Critical Review—Identifying Critical Gaps for Polymer Electrolyte Water Electrolysis Development

The principal design of a PEWE is shown in the Figure 1. The catalyst coated membrane (CCM) comprises a polymer electrolyte, which ensures selective conduction of protons and separation of product gases while providing electrical insulation and gas separation between the electrodes. The CCM is compressed between two porous transport layers (PTLs), which provide mass transport pathways for the water and product gases and electrical and thermal conductivity between the bipolar plates (BPPs) and the catalyst layers of anode and cathode (CL_a, CL_c). The BPPs are often structured with flow channels to ensure an even flow of the water over the PTL and easy removal of product gases. Liquid water is generally introduced to the cell at the anode. During operation, water reaching the anodic CL through the anodic PTL is oxidized into protons, electrons and oxygen (Equation 1). The produced oxygen is transported through the anodic PTL and removed from the cell. Protons from the oxygen evolution reaction (OER) at the anode are transported through the membrane to the cathodic CL and are reduced to molecular hydrogen in the hydrogen evolution reaction (HER) (Equation 2). Movement of protons through the membrane is accompanied by a simultaneous transport of water as a result of the electro-osmotic drag. Produced hydrogen is then transported through the cathodic PTL and removed from the cell. PEWE designs often enable water recirculation through the anodic and cathodic cell compartments for product gas removal and temperature control. The overall reaction is the sum of the two half-cell reactions (Equation 3).

Zoom In Zoom Out Reset image size

The electrochemical splitting of water from its liquid state has a standard free energy of reaction Δ_RG^*° of 237 kJ/mol at standard conditions (25°C, 1 bar). It represents the minimum electrical work needed for splitting water at the corresponding theoretical cell potential of E^*0 = 1.23 V. If no external heat source is present, the total energy, corresponding to the reaction enthalpy Δ_RH^*° of 286 kJ/mol, needs to be provided through electricity for the reaction to take place. The associated potential is higher than E^*0 and is called the thermoneutral potential E^tn = 1.48 V. In practical systems the operating potential is higher at technically and economically viable reaction rates for hydrogen production. The excess energy in the form of overpotential is converted into heat during electrolysis, and needs to be properly managed. The electrical energy required to split water Δ_RG^*° decreases with increasing temperature, and a higher fraction of the total energy can be provided in the form of heat TΔS. Liquid water can still be fed to the electrolyzer at temperatures above 100°C at elevated system pressures (Figure 2).

Zoom In Zoom Out Reset image size

Reversible anode and cathode potentials are estimated using the Nernst equation, as a function of temperature, pressure and the activities of hydrogen and oxygen at standard conditions (Equations 4 and 5).

The efficiency η of a PEWE cell is an important parameter to assess the energy costs of the process. The voltage efficiency η_V relates the theoretically required amount of energy for the reaction, i.e., the reaction enthalpy Δ_RH^*°, to the actual electrical energy input into the cell according to

where E_cell is the operating cell voltage and Δ_RH^*° is the enthalpy of the water splitting reaction and the same as the heating value of hydrogen (cf. Table I). The faradaic efficiency η_F takes the losses of hydrogen (and, if of interest, oxygen) due to crossover through the membrane into consideration (cf. below). Conversion efficiency of the water electrolyzer can be referred to the lower heating value (LHV) or the higher heating value (HHV) of hydrogen. The LHV is typically used when the focus of the discussion is on the use of hydrogen, since many applications do not benefit from the condensing enthalpy of water (ΔH_vap = 44 kJ/mol at 25°C). The HHV is a preferred reference when considering the heat balance of the electrolyzer, since it is fed with liquid water. The use of the specific electricity requirement in terms of kWh/Nm³(H₂) or kWh/kg(H₂) as a practical measure for efficiency is unambiguous in this respect, as it comprises overall energy input for a given output of H₂ on the level of single cell, stack or system.^d

On the system level, the specific energy requirement needs to include balance-of-plant (BoP) components. AC power needs to be converted to DC, produced hydrogen is humid, and has to be dried and eventually cleaned as the purity of gases depends on operating conditions and stack components. This requires recirculation and feed pumps, heaters, power electronics etc., and their energy consumption needs to be taken into account. The common measure of the stack as well as system efficiency in the electrolyzer community is the specific electricity demand given in kWh/kg or kWh/Nm³ of H₂ produced, which encompasses all the losses on the cell level (voltaic, faradaic) and, if quoted on the system level, losses related to the BoP components.

