Adsorption behavior of low concentration carbon monoxide on polymer electrolyte fuel cell anodes for automotive applications

Journal of Power Sources 318 (2016) 1e8

Contents lists available at ScienceDirect

Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour

Adsorption behavior of low concentration carbon monoxide on polymer electrolyte fuel cell anodes for automotive applications Yoshiyuki Matsuda a, b, *, Takahiro Shimizu a, Shigenori Mitsushima b, c a

Japan Automobile Research Institute, 2530 Karima, Tsukuba, Ibaraki, 305-0822, Japan Green Hydrogen Research Center, Yokohama National University, 79-5 Tokiwadai, Hodogaya-ku, Yokohama, 240-8501, Japan c Institute of Advanced Sciences, Yokohama National University, 79-5 Tokiwadai, Hodogaya-ku, Yokohama, 240-8501, Japan b

h i g h l i g h t s
CO, CO2 and O2 in the anode exhaust were measured during the PEFC operation.
CO coverage was estimated from gas analysis and CO stripping voltammetry.
The CO coverage at low CO concentration followed a Temkin-type isotherm.
The CO coverage was 0.6 at 0.2 ppm CO and 0.11 mg cm2 anode loading at 60  C.
Permeated O2 should have an important role for CO oxidation at low CO concentration.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 14 January 2016 Received in revised form 28 March 2016 Accepted 29 March 2016

The adsorption behavior of CO on the anode around the concentration of 0.2 ppm allowed by ISO 146872 is investigated in polymer electrolyte fuel cells (PEFCs). CO and CO2 concentrations in the anode exhaust are measured during the operation of a JARI standard single cell at 60  C cell temperature and 1000 mA cm2 current density. CO coverage is estimated from the gas analysis and CO stripping voltammetry. The cell voltage decrease as a result of 0.2 ppm CO is 29 mV and the CO coverage is 0.6 at the steady state with 0.11 mg cm2 of anode platinum loading. The CO coverage as a function of CO concentration approximately follows a Temkin-type isotherm. Oxygen permeated to the anode through a membrane is also measured during fuel cell operation. The exhaust velocity of oxygen from the anode was shown to be much higher than the CO supply velocity. Permeated oxygen should play an important role in CO oxidation under low CO concentration conditions. © 2016 Elsevier B.V. All rights reserved.

Keywords: Polymer electrolyte fuel cell Carbon monoxide Gas analysis Oxygen permeation

1. Introduction

hydrocarbons [1,2]. In the case of methane, the steam reforming is

Polymer electrolyte fuel cells (PEFCs) are promising power devices for stationary, portable and transportation applications because of their high power density and efficiency. In particular, fuel cell vehicles (FCVs) are desirable because of the advantage of lower pollution emissions than internal combustion engine vehicles, and they have longer cruising distances than battery electric vehicles. Development and establishment of the hydrogen infrastructure is also in progress for the widespread use of FCVs. At present, most hydrogen is produced by steam reforming from

CH4 þ H2 O4CO þ 3H2 :

* Corresponding author. Japan Automobile Research Institute, 2530 Karima, Tsukuba, Ibaraki, 305-0822, Japan. E-mail address: [email protected] (Y. Matsuda). http://dx.doi.org/10.1016/j.jpowsour.2016.03.104 0378-7753/© 2016 Elsevier B.V. All rights reserved.

(1)

After the steam reforming, carbon monoxide (CO) in the reformate gas is removed by the water gas shift reaction in Equation (2) and purified by pressure-swing adsorption or CO selective oxidation in Equation (3):

CO þ H2 O4CO2 þ H2 ;

(2)

1 CO þ O2 /CO2 : 2

(3)

The CO concentration in the reformate gas after those reactions is reduced to a level of 10 ppm. Then, the hydrogen gas for the fuel cell is obtained by the separation of CO and other impurities in the

