An alkaline microfluidic fuel cell based on formate and hypochlorite bleach

Article abstract of DOI:10.1016/j.electacta.2008.07.009

An alkaline microfluidic fuel cell is demonstrated employing an alkaline version of a formic acid anode and a sodium hypochlorite cathode. Both sodium formate fuel and sodium hypochlorite oxidant are available and stable as highly concentrated solutions, thereby facilitating fuel cell systems with high overall energy density. Sodium hypochlorite is commonly available as hypochlorite bleach. The alkaline anodic half-cell produces carbonate rather than the less-desirable gaseous CO₂, while sustaining the rapid kinetics associated with formic acid oxidation in acidic media. Both half-cells provide high current densities at relatively low overpotentials and are free of gaseous products that may otherwise limit microfluidic fuel cell performance. The microfluidic fuel cell takes advantage of a recently developed membraneless architecture with flow-through porous electrodes. Power densities up to 52 mW cm^-2 and overall energy conversion efficiencies up to 30% per single pass are demonstrated at room temperature using 1.2 M formate fuel and 0.67 M hypochlorite oxidant. The alkaline formate/hypochlorite fuel and oxidant combination demonstrated here, or either one of its individual half-cells, may also be useful in conventional membrane-based fuel cell designs.

Full text of DOI:10.1016/j.electacta.2008.07.009

Electrochimica Acta 54 (2008) 698–705
Contents lists available at ScienceDirect
Electrochimica Acta
journal homepage: www.elsevier.com/locate/electacta
An alkaline microfluidic fuel cell based on formate and hypochlorite bleach
Erik Kjeanga,c, Raphaelle Michela,c, David A. Harringtonb,c, David Sintona,c,∗, Ned Djilalia,c
a Department of Mechanical Engineering, University of Victoria, 3800 Finnerty Road, Victoria, BC V8W 2Y2, Canada
b Department of Chemistry, University of Victoria, 3800 Finnerty Road, Victoria, BC V8W 2Y2, Canada
c Institute for Integrated Energy Systems (IESVic), University of Victoria, 3800 Finnerty Road, Victoria, BC V8W 2Y2, Canada
a r t i c l e i n f o
a b s t r a c t
Article history:
An alkaline microfluidic fuel cell is demonstrated employing an alkaline version of a formic acid anode and
a sodium hypochlorite cathode. Both sodium formate fuel and sodium hypochlorite oxidant are available
and stable as highly concentrated solutions, thereby facilitating fuel cell systems with high overall energy
density. Sodium hypochlorite is commonly available as hypochlorite bleach. The alkaline anodic half-cell
produces carbonate rather than the less-desirable gaseous CO2, while sustaining the rapid kinetics associ-
ated with formic acid oxidation in acidic media. Both half-cells provide high current densities at relatively
low overpotentials and are free of gaseous products that may otherwise limit microfluidic fuel cell per-
formance. The microfluidic fuel cell takes advantage of a recently developed membraneless architecture
with flow-through porous electrodes. Power densities up to 52 mW cm−2 and overall energy conversion
efficiencies up to 30% per single pass are demonstrated at room temperature using 1.2 M formate fuel and
0.67 M hypochlorite oxidant. The alkaline formate/hypochlorite fuel and oxidant combination demon-
strated here, or either one of its individual half-cells, may also be useful in conventional membrane-based
fuel cell designs.
Received 5 February 2008
Received in revised form 30 June 2008
Accepted 1 July 2008
Available online 12 July 2008
Keywords:
Microfluidic fuel cell
Alkaline fuel cell
Formate
Hypochlorite bleach
Flow-through porous electrode
© 2008 Elsevier Ltd. All rights reserved.
1. Introduction
microchannel; fuel and/or oxidant crossover can be mitigated by
adjusting the flow rate of the co-laminar streams; no ion exchange
A microfluidic fuel cell is defined as a device that incorporates all
fundamental components of a fuel cell within a single microfluidic
channel and its walls. Since its invention in 2002, microfluidic fuel
cell technology has developed rapidly [1–22] and is now considered
a candidate for commercial small-scale portable power generation.
Microfluidic fuel cells typically operate without a membrane, and
the most common configurations rely on the laminar nature of
flow in microstructures to maintain sufficient separation of fuel
and oxidant streams, flowing side-by-side in a co-laminar format.
Supporting electrolyte contained in both co-laminar streams facili-
tates ionic conduction between the electrodes. Mixing by diffusion
is restricted to an interfacial width at the center of the channel,
depending on mean velocity and channel geometry. The electrode
spacing is typically an order of magnitude larger than the inter-
diffusion width. Microfluidic fuel cells have several advantages as
compared to traditional proton exchange membrane (PEM)-based
fuel cells: fuel and oxidant streams may be combined in a single
membrane is required; sealing, manifolding, and fluid delivery
infrastructure requirements are reduced; and issues related to
membrane hydration and water management are eliminated. In
addition, microfluidic fuel cells may be manufactured using inex-
pensive microfabrication methods and low-cost materials.
Microfluidic fuel cells with room-temperature power densities
on the order of 100 mW cm−2 have been demonstrated [15,16].
