Title: Review of ambient CO2 effect on anion exchange membranes fuel cells
Authors: Noga Ziv; William E Mustain; Dario R Dekel
This is the author manuscript accepted for publication and has undergone full peer review but has not been through the copyediting, typesetting, pagination and proofrea-ding process, which may lead to differences between this version and the Version of Record.
To be cited as: 10.1002/cssc.201702330
Review of ambient CO
effect on anion exchange membranes fuel
, William E. Mustain*[b]
and Dario R. Dekel*[a]
N. Ziv, Prof. D. R. Dekel
The Wolfson Department of Chemical Engineering and The Nancy & Stephan Grand Technion Energy Program (GTEP) Technion – Israel Institute of Technology
Haifa 3200003, Israel E-mail: email@example.com [b] Prof. W. E. Mustain
Department of Chemical Engineering University of South Carolina
Columbia, SC 29208, USA E-mail: firstname.lastname@example.org
Prof. Dekel received his M.Sc. in Chemical Engineering, and his PhD and MBA from Technion – Israel Institute of Technology. After working 10 years in the battery industry, he co-founded CellEra (today PO-CellTech), where he served as Vice President for R&D and Engineering, leading the development of Anion Exchange Membrane Fuel Cell technology. In 2015 Dr. Dekel joined the Technion, as an Associate Professor. Prof. Dekel published more than
40 patents and peer-reviewed papers on battery and fuel cell technologies. He currently holds several governmental and company research grants from the Israel, European Community, and bi-national Foundations in Europe and USA.
William E. Mustain received his Ph.D. in Chemical Engineering from the Illinois Institute of Technology in 2006. He then moved to Georgia Tech for his postdoctoral studies. He now runs the Laboratory for Electrocatalysts and Fuels as a Full Professor at the University of South Carolina. His group focuses on the development of novel catalysts for
electrochemical synthesis, acidic and alkaline fuel cells and electrolyzers, Li-ion and aqueous batteries, etc. He has been the recipient of several awards including the Illinois Institute of Technology Young Alumnus Award, U.S. Department of Energy Early Career Award and Fulbright Scholar Fellowship.
Over the past 10 years, there has been a surge of interest in anion exchange membrane fuel cells (AEMFCs) as a potentially lower cost alternative to proton exchange membrane fuel cells (PEMFCs). Recent work has shown that AEMFCs achieved nearly identical performance to state-of-the-art PEMFCs; however, much of that data has been collected while feeding CO2-free air or pure oxygen to the cathode. Usually, removing CO2 from the oxidant is done in order to avoid the detrimental effect of CO2 on AEMFC performance, due to carbonation where CO2 reacts with the OH- anions, forming HCO3- and CO32-. In spite of the crucial importance of this topic for the future development and commercialization of AEMFCs, unfortunately there have been very few investigations devoted to this phenomenon and its effects. Much of the data available in the literature is widely spread out and there currently does not exist a resource that researchers in the field, or those looking to enter the field, can use as a reference text that explains the co mplex influence of CO2 and HCO3-/CO32- on all aspects of AEMFC performance. The purpose of this review is to summarize the experimental and theoretical work that has been reported in the literature for the effect of ambient CO2 on AEMFCs. This systematic review intends to create a single comprehensive account of what is known regarding how CO2 behaves in AEMFCs to date, as well as identify the most important areas for future work in this field.
Anion-exchange membrane fuel cells (AEMFCs) have gained significant interest recently due to their potential advantages over the current technology based on proton-exchange membranes (PEMs). Among them, the high pH environment of AEMFCs enables, in principle, the use of low cost catalysts[2–5] and bipolar plate materials. It also facilitates the use of polyhydrocarbon-based chemistries as the ionomeric membrane, which in turn significantly reduces the fuel crossover in the cell. Furthermore, since the electrolyte is a polymer membrane rather than concentrated KOH solution – as is the case with alkaline fuel cell (AFC) technology – the issues of carbonate salt precipitation are completely avoided.
Anion-exchange membranes (AEMs) are made of polymeric materials containing fixed positively charged groups (as opposed to negatively charged groups in PEMs), which serve to transport anions through the membrane from the cathode electrode to the anode electrode (as opposed to transport of cations from anode to cathode in PEM fuel cells), Figure 1. In AEMFCs, the most common fuel that is oxidized at the anode is hydrogen and the most widely employed oxidant supplied to the cathode is oxygen. The oxygen is reduced through a four-electron process with two water molecules to form the OH- (hydroxide) that carries the charge through the electrolyte and reacts with the H2 at the anode (to form H2O; two molecules are formed per reacted H2, hence four per reacted O2). When pure oxygen is fed to the cathode, OH- alone is present and carries the charge. However, when ambient air is used instead of pure oxygen, carbonation processes occur where OH- as counter anion reacts with CO2 present in the air to produce CO32- (carbonate) and HCO3- (bicarbonate). This represents a significant challenge for AEMFC technology. The main result of this carbonation process is a significant decrease in the effective anion conductivity in the AEM[9,10] and, in turn, a reduction of AEMFC power output.[11,12]
The reduction in AEM effective anion conductivity may be due to several factors, among them are the larger size of the carbonate anions as compared to the size of the hydroxide anion, as well as the different mechanisms that are available for carbonate versus hydroxide anion transport in presence of water. For instance, in diluted aqueous solutions, HCO3-/CO32- have ~1/5 of the ionic conductivity of OH-.
Figure 1. Schematic representation of electrode reactions, ion and water transport in A) AEMFC and B) PEMFC. (Reprinted from Omasta et al.)
Due to this challenge, most present research in AEMs involves fuel cells operating on pure O2 or CO2-free air as the oxidizer in the cathode, avoiding the carbonation issue. In the past couple of years, some researchers have been able to achieve high AEMFC performance with cells operated on oxygen and air; however, performance data with ambient air represents only 3% of the cell performance data published in the AEMFC literature, as was recently reviewed. For practical applications of AEMFCs, the final goal of this technology requires the use of ambient air, which contains ~400ppm CO2. To do so, further research is necessary, aiming to understand the carbonation process and mitigate the detrimental effects of CO2 on the operation of AEMFCs. To date, several
studies have been directly or indirectly devoted to this issue; however, no effort has been done to compile all of the available information in this field. This review summarizes the research done so far on this subject, describing the implications of us ing CO2 -containing air on the AEM characteristics and on the performance of the AEMFCs.
2. Effect of CO2
When an AEM in its OH--form is exposed to ambient air (containing ~400ppm of CO2), the OH- anions present in the membrane are chemically transformed through acid-base reactions to create bicarbonate and carbonate anions, according to (forward reactions): OH⁻ + CO₂⇄ HCO₃⁻ ( 1 )
HCO₃⁻ + OH⁻⇄ CO₃²⁻ + H₂O ( 2 )
It is widely understood that these reactions occur quickly.[10,17–20] In fact, it can be problematic to accurately calculate their intrinsic rate since in relevant systems the transformation may be CO2 mass transport limited. Yanagi and Fukuta found that for an A201 AEM (Tokuyama Corp., Japan), the OH- concentration decreases by ca. 90% in the first 10 minutes of exposure to air, forming CO
32-. After that time, the CO32- concentration slowly decreases while the HCO3- concentration starts to increase. This is a result of reaction 1 (forward) and reaction 2 (backward) taking place when the OH- concentration is low, and the CO2 concentration remains constant. Over time, the OH- concentration continues to decrease until essentially all of it is consumed after 30 minutes. After 2 hours, it was shown that the AEM's anion content is a mixture of bicarbonate and carbonate, with ca. 60% HCO3- and 40% CO3 2-(Figure 2).