2

Y. Matsuda et al. / Journal of Power Sources 318 (2016) 1e8

pressure-swing adsorption process. However, a trace amount of impurities may remain in the hydrogen obtained as above. If the hydrogen purity is inadequate, the impurities may decrease the fuel cell's performance. On the other hand, excessively high hydrogen purity will lead to an increase in the hydrogen production cost. Therefore, a hydrogen quality specification for FCVs was discussed at ISO/TC197/WG12, which is the working group on hydrogen fuel specification, and was issued as an international standard as ISO 14687-2 in 2012 [3]. The allowable concentration of impurities in hydrogen fuel was specified in consideration of the effect of impurities on single cell performance or degradation of the constituent material. Methods for cell performance recovery after poisoning, analytical methods and accumulation behavior in the hydrogen circulation system were also considered [4]. This international standard applies to the early stages of commercial FCVs. Fuel cell cost reduction will be accelerated with the progress of technical development of fuel cell components, such as lowering of the platinum loading of electrocatalysts [5,6] and thinning of the electrolyte membrane. Preparation for the revision of the specification for hydrogen fuel is under way, looking ahead to the widespread usage stage. Among the impurities specified in the standard, CO is known as having one of the greatest potentials to decrease fuel cell performance. In addition, a CO concentration that was almost the same as the upper limit of the CO standard has been reported at a hydrogen station [7,8]. So, the allowable concentration of CO is still an important subject for discussion for the revision of the hydrogen fuel specification. There are many reports on the effect of CO poisoning on the performance of PEFCs [5,6,9e28] and on methods to decrease CO poisoning, such as by PteRu electrocatalysts [5,9e15] and air-bleed systems [16e20] in PEFCs. However, application of these antipoisoning techniques is difficult for FCVs because the PteRu electrocatalyst has a low tolerance for sulfur-containing impurities [9], and the air bleed should cause nitrogen accumulation in the hydrogen circulation system of the FCVs, or membrane degradation by the formation of H2O2 in the anode potential range [30]. Consequently, the evaluation of CO at low concentration with low anode platinum loadings is necessary for a discussion of the hydrogen fuel specification for the FCVs. Gas emission behaviors when a relatively low concentration of CO is added to the fuel cell have been reported in the literature [6,24,25]. However, the influence of CO has not been well understood because the change of CO coverage during the PEFC operation was not clear. To discuss the allowable concentration of CO, which considers the MEA specification and the operating conditions with the widespread use of FCVs, it is important to understand the basic phenomena, such as the relationship between CO coverage and performance degradation of the PEFC. In addition, permeated oxygen from the cathode would react with hydrogen or CO on the anode electrocatalyst. If permeated oxygen exists in the anode exhaust, oxygen will accumulate in the hydrogen circulation system of the FCVs, and subsequently it will make a possible contribution to mitigating the effect of CO, as in airbleed systems [16e22]. However, oxygen concentration in the anode exhaust has not been reported. In this study, the adsorption behavior of CO on the anode during PEFC operation has been investigated to understand the progress of CO poisoning near the upper limit of ISO 14687-2 and recovery by high-purity hydrogen. Both CO and CO2 concentrations in the PEFC exhaust were measured to evaluate the carbon balance of CO in the supplied hydrogen fuel. The oxygen concentration in the anode exhaust was also measured to evaluate the permeated oxygen content from the cathode.

2. Theoretical In general, two types of bonding for CO adsorption are known on the platinum surface, linear and bridge bonds. The linear-bonded CO occupies one platinum site, while the bridge-bonded CO occupies two sites. Igarashi et al. reported that the type of CO bonding on Pt sites was dependent on the CO coverage, and that linearbonded CO is dominant near full coverage [26]. To estimate the CO coverage in a single cell under operation with hydrogen containing a low concentration of CO, the two possible CO coverage scenarios, which are only linear-bonded CO (qL) and both linearand bridge-bonded CO (qLþB), are considered here. The CO coverage is estimated by determination of the two types of CO adsorption. One is measured by CO stripping voltammetry and the other is measured by anode exhaust analysis during cell operation. In CO stripping voltammetry, if the platinum catalyst is exposed to a high concentration of CO, linear-bonded CO dominates and the CO coverage reaches almost 1.0 [26]. Then the electrochemical oxidation of CO is expressed by the following equation.

Pt  CO þ H2 O/Pt þ CO2 þ 2Hþ þ 2e

(4)

The amount of CO adsorption n (mol) is estimated using the CO oxidation charge QCO (C) and geometric electrode area ACell (cm2) as follows:

n¼

QCO ACell ; 2F

(5)

where F is the Faraday constant (¼ 9.65  104 C mol1). The whole platinum surface area SCO (m2) is represented by the charge needed to strip a monolayer of CO on platinum, 420 mC cm2:

SCO ¼

QCO ACell : 420

(6)

The CO and CO2 concentrations in the anode exhaust of a single cell were measured to estimate the amount of CO adsorption. The CO exhaust velocity vCO,out,t (mol h1) and the CO2 exhaust velocity vCO2,out,t (mol h1) at time t (h) can be represented as follows:

vCO;out;t ¼

1

vCO2 ;out;t ¼

1 Sf

1

!

1 Sf

!