The use of commercially widespread perfluoroalkylsulfonic acid (PFSA) type membranes, such as Nafion, Flemion or Aquivion, limits the choice of electrocataysts and structure materials in the PEWE cell for stability reasons. Electrocatalysts with high activity and stability can only be selected from platinum-group-metals (PGM). At the cathode, usually platinum black or carbon supported platinum based catalysts are used for the HER, similar to the hydrogen oxidation reaction (HOR) catalysts in the polymer electrolyte fuel cell (PEFC).

Iridium oxide and/or iridium-ruthenium oxide blacks are well performing catalysts for the OER. Only iridium based catalyst seems to be sufficiently stable for long term operation. The high electrochemical potential on the anode side of >1.4 V precludes the use of carbon as a catalyst support material, since carbon readily oxidizes under these conditions.¹⁵ Ir and Ru oxides show the highest activity toward OER in acidic media.¹⁶ Although RuO₂ shows lower overpotentials for the OER,¹⁷ IrO₂ remains the material of choice for the PEWE anodic electrocatalyst, since the rate of RuO₂ dissolution and corrosion is inacceptably high.¹⁸

The effort in the research community is targeting the reduction in the PGM loading of the anodic CL using high surface area supported catalysts or multi-metal oxide catalysts. Titanium or tin oxide supports are possible alternatives to lower the noble metal catalyst loading without impairing the performance significantly.^19–21 Promising performance and stability were demonstrated using 0.1 mg/cm² of Ir supported on tungsten-doped titanium oxide.²² There is a potential to increase the active catalyst surface area using improved catalyst manufacturing techniques. Recently, 0.1 mg/cm² of IrOx loading on the anode and a measured cell potential with corresponding N117-based CCM of 2 V at 1.8 A/cm² and 50°C were reported.²³ 3M's nanostructured thin film (NSTF) catalysts with extended surface area have been used for PEWE with an Ir loading of 0.25 mg/cm².²⁴ Additionally, new catalyst synthesis approaches for high surface area IrO₂ reaching specific surface areas of 150 m²/g with significantly increased OER activity have been recently reported.²⁵ Carmo et al. give a historic overview of achieved cathode and anode catalyst types and loadings, with the corresponding performance.²⁶ A detailed overview of the state-of-the-art catalysts used in acidic electrolyzers and prospects are given in Ref. 27. Considering the relatively low contribution of the anode catalyst to the overall stack cost of 6% (cf. Figure 10), the argument for reducing the Ir-loading is not primarily the lowering of the stack cost, but the limited worldwide production of iridium metal, which is a secondary metal, to enable the deployment of electrolysis plants at the scale of GW per year (cf. Degradation phenomena section). Although recycling infrastructure for iridium will moderate issues of its abundance, the iridium loading still influences the overall amount of noble metal in circulation.

Zoom In Zoom Out Reset image size

The membranes typically used in state-of-the-art PEWE cells are relatively thick (5–7 mil; 1 mil = 25.4 μm) PFSA membranes, such as Nafion 115 and 117. This is a compromise between area-specific resistance, low gas (H₂, O₂) crossover, and mechanical robustness.²⁸ Also, chemical degradation of the membrane, resulting in thinning, is less critical in this case. A number of recent reports have highlighted the use of thinner PFSA membranes, yielding lower ohmic losses and thus enabling higher current densities at the same conversion efficiency.^12,24 However, thin membranes are associated with higher gas crossover and susceptibility to mechanical failure.^12,29,30 The latter can be alleviated using reinforced membranes.¹² The higher gas crossover of thinner membranes can be addressed by incorporating a H₂-O₂ recombination catalyst, e.g. Pt or Pd, into the membrane.^31,32 Although this reduces the safety hazard caused by excessive H₂ in the O₂ stream (O₂ in the H₂ product stream is less problematic, because cross-diffused O₂ readily reacts with H₂ on the Pt cathode catalyst), it does not tackle the faradaic loss associated with gas crossover. Alternative, non-PFSA membranes are being developed to tackle some of the issues associated with PFSA membranes. Hydrocarbon polyarylene type membranes, such as sulfonated poly(ether ketones), poly(ether sulfones), polybenzimidazoles, etc., and partially fluorinated membranes comprising a fluoropolymer backbone and grafted styrenic polyelectrolyte generally show a much lower gas permeability compared to PFSA membranes and, consequently, a more favorable combination of resistance and gas crossover characteristics.^29,33 To improve mechanical robustness, in particular for thin membranes, and to improve the creep resistance of the material, mechanical reinforcements are introduced.³⁴ Many of these alternative membranes offer the prospect of lower cost compared to PFSA membranes.³⁵ However, both chemical and mechanical durability of alternative membranes need to be carefully assessed to judge their suitability for use in the electrolyzer.

