You are on page 1of 8

Journal of Power Sources Advances 3 (2020) 100012

Contents lists available at ScienceDirect

Journal of Power Sources Advances


journal homepage: www.journals.elsevier.com/journal-of-power-sources-advances

Cathode starvation as an accelerated conditioning procedure for


perfluorosulfonic acid ionomer fuel cells
Emmanuel Balogun a, Alejandro Oyarce Barnett b, c, Steven Holdcroft a, *
a
Holdcroft Research Group, Department of Chemistry, Simon Fraser University, 8888 University Drive, Burnaby, BC V5A 1S6, Canada
b
SINTEF Industry, New Energy Solutions, Sem Sælands Vei 12, Trondheim, 7034, Norway
c
Department of Energy and Process Engineering, Norwegian University of Science and Technology, NO-7491, Trondheim, Norway

A R T I C L E I N F O A B S T R A C T

Keywords: Freshly assembled proton exchange fuel cells (PEMFC) require conditioning to reach maximum power density.
PEMFC This process may last up to tens of hours and adds to the cost of commercial fuel cell technology. We present an
Fuel cell conditioning accelerated conditioning procedure involving starving the cathode of oxidant. In single cells, this procedure
Perfluorosulfonic acid ionomer
conditions a membrane electrode assembly (MEA) within 40 min, without compromising durability. The per-
Accelerated degradation
formance and durability of MEAs conditioned using this technique are compared with US Department of Energy
(DOE) and European Union (EU) harmonized protocols, and to an amperometric conditioning protocol. The time
to reach peak power density using cathode starvation conditioning is <10% of the time required for DOE, EU, and
amperometric protocols. Conditioned MEAs were subjected to accelerated degradation by cycling the cell voltage
between 0.6 V and open-circuit voltage under low relative humidity. Degradation was found to be caused by loss
of electrochemical surface area of the cathode, which in turn increases the charge transfer resistance of the MEA.
MEAs conditioned using cathode starvation experienced only a 15% loss in performance; in contrast to 19, 17 and
17% losses in performance caused by the DOE, EU, and amperometric protocols, respectively.

1. Introduction to several hours or even days, and requires lengthy deployment of


expensive test benches and consumption of hydrogen fuel. Conditioning
Fuel cell technology has great potential in the portable, transportation increases the cost of PEMFC technology and bottlenecks in their mass
and stationary sectors. Of these, polymer electrolyte membrane fuel cells manufacture and deployment. One way to reduce PEMFC manufacturing
(PEMFC) have proven potential as zero-emission energy conversion de- cost is to establish new and efficient conditioning procedures to minimize
vices due to their rapid refueling, compactness, low noise, and scalability the time required for activation of MEAs without compromising
[1]. However, the technology could benefit from improvements in maximum power density and fuel cell durability.
durability and a reduction in materials cost [2]. While improvements in Several reports present ex-situ and in-situ techniques to maximize
the overall PEMFC performance have been realized, there is still a need to power density of the PEMFC within the shortest time possible. Off-line
implement other cost mitigating strategies to be able to meet the US conditioning procedures generally involve taking steps to activate the
Department of Energy (DOE) target of $30/kW by 2030 [2,3]. cell before operation. Qi et al. [6], and Anderson et al. [8] report a
There is a need to condition the fuel cell prior to its intended use conditioning procedure that entails the steam treatment of the MEA.
(referred to as initial conditioning, activation or break-in). Conditioning Palanichamy et al. [9], proposed a method that involves physically dis-
hydrates the membrane and catalyst layers of the membrane-electrode- engaging impurities from the electrode surface through a process of
assembly (MEA) that may dry out during manufacturing [4]. The phys- electrochemical treatment. In-situ conditioning procedures typically
ical processes behind conditioning are not fully understood, but it is entail the use of current control [10–12], voltage control [13,14] or both
assumed that during the conditioning process the number of active [13], and are by far the most commonly applied conditioning protocols.
catalyst sites is increased, the polymer membrane and ionomer in the Other in-situ conditioning protocols include short-circuiting [15], CO
catalyst layer are humidified, improving protonic pathways through the stripping [16], temperature control [17], air braking [18], and thermal
MEA [5–7]. The initial conditioning procedure can be lengthy, taking up cycling (for Pt/NSTF catalyst layers [19,20]. Kato et al. [21] proposed

* Corresponding author.
E-mail address: holdcrof@sfu.ca (S. Holdcroft).

https://doi.org/10.1016/j.powera.2020.100012
Received 3 April 2020; Received in revised form 7 May 2020; Accepted 12 May 2020
2666-2485/© 2020 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
E. Balogun et al. Journal of Power Sources Advances 3 (2020) 100012

