Download Carbon Based Electrodes for Upscaling Microfluidic Fuel Cells

Proceedings of the ASME 2012 10th Fuel Cell Science, Engineering and Technology Conference
FuelCell2012
July 23-26, 2012, San Diego, CA, USA
FuelCell2012-91043
CARBON BASED ELECTRODES FOR UPSCALING MICROFLUIDIC FUEL CELLS
D. Fuerth
Mechanical and Industrial Engineering
University of Toronto
Toronto, ON, Canada
ABSTRACT
In this work, we present an experimental microfluidic
fuel cell with a novel up-scaled porous electrode architecture
that provides higher overall power output compared to
conventional microfluidic fuel cells and a methodology for
electrode material evaluation to inform designs for improved
performance. Our proof-of-concept architecture is an up-scaled
version of a previously presented flow-through cell with more
than nine times the active electrode surface area. We employed
0.04M formic acid and 0.01M potassium permanganate as fuel
and oxidant, respectively, dissolved in a 1M sulfuric acid
electrolyte. Platinum black was employed as the catalyst for
both anode and cathode. Carbon based porous electrodes
including felt, cloth, fibre, and foam were compared to
traditional Toray carbon paper in order to characterize their
respective performances. We also discussed current densities
normalized by electrode volume, which is appropriate for
comparison of flow-through architectures. The traditional
method of current normalization by projected electrode surface
area is also presented.
1.0 INTRODUCTION
With an increased emphasis towards miniaturization of
electronic devices and lab-on-a-chip technologies, increased
attention has been placed on high energy density conversion
devices. Microfluidic fuel cells (MFCs) pose a viable solution
to meet the rising energy demand due to their potential for high
energy densities and compact design [1]. Since the introduction
of MFCs in 2002 [2], many advances on their design and
performance have been made in an attempt to bring them closer
to commercialization [3]; however, there still remains limiting
factors. Low power densities, high ohmic resistance, mass
transport limitations, and scalability [1] remain important
challenges in the commercialization of MFCs.
Early MFC refinement focused on the Y-channel
A. Bazylak
Mechanical and Industrial Engineering
University of Toronto
Toronto, ON, Canada
architecture [2, 4-12], where current was normalized by the
projected active electrode surface area in which reactions
occurred. The performance of single Y-channel cells were
limited, due to diffusive mixing of fuel and oxidant leading to
crossover, fuel depletion layers across the electrode reaction
surfaces, low maximum power, and low fuel utilization.
A circularly designed MFC architecture presented by
Solloum et al. [13] incorporated flow through porous electrodes
that originated in the centre and travelled radially outward. The
cell was normalized by a projected electrode surface area
resulting in a power density of 2.8 mW cm-2 and fuel utilization
of 58%; however, the total surface area involved in the chemical
reactions was higher due to interior electrode surface area
contributions. This MFC demonstrated the benefits of porous
electrodes for high internal surface areas.
A significant advancement was made by Kjeang et al.
[14] when they presented a flow-through, porous electrode
MFC employing Toray carbon paper electrodes. The cell
housed independently fed fuel and oxidant through the anode
and cathode, respectively, before meeting in the centre channel
for ionic transport. The cell exhibited the highest reported
power density at high flow rates (120 mW cm-2 at 300 μL min-1)
and the highest reported fuel utilization at low flow rates (94%
at 1 μL min-1) when compared to previously presented porous
electrode architectures. This cell design addressed issues with
mass transport, fuel utilization, and power density, which were
typical limitations previously encountered in MFCs.
The flow-through porous electrode architecture
achieves large surface areas for electrochemical reactions, while
demonstrating high porosity for reactant and product transport.
Since microfluidic fuel cells are driven by surface based
reactions and are attractive due to their compact nature,
employing porous electrodes increases the surface area-tovolume ratio in a fuel cell. Previous works in flow-through
microfluidic fuel cell architectures have exclusively employed
Toray carbon paper as the porous anode and cathode material.
1
Copyright © 2012 by ASME
Alternative electrode materials have not been reported in flowthrough MFC literature. Material properties such as porosity
and tortuosity define the flow patterns within a porous material
that affect the fuel cell performance. Materials that allow fuel
and oxidant to effectively reach reaction sites will increase fuel
utilization and mitigate mass transport limitations.
