is offered for a sun-driven power generation in
(remote) areas with high solar inputs without the need for
scarce and costly construction materials such as silicon used
to fabricate photovoltaic cells. This nondestructive energy
production from biomass occurs without the need of costly
and/or energy consuming conversion techniques. Whereas
a typical sediment MFC is furthermore limited by diffusion
to the anode, a living plant can continuously deliver substrate
close to the anode, allowing continuous power production
and increasing the attainable production from a typical
nonplanted SMFC. Alternatively, this type of SMFC might be
applied to influence redox-dependent processes in the
rhizosphere. For example, through the presence of the
oxidizing anode, an electron acceptor for respiration pathways
is continuously available in the rhizosphere, allowing
a full oxidation of the plant substrates. The technique could
hence offer the prospect to mitigate the sediment redox
potential and abate undesirable processes such as methylation
of metals and emission of methane, which will be
investigated in subsequent experiments.
Acknowledgments
L.D.S. is supported through a Ph.D. grant from the Bijzonder
Onderzoeks Fonds of Ghent University (Grant No. 01D24405).
K.R. is supported by the UQ Postdoctoral Research Fellow
scheme and the ARC Discovery program. The supply of rice
seeds and advice on rice growth from David De Vleesschauwer
were greatly appreciated. The useful comments of Peter
Aelterman, Peter Clauwaert, Hai The Pham, Jorge Sanchez
Martinez, Michael Friedrich, and Angela Cabezas are kindly
acknowledged. The supply of a PAR sensor and climatic data
by the Laboratory of Plant Ecology and the Laboratory of
Wood Technology, Ghent University, was highly appreciated.
Supporting Information Available
Table S1 containing an overview of the rice reactor names
and configurations, Table S2 giving an overview of reactor
performances), Figure S1 showing cumulative electron
transfer, Figure S2 illustrating power ratios, Figure S3
picturing anodic substrate, and an extended version of the
energy considerations from Results and additional literature.
This material is available free of charge via the Internet at
http://pubs.acs.org.
Literature Cited
(1) Lu, Y. H.; Watanabe, A.; Kimura, M. Carbon dynamics of
rhizodeposits, root- and shoot-residues in a rice soil. Soil Biol.
Biochem. 2003, 35, 1223–1230.
(2) Grayston, S. J.; Vaughan, D.; Jones, D. Rhizosphere carbon flow
in trees, in comparison with annual plants: The importance of
root exudation and its impact on microbial activity and nutrient
availability. Appl. Soil Ecol. 1997, 5, 29–56.
(3) Leake, J. R.; Ostle, N. J.; Rangel-Castro, J. I.; Johnson, D. Carbon
fluxes from plants through soil organisms determined by field
13
CO 2 pulse-labelling in an upland grassland. Appl. Soil Ecol.
2006, 33, 152–175.
(4) Marschener, H. Role of root growth, arbuscular mycorrhiza,
and root exudates for the efficiency in nutrient acquisition. Field
Crops Res. 1998, 56, 203–207.
(5) Kimura, M.; Murase, J.; Lu, Y. H. Carbon cycling in rice field
ecosystems in the context of input, decomposition and translocation
of organic materials and the fates of their end products
(CO 2 and CH 4 ). Soil Biol. Biochem. 2004, 36, 1399–1416.
(6) Intergovernmental Panel on Climate Change. Climate Change
2001: The Scientific Basis. Available at http://www.grida.no/
climate.
(7) Tender, L. M.; Reimers, C. E.; Stecher, H. A.; Holmes, D. E.;
Bond, D. R.; Lowy, D. A.; Pilobello, K.; Fertig, S. J.; Lovley, D. R.
Harnessing microbially generated power on the seafloor. Nat.
Biotechnol. 2002, 20, 821–825.
(8) Rabaey, K.; Verstraete, W. Microbial fuel cells: novel biotechnology
for energy generation. Trends Biotechnol. 2005, 23, 291–
298.
(9) Bond, D. R.; Holmes, D. E.; Tender, L. M.; Lovley, D. R. Electrodereducing
microorganisms that harvest energy from marine
sediments. Science 2002, 295, 483–485.
(10) Ryckelynck, N.; Stecher, H. A.; Reimers, C. E. Understanding
the anodic mechanism of a seafloor fuel cell: Interactions
between geochemistry and microbial activity. Biogeochemistry
2005, 76, 113–139.
(11) Jones, J. B. A Guide for the Hydroponic and Soilless Culture
Grower; Timber Press: Portland, Oregon, 1983.
(12) Aelterman, P.; Rabaey, K.; Pham, H. T.; Boon, N.; Verstraete, W.
Continuous electricity generation at high voltages and currents
using stacked microbial fuel cells. Environ. Sci. Technol. 2006,
40, 3388–3394.
(13) Fisher, P. R.; Donnelly, C. S.; Faust, J. Evaluating supplemental
light for your greenhouse. Ohio Florist’s Assoc. Bull. 2001, 858,
4–7.
(14) Logan, B. E.; Hamelers, B.; Rozendal, R.; Schroder, U.; Keller,
J.; Freguia, S.; Aelterman, P.; Verstraete, W. Microbial fuel cells:
methodology and technology. Environ. Sci. Technol. 2006, 40,
5181–5192.
(15) Greenberg, A.; Clesceri, L. S.; Eaton, A. D. Standard Methods for
the Examination of Water and Wastewate; American Public
Health Association: Washington, D.C., 1992.
(16) Reimers, C. E.; Girguis, P.; Stecher, H. A.; Tender, L. M.;
Ryckelynck, N.; Whaling, P. Microbial fuel cell energy from an
ocean cold seep. Geobiology 2006, 4, 123–136.
(17) Holmes, D. E.; Bond, D. R.; O’Neill, R. A.; Reimers, C. E.; Tender,
L. R.; Lovley, D. R. Microbial communities associated with
electrodes harvesting electricity from a variety of aquatic
sediments. Microb. Ecol. 2004, 48, 178–190.
(18) Kerdchoechuen, O. Methane emission in four rice varieties as
related to sugars and organic acids of roots and root exudates
and biomass yield. Agric. Ecosyst. Environ. 2005, 108, 155–163.
(19) Neue, H. U.; Wassmann, R.; Lantin, R. S.; Alberto, M.; Aduna,
J. B.; Javellana, A. M. Factors affecting methane emission from
rice fields. Atmos. Environ. 1996, 30, 1751–1754.
(20) Waters, I.; Armstrong, W.; Thompson, C. J.; Setter, T. L.; Adkins,
S.; Gibbs, J.; Greenway, H. Diurnal changes in radial oxygen loss
and ethanol-metabolism in roots of submerged and nonsubmerged
rice seedlings. New Phytol. 1989, 113, 439–451.
(21) Barber, D. A.; Lynch, J. M. Microbial growth in the rhizosphere.
Soil Biol. Biochem. 1977, 9, 305–308.
(22) Rosenbaum, M.; Schröder, U.; Scholz, F. In situ electrooxidation
of photobiological hydrogen in a photobioelectrocemical fuel
cell based on Rhodobacter sphaeroides. Environ. Sci. Technol.
2005, 39, 6328–6333.
(23) Lam, K. B.; Johnson, E. A.; Chiao, M.; Lin, L. A MEMS
photosynthetic electrochemical cell powered by subcellular
plant photosystems. J. Microelectromech. Syst. 2006, 15,
1243–1250.
ES071938W
F 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. xxx, NO. xx, XXXX