methods of activating single cells and stacks using series of cyclic po- recommended by Taghiabadi et al. [28].
tential sweeps. Similarly, Choo et al. [22] put forward a method of
accelerated activation for fuel cell stacks by applying high current for a 2. Experimental
designated time. Although, there is no standard measurement to verify
the effectiveness of a particular PEMFC conditioning protocol, there are 2.1. MEA preparation and fuel cell operation
recommendations on ways to compare conditioning efficiencies using
maximum output power and polarization curve performance [23]. Much Catalyst ink consisting of a homogeneous mixture of Pt/C catalyst
information on conditioning commercial fuel cells is often proprietary, powder (TEC10E50E, lot 109–0111, 46.4% Pt), Nafion PFSA polymer
inaccessible to the public, and specific to each fuel cell company tech- dispersion D520 (Ion Power Inc., lot SGA-12-02CS) and methanol (re-
nology platform. agent grade, Fischer Scientific) was spray-coated onto Nafion NR-N211
In an attempt to standardize and accelerate the conditioning process, membrane (DuPont, TE143904) using a spray coater (Sono-tak Exacta-
the US Department of Energy (DOE) proposed a method of conditioning Coat). The Pt loading on both anode and cathode electrodes was
which entails applying a cyclic potential between the OCV and 0.55 V 0.4 mg Pt/cm2. The catalyst coated membrane (CCM) was sandwiched
[24]. Similarly, the European Union (EU) fuel cell community proposed a between two 5 cm2 (2.24  2.24 cm) PTFE treated gas diffusion layers
harmonized test protocol that involves a series of relative humidity-, (Sigracet 29BC, FuelCellStore) and compressed in fuel cell hardware
stepwise current-, and potential-cycles [25]. These standardized pro- (AHNS Co.) using a torque wrench, with fuel cell compression deter-
tocols helped uniformity of MEA performance but require in excess of 7 h mined by pressure-sensitive paper (5.7 Nm).
and 11 h, respectively. There have emerged other accelerated condi- The cell hardware was connected to a fuel cell test station (Scribner
tioning procedures that reportedly activate MEAs in less than 1 h. One 890CL). MEAs were conditioned using our novel cathode starvation
such protocol involves a short-circuiting procedure [15], which yields procedure and compared to the standardized DOE conditioning proced-
good I–V performance but durability is compromised due to increased ure, EU harmonized conditioning procedure, and an amperometric pro-
degradation rates caused by corrosion of the catalyst carbon support [26, cedure. The fuel cell temperature was set at 80  C and 100% relative
27]. A steam-treatment conditioning protocol [6,8] is reported to acti- humidity (RH) at both electrodes. The inlet gas flows were 0.5 slpm
vate MEAs within 1 h but MEA performance and durability are hydrogen at the anode and 1.0 slpm oxygen at the cathode.
compromised compared to other protocols [26]. In a recent report by 12 MEAs samples were prepared under identical conditions and
Taghiabadi et al. [28], an accelerated stress test (AST) was introduced to divided into four sets of 3 for evaluation using 4 different conditioning
investigate the effect of different current- and potential-based condi- procedures. Each of the conditioning procedures described below was
tioning procedures on PEMFC durability. It was shown that a constant repeated three times on individually prepared MEAs. For each of the
current conditioning protocol may require a longer time to activate the three tests carried out using the same conditioning procedure, the per-
MEA and causes rapid degradation of the catalyst layer. formance and degradation profile obtained were found reproducible,
In addition to considerations of FC conditioning, the mitigation and/ showing the same trend. Further information on how the accuracy of
or recovery of reversible (transient) degradation of PEMFCs to ensure comparison is ensured is included in the SI (Fig. S1- S6, Table S5).
their prolonged durability are important areas of research. Several
methods of recovering transient degradation have been proposed, 2.2. Conditioning procedures
ranging from overnight rest, various shutdown procedures, and contin-
uous in-situ testing. In a recent report on FC shutdown procedures, Pivac The conditioning procedures investigated in this study are listed
and Barbir [29] illustrated that a shorter soak time following the shut- below:
down of a PEMFC results in higher performance when the FC is rejuve-
nated. Oyarce et al. [30], compared the effects of different shutdown (a) Cathode starvation conditioning protocol. The cell is operated
procedures on PEMFC degradation and found two methods, (i) H2 purge under H2/O2 configuration and the backpressure is increased from
of the cathode, and (ii) O2 consumption under load were most effective, ambient to 1.5 bar. The current is held constant as the cathode gas
resulting in the least amount of degradation. Holmstr€ om et al. [31], feed is switched off, then turned back on as the voltage reaches
subsequently applied it as a start-up procedure. There have been studies 0.2 V. This process is repeated in cycles until the difference be-
on the impact of reactant starvation on PEMFC performance. Gerad et al. tween the last two cycles is < 5 mV.
[32] had earlier studied the effect of oxygen starvation on PEMFC (b) Standardized DOE conditioning procedure: The voltage is
durability, Colbow et al. [33] proposed a technique of PEMFC operation switched between OCV (held for 1 s) and 0.55 V (held for 1 h)
using periodic reactant starvation. Reiser et al. [34] and Gould et al. [35] until the difference in current (recorded immediately after
introduced methods of cyclic oxidant starvation for PEMFC performance returning to 0.55 V) between 2 successive cycles is < 50 mA
recovery. (10 mA/cm2).
The accelerated conditioning protocol used in this study is modified (c) Standardized EU conditioning protocol: (i) The power density
from an approach by Eickes et al. [18] to condition DMFCs. In this study, response is monitored as the current density is ramped from 0.5 A
we report for the first time an in-depth analysis of an accelerated cathode (100 mA/cm2) to 4A (800 mA/cm2) and allowed to stabilize for
starvation conditioning procedure for PFSA-based MEAs, contrasting its 6 h. (ii) The voltage response is monitored at a constant current of
performance and durability to established state-of-the-art conditioning 5 A (1000 mA/cm2) for 2 h at 50% RH. (iii) The voltage is sub-
procedures. The cathode starvation protocol involves starving the cath- sequently cycled between 0.75 V and 0.45 V until the difference in
ode of oxidant while allowing H2 to diffuse from the anode to cathode. the current response between the last 2 steps is < 5 mA (1 mA/
MEAs conditioned using this cathodic starvation protocol are favorably cm2).
compared with MEAs conditioned using the standardized DOE, EU and (d) In-house, amperometric conditioning protocol. The current is
an in-house amperometric protocol which is similar to those conditioning linearly increased from 2 A (400 mA/cm2) to 15 A (3000 mA/cm2)
procedures reported by Bi et al. [11] and Shan et al. [12]. The time for over an 18 h period. The procedure is repeated until attaining
conditioning is determined by observing the time taken to reach peak optimal power density performance (i.e., the difference between
power density, and the current density of the cell at 0.6 V, as recom- the last 2 current scans is < 50 mA (10 mA/cm2)).
mended by Murphy et al. [36] The durability of MEAs is investigated by
comparing the peak power density, polarization curves, ionic resistances 2.3. Electrochemical characterization
of the membrane and catalyst layer, the charge transfer resistance, and
electrochemical surface area before and after 9000 aging cycles, as The electrochemical characteristics of the MEAs were investigated