In this work we present a proof-of-concept up-scaled,
flow-through microfluidic fuel cell architecture with over nine
times the projected active electrode surface area than originally
presented [14]. The up-scaled design is achieved by employing
larger electrodes, which facilitate a higher number of reaction
sites for energy conversion (compared to conventional MFC
counterparts), allowing for higher overall power and fuel
utilization for a single pass through the cell, while still
embodying the high ratios of surface area to volume within the
reactive pores that makes MFCs so attractive. Achieving more
power through a single cell reduces cell complexity and overall
volume of cell housing, which provides benefits when MFC
stacks are implemented. We employ novel electrodes including
carbon foams of various porosities, carbon fibre, and carbon
cloth in the up-scaled architecture to benchmark against the
previously utilized Toray carbon paper. Each electrode material
exhibits microstructured pores that affect the flow patterns and
chemical reactions within the cells. The larger electrode size
will create an emphasis on the electrode associated losses in the
cell and allow us to characterize their impact for potential
replacement of Toray carbon paper. Also presented is a novel
electrode volume normalization of the data allowing for an
exclusive comparison of flow-through architecture fuel cells in
addition to the traditional normalization based on projected
active electrode surface area. Electrode volume normalization
incorporates the projected active electrode surface area as well
as electrode height to establish the electrode volume. Since the
electrode volume has a large impact on the MFC architecture, a
reduced electrode volume would lead to a more dense overall
power density.
height of 300 μm. To promote laminar flow, reduce liquid
diffusion, and secure the electrode, the centre channel was
reduced to 150 μm. To accommodate larger electrodes, an
additional master was micromachined with an altered inlet
channel and electrode housing, both with a height of 2mm.
Polydimethylsiloxane (PDMS) (Sylgard 184 Silicone
Elastomer Kit, Dow Corning Corporation, Midland MI) was
poured over the master and fully cured at 75ᵒC for 2 hours to
create the top layer of the cell. Inlet and outlet holes were
punched in the top layer where 1/16 inch inner diameter tubing
was inserted for liquid interconnects. Two additional holes were
made over the dry region of the electrode housing
compartments of the top layer in preparation for electrical
connectivity. Electrodes were positioned in the electrode
housing compartments. A thin layer of PDMS, to be used as the
adhesion layer, was poured over a glass slide and cured at 75ᵒC
until it reached a partially cured state. The top layer was pressed
against the adhesion layer to create an irreversible seal.
The exposed dry regions of the electrodes were painted
with TIGA Silver 920H (Resin Technology Group L.L.C.,
South Easton, MA) to assist in the electrical connection to 23
Standard Wire Gauge (SWG) wire. The entire cell was baked at
100ᵒC for 24 hours to cure the silver. Epoxy was used over the
inlets, outlet, and electrical connections to achieve mechanical
stability and liquid sealing. The fabricated fuel cell is shown in
Figure 1.
2.0 EXPERIMENTAL
2.1 FUEL CELL FABRIACTION
The fuel cell presented is an up-scaled flow-through
architecture based on work previously presented by Kjeang et
al. [14] with a modified assembly and dimensions. Fuel and
oxidant separately flow into the cell from the two inlets. The
fuel and oxidant pass through the porous anode and cathode,
respectively, and remain laminar as they reach the outlet with
minimal diffusive mixing. The electrodes were extended to
include a dry region for electrical connectivity. The cell consists
of a top layer to house the channels and electrodes, an adhesion
layer to prevent leakage, and a glass slide layer for support.
A multi-height master was created using negative
photoresist (SU8, Micro-Chem, Newton, MA) and standard
photolithography techniques for the casting of the channels. The
inlet channels and electrode housings were microfabricated to a
Figure 1 - Photograph of a completed microfluidic fuel cell
from a) top view and b) bottom view.
