Hydrogen Crossover in PEM and Alkaline Water Electrolysis

F502 Journal of The Electrochemical Society, 165 (7) F502-F513 (2018)
0013-4651/2018/165(7)/F502/12/$37.00 © The Electrochemical Society
Hydrogen Crossover in PEM and Alkaline Water Electrolysis:

Mechanisms, Direct Comparison and Mitigation Strategies
P. Trinke,1,= P. Haug,2,= J. Brauns,2 B. Bensmann, 1,z
R. Hanke-Rauschenbach,1
and T. Turek2
1 Institute
of Electric Power Systems, Leibniz Universität Hannover, 30167 Hanover, Germany
2 Institute
of Chemical and Electrochemical Process Engineering, Clausthal University of Technology, 38678
Clausthal-Zellerfeld, Germany
This study provides a direct comparison of hydrogen crossover in PEM (Nafion 117) and alkaline water electrolysis (Zirfon) at a
temperature of 60◦ C applying state-of-the-art separating unit materials. To this end, occurring crossover mechanisms are described
first, before experimental data of the anodic hydrogen content are shown in dependence of current density, system pressure and
process management strategy. The results suggest that permeation in PEM electrolyzers is mainly governed by diffusion due to a
supersaturated concentration of dissolved hydrogen within the catalyst layer, showing a share of 98% of the total permeation flux at
1 A cm−2 and atmospheric pressure. Permeation in alkaline electrolyzers also exhibits a significant influence of supersaturation, but
the overall crossover is mainly influenced by mixing the electrolyte cycles, which makes up a share of 90% at 0.7 A cm−2 and 1 bar.
Generally it becomes evident that hydrogen permeation across the separating unit is more than one order of magnitude smaller in
alkaline electrolysis, which is mainly a consequence of the significantly lower hydrogen solubility in concentrated KOH electrolyte.
Finally, this study concludes with an assessment of the impact of separating unit thickness and provides mitigation strategies to
reduce hydrogen crossover.
© 2018 The Electrochemical Society. [DOI: 10.1149/2.0541807jes]
Manuscript submitted March 14, 2018; revised manuscript received April 19, 2018. Published May 16, 2018.
Proton exchange membrane (PEM) and alkaline water electrolysis water. Ito et al.9 reported a decrease of the measured hydrogen frac-
(AEL) represent two promising technologies, which are capable of the tion if a platinum based reversed catalyst coated membrane (CCM)
future production of renewable hydrogen. It has already been demon- was used for fuel cell and electrolysis operation. However, it is im-
strated that both technologies can react quickly to dynamic power portant to emphasize that recombination catalysts only reduce the risk
profiles from renewable energy sources.1,2 Generally, both electrol- of safety issues, but do not address the origin of crossover directly.
ysis technologies are operated in a similar range of parameters to Two further negative side effects of gas crossover, degradation10,11
split water into the constituents hydrogen and oxygen (Eq. 1). Thus, and reduction of Faraday efficiency,8,12 are still present. However, es-
temperatures between 50 and 80◦ C or system pressures up to 30 bar pecially degradation, which has been observed in PEM electrolysis,
represent the current state of the art.3 should not be neglected as the trend toward thinner membranes may
further enhance gas crossover. So far, for AEL no publication could
H2 O −→ H2 + 0.5O2 [1]
be found that establishes a link between gas crossover and electrode
The main difference between these two technologies lies in the ap- degradation, although this dependency also could exist. Nevertheless,
plication of either an acid or alkaline electrolyte. In PEM electrolysis the safety problem still represents a major issue in AEL.
protons are used for the charge transport as it is shown in the anodic Most recent publications on gas crossover in PEM electrolysis
and cathodic half cell reactions (Eqs. 2 and 3). are focused on the investigation of the current density influence. It
could be shown so far that hydrogen and oxygen crossover increases
Anode : H2 O −→ 2H+ + 0.5O2 + 2e− [2] linearly with the applied current density.5,13,14 The publications on
+ −
Cathode : 2H + 2e −→ H2 [3] AEL in contrast are mostly focused on the influence of separator
and membrane materials, as well as process conditions on the anodic
On the contrary, hydroxide ions are responsible for the charge ex- hydrogen fractions. Furthermore, strategies are provided, which may
change in AEL (Eqs. 4 and 5). be used for an improved anodic gas purity.7,15,16
Anode: 2OH− −→ 0.5O2 + H2 O + 2e− [4] The present contribution firstly provides a summary on the occur-
− −
ring crossover mechanisms in PEM and alkaline water electrolysis.
Cathode : 2H2 O + 2e −→ H2 + 2OH [5] Subsequently, experimental data of the influence of current density,
However, both technologies face similar issues, which need to be system pressure and various process management possibilities on the
solved for a successful coupling with renewable energy sources. Thus, anodic hydrogen content is shown for both electrolysis technologies.
gas crossover represents one of the main issues in PEM and alkaline In order to create comparable crossover data, the experiments of both
water electrolysis so far. Especially the permeation of hydrogen into technologies are carried out with state-of-the-art separating unit ma-
the anodic half cell is of special importance, as explosive gas mix- terials (PEM: Nafion 117, AEL: Zirfon) and at identical process tem-
tures can form in the part-load operation range of the electrolyzers, peratures. The obtained results are then used to identify the most
e.g. Refs. 4–7. However, this problem can be solved relatively easily influential crossover mechanism in PEM and alkaline electrolysis. Fi-
by the application of recombination catalysts as it has already been nally, the study provides an estimation of the influence of separating
demonstrated for PEM electrolysis. Grigoriev et al.8 showed that the unit thickness on hydrogen crossover and makes proposals on possible
anodic hydrogen content could be significantly reduced using a recom- crossover mitigation strategies.
bination catalyst, which was applied at the back side of the porous
transport layer (PTL). A further improvement of the anodic gas pu- Experimental Setup and Method
rity could be achieved when the recombination catalyst was placed in
the product gas conduits downstream of the gas separators. Another Cell design.—PEM electrolysis.—The PEM water electrolysis ex-
possibility is the utilization of electrocatalysts, which also promote periments in this study were conducted with a commercially avail-
the oxidation of hydrogen and recombination with evolved oxygen to able electrolysis cell (Sylatech Analysetechnik GmbH, type ZE 200),
which is shown in Fig. 1a. The cell has a circular design with an
active area of 62 cm2 without flow field structures. On the anode side
=
These authors contributed equally to this work. a titanium mesh serves as a water distributor. Between the mesh and
z
E-mail: [email protected] the CCM a PTL is incorporated, which consists of sintered titanium
Downloaded on 2018-05-16 to IP 137.207.120.173 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
Journal of The Electrochemical Society, 165 (7) F502-F513 (2018) F503
(a) cooling step of the gas after the outlet of the first separator. The water
of the anode is cycled, whereas on the cathode the separated water is
drained.
Pressure control valves regulate the absolute cathodic gas pressure
to 1, 10 and 20 bar, whereas on the anode side pressures of 1 and
10 bar are applied. The error of this regulation is less than ± 0.2 bar.
Every three minutes a micro-GC (Agilent 490) takes a gas sample of
the anodic product gas to measure the concentration of hydrogen in
oxygen. This micro-GC is equipped with a 10 m long 5Å molesieve
column and a thermal conductivity detector for permanent gas sepa-
ration and detection. Test gas mixtures with hydrogen concentrations
(b) of 0.1, 1 and 2.5 vol.% in oxygen (accuracy of reading ± 2%, Linde)
were used for calibration of the GC. The calibration measurements
indicated a standard deviation of ± 0.006 vol.%. Furthermore, it is
possible to directly feed the electrolysis cell with oxygen (Linde 5.0)
and hydrogen (Linde 5.0) to the corresponding cell inlet with an accu-
racy of ± 1% of the reading. This gas feed is mandatory for reference
measurements, which will be described later in this study.
Alkaline electrolysis.—The alkaline test cell was connected to a