Figure 2. The change in alkaline anion content when an OH--form Tokuyama A201 AEM is exposed to air (wet conditions); OH-, CO32-, HCO3-. (Adapted from ref. ) [a]
Identical carbonation behavior was also found in a model developed by Myles et al. The authors divided the carbonation process into two regions: hydroxide depletion (corresponding to reactions 1 and 2), followed by bicarbonate accumulation (corresponding to backward reaction 2), until equilibrium is achieved. During this process, both carbonate and bicarbonate are present in the membrane. When equilibrium is reached, they concluded that CO32- is the dominant anion in an air-exposed AEM. However, experimental work shows that after long AEM exposure times to ambient air, the conductivity of an OH--form AEM decreased and reached that of AEMs that have been purposely exchanged to the HCO3--form.[23,24]
Though the rate of OH- reacting with gaseous CO2 is fast, it has become apparent that reaching the final equilibrium condition in a full AEM can take a relatively long time. Yanagi et al. reported that initial OH- anions were completely replaced by either CO32- or HCO3- in 30 minutes. Similarly, in a different experimental report, Suzuki et al. found that half of the initial amount of OH- anions was replaced in the first 10 minutes of exposure of the AEM to ambient air, but complete replacement of all OH- anions took 6 hours. This was consistent also with the work of Kizewski et al., which showed that OH- was fully replaced after 1 hour, and after 4 hours the AEM was in a mixed HCO3-/CO32- form. All the reported data on ex-situ AEM carbonation is summarized in Table 1.
One of the interesting findings in Table 1 is that different exchange conditions on the same AEM, i.e. Tokuyama A201 by Yanagi et al. and by Suzuki et al., affects the time required for full carbonation. The longer duration (6 hours compared to 30 minutes) needed to achieve full consumption of OH- in Suzuki’s work
was likely due to conditioning the AEM in ambient air, as opposed to immersing it in water that is exposed to air. It has been reported previously that AEMs display higher water uptake in liquid water than in humidified gas environment,[25,26] which leads to larger amounts of water within the polymer pores and easier access of dissolved CO2 to the functional groups. Therefore, CO2 absorption and OH- reaction kinetics are faster when the membrane is wet (not just fully humidified) – though it should be noted that neither condition truly reflects the local conditions in an operating AEMFC.
Republished with permission of Electrochemical Society, Inc, from "Anion Exchange Membrane and Ionomer for Alkaline Membrane Fuel Cells (AMFCs)", H. Yanagi and K. Fukuta, vol. 16, 2008; permission conveyed through Copyright Clearance Center, Inc.
Table 1. Summary of OH--form AEM neutralization to HCO3
Ref. AEM T Time until OH- is fully replaced[a]
Type of AEM exposure to CO2
Yanagi et al.
Tokuyama A201, immersed in water
20°C 30 min Immersion in water in contact with ambient air
Suzuki et al.
Tokuyama A201 R.T. 6 h Exposed directly to ambient air
Kizewski et al.
ETFE–based radiation–grafted AEM
--- 120 min Exposed directly to ambient air
Marino et al.
FAA-3 --- ~ 100 min (less than 10%) Exposed directly to ambient air
Divekar et al.
30°C 15 min Exposed directly to humidified ambient air (95% RH[b])
[a] Anion content in the AEM was measured by Ion exchanging to Cl
and titration based on Warder Method [b] RH - Relative Humidity
3. The effect of HCO3-
on ex-situ measured conductivity
It is important to first understand the effect of CO2-containing air (and thereby HCO3-/CO32- anions) on the AEM characteristics without the added effects of electric current and potential gradients of the fuel cell. This section addresses the literature data related to the effect of CO2 on the conductivity of AEMs outside a fuel cell (ex-situ). The impact of CO2 on an operating fuel cell (in-situ) will be explored later in Section 5.
To understand the effect of different anions on the ionic conductivity of the membrane, the link between the chemical propert ies of each anion and its mobility must be made. To begin, a general (approximate) expression for the diffusion coefficient of ion i in a solution can be found from the Stokes-Einstein relation:
( 3 )
where kB is Boltzmann's constant, T is the temperature, is the dynamic viscosity and ri is the ionic radius. In a membrane (M), the
diffusive relationship of species i is more complicated, and can be theoretically described by kinetic theory as the diffusion coefficient between two species, i (anion) and M (membrane), Equation (4):[27,28]
√ ( 4 )
where mi,M is the reduced mass of the two diffusing particles, ri,M is the collision radius (related to ionic radius) and Ni,M is the total
number density of the system (related to concentration of particles). This expression shows that the diffusion coefficient is inversely related to the square of the ionic radius; meaning that as the radius increases, the diffusion coefficient decreases significantly. The ionic mobility of species i, i, is related to the diffusion coefficient by the Einstein relation:
( 5 )
where qi is the ionic charge and Di is the ion's diffusion coefficient. Since qi is constant for each anion, the mobility is directly
proportional to the diffusion coefficient, and therefore inversely related to the square of the radius. Finally, an expression for membrane conductivity is given by:
where zi is the ionic charge, F is faraday's constant and Ci is the ion's concentration. Since zi and F are constants, the conductivity
depends mainly on the anion species concentration and its mobility.
As shown in the previous expressions, CO32- and HCO3- anions have a detrimental effect on the AEM's effective conductivity since they have a larger ionic radius than OH- anions, and therefore have lower diffusion coefficients and lower mobility. This can be seen experimentally in aqueous solutions where ionic mobilities were reported to be 20.64, 7.46 and 4.61 *108 m2s-1V-1 for OH-, CO
32- and HCO3- respectively. Furthermore, a mathematical model for ion transport in AEMs developed by Kiss et al. calculated that the ion-membrane diffusion coefficient of OH- is up to 1.3 times higher than the CO32- coefficient and 1.5 times higher than HCO3-. Theoretically, according to Equation (5), the ratio between diffusion coefficients of different anions should be the same as the ratio between their mobilities (directly proportional to each other). However, there may be some deviation from theory due to structural factors in the membrane; for example, the motion of ions in the polymer may be slowed down by their chemical interactions with the polymer framework, their association with the cationic fixed groups and morphological arrangement of the water regions. For instance, Amel et al. found that HCO
3- diffusion coefficient in an AEM is 90% lower than its diffusion coefficient in an aqueous solution.
In addition, the anion transport through the membrane is performed by several mechanisms including diffusion/migration, convec tion and the Grotthuss (proton hopping) mechanism, and their movement is strongly dependent on the membrane's level of hydration. The Grotthuss mechanism is available only for OH- (out of the 3 anions), and involves the breaking and forming of O-H bonds in the water molecules rather than physical movement of the anions. The only transport mechanisms available for HCO3- and CO32- are diffusion and convection which are dependent on the anion's size and charge (as described in Equation (4) and (5)) and theref ore, OH- conductivity is further increased compared to HCO3- and CO32-.