The use of anion exchange membranes instead of a proton exchange membrane represents a promising approach in terms of combining advantageous features of a membrane electrolyzer (high current density due to small electrolyte gap, absence of liquid electrolyte, possibility to operate at high differential pressures) and an alkaline cell type (non-noble metal catalyst, low-cost structural materials), yet the current membrane technology suffers from stability issues of the anionic head groups, which limits the lifetime of the cell.³⁶ However, with rapid progress in chemistry and stability of alkaline anion exchange membranes,³⁷ it is worth keeping an eye on this technology for future developments.

The major tasks of the porous transport layer (PTL) is the simultaneous transport of charge and heat between electrode and BPP in the solid structure and gas and water in the pore space. In case differential pressure is applied, the PTL has to furthermore provide mechanical support for the membrane. It is conceivable to combine BPP and a PTL into a single component at the expense of a well-designed flow field. At the cathode, carbon fiber based materials in the form of papers or cloths can be used. These materials have so far mainly been developed for fuel cells (e.g. Ref. 38). However, it should be noted that the conditions in electrolyzers and fuel cells in terms of flow directions for the two-phase flows are different and the fuel cell know-how is therefore not necessarily directly applicable.

At the anode, the choice of materials is usually limited to titanium based porous materials, in the form of sintered powders, fibers or meshes due to the high electrochemical potential. However, with suitable coatings, such as TiN, Pt or Au, also stainless steel materials may become viable. The use of an additional microporous layer on coarse PTLs, a concept adapted from PEFC technology, may improve performance by reduction of the interfacial contact resistance.³⁹ In unitized reversible fuel cells (URFCs) typically also titanium PTLs are used at the oxygen electrode. PTFE-coatings of titanium felt⁴⁰ or additional titanium powder at the PTL/CCM interface⁴¹ were investigated. However, in contrast to fuel cell operation, little/no improvement was observed in electrolysis mode.

Modern manufacturing approaches such as electron beam melting (EBM) can be used to fabricate 3-D porous structures from various grades of Ti with lower cost and more flexibility in the design. Mo et al. have demonstrated that a PTL produced this way can lead to a performance improvement as compared to woven mesh Ti PTLs, attributed to the better interfacial contact between the EBM parts and the catalyst layer.⁴² Another approach toward thin PTLs with tunable porosity is to use photochemical machining of Ti foils. PTLs fabricated using this technique have smaller pore openings compared to EBM PTLs. The improvement in performance is again attributed to lower interfacial contact resistance, as the Ti surface between the pores is flat.⁴³

BPPs should have sufficient electrical and thermal conductivity. Often a flow field based on a channel structure, typically in the mm-range, is used to distribute the reactant water evenly over the active area and remove product gases and waste heat. Quite naturally, parallel channels seem to be preferred over serpentine patterns due to a better water and temperature distribution and a lower pressure drop.⁴⁴ But also parallel channel structures may result in non-uniform water flow rate and temperature distributions over the active area.⁴⁵ Considering that BPPs make up a significant share of the stack cost (cf. Figure 10), approaches to reduce the cost of BPP materials and processing should be given a high development priority. There are novel approaches reported in the literature to reduce cell costs by additive manufacturing of the PTLs and BPPs.^42,46 The reported performance compared to the state-of-the-art components is promising, but it remains open if the stability of such components made from Ti and its alloys is comparable to that of the pure metal components.