Greenlight test station (E40), which continuously supplies the anodic
and cathodic half cell with a 32 wt% KOH electrolyte solution at
a flow rate of 0.350 ± 0.002 L min−1 , a temperature of 60◦ C and
Figure 1. Cell designs of the applied PEM (a) and alkaline (b) electrolysis absolute system pressures of 1, 10 and 20 bar. Here, both half cell
cells (pictures are not scaled). compartments are always operated at equalized pressure. With the
given volumetric flow rate, the electrolyte of each half cell is replaced
70 times per hour, which is defined as the recirculation rate within this
fibers. In the cathodic half cell a porous graphite plate is implemented,
work. The electrolysis power is supplied by an Ametek Sorensen XG
whereas an O-ring is used for sealing of the electrolysis cell. The CCM
12–140 power supply, which is used for galvanostatic measurements
is based on a Nafion 117 membrane, which was manufactured by HIAT
with an accuracy of ± 0.2% of the rated output current.
gGmbH with anode and cathode catalyst loadings of 2 mgIr cm−2 and
Downstream of the electrolysis cell the gas-liquid mixtures are fed
1 mgPt cm−2 , respectively. The dry Nafion membrane has a thickness
to gas separators, where the evolved gaseous products are removed
of approximately 180 μm, the porosity amounts to ε = 0.37 (water
from the electrolyte solution. The rising gas flows are then cooled
volume) and the channel diameter is roughly d = 2.5 nm.a,17 The
down to approximately 25◦ C to reduce the water content of the product
whole cell was encased with thermal insulating fabric for isothermal
gases. The condensed water is collected in a tank and fed back into the
operating conditions.
gas separators. The remaining gaseous water content is then further
reduced by desiccant dryers, before the composition of the product gas
Alkaline electrolysis.—The AEL experiments were conducted with
streams is analyzed in an online gas chromatograph (Agilent 7820A)
a custom-built single electrolysis cell, which consists of two end
every 5 minutes, which is also equipped with a molesieve column and
plates, current collectors, electrode mounts, electrodes and a sepa-
a thermal conductivity detector. Calibration of the GC was carried out
rator for keeping the product gases apart. Fig. 1b depicts a choice of
with test gas compositions of 0.02, 0.2 and 1 vol.% hydrogen in argon,
these cell components. The end plates are made of nickel coated 304
which could be determined with a maximum error of ± 0.01 vol.%.
steel and are used for the electrolyte supply and removal. Addition-
The liquid electrolyte from the gas separators, however, is led back to
ally, each end plate provides a circular electrolyte volume of roughly
the mixing point of the electrolyte cycles in order to compensate the
300 mL. The electrolysis voltage can be applied through current col-
concentration difference caused by the electrode reactions (Eqs. 4 and
lectors out of 316Ti steel, which are connected to the electrode mounts
5). Besides the classical electrolyzer operation with mixed electrolyte
via bolted connections.
cycles, this test station also possesses the possibility to separate the
The active electrodes themselves consist of quadratic Nickel (Alloy
cycles by changing the valve position of the 3/2-way valves, which
201) expanded metal with a geometrical area of 100 cm2 and are
are shown in the flow sheet in Fig. 2b. It can further be seen in the
welded on the electrode mounts of the same material. These electrodes
flow sheet that it is possible to feed the cell via Bronkhorst mass flow
are directly pressed onto a circular Zirfon Perl UTP 500 separator with
controllers with hydrogen (Linde 5.0) to the cathodic and oxygen
a geometrical area of 227 cm2 , so that a zero-gap cell arrangement is
(Linde 3.5) to the anodic half cell with an accuracy of ± 0.5% of the
achieved. The manufacturer AGFA specifies the separator thickness
desired setpoint.
with δsep = 500 ± 50 μm, the separator porosity with ε = 0.5 ± 0.1
and the pore diameter with d = 150 ± 50 nm.18 Furthermore the
cell contains 3D-printed polypropylene inlays, which are necessary
Methods.—In this study the stationary anodic hydrogen contents
for the electrical insulation of the current collectors from other cell
at a fixed temperature of 60◦ C, absolute pressures of 1, 10 and 20 bar
components.
and various current densities in the range of 0.05 to 0.9 A cm−2 were
determined by online gas chromatography for PEM and alkaline water
Test station.—PEM electrolysis.—A Greenlight test station (E100)
electrolysis. In both experimental setups stationary data was achieved
supplies the anodic compartment of the PEM electrolysis cell with
in such way that a specific current density was applied to the elec-
150 ± 5 mg min−1 deionized water at a temperature of 60◦ C. The gal-
trolysis cell and chromatograms were recorded simultaneously. The
vanostatic steps are applied by an Ametek Sorensen XG 6-220 power
current density was kept constant until the measured hydrogen frac-
supply with an accuracy of ± 0.2% of the output current reading. The
tion remained unchanged for at least 1 h before it was changed to the
flow sheet of the test station is shown in Fig. 2a. On each side the gas-
next operating point. For determination of the hydrogen content with
water mixture is separated by two gas separators and an intermediate
separated electrolyte cycles in AEL, the cycles were remixed after
each operating point to account for the arisen electrolyte concentra-
a
porosity and diameter were measured for a Nafion 115 membrane tion difference.
Figure 2. Schematic flow sheets of the PEM (a) and AEL (b) test stations applied in this study.
Similarly, experiments were carried out in which the PEM and (Eq. 9).
alkaline electrolysis cells were disconnected from the power supply, i H 2 Ael
perm
but fed with hydrogen and oxygen according to their volumetric flow N H2 = [9]
rates representing a current density range from 0.05 to 0.9 A cm−2 . In 4F 1 − H2 Asep
the following these experiments are referred to as reference measure-
ments, which have only been conducted at atmospheric pressure and Pressure drop method.—For the PEM setup further crossover mea-
60◦ C. surements were conducted with the so called pressure drop method.
With the experimental data from the electrolysis and reference Therefore, the cathode is pumped up with hydrogen gas. After reach-
measurements it is possible to derive the hydrogen permeation rate ing the targeted pressure, a valve on the cathode side is closed. The
through the membrane or separator while presuming that the anodic pressure drop is recorded by a pressure sensor (P-30: 0–40 bar, Wika).
catalyst materials are inactive for the oxidation of hydrogen. Thus, all The gas leakage rate of the applied cell was negligibly low in compar-
the permeating hydrogen can be measured in the anodic product gas ison to the crossover. Consequently, the measured pressure decreasing
stream. For the AEL this is only valid for experiments with separated rate is proportional to the hydrogen crossover that can be calculated
electrolyte cycles. Additionally, it has to be assumed that the oxygen by the cathodic mass balance. This method can be classified as a
permeation rate is negligibly small compared to the evolution rate of dynamical pressure measurement with a constant volume.21
oxygen. This assumption can be made for PEM electrolysis as the
oxygen permeation through a Nafion membrane is reported to be at
least two times smaller than its hydrogen counterpart.19 For AEL this Crossover Mechanisms
assumption is also admissible as the hydrogen diffusion coefficient in For both electrolysis technologies several possibilities exist, which
32 wt% KOH at 60◦ C is about three times larger than that of oxygen.20 cause hydrogen crossover. In general these permeation routes can
The measured water-free anodic hydrogen fraction H2 can then be be divided into diffusive and convective mass transfer mechanisms.
described according to Eq. 6. This chapter will give an overview on the occurring mass transfer
perm mechanisms and will rather point out, which of them are technology
N H2 specific.
H2 = perm [6]
NOano
2
+ N H2
Diffusion.—One of the possible crossover mechanisms is the dif-
perm fusion of electrolysis products across the membrane or separator into
Within Eq. 6 NH2 denotes the hydrogen permeation flux, whereas
the anodically evolving oxygen flux is represented by NOano . This the opposite half cell compartment for both technologies. Generally,
2
oxygen flux in turn can be obtained by insertion of the applied current the products may diffuse through the solid and aqueous phase of the
density i into Faraday’s law (Eq. 7) for the anodic half cell reaction. separating unit. However, it is reported that diffusion through the
solid phase of a fully hydrated Nafion membrane is roughly one or-
i der of magnitude smaller compared to its liquid phase.22,23 Similar
NOano
2
= [7] information can be found for Zirfon as the diffusion through the sep-
4F
arator’s solid phase is also assumed to be negligible. Additionally, the
With these two equations it is now possible to derive the hydrogen separator is further presumed to be impermeable for gas bubbles at
permeation flux from the experimentally determined anodic hydrogen atmospheric pressure.24 Thus, it is comprehensible for both technolo-
content. Therefore, Eq. 7 needs to be inserted into Eq. 6 and solved gies that only species that are dissolved in water or KOH solution are
perm
for the hydrogen permeation flux NH2 . considered for diffusion across the separating unit. This is typically
done by application of Fick’s law (Eq. 10) in PEM4,25–28 and alkaline
perm i H 2 water electrolysis:16
N H2 = [8]
4F 1 − H2 cH2
NHdiff = DHeff2 [10]
2
δsep
It has to be kept in mind that this equation is only valid if the
separating unit surface area equals the geometrical area of the elec- Here, cH2 represents the dissolved hydrogen concentration gradi-
trodes. Otherwise Eq. 8 needs to be multiplied by the ratio of electrode ent across the separating unit with the thickness δsep , whereas DHeff2 de-
and separating unit area to account for the additional permeation area notes the effective diffusion coefficient of hydrogen in the membrane
or separator. The estimation of the effective diffusion coefficient in the mathematical description of this transport mechanism commonly
polymer electrolyte membranes29 or porous media30 is typically done Darcy’s law (Eq. 17) is applied, e.g. Refs. 24, 25, 28 and 37.
by correction of the molecular diffusion coefficient in the aqueous K sep p
solvent DH2 with the porosity ε and tortuosity τ of the separating unit dp
N H2 = SH2 pHcat2 [17]
(Eq. 11). η δsep
ε Here, K sep denotes the permeability of the separating unit, η is the
DHeff2 = DH2 [11] dynamic viscosity of the solvent, whereas p describes the absolute
τ
pressure difference between the cathodic and anodic compartment.
In PEM electrolysis the concentration gradient across the mem- The concentration of dissolved hydrogen is again estimated by the
brane cH2 can be estimated with the cathodic concentration of dis- insertion of Henry’s law (Eq. 13). The hydraulic permeability K sep
solved hydrogen cHcat2 as the anodic hydrogen concentration cHano 2
is of porous media such as membranes can be estimated by the Hagen-
approximately zero. However, this assumption becomes also applica- Poiseuille (Eq. 18) or Kozeny-Carman (Eq. 19) equations.38
ble for AEL if the electrolyzer is operated with separated electrolyte
cycles. εd 2
K sep = [18]
0 32τ
cH2 ≈ cHcat2 − >