The reduction in conductivity is not dependent on ionic radius alone, but also on the number of hydration layers. The hydration layer is different for each anion and depends on the anion surface charge density. As the ionic charge is larger and ionic radius s maller, the surface charge density increases and therefore the hydration shell grows and the ion mobility decreases. The number of water molecules present in the hydration shell ranges from 4 around OH-, to 6.9 around HCO3-, and 8.7 to 9.1 around CO32-. Subsequently, the hydration radius for solvated OH-, CO32- and HCO3- in an aqueous solution has been reported as 3Å, 3.94Å and 5.6Å respectively. HCO3-, therefore, most likely has the biggest influence on the decrease in the effective conductivity of the AEM, since it has the highest hydration radius and carries a charge of only -1 (as opposed to CO32-). According to the previously mentioned model by Kiss et al., CO32- contributes most to the mixed-form membrane's conductivity given it has the highest concentration in the membrane and carries more charge since it is divalent. It should be noted again that inside the AEM other hydrodynamic and electrostatic factors exist that cause lower ion mobility than in aqueous solutions.
Another aspect of the effect of HCO3-/CO32- anions on the AEM effective conductivity is the dissociation equilibrium of anions and fixed cationic groups on the AEM, which may be considered as a representation of the thermodynamic barrier of the interaction s between the anions and fixed cationic groups. The higher the degree of dissociation, the greater the number of ions that are free to serve as charge carriers through the membrane. Based on theoretical calculations,[27,35] in an OH--form AEM, ca. 32% of the OH -anions present in the membrane are dissociated from the cationic groups, while in a mixed-form AEM (treated with 0.5M Na2CO3/0.5M NaHCO3 solution) roughly 40% of OH- and HCO3- anions are dissociated and only 20% of CO32-. Subsequently, the fraction of total charge that is free was calculated; in an OH--form AEM ~32% of total charge is free while that number is reduced to only 25% in the mixed CO32-/HCO3- AEM. A similar conclusion was reached by Amel et al., where their computational analysis of experimental results estimated that in their AEM in HCO3- form, 30-40% of the anions are free.
Figure 3 shows AEM conductivity as a function of the relative fraction of functional groups occupied by CO32- and HCO3-, as measured by Suzuki et al. It can be seen that as the fraction of carbonate and bicarbonate species in the AEM increases, the conductivity decreases, which is to be expected from the expression for the overall conductivity given by Equation (6) (the weighted average of individual conductivities of each anion species).
Figure 3. Conductivity of AEM (A201, Tokuyama Co.) at 50°C, 95% RH as a function of the equivalent fraction of carbonate ion species (CO32- + HCO3-) to quaternary-ammonium base. (Adapted from Suzuki et al.) [a]
Reprinted from Electrochimica Acta, Vol. 88, S. Suzuki, H. Muroyama, T. Matsui, and K. Eguchi, Influence of CO2 dissolution into anion exchange membrane on
In addition to the effect of the anion species, there are several structural and environmental factors that can influence the membrane's conductivity; i.e. ion exchange capacity (IEC), water uptake, and temperature. As the IEC increases, the density o f fixed cationic groups increases and therefore anion concentration in the membrane increases, which leads to higher conductivity.[30,36] Furthermore, HCO3- conduction relies strongly on the presence of loosely-bound water and hence higher water uptake (which in turn also depends on IEC) leads to higher conductivity. At higher temperatures, the anions' diffusion coefficient (and mobility) increases and therefore the AEM's conductivity increases (Equations (5) and (6)). In addition, as temperature rises, the solubility of CO2 in water decreases and so HCO3-/CO32- concentration in the AEM is lower, increasing the conductivity.[22,38,39]
The activation energy for anion transport can be used to understand the possibility of increasing the conductivity by increas ing the temperature. Arrhenius plots of conductivity as a function of temperature suggest that the activation energy was 1.2-1.5 times higher for HCO3-/CO32- conduction than for OH- conduction.[22,23,37] This result is consistent with the fact that the mobility of both CO32- and HCO3- are lower than OH-. Therefore, as expected theoretically, the conductivity of an AEM in its OH- form decreases immediately when in contact with CO2-containing gas, including ambient air (see Figure 4).[9,23,40] In practice, ex-situ measurements of AEM conductivity in their OH--form yield 1.5 upto 10 times higher conductivity than AEMs in their CO
32--form or HCO3--form (Table 2, Figure 5). Typically, researchers find that the ratio of OH- and HCO3-/CO32- AEM conductivities is between 2 and 5, with ~4 being the most common; the higher ratios have been reported in cases of very low conductivity AEMs. As mentioned previously, conductivi ty ratios in aqueous solutions are slightly lower than those in AEMs, mainly due to limited hydrated regions, interactions with the polymer structure and different transport mechanisms.
Table 2. AEM ionic conductivities in different anion forms
Ref. AEM T - [mScm-1 ] - [mScm-1 ] - [mScm-1 ]
CRC Handbook of
Chemistry and Physics Aqueous solution 20C 192 (1M) 68 (1M) 138 (1M) 2.8 1.4
Suzuki et al. Tokuyama A201 50°C ~70 ~15 --- 4.7
Janarthanan et al. Aminated trimethyl poly(phenylene) (ATMPP) 50°C ~36 --- ~10 3.6
Zhou et al.
Poly(arylene ether sulfone) containing fluorenyl groups and functionalized with benzyltrimethylammonium groups (QAPSF)
50°C ~36 --- ~24 1.5
Vega et al. Ralex AMH-PAD (Mega a.s.)
~27 ~6[a] 4.5[a]
MA-3475 (Lanxess Sybron) ~13.5 ~3[a] 4.5[a]
Excellion I-200 (SnowPure) ~21.5 ~2[a] 10.8[a]
AMB-SS (ResinTech) ~13.5 ~2.5[a] 5.4[a]
AMI-7001S (Membranes International) ~12 ~3[a] 4[a]
Unlu et al. Poly(arylene ether sulfone) (PSF) 25°C 21.21 --- 9.91 2.1
Kimura et al. Tokuyama A201 25°C 23.8 9.3[b] 2.6
Yanagi et al. Tokuyama A201 20°C 42 --- 10 4.2
Pasquini et al. PSU-TMA 25°C 12 0.8 1.6 15 7.5
PSU-DABCO 25°C 8.8 1.1 1.1 8 8
Liu et al. Self-crosslinked PPO
14.5 3.2 --- 4.5
PPO 28 7.2 --- 3.9
Self-crosslinked PS 26.3 6.5 --- 4.1
PS 42.5 9.8 --- 4.3
Xue et al. Crosslinked PPO-BTMA 20°C[d] 10.5 2.6 --- 4
PPO-BTMA 22.3 4.9 --- 4.6
Fukuta Tokuyama A201 40°C 62.1 15.2 --- 4.1
Wright et al. HMT-PMBI 30C[e] 23 8.1[a] 2.8
Ge et al.
HB-QPVBC (hyper-branched quaternized poly-4-(chloro- methyl)styrene)
20C[d] 39.6 17.7 2.2
60C[d] 84.5 34.5 2.5
Zha et al. Metal-cation- based AEM
30C[d] 28.6 6.4 4.5
Robertson et al.
Tetraalkyl ammonium crosslinkers based AEM 50C[d]
111 22 28 5 4
[a] Exchanged to mixed HCO3
form [b] Equilibrated with air containing 38% CO2 [c] Equilibrated with pure O2 [d] In water [e] 95% RH
Figure 4. Change in OH--form AEM (HMT-PMBI) conductivity with time from the beginning of exposure to ambient air containing CO2. (Adapted from Ref  with permission of The Royal Society of Chemistry. http://dx.doi.org/10.1039/c6ee00656f)
Figure 5. Conductivity data reported in the literature for AEMs in the three different anion forms. (Adapted from ref. [9,10,17,22,29,37,38,40– 49]
Another important consideration for AEMFC operation is what happens to the carbonation in the membrane when the gaseous CO2 is removed from the environment. It has been shown that the conductivity of the AEMs can be partially recovered – increasing back to 50% of its initial value (right after ion exchange) – after several hours of CO2 removal.  This occurs as dissolved CO2 is released from the AEM and HCO3- decomposes to create OH- (the reverse of Equation (1)).