2
E. Balogun et al. Journal of Power Sources Advances 3 (2020) 100012

using polarization data, EIS and cyclic voltammetry (CV). Polarization and electrode thickness, respectively, while LFR is the low-frequency
data were recorded from OCV to a shutoff potential of 0.3 V in 200 mA resistance.
steps, measuring 5 min⋅point1. Higher resolution in the kinetic region
was attained using a current scan from 0.00 to 0.20 A (40 mA/cm2) in 2.4. Accelerated degradation
0.01 A steps at 1 min⋅point1. Similarly, the Ohmic region of the I–V
curves was evaluated from 0.50 A (100 mA/cm2) to 1.50 A (300 mA/ The accelerated degradation protocol is described in Ref. [28].
cm2), in 0.50 A (100 mA/cm2) steps of 5 min⋅point1. The latter region of Briefly, at 80  C, and 100% RH and 30% RH at the anode and cathode,
the polarization curve was obtained by scanning through 2A (400 mA/ respectively, the aging process is accelerated via the following steps: (i)
cm2) to 15A (3000 mA/cm2) using 1 A (200 mA/cm2) steps at maintain the cell at OCV for 8 sec; (ii) perform a voltage scan in the range
5 min⋅point1. OCV → 0.6 V → OCV, at a scan rate of 50 mV/s. Steps i and ii are repeated
A VersaStat 4 Potentiostat/Frequency Response Analyzer (FRA) was for up to 9000 cycles. The cell performance was characterized after 3000
used for electrochemical characterizations. Cyclic voltammograms (CV) cycle intervals.
were measured by sweeping the potential between 0.04 V and 0.80 V vs.
RHE at a scan rate of 50 mV/s, after an initial potential hold at 0.4 V vs. 3. Results and discussion
RHE for 45 sec. Measurement conditions were 80  C and 100% RH, with
inlet gas flows of 0.5 slpm H2 at the anode and 1.0 slpm O2 (0.5 slpm N2 Cathode Starvation Conditioning Procedure. The approach used
for CL investigation) at the cathode. here is similar to the proton-pump configuration. This protocol focuses
The electrochemical surface area associated with hydrogen adsorp- not just on the hydration of the MEA but also on reclaiming dehydrated
tion was estimated using the following relation: and blocked catalyst sites in the cathode. This is achieved by allowing
crossover of hydrogen gas from the anode to the cathode, lowering the
Q
ECSA ¼ (1) cathode potential and thereby promoting reducing conditions at the
L * μPt cathode. The proposed cathode starvation procedure entails a series of
steps. First, the backpressure on the anode is increased from ambient to
where ECSA is the electrochemically active surface area, Q is the charge
1.5 bar to aid the crossover of H2 from anode to the cathode. Thereafter,
density of the atomic hydrogen adsorption, L is the Pt loading (0.4 mg Pt/
the current is increased in steps until the cell voltage is 0.3 V. The cell is
cm2) and μpt is the charge required to reduce a monolayer of protons on a
maintained at this voltage for 2–5 min in order for the current response to
polycrystalline Pt surface of 1 cm2 (210 mC cm2 Pt).
stabilize. The current is then reduced to 1 A (200 mA/cm2). The supply of
EIS analysis was performed on cells equilibrated under 0.5 slpm H2 at
cathode gas is stopped while maintaining the current at 1 A. The voltage
the anode and 0.5 slpm N2 at the cathode, once a stable potential of
of the cell is observed to drop since there is no supply of oxygen to the
<0.15 V was achieved. Under these conditions, an AC voltage amplitude
cathode. At the point where the voltage reaches 0.2 V, the cathode gas
of 10 mV (RMS) was applied, and frequency scans were performed from
flow is quickly re-supplied. The process is repeated. A more rapid drop
100 kHz to 100 mHz. EIS characterization was performed potentiostati-
voltage is observed overtime when the cathode gas is shut off. At the end
cally at 0.8 V. A Randle's equivalent circuit model was used to extract
of the conditioning procedure, the backpressure at both electrodes is
meaningful information by modeling the different regions of the fuel cell
returned to ambient. This protocol is illustrated in Fig. 1a and the
as an electrical circuit having ideal resistors, capacitors, and inductors
resultant maximum power density obtained as a function of these cycles
[37–39]. The corresponding impedance data were plotted in a Nyquist
is listed in Table S1. The MEA is considered conditioned when the dif-
plot, and fitted in Zview software using a simple Randles circuit [40].
ference between the last two cycles is < 5 mV. Full conditioning is
To determine the ionic resistance of the catalyst layer, a double par-
attained in <40 min. Moreover, > 90% of the maximum current (density)
allel resistive rail was used, as detailed by Qi et al. [41], and Lefebvre
is achieved after one conditioning cycle (12 min). The reproducibility of
et al. [42]. One of the parallel resistive rails represents the electron
the polarization curves using three individual MEAs is illustrated in
transport through the conducting carbon support, and the other repre-
Fig. S1, which shows overlaid I–V and power density plots immediately
sents ion transport through the inter-particle regions corresponding to
after their conditioning.
ion migration through the catalyst layer [41,42]. The low-frequency
DOE Conditioning Procedure. The DOE conditioning procedure
resistance (LFR) is represented by an upward curve in the impedance
entails applying multiple cycles of potential [24]. In this protocol, as
plot, to a limiting capacitive response (vertical) which corresponds to the
illustrated in Fig. 1d, the fuel cell is switched between its open-circuit
total capacitance and resistance of the catalyst layer. The ionic resistance,
voltage and 0.55 V repeatedly every hour until the last two measured
Rionic, was obtained from the length of the Warburg-like region projected
current values are < 50 mA (10 mA/cm2). The MEA is expected to reach
onto the real impedance (Z') axis (Rionic/3) [40–43]. The membrane's
its peak performance after 6 h [20]. The current response to the condi-
ionic resistance was obtained from the high-frequency intercept (HFR)
tioning cycles is illustrated in Fig. 1b and the current and current den-
with the x-axis on the Nyquist plot [27].
sities after each conditioning cycle is listed in Table S2, which shows
L MEAs did not meet the assigned metric for conditioning until 8 condi-
σm ¼ (2) tioning cycles had been carried out, i.e., the equivalent of 8 h condi-
A * Rionic
tioning time. It is noted that >90% of the maximum current (density) is
Similarly, from the electrode's ionic resistance was calculated using achieved after one conditioning cycle (1 h). The reproducibility of the
the equation below [31], polarization curves using three individual MEAs is demonstrated in
REionic Fig. S2, which shows overlaid I–V and power density plots immediately
LFR ¼ after their conditioning.
3
EU Harmonized Conditioning Procedure. The EU procedure in-
and, volves relative humidity cycling, together with stepwise current and
potential cycling [25]. As illustrated in Fig. 1c, the current density is
LE 3 * LE increased in steps of 100 mA/cm2 up to 800 mA/cm2 while keeping the
σE ¼ ¼ (3)
A * REionic A * LFR cell voltage >400 mV. The cell is allowed to stabilize for 6 h (region i).
The relative humidity of the reactants is then reduced to 50% RH while
where σm and σE are the membrane and electrode's conductivity
increasing the current and keeping the cell voltage >400 mV, up to 1
respectively. Rionic and RE-ionic represent the membrane and electrode's
A/cm2. The current is maintained under this condition for 2 h (region ii).
ionic resistance. A is the active cell area, and L and LE are the membrane
Following this, potential cycling is carried out between 0.75 V and