cHano [12]
2 ε3
K sep = [19]
Henry’s law (Eq. 13) states that a species’ dissolved concentration K koz a 2 (1 − ε)2
is directly proportional to its partial pressure in the gas phase. There-
fore, this approach can be applied for the calculation of the cathodic Here, ε represents the porosity, d the pore diameter, τ the tortuosity
dissolved hydrogen concentration. Within the following Eq. 13 SH2 and a the specific surface area of the separating unit, while K koz de-
denotes the hydrogen solubility in the solvent, whereas the cathodic notes the Kozeny constant, which depends on the porous media.39 For
hydrogen partial pressure is represented by pHcat2 . a rough estimation of the hydraulic permeability K sep in this work the
Hagen-Poiseuille equation (Eq. 18) is used. The necessary porosities
cHcat2 = SH2 pHcat2 [13] and pore diameters of both applied separating units are given within
the setup section. In order to make a worst case estimate the tortuosity
Data of the hydrogen solubility in pure water was published by is chosen to be τ = 1.5 for both systems. Therewith the permeability
Young et al.31 for atmospheric pressure conditions. Further data for for the Nafion membrane and Zirfon separator are approximated to
pressures ranging from 25 to 1000 atm can be found in the publi- be 5 · 10−20 m2 and 2 · 10−16 m2 , respectively. These estimates match
cation by Wiebe and Gaddy.32 However, literature for the hydrogen literature data of both separating units quite well, e.g. Refs. 17 and 40.
solubility in concentrated potassium hydroxide solution is scarce. Hence, AEL with a porous separator is theoretically more prone to
Ruetschi and Amlie33 reported data for various electrolyte concen- convective permeation. But it has to be noted that it is also more con-
trations at a temperature of 30◦ C and atmospheric pressure, whereas venient to mix the electrolyte cycles at identical pressure levels, which
Knaster and Apel’baum34 provided further values at temperatures of is also an important reason for balanced pressure operation. However,
21, 45 and 75◦ C. if it is assumed that commonly applied back pressure control valves
The cathodic hydrogen partial pressure can be obtained if it is are capable of controlling the anodic and cathodic pressures with an
assumed that the cathodic oxygen partial pressure is negligible and accuracy of roughly 1%, it becomes comprehensible that a differential
that the hydrogen is saturated with water vapor. Then the following pressure across the separator may be formed. Thus, it is conceivable
Eq. 14 applies, where pcat denotes the total cathodic pressure: that either dissolved hydrogen or oxygen could convectively be trans-
pHcat2 = pcat − pHcat2 O [14] ported into the opposite half cell.
In contrast, literature suggests that no pressure-driven permeation
The calculation of the water vapor pressure may be performed by is evident for Nafion membranes,23,41 which is supported by the low
application of the Antoine equation with parameters for pure water, calculated permeability of the membrane used in this study. This al-
e.g. Ref. 35. However, the estimation of the water vapor pressure above lows the operation of PEM electrolyzers under asymmetrical pressure
a potassium hydroxide solution necessitates the empirical correlations conditions. Here, the cathode is typically pressurized, whereas the
by Balej.36 anodic cycle stays at atmospheric pressure, which can be favorable
Finally, the diffusional hydrogen flux across the separating unit in terms of energy demand, e.g. Refs. 26, 42, and 43. However, if
can be described with the following Eq. 15 if the aforementioned alternative membrane materials are applied, this operation mode may
assumptions apply and Eq. 13 is inserted into Eq. 10. also promote a permeation flux as the resulting pressure-driven water
pHcat2 flow carries dissolved hydrogen into the anodic compartment.28
NHdiff = DHeff2 SH2 [15]
2
δsep Electro-osmostic drag.—A further possibility to cause convective
The product of the effective diffusion and solubility coefficient permeation is the electro-osmotic drag. Due to the electric field and
DHeff2 SH2 is frequently provided in form of the permeability coefficient the associated movement of ions, the electroneutral solvent can be
K H2 , which is a classical material property for separating units. dragged with them across the separating unit. Hence, dissolved gas
may also be transported through it.26,28,44 Thus, in PEM electrolysis
Convection.—Convection represents another general cause of dissolved oxygen may be dragged along with the transport of protons
crossover, which can be divided into several further mechanisms. from the anodic into the cathodic half cell. In contrast, the electro-
Generally, convective mass transport is mathematically expressed by osmotic drag could also be capable of reducing hydrogen permeation
Eq. 16. There, vsolv describes the velocity of solvent (PEM: water, as it may transport dissolved hydrogen back to the cathode. On the
AEL: KOH solution) moving perpendicular to the separating unit, contrary, OH− ions are responsible for the charge transport in AEL.
whereas cH2 denotes the concentration of dissolved gas within the According to the half cell reactions (Eqs. 4 and 5) these hydroxide
solvent. ions are transported from the cathodic to the anodic compartment,
which therefore may enhance hydrogen and reduce oxygen crossover.
NHconv
2
= vsolv cH2 [16] However, no publication was found that could confirm or quantify this
mechanism. A rough estimation of the electro-osmotic crossover flux
can be carried out with the following Eq. 20:
Differential pressure.—One possible reason for convective perme-
ation is the transport of electrolyte and dissolved species across the drag SH2 pHcat2 n drag i
separating unit due to the presence of total pressure gradients. For N H2 = [20]
csolv F
Here, csolv is the concentration of solvent within the separating solved form or to the gas-liquid interface of gas bubbles present in
unit and n drag is the electro-osmotic drag coefficient, which describes the electrode boundary layer.53 However, dissolved hydrogen may
the ratio between the flux of dragged solvent molecules to that of the also be transported through the separating unit of the electrolysis
charged ions. Jacobson et al.45 compared drag coefficients of proton- cell.5
conducting Nafion membranes to that of a hydroxide-conducting As theory of the aforementioned diffusive and convective crossover
Tokuyama A201 membrane for the application in acid and alka- mechanisms only supposed hydrogen equilibrium concentration
line fuel cells. Their results showed a smaller value for the anion- (Henry’s law), supersaturation enhances these crossover mechanisms
conducting membrane, which may be attributed to the different charge in PEM and zero-gap alkaline electrolysis. However, the influence of
carrier or water domain. A temperature-dependent correlation for the supersaturation on crossover can only be accounted, if the previously
estimation of the drag coefficient in a Nafion membrane used for PEM mentioned equations are expressed in their concentration forms in-
electrolysis can be found in the publication by Onda et al.46 stead of the gas pressure expressions. Then Eq. 21 can be used for
the calculation of the dissolved gas concentration within the catalyst
Electrolyte mixing.—The electrochemical reactions in AEL (Eqs. 4 layer in PEM electrolysis:5
and 5) cause a change in electrolyte concentration since water is i
+ kL pHcat2 SH2
consumed at the cathode, whereas it is produced in the anodic half cHcat2 = 2F
eff
[21]
DH
cell. Therefore, the anodic and cathodic electrolyte cycles are usually kL + 2
δsep
mixed together to balance this concentration gradient.47 However,
this process management leads to a decrease of the resulting product Here kL denotes the mass transfer coefficient, which includes sev-
gas purity as the electrolyte is saturated with dissolved electrolysis eral transport and transfer steps: beginning with the desorption of the
products. Thus, the electrolysis cell is continuously fed with dissolved electrolysis product from the catalyst particles up to the transfer into
hydrogen and oxygen from the gas separators, where they are then the gas phase within the pore space. For the PEM catalyst layers it is
able to outgas. Of course, separating the electrolyte cycles prevents suggested that this mass transfer coefficient is significantly affected
this crossover mechanism at all. But due to the shift in the anodic and by the diffusion of the dissolved gas from the catalyst particles to the
cathodic electrolyte concentrations the cell efficiency may decrease, pore space through the ionomer. Already small limitations can lead to
as the electrolyte conductivity is reduced with prolonged time. a significant increase of the dissolved gas concentration.5 Addition-
ally, Eq. 21 reveals that supersaturation also increases toward higher
Supersaturation.—It is generally assumed for gas evolving elec- current densities. So, it is necessary to consider that the concentration
trodes that the electrolysis products are produced in dissolved form48 of dissolved gas is not solely a function of gas solubility, system pres-
before gas bubbles grow at active nucleation sites. Nucleation sites sure and the electrode specific mass transfer coefficient, but also of
are small electrode surface irregularities, which depend on the ma- current density: cHcat2 = f ( pHcat2 , SH2 , kL , i).
terial and its roughness.49 For the nucleation sites to become ac-
tive a sufficient deviation from equilibrium concentration of the Summary and comparison of crossover mechanisms.—The pre-
generated product is mandatory.50 Therefore, the electrolyte be- viously described crossover mechanisms of both technologies are
comes supersaturated, which describes a higher concentration of summarized and directly compared in Table I. It is stated, which
dissolved hydrogen within the electrode boundary or catalyst layer impact each individual crossover mechanism has on the overall
than it would be expected through Henry’s law (Eq. 13). The ex- crossover flux of the respective technology. The assessment of the
istence of this supersaturated concentration of dissolved species individual influences is supported by a direct comparison of the key pa-
could already be proven experimentally, e.g. Refs. 51 and 52. rameters controlling the mechanisms. Therefore, the crossover equa-
Consequently, the subsequent mass transfer of dissolved hydrogen tions are given in their concentration forms.
is mainly controlled by two competing mechanisms. Thus, dis- For diffusive crossover through the separating unit every rele-
solved hydrogen is either transported to the electrolyte bulk in dis- vant parameter effects a higher permeation rate for PEM electrolysis
Table I. Comparison of the crossover mechanisms with key parameters and evaluation of the impact on the overall gas crossover of the respective
technology. If no special information given, stated values are valid for a temperature of 60◦ C and in case of AEL for a KOH concentration of
30 wt%.
Mechanism Equation: Parameter/Symbol/Unit PEM (Nafion) AEL (Zirfon)

cat ( p cat ,S ,k ,i)
cH H2 L
= DH
H
Diffusion NHdiff
2
eff 2
2
2
δsep very high low
thickness/δsep /μm 20–250 < 500
solubility coeff./SH2 /mol Pa−1 m−3 3.8e-6a,33 3.6e-7b,33
diff. coeff. in solvent/DH2 /m2 s−1 8.9e-9c,54 > 3.3e-9c,20
Convective
Differential pressure
dp K cat ( p cat , S , k , i) p
NH2 = ηsep cH low possibly high
2 H2 H2 L δsep
hydr. permeability/K sep /m2 5e-20d 2e-16d
cat ( p cat ,S ,k ,i)
cH
drag H L n i
NH2 = 2 c2solv 2 >
H drag
Electro-osmotic drag F lowe even lowere
drag coeff./n drag /– 2.845 –4.546 > 1.3f,45
solvent conc./csolv /mol m−3 55e3g ≈ 56e3g
Electrolyte mixing NHmix
2
= f ( pH
cat , S , k , i, V̇ )
2
H2 L no mixing very high
a value for 30◦ C in pure water.