Operation at high power densities has, however, resulted in low
overall energy conversion efficiency due to low single-pass fuel
utilization. To mitigate this problem, a new microfluidic fuel cell
architecture with flow-through porous electrodes was recently
developed by our group [15]. The flow-through design is based
on cross-flow of reactant through the porous electrodes into a co-
laminar center channel with orthogonally directed flow in which
the waste solutions facilitate ionic charge transfer in a mem-
braneless configuration. This cell architecture enables improved
utilization of the three-dimensional active area inside the porous
electrodes and provides enhanced rates of convective/diffusive
transport to the active sites without increasing the parasitic loss
required to drive the flow. When operated with 2 M vanadium solu-
tions, the flow-through design demonstrated a high level of overall
energy conversion efficiency, as a relatively high level of fuel uti-
lization and cell voltage were achieved concurrently. With respect
∗
Corresponding author at: Department of Mechanical Engineering, and the Insti-
tute for Integrated Energy Systems (IESVic), University of Victoria, 3800 Finnerty
Road, Victoria, V8W 2Y2, Canada.
E-mail address: dsinton@me.uvic.ca (D. Sinton).
0013-4686/$ – see front matter © 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.electacta.2008.07.009

E. Kjeang et al. / Electrochimica Acta 54 (2008) 698–705
699
to overall energy density, the solubility and concentration of ions
is a limitation of microfluidic fuel cell systems based on vanadium
redox species; in addition, the associated redox reactions provide
only one electron per ion.
porous electrodes are formed via electrodeposition on a porous
carbon substrate. The electrochemical responses of the porous elec-
trodes are characterized in a stationary three-electrode cell prior to
in situ tests and fuel cell implementation.
In the context of fuel and oxidant combinations for microflu-
idic fuel cell systems with practical energy density, the following
attributes are desirable: both reactants and products are available
and stable at high concentration in the liquid phase; the reactants
provide at least two electrons per molecule; spontaneous and/or
electrochemically activated decomposition into gaseous products
are prevented; the fuel and oxidant do not react upon mixing; and if
the anodic and cathodic supporting electrolytes are different, both
fuel and oxidant species are stable in both electrolytes.
Several different fuels have been employed in microfluidic fuel
cells presented to date. These include hydrogen [7,18,19], methanol
[5,6,11], formic acid [4,8,12,13], vanadium redox species [9,14–16],
and hydrogen peroxide [10]. Most liquid hydrocarbon fuels produce
carbon dioxide (CO2) as the end product after complete electro-
chemical conversion. Typically, CO2 evolves as a gaseous product
in acidic or neutral media [13] unless the current densities are low
enough to fully dissolve it in the fuel or electrolyte stream [4].
Formic acid is an established fuel with high energy density that
has previously been used in both PEM-based direct formic acid fuel
cells [23,24] and microfluidic fuel cells [4,8,12,13], and its electro-
chemical kinetics on palladium catalyst are fast, thereby allowing
much higher power densities than other hydrocarbon fuels such
as methanol [25]. In contrast to acidic and neutral media, alkaline
media have the capability of absorbing large amounts of CO2 as
carbonates with the additional benefit of automated carbon seques-
tration. An alkaline formate fuel from concentrated formic acid
therefore accomplishes all the criteria for an all-liquid microfluidic
fuel cell system with high energy density.
2. Experimental
2.1. Preparation of solutions
Alkaline formate fuel solution was obtained by adding 5 wt.%
concentrated formic acid (HCOOH; Fisher Sci., Fair Lawn, NJ) to a
10 wt.% (2.8 M) sodium hydroxide (NaOH) electrolyte prepared by
dissolving NaOH pellets (EMD Chemicals, San Diego, CA) in Milli-
pore Milli-Q water (Millipore, Billerica, MA). The resulting solution
contained 1.2 M formate and 1.6 M OH−. Alkaline hypochlorite
(ClO−) oxidant solution (0.67 M) was prepared by dissolving 10 wt.%
(2.8 M) NaOH pellets directly in a 5% sodium hypochlorite solu-
tion (NaOCl; Caledon, Georgetown, Canada) or 5.25% hypochlorite
bleach (Javex-5; Colgate-Palmolive, Toronto, Canada). The gold
(Au) plating solution consisted of 17 mM potassium gold cyanide
(KAu(CN)2; Degussa, Frankfurt, Germany) dissolved in Millipore
Milli-Q water, followed by the successive addition of 0.4 M citric
acid (C6H8O7; Matheson Coleman & Bell, Montreal, Canada) and
0.4 M sodium citrate (Na3C6H5O7; Merck, Montreal, Canada) to
a final pH of 3–5. The palladium (Pd) plating solution consisted
of 1.0 wt.% palladium(II) chloride (PdCl2; Sigma–Aldrich) in 1 M
hydrochloric acid (HCl; Anachemia, Montreal, Canada) in Milli-
pore Milli-Q water. All chemicals used were ACS grade except the
hypochlorite bleach.
2.2. Electrodeposition
The oxidants employed in microfluidic fuel cells to date have
been oxygen [4–8,18,19], air [11,12], vanadium redox species
[9,14–16], and hydrogen peroxide [10,13]. Among these, only hydro-
gen peroxide is available as a highly concentrated liquid. Direct
hydrogen peroxide reduction on common catalysts such as plat-
inum and palladium is however frequently accompanied by oxygen
gas evolution from its decomposition [13]. Alternative liquid oxi-
dants that have been employed in fuel cells or flow batteries
include permanganate [4,26] and hypochlorite [27,28]. Perman-
ganate is a strong oxidant that enables high current densities and
highly positive reduction potentials in acidic media [29]. Reports
have shown, however, that product manganese dioxide precipi-
tate can clog the pores of a microporous electrode in the absence
of stirring [26]. The use of permanganate oxidant in microflu-
idic fuel cells that incorporate both microporous electrodes and
microscale channels should therefore be avoided. Hypochlorite, in
contrast, does not produce any precipitate, and has shown benign
electrochemical characteristics as an oxidant in magnesium, zinc
and aluminum semi-fuel cells [27,28] using palladium/iridium and
platinum catalysts. Hypochlorite oxidant has many useful charac-
teristics for microfluidic fuel cell implementation: relatively high
standard reduction potential and rapid kinetics; high solubility in
aqueous media; highly soluble reduction product (chloride); and
finally, no gaseous decomposition reactions take place when the
hypochlorite is used in an alkaline electrolyte [27]. Furthermore,
hypochlorite solution is produced in large quantities as domestic
bleach. Sodium hypochlorite bleach is considered relatively safe
and is available at very low cost. Hypochlorite therefore meets all
the criteria of an all-liquid microfluidic fuel cell with high energy
density. Additionally, the product sodium chloride is common and
non-toxic.