Another aspect affecting AEM conductivity over time is the alkaline stability of the membrane functional groups. In many cases a temporal decline in conductivity of OH--form AEMs was measured when soaked in OH- solutions, as a result of the degradation of the membrane's cationic groups.[50,51] In contrast, the conductivity of AEMs soaked in HCO3-/CO32- solutions remained constant (though significantly lower) over time.[38,42] For example, Vega et al. measured a decrease of over 25% in conductivity of a membrane after 30 days in 1M KOH, while membranes immersed in 0.5M Na2CO3/0.5M NaHCO3 saw no measurable decline in the conductivity after 30 days. This suggests that HCO3-/CO32--form AEMs may have higher chemical stability. This makes sense from a chemistry perspective since carbonates are weaker nucleophiles than hydroxide, meaning that membranes are expected to be less prone to chemical attack (and hence a degradation in physical and electrochemical properties during operation) in carbonate environments than in hydroxide environments.
effect on electrochemical reactions in the electrodes
In addition to its effect on AEM conductivity, CO2 may also have an effect on the electrochemical reactions taking place at the electrodes. This section explores the data in the literature related to electrochemical measurements of these reactions and their response to CO2. The data related to the electrodes in an operating fuel cell is discussed in section 5.
4.1. Oxygen Reduction Reaction
The oxygen reduction reaction (ORR) in AEMFCs operating with pure oxygen is: -
( 7 )
However, in the presence of CO2 the ORR may also occur through:[38,53] - - ( 8 )
This most likely occurs as a 2-stage reaction where OH- produced according to Equation (7), and then reacts with CO2 according to Equations (1) and (2).
Several cyclic voltammetry experiments were conducted using Pt rotating disk electrodes (RDEs) in aqueous solutions of OH- and CO32-, saturated with oxygen. The kinetic current density of the ORR has occasionally been reported to be higher in CO32- solution than in OH- solution.[55–57] However, O2 solubility is much lower in CO32- solutionsthan OH- solutions, and both oxygen solubility and the oxygen diffusion coefficient diminish with increased salt concentration as well.[54,58] Additional results, such as a slight change in the Tafel slope, might indicate CO32-/HCO3-anions are adsorbed on the electrode and reduce electrode active area, decreasing in turn the observed ORR rate.[54,59] The result is a strongly negative impact on the reaction rate and the rate of the ORR decreases with higher concentrations of HCO3-/CO32-.
The observation of decreasing ORR reaction rate with increased carbonation was also observed in reacting environments more similar to operating AEMFCs. For example, Gunasekara et al. studied the effect of carbonation on the ORR at the interface with a membrane as opposed to the traditional tests in aqueous solutions; a special electrochemical cell was constructed, consisting of flat-tip micro electrodes pressed against an AEM. Oxygen gas was supplied to the cell through a porous glass filter. It was found that although the exchange current density with a carbonated AEM was slightly lower than in the case of an AEM with OH- only (carbonate-free), overall the carbonates have little effect on the intrinsic ORR kinetics. However, the study also reported a significant decrease in the O2 diffusion coefficient (one order of magnitude lower) with a CO32--form AEM compared to an AEM in the OH- form, which greatly affected the electrochemical reaction rate. This small effect of CO2 on the intrinsic ORR kinetics was confirmed experimentally by Inaba et al., who studied the effect of CO
2 in the cathode gas inlet on the cathode performance in a fuel cell. They found that in the range of 500-5000 ppm, adding CO2 in the cathode gas has little to no effect on the electrode overpotential (see Figure 6).
Figure 6. Effects of CO2 concentration in the cathode inlet gas on polarization curves of the AEMFCs at the anode and cathode. (Reprinted from Inaba et al.)
4.2. Hydrogen Oxidation Reaction
Since OH- is a direct reactant in the hydrogen oxidation reaction (HOR), it is reasonable to expect that its depletion and the emergence of carbonates might disrupt the anode kinetics and activity. The HOR in alkaline medium in a CO2-free environment is:[61,62]
- - ( 9 )
However, the HOR in the presence of HCO3-/CO32- (as part of the carbonate cycle) may proceed as follows:[63,64] - - ( 10 )
- ( 11 )
Similar to the case of the ORR, these oxidation reactions likely take place as a 2-stage reaction (evidence for this based on literature results will be presented below) including the reverse of Equation (1) and/or Equation (2), followed by Equation (9).[63,64]
As OH- is continuously produced in the cathode during fuel cell operation, high HCO3-/CO32- concentrations are formed in the anode, which creates a lower pH environment than in the cathode. This increases the anode potential relative to the cathode, though this is not necessarily manifest as a reduction in the cell voltage, because a junction potential of equal magnitude exists across the gradient as has been shown in similar membrane systems.[1,65] In fact, the shift in the anode potential can be rather significant. Wang et al. estimated that the pH shift can be as high as 6.1 pH units (~360mV) based on estimates of near-membrane concentrations. Recently, Krewer et al. used a physio-chemical model and found that the shift in the anode was closer to 3 pH units (~180mV) due to enrichment in CO2 near the anode.
Additionally, Inaba et al. measured the effect of CO2 on the electrodes in an operating AEMFC. The impact on each of the electrodes was determined using electrode voltammetry experiments. Agreeing with the discussion in the previous section, the cathode overpotential was hardly changed by the presence of CO2 in the cathode gas (100, 1000 and 5000ppm CO2 in O2). In contrast, the anode overpotential (operated with pure H2) was significantly affected by the CO2 fed to the cathode, especially at low current densities, as shown in Figure 6. However, some RDE experiments in aqueous solutions showed that the intrinsic kinetics (the exchange current density and Tafel slope) are not appreciably affected by the presence of carbonates.[63,64,68] These results imply that: i) the anode reaction with carbonate is most likely a two-stage process – first to decarbonate and release OH-, followed by OH -oxidation of H2; and ii) HCO3-/CO32- anions formed in the cathode accumulate in the anode during fuel cell operation, causing a decrease in both the conductivity and activity of OH- that work in concert to increase the anode overpotential.
All of these studies demonstrate the challenges involving CO2 enrichment close to the anode, and the importance of either its efficient release or the development of new strategies in the catalysts and membrane that minimize the conversion of OH- to carbonates in the presence of CO2 in the cathode. However, the HOR reactions in the presence of HCO3-/CO32- also show the potential to purge carbonates from the membrane during AEMFC operation by releasing CO2 and thus increasing the local AEM conductivity. This proposed purging mechanism will be discussed in Section 6. However, it can be said here that in an extreme case, it has been proposed to use these carbonate purging reactions purposefully to create electrochemical CO2 separation cells. These CO2 separation measurements have implied that CO2 is transferred through the AEM primarily by migration, and not gas-phase diffusion. Finally, with respect to the effect of the carbonation process on the electrochemical reactions over time, Gülzow et al. concluded that CO2 does not induce long-term degradation of the electrodes, since they observed a similar decrease in the electrodes' electrochemical activity both with pure and with CO2-containing gases during extended operation.