b value for 30◦ C in 32.3 wt% KOH.
c diff. coeff. for water and 30 wt% KOH – the eff. coeff. can be estimated via Eq. 11.
d calculated within this work via Eq. 18.
e absolute convective drag may be lower for AEL, but due to opposite ion migration hydrogen permeation is increased in AEL and decreased in PEM.
f value for a Tokuyama A201 anion exchange membrane.
solv
solv ≈ Msolv .
g estimated by the following equation: c
Figure 3. Measured anodic hydrogen content as

a function of current density for the PEM (a) and
alkaline (b) electrolysis setup at a temperature
of 60◦ C, different system pressures and process
management strategies.
compared to alkaline cells. The membrane is thinner and the diffu- vidual crossover mechanisms of both technologies is shown. Finally,
sion coefficient of hydrogen in pure water is larger than in a KOH the chapter is concluded by a study of the influence of the separating
solution. Certainly, the solubility of hydrogen in pure water is also unit thickness on the anodic hydrogen content.
one order of magnitude higher than in a 30 wt% KOH electrolyte.
For lower concentrations of KOH the diffusion and solubility coeffi-
cients increase toward the values of water. Consequently, the diffusive Anodic hydrogen impurity.—Figs. 3a and 3b summarize the mea-
crossover increases in AEL with lower concentrated potassium hy- sured anodic hydrogen fractions of the experiments conducted in this
droxide electrolytes. work for a current density range from 0.05 to 1 A cm−2 , pressures
The convective crossover is divided into the three different mecha- ranging from 1 to 20 bar, a temperature of 60◦ C and different process
nisms: differential pressure, electro-osmotic drag and electrolyte mix- management possibilities. Additionally, the results of the reference
ing. As it was discussed previously, dissolved hydrogen can be carried measurements without electrolysis operation are shown.
into the anodic half cell due to a pressure-driven water or electrolyte Firstly, it becomes obvious for both electrolysis technologies that
flow. Therefore, low hydraulic permeabilities are essential to avoid higher applied current densities lead to a decrease of the anodic hy-
this crossover mechanism. In this case, the Zirfon separator applied drogen contamination. This can be explained with a rising oxygen
in AEL shows a 3 to 4 orders of magnitude higher hydraulic perme- evolution rate NOano2
toward higher current densities, which dilutes the
ability than the Nafion membrane. Although hydrogen solubility in permeating hydrogen and therefore reduces its fraction (s. Eqs. 6 and
concentrated alkaline electrolytes is low, the estimated permeability of 7).
the Zirfon separator allows a distinct crossover at moderate pressure It can further be seen in both experimental data sets that an increase
gradients already. of the operating pressure effects a drastic increase of the anodic hy-
In contrast, the second convective crossover through the separating drogen content. Thus, in PEM electrolysis a pressure change from 1
unit caused by the electro-osmotic drag is higher for PEM electroly- to 20 bar leads to an increase of the anodic hydrogen content from
sis as the water drag coefficient and solubility is larger in this case. 0.654 to 2.08 vol.% at a current density of 0.7 A cm−2 . Accordingly,
However, the electro-osmotic drag is assumed to reduce hydrogen, in AEL equal operating conditions caused an increase from 0.018 to
but increase oxygen crossover in PEM electrolysis. This is different 0.130 vol.% with separated electrolyte cycles, whereas mixed cycles
for AEL as the ions migrate into opposite directions in the two tech- yielded a change from 0.178 to 2.564 vol.%. So, it becomes clear
nologies. Nevertheless, the influence of this crossover mechanism is that alkaline process management possibilities reveal distinct differ-
assumed to be small, since the estimation only suggests a small flow ences in the resulting anodic hydrogen fraction. This huge deviation
of dragged water or electrolyte with low concentrations of dissolved is caused by the mixing of the anodic and cathodic electrolyte cycles.
gas. However, at high current densities and high system pressures Thus, depending on the separation performance of the gas separators,
this mechanism can become more important as the concentration of dissolved or even gaseous hydrogen is transported into the anodic
dissolved gas is increased. half cell where it contaminates the evolving oxygen. This effect be-
Since mixing of the anodic and cathodic electrolyte cycles is not comes more relevant toward rising system pressures as the amount
performed in PEM electrolysis, this represents no crossover mecha- of dissolved gas within the electrolyte increases, whereas the mean
nism here. In contrast, this is one of the main sources of crossover in gas bubble diameter decreases, which further complicates a proper
AEL. The mixing of gas saturated electrolyte cycles leads to a high gas separation. However, it has to be stated that hydrogen crossover
exchange of both gases. This phenomenon is mainly influenced by due to electrolyte cycle mixing becomes less important with a grow-
the dissolved gas concentrations and the electrolyte flow rates, but ing electrolysis plant size if the electrolyte recirculation rate is kept
also by the applied gas separators. Consequently no single equation constant.7,55 The alkaline electrolyzer in this study was operated with a
can be used to calculate hydrogen crossover through electrolyte mix- recirculation rate of 70 electrolyte replacements an hour. For minimum
ing. Nevertheless, a mathematical model was presented in a previous crossover through electrolyte mixing, the recirculation rate should al-
publication,16 which can be used for an estimation of this crossover ways be kept as low as possible. It must be ensured though that the flow
effect. rate is high enough to keep the temperature increase within the cell
below certain limits. Furthermore, it is important to emphasize that a
reduction of the electrolyte recirculation may lead to an increase of the
electrode bubble coverage, which effects an increase of cell voltage.
Results and Discussion
In PEM electrolysis, by contrast, no significant distinction between
Firstly, this section illustrates the influence of current density and differential and balanced pressure operation is noticeable at a system
system pressure on the measured gas purity in PEM and alkaline elec- or cathodic pressure of 10 bar.
trolysis. Subsequently, the permeation fluxes through the separating As it was described earlier, both electrolysis test cells were also
units are derived from these experimental results in order to clarify the fed with hydrogen and oxygen according to their volumetric flow
influence of current density. Following this, a breakdown of the indi- rates in a current density range from 0.05 to 0.9 A cm−2 while being
Figure 4. Determined hydrogen crossover in the