CCO;out;t FH2 ; Vm CCO2 ;out;t FH2  vCO2 ;b;t ; Vm

(7)

(8)

where CCO,out,t (ppm) and CCO2,out,t (ppm) are the cell outlet CO and CO2 concentrations at time t (h), respectively. Sf, FH2 (L h1), Vm (¼22.4 L mol1) and vCO2,b,t are the anode stoichiometry, fuel flow rate, mole volume and CO2 exhaust velocity baseline, respectively. All CO supplied to the anode is thought to be exhausted as CO or CO2 at the steady state. Here, the CO and CO2 exhaust velocity at the steady state are defined as vCO,out,s (mol h1) and vCO2,out,s (mol h1), respectively. Therefore, the CO adsorption velocity vCO,cell,t (mol h1) at time t (h) can be represented as follows:

    vCO;cell;t ¼ vCO;out;s þ vCO2 ;out;s  vCO;out;t þ vCO2 ;out;t :

(9)

Then, the cumulative CO adsorption amount per platinum surface area qCO (mol m2) is

qCO ¼

1 SCO

Zt 0

  vCO;cell;t dt:

(10)

Y. Matsuda et al. / Journal of Power Sources 318 (2016) 1e8

If the CO adsorption is only linear-bonded, the CO coverage qL is obtained by dividing qCO by the saturated amount of CO adsorption (n/SCO), which can be obtained from Equations (5) and (6); that is

qL ¼

qCO Fq ¼ CO n=SCO 210

(11)

Then, the CO coverage that takes into account both linear- and bridge-bonded CO (qLþB) can be estimated as follows. The mole fractions of linear- and bridge-bonded CO are denoted by fL and fB, and their CO adsorption amounts are denoted by qLCO and qBCO, respectively. The relationship between the linear- and bridgebonded CO amounts and mole fractions is given by

fL þ fB ¼ 1;

(12)

qCO ¼ qCO L þ qCO B ;

(13)

fL ¼

qCO L ; qCO

fB ¼

qCO B : qCO

(14)

The number of Pt sites NCO (mol m2) that are covered by CO is given by

NCO ¼ qCO L þ 2qCO B ¼ qCO þ qCO B :

(15)

The CO coverage qLþB, which is assumed to include both linearand bridge-bonded CO, can be obtained by dividing both sides of Equation (15) by n/SCO; that is

qLþB ¼

NCO q ¼ CO n=SCO n=SCO

1þ

qCO B qCO

3

3.2. Experimental apparatus The single-cell tests were conducted by using a fuel cell test station (Toyo Corporation) with an electrical load (890CL, Scribner Associates Inc.) (current accuracy of ±0.15%, voltage accuracy of ±5 mV). A mixture of high-purity hydrogen (99.99999%) and COcontaining hydrogen was used as the anode gas. The air, which was purified by an air purifier (Model-JAR, Japan Pionics Co. Ltd), was used as the cathode gas. The flow rate of each gas was adjusted using mass flow controllers (accuracy of ±1% F.S.). The hydrogen and air were humidified with bubblers. Some CO2 from the atmosphere would exist in the humidification water and affect the analysis of trace amounts of CO2. Therefore, hydrogen was used to purge CO2 from the water more than 5 h before the single cell operation test. Moreover, no additional water was supplied to the humidifier during the examination. A gas chromatograph (Agilent Technologies 7890A series) was employed for the analysis of CO, CO2 and O2 in the anode exhaust gas. This gas chromatograph has a Valco pulsed-discharge helium ionization detector (PDHID), so the trace amounts of CO and CO2 could be analyzed. Almost all compounds can be detected with the PDHID, and the detection limit is lower than that of conventionally used detectors such as the thermal conductivity detector (TCD) and flame ionization detector (FID) [32]. A backflush was performed to shorten the time per analysis. The heart-cut technique was used to separate the CO and CO2 peaks clearly from the hydrogen gas. The lower detection limits of CO and CO2 in this system are 30 and 40 ppb, respectively. The O2 concentration in the hydrogen was measured using the TCD.

! ¼ qL ð1 þ fB Þ:

(16)

The relationship between the bridge-bonded (or linear-bonded) CO mole fraction and the CO coverage has been reported by Igarashi et al. [26]. We applied their reports in Equation (16) to estimate qLþB.