Porous carbon electrodes subject to catalyst deposition were cut
to 25 mm × 1 mm strips from sheets of Toray carbon paper (B-2
Designation TGPH-090; E-TEK, Somerset, NJ) with typical thickness
300 ␮m (measured), typical density 0.49 g cm−3 and 78% porosity. A
20 mm long section of the carbon strips were rendered hydrophilic
by annealing (∼1 s) in a propane flame, and the hydrophilic elec-
trodes were then supported on glass slides with wires attached
using PELCO® conductive silver 187 (Ted Pella Inc., Redding, CA)
covered by regular epoxy. The strip electrodes were rinsed thor-
oughly in Millipore Milli-Q water before and after each plating
sequence. Electrodeposition of Au and Pd on the porous carbon
electrodes was performed by immersing the hydrophilic electrode
parts in a bath containing either Au or Pd plating solution, together
with a Pt mesh counter electrode and a saturated calomel refer-
ence electrode (SCE). The carbon working electrode was aligned
with the counter electrode to obtain a uniform current distribu-
tion during plating. All electrochemical experiments in this study
were driven by a PARSTAT 2263 potentiostat (Princeton Applied
Research, Oak Ridge, TN), and all individual electrode potentials are
given versus the SCE (sat. KCl) reference electrode (0.241 V vs. SHE).
Each electrodeposition experiment started with a cyclic voltam-
mogram (3 cycles) of a bare porous carbon electrode in the Au
plating solution from the open-circuit potential (OCP) to −1.1 V,
followed by electroplating under potentiostatic control at −1.1 V
for 60 min. The Au loading of the obtained porous gold electrodes
was not assessed. Porous Pd electrodes were obtained by electrode-
position of Pd on the fresh Au layer by a similar procedure [13]: a
cyclic voltammogram (3 cycles) was measured with the porous Au
electrode in the Pd plating solution from the OCP to −0.1 V, fol-
lowed by electroplating under potentiostatic control at 0.0 V until
a desired loading of 5 mg cm−2 had been deposited, typically after
20–30 min, calculated in real time by integrating the current pro-
file and assuming a 60% coulombic plating efficiency [30]. Scanning
In this work, a microfluidic fuel cell based on alkaline formate
and hypochlorite bleach is presented. Both palladium and gold

700
E. Kjeang et al. / Electrochimica Acta 54 (2008) 698–705
electron micrographs were captured by a Hitachi S-3500N scanning
electron microscope with a tungsten filament operated at 15 kV.
0.5 Hz with 30 log spaced data points. The ohmic resistance was
calculated as the high-frequency real axis intercept of the Nyquist
plot of impedance. For in situ EIS measurements, the flow rate was
2.3. Fuel cell fabrication
fixed at 60 ␮L min−1
.
A collection of microfluidic fuel cells with flow-through porous
electrodes was assembled using in-house developed microfabri-
cation techniques detailed in a previous study [15]. In short, one
20 mm long porous Pd anode and one 20 mm long porous Au cath-
ode were cut from the electrodeposited strip electrodes and fitted
in custom-sized grooves (20 mm × 1 mm × 300 ␮m, separated by
1 mm), fabricated in a poly(dimethylsiloxane) (PDMS; Dow Corn-
ing, Midland, MI) part. Of the 20 mm length of electrode, only
12 mm was exposed to the solutions; the remaining 8 mm allowed
for electrical contact and sufficient separation between the contact
and the electrolyte. The PDMS part had a dual layer structure: a
300 ␮m deep section that defined the two electrode grooves plus
the two inlet channel structures, and a 150 ␮m deep layer accom-
modating the co-laminar flow channel between the electrodes,
leading towards the outlet. The fourth wall of the microfluidic
device was provided by a flat layer of PDMS, with previously
punched holes for the inlets, outlet, and electrical contacts, which
was sealed irreversibly on top of the other PDMS part after plasma-
treating. External wiring was contacted to the exposed ends of the
carbon strip electrodes using PELCO® conductive silver 187 covered
with regular epoxy.
3. Results and discussion
3.1. Reaction scheme
At the pH used here, the fuel exists predominantly as formate
(HCOO−), the oxidant as hypochlorite (ClO−) and the CO2 as car-
bonate. The reactions, written in terms of these species, lead to the
following standard electrode potentials (vs. SCE) at 298 K ((1) cal-
culated from Gibbs’ energies of formation in [31] and (2) taken from
[32]):
Anode:
HCOO− + 3OH− ꢀ CO32− + 2H2O + 2e−, E0 = −1.17 V
Cathode:
(1)
(2)
(3)
ClO− + H2O + 2e− ꢀ Cl− + 2OH−, E0 = 0.57 V
Overall cell reaction:
HCOO− + ClO− + OH− ꢀ CO32− + Cl− + H2O, Ec0ell=1.74 V
Note that the actual half-cell and cell reversible potentials also
depend on pH and concentration of the aqueous species according
to the Nernst equation. The cell reaction involves some net con-
sumption of OH−, which is available at high concentration in the
alkaline electrolyte.