2
Copyright © 2012 by ASME
2.2 CHEMICAL PREPARATION
Both fuel and oxidant were suspended in a 1M sulfuric
acid electrolyte solution created by diluting concentrate sulfuric
acid (University of Toronto Chemical Stores, Toronto, ON) with
de-ionized water. The fuel and oxidant used were 0.04M formic
acid (Formic Acid >95%, Sigma-Aldrich, St. Louis, MO) and
0.01M potassium permanganate (Fisher Scientific Group,
Ottawa, ON) respectively. The presence of the potassium
permanganate in the sulfuric acid exhibits a distinguishable
purple hue, while the formic acid suspended in sulfuric acid
appears as a clear liquid (Figure 2).
Figure 2 – Clear formic acid fuel (left) and purple
potassium permanganate oxidant (right) both suspended in
sulfuric acid electrolyte.
2.3 ELECTRODE PREPARATION
Electrodes were cut from bulk samples of carbon
paper, cloth, fibre, and foam to dimensions of 36mm in length
and 4mm in width. Both 80 and 100 pore per linear inch (PPI)
carbon foams were used, resulting in a total of five distinct
electrode types. The thicknesses of the carbon paper, cloth, and
fibre were acquired from a ten measurement average using a
Mastercraft electronic caliper with an accuracy of 20 μm. Due
to the large pore size of the foam, the electrodes were cut to a
minimal thickness that still allowed for structural stability by an
electrical discharge machine (EDM).
An ink catalyst was prepared using the recipe
presented by M. S. Wilson and S. Gottesfeld [15]. Nafion (5%
wt., Sigma-Aldrich, St. Louis, MO), 20% Pt on carbon (Alfa
Aesar, Ward Hill, MA), deionized water, and glycerol (SigmaAldrich, St. Louis, MO) were mixed to a ratio of 1:3:20:80,
respectively. The solution was mixed ultrasonically to create a
homogeneous solution with fully suspended particles.
Electrodes were immersed in the catalytic ink solution and dried
at 135ᵒC for 8 hours.
Catalyst loadings for each electrode, normalized by the
projected surface area of the electrode, varied due to electrode
volume and available surface area for coating. The carbon fibre
electrode had a small volume and extremely dense structure
limiting the catalyst ink penetration in comparison to the other
electrodes leading to the lowest catalyst loading of 1.4 mg Pt
cm-2. The carbon cloth and Toray carbon paper electrodes had
catalyst loadings of 3.4 mg Pt cm-2 and 3.9 mg Pt cm-2,
respectively, due to a greater electrode volume. The cloth
electrode structure is less dense with less available surface area
for coating when compared to carbon paper leading to a lower
catalyst loading even though the volume is greater. The 100 PPI
and 80 PPI carbon foams had the largest catalyst loadings of 5.0
mg Pt cm-2 and 6.1 mg Pt cm-2, respectively, due to their large
volumes. The 80 PPI carbon foam had a higher catalyst loading
due to a higher density which led to more available surface
areas for coating, when compared to the 100 PPI carbon foam.
2.4 FUEL CELL DIAGNOSTICS
Prior to operation, the cells were flushed with deionized water to encourage flow paths for fuel and oxidant to
follow and to prevent initial fuel and oxidant cross-over
contamination. Fuel and oxidant were supplied to the cell at a
rate of 100 μL min-1 using a Harvard Pump Plus Dual Syringe
(Harvard Apparatus, Holliston, MA). Data was acquired using a
VersaSTAT 3 potentiostat (Princeton Applied Research, Oak
Ridge, TN) with VersaStudio software.
Current densities for MFCs are traditionally
normalized with respect to the exposed surface area of the
electrode. In flow-over fuel cells [2, 4-12] this would be the
projected surface area of the electrode in which fluid flows over
and chemical reactions take place. In the flow-through
architecture chemical reactions also take place in the interior
surface area of the electrode. In this architecture the electrode
thickness along with the surface area will impact the
performance of the cell. Current densities normalized by both
the projected electrode surface area, as well as the electrode
volume are used to present the data in this study for
comparative purposes. A summary of the values used to
normalize the data is presented in Table 1.
Thickness
(cm)
Area
(cm2)
Volume
(cm3)
Carbon Foam (80 PPI)
0.200
1.12
0.22400
Carbon Foam (100 PPI)
0.200
1.12
0.22400
Carbon Cloth
0.055
1.12
0.06160
Material
Toray Carbon Paper
0.038
1.12
0.04256
Carbon Fibre
0.024
1.12
0.02688
Table 1 - Values used to normalize current data for each
carbon material.