3. Experiment 3.1. Catalyst-coated membrane (CCM) preparation and cell assembly The catalyst paste was obtained by mixing the Pt/C catalyst (TEC10E50E, 46 wt% Pt, Tanaka Kikinzoku Kogyo K.K.) and Nafion® dispersion solution (DE2020 CS, DuPont) with bead mills. The weight ratio of the carbon support to ionomer content was 1.0. The catalyst layer was coated uniformly on polytetrafluoroethylene (PTFE) sheets. The platinum loading on the anode and cathode were 0.11 and 0.3 mg cm2, respectively. Then, the catalyst layer was hotpressed on both sides of an electrolyte membrane (Nafion® NR-211, DuPont), 25 mm in thickness, at a temperature of 135  C. The geometric area of the electrodes was 25 cm2. The 24BCH and 25BCH (SGL Technologies GmbH) gas diffusion layers were used on the anode and cathode, respectively. A heating medium temperature controlled type JARI standard single cell, which can maintain a low temperature even under high current density operation, was used for the single-cell tests [31]. This cell has a serpentine water channel opposite the gas channel at the carbon separator, and the cell temperature is controlled by the water flow at a constant temperature. The CCM and gas diffusion layers were assembled on the cell. The gasket was 0e10 mm thicker than the sum of the height of the gas diffusion layer and the catalyst layer.

3.3. Experimental conditions for the CO exposure test The experimental conditions are listed in Table 1. Before single cell operation, the hydrogen and nitrogen were supplied to the anode and cathode, respectively. Both gas flow rates were 200 mL min1. The cathode gas was switched to air after the CO2 concentration of the anode exhaust reached the steady state. The gas analysis during the operation test was performed every 15 min. The preconditioning was conducted at 1000 mA cm2 using highpurity hydrogen as fuel for 5 h, and then the CO-mixed hydrogen was fed to the anode. After 30 or 50 h, the anode gas was switched to high-purity hydrogen again, and operated for 5 h to confirm the performance recovery by high-purity hydrogen. The CO, CO2 and O2 concentrations were measured using the GCePDHID or GCeTCD in the anode exhaust before and during the cell operation test. The oxygen concentration in the anode exhaust was also measured without current loading. The experimental conditions are shown in Table 2. Condition (1) was no air in the cathode and no current loading, to confirm there was no leakage from the atmosphere into the cell or the gas chromatograph system. In condition (2), the cathode atmosphere was air, again without current loading. The anode gas was nitrogen, so as not to react with permeated oxygen. Finally, condition (3) was oxygen, measured during the cell operation test.

Table 1 Operating conditions of the single cell. Current density/mA cm2 Fuel/air stoichiometry Cell temperature/ C Anode/cathode dew point/ C Pressure CO concentration/ppm

1000 1.43/2.50 60 47/40 Atmospheric at cell outlet 0, 0.2, 0.4, 1.0

4

Y. Matsuda et al. / Journal of Power Sources 318 (2016) 1e8

Table 2 Experimental conditions of oxygen measurement in the anode exhaust. Conditions

(1) without cathode air and current loading

(2) with cathode air, without current loading

(3) with cathode air and current loading

Current density/mA cm2 Anode gas Anode inlet flow rate/mL min1 Anode outlet flow rate/mL min1 Cathode gas Cathode inlet flow rate/mL min1

0 H2 75 75 N2 200

0 N2 75 75 Air 1036

1000 H2 249 75 Air 1036

3.4. CO stripping voltammetry The CO stripping voltammetry of the MEA anode was conducted to estimate the saturated adsorption amount of CO. A potentiostat (PGSTAT302N, Metrohm Autolab B. V.) was employed for the electrochemical measurements. The cell temperature and dew point were 25  C. The 1000 ppm CO balanced with nitrogen was supplied for 30 min at a flow rate of 200 mL min1, a concentration that should be enough to cover the platinum site completely [33,34]. Then, the anode gas was switched to nitrogen at a flow rate of 200 mL min1 to purge the remaining CO in the anode. The hydrogen was supplied to the cathode and used as a common electrode of a reversible hydrogen electrode (RHE) and a counter electrode. The anode potential was held at 0.10 V vs. RHE during the CO adsorption and nitrogen purge steps. After 15 min, the supply of nitrogen to the anode was stopped, and then the anode potential was scanned from 0.05 to 0.90 V vs. RHE at a scan rate of 50 mV s1 for three cycles. The amount of adsorbed CO molecules, N, was calculated from the area of the CO oxidation charge (QCO) at the first cycle.