2.4. Fuel cell testing
The co-laminar flow of formate and hypochlorite solutions
through the fuel cell was driven by a syringe pump (PHD 2000;
Harvard Apparatus, Holliston, MA) via Teflon® tubing (1/16 in.
diameter; S.P.E. Limited, North York, ON, Canada) to the inlets and
through a larger Tygon® tube (3 mm diameter; Fisher Sci., Pitts-
burgh, PA) from the outlet to the outlet reservoir. The desired flow
rate was adjusted and the co-laminar flow was stabilized before ini-
tiation of the electrochemical measurements. Cell polarization data
were measured by chronoamperometry under stepwise potentio-
static control from the open-circuit cell voltage down to 0.0 V by
−0.2 V increments, using the PARSTAT potentiostat. The current
transient generated by the cell was monitored until steady state
was reached, which typically occurred within 180 s. Polarization of
individual electrodes was measured versus an external SCE refer-
ence electrode placed in the outlet reservoir, using the opposite fuel
cell electrode as counter electrode.
The definition of an area with which to characterize the perfor-
mance of flow-through cells is somewhat ambiguous due to the
three-dimensional architecture. The porous electrodes employed
here were 1 mm wide, 300 ␮m deep, and a length of 12 mm was
exposed to the solutions. Utilization of the high internal surface
area (estimated at 3 cm2) is a feature of this fuel cell architec-
ture. To make meaningful performance comparison with previous
microfluidic fuel cells, however, current densities and power densi-
ties reported here were calculated based on the vertically projected
area of the electrodes (1 mm × 12 mm = 0.12 cm2). This is the geo-
metric area in the fluidic plane of the fuel cell, and is significantly
larger than the geometric area in the flow-normal direction. As
noted previously [15], power-per-volume, while unconventional,
is ultimately a more pertinent performance metric from a device
perspective.
3.2. Electrodeposition of porous electrodes
To reduce the activation overpotentials of the proposed electro-
chemical reaction scheme, suitable catalysts need to be added to
the electrodes. In this study, palladium (Pd) and gold (Au) were
selected as the catalysts for formate oxidation and hypochlorite
reduction, respectively. Pd and Au were electrodeposited on porous
carbon paper substrates in order to develop porous Pd and Au elec-
trodes compatible with the unique co-laminar flow mechanism of
the microfluidic fuel cell with flow-through porous electrodes [15].
Coating of the carbon fibers with Au may also be advantageous
towards reducing the overall ohmic resistance of the electrodes,
which was relatively high for the bare porous carbon electrodes
used previously with this cell architecture [15]. Fig. 1a shows a
typical cyclic voltammogram of a bare porous carbon electrode
in the gold cyanide plating bath. The overpotential for Au plating
is relatively high; plating effectively started at −0.6 V, which was
about 800 mV negative of the open-circuit potential (OCP). Hydro-
gen evolution started at about −1.0 V on the first scan and at −0.8 V
on consecutive scans. The difference between the three cycles is
associated with the fresh Au deposit on the carbon electrode that
reduced the overpotential for hydrogen evolution, which is rela-
tively high on bare carbon. Au plating below −0.8 V was therefore
accompanied by hydrogen evolution. The viability of potentiostatic
Au plating on porous electrodes was evaluated at three different
potentials: −0.7 V, −0.9 V and −1.1 V, with results shown in Fig. 1b.
The plating current at −0.7 V was essentially free from hydrogen
evolution and therefore quite low. Although not readily apparent on
the scale of the plot in Fig. 1b, the current showed a linear increase
from 0.15 to 0.45 mA/cm2 (over the 1 h test period), attributed to
the nucleation and growth of Au grains. The plating currents mea-
sured at −0.9 V and −1.1 V were dominated by hydrogen evolution,
which also increased over time as the surface area of the Au deposit
The ohmic resistance of the fuel cell and its individual elec-
trodes was measured by electrochemical impedance spectroscopy
(EIS), using the PARSTAT 2263 potentiostat. Impedance spectra
were recorded at the open-circuit cell voltage, by applying an ac
amplitude of 5 mV rms over the frequency range from 50 kHz to

E. Kjeang et al. / Electrochimica Acta 54 (2008) 698–705
701
observable fibers throughout the 3-D porous matrix showed effec-
tively complete coverage (Fig. 3b); and the surface morphology had
dendritic features with size <1 ␮m indicating a high specific surface
area (Fig. 3c). The improved 3-D coverage at high overpotentials
is attributed to the continuous hydrogen gas evolution introduc-
ing a local two-phase flow component. Specifically, as hydrogen
bubbles were released from the porous matrix, fresh plating solu-
tion was introduced into the matrix. Although not evaluated here,
convection-assisted plating by stirring or ultrasonic vibration is
expected to have a similar effect. In sum, electrodeposition of Au on
porous carbon electrodes was shown to be very effective at −1.1 V.