3.0 RESULTS
The OCV of each cell was measured after allowing the
cell to operate for 15 minutes (break-in procedure), until steady
3
Copyright © 2012 by ASME
state was reached. The OCV values were recorded using the
open circuit setting in the VersaStudio software. The
potentiostatic setting in the VersaStudio software held the
voltage of the MFCs constant while corresponding current
readings were recorded. This data was separately normalized by
the projected electrode surface area and electrode volume to
create polarization and power density curves presented in
Figure 3 for a comparative evaluation of the MFCs. The OCV,
peak power voltage, lowest voltage, maximum current density,
and maximum power density for each MFC is summarized in
Table 2.
Figure 3 - Polarization curves normalized by a) projected electrode surface area and b) electrode volume and power density
curves normalized by c) projected electrode surface area and d) electrode volume for 80 PPI carbon foam ( ), 100 PPI carbon
foam ( ), fibre ( ), carbon paper ( ), and carbon cloth ( ).
Material
OCV
(V)
Voltage at
Maximum
Power Density
(V)
Maximum
Current
Density
(mA cm-2)
Maximum
Current
Density
(mA cm-3)
Maximum
Power Density
(mW cm-2)
Maximum
Power Density
(mW cm-3)
Carbon Foam (80 PPI)
1.02
0.50
0.7731
3.865
0.3651
1.826
Carbon Foam (100 PPI)
0.80
0.40
0.8941
4.392
0.2851
1.426
Carbon Cloth
0.57
0.35
0.1867
3.394
0.0383
0.697
Toray Carbon Paper
0.98
0.60
0.4421
11.635
0.2562
6.742
0.60
0.4102
17.091
0.2016
8.402
Carbon Fibre
1.04
Table 2 - Summary of electrode performance.
4
Copyright © 2012 by ASME
4.0 DISCUSSION
4.1 OPEN CIRCUIT VOLTAGE
A comparison between the OCVs reported in Table 2
show a variation between the cells. The OCVs of the five cells
are expected to be identical, as each cell is using formic acid
and potassium permanganate as the fuel and oxidant,
respectively, and each cell follows the same fabrication
methodology. Lower OCV values were reported for the carbon
cloth cell and the 100 PPI carbon foam cell. The lower OCV
values reported from these cells is attributed to the crossover
contamination of the fuel and oxidant. A visual inspection of the
flow patterns from the distinct colors of the fuel and oxidant
during operation revealed inconsistencies between the five cells.
The carbon cloth cell exhibited a high concentration of fuel
entering the centre channel close to the outlet, leaving the
remainder of the centre channel dominated by oxidant. The
oxidant was able to reach the anode causing severe crossover.
The 100 PPI carbon foam cell was limited by the centre
channel. The centre channel was on average, 75% occupied
with an air pocket, with the remaining centre channel containing
a gradient of fluid rather than a distinction between the two
liquids. Mitigation of these irregular flows is currently being
investigated.
4.2 DATA TREND TERMINATION
A time threshold of 5 minutes at each voltage was used
for raw data collection to allow for steady current readings. At
low currents the data stabilized well below the time threshold.
At high currents, the data failed to stabilize which lead to
inaccurate readings. When the time threshold was reached due
to cell instability, the cell was stopped. Each cell reached the
threshold at a different current. This is evident from the
polarization and power density curves (Figure 3) terminating at
different locations.
Repeated experiments with individual cells could not
be conducted due to internal unintended chemical reactions
occurring close to the electrical interconnects. The instability of
the cell at high currents was attributed to this deterioration.
Figure 4 displays an image of a cell after operation with
deteriorated interconnects. Modification of cell construction is
currently underway to improve the seal of the electrical
interconnects.
4.3 POLARIZATION LOSSES
The polarization curves displayed in Figures 3 a and b
are both applicable to the present discussion since losses
associated with the cells were independent of the area/volume
normalization process. The dominant linear decrease in voltage
with increasing current was mainly attributed to ohmic losses in
each cell. This was the only dominant loss that was present in
the 80 PPI carbon foam and Toray carbon paper cell. The term
dominant does not indicate that other losses were not present in
cell operation, but that ohmic losses are the main influence
based on the trends presented.