The increase in operating temperature leads to the reduction of the ohmic and activation losses in the cell, primarily with the membrane having higher ionic conductivity and to a lesser extent due to faster kinetics. The influence of temperature on the electrolyzer performance and the ohmic resistance is exemplarily shown in Figure 3.

Zoom In Zoom Out Reset image size

The temperature limitation for conventional PEWE stems from material stability issues at temperatures above 100°C. The tradeoff between improved performance with increased temperature⁴⁸ and the thermal limitations of the PFSA materials leads to maximum operation temperatures in the range between 70–90°C. The deterioration of mechanical stability of the polymer and higher hydrogen permeation at increased temperatures^49,50 contribute to a safety risk in long-term operation. However, membranes exhibiting good mechanical and gas barrier properties at elevated temperatures and pressures could open a window for the reduction in the PGM-based catalyst loading if the cell performance can be kept at the same level of those operating below 90°C. PFSA membranes with short side-chains, owing to their higher crystallinity at given equivalent weight compared to long side-chain PFSA membranes, could be more suitable for operation and higher temperature.^51,52 In today's commercial systems often the temperature is kept even lower (60–70°C) due to stability issues with ion-exchange resins used to maintain the purity of the process water.

The minimum compression work needed for an ideal isothermal or adiabatic compression can be calculated using Equations 7 and 8, respectively.

where n is the number of moles of gas, R is the ideal gas constant, T is the temperature, p is the pressure of the initial state 1 and the final state 2, and γ is the isentropic expansion factor equal to 1.41 for hydrogen at room temperature.

In Figure 4 ideal isothermal and adiabatic compression losses, normalized to the lower heating value (LHV) of hydrogen of 237 kJ/mol, are plotted against the final pressure. Generally, the isothermal compression is preferable to the adiabatic in terms of compression efficiency by more than an order of magnitude of pressure.

Zoom In Zoom Out Reset image size

Real compressors will have higher losses than the theoretical limits. New 5-stage ionic compressors of Linde⁵³ are claimed to consume 2.7 kWh/kg for compression of hydrogen from 5 to 900 bar, which corresponds to a loss of about 8.1%, which is quite close to the theoretical loss of 5.3% for isothermal compression by a factor of 180.

For the electrochemical hydrogen compressor (EHC), in which hydrogen is concentrated and/or compressed from the anode to the cathode, an isothermal behavior is expected as well.^54–56 For instance, for a single-stage EHC, for compression from 10 to 400 bar, an energy requirement of 3 kWh/kg was reported,⁵⁷ which corresponds to a loss of about 9.0% compared to the theoretical isothermal loss of 3.8% at room temperature. However, with thinner and/or less hydrogen permeable membranes the efficiency of the EHC could improve, making it industrially viable.

Today's commercial electrolyzers produce hydrogen (and oxygen) at gas pressures in the range of 30–50 bar. However, on the prototype level direct electrochemical compression in PEWEs up to 700 bar (differential) pressure has been demonstrated.^12,58,59

Theoretically, the compression loss in a PEWE can be predicted by the Nernst equation (see Equations 4 and 5), where for the calculation of the equilibrium potential assumptions for the activity of gases and water need to be made. For the solubility of hydrogen and oxygen in water up to 100 bar ideal gas behavior for both H₂ and O₂ can be assumed,^60–62 because at thermodynamic equilibrium conditions only the dissolved gas in the water is in contact with the electrodes. Consequently, the activities can be expressed by the partial pressures. The activity of water is assumed as unity.

According to the Nernst equation (see Equations 4 and 5), the thermodynamic cell voltage at 60°C increases by 33 mV per decade of pressure for differential pressure operation (O₂ at atmospheric pressure) and by 50 mV per decade for balanced pressure operation (both gases at higher pressure). While the increasing thermodynamic cell voltage can largely be seen in the differential pressure operation, demonstrated up to 700 bar,^12,58,59,63 as well as in electrochemical hydrogen compressors,^56,57 in experimental data of pressurized PEWE with balanced gas pressures, at relevant current densities, the cell voltage increase with operating pressure is often not observed.^64–66 In this case, processes with a negative overpotential vs. pressure relation are compensating the higher thermodynamic cell voltage. These processes may be kinetic phenomena due to a higher apparent exchange current density and/or improved two-phase flow in the porous structures due to lower gas volumes.^64,67 Gas compression in the electrolyzer at balanced pressure seems energetically inexpensive. However, for efficiency considerations the gas crossover increasing with pressure and reducing the faradaic efficiency⁶⁸ needs also to be taken into account.