5. Effect of CO2
on fuel cell performance
The previous sections dealt with the effect of CO2 on the properties of individual components of the AEMFC. In this section we summarize the existing literature data for operating fuel cells exposed to CO2 in the cathode. Our analysis aims to describe the effect of CO2 on the AEMFC voltage, power density and overall cell resistance.
Taking into account the carbonation effect in the electrodes, and since current is dependent on the transport of anions through the membrane and catalyst layers, the AEMFC performance is expected to be affected by the anion species that are present in the cell. A recent study that reviews the AEMFC state-of-the-art indicates that 97% of the data reported on cell performance is based on pure oxygen or synthetic air (CO2-free air). Very few studies have directly reported the effect of CO2 on cell performance. The data in the literature can be divided into two parts: 1) fuel cells operating with pre-treated AEMs in the CO32-/ HCO3- form and O2/clean air (without CO2) as the cathode gas, and 2) fuel cells operating with uniformly treated AEMs (usually OH- form) and ambient air (containing CO2) as the cathode gas.
5.1. Influence of pre-carbonating AEMs to their CO32-/ HCO3- form on AEMFC performance
The available data in the literature unequivocally shows that the performance of H2/O2 fuel cells operating with HCO3-/CO32- form AEMs is significantly lower than that using OH--form AEMs. According to Suzuki et al., an AEMFC operated with H
2 and clean air with an AEM in the OH- form can have a cell voltage up to 1.3 times higher at the same current density (0.7V compared to 0.55V), and ca. 0% low r c ll r sistanc (0. Ωcm2 compar d to 0. 5Ωcm2) compared to a cell with an AEM in the HCO3- form. Moreover, Li et al. showed that cell resistance is up to twice as high in a HCO3- -form AEMFCs than the same systems in the OH--form (0.0 Ωcm2 v rsus 0.0 Ωcm2, including both the resistance related to ionic conductivity and the resistance related to the electrochemical reactions). Additionally, the authors showed that the fuel cell with the AEM in the OH- form reached a maximal power density 1.4 times higher than in the HCO3- form (610mW/cm2 compared to only 430mW/cm2).
5.2. Effect of CO2 in the cathode gas on fuel cell performance
As would be expected based on the discussions above, AEMFC operation with pure O2 or clean (CO2-free) air in the cathode leads to higher cell performance than with ambient air.[9,11,12,60,71,72] The data is summarized in Table 3, and Figure 7a demonstrates clearly by polarization curves the effect of using ambient air instead of clean air. For example, Li et al. measured a maximum power density of 610mW/cm2 with pure O2 and 410mWcm-2 with clean air, almost twice and 1.3 times, respectively, higher than the peak power density of 320mW/cm2 in a fuel cell operating with ambient air. This shows that performance is not hindered by lower O2 concentration in air alone, but also by the presence of CO2. In addition, the resistance related to ionic conductivity was 2.2 times higher and the resistance related to the electrochemical reactions was up to 4 times higher in a fuel cell with ambient air t han with high purity O2. These values are relatively close to those summarized in Section 5.1, strengthening the previous observations that i) the changes in anion composition during AEMFC operation significantly affect the ionic conductivity and electrochemical react ions; and ii) the pre-treatment of the AEM with different anions does not significantly influence in long-term operation.
Moreover, an increase in CO2 concentration in the cathode gas was found to cause an increase in cell resistance and a decrease in cell potential.[10,60] Gottesfeld et al., showed that by switching the cathode gas from CO2-free air to ambient air the cell voltage dropped by nearly 50% (at constant current density of 400mAcm-2). Similarly, other studies reported drops of ca. 30% when switching from clean/synthetic air to ambient air.[11,71] Figure 7b clearly illustrate the immediate effect of CO2 on the cell performance. In addition, Inaba et al. found that the cell resistance of an OH--form fuel cell increased by a factor of 1.7 when adding 100ppm CO
2 to pure O2, and by 2.3 when adding 5000ppm CO2. Similarly, the cell voltage recorded in separate polarization curves decreased by 18% and 44% when adding 100ppm and 5000ppm CO2, respectively (see Table 3). This means that even when present in a small amount such as 100ppm (which is much lower than ambient air), CO2 greatly effects the anion composition in the AEM as the HCO3 -/CO32- concentration increases at the expense of OH-.
However, based on the ability for AEMFCs to self-purge carbonates, it is expected that the operating current density would play a role in dictating the CO2 threshold that a cell is able to tolerate, though that effect has been recently elucidated theoretically by Krewer et al., but not yet experimentally. The effect of CO2 in the cathode gas on the long-term performance of the fuel cell has not been thoroughly explored. Tests performed so far indicate no significant degradation in performance, though those experiments have been run for only 50 hours. It is imperative to study the long-term effect of ambient air compared to CO
2-free air on AEMFC performance, as carbonate species anions may contribute to the AEM chemical stability (as discussed previously in Section 3).
In contrast to results presented above, a few experiments found that adding CO2 in a large amount to the O2 in the cathode gas of CO32--form AEM fuel cell increases current density (or cell voltage when load is constant).[38,43,53,54] However, the cells studied when this was the case had very low performance, and therefore, the impact of the carbonation on the cell performance may be affected by other non-related factors. However, all experiments at reasonable current and power densities involving CO2 in the cathode show that CO2 has a definitive negative effect on fuel cell performance.
Table 3. Effect of CO2 on AEMFC performance
Pre-treatment Operation conditions Fuel cell performance
Ref. AEM Anion form
Tempe-rature Cathode Gas
Cell Voltage[a] [V] Cell Resistance [mΩcm2 ] Max. Power Density [mWcm-2] Limiting current density [mAcm-2] Inaba et al. Tokuyama A901 OH- 50°C O2 0.62 [b] 230[b] O2 + 100ppm CO2 0.51 [b] ~400[b] O2 + 1000ppm CO2 0.42 [b] ~450[b] O2 + 5000ppm CO2 0.35[b] ~520[b] Piana et al. Benchmark commercial membrane
N/A 50°C Clean air 0.8[c]
320 600 Ambient air 0.4[c] 83 260 Li et al. Self-aggregated quaternary ammonium polysulfone (aQAPS-S8) OH -60°C O2 0.82[b] 16[c] 610 2200 Clean air 0.77[b] 22[c] 410 1500 Ambient air 0.74[b] 35[c] 320 1400 Topal et al. Tokuyama A201 OH -55°C O2 22 60 Synthetic air 16 45 Ambient air 8 30 Fukuta et al.
Tokuyama A201 HCO3- 50°C Clean air 290 1100
Ambient air 125 700 Wright et al. HMT-PMBI OH -60°C O2 180 Ambient air 70
Lang et al. AFN AEM CO32-/ HCO3 - 26°C O2+CO2 [e] 0.1[d] 0.54 44°C 0.15[d] 0.68 55°C ~0.5[d] ~0.52 Zhou et al. QAPSF CO3 2-25°C O2 2.4 O2+CO2 [e] 4 Unlu et al. QAPSF CO3 2-25°C O2 2.4 O2+CO2[e] 4.4
[a] Extracted from polarization curves [b] @200mAcm-2 [c] @100mAcm-2 [d] @4mAcm-2 [e] Volume ratio of 1:2
Figure 7. Effect of CO2 on fuel cell performance: (a) Fuel cell polarization curves using ambient air compared to clean air/pure O2. (Data adopted from [9,11,12,71,72]) (b) Immediate effect of switching (at t=0) from CO2-free air to ambient air in the cathode of a fuel cell operating at constant current density. The data shown was normalized by the initial power density or voltage. (Data adapted f rom [11,73]
6. Increasing AEMFC performance by reducing HCO3-
concentration via self-purging
As mentioned in previous sections, it is possible to remove HCO3-/CO32- anions from the AEM during the operation of the fuel cell through the release of CO2 gas in the anode. This section aims to explain the details of this process and presents experimental proof from the literature.