PEM (a) and alkaline (b) electrolysis setup at
different pressure conditions and 60◦ C. Hollow
symbols at the y-axis represent pressure drop and
literature data.
disconnected from the power supply. In the following, these exper- arating unit primarily increases with rising current density and also
iments will be referred to as reference measurements. The obtained system pressure during electrolysis operation.
results are also included in Fig. 3. It can easily be seen that the mea- Thus, Fig. 4a reveals a clear linear dependency of the hydrogen
sured hydrogen fractions are well below the results of the electrolysis crossover on the current density for the PEM electrolysis measure-
measurements. While only a value of 0.022 vol.% was measured in the ments, which has also been observed in the literature, e.g. Refs. 5 and
PEM setup, the alkaline cell yielded a concentration of 0.004 vol.% 14. This linear growth can be explained with mass transport resistances
with separated electrolyte cycles at a current density of 0.5 A cm−2 . within the cathode catalyst layer that cause an increasing supersatu-
In summary, the results indicate clearly that safety issues limit the ration of dissolved hydrogen with increasing current density.5 The
pressurized operation of the PEM and classical alkaline electrolysis slope of the crossover is approximately equal for all the investigated
setup with mixed electrolyte cycles as the anodic hydrogen content pressure stages, since the mass transport resistance is assumed to be
rises above 2 vol.%, which corresponds to approximately 50% of mainly affected by geometrical and/or structural parameters and less
the lower explosion limit.47 Consequently, electrolyzers are typically by operating conditions.5,13 As the linear extension of the determined
shut down as soon as this concentration is exceeded.1,6 Therefore, hydrogen crossover matches the zero current values excellently, a lin-
both electrolysis test stations cannot be operated safely under specific ear dependency of the hydrogen crossover is also assumed for even
current density and pressure conditions if no measures are taken to smaller current densities. Finally, it can also be seen that the balanced
reduce the crossover or the hydrogen content within the anodic half and differential pressure operation at a system or cathodic pressure of
cell. Subsequently to this chapter different mitigation strategies are 10 bar show no significant difference in hydrogen crossover. This was
discussed. expected, since literature also suggests no pressure-driven convective
crossover for Nafion membranes at this pressure gradient.23,41
Hydrogen permeation through the separating unit.—With the The AEL results are shown in Fig. 4b, where no specific trend
previously shown hydrogen fractions it is hard to assess the correlation becomes obvious at first. However, the data suggests that hydrogen
between hydrogen crossover and current density. However, with Eq. 8 crossover initially also increases with rising current density until a
or Eq. 9 and the measured hydrogen content, the hydrogen crossover plateau is formed. Additionally, increasing system pressures seem to
perm
NH2 can be achieved for the different applied current densities. In shift the formation of the plateau toward higher current densities. This
Figs. 4a and 4b the determined hydrogen crossover through the Nafion effect may be explained with an increase of the hydrogen mass trans-
117 membrane and the Zirfon separator is shown as a function of fer coefficient from the liquid electrolyte into the gas phase at higher
current density. In case of the alkaline system, measurement data current densities. As previously mentioned, electrolysis products are
with mixed electrolyte cycles are excluded, so that only hydrogen assumed to be generated in dissolved form before gas is evolved in
permeation through the separator is evaluated. a subsequent physical step. Therefore, dissolved hydrogen may leave
Furthermore, Fig. 4 also shows the results of the reference mea- the electrode either in gaseous state through the transport into adher-
surements. It can be seen for both technologies that the determined ing bubbles and subsequent bubble detachment or in dissolved form
crossover fluxes remain constant with increasing current density and through transport into the electrolyte bulk or across the separator. For
that the values are small compared to the results during electrolysis the description of this phenomenon Vogt56 introduced the gas evolu-
operation. These constant measurement values can be explained by tion efficiency, which describes the ratio of product transported into
the fact that the electrolyte is gas saturated by the continuous supply of gas bubbles at the electrode to the total generated flux. Literature
hydrogen and oxygen, which results in a constant diffusive flux across values for the gas evolution efficiency57–59 show a strong increase at
the separating unit (s. Eq. 15), whereas no other crossover mechanisms low current densities (up to approx. 0.2 A cm−2 ) until a linear de-
occur as the experiments were conducted under atmospheric pressure pendency can be observed. Hence, at very small current densities
conditions and no supersaturation can be formed. Additionally, Fig. 4 nearly the total amount of product leaves the electrode in dissolved
provides values for the purely diffusive hydrogen crossover at zero form. As the current density increases, the concentration of dissolved
current, which are represented by hollow symbols. For the PEM elec- hydrogen becomes high enough to activate nucleation sites, which
trolysis experiments these values were determined according to the enable the evolution of gas bubbles at the electrode.50 Therefore, the
aforementioned pressure drop method, whereas the values for the fraction of product leaving the electrode in dissolved form decreases
alkaline system were calculated from data provided by Schalenbach with higher current densities, which consequently limits the flux of
et al.24 Here it is recognizable that the reference measurements of both dissolved hydrogen across the separator. Furthermore, it is assumed
electrolysis systems show excellent agreement with the zero current that an increase in pressure moves the gas evolution efficiency toward
values of the pressure drop and literature data at atmospheric pressure, higher current densities, which would explain the later plateau forma-
which therefore confirms the purely diffusive crossover occurring in tion. Thus, a pressure increase enhances the mandatory dissolved gas
these measurements. concentration for activation of nucleation sites and reduces the mass
In contrast to the data of the reference measurements, the results transfer coefficient (kL = f (Re)) of dissolved hydrogen into adher-
of both technologies show that hydrogen crossover through the sep- ing bubbles, as the Reynolds number Re = u · db /ν is decreased by
Figure 5. Crossover breakdown for the PEM (a)

and classical alkaline (b) electrolysis experiments
under atmospheric pressure conditions. Since
crossover through electrolyte mixing does not rep-
resent a cell-specific quantity, the y-axis of the
alkaline experiments shows the actual molar flow
rates.
smaller bubble diameters. Of course, it may also be possible that the crossover fluxes need to be multiplied by the surface area of the sepa-
determined crossover is superimposed by the convective transport of rator as crossover by mixing the electrolyte cycles does not represent
dissolved hydrogen across the separator. However, this is questioned a cell-, but electrolyzer-specific value.
with the current state of knowledge as the anodic cycle showed an Therefore, Eq. 9 is used for the estimation of the overall crossover
approximately 20 mbar higher pressure at elevated pressure stages, with mixed electrolyte cycles and then multiplied by the surface area of
which would have suppressed convective hydrogen permeation. Fi- the separator. In order to simplify the evaluation, it is assumed that the
nally, in contrast to PEM electrolysis, where a linear dependency of diffusive crossover due to electrolyte saturation and supersaturation,
the crossover can be assumed for the complete investigated current which were determined with separated cycles, can also be transferred
density range, the analysis of the alkaline zero current values suggests to mixed electrolyte cycles. However, the diffusive crossover should
a strong increase of the crossover at small current densities. However, be slightly smaller with mixed electrolyte cycles, since the electrolyte
further investigations are necessary to confirm this presumption. entering the half cells is already saturated with hydrogen and oxygen,
The overall comparison of the electrolysis results shows that diffu- thus reducing the concentration gradient across the separator. There-
sive crossover in PEM electrolysis is more than one order of magnitude fore, the share of crossover due to electrolyte mixing might be even
higher than in AEL, which is mainly a consequence of the lower hy- larger for this specific test station than shown in the following. The
drogen solubility in concentrated potassium hydroxide solution. So, trend of the diffusive crossover due to saturation and supersaturation
in comparison to pure water the solubility is decreased by a factor of 8 of the electrolyte was already discussed in the previous section. At a
in 30 wt% KOH at a temperature of 30◦ C and a gas partial pressure of current density of 0.7 A cm−2 their shares amount to 1.4% and 8.6%,
1.013 bar.7 Similar information was published by Schalenbach et al.,60 respectively. Here, similar to PEM electrolysis, a slight increase of
who reported a 38 times lower hydrogen permeability coefficient in the diffusional crossover due to supersaturation can be observed with
Zirfon soaked with 30 wt% KOH compared to Nafion 117 immersed rising current density as the share at 0.1 A cm−2 is determined to be
in water at a cell temperature of 80◦ C. 6.0%.
However, it is obvious that the mixing of the electrolyte cycles
represents the largest source for transferring hydrogen into the anodic
Crossover breakdown.—In this chapter the respective share of the half cell. This crossover flow remains approximately constant over
different crossover mechanisms is graphically depicted in Fig. 5 for the whole investigated current density range as regardless of the ap-
PEM and classical alkaline water electrolysis with mixed electrolyte plied current density gas-saturated electrolyte is always fed into the
cycles at atmospheric pressure conditions. half cells. Only at smaller current densities a minor increase of this
Fig. 5a shows the breakdown of the achieved PEM electrolysis mechanism is recognizable, which might be regarded to not only dis-
results. Here, the hydrogen crossover can be divided into two different solved species, but also gas bubbles being transported back into the
areas. Firstly, a small proportion of the overall crossover is caused electrolysis cell. The percentage of crossover due to electrolyte mix-
by diffusion due to saturated dissolved hydrogen, which has been ing is the highest for all investigated pressure stages, yielding 90% at
determined with the reference measurements. The second and major atmospheric pressure, 88% at 10 bar and 95% at 20 bar for a current
part is also a result of diffusion, but is caused by a supersaturated density of 0.7 A cm−2 . These high values exemplify once more that an
hydrogen concentration in the cathodic half cell. The percentage of improvement of the anodic hydrogen concentration can in particular
this effect increases with current density. Whereas 86% are transported be achieved by an optimized lye circulation strategy.
across the membrane at 0.1 A cm−2 , a current density of 1 A cm−2
shows a share of 98% since supersaturation increases linearly with
current density as it was stated before. Variation of separating unit thickness.—The experimental results
Furthermore, the results indicate that the percentage of crossover showed that safety issues can limit the operating ranges of the elec-
due to supersaturation decreases with increasing cathodic pressure. trolyzers under certain conditions with the investigated state-of-the-art
The values for the higher cathodic pressures are based on the results separating unit materials. In the following section, the influence of the
of the pressure drop method. At a pressure of 10 bar the share is separating unit thickness on the anodic hydrogen content is studied.
70%, while only 53% are determined at 20 bar for a current density of Therefore, the anodic gas contamination is estimated for thinner sep-
1 A cm−2 . This can be explained by the fact that diffusive crossover arating units on basis of the obtained experimental data.
due to saturated dissolved hydrogen increases with raising pressure, It is assumed for the PEM system that neither supersaturation nor
whereas the supersaturation is nearly pressure independent. Thus, the the permeability coefficient are influenced by the membrane thick-
current density effect loses importance at very high cathodic pressures. ness. Thus, the overall crossover flux NHcross
2
for different membrane
The crossover breakdown for the atmospheric AEL experiments thicknesses can be calculated according to Eq. 22.
is shown in Fig. 5b. Since the classical process management with
mixed electrolyte cycles is chosen for the evaluation of the individ- perm δN117
NHcross
2 ,PEM
= NH2 , exp [22]
ual crossover fractions, the previously shown area-related diffusive δsep
Figure 6. Anodic hydrogen content as a function