3
E. Balogun et al. Journal of Power Sources Advances 3 (2020) 100012

Fig. 1. Plots showing the current and


power density response as a function of
conditioning time with current (black
dotted line) and power density (red)
response for MEAs consisting of Nafion®
NR-211, Nafion® D520 ionomer in the
catalyst layer and Pt loadings of
0.4 mg Pt/cm2 for cathode and anode:
(a) Cathode starvation conditioning (b)
DOE conditioning, (c) EU conditioning,
(d) amperometric conditioning. (For
interpretation of the references to colour
in this figure legend, the reader is
referred to the Web version of this
article.)

0.45 V, and the cell is allowed to stabilize for 30 min at 0.45 V. This cycle maintained a maximum power density of 1230 mW/cm2 (15% drop), the
is repeated until the cell current becomes stable. The recommended EU-conditioned MEA yielded 1190 mW/cm2 (17% drop), and the
stability criterion is based on the cell voltage fluctuation and is consid- amperometric conditioned-MEA yielded 1180 mW/cm2 (17% drop) as
ered fulfilled when lower than 5 mA (1 mA/cm2) in the final 2 steps depicted in Fig. 4. Table 1 shows the values of the maximum power
(region iii) [21]. The current (density) after successive conditioning density and current density at 0.6 V for the MEAs conditioned by the
potential cycles are listed in Table S3. MEAs are considered fully condi- different conditioning procedures. The reproducibility of the polarization
tioned after 11 h. Overlaid I–V curves for three individual MEAs imme- curves using three individual MEAs for each of the four condition pro-
diately after their conditioning are shown in Fig. S3 to demonstrate tocols are illustrated in Figs. S1–S4, which shows overlaid I–V and power
experimental reproducibility. density plots immediately after 9000 degradation cycles.
Amperometric Conditioning. This is a common conditioning pro- To further characterize MEAs, impedance spectra were analyzed
cedure and one we had adopted for several years in our laboratory. As using EIS and the corresponding frequency response plotted as Nyquist
illustrated in Fig. 1d, the entire conditioning process involves linear plots (shown in Figs. S7 and S8). Data for unconditioned MEAs are shown
sweeps of current ranging from 2A (400 mA/cm2 initial point) to 15 A for comparison and to exemplify the importance of conditioning
(3.0 A/cm2 final point) at 5 min.point1 and 0.25 A (50 mA/cm2) steps
per point. The procedure was repeated four times before attaining
optimal power density performance (i.e., the difference between the last
2 current scans is < 50 mA (10 mA/cm2)). The current (density)
response as a function of the current cycle is illustrated in Fig. 1c and the
resultant maximum power density obtained as a function of these cycles
is listed in Table S3. From these data, MEA is considered fully condi-
tioned after 18 h. Overlaid I–V curves for three individual MEAs condi-
tioned amperometrically are shown in Fig. S4.
Fig. 2 shows graphically the time taken to attain optimum condi-
tioning when each of the conditioning procedure is used. As observed
from the polarization and power density curve shown in Fig. 3a, the DOE
conditioning protocol yields the highest initial performance immediately
after conditioning with a maximum power density of 1530  20 mW/
cm2. This is followed by MEAs conditioned with the cathode starvation
protocol with 1450  10 mW/cm2; followed by the EU protocol, with
1430  6 mW/cm2; and the amperometric conditioning protocol, with
1420  6 mW/cm2 (see Table 1).
After 9000 degradation cycles, the DOE-conditioned MEA yields a Fig. 2. Time for the fuel cell to reach maximum power density using various
power density of 1245 mW/cm2 representing a 19% drop in performance conditioning protocols. Error bars represent the upper and lower boundaries of
from its initial value. The cathode starvation conditioned-MEA three separate measurements.

4
E. Balogun et al. Journal of Power Sources Advances 3 (2020) 100012

Fig. 3. Polarization (black) and power density plots (red) of MEAs subjected to the four conditioning procedures: (a) immediately following conditioning, (b) after
9000 degradation cycles. FC polarization data were obtained at 80  C, H2 anode and O2 cathode, 100% RH, 1 atm pressure. (For interpretation of the references to
colour in this figure legend, the reader is referred to the Web version of this article.)