The gas crossover can be expressed by a current density equivalent as defined in Equation 9:

where P is the gas permeability, δ is the membrane thickness, z is the stoichiometric factor (z = 2 for H₂, z = 4 for O₂) and p is the partial pressure of the respective product gas. For operating pressures higher than 5 bar_a, the water vapor pressure is small compared to the partial pressure of the gas, hence the latter can be approximated by the total pressure.

According to Schalenbach et al.⁶⁹ the gas permeability for PFSA membranes (e.g. Nafion) increases by about one order of magnitude from the dry to the wet state. Furthermore for wet Nafion, with a temperature increase from 40 to 80°C the gas permeability increases about threefold. With reported values for gas permeability for wet Nafion at 80°C of 5.32·10⁻¹¹ mol·cm⁻¹·s⁻¹·bar⁻¹, the hydrogen gas crossover current density equivalent (normalized to the partial pressure) for a 7 and a 2-mil thick Nafion membrane is about 0.49 and 1.63 mA·cm⁻²·bar⁻¹, respectively. Finally, the faradaic losses due to crossover at elevated pressure, where only the hydrogen gas crossover is considered, can be calculated (with p₂ as the hydrogen pressure), Eq. 10:

The corresponding compression losses, including permeation, for four current density/membrane combinations (2 and 7 mil PFSA membrane at 1, 2 and 10 A/cm²) are compared in Figure 4b. Generally, the losses due to crossover decrease with increasing current density (assuming that the crossover is current density independent). Exemplarily, at 100 bar the losses for cells with a 7 mil PFSA membrane at 2 A/cm² and a 2 mil PFSA membrane at 10 A/cm² are around 2.5% and 1.6%, respectively. In case of balanced pressure operation, oxygen crossover of about half of the hydrogen value should be taken into account as well.⁷⁰ Permeated oxygen recombines partly or completely at the cathodic platinum catalyst, reducing the faradaic efficiency further.

When assuming that hydrogen is produced by electrolysis and neglecting any possible additional losses or safety issues, it is possible to directly compare the compression losses for mechanical and electrochemical compression as shown in Figure 4. It can be seen that pressurized PEWE is energetically favorable as compared to ambient PEWE combined with downstream isothermal compression over a wide pressure range, e.g. up to about 100 bar for a 7 mil PFSA membrane and a current density of 1 A/cm².

Besides efficiency, also safety aspects need to be discussed for operation at elevated pressures. Therefore in Figure 4b the expected H₂ concentrations in O₂ for the four current density/membrane combinations, as well as the corresponding lower explosion limit (LEL) of H₂/O₂ mixtures for differential and balanced electrolysis are plotted. The LEL increases with increasing pressure and slightly decreases with increasing temperature. Here experimental data of Schröder et al.⁷¹ for 80°C up to 200 bar is extrapolated toward 1000 bar.

Production of gases in the PEWE cell is directly proportional to the current density applied. Increasing the current density while maintaining conversion efficiency by using thinner membranes is an approach for increasing the rate of H₂ production per unit cell area of the PEWE. Current densities up to 19 A/cm² have been reported using a 50 μm thick membrane, with a cell potential of 3 V.²⁴ The associated power density is on the order of 50 W/cm². The excess heat from operation at elevated current densities needs to be properly managed to prevent or alleviate degradation. Operation at elevated current densities is also an option to control the effect of gas crossover. Hydrogen gas produced will permeate across the CCM to the anode side driven by diffusion and differential pressure,^68,69 and mix with the oxygen produced at the anode creating a safety hazard as discussed before. At high current densities hydrogen in oxygen gas will be diluted by the increased oxygen production rate.^68,72 However, the hydrogen permeation rate as function of current density is not fully understood and might be influenced by CCM heterogeneities⁷³ or other effects, such as the electro-osmotic drag of water. Variable electrolyzer stack operation invariably calls for designs and engineering solutions that allow operation over a broad range of current densities. The hydrogen in oxygen content is mainly a tradeoff between membrane thickness, gas pressure and current density and to a lesser extent also to temperature (higher cross-over at higher temperatures, see above). Thinner membranes allow for higher current density at a given cell voltage, and thereby effectively decrease the investment costs for the electrolyzer stack, but with the challenges related to the gas crossover, as illustrated in Figure 5b showing simulated polarization curves of an electrolyzer with PFSA membranes of different thickness.