of separating unit thickness and current density
for PEM (a) and alkaline electrolysis (b) at at-
mospheric pressure, a temperature of 60◦ C and
32 wt% KOH. Calculations are based on the ob-
tained experimental data for the N117 membrane
and the Zirfon separator.
The same assumptions also apply for the alkaline system. Further- very small share. So, the alkaline cell, equipped with a Zirfon sepa-
more, crossover through electrolyte mixing is also independent on rator, shows that the reduction of the separator thickness offers high
separator thickness. Therefore, Eq. 23 may be used for the calculation potential for improving ohmic losses if the mechanical stability can
of the overall crossover flux at alternative separator thicknesses. be maintained and this is technically feasible. However, it has to be
said that convective crossover gains in importance with a decreasing
perm δZirfon
NHcross
2 ,AEL
= NHmix
2 , exp
+ NH2 , exp [23] separator thickness, which necessitates a complex pressure control to
δsep keep the differential pressure at a minimum.
Subsequently, the hydrogen content can be calculated with Eq. 6
for both technologies in variation of the separating unit thickness. Mitigation Strategies
The obtained results are depicted in Fig. 6 for atmospheric pressure
In this section strategies for the mitigation of crossover and cor-
conditions and a temperature of 60◦ C. The ohmic resistances Rsep
responding side effects are discussed for both technologies. These
shown there are calculated according to Eq. 24.
strategies are summarized in Table II and divided into five categories.
δsep The different strategies are evaluated regarding the five following cri-
Rsep = [24] teria: safety issues (H2 in O2 ), degradation (catalyst layer/electrode,
σsep
membrane) due to crossover, Faraday efficiency, electrochemical per-
Here, σsep denotes the ionic conductivity of the separating units, formance and costs.
which were achieved from the literature. For a Nafion 117 membrane
immersed in water the conductivity is given with roughly 13.5 S m−1 Separating unit.—Particularly in PEM electrolysis the use of
at 60◦ C.61 Vermeiren et al.62 provide a value of 0.166 cm2 for a highly ion-conductive membranes with lower hydrogen permeabil-
Zirfon separator with a thickness of 500 μm at a temperature of 60◦ C ity is an interesting possibility to solve existing safety issues. This
and a potassium hydroxide concentration of 30 wt%. Correspondingly, may not only increase the Faraday efficiency, but the overall perfor-
the insertion into Eq. 24 yields an ionic conductivity of 30.1 S m−1 . mance of the electrolyzer. In literature some promising candidates can
Although the electrolyte applied in this study is slightly higher con- be found, e.g. radiation grafted membranes63 or hydrocarbon-based
centrated, the value of Vermeiren et al. can still give an approximate materials.64 In contrast, for AEL the previous results suggest that
estimation. diffusive crossover may not be significantly reduced with advanced
The heatmaps of Fig. 6 show the correlation between the separating materials, as the diffusional flux is already very limited due to the
unit thickness on the bottom or the respective electrical resistance on poor solubility of hydrogen and oxygen in the highly concentrated
the top x-axis and current density on the y-axis. The hydrogen content electrolyte. However, if lower electrolyte concentrations are to be
itself is depicted in form of contour lines. A comparison of the two used in the future, this approach will become more important as better
technologies shows that the contour lines point in different directions. cell efficiencies can be achieved with ion-conductive membranes in
Thus, in PEM electrolysis the lines are ordered vertically, while the this operating regime, e.g. Refs. 65 and 66.
hydrogen content in AEL roughly shows a horizontal curve. Another possibility for a reduction of the anodic hydrogen content
The contour plot for PEM electrolysis reveals that the hydrogen would be the increase of the membrane thickness, which of course
content strongly increases with a decreasing membrane thickness. would be linked to a reduction of the electrolyzer performance. So,
This results from the purely diffusive character of hydrogen crossover this is not a real option and also does not reflect the trend of research
in the PEM setup, which increases inversely proportional with a de- and development.
creasing membrane thickness (s. Eq. 10). Consequently, the trend
toward thinner membranes to reduce ohmic losses can only be re- Recombination catalyst.—A totally different approach is the ap-
alized with a simultaneous reduction of the hydrogen content in the plication of recombination catalysts, which enable the reaction of
anode or the crossover itself. permeated hydrogen back to water. For PEM electrolysis, it has al-
Fig. 6b shows a totally different trend for the hydrogen content ready been shown that the incorporation of platinum is able to reduce
with a decreasing separator thickness in AEL. Here, the hydrogen the anodic hydrogen content significantly if the catalyst is applied to
content is nearly independent on the separator thickness. An increase the back side of the current collectors. Even better results were ob-
of the hydrogen concentration due to diffusion can only be observed tained, when the recombiner was positioned downstream of the gas
at thicknesses below 100 μm. This can be explained with the fact separators as the optimal operation necessitates dry gases.8 Addition-
that the main portion of hydrogen crossover is caused by mixing of ally, it is also conceivable to integrate the catalyst directly into the
the electrolyte cycles, whereas diffusive crossover only represents a electrode9 or separating unit in order to provide a large contact area
Table II. Approaches for crossover reduction and negative consequences in PEM (left)/AEL (right). Meaning of the symbols: ++ very effective,
+ good, o no effect or slightly positive/negative, - negative, - - very negative, ? effect cannot be estimated.
H2 in O2 (safety) Degradation Faraday Performance Cost