Fig. 1. Electrodeposition of gold on porous carbon electrodes: (a) cyclic voltammo-
gram of a bare substrate in gold cyanide plating solution measured at 50 mV s−1, and
(b) potentiostatic deposition curves of 1 h duration at three different potentials (as
indicated vs. SCE).
was enlarged. Images of the final Au deposits captured by scanning
electron microscopy (SEM) are presented in Fig. 2. These images
provide a useful illustration of the microstructure and degree of
coverage achieved in the deposition of Au on the porous carbon
electrodes. With respect to the electrochemical performance of a
porous electrode, both high coverage and high specific surface area
are desirable characteristics. Fig. 2a shows the granular Au structure
obtained at −0.7 V. At −0.9 V (Fig. 2b), a similar granular structure
was observed with significantly higher loading. The image of the
deposit obtained at −1.1 V (Fig. 2c) indicates a high degree of cov-
erage with different surface morphology. Most importantly, the Au
deposit appeared much more uniform over the observable depth
of the porous electrode at −1.1 V compared to the other poten-
tials.
The porous Au electrode deposited at −1.1 V was further ana-
lyzed with images obtained at three different magnifications,
shown in Fig. 3. These images demonstrate the quality of the
obtained Au deposit in the context of fuel cell implementation: the
outer surfaces of the electrode were uniformly coated (Fig. 3a); all
Fig. 2. Scanning electron micrographs showing the microstructure of the porous
gold electrodes electrodeposited on carbon paper substrates at (a) −0.7 V, (b) −0.9 V
and (c) −1.1 V (SCE), captured at 2000× magnification.

702
E. Kjeang et al. / Electrochimica Acta 54 (2008) 698–705
implementation. Cyclic voltammograms measured ex situ for a
porous Pd anode, a porous Au cathode, and a porous Pd cathode
are shown in Fig. 4. The Pd anode produced high current densities
up to 340 mA cm−2 at quite low overpotentials, in the absence of
convective motion. As expected, no gas evolution was observed in
this anodic half-cell, indicating that the product is carbonate, with-
out any CO2. The OCP of this half-cell was measured to −0.99 V,
within 0.2 V of the standard potential of reaction (1). (The reversible
potential cannot be calculated because the product concentrations
are unknown.) The ohmic resistance of the Pd anode was mea-
sured as 3.6 ꢀ by electrochemical impedance spectroscopy (EIS),
as described in Section 3.4.
The Au and Pd cathodes also delivered notably high current
densities up to 230 and 330 mA cm−2, respectively, despite the
lack of convective transport. The Au cathode, however, exhibited a
250 mV higher overpotential than the Pd cathode for the hypochlo-
rite reduction reaction (2). As shown, Au is not an ideal catalyst
for hypochlorite reduction, but it does have a significant advantage
with respect to integration in microfluidic fuel cells: in contrast
to the Pd electrode that produced a small amount of gas during
these tests, no gas evolution was observed on the Au electrode.
The OCPs for hypochlorite reduction were measured as 0.38 V (Au)
and 0.46 V (Pd). The oxidation of water to oxygen can occur above
0.38 V at pH 14.4, the nominal pH of this solution, explaining the
gas evolution in the case of Pd. The measured OCP was again within
0.2 V of the standard reduction potential (0.57 V). While not ideal
for catalysis of this reaction, the alkaline electrolyte is required to
stabilize the hypochlorite ion in solution. The ohmic resistance of
the cathodes was measured by EIS as 3.0 ꢀ (Au) and 2.9 ꢀ (Pd). The
electrical resistance of the Au coated electrodes was thus reduced
by more than 50% compared to the bare carbon strip electrodes
(7–9 ꢀ). Overall, these half-cell experiments indicate that the for-
mate anodic half-cell and the hypochlorite cathodic half-cell are
well-suited to implementation in a fuel cell.
An additional half-cell test was performed with the ACS-grade
hypochlorite solution replaced with a low-cost household bleach
product (Javex-5; Colgate-Palmolive), containing 5.25% sodium
hypochlorite. The obtained cyclic voltammograms closely matched
those in Fig. 4.
Fig. 3. Scanning electron micrographs of a porous gold electrode electrodeposited
at −1.1 V (SCE), captured at (a) 60×, (b) 800× and (c) 6000× magnification.
Similarly fabricated porous Au electrodes were also used to make
porous Pd electrodes, by coating an additional Pd layer (5 mg cm−2
)
following our previously described Pd electrodeposition method-
ology [13], with details outlined in Section 3.2.
3.3. Ex situ half-cell characterizations
Fig. 4. Cyclic voltammograms measured ex situ for formate oxidation on a porous
palladium anode and hypochlorite reduction on porous gold and palladium cathodes
in a stationary three-electrode electrochemical cell. The anolyte and catholyte con-
sisted of 1.2 M formate and 0.67 M hypochlorite, respectively, in alkaline solution.
The porous Pd anodes and Au cathodes were characterized elec-
trochemically in a stationary three-electrode electrochemical cell
in the alkaline formate or hypochlorite solutions, respectively, to
analyze their performance and suitability for microfluidic fuel cell
The scan rate was 50 mV s−1
.

E. Kjeang et al. / Electrochimica Acta 54 (2008) 698–705
703
Table 1
3.4. Microfluidic fuel cell implementation
Selected resistances measured by electrochemical impedance spectroscopy (EIS): ex
situ electrode resistances measured in a stationary three-electrode electrochemical
cell, and in situ electrode and cell resistances measured in the fuel cell assembly
during operation at 60 ␮L min−1 flow rate
A microfluidic fuel cell with flow-through porous electrode
architecture was fabricated, incorporating an electrodeposited
porous Au electrode as cathode and an electrodeposited porous
Pd/Au electrode as anode. A schematic of the microfluidic fuel cell
architecture is given in Fig. 5a, and the assembled cell is shown in
Fig. 5b and c. The transparency of the PDMS enables observation of
the flow, and detection of any intermittent disturbances. As gaseous
products can destabilize the co-laminar flow [13], the Au cathode
was chosen over the Pd cathode (that produced a small amount of
gas in the ex situ experiments) for implementation in the fuel cell.