Table 1
Maximum power density, and current density at 0.6 V, for MEAs conditioned under different procedures and after accelerated degradation cycles. Std. Devs. provided in
Table S5.
Conditioning Maximum power density (mA/cm2) Current density at 0.6 V (A/cm2)
Procedure
After conditioning (0 After 3000 After 6000 After 9000 After conditioning (0 After 3000 After 6000 After 9000
cycles) cycles cycles cycles cycles) cycles cycles cycles

Cathode Starvation 1450 1380 1310 1230 2000 1650 1450 1330
DOE 1530 1420 1360 1245 2100 1750 1650 1450
EU 1430 1390 1290 1190 1800 1650 1430 1280
Amperometric 1420 1320 1240 1180 1800 1600 1430 1230

30% increase and a 23% decrease in membrane resistances (Rmea before


and after degradation) were observed for MEAs conditioned using the EU
and amperometric protocols, respectively. This is an indication that for
MEAs conditioned by the DOE and cathode starvation protocols, the
membrane is well hydrated and that there is no significant degradation of
the membrane occurring even after 9000 degradation cycles. For the EU
conditioning protocol, a significant increase in membrane resistance is an
indication of accelerated degradation of the membrane. Similarly, the
reduction in membrane resistance observed in the amperometric pro-
cedure after 9000 cycles indicate that the membrane was not fully
conditioned initially. This is interesting to note considering that the
amperometric protocol required the longest time to attain its peak
performance.
Similarly, the charge transfer resistance (Rct) recorded for uncondi-
tioned MEAs is observed (Fig. S7, Table 2) to be > 485 mΩ cm2, this is
more than double the Rct values for conditioned cells (Table 2) irre-
spective of the conditioning procedure. Rct values recorded immediately
Fig. 4. Degradation data comparing the performance loss observed as a function after conditioning and after 9000 degradation cycles are observed to
of percentage loss in maximum power density after various accelerated degra- almost double, as listed in Table 2. This implies that the degradation of
dation cycles for MEAs conditioned by the different procedures. the MEA is prominent in the catalyst layer. It is observed that the cathode
starvation conditioning protocol yields MEAs with the lowest Rct
(Fig. S7). The data extracted from Nyquist plots of unconditioned MEAs immediately after conditioning while the DOE protocol yields the largest
are shown in Table S5. Unconditioned MEAs are observed to possess a charge transfer resistance. This suggests that the cathode starvation
relatively large membrane resistance (Rmea) of 99.5 mΩ cm2 which is protocol promotes the rate of ORR and by extension leads to the reduc-
almost double that of conditioned cells, as can be observed from the tion in the activation polarization and improvement in overall MEA
extracted resistance data presented in Table 2. The cathode starvation performance. Noting that Pt oxides and hydroxides may exist on the
procedure is observed to yield MEAs with the lowest membrane resis- catalyst surface, formed in the presence of humidified O2, and that the
tance immediately after conditioning. The highest Rmea value is found for rate of the O2 reduction reaction is higher on Pt than on PtO or Pt(OH)4
the amperometric procedure. After 9000 cycles, there were no significant [44], we speculate that the cathode starvation protocol, which is asso-
changes in membrane resistances before and after degradation for MEAs ciated with the H2 fuel being consumed at the anode and produced at the
conditioned by the DOE and cathode starvation procedures. However, a cathode produces sufficiently strong reducing conditions at the cathode

5
E. Balogun et al. Journal of Power Sources Advances 3 (2020) 100012

Table 2
MEA resistance data immediately after conditioning and after 9000 degradation cycles for MEAs conditioned using different procedures.
Conditioning procedure Unconditoned MEA Cathode starvation DOE EU Amperometric
2
After conditioning (0 cycles) Rmea (mΩ cm ) 99.5  10.1 37.3  3.78 41.8  4.24 39.3  3.99 51.5  4.85
Rct (mΩ cm2) 488  5.22 193  6.52 242  5.01 210  4.02 209  8.04
After 9000 cylces Rmea (mΩ cm2) – 37.4  3.24 42.4  2.19 56  4.62 39.6  3.23
Rct (mΩ cm2) – 428  5.84 355  3.22 465  5.20 489  6.08

CL Ionic resistance after conditioning Rionic/3(mΩ cm2) 143  4.98 41.6  2.32 51.5  3.24 55.5  1.08 58.5  3.06
Rionic (mΩ cm2) 429  10.1 125  6.96 155  9.72 166  3.24 175  9.18
CL Ionic resistance 9000 cylces Rionic/3(mΩ cm2) – 40.8  2.24 55  3.92 46.4  2.98 47.5  4.02
Rionic (mΩ cm2) – 123  6.72 166  11.8 139  8.94 143  12.1

electrode to reduce Pt oxides and hydroxide to bare Pt.