4. Results and discussion 4.1. Gas analysis during the CO exposure tests The single-cell tests were conducted using hydrogen or hydrogen mixed with CO. The cell voltage and the concentration of CO and CO2 in the anode exhaust with the high-purity hydrogen and 0.2 ppm CO-containing hydrogen as a fuel are shown in Fig. 1(a) and (b), respectively. When high-purity hydrogen was used as a fuel, the voltage hardly changed throughout the 30 h. CO was not detected in the anode exhaust (<30 ppb), whereas CO2 was exhausted constantly at a concentration of 0.3 ppm. The CO2 generation in the anode exhaust may be attributed to the oxidation of the support of the anode electrocatalyst by hydrogen peroxide

under low potential conditions [35]. In case of 0.2 ppm CO-containing hydrogen, the cell voltage decreased, and the CO2 concentration increased after CO supply. After 40 h of CO injection, the cell voltage reached a steady state, and the concentration of exhaust gases also became constant. The voltage drop after 50 h was 29 mV. The cell voltage recovered after the high-purity hydrogen supply was reinstated at 50 h. The CO concentration decreased quickly to 0 ppm. The CO2 concentration also decreased, but the concentration was higher than 0.5 ppm, which was 0.2 ppm larger than the baseline even after a lapse of 5 h. The single-cell tests were also performed under the conditions of 0.4 and 1.0 ppm CO concentration in addition to 0.2 ppm. The exhaust velocities (mmol h1) of the CO and CO2 obtained from Equations (7) and (8) are shown in Fig. 2(a) and (b), respectively. In the case of 0.2 ppm, most CO was oxidized to CO2. The exhaust velocity of the CO increased with the rate of CO feed. The CO exhaust velocity decreased soon after changing to the high-purity hydrogen feed at 30 or 50 h, while the CO2 concentration gradually decreased. Fig. 3 shows the CO and CO2 exhaust velocities at the steady state over 30 or 50 h as a function of the inlet CO concentration. The dashed line shows the CO supply velocity, so the sum of the CO and CO2 exhaust velocities must be the same as the line at the steady state. The CO supply velocity was almost the same as the sum of the CO and CO2 exhaust velocities. In the case of the 0.2 ppm CO feed, most of the CO was oxidized to CO2. The CO exhaust velocity relative to CO2 increased with the CO supply velocity.

4.2. Amount of CO adsorption and coverage Fig. 4 shows the amount of CO adsorption as a function of time, derived from Equation (10). At first, the CO adsorption was proportional to time, and the slopes of the plots shown in Fig. 4(b) corresponded to the CO concentration. These results indicate that most of the CO adsorbed on the anode in the initial few hours. After

Fig. 1. CO and CO2 concentration in the anode exhaust and the cell voltage over time during (a) high-purity hydrogen and (b) CO (0.2 ppm)-contaminated hydrogen feeding at 60  C cell temperature with 1000 mA cm2 current density.

Y. Matsuda et al. / Journal of Power Sources 318 (2016) 1e8

5

Fig. 2. (a) CO and (b) CO2 exhaust velocities as a function of time at various CO concentrations at 60  C with 1000 mA cm2 current density.

Fig. 3. Mass balance of the supplied CO and exhaust for the sum of CO and CO2 at 60  C cell temperature with 1000 mA cm2 current density. The dashed line represents the CO supply velocity.

the fuel was switched to high-purity hydrogen again, the amount of CO adsorption decreased, but CO adsorbed on the anode remained for 5 h of continuous operation.

The saturated amount of CO adsorption (n/SCO) and the amount of steady state CO adsorption (qCO) as a function of CO concentration are shown in Fig. 5. The amount of full CO adsorption on the Pt surface was estimated as 22 mmol m2 Pt, which can be obtained by dividing Equation (5) by Equation (6). The amount of CO adsorption linearly increased with the logarithm of the CO concentration, which corresponds to a Temkin isotherm. This result is the same tendency as reported in the literature [16,17,29]. The mole fraction of linear- and bridge-bonded CO, the CO coverage which were obtained from Equations (11) and (16) and the cell voltage change as a function of CO concentration at the steady state are shown in Table 3. The mole fraction of linear-bonded CO increased with the CO concentration. The qL and qLþB were 0.51 and 0.76, respectively, at 0.4 ppm CO with a 46 mV cell voltage drop at 1000 mA cm2. CO coverage of 0.5 with CO-containing fuel should correspond to half the platinum catalyst loading with pure hydrogen. However, the cell voltage hardly changes after reducing the anode platinum loading from 0.1 to 0.05 mg cm2 at 1000 mA cm2 current density [5,6]. Therefore, the linear- and bridge-bonded model should be better to explain the voltage drop at 0.4 ppm CO compared with the linear-bonded model. The CO coverage qLþB would be about 0.6 at 0.2 ppm CO with a bridgebonded model. Our results reveal that 60% of the Pt would be covered by CO even at the 0.2 ppm level prescribed in ISO 14687-2 at 60  C temperature and 1000 mA cm2 current density. Fig. 6 shows the relationship between the voltage change and CO coverage qLþB from Equation (16). The shapes of these curves were almost the same despite the different CO concentrations. The voltage change was small at low CO coverage, but the voltage