The fuel and oxidant streams enter the microfluidic fuel cell device
(Fig. 5) through separate inlets and approach the electrodes via the
deep (300 ␮m) distribution channels. From the distribution chan-
nels, the flow is directed orthogonally into the porous electrodes.
The cross-flow through the porous media has two orders of magni-
tude higher pressure drop than the flow in the distribution channels
[15], and is therefore expected to be uniformly distributed through-
out the available cross-sectional area of the electrodes. In addition,
the high cross-sectional area open to the cross-flow provides low
mean velocities inside the porous electrodes, which enables ample
opportunity for high coulombic fuel utilization. Both anodic and
cathodic streams enter the center channel through the electrodes
on both sides at the same flow rate. The two streams are directed
towards the outlet in a co-laminar format. The center channel is
relatively shallow (150 ␮m) and has a small cross-sectional area in
Resistance [ꢀ]
Ex situ
In situ
Anode (Pd)
Cathode (Au)
Cell
3.6
3.0
–
7.0
12.8
24.1
order to provide high mean velocities and reliable co-laminar flow.
In this way mixing of the waste streams is limited to a relatively
small diffusion width in the centre, while the electrolyte provides
effective ionic transport. The flow-through electrode architecture
thus achieves utilization of the full depth of the electrode and asso-
ciated active area, which is 20–30 times larger than the vertically
projected area of the electrode [16], and provides enhanced species
transport from the bulk to the active sites as well as effective ionic
transport in a membraneless configuration.
3.5. Microfluidic fuel cell performance
The assembled microfluidic fuel cell device with flow-through
porous electrodes was operated at flow rates ranging from 2 to
300 ␮L min−1 per stream, using formate anolyte and hypochlorite
catholyte. The cell exhibited very stable co-laminar flow without
any gas evolution or other disturbances and the open-circuit cell
potential was steady between 1.37 and 1.42 V. The OCP increased
somewhat with flow rate, perhaps because the OCP at the cath-
ode is a mixed potential with a component of oxygen evolution
and is more easily perturbed than a true reversible potential. These
cell potential levels are similar to the estimated cell potential from
the ex situ experiments (1.37 V), calculated as the difference in OCP
between the Pd anode and the Au cathode. The combined ohmic
resistance of the cell was measured using EIS at open circuit and
60 ␮L min−1 as 24.1 ꢀ. This resistance measurement includes the
solution resistance with respect to ionic transport in the co-laminar
center channel and the electrical resistance in electrodes, contacts
and wires. By incorporation of an external reference electrode (SCE)
in the outlet reservoir, we were also able to measure the ohmic
resistance of the individual electrodes in situ using the same tech-
nique, resulting in 7.0 ꢀ for the anode and 12.8 ꢀ for the cathode. A
summary of the resistance measurements is provided in Table 1.
The in situ values are somewhat higher than the corresponding
ex situ measurements (3.6 ꢀ and 3.0 ꢀ, respectively), primarily
attributed to high electrical contact resistance on the chip. The solu-
tion resistance was estimated to 4.3 ꢀ by subtracting the electrical
resistance from the combined ohmic cell resistance, which is in
agreement with previous studies using acidic electrolyte [14].
Room-temperature polarization data were obtained under
potentiostatic control at steady state conditions. Fig. 6 presents
three different polarization curves obtained at 2, 10 and
60 ␮L min−1. The current densities increased with flow rate up
to a maximum 230 mA cm−2 at 60 ␮L min−1, indicating partial
mass transport control and relatively fast electrochemical kinetics;
however, increasing the flow rate above 60 ␮L min−1 (polariza-
tion curves not shown here) resulted in no significant performance
improvements. Basedon theseobservations, electrochemical kinet-
ics and ohmic resistance were dominating at high flow rates.
The parasitic combined ohmic resistance of the cell contributed
almost 50% to the average slope of the polarization curve (51 ꢀ) at
60 ␮L min−1, and about 70% to the linear part of the curve at practi-
cal cell voltages of 0.8 V and below. At low flow rates, the potential
loss associated with ohmic resistance was not as severe: it was less
than 30% at 10 ␮L min−1 and only about 8% at 2 ␮L min−1. At these
Fig. 5. The microfluidic fuel cell with flow-through porous electrodes: (a) schematic
of the fuel cell architecture in cross-section; (b) image of the assembled fuel cell
with porous palladium and gold electrodes with microchannel flow arrows super-
imposed; and (c) labeled image of the assembled cell.

704
E. Kjeang et al. / Electrochimica Acta 54 (2008) 698–705
Fig. 6. Fuel cell polarization curves measured experimentally at steady state and
room temperature, for operation of the formate/hypochlorite microfluidic fuel cell
at three different flow rates (as indicated).
Fig. 8. Power density curves for formate/hypochlorite fuel cell operation at steady
state and room temperature, measured at three different flow rates (as indicated).
and cathodic polarization curves are surprisingly well-balanced.
The symmetry outlined by the two curves justifies the viability of
the formate/hypochlorite fuel and oxidant combination and also
confirms the uniform transport characteristics of the flow-through
architecture. Both electrodes responded well to fuel cell operation
at cell voltages below 1.0 V, where neither of them caused an over-
all restriction to the cell. The activation overpotential was shared
by the two half-cells but dominated by the cathode. These results
are in good agreement with the ex situ measurements, where the
activation overpotential was quite high on the Au cathode. The flow-
through fuel cell architecture was endurance tested previously [15].