To determine the ionic resistance of the CL, a transmission line
Randles equivalent electric circuit model was used to model the Nyquist
plot as shown in Fig. S9 and Fig. S10. The ionic resistance of an uncon-
ditioned catalyst layer in an MEA is found to be 428 mΩ cm2 (Fig. S8,
Table S5). Upon conditioning, this resistance drops by 200%. The
reduced CL ionic resistance observed in the CL after conditioning is
believed to be associated with hydration and improved alignment of
hydrophilic domains leading to well-connected protonic pathways
within the CL.
Comparing the various conditioning protocols, from Table 2, it is
observed that CL ionic resistance is lowest for the cathode starvation
protocol both immediately after conditioning (Cathode starvation,
125 mΩ cm2; Amperometic, 175 mΩ cm2; DOE, 155 mΩ cm2; EU,
166 mΩ cm2) and after 9000 degradation cycles (Cathode starvation,
122 mΩ cm2; Amperometric, 142 mΩ cm2; DOE: 166 mΩ cm2; EU:
139 mΩ cm2). The reduction in Rionic observed after degradation is
corroborated by an earlier report by Farhana et al. [45], which is an
indication that potential cycling does not lead to ionomer degradation
under these aging conditions. From the values of proton resistance within
the CL, it is observed that for the cathode starvation protocol, there is
little difference between the ionic resistance of the CL immediately after
activation and after 9000 degradation cycles. It shows that the cathode
starvation conditioning procedure yields an MEA that is well hydrated
and maintains its performance over a longer period. Hydration within the
CL is key to proton conductivity and reduced resistance [45,46], the
cathode starvation protocol promotes the hydration of the CL, which is
necessary for efficient proton transfer within the CL, and between the CL
to the membrane.
The CV-determined electrochemical surface area (ECSA) of the Fig. 5. Plot showing (a) Electrochemical Surface Area (ECSA) for MEA cathodes
cathode for MEAs conditioned using different procedures are shown in conditioned with different procedures and subjected to increasing accelerated
Fig. 5a. The ECSA recorded for each MEA after conditioning was similar degradation cycles. (b) An inverse relationship between the ECSA (m2/g pt) and
to values reported in earlier studies for MEAs with similar configurations the MEA charge transfer resistance (mΩ cm2) for MEAs conditioned using four
different conditioning procedures. (A) Data obtained immediately after condi-
[45,47,48]. MEAs conditioned using the cathode starvation protocol
tioning, (B) after 3000 degradation cycles, (C) after 6000 degradation cycles,
yielded the highest ECSA values after conditioning and throughout the
and (D) after 9000 degradation cycles. Conditions for ECSA measurement:
degradation process. Similarly, contrasting the ECSA lost after 9000 cy- 80  C, H2 anode and N2 cathode, 100% RH, 1 atm pressure.
cles, MEAs conditioned by cathode starvation, the amperometric pro-
cedure, the DOE procedure, and the EU procedure lost 40%, 42%, 40%,
hydrophilic phase, it is clear from the higher double-layer capacitance
and 42% of their value, respectively. The reduction in ECSA observed
observed that the conditioning procedure increases the total electroni-
after degradation is believed the result of carbon substrate corrosion,
cally conductive surface area, which is consistent with the observed in-
dissolution of the Pt catalyst and/or washing away of the catalyst as a
crease in ECSA [45,47]. Similarly, from Fig. S11, the cathode starvation
result of excess water formation at the electrode [26].
conditioning procedure also showed the highest Cdl value, which is
The catalyst layer was also analyzed for double-layer capacitance
indicative of a larger ECSA as observed from Fig. 5a. Furthermore, it is
formation using a plot of 1/-Im Z*ω vs Re Z as shown in Fig. S11. The
observed that an inverse relationship between the charge transfer resis-
slope of the plot for the MEA prior to conditioning was relatively low,
tance (Rct) of the MEA and the ECSA exists, as illustrated in Fig. 5b. As the
indicating that the catalyst layer possessed a higher resistivity [47]. The
ECSA decreases, the charge transfer resistance (Rct) of the MEA increases.
MEA conditioned by the cathode starvation procedure was observed to
Scanning electron micrographs (SEM) of cross-sections the different
yield the steepest slope (lowest resistivity), followed by the DOE,
MEAs are shown in Fig. S12. It is observed that the thickness of the
amperometric and EU procedures, respectively. Furthermore, it is
membranes of the MEAs after undergoing accelerated degradation is
observed from Fig. 5a and Fig. S11, that the double layer capacitance
relatively unchanged, with the membrane thickness ranging between
increases with increasing ECSA. Since the charge double capacitance
24.7 and 25.7 μm regardless of the conditioning procedure used. How-
(Cdl) reports on the total electronically conductive surface area (catalyst
ever, differences in catalyst layer thicknesses were observed based on the
þ conductive support) in the catalyst layer that is in contact with the