The self-purging mechanism is the process by which the HCO3-/CO32- can be removed from the AEMFC under high current conditions, which may help improve cell performance, has been demonstrated in several studies.[8,13,16,17,72] The process may be explained as follows: at relatively high current densities, a large amount of OH- anions are formed as a result of the high rate of ORR in the cathode (Equation (7)). The existing charge, which includes HCO3- and CO32-, is transported through the AEM to the anode electrode. It is not known whether these ions react directly with H2 (Equations (10) and (11)) or whether OH- remains the active anion (Equation (9)) and under hydroxide consumption equilibrium is shifted in the reverse direction of Equations (1) and (2). As discussed above, anode studies alone suggest that it is the latter. In either case, the result is that CO2 gas is readily released to the anode and purged from the system. This concept is schematically shown in Figure 8. At increased current densities, the rate of OH- production in the cathode overcomes the rate of CO2 absorption into the membrane, and therefore the overall result is the increase in OH -concentration at the expense of HCO3-/CO32- concentration. It is possible that in higher CO2 concentrations, or at low operating current densities, the rate of CO2 absorption will surpass the production of OH- and self-purging will not occur.
Figure 8. Schematic representation of AEMFC current density on the self-purging process in the AEM. On the left, at low current density OH- is slowly formed in the cathode but is quickly converted to HCO3-/CO32-. On the right, at high current density OH- surpasses the carbonation process, and HCO3-/CO32- are released at the anode as CO2.
Evidence of the self-purging mechanism was observed in several ways. FT-Raman analysis of a CO32--form AEM performed by Adams et al. after operation in a H2-air AEMFC revealed that the CO32- content of the AEM decreased. Also, Yanagi et al. measured the anion species concentration in a HCO3--form AEM before and after operation in H2-clean air AEMFC. While prior to operation the AEM contained only HCO3-, after 6 hours operation at constant voltage (0.2V) it contained only OH- and CO3-2 (~1:1 ratio), showing that HCO3-/CO32- can be purged from the AEM.
In addition, CO2 formation and release in the anode exhaust was shown in several experimental studies.[10,13,43,74] For example, by flowing the anode product stream through a Ca(OH)2 solution, Unlu et al. observed the formation of a milky precipitate (CaCO3) seconds after the start of fuel cell operation (using O2+CO2 in the cathode at volume ratio of 1:2). The author showed that the amount of CaCO3 precipitate increased with time as the cell operated. Other works used analytical methods such as infrared,[69,72,75] mass spectrometry[13,74] and gas chromatography to measure the concentration of CO2 in the anode exhaust gas of an AEMFC in operation. Suzuki et al. found that the flux of CO2 in the anode exhaust of AEMFCs increases with CO2 concentration in the cathode and also with cellcurrent density. It has been found that CO2 emission and cell current are closely related; as current increases, the flow of CO2 at the anode outlet increases,[72,75] meaning less CO2 is accumulated in the AEM.
Moreover, several experiments showed that the Ohmic resistance of AEM fuel cells in the HCO3-/CO32- form decreases at high current densities.[10,12] Inaba et al. measured a resistance approaching that of the original AEM in the OH--form after high current density operation, proving the replacement of HCO3-/CO32- by OH- through the self-purging mechanism was possible. Similarly, Kimura et al. measured an increase in ex-situ conductivity of an AEM which was exposed to CO
2 by applying external current in a pure O2 environment.
Several theoretical and experimental[13,69,72] studies have suggested that during fuel cell operation, an equilibrium state of HCO3 -/CO32- concentration exists in the AEM – with the equilibrium state being determined by the current density. During fuel cell operation with ambient air, when the current density is increased there is a temporary increase in the CO2 concentration in the anode exhaust (Figure 9). According to this, the amount of CO2 that is released after a change in current density is the difference in CO32- content of the AEM between the two equilibrium states, and at constant current, the amount of CO2 adsorbed in the cathode is equal to the amount released in the anode.[13,72]
Figure 9. CO2 emission from anode exhaust of an AEMFC operated with ambient air while stepwise increasing current density. (Adapted from ref. ) [a]
Based on the data above, it can be suggested that a start-up protocol in the operation of the fuel cell should include operating at high current density/low voltage (i.e. 0.1-0.2V) for a brief period, in order to purge HCO
3-/CO32- anions from the AEM, and begin the operation with an AEM in the OH--form in ord r to yi ld high r p rformanc . Similar protocol was also succ ssfully us d to ‘activat ’ AEMFCs operated with air containing low concentrations of CO2 (filtered air).
7. AEMFC models simulating HCO3-
Republished with permission of Electrochemical Society, Inc, from "In-situ Observation of CO2 through the Self-purging in Alkaline Membrane Fuel Cell
Mathematical models to study the mechanism of mass and heat transfer are also of great interest and importance. An analytical model of an AEMFC can provide direct relations between performance, design parameters and operating conditions, and it can be time and computationally efficient. Several models have been created to simulate and understand water management during operation of an AEMFC.[76–81] Those models were validated at low and moderate current densities, where the carbonate process effect may be significant. A recent study reported a numerical model able to simulate AEMFC experimental polarization data as well as transient data for cells operating at high current densities.[82,83] However, all of these models do not consider the effect of atmospheric CO2 on the AEM and assume pure OH--form AEM and pure O2 cathode gas; hence, they are not able to take into account the carbonation process that may occur when the cell is operated with ambient air containing CO2. There have been a few models that do account for the effects of CO2, which can be divided into two types: i) models dealing with the properties of an AEM which is not in an operating fuel cell (ex-situ); and ii) models dealing with the effect of CO2 on the fuel cell performance. The comparison between different aspects of each model is presented in Table 4.
7.1. Models including the effect of CO2 on AEMs ex-situ
In order to address the carbonation issue, modeling techniques have been employed to simulate the concentration profile and ion exchange process between the OH- and HCO3-/CO32- anions in the AEM.[16,21,22,27,30,67] The conductivity of AEMs equilibrated with O2 containing 1ppm CO2 was comparable to that of an air-equilibrated AEM at low temperatures, meaning that even a small concentration of CO2 may cause significant decrease in OH- concentration and membrane conductivity (Figure 10a). However, at fuel cell-relevant temperatures (> 60oC, and especially > 80oC), 1ppm CO2 had much less of an effect. It should also be noted that this model does not account for electrochemical reactions and the self-purging mechanism.
The anion-membrane diffusion coefficients were calculated computationally and as expected, OH- had the largest value, followed by CO32- and HCO3-. Subsequently, the membrane's conductivity and its dependence on water content was calculated for AEMs in the OH- form and mixed anion form (containing CO
32- and HCO3-, after exposure to ambient air). Results showed higher conductivity for OH--form, and strong dependence of both forms' conductivity on water level in the membrane, in agreement with experimental results (Figure 10b).