Fig. 4. Amount of CO adsorption as a function of time at various CO concentrations at 60  C cell temperature with 1000 mA cm2 current density. Fig. 4(a) represents the whole change in the amount of CO adsorption during CO exposure and hydrogen recovery. A more detailed view of the change in the CO adsorption up to 6 h is shown in Fig. 4(b). The slopes (solid lines) represent the cumulative CO supply amount.

6

Y. Matsuda et al. / Journal of Power Sources 318 (2016) 1e8

Fig. 5. Amount of steady state CO adsorption as a function of CO concentration at 60  C cell temperature with 1000 mA cm2 current density. The dashed line represents the saturated amount of CO adsorption (n/SCO) calculated from Equations (5) and (6).

drastically dropped when the CO coverage was more than 0.6. Hysteresis between the voltage change and CO coverage was observed during CO poisoning and hydrogen recovery. From the analysis of segmented cell system, the beginning of CO poisoning on the anode electrocatalyst progresses from the anode upstream; at this time, temporary and significant current decrease at the cell inlet was occurred which was not observed in the hydrogen recovery [22]. The electrochemical impedance study showed that the high frequency resistance and charge transfer resistance both the anode and the cathode increased during the CO exposure [22,23]. Those resistance increase should cause an increase of anode and cathode overpotential. Those involved change of anode and cathode overpotential at the early stage of CO poisoning and the hydrogen recovery may be a cause of the hysteresis between the voltage change and CO coverage. The values of the CO coverage at each CO concentration were still high after the fuel was switched to high-purity hydrogen. This result implies that CO still remains on the anode catalyst even though the voltage apparently recovered.

Fig. 6. Relationship between CO coverage qLþB and voltage change during CO poisoning (dashed lines) and hydrogen recovery (solid lines). The cell temperature was 60  C and current density was 1000 mA cm2.

again without loading, and the anode oxygen exhaust velocity was 92 mmol h1, which permeated through the membrane. The oxygen permeation coefficient from the result was 1.0  1014 mol m1 s1 Pa1. This value is almost the same as the 1.1  1014 mol m1 s1 Pa1 at 55  C and 100% relative humidity found for Nafion® 117 and the 1.0  1014 mol m1 s1 Pa1 at 60  C and 95% relative humidity found for Nafion® NRE 212 [36,37]. Oxygen was detected even in the operating anode exhaust, and its concentration was 12 mmol h1, but the value was almost 1/8 that of H2/N2 for the anode/cathode. Most of the permeated oxygen should react with hydrogen to form water. Fig. 7 shows the oxygen exhaust velocity as a function of time during CO exposure. The oxygen exhaust velocity increased after

4.3. Oxygen concentration in the anode exhaust The oxygen concentration in the anode exhaust was measured for various operating conditions represented in Table 2. Firstly, oxygen was not detected at the anode with H2/N2 for the anode/ cathode, respectively, without loading. Therefore, there was no leakage into the gas analysis system or the cell from the atmosphere. Then, the anode/cathode gas was set to N2/air, respectively, Table 3 The mole fraction of linear- and bridge-bonded CO, the CO coverage and the cell voltage change as a function of CO concentration at the steady state. CO concentration/ppm

0.2

0.4

1.0

Mole fraction of linear CO, fL Mole fraction of bridge CO, fB The CO coverage qL The CO coverage qLþB Voltage change

0.20 0.80 0.34 0.61 29

0.49 0.51 0.51 0.76 46

0.63 0.37 0.59 0.80 216

Fig. 7. Oxygen exhaust velocity change at the cell outlet over time during CO exposure. The cell temperature was 60  C and current density was 1000 mA cm2.

Y. Matsuda et al. / Journal of Power Sources 318 (2016) 1e8

7

Fig. 8. Schematic explanation of the oxygen exhaust velocity change with and without CO on the anode.