Figure 4 - Electrode deterioration causing instability in the
MFCs performance.
Since three cells (Toray carbon paper, carbon fibre,
and 80 PPI carbon foam) met a significantly early termination
due to instability from deterioration, the mass transport losses
regions were not reached for these cells. The final data point in
the Toray carbon paper trend appears to have been influenced
by a mass transport limitation, but is more likely attributed to
unstable data acquisition near the cell threshold. The MFCs
containing fibre and cloth electrodes appear to have
experienced activation losses noted by the rapid decrease in
performance before reaching the region dominated by ohmic
losses. Mass transport losses were present in the cell containing
100 PPI carbon foam, but did not make a significant impact on
the cell. The drop in voltage was not drastic which indicates the
cell had not reached the current limit.
4.4
CELL
COMPARISON
BY
PROJECTED
ELECTRODE AREA NORMALIZATION
The maximum power density acquired was 0.365 mW
cm-2 at 0.5 V in the MFC containing 80 PPI carbon foam. Cells
containing 100 PPI carbon foam, Toray carbon paper, carbon
fibre, and carbon cloth achieved 78.1%, 70.1%, 55.3%, and
10.4% of the power density obtained by the 80 PPI foam
sample.
The data normalized by the projected electrode surface
area served as a good comparison to the overall total power
produced by each cell. Since the normalizing area was constant,
power density was scaled equally with total power output for
each MFC. The data concluded that for a cell containing 80 PPI
carbon foam, the greatest total power output will be achieved.
The data reported under this normalization allows for
benchmarking with previous work; however, there is no
literature which utilizes the flow-through architecture paired
with the fuel and oxidant employed in this work.
4.5 CELL COMARISON BY ELECTRODE VOLUME
NORMALIZATION
The maximum power density acquired was 8.40 mW
cm-3 at 0.6 V in the MFC containing carbon fibre. Cells
5
Copyright © 2012 by ASME
containing Toray carbon paper, 80 PPI carbon foam, 100 PPI
carbon foam, and carbon cloth achieved 80.2%, 21.7%, 16.9%,
and 8.3% of the power density obtained by the fibre sample.
The data normalized by electrode volume incorporates
the impact of electrode thickness on the cell performance. The
electrode thickness governs the cell housing dimensions,
leading to an overall reduction in MFC volume. The cells
containing carbon fibre and Toray carbon paper electrodes
greatly outperformed the other cells due to their thin
manufactured bulk sample. Since other researchers do not
report flow-through architectures in an electrode volume power
density manner, it is difficult to classify the performance of the
cells with this method.
5.0 CONCLUSION
Presented was a proof-of-concept microfluidic fuel cell
architecture which demonstrated the ability to upscale MFCs for
larger total power output in a single. The enlarged electrode
material allowed for the impact of their properties to enhance
the cell performance. The enhanced impact allowed for a
comparative study into four novel carbon based electrode
materials against the traditionally employed Toray carbon paper.
These materials included 100 PPI carbon foam, 80 PPI carbon
foam, carbon cloth, and carbon fibre.
On a projected electrode surface area normalization,
the cell containing 80 PPI carbon foam produced the highest
power density with a value of 0.365 mW cm-2 at a flow rate of
0.1 μL min-1 and voltage of 0.55V. Since the projected electrode
surface area was identical for each cell, this cell also produced
the highest overall power density of the electrode materials.
With electrode volume normalization, the cell containing
carbon fibre produced the highest power density with a value of
8.40 mW cm-3 at a flow rate of 0.1 μL min-3 and voltage of
0.6V. The volumetric based normalization accounted for each
electrode thickness which influenced the overall space required
to house the electrodes and cell. This novel method of
normalization hopes to set precedent for future work to provide
an additional comparative measure.
Current work is underway to control flow patterns
within the operating cell to decrease fuel crossover. This will
lead to a higher overall open circuit voltage value and a steadier
operation. Sealing techniques around the electrode
interconnects and anti-deterioration methods are also under
investigation. Validation of the results will be acquired with
repeatable future experiments that take into account these
solutions.