Figure 10. Modelling results for the effect of operating conditions on AEM ex-situ conductivity: (a) Model predictions by Grew et al. for Tokuyama A201 AEM ionic conductivity in equilibrium with gas containing 0 ppm (pure O2), ---1 ppm, and --385 ppm CO2.[a] (b) Model predictions for AEM conductivity at different hydration levels by Kiss et al. The hydroxide () and mixed (combined carbonate and bicarbonate) (---) form are plotted.[b]
Table 4. Summary of computational models for AEM properties.
on Model predicts
Reference Model Ex-situ /
Fuel cell RH T Anion concentration Water content/ water uptake Current density range [mAcm-2 ] CO2 issue Grew et al.
A Dusty Fluid Model (DFM) Predicting OH -Conductivity
Republished with permission of Electrochemical Society, Inc, from "Effects of Temperature and Carbon Dioxide on Anion Exchange Membrane Conductivity ", K.N. Grew, X. Ren, D. Chu, vol. 14, Copyright 2011; permission conveyed through Copyright Clearance Center, Inc.
Republished with permission of Electrochemical Society, Inc, from "Carbonate and Bicarbonate Ion Transport in Alkaline Anion Exchange Membranes", A.M. Kiss, T.D. Myles, K.N. Grew, A.A. Peracchio, G.J. Nelson, W.K.S. Chiu, vol. 160, Copyright 2013; permission conveyed through Copyright Clearance Center, Inc.
Grew et al.
Ionic Equilibrium and Transport model Ex-situ
Grew et al.
Effects of Temperature and CO2 on Conductivity Ex-situ
Kiss et al. Carbonate and Bicarbonate Ion Transport Ex-situ
Myles et al.
Transient ion exchange of AEM exposed to ambient air
Siroma et al.
Modeling of the Concentration Profile of Carbonate Ions in operating FC (steady state)
Huo et al. 3D half-cell model for water management in the anode
Deng et al.
3D Transient half-cell model of water transport in the anode and performance
Jiao et al. 3D multiphase non-isothermal model of FC performance and water management
Jiao et al.
Analytical model of cell performance Fuel cell
Deng et al.
3D model with interfacial effect and water management optimization
Krewer et al.
Modeling of the Concentration Profile of different anions in air-based operating FC (steady state)
Wrubel et al.
transient, spatially-averaged theoretical model for anion concentration during air-based FC operation
[a] Degree of dissociation [b] Including diffusion coefficients
the OH- depletion region followed by the bicarbonate accumulation region. In the OH- depletion region, CO2 is absorbed into the membrane but does not significantly accumulate in the membrane bulk, since itis primarily consumed close to the air-membrane interface. As time goes by, CO2 reacts with OH- closer to the center of the membrane where OH- is still present, and HCO3- concentration closer to the air interface increases. Finally,in the bicarbonate accumulation region, there is no OH- in the membrane and only CO32- and HCO3- anions exist. At this point, dissolved CO2 is starting to accumulate throughout the bulk of the membrane. This is consistent with theoretical models for an operating fuel cell, which showed increased concentration of dissolved CO2 in membrane at low current densities.
When simulating the effect of temperature on the anion composition in the AEM, Grew et al. showed that at high temperature the concentration of HCO3− and CO32- decreases as a result of the decrease in CO2 solubility, and the membrane's conductivity increases. A similar positive effect of high temperature on FC performance was found in a model for operating AEMFC, which is discussed in the next sub-section. These results may indicate that, from this particular point of view, operation of AEMFCs at higher temperatures is recommended.
7.2 Models including the effect of feeding CO2 to operating AEMFCs
Very few models have been developed to study the effect of CO2 on the performance of an operating fuel cell. Siroma et al. studied the membrane's self-purging mechanism and attempted to predict the anion distribution in a HCO3--form AEMFC while operating with O2. The results were in agreement with CO2 emission experiments performed by the authors; after current density is increased, anion concentrations in the AEM change, and a new equilibrium state exists, which includes OH-,HCO
3- and CO32. The authors showed that as the current increases, the overall fraction of CO32- in the membrane decreases, and the spatial gradient of CO32- concentration is larger due to faster OH- formation in the cathode. In addition, it was found that for a given constant current density, as the AEM thickness increases, the equivalent ratio of CO32- in the membrane decreases. This is due to the differences in hydration profiles across the membrane obtained in AEMs of different thicknesses, which shifts the equilibrium towards OH- formation.
Recently, a model was developed by Wrubel et al., aiming to predict and demonstrate the transient changes in carbonate concentrations in the AEM (which was initially exposed to ambient air) during the start of operation in a fuel cell. The results showed an increase in OH- concentration with time until reaching an equilibrium state, as implied by experimental fuel cell CO
2 emission results. As the current density increases, the time to reach equilibrium is shorter and the final OH- concentration is higher.
Another recent, and more comprehensive, model was developed by Krewer et al. to predict anion concentration profiles in the AEM of an operating fuel cell, as well as their impact on fuel cell performance. It has been shown that when operating an AEMFC with ambient air, CO2 is absorbed at the cathode and transported through the membrane as HCO3- and CO32- anions. The concentration profile reveals a carbonate species enriched zone closer to the anode and high OH- concentration near the cathode (Figure 11). This is a result of the slow rate of CO2 gas release at the anode outlet. The model further shows that as the current density increases, the enrichment zone becomes smaller (and the membrane contains less CO2) due to faster production of OH-. The authors suggest that for current densities higher than 0.5 Acm-2, less than 10% of the anions are HCO
3- or CO32-. The study concluded that the entire impact of CO2 on AEMFC performance can be significantly reduced by operating the cell at current densities above 1 Acm-2.
Figure 11. Computational model's results for profiles of equivalent anion ratio of (a) OH-, (b) CO32- and (c) HCO3- across the membrane between anode (0 µm) and cathode (28 µm) for different current densities (100, 500, 1000 and 1500 mAcm-2). (Reprinted from Krewer et al.)
Anion exchange membrane fuel cells are a promising fuel cell technology that can enable the future use of very low-cost cell and system level components that are not possible in PEMFCs. Recent work has shown that AEMFCs can achieve comparable initial performance to state-of-the-art PEMFCs. However, one of the major challenges in the development of this technology is that during fuel cell operation with ambient air, the OH- anions that
are formed in the cathode electrode react with the ambient CO2 to form bicarbonate (HCO3-) and carbonate (CO32-) anions, affecting the anion composition in the AEM and therefore, significantly reducing the cell performance. Despite the widespread view that the presence of carbonates is an issue that must be resolved in order to achieve high performance AEMFCs, there has been a shortage of studies in the literature that explicitly focus on the effect of CO2 in this emerging type of fuel cells. The lack of targeted investigations pointed to the need to summarize what has been done and reported in the literature to date in this field. This work presents, for the first time, a comprehensive summary of the effect of CO2 on both AEMs and AEMFCs, reported in the literature through the last decade.
Within the scope of the work done so far in the subject, researchers have established theoretical and experimental frameworks that allow the field to begin understanding the carbonate dynamics in the AEM and AEMFC – from their formation and balance with the gas environment, to their dissociation from the stationary membrane cationic groups and mobility through the polymer electrol yte. Following are the main observations we can summarize in this work from the data in literature regarding the carbonation process in AEMFCs:
The CO2 from the ambient air reacts at the cathode electrode to form bicarbonate and carbonate anions. The anion profiles across the AEM depend on the operating conditions of the fuel cell as well as on the cell characteristics.