6
E. Balogun et al. Journal of Power Sources Advances 3 (2020) 100012

conditioning procedure employed. Membranes conditioned using the [2] J. Peron, Z. Shi, S. Holdcroft, Hydrocarbon Proton Conducting Polymers for Fuel
Cell Catalyst Layers, 2011, https://doi.org/10.1039/c0ee00638f.
cathode starvation and DOE procedures showed a similar cathode cata-
[3] D. Papageorgopoulos, Fuel cell R&D overview, in: Annu. Merit Rev. Peer Eval.
lyst thickness of 9.9–10.2 μm and similar anode thickness of Meet., 2019 (2019) 33.
11.2–11.7 μm. MEAs conditioned using the EU protocol showed a thinner [4] T.W. Lim, S.H. Kim, S.Y. Ahn, B.K. Hong, B.K. Ahn, (12) United States Patent, 2013,
electrode thickness of 9.84–9.88 μm for the cathode and 10.1–10.9 μm US 2010/0129689 A.
[5] J. Peron, A. Mani, X. Zhao, D. Edwards, M. Adachi, T. Soboleva, Z. Shi, Z. Xie,
for the anode. MEAs conditioned using the amperometric procedure T. Navessin, S. Holdcroft, Properties of Nafion® NR-211 membranes for PEMFCs,
showed the thinnest electrode thicknesses, with 9.54–9.56 μm for the J. Membr. Sci. 356 (2010) 44–51, https://doi.org/10.1016/j.memsci.2010.03.025.
cathode and 9.80–9.84 μm for the anode. According to Oyarce et al. [30], [6] Z. Qi, A. Kaufman, Quick and effective activation of proton-exchange membrane
fuel cells, J. Power Sources 114 (2003) 21–31, https://doi.org/10.1016/S0378-
changes in electrode thickness during degradation may be attributed to 7753(02)00587-6.
the loss of carbon support as a result of carbon corrosion at high oxidative [7] D. Bezmalinovic, J. Radosevic, F. Barbir, Initial conditioning of polymer electrolyte
potentials. It is observed that both the cathode starvation and the DOE membrane fuel cell by temperature and potential cycling, Acta Chim. Slov. 62
(2015) 83–87, https://doi.org/10.17344/acsi.2014.730.
protocol provides superior protection to the electrode against such losses [8] M. Bradley P. Anderson, Andover, United States Patent 12 (2009). US7608118B2.
compared to the other two activation protocols. [9] K. Palanichamy, A.K. Prasad, S.G. Advani, 214th ECS meeting, abstract #1739,
Electrochem. Soc. 410 (2008) 1739.
[10] Z. Xie, X. Zhao, M. Adachi, Z. Shi, T. Mashio, A. Ohma, K. Shinohara, S. Holdcroft,
4. Conclusion T. Navessin, Fuel cell cathode catalyst layers from “green” catalyst inks, Energy
Environ. Sci. 1 (2008) 184–193, https://doi.org/10.1039/b808613n.
The performance and durability of MEAs subjected to a new accel- [11] W. Bi, G.E. Gray, T.F. Fuller, PEM fuel cell PtC dissolution and deposition in Nafion
electrolyte, Electrochem. Solid State Lett. 10 (2007) 101–104, https://doi.org/
erated conditioning procedure, involving cathode starvation, was 10.1149/1.2712796.
investigated and compared with the standardized DOE and EU harmo- [12] J. Shan, X. Yan, X. Sun, Z. Hou, P. Qi, P. Ming, Chinese Patent 2010, CN
nized conditioning procedures and against an in-house-adapted amper- 10010014.3.
[13] USFCC single cell test protocol, Available from: http://www.fchea.org/core/impor
ometric procedure. The time required to condition MEAs using the t/PDFs/TechnicalResources/MatCompSingleCellTestProtocol05-014RevB.207130
cathode starvation procedure, as determined by the time to reach peak 6.pdf. (Accessed 10 December 2019).
power density is 40 min. Greater than 90% of the maximum current [14] L.N.( US, Ryan, Hallum, (12) United States Patent, 2003. US 6576356 B1.
[15] C.O. Seung, E.Y. Seung, H.K. Myong, H.L. Jong, J.K. Jun, J.I. Son, Chinese Patent
(density) is achieved after 12 min. However, unlike other accelerated 2009, CN 101340004A.
conditioning procedures reported in the literature that are purported to [16] Z. Xu, Z. Qi, A. Kaufman, Activation of proton-exchange membrane fuel cells via CO
condition MEAs in under an hour, our cathode starvation procedure was oxidative stripping, J. Power Sources 156 (2006) 281–283, https://doi.org/
10.1016/j.jpowsour.2005.02.094.
found to lead to MEAs that were more durable under an accelerated [17] K. Fumio, M.Naoya, S. Ryoichi, European patent, 2005, EU05004688.7.
degradation test. The cathode starvation procedure results in a significant [18] C. Eickes, P. Piela, J. Davey, P. Zelenay, Recoverable cathode performance loss in
reduction in ionic resistance of the catalyst layer, increased ECSA and direct methanol fuel cells, J. Electrochem. Soc. 153 (2006) A171, https://doi.org/
10.1149/1.2136073.
decreased charge transfer resistance. This is attributed to the low cathode
[19] M.K. Debe, Tutorial on the fundamental characteristics and practical properties of
potential and H2 crossover from anode to cathode during the condi- nanostructured thin film (NSTF) catalysts, J. Electrochem. Soc. 160 (2013)
tioning that promotes proton transport within the CL and provides F522–F534, https://doi.org/10.1149/2.049306jes.
reducing conditions at the cathode to promote sufficiently strong [20] A. Goleman, D. Richard, M. Perdana, J. Chem. Inf. Model. 53 (2018) 1689–1699,
https://doi.org/10.1017/CBO9781107415324.004.
reducing conditions at the cathode electrode to reduce Pt oxides and [21] B.Gould, K. Lyons, O. Baturina, United States Patent, 2011, US 20110008686A1.
hydroxide to bare Pt. Future work should be undertaken to address the [22] T. Kato, H. Tsuge, T. Yamada, United States Patent, 2013, US2013171529A1.
validity of these processes. [23] X.Z. Yuan, S. Zhang, J.C. Sun, H. Wang, A review of accelerated conditioning for a
polymer electrolyte membrane fuel cell, J. Power Sources 196 (2011) 9097–9106,
For the MEAs conditioned by cathode starvation, the smallest per- https://doi.org/10.1016/j.jpowsour.2011.06.098.
centage change in CL ionic resistance and ECSA was found during [24] U.S. Department of Energy, DoE Procedures for Performing PEM Single Cell Testing,
accelerated degradation, with no significant difference in ionic resistance 2009, p. 47. # DE-FC36-06GO16028 April 8.
[25] G. Tsotridis, A. Pilenga, G. De Marco, T. Malkow, EU Harmonised Test Protocols for
and smallest percentage loss of active sites before and after degradation, PEMFC MEA Testing in Single Cell Configuration for Automotive Applications, JRC
compared to the other conditioning procedures examined. It can be Science for Policy report, 2015, https://doi.org/10.2790/54653.
concluded that the new cathode starvation conditioning procedure has [26] X.Z. Yuan, J.C. Sun, H. Wang, H. Li, Accelerated conditioning for a proton exchange
membrane fuel cell, J. Power Sources 205 (2012) 340–344, https://doi.org/
proven to yield MEAs exhibiting high performance and durability while 10.1016/j.jpowsour.2012.01.039.
taking 90% less amount of time compared to other standardized condi- [27] T. Mittermeier, A. Weiß, F. Hasche, G. Hübner, H.A. Gasteiger, PEM fuel cell start-
tioning protocols. The positive attributes of this conditioning protocol up/shut-down losses vs temperature for non-graphitized and graphitized cathode
carbon supports, J. Electrochem. Soc. 164 (2017) F127–F137, https://doi.org/
warrant further investigation to determine if the same attributes apply to
10.1149/2.1061702jes.
larger MEAs, fuel cell stacks, and state-of-the-art MEAs used in the in- [28] M.M. Taghiabadi, M. Zhiani, V. Silva, Effect of MEA activation method on the long-
dustry which operate with stabilized materials, thinner membranes, anti- term performance of PEM fuel cell, Appl. Energy 242 (2019) 602–611, https://
oxidant additives, lower Pt loadings, and. multi-component catalysts. doi.org/10.1016/j.apenergy.2019.03.157.
[29] I. Pivac, Barbir, Rejuvenation Techniques for Automotive Fuel Cells, 2019, p. 20. htt
ps://www.semanticscholar.org/paper/Rejuvenation-techniques-for-automotive-fue
Acknowledgments l-cells-Pivac-Barbir/50a7ea5eef9dcd906c93ef199f4478da9aea19d9.
[30] A. Oyarce, E. Zakrisson, M. Ivity, C. Lagergren, A.B. Ofstad, A. Boden, G. Lindbergh,
Comparing shut-down strategies for proton exchange membrane fuel cells, J. Power
This work was supported by the Natural Sciences and Engineering Sources 254 (2014) 232–240, https://doi.org/10.1016/j.jpowsour.2013.12.058.
Research Council of Canada (NSERC) and made use of facilities in 4D [31] N. Holmstr€ om, K. Wiezell, G. Lindbergh, Studying low-humidity effects in PEFCs
using EIS, J. Electrochem. Soc. 159 (2012) F369–F378, https://doi.org/10.1149/
LABS, SFU. 2.005208jes.
[32] M. Gerard, J. Poirot-Crouvezier, D. Hissel, M. Pera, Oxygen starvation effects on
pemfc durability, in: Proceedings of the ASME 2010 Eighth International Fuel Cell
Appendix A. Supplementary data
Science, Engineering and Technology Conference, 2010, pp. 1–8.
[33] K. Colbow, M. Van der Geest, C. Longley, J. Muller, J. Roberts, J. St[HYPHEN]
Supplementary data to this article can be found online at https Pierre, P. Urba, R. Wezel and D. Wilkinson, J. Zhang. US 6,472,090 B1. Oct, 29,
://doi.org/10.1016/j.powera.2020.100012. 2002.
[34] C. Reiser, R. Balliet, United States Patent, 2003, US 2003224228A1.
[35] Hyundai Motor Company, United States Patent, US 20170271693A1, 2017.
References [36] W.H. Zhu, R.U. Payne, B.J. Tatarchuk, PEM stack test and analysis in a power
system at operational load via ac impedance, J. Power Sources 168 (2007)
211–217, https://doi.org/10.1016/j.jpowsour.2007.02.071.
[1] R. Souzy, B. Ameduri, Functional fluoropolymers for fuel cell membranes,
[37] A. Baricci, R. Mereu, M. Messaggi, M. Zago, F. Inzoli, A. Casalegno, Application of
Fluorinated Mater. Energy Convers. (2005) 469–511, https://doi.org/10.1016/
computational fluid dynamics to the analysis of geometrical features in PEM fuel
B978-008044472-7/50049-7.