ACKNOWLEDGMENTS
The authors would like to acknowledge the guiding
hand of the Microscale Energy Systems Transport Phenomena
Laboratory for their collective assistance and encouragement.
Financial support to the authors from The Natural Sciences and
Engineering Research Council of Canada (NSERC), David
Sanborn Scott and Ron D. Venter Graduate Fellowship, Bullitt
Foundation, Canada Foundation for Innovation (CFI), and
University of Toronto are also gratefully acknowledged.
REFERENCES
[1] Kjeang, E., Djilali, N., and Sinton, D., 2009, "Microfluidic
Fuel Cells: A Review," Journal of Power Sources, 186(2) pp.
353-369.
[2] Ferrigno, R., Stroock, A., Clark, T., 2003, "Membraneless
Vanadium Redox Fuel Cell using Laminar Flow (Vol 124, Pg
12930, 2002) RID C-1299-2010," Journal of the American
Chemical Society, 125(7) pp. 2014-2014.
[3] Mousavi Shaegh, S. A., Nguyen, N., and Chan, S. H., 2011,
"A Review on Membraneless Laminar Flow-Based Fuel Cells,"
International Journal of Hydrogen Energy, 36(9) pp. 56755694.
[4] Phirani, J., and Basu, S., 2008, "Analyses of Fuel Utilization
in Microfluidic Fuel Cell," Journal of Power Sources, 175(1)
pp. 261-265.
[5] Sprague, I. B., Byun, D., and Dutta, P., 2010, "Effects of
Reactant Crossover and Electrode Dimensions on the
Performance of a Microfluidic Based Laminar Flow Fuel Cell,"
Electrochimica Acta, 55(28) pp. 8579-8589.
[6] Kjeang, E., Proctor, B. T., Brolo, A. G., 2007, "HighPerformance Microfluidic Vanadium Redox Fuel Cell,"
Electrochimica Acta, 52(15) pp. 4942-4946.
[7] Bazylak, A., Sinton, D., and Djilali, N., 2005, "Improved
Fuel Utilization in Microfluidic Fuel Cells: A Computational
Study," Journal of Power Sources, 143(1–2) pp. 57-66.
[8] Ebrahimi Khabbazi, A., Richards, A. J., and Hoorfar, M.,
2010, "Numerical Study of the Effect of the Channel and
Electrode Geometry on the Performance of Microfluidic Fuel
Cells," Journal of Power Sources, 195(24) pp. 8141-8151.
[9] Choban, E. R., Markoski, L. J., Wieckowski, A., 2004,
"Microfluidic Fuel Cell Based on Laminar Flow," Journal of
Power Sources, 128(1) pp. 54-60.
[10] Choban, E. R., Spendelow, J. S., Gancs, L., 2005,
"Membraneless Laminar Flow-Based Micro Fuel Cells
Operating in Alkaline, Acidic, and acidic/alkaline Media,"
Electrochimica Acta, 50(27) pp. 5390-5398.
[11] Jayashree, R. S., Yoon, S. K., Brushett, F. R., 2010, "On
the Performance of Membraneless Laminar Flow-Based Fuel
Cells," Journal of Power Sources, 195(11) pp. 3569-3578.
[12] Yoon, S. K., Fichtl, G. W., and Kenis, P. J. A., 2006,
"Active Control of the Depletion Boundary Layers in
Microfluidic Electrochemical Reactors," Lab on a Chip, 6(1)
pp. 1516-1524.
[13] Salloum, K. S., Hayes, J. R., Friesen, C. A., 2008,
"Sequential Flow Membraneless Microfluidic Fuel Cell with
Porous Electrodes," Journal of Power Sources, 180(1) pp. 243252.
[14] Kjeang, E., Michel, R., Harrington, D. A., 2008, "A
Microfluidic Fuel Cell with Flow-through Porous Electrodes,"
Journal of the American Chemical Society, 130(1) pp. 40004006.
[15] Wilson, M. S., and Gottesfeld, S., 1992, "Thin-Film
Catalyst Layers for Polymer Electrolyte Fuel Cell Electrodes,"
Journal of Applied Electrochemistry, 22(1) pp. 1-7.
6
Copyright © 2012 by ASME