CO injection to the anode. The oxygen exhaust velocity at the steady state increased with the CO feed concentration. A schematic explanation of the oxygen exhaust velocity change is illustrated in Fig. 8. If there is no CO in the hydrogen fuel, permeated oxygen reacts with hydrogen. On the other hand, if CO contaminates the hydrogen fuel, permeated oxygen reacts with hydrogen and CO. The CO oxidation velocity seems to be slower than that of hydrogen. If the CO coverage is high, the oxygen consumption velocity may be low. Therefore, the oxygen exhaust velocity at the anode should be high during the CO contamination feeding. The CO supply velocity was 0.13 mmol h1 when the CO concentration was 0.2 ppm in this experiment, whereas the permeated O2 velocity from the cathode and exhausted O2 velocity in the anode were 92 and 10 mmol h1, respectively. These results show that O2 in the anode exists in large excess compared with CO. Thus, permeated oxygen should play an important role in the oxidation of low-concentration CO via a non-electrochemical reaction, as reported in the literature [20e22].

2Pt þ O2 42Pt  O

(17)

Pt  O þ Pt  CO42Pt þ CO2

(18)

In addition, we examined the effect of CO in the one-way pass through which the fuel at the cell outlet was exhausted to the system. However, FCVs usually have a hydrogen circulation system to use hydrogen in the exhaust. Therefore, exhausted oxygen in the anode will also circulate, and it will make a possible contribution to mitigating the effect of CO, as in air-bleed systems. 5. Conclusion We have investigated the CO adsorption behavior on the anode during PEFC operation at around 0.2 ppm CO, the maximum concentration allowed in ISO 14687-2. Even at 0.2 ppm CO, the cell voltage decreased by 29 mV at the steady state under the condition of 0.11 mg cm2 anode platinum loading, 60  C cell temperature and 1000 mA cm2 current density. Under this condition, the CO coverage was estimated as 0.6 with both linear- and bridge-bonded CO adsorption. The amount of CO adsorption increased with inlet CO concentration, following the Temkin isotherm. When the fuel was switched to high-purity hydrogen after CO poisoning, the cell voltage apparently recovered, but most of the CO seemed to remain on the anode. This result means that the drop in the cell voltage for a recovered cell would occur earlier than that for a CO-free cell if a relatively high concentration of CO was fed to the anode. We also measured permeated oxygen from the cathode to the anode during cell operation. A large excess of O2 was observed in

the anode compared with CO. Permeated oxygen should play an important role in CO oxidation under low CO concentration conditions. FCVs usually have a hydrogen circulation system to use hydrogen in the exhaust. Therefore, permeated oxygen from the cathode will also circulate, and it will make a possible contribution to mitigating the effect of CO, as in air-bleed systems. Acknowledgments This study was supported by the “PEFC evaluation project” from the New Energy and Industrial Technology Development Organization (NEDO) of Japan. The authors would like to thank Shinjiro Nakamura and Naomi Tsurumi for their significant efforts in experimental support and discussions of the single-cell tests and gas analysis. List of symbols ACell (cm2) Geometric electrode area CCO,out,t (ppm) Cell outlet CO concentration at time t CCO2,out,t (ppm) Cell outlet CO2 concentration at time t fB Mole fractions of bridge-bonded CO fL Mole fractions of linear-bonded CO F Faraday constant (¼9.65  104 C mol1) FH2 (L h1) Fuel flow rate n (mol) Amount of CO adsorption NCO (mol m2) The number of Pt sites qBCO (mol)Bridge-bonded CO adsorption amount qLCO (mol)Linear-bonded CO adsorption amount qCO (mol m2) Cumulative CO adsorption amount per platinum surface area QCO (C) CO oxidation charge from CO stripping voltammetry Sf Anode stoichiometry SCO (m2) Whole platinum surface area vCO2,b,t (mol h1) CO2 exhaust velocity baseline vCO,cell,t (mol h1) CO adsorption velocity at time t vCO,out,s (mol h1) CO exhaust velocity at the steady state vCO2,out,s (mol h1) CO2 exhaust velocity at the steady state vCO,out,t (mol h1) CO exhaust velocity at time t vCO2,out,t (mol h1) CO2 exhaust velocity at time t Vm Mole volume (¼22.4 L mol1) qL CO coverage which considered only linear-bonded CO qLþB CO coverage which considered both linear- and bridgebonded CO References [1] A.L. Dicks, J. Power Sources 61 (1996) 113e124. [2] L. Barelli, G. Bidini, F. Gallorini, S. Servili, Energy 33 (2008) 554e570.