1. Separating unit
a) advanced materials with high ionic conductivity and low permeability63,64 +/o ?/? +/o ++/+ ?/?
b) thicker separating units +/o +/o +/o - -/- - -/-
2. Recombination catalyst
a) external recombination8 +/+ o/o o/o o/o -/-
b) on the backing layers 8 (PTL, bipolar plate) +/+ o/o o/o o/o -/-
c) within the catalyst layer/electrode9 ++/+ ?/? ?/o o/o -/-
d) within the membrane/separator +/+ ?/o o/o -/o -/-
3. Supersaturation
a) improved catalyst layers/electrodes, less transport resistances ++/+ +/o +/+ ++/++ o/o
b) ultrasonic field67 +/+ ?/? +/+ +/+ - -/- -
c) magnetic field51,68,73–76 +/? ?/? +/? +/+ - -/- -
d) super gravity field69,77 ?/? ?/? ?/? +/+ - -/- -
4. Interlayer
a) thin layers/sheets that are capable for the ionic transport, but gas-tight70,71 ++/o +/o ++/o ?/o -/-
b) 3rd/4th electrode72 ++/+ ?/? +/+ -/- - -/- -
5. Process management (only AEL)
a) advanced gas/liquid separators /+ /o /+ /o /-
b) mixing strategies7,55 /++ /o /+ /o /o
c) advanced mixing appliances /++ /o /++ /o /- -
between the permeating gases and the recombination catalyst. But it supplied by an additional power source. In case of PEM water elec-
is important to emphasize that the introduction of recombination cata- trolysis the permeating hydrogen can be oxidized to protons at the
lysts only reduces the risk of explosive mixtures, whereas the Faraday third electrode, which will then be pumped back to the cathode elec-
efficiency is not improved as this measure does not address the origin trochemically. This method was already successfully demonstrated
of crossover. by Schalenbach et al.72 They also showed that this third electrode can
work in combination with the oxygen evolving anode. In this case per-
Reduction of supersaturation.—As it was discussed before, super- meating hydrogen is oxized at the additional electrode, while the anode
saturation strongly enhances the diffusional crossover in both tech- works as a fuel cell locally. Furthermore, a four electrode arrangement
nologies. Therefore, especially when membrane electrode assemblies supplied by two additional power sources could also be used to send
are applied, it should be examined whether transport resistances that back permeating hydrogen and oxygen in AEL. Here, one electrode
contribute to the formation of supersaturation can be reduced by im- each would have to be connected to the anode and cathode of the
proved geometrical and structural properties of the electrodes or cat- electrolysis cell, whereby the reverse reactions of the corresponding
alyst layers. The comparison of literature data for PEM electrolysis gas-evolving electrode would then take place there. However, placing
reveals different slopes for the crossover increase with rising cur- additional electrodes is expensive and further complicates cell stack
rent density, which is presumed to be mainly an effect of different design.
geometrical or structural properties.5 Finally, it is also possible to apply a recombination catalyst as an
Furthermore, the influence of external impacts on the electrolysis interlayer, which has already been described before (s. Table II 2.d).
cell on supersaturation of the electrolyte could be investigated. Li
et al.67 reported that they were able to reduce the energy demand in Process management.—It could be shown for AEL that mixing of
AEL, when the cell was operated under the influence of ultrasound the electrolyte cycles represents the most important crossover mech-
as the evolving gas bubbles were removed from the electrode sur- anism. For this reason advanced gas separators should ensure that no
face more quickly. Here, it is also conceivable that an ultrasonic field gaseous species are transported back into the cell even at high system
could promote the formation of gas bubbles from the supersaturated pressures.
electrolyte, which consequently would reduce the dissolved concen- Furthermore, Schug55 reported that the amount of recycled dis-
tration of electrolysis products. Similar approaches are the application solved species can be reduced with an adequate electrolyte circulation
of magnetic68 or gravitational69 fields, which have already been shown control. Here, the recirculation rate should be chosen as low as pos-
to reduce cell voltage in AEL. It is difficult to estimate in which way sible to minimize the convective crossover of electrolysis products
these technologies influence supersaturation of dissolved gas within through mixing anolyte and catholyte. Furthermore, this crossover
the respective electrolysis technology, but certainly represent an inter- mechanism loses importance with a growing electrolysis plant size
esting field of research, since performance, crossover and also Faraday or electrode area as the gas production rate is increased, whereas the
efficiency could be improved. amount of recycled species stays equal at identical electrolyte flow
rates.
Interlayer.—Another option to reduce crossover or the correspond- Additionally, it could be demonstrated before7 that periodic switch-
ing negative effects is placing thin interlayers within the separator or ing between mixed and separated cycles is capable of reducing the
between the electrode and the separating unit. These interlayers could mean anodic hydrogen content while maintaining an equal electrolyte
consist of a material that is characterized by a low permeability coef- concentration in the anolyte and catholyte. It may also be conceivable
ficient, but capable for the ionic transport. For instance, this approach that electrolysis is carried out with separated cycles under part-load
was shown using hexagonal boron nitride and graphene in methanol operation, whereas a merging of the cycles is initiated as soon as a
fuel cells, e.g. Ref. 70. For PEM fuel cells this was also successfully specific minimum current density is exceeded.
demonstrated through the application of a graphene oxide/cerium ox- Eventually an advanced mixing appliance is proposed in
ide interlayer.71 This method could be very effective, because of the Table II. This could be realized by an additional electrolyte mix-
reduction of crossover at its origin and therefore all negative effects. ing tank equipped with an ion-exchange membrane that still enables
Another interlayer could be realized in form of a third electrode an equalization of the anodic and cathodic electrolyte concentration.
that is electrically connected to the hydrogen evolving cathode and Consequently, the transport of cycled gas bubbles could be prevented.
Furthermore, crossover due to the transport of dissolved species into thickness, a strong increase of the anodic hydrogen contamination
the opposite half cell can be reduced in comparison to the direct mix- is evident with decreasing membrane thicknesses in PEM electrol-
ing of anolyte and catholyte as the applied membrane serves as an ysis. Contrarily, the estimation shows a huge optimization potential
additional diffusion barrier. for a reduction of the separator thickness in AEL as no meaningful
increase of the anodic hydrogen concentration is observable down to
thicknesses of 100 μm as crossover is mainly influenced by electrolyte
Conclusions
mixing.
This study provides a summary of the crossover mechanisms oc- Finally, this study provides a summary of possible mitigation
curring in PEM and alkaline water electrolysis and experimentally strategies that could help to reduce hydrogen crossover or at least
investigates the influence of current density, system pressure and vari- the anodic foreign gas content. To this end, literature references are
ous process management strategies on the resulting gas purity. In addi- mentioned in which initial investigations have already been reported.
tion, reference measurements for the determination of purely diffusive
crossover in both technologies are conducted. For a fair comparison Acknowledgments
of the two technologies, a uniform operating temperature of 60◦ C
and state-of-the-art separating unit materials, Nafion 117 in PEM and The authors gratefully acknowledge the financial support by the
Zirfon in alkaline electrolysis, were chosen. Federal Ministry of Education and Research of Germany in the frame-
Generally the results of both technologies indicate that the anodic work of PowerMEE (project number 03SF0536B) and the DFG in the
gas contamination increases significantly with decreasing current den- framework of the projects INST 189/182-1 FUGG and TU 89/18-1.
sity and rising system pressure, which is mainly a consequence of
increased dissolved gas concentration at elevated pressures. In fact, List of Symbols
under specific operating conditions the determined hydrogen concen- a specific surface area m−1
trations even exceeded 2 vol.% in the anodic product gas stream, so Ael cross-sectional electrode area m2
that a safe plant operation was no longer possible. Here, the compari- Asep cross-sectional separating unit area m2
son of different process management strategies in AEL showed that the cH2 dissolved hydrogen concentration mol m−3
anodic hydrogen concentration is particularly influenced by mixing csolv concentration of solvent within mol m−3
anolyte and catholyte, since a safe operation of the electrolyzer with separating unit
separated cycles could be ensured at all investigated system pressures. d pore diameter of separating unit m
Furthermore, the PEM experiments revealed that no distinction in db gas bubble diameter m
the anodic hydrogen concentration is recognizable between balanced D H2 diffusion coefficient of dissolved m2 s−1
and differential pressure operation at an applied pressure gradient of hydrogen in solvent
10 bar. F Faraday constant 96485A s mol−1
In a subsequent step the measured anodic hydrogen concentrations i current density A m−2
of both technologies were converted into permeation fluxes to illus- kL mass transfer coefficient m s−1
trate the influence of current density and system pressure more clearly. K H2 hydrogen permeability coefficient mol Pa−1 m−1 s−1
As a result a linear increase of hydrogen permeation toward rising cur- K koz Kozeny constant –
rent densities could be observed in PEM electrolysis at all pressure K sep hydraulic permeability m2
levels. It is assumed that this linear growth is caused by a super- n drag drag coefficient –
saturation of dissolved hydrogen, which is caused by mass transport N H2 hydrogen flux mol m−2 s−1
resistances within the cathode catalyst layer. The supersaturation in- N O2 oxygen flux mol m−2 s−1
creases with current density, thus enhancing the diffusive permeation p absolute pressure Pa
pi partial pressure of comp. i Pa
across the membrane. In comparison to the reference measurements,
Rsep ohmic resistance of separating unit cm2
which showed no current density dependence of permeating hydro-
Re Reynolds number –
gen, supersaturation multiplied the diffusive crossover by a factor of
SH2 solubility of hydrogen in solvent mol Pa−1 m−3
20 at a current density of 1 A cm−2 and a system pressure of 1 bar.
T temperature K
However, the influence of supersaturation is seen to decrease with u velocity m s−1
increasing system pressure. While the share on the overall crossover vsolv hydraulic velocity of solvent m s−1
flux amounts to 98% at 1 A cm−2 and 1 bar, only 53% are determined V̇ volumetric flow rate of electrolyte m3 s−1
at a pressure of 20 bar.
In contrast, the hydrogen permeation data of the alkaline experi- Greek
ments did not show such a clear dependency on the current density. δsep separating unit thickness m
However, it is observed at all investigated pressure stages that hydro- ε porosity –
gen crossover initially also increases with rising current density until η dynamic viscosity of solvent Pa s
a maximum permeation flux is reached. Similarly to PEM electrolysis ν kinematic viscosity of solvent m2 s−1
it is assumed that the diffusive crossover is enhanced by a super- σsep ionic conductivity S m−1
saturated electrolyte. But this effect is presumed to lose influence at τ tortuosity –
higher current densities as a greater proportion of evolved hydrogen is H2 anodic hydrogen fraction –
directly transported into gas bubbles adhering to the electrode, which Subscripts/Superscripts
consequently limits the diffusive crossover flux. Electrolyte mixing
ano anode
represents the most influential crossover mechanism with a share of
cat cathode
90% under atmospheric pressure conditions and a current density of
conv convective
0.7 A cm−2 . Consequently diffusion due to saturation and supersatu- cross crossover
ration only yields 1.4% and 8.6% for this experimental setup. diff diffusive
Overall, PEM electrolysis showed a more than one order of mag- dp differential pressure
nitude higher diffusive crossover flux than AEL, which is particularly drag drag
a consequence of the significantly lower hydrogen solubility in con- eff effective
centrated potassium hydroxide solution. Therefore, this study also mix electrolyte mixing
provides an estimation of the influence of the separating unit thick- perm permeation
ness on the expected anodic gas contamination. Due to the inversely sep separator
proportional dependency of the hydrogen crossover on the membrane solv solvent
ORCID 35. D. R. Stull, Ind. Eng. Chem., 39(4), 517 (1947).