Zoom In Zoom Out Reset image size

Besides pressurized operation and associated faradaic losses, also the operational range, which is also referred to as 'turndown ratio', is of interest. For illustration, the turndown ratio of a PEWE cell with balanced pressure is calculated for an operating pressure of 30 bar and at 80°C as a function of the membrane thickness (Figure 5). The upper bound for the current density corresponds to a maximum cell voltage of 2 V, whereas the lower bound is determined by a maximum of 2% H₂ in O₂. Thus the current density range is limited between a minimum determined by gas crossover and a maximum as a result of the corresponding overpotentials. In this particular case the turndown ratio for 7, 2 and 1 mil PFSA membranes is limited to a factor of 2.6, 1.7 and 1.2 or 61%, 41% and 14% of the maximum current density, respectively. Thick membranes do not allow high operating current densities due to the larger ohmic overpotentials, but have good gas barrier properties. On the other hand, thin membranes are limited in the lower range of the operating currents, since at low current densities the mixture of anodic gases can be explosive. Increasing the pressure imposes another limit on the lower current density due to the higher gas crossover. It is of course possible to expand the operation range by increasing the upper current density at the expense of voltage efficiency, or by decreasing the lower boundary by reducing the gas crossover by using less permeable membranes (cf. Materials and components section) or by employing gas recombiners, e.g. by coating PTLs or flow fields with Pt or Pd.⁷²

So far, different aspects addressing the operational conditions (temperature, pressure and current density) and related challenges were presented. Furthermore, considering technical applications of PEWE a number of boundary conditions have to be taken into account (e.g. H₂ and/or O₂, purity required, final pressure, investment vs. operating cost, etc.) which would be a study on its own. Here we discuss the general advantages and disadvantages of ambient, differential (only H₂ pressurized) and balanced (H₂ and O₂ pressurized) PEWE using key characteristics, summarized in Table II.

For hydrogen at ambient pressure, efficiency and gas purity will be best with ambient electrolysis at the expense of increased drying efforts with the large gas volumes. For pressurized hydrogen, differential pressure PEWE is preferable due to higher gas purity and higher faradaic efficiency compared to balanced operation and lower gas drying costs compared to ambient operation combined with mechanical compression.¹² In case pressurized oxygen is of interest too, balanced pressure PEWE is the method of choice, because of lower efforts for drying and compression for both gases.

For the porous transport layers ex-situ diagnostic tools are commonly used to characterize their structure. As the anodic PTL usually only consists of titanium the porosity can easily be obtained by weighing. For composites based on different types of titanium or even based on carbon including binders, the issue is more complex.

Pore size distributions can be measured by mercury intrusion porosimetry (MIP), capillary flow porometry and/or X-ray tomographic microscopy (XTM) combined with image analysis.^75,76 Especially XTM is a powerful tool to obtain data for anisotropies, tortuosities and particle size distributions among others.⁷⁶ Figure 6 shows XTM derived data for sintered powder materials made of titanium (porosity of 46% and a mean pore diameter of 17 μm).

Zoom In Zoom Out Reset image size

With known geometric parameters transport properties, such electrical and thermal conductivities, as well as permeabilities can be derived. Zielke et al.⁷⁶ reported two characteristic relationships: i) an exponential relation between porosity and thermal conductivity and ii) an exponential relation between through-plane electrical conductivity and water permeability.