The performance over 10 h was very stable, with no observable
loss in performance over that period. Similarly the alkaline cell
tested here showed no detectable degradation over the course of
this study.
flow rates, the current densities were primarily controlled by the
rate of reactant supply. The cell was not operated below 2 ␮L min−1
,
although the stability of the co-laminar flow is expected to be main-
tained at least down to 0.3 ␮L min−1 [15].
The relatively low current densities produced in the high cell
voltage range (1.0–1.4 V) indicate significant activation overpoten-
tial for the electrochemical reactions, which is common to most
fuel cell systems. This effect was further analyzed by employing
the external reference electrode (SCE) to measure the performance
of the individual electrodes in situ under the same flow condi-
tions, using the opposite electrode as counter electrode. Separate
anodic and cathodic polarization curves are presented in Fig. 7.
This type of analysis is often used to diagnose overall fuel cell per-
formance and to identify half-cell-specific limitations. Given the
significant differences related to the electrochemistry of the for-
mate oxidation and hypochlorite reduction reactions, the anodic
Power density curves calculated based on the polarization data
at 2, 10 and 60 ␮L min−1 are provided in Fig. 8. The overall power
density increased significantly with flow rate, which again con-
firms the partial mass transport control. The highest power density
measured in this study was 52 mW cm−2 at 0.4 V and 60 ␮L min−1
.
This level of power density compares favorably to most other
microfluidic fuel cells reported to date and demonstrates the high
capacity of the proposed formate/hypochlorite fuel and oxidant
combination. The power densities measured here did not how-
ever reach the levels obtained with the vanadium redox system
in the same cell architecture [15] due to the lower cell voltages and
reactant concentrations. The feasibility of the formate/hypochlorite
system in the flow-through electrode architecture is more evi-
dent in the low flow rate regime. Even at the lowest flow rate
(2 ␮L min−1), a useful peak power density of 18 mW cm−2 was gen-
erated at 0.6 V cell voltage. Fuel cell operation in this regime is
very efficient due to the high levels of fuel utilization enabled
by the flow-through electrode architecture and practical cell volt-
ages achieved concurrently. The fuel utilization with respect to
the hypochlorite component, which had a lower molar concen-
tration (0.67 M) than the formate solution (1.2 M), was as high
as 85% at the point of peak power and effectively 100% at low
cell voltages. Based on the theoretical standard cell potential of
1.74 V, the overall energy conversion efficiency at 0.6 V was ∼30%
per single pass. The highly energy efficient microfluidic fuel cell
performance previously demonstrated with the vanadium redox
Fig. 7. Fuel cell polarization data of anode and cathode measured in situ at room
temperature and 60 ␮L min−1 flow rate, using the opposite fuel cell electrode as
counter electrode and an external reference electrode (SCE).

E. Kjeang et al. / Electrochimica Acta 54 (2008) 698–705
705
system was thus maintained with the new formate/hypochlorite
system.
ide electrolyte and optimizing its concentration, and evaluating
other liquid hydrocarbon fuels such as methanol. The new alka-
line formate and hypochlorite fuel cell concept demonstrated here,
or either one of its individual half-cells, may also find applications
using conventional membrane-based fuel cell designs.
The fuel cell performance measured here can also be compared
to the ex situ half-cell characterizations provided by the cyclic
voltammograms in Fig. 4. In contrast to the steady state fuel cell
data, these predictions are based on unsteady operation (50 mV s−1
scan rate) without flow, and the ohmic resistance was only about
3 ꢀ per electrode. A fuel cell based on the performance measured
in the ex situ experiments would theoretically have an open-circuit
voltage of 1.37 V (Au cathode) or 1.45 V (Pd cathode), and would
produce peak power densities of 81 mW cm−2 (Au cathode) and
143 mW cm−2 (Pd cathode) at 0.35 V and 0.55 V cell voltage, respec-
tively. These power density levels are much higher than those
obtained in the microfluidic fuel cell, despite the lack of convective
transport. This comparison indicates that the high capacity inher-
ent to the formate/hypochlorite system was not fully exploited in
the proof-of-concept microfluidic fuel cell, primarily due to its high
parasitic ohmic resistance. While electrode contamination from the
carbonate species is a possibility at high fuel concentrations, no
detectable deterioration was observed in these tests.
Acknowledgments
The authors would like to acknowledge project funding by
the Natural Sciences and Engineering Research Council of Canada
(NSERC) and Angstrom Power Inc., and infrastructure funding
from Canada Foundation for Innovation (CFI) and British Columbia
Knowledge Development Fund (BCKDF).
References
[1] A. Bazylak, D. Sinton, N. Djilali, Journal of Power Sources 143 (2005) 57.
[2] M.H. Chang, F. Chen, N.S. Fang, Journal of Power Sources 159 (2006) 810.
[3] F.L. Chen, M.H. Chang, M.K. Lin, Electrochimica Acta 52 (2007) 2506.
[4] E.R. Choban, L.J. Markoski, A. Wieckowski, P.J.A. Kenis, Journal of Power Sources
128 (2004) 54.
[5] E.R. Choban, J.S. Spendelow, L. Gancs, A. Wieckowski, P.J.A. Kenis, Electrochimica
Acta 50 (2005) 5390.
Finally, we also assembled a prototype microfluidic fuel cell with
a Pd cathode, to evaluate the performance gains compared to the
cell with a Au cathode, envisaged by the ex situ measurements. The
cell generated an initial peak power density of 59 mW cm−2 (at a
flow rate of 60 ␮L min−1), 13% higher than the previous cell, but the
gas bubbles growing on the Pd cathode made its operation unsta-
ble shortly thereafter. To enable reliable fuel cell operation with
a Pd cathode, some design modifications need to be undertaken.