36. J. Balej, Int. J. Hydrogen Energ., 10(4), 233 (1985).
B. Bensmann http://orcid.org/0000-0001-8685-7192 37. P. Medina and M. Santarelli, Int. J. Hydrogen Energ., 35(11), 5173 (2010).
38. M. Mulder, Characterisation of membranes, in: M. Mulder (Ed.), Basic Principles of
Membrane Technology, Springer Netherlands, Dordrecht, pp. 157–209 1996.
References 39. T. Ozgumus, M. Mobedi, and U. Ozkol, Eng. Appl. Comp. Fluid, 8(2), 308 (2014).
40. M. Schalenbach, J. Electrochem. Soc., 164(14), X23 (2017).
1. L. M. Gandia, R. Oroz, A. Ursua, P. Sanchis, and P. M. Dieguez, Energ. Fuel., 21(3), 41. F. Barbir, Sol. Energy, 78(5), 661 (2005).
1699 (2007). 42. B. Bensmann, R. Hanke-Rauschenbach, I. K. Peña Arias, and K. Sundmacher, Elec-
2. R. E. Clarke, S. Giddey, and S. P. S. Badwal, Int. J. Hydrogen Energ., 35(3), 928 trochim. Acta, 110, 570 (2013).
(2010). 43. B. Bensmann, R. Hanke-Rauschenbach, G. Müller-Syring, M. Henel, and
3. M. Carmo, D. L. Fritz, J. Mergel, and D. Stolten, Int. J. Hydrogen Energ., 38(12), K. Sundmacher, Appl. Energ., 167, 107 (2016).
4901 (2013). 44. S. A. Grigoriev, A. A. Kalinnikov, P. Millet, V. I. Porembskiy, and V. N. Fateev, J.
4. S. A. Grigoriev, V. I. Porembskiy, S. V. Korobtsev, V. N. Fateev, F. Auprête, and Appl. Electrochem., 40(5), 921 (2010).
P. Millet, Int. J. Hydrogen Energ., 36(3), 2721 (2011). 45. L. C. Jacobson, X. Ren, and V. Molinero, J. Phys. Chem. C, 118(4), 2093 (2014).
5. P. Trinke, B. Bensmann, and R. Hanke-Rauschenbach, Int. J. Hydrogen Energ., 46. K. Onda, T. Murakami, T. Hikosaka, M. Kobayashi, R. Notu, and K. Ito, J. Elec-
42(21), 14355 (2017). trochem. Soc., 149(8), 1069 (2002).
6. W. Hug, J. Divisek, J. Mergel, W. Seeger, and H. Steeb, Int. J. Hydrogen Energ., 47. V. Schroeder, B. Emonts, H. Janssen, and H.-P. Schulze, Chem. Eng. Technol., 27(8),
17(9), 699 (1992). 847 (2004).
7. P. Haug, M. Koj, and T. Turek, Int. J. Hydrogen Energ., 42(15), 9406 (2017). 48. J. Eigeldinger and H. Vogt, Electrochim. Acta, 45(27), 4449 (2000).
8. S. A. Grigoriev, P. Millet, S. V. Korobtsev, V. I. Porembskiy, M. Pepic, C. Etievant, 49. R. J. Balzer and H. Vogt, J. Electrochem. Soc., 150(1), 11 (2003).
C. Puyenchet, and V. N. Fateev, Int. J. Hydrogen Energ., 34(14), 5986 (2009). 50. H. Vogt, K. Kreysa, S. Vasudevan, R. Wüthrich, J. D. Abou Ziki, and R. El-Haddad,
9. H. Ito, N. Miyazaki, M. Ishida, and A. Nakano, Int. J. Hydrogen Energ., 41(45), Electrochemical reactors, Ullmann’s encyclopedia of industrial chemistry, Wiley-
20439 (2016). VCH, 2013.
10. M. Inaba, T. Kinumoto, M. Kiriake, R. Umebayashi, A. Tasaka, and Z. Ogumi, 51. H. Matsushima, D. Kiuchi, and Y. Fukunaka, Electrochim. Acta, 54(24), 5858
Electrochim. Acta, 51(26), 5746 (2006). (2009).
11. M. Chandesris, V. Médeau, N. Guillet, S. Chelghoum, D. Thoby, and F. Fouda-Onana, 52. K. Kikuchi, H. Takeda, B. Rabolt, T. Okaya, Z. Ogumi, Y. Saihara, and H. Noguchi,
Int. J. Hydrogen Energ., 40(3), 1353 (2015). J. Electroanal. Chem., 506(1), 22 (2001).
12. M. Schalenbach, Int. J. Hydrogen Energ., 41(1), 729 (2016). 53. H. Vogt, Electrochim. Acta, 56(3), 1409 (2011).
13. P. Trinke, B. Bensmann, and R. Hanke-Rauschenbach, Electrochem. Commun., 82, 54. R. T. Ferrell and D. M. Himmelblau, AIChE J., 13(4), 702 (1967).
98 (2017). 55. C. A. Schug, Int. J. Hydrogen Energ., 23(12), 1113 (1998).
14. S. Al Shakhshir, X. Cui, S. Frensch, and S. K. Kær, Int. J. Hydrogen Energ., 42(34), 56. H. Vogt, Electrochim. Acta, 29(2), 167 (1984).
21597 (2017). 57. H. Vogt, Electrochim. Acta, 30(2), 265 (1985).
15. A. Manabe, H. Domon, J. Kosaka, T. Hashimoto, T. Okajima, and T. Ohsaka, J. 58. J. M. Chin Kwie Joe, L. J. J. Janssen, S. J. D. van Strelen, J. H. G. Verbunt, and
Electrochem. Soc., 163(11), 3139 (2016). W. M. Sluyter, Electrochim. Acta, 33(6), 769 (1988).
16. P. Haug, B. Kreitz, M. Koj, and T. Turek, Int. J. Hydrogen Energ., 42(24), 15689 59. J. St-Pierre, N. Massé, and M. Bargeron, Electrochim. Acta, 40(8), 1013 (1995).
(2017). 60. M. Schalenbach, G. Tjarks, M. Carmo, W. Lueke, M. Mueller, and D. Stolten, J.
17. Q. Duan, H. Wang, and J. Benziger, J. Membrane Sci., 392-393, 88 (2012). Electrochem. Soc., 163(11), 3197 (2016).
18. Brochure Zirfon Perl UTP 500, https://www.agfa.com/sp/global/en/binaries 61. H. Ito, T. Maeda, A. Nakano, and H. Takenaka, Int. J. Hydrogen Energ., 36(17),
/ZirfonPerl_UTP500_tcm611-56748.pdf, Accessed: 2018-01-15. 10527 (2011).
19. T. Sakai, H. Takenada, N. Wakabayashi, Y. Kawami, and E. Torikai, J. Phys. 62. P. Vermeiren, W. Adriansens, J. Moreels, and R. Leysen, Int. J. Hydrogen Energ.,
Chem.-US, 132(6), 1328 (1985). 23(5), 321 (1998).
20. M. K. Tham, R. D. Walker, and K. E. Gubbins, J. Phys. Chem.-US, 74(8), 1747 63. A. Albert, A. O. Barnett, M. S. Thomassen, T. J. Schmidt, and L. Gubler, ACS Appl.
(1970). Mater. Inter., 7(40), 22203 (2015).
21. B. Bensmann, R. Hanke-Rauschenbach, and K. Sundmacher, Int. J. Hydrogen Energ., 64. C. K. Mittelsteadt and J. A. Staser, Electrolyzer membranes, in: Polymer Science: A
39(1), 49 (2014). Comprehensive Reference, Elsevier, 2012, pp. 849–871.
22. T. Sakai, H. Takenaka, and E. Torikai, J. Electrochem. Soc., 133(1), 88 (1986). 65. L. A. Diaz, J. Hnat, N. Heredia, M. M. Bruno, F. A. Viva, M. Paidar, H. R. Corti,
23. M. Schalenbach, T. Hoefner, P. Paciok, M. Carmo, W. Lueke, and D. Stolten, J. Phys. K. Bouzek, and G. C. Abuin, J. Power Sources, 312, 128 (2016).
Chem. C, 119(45), 25145 (2015). 66. M. R. Kraglund, D. Aili, K. Jankova, En. Christensen, Q. Li, and J. O. Jensen, J.
24. M. Schalenbach, W. Lueke, and D. Stolten, J. Electrochem. Soc., 163(14), 1480 Electrochem. Soc., 163(11), 3125 (2016).
(2016). 67. S. Li, C. Wang, and C. Chen, Electrochim. Acta, 54(15), 3877 (2009).
25. F. Marangio, M. Santarelli, and M. Cali, Int. J. Hydrogen Energ., 34(3), 1143 (2009). 68. M. Lin, L. Hourng, and C. Kuo, Int. J. Hydrogen Energ., 37(2), 1311(2012).
26. M. Schalenbach, M. Carmo, D. L. Fritz, J. Mergel, and D. Stolten, Int. J. Hydrogen 69. M. Wang, Z. Wang, and Z. Guo, Int. J. Hydrogen Energ., 35(8), 3198 (2010).
Energ., 38(35), 14921 (2013). 70. S. M. Holmes, P. Balakrishnan, V. S. Kalangi, X. Zhang, M. Lozada-Hidalgo,
27. H. Kim, M. Park, and K. S. Lee, Int. J. Hydrogen Energ., 38(6), 2596 (2013). P. M. Ajayan, and R. R. Nair, Adv. Energy Mater., 7(5), 1601216 (2017).
28. P. Trinke, B. Bensmann, S. Reichstein, R. Hanke-Rauschenbach, and K. Sundmacher, 71. M. Breitwieser, T. Bayer, A. Büchler, R. Zengerle, S. M. Lyth, and S. Thiele, J. Power
J. Electrochem. Soc., 163(11), 3164 (2016). Sources, 351, 145 (2017).
29. J. Fimrite, B. Carnes, H. Struchtrup, and N. Djilali, J. Electrochem. Soc., 152(9), 72. M. Schalenbach and D. Stolten, Electrochim. Acta, 156, 321 (2015).
1815 (2005). 73. H.-B. Lee, J.-C. Tsau, and C.-Y. Lee, J. Nanomater., 2013(11), 1 (2013).
30. R. B. Evans and G. M. Watson, J. Chem. Phys., 35(6), 2076 (1961). 74. H. Matsushima, T. Iida, and Y. Fukunaka, Electrochim. Acta, 100, 261 (2013).
31. C. L. Young, Hydrogen and deuterium, Solubility data series, 5/6., Pergamon Press, 75. D. Baczyzmalski, F. Karnbach, X. Yang, G. Mutschke, M. Uhlemann, K. Eckert, and
1981. C. Cierpka, J. Electrochem. Soc., 163(9), E248 (2016).
32. R. Wiebe and V. L. Gaddy, J. Am. Chem. Soc., 56(1), 76 (1934). 76. L. Elias and A. Chitharanjan Hegde, Electrocatalysis, 8(4), 375 (2017).
33. P. Ruetschi and R. F. Amlie, J. Phys. Chem.-US, 70(3), 718 (1966). 77. H. Matsushima, D. Kiuchi, Y. Fukunaka, and K. Kuribayashi, Electrochem. Commun.,
34. M. B. Knaster and L. A. Apel’baum, Russ. J. Phys. Ch., 38, 120 (1964). 11(8), 1721 (2009).

Hydrogen Crossover in PEM and Alkaline Water Electrolysis

Uploaded by

Copyright:

Available Formats

Hydrogen Crossover in PEM and Alkaline Water Electrolysis

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Hydrogen Crossover in PEM and Alkaline Water Electrolysis

Uploaded by

Copyright:

Available Formats

F502 Journal of The Electrochemical Society, 165 (7) F502-F513 (2018)

0013-4651/2018/165(7)/F502/12/$37.00 © The Electrochemical Society

Hydrogen Crossover in PEM and Alkaline Water Electrolysis:

Alkaline electrolysis.—The alkaline test cell was connected to a

Mechanism Equation: Parameter/Symbol/Unit PEM (Nafion) AEL (Zirfon)

a value for 30◦ C in pure water.

Figure 3. Measured anodic hydrogen content as

Figure 4. Determined hydrogen crossover in the

Figure 5. Crossover breakdown for the PEM (a)

Figure 6. Anodic hydrogen content as a function

H2 in O2 (safety) Degradation Faraday Performance Cost

ORCID 35. D. R. Stull, Ind. Eng. Chem., 39(4), 517 (1947).

You might also like