Gregoriev et al.⁷⁷ and Ito et al.^48,75 investigated sintered titanium spherical powders and titanium felts decorated with titanium powder on the top surface. Based on electrochemical measurements at atmospheric gas pressure and at temperatures of 80 and 90°C, even for distinctly different structures in terms of spherical powder and fibers, agreement of key parameters and their influence on performance was reported. The porosity of the PTL is claimed to be optimal between 30 and 50%.^48,75,77 Theoretically, a higher porosity should facilitate gas transport and water supply at the expense of a higher ohmic resistance and higher thermal resistance. A mean pore diameter (MPD) larger than 10 μm, but close to this size, was considered optimal. Larger MPD lead to increased parasitic ohmic losses due to inhomogeneous distribution of current in the catalyst layer. The mean particle size (MPS) of a spherical powder seems to have an optimum between 50 and 75 μm. Larger MPS might increase the contact resistance. Smaller MPS tend to increase the mass transport losses due to capillary effects. For fibrous materials, Ito et al.^48,75 reported that a mean fiber diameter of 20 μm seems to be preferable to 80 μm due to a better electrical contact between PTL and CL.

Only few reports about the two-phase flow are available, with the main focus on the channels in the flow field.

Simulating an electrolysis environment using different microfluidic chips as model structures for Ti felt, foam and sintered powder Arabi et al.⁷⁸ showed that the transport mechanism of air bubbles is capillary dominated even at high gas flow rates. Furthermore the (produced) gas is released in pathways within the PTL which are independent of the water flow rate in the channels.⁷⁸ Thus the gas bubbles grow at the electrode surface until the buoyancy and shear forces acting on a bubble exceed the adhesion force.⁴⁴

In an electrolyzer, larger pores in the PTL tend to change the flow regime in the channels of the flow field from a preferred dispersed bubbly into a slug flow,⁷⁵ meaning that the gas bubbles coalesce, which can be avoided by higher water velocities in the channel.⁷⁹

Using high-speed photography and an optically transparent cell, Dedigama et al.⁷⁹ noticed a transition from bubble to slug flow in the channels of the flow field at higher current densities and with progressive channel length, which was associated with improved mass transport. Furthermore, the authors underlined this conclusion by the reduction or nonexistence of a second impedance arc in the electrochemical impedance spectra at low frequencies at high current densities. In subsequent work using current density mapping, an increased current density was measured toward the channel outlet.⁸⁰ However, there is no comprehensive understanding of the current density distribution as van der Merwe et al.⁸¹ reported a decreasing current density along the channel, interpreted as a result of increasing oxygen gas to liquid water ratio. Such inconsistences might be explained by the differences in thermal boundary conditions, flow field or PTL structures.

Through-plane neutron radiography and high resolution optical visualization were used to show that gravity and buoyancy affect the water distribution across the cell as gas bubbles tended to accumulate in the higher regions of the cell.⁸²

With the same imaging technique Hoeh et al.⁸³ quantified the amount of gas in the anodic channel as a function of current density and water flow rate, results are exemplarily shown in Figure 7. As expected, the ratio of gas to water decreases with increasing water flow rate and increases with increasing current density.

Zoom In Zoom Out Reset image size

First approaches in a real electrolyzer were performed by using X-ray radiography in in-plane direction to investigate the cathodic interface of PTL and BPP and in through-plane direction for the anode.⁸⁴ The authors confirmed the presence of gas transport pathways whose number increases with increasing current density.

Using in-plane neutron radiography for a small 1 cm² differential cell, Seweryn et al.⁸⁵ observed that the water/gas distribution inside the PTL is unaffected by the current density up to 2.5 A/cm², which is interesting in terms of possible transport losses. This is the only work which is investigating the PTL under real electrolyzer operating conditions (temperature and pressure). Entirely missing are reports addressing the interface of the PTL/CL or the CL itself.

Compared to the PEFC literature, PEWE degradation studies are scarce. Most of the literature is focused on the degradation phenomena related to the CCM.^88–94 Typical membrane degradation originates from thermal, chemical and mechanical stressors present during operation.⁹⁵ Polymer membranes, usually PFSA membranes, e.g. Nafion, are susceptible to radical induced attack and degradation, which leads to the gradual decomposition of the ionomer. Reactive intermediates, such as hydrogen peroxide (H₂O₂), hydroperoxyl (HO₂^•) and hydroxyl (HO^•) radicals, are formed primarily by the crossed-over gases on the catalytic surface,⁹⁵ of which HO^• has the highest oxidative strength.