Potentially a grooved channel geometry for gas removal could be
incorporated, as previously demonstrated [13]. An alternative strat-
egy is to evaluate other catalysts for the hypochlorite reduction that
do not produce gas. Both platinum [27] and a combination of palla-
dium and iridium [28] have been used effectively for hypochlorite
reduction but their application in microfluidic fuel cells needs to
be assessed.
[6] E.R. Choban, P. Waszczuk, P.J.A. Kenis, Electrochemical and Solid State Letters 8
(2005) A348.
[7] J.L. Cohen, D.J. Volpe, D.A. Westly, A. Pechenik, H.D. Abruna, Langmuir 21 (2005)
3544.
[8] J.L. Cohen, D.A. Westly, A. Pechenik, H.D. Abruna, Journal of Power Sources 139
(2005) 96.
[9] R. Ferrigno, A.D. Stroock, T.D. Clark, M. Mayer, G.M. Whitesides, Journal of the
American Chemical Society 124 (2002) 12930.
[10] S. Hasegawa, K. Shimotani, K. Kishi, H. Watanabe, Electrochemical and Solid
State Letters 8 (2005) A119.
[11] R.S. Jayashree, D. Egas, J.S. Spendelow, D. Natarajan, L.J. Markoski, P.J.A. Kenis,
Electrochemical and Solid State Letters 9 (2006) A252.
[12] R.S. Jayashree, L. Gancs, E.R. Choban, A. Primak, D. Natarajan, L.J. Markoski, P.J.A.
Kenis, Journal of the American Chemical Society 127 (2005) 16758.
[13] E. Kjeang, A.G. Brolo, D.A. Harrington, N. Djilali, D. Sinton, Journal of the Elec-
trochemical Society 154 (2007) B1220.
[14] E. Kjeang, J. McKechnie, D. Sinton, N. Djilali, Journal of Power Sources 168 (2007)
379.
[15] E. Kjeang, R. Michel, D.A. Harrington, N. Djilali, D. Sinton, Journal of the Amer-
ican Chemical Society 130 (2008) 4000.
[16] E. Kjeang, B.T. Proctor, A.G. Brolo, D.A. Harrington, N. Djilali, D. Sinton, Elec-
trochimica Acta 52 (2007) 4942.
4. Conclusions
[17] E. Kjeang, D. Sinton, D.A. Harrington, Journal of Power Sources 158 (2006) 1.
[18] S.M. Mitrovski, L.C.C. Elliott, R.G. Nuzzo, Langmuir 20 (2004) 6974.
[19] S.M. Mitrovski, R.G. Nuzzo, Lab on a Chip 6 (2006) 353.
[20] K.G. Lim, G.T.R. Palmore, Biosensors & Bioelectronics 22 (2007) 941.
[21] C.M. Moore, S.D. Minteer, R.S. Martin, Lab on a Chip 5 (2005) 218.
[22] M. Togo, A. Takamura, T. Asai, H. Kaji, M. Nishizawa, Electrochimica Acta 52
(2007) 4669.
This work demonstrated the feasibility of an alkaline formate
anode coupled with an alkaline hypochlorite cathode in a microflu-
idic fuel cell architecture with flow-through porous electrodes. Both
half-cells were unique in terms of fuel cell implementation. In con-
trast to the vanadium redox system, formate and hypochlorite are
both available in highly concentrated solutions, thereby enabling a
microfluidic fuel cell system with high overall energy density. The
proof-of-concept formate/hypochlorite microfluidic fuel cell con-
currently achieved high power densities and near-complete fuel
utilization at relatively high cell voltages, and therefore enabled
high overall energy conversion efficiency. The performance would
be enhanced further by reducing the combined ohmic resistance of
the cell, specifically the on-chip electrical contact resistance, and by
performing a catalyst screening study for hypochlorite reduction to
reduce the activation overpotential and enhance the electrochemi-
cal kinetics while preventing gas evolution. Other advancements
under investigation include employing higher fuel and oxidant
concentrations in the fuel cell, switching to potassium hydrox-
[23] A. Kundu, J.H. Jang, J.H. Gil, C.R. Jung, H.R. Lee, S.H. Kim, B. Ku, Y.S. Oh, Journal
of Power Sources 170 (2007) 67.
[24] J.D. Morse, International Journal of Energy Research 31 (2007) 576.
[25] Y.M. Zhu, Z. Khan, R.I. Masel, Journal of Power Sources 139 (2005) 15.
[26] S. Licht, Electrochemistry Communications 1 (1999) 33.
[27] A.M. Cardenas-Valencia, C.J. Biver, L. Langebrake, Journal of Power Sources 166
(2007) 273.
[28] M.G. Medeiros, E.G. Dow, Journal of Power Sources 80 (1999) 78.
[29] S.J. You, Q.L. Zhao, J.N. Zhang, J.Q. Jiang, S.Q. Zhao, Journal of Power Sources 162
(2006) 1409.
[30] J.J. Whalen, J.D. Weiland, P.C. Searson, Journal of the Electrochemical Society
152 (2005) C738.
[31] D.D. Wagman, W.H. Evans, V.B. Parker, R.H. Schumm, I. Halow, S.M. Bailey, K.L.
Churney, R.L. Nuttall, Journal of Physical and Chemical Reference Data 11 (Suppl.
(2)) (1982).
[32] D.R. Lide, CRC Handbook of Chemistry and Physics, CRC Press, Boca Raton, FL,
2002.