7
E. Balogun et al. Journal of Power Sources Advances 3 (2020) 100012

cells flow fields with the aid of impedance spectroscopy, Appl. Energy 205 (2017) IntechOpen , the World ’ S Leading Publisher of Open Access Books Built by
670–682, https://doi.org/10.1016/j.apenergy.2017.08.017. Scientists , for Scientists TOP 1 %, 2012, p. 13, https://doi.org/10.1016/
[38] Scribner Associates Inc, Electrochemical impedance spectroscopy ( EIS ): a powerful j.colsurfa.2011.12.014. Intech. i.
and cost- effective tool for fuel cell diagnostics, electrochem. Impedance spectrosc. [43] F.S. Saleh, E.B. Easton, Diagnosing degradation within PEM fuel cell catalyst layers
A powerful cost- eff, Tool Fuel Cell Diagnostics 35 (2014) 1–5, https://doi.org/ using electrochemical impedance spectroscopy, J. Electrochem. Soc. 159 (2012)
10.1016/0013-4686(90)80007-B. B546–B553, https://doi.org/10.1149/2.098205jes.
[39] Z. Qi, P.G. Pickup, High performance conducting polymer supported oxygen [44] S. Holdcroft, Fuel cell catalyst layers: a polymer science perspective, Chem. Mater.
reduction catalysts, Chem. Commun. (1998) 2299–2300, https://doi.org/10.1039/ 26 (2014) 381–393, https://doi.org/10.1021/cm401445h.
a805322g. [45] O'Rian O. Reid, Farhana S. Saleh, E. Bradley Easton, Application of the transmission
[40] M.C. Lefebvre, R.B. Martin, P.G. Pickup, Characterization of ionic conductivity line EIS model to fuel cell catalyst layer durability, ECS Transactions 457 (2014)
profiles within proton exchange membrane fuel cell gas diffusion electrodes by 356–359, https://doi.org/10.7868/s0869565214210269.
impedance spectroscopy, Electrochem. Solid State Lett. 2 (1999) 259–261, https:// [46] R. Alipour Moghadam Esfahani, E.B. Easton, Exceptionally durable Pt/TOMS
doi.org/10.1149/1.1390804. catalysts for fuel cells, Appl. Catal. B Environ. 268 (2020) 118743, https://doi.org/
[41] S. Touhami, J. Mainka, J. Dillet, S.A.H. Taleb, O. Lottin, Transmission line 10.1016/j.apcatb.2020.118743.
impedance models considering oxygen transport limitations in polymer electrolyte [47] E.B. Easton, P.G. Pickup, An electrochemical impedance spectroscopy study of fuel
membrane fuel cells, J. Electrochem. Soc. 166 (2019) F1209–F1217, https:// cell electrodes, Electrochim. Acta 50 (2005) 2469–2474, https://doi.org/10.1016/
doi.org/10.1149/2.0891915jes. j.electacta.2004.10.074.
[42] I.C. Yu, R. Chen, J.J. Li, J.J. Li, M. Drahansky, M. Paridah, A. Moradbak, [48] A.J. Steinbach, M.K. Debe, J.L. Wong, M.J. Kurkowski, A.T. Haug, D.M. Peppin,
A. Mohamed, H. abdulwahab taiwo Owolabi, FolaLi, M. Asniza, S.H. Abdul Khalid, S.K. Deppe, S.M. Hendricks, E.M. Fischer, A new paradigm for PEMFC ultra-thin
T. Sharma, N. Dohare, M. Kumari, U.K. Singh, A.B. Khan, M.S. Borse, R. Patel, electrode water management at low temperatures, E.C.S. Transactions 33 (2010)
A. Paez, A. Howe, D. Goldschmidt, C. Corporation, J. Coates, F. Reading, We Are 1179–1188.

You might also like