8

Y. Matsuda et al. / Journal of Power Sources 318 (2016) 1e8

[3] ISO 14687-2 (2012). [4] Y. Matsuda, Y. Hashimasa, D. Imamura, M. Akai, S. Watanabe, Rev. Automot. Eng. 30 (2009) 167e172. [5] H.A. Gasteiger, J.E. Panels, S.G. Yan, J. Power Sources 127 (2004) 162e171. [6] Y. Hashimasa, Y. Matsuda, M. Akai, ECS Trans. 26 (1) (2010) 131e142. [7] Petroleum Energy Center, Japan Automobile Research Institute, Engineering Advancement Association of Japan, The Japan Gas Association, Stage 2 JHFC Project (JHFC2) Final Report, 2011, p. 24. [8] National Renewable Energy Laboratory (NREL) website: http://www.nrel.gov/ hydrogen/cdp_topic.html#stations. [9] D. Imamura, D. Ebata, Y. Hashimasa, M. Akai, S. Watanabe, SAE Technical PaperseJSAE/SAE International Fuels and Lubricants Meeting, 2007-01-2010, 2010. [10] G.A. Camara, E.A. Ticianelli, S. Mukerjee, S.J. Lee, J. McBreen, J. Electrochem. Soc. 149 (2002) A748eA753. [11] C.G. Farrell, C.L. Gardner, M. Ternan, J. Power Sources 171 (2007) 282e293. [12] N. Wagner, M. Schulze, Electrochim. Acta 48 (2003) 3899e3907. [13] J. Divisek, H.-F. Oetjen, V. Peinecke, V.M. Schmidt, U. Stimming, Electrochim. Acta 43 (1998) 3811e3815. [14] M. Watanabe, S. Motoo, J. Electroanal. Chem. 60 (1975) 275e283. [15] S.J. Lee, S. Mukerjee, E.A. Ticianelli, J. McBreen, Electrochim. Acta 44 (1999) 3283e3293. [16] T.E. Springer, T. Rockward, T.A. Zawodzinski, S. Gottesfeld, J. Electrochem. Soc. 148 (2001) A11eA23. [17] J.J. Baschuk, X. Li, Int. J. Energy Res. 27 (2003) 1095e1116. [18] M. Murthy, M. Esayian, A. Hobson, S. MacKenzie, W.K. Lee, J.W. Van Zee, J. Electrochem. Soc. 148 (2001) A1141eA1147. [19] M. Murthy, M. Esayian, W.K. Lee, J.W. Van Zee, J. Electrochem. Soc. 150 (2003)

A29eA34. [20] W. Wang, J. Power Sources 191 (2009) 400e406. [21] J.X. Zhang, T. Thampan, R. Datta, J. Electrochem. Soc. 149 (2002) A765eA772. [22] T.V. Reshetenko, K. Bethune, R. Rocheleau, J. Power Sources 218 (2012) 412e423. [23] N. Wagner, E. Gülzow, J. Power Sources 127 (2004) 341e347. [24] M.S. Angelo, J. St-Pierre, K.P. Bethune, R.E. Rocheleau, ECS Trans. 35 (32) (2011) 167e178. [25] L.C. Perez, P. Koski, J. Ihonen, J.M. Sousa, A. Mendes, J. Power Sources 258 (2014) 122e128. [26] H. Igarashi, T. Fujino, M. Watanabe, J. Electroanal. Chem. 391 (1995) 119e123. [27] S.K. Das, A. Reis, K.J. Berry, J. Power Sources 193 (2009) 691e698. [28] M. Watanabe, S. Motoo, J. Electroanal. Chem. 206 (1986) 197e208. [29] H.P. Dhar, L.G. Christner, A.K. Kush, H.C. Maru, J. Electrochem. Soc. 133 (1986) 1574e1582. [30] M. Inaba, M. Sugishita, J. Wada, K. Matsuzawa, H. Yamada, A. Tasaka, J. Power Sources 178 (2008) 699e705. [31] Y. Hashimasa, T. Numata, K. Moriya, S. Watanabe, J. Power Sources 155 (2006) 182e189. [32] D.S. Forsyth, J. Chromatogr. A 1050 (2004) 63e68. [33] T. Ioroi, K. Yasuda, Y. Miyazaki, Phys. Chem. Chem. Phys. 11 (2002) 2337e2340. [34] K. Shinozaki, T. Hatanaka, Y. Morimoto, ECS Trans. 11 (1) (2007) 497e507. [35] S. Maass, F. Finsterwaldner, G. Frank, R. Hartmann, C. Merten, J. Power Sources 176 (2008) 444e451. [36] K. Broka, P. Ekdunge, J. Appl. Electrochem 27 (1997) 117e123. [37] J. Zhang, H.A. Gasteiger, W. Gu, J. Electrochem. Soc. 160 (2013) F616eF622.