The presence of HCO3- and CO32- has a negative impact on ionic conductivity of the AEMs, mainly due to the lower charge mobility and slower transport mechanisms as compared to the OH-. Most of the work done so far report a ratio of 2-4 between ex-situ AEM conductivities in the forms of OH- versus HCO
The presence of HCO3- and CO32- appears to negatively affect both reactions, but the anode to a larger extent. The intrinsic kinetics and reaction mechanisms for the ORR and HOR do not appear to be changed, but the presence of HCO3-/CO32- leads to mass transport limitations in the cathode and a (likely significant) reduction in the OH- activity at the anode.
The presence of CO2 in the oxidant gas leads to up to 50% lower AEMFC voltage and power density.
It has been found, both theoretically and experimentally, that HCO3- and CO32- can be removed during high current operation through a self-purging mechanism – with higher currents being more effective in removing them from the system. The HCO3 -/CO32- is expelled from the anode electrode in the form of gaseous CO2. A recent numerical model predicts that operating an AEMFCs above 1 Acm-2 would significantly reduce the amount of HCO
3-/CO32- in the AEM, and therefore make the effect of the ambient CO2 in the air on the cell performance negligible.
However, it should be noted that the field of carbonation in AEMFCs remains in its infancy and there is a significant amount of work that remains:
Despite the understanding of membrane dynamics, very little is known about the carbonation and de-carbonation in the electrodes, whose structure is highly complex and the mass transport dynamics cannot be easily untangled experimentally or theoretically.
It is also unknown what the limitations are for the self-purging mechanism; this is particularly important in the light of the many studies that were unable to completely remove the effects of carbonation – even after prolonged exposure and operation with CO2-free cathode gases. This raises new questions about the reversibility of carbonation as well as the dynamics of carbonate removal and association with the membrane groups as its concentration is depleted.
Additionally, it is truly unknown how stable are AEMFCs operating during time under ambient air at cathode feed. The effect of temperature on the carbonation dynamics and impacts is also unclear.
Though the answers to these questions are not yet available, it is important for researchers in the field to begin systematically exploring the CO2 and HCO3-/CO32- effects in AEMFCs, so they can begin to develop new strategies to limit their negative impact and design ultra-high performance AEMFCs operating on ambient air.
This work was partially funded by the Grand Technion Energy Program (GTEP); by the European Union's Horizon 2020 research and innovation program [grant number 721065]; by the Ministry of Science, Technology & Space of Israel through the M.era-NET Transnational Call 2015, NEXTGAME project [grant number 3-12948]; by the 2nd Israel National Research Center for Electrochemical Propulsion (INREP2-ISF); by the Ministry of National Infrastructure, Energy and Water Resources of Israel [grant no. 3-13671 and as part of the scholarship program for undergraduate and graduate students in the field of energy]. The contributions of W.E. Mustain to this manuscript was made possible by funding from US DOE Early Career Program (Award Number DE-SC0010531).
Keywords: CO2 • carbonation • hydrog n fu l • ambi nt air • anion xchang m mbran fu l c lls
 J. R. Varcoe, P. Atanassov, D. R. Dekel, A. M. Herring, M. A. Hickner, P. A. Kohl, A. R. Kucernak, W. E. Mustain, K. Nijmeijer, K. Scott, et al., Energy Environ. Sci.2014, 7, 3135–3191.
 S. Lu, J. Pan, A. Huang, L. Zhuang, J. Lu, Proc. Natl. Acad. Sci.2008, 105, 20611–20614.
 H. A. Miller, F. Vizza, M. Marelli, A. Zadick, L. Dubau, M. Chatenet, S. Geiger, S. Cherevko, H. Doan, R. K. Pavlicek, et al., Nano Energy2017, 33, 293–305.
 M. Alesker, M. Page, M. Shviro, Y. Paska, G. Gershinsky, D. R. Dekel, D. Zitoun, J. Power Sources2016, 304, 332–339.
 H. A. Miller, A. Lavacchi, F. Vizza, M. Marelli, F. Di B n d tto, F. D’Acapito, Y. Paska, M. Pag , D. R. D k l, Angew. Chemie2016,
128, 6108–6111; Angew. Chemie Int. Ed.2016, 55, 6004–6007.
 D. R. Dekel, ECS Trans.2013, 50, 2051–2052.
 D. Dekel, in Encycl. Appl. Electrochem., Springer New York, New York, NY, 2014, pp. 33–45.
 L. A. Adams, S. D. Poynton, C. Tamain, R. C. T. Slade, J. R. Varcoe, ChemSusChem2008, 1, 79–81.
 A. G. Wright, J. Fan, B. Britton, T. Weissbach, H.-F. Lee, E. A. Kitching, T. J. Peckham, S. Holdcroft, Energy Environ. Sci.2016, 9, 2130–2142.
 S. Suzuki, H. Muroyama, T. Matsui, K. Eguchi, Electrochim. Acta2013, 88, 552–558.
 M. Piana, M. Boccia, A. Filpi, E. Flammia, H. A. Miller, M. Orsini, F. Salusti, S. Santiccioli, F. Ciardelli, A. Pucci, J. Power Sources
2010, 195, 5875–5881.
 G. Li, Y. Wang, J. Pan, J. Han, Q. Liu, X. Li, P. Li, C. Chen, L. Xiao, J. Lu, et al., Int. J. Hydrogen Energy2015, 40, 6655–6660.
 S. Watanabe, K. Fukuta, H. Yanagi, in ECS Trans., The Electrochemical Society, 2010, pp. 1837–1845.
 T. J. Omasta, L. Wang, X. Peng, C. A. Lewis, J. R. Varcoe, W. E. Mustain, J. Power Sources2018, 375, 205–213.
 D. R. Dekel, J. Power Sources2018, 375, 158–169.
 Z. Siroma, S. Watanabe, K. Yasuda, K. Fukuta, H. Yanagi, J. Electrochem. Soc.2011, 158, B682.
 H. Yanagi, K. Fukuta, in ECS Trans., ECS, 2008, pp. 257–262.
 J. Kizewski, N. Mudri, R. Zeng, S. Poynton, R. C. T. Slade, J. R. Varcoe, in ECS Trans., The Electrochemical Society, 2010, pp. 27– 35.
 M. G. Marino, J. P. Melchior, A. Wohlfarth, K. D. Kreuer, J. Memb. Sci.2014, 464, 61–71.
 A. G. Divekar, A. M. Park, Z. R. Owczarczyk, S. Seifert, B. S. Pivovar, A. M. Herring, ECS Trans.2017, 80, 1005–1011.
 T. D. Myles, K. N. Grew, A. A. Peracchio, W. K. S. Chiu, J. Power Sources2015, 296, 225–236.
 K. N. Grew, X. Ren, D. Chu, Electrochem. Solid-State Lett.2011, 14, B127.
 J. Yan, M. A. Hickner, Macromolecules2010, 43, 2349–2356.
 T. P. Pandey, A. M. Maes, H. N. Sarode, B. D. Peters, S. Lavina, K. Vezzù, Y. Yang, S. D. Poynton, J. R. Varcoe, S. Seifert, et al.,
Phys. Chem. Chem. Phys.2015, 17, 4367–4378.
 Q. Duan, S. Ge, C.-Y. Wang, J. Power Sources2013, 243, 773–778.
 Y. S. Li, T. S. Zhao, W. W. Yang, Int. J. Hydrogen Energy2010, 35, 5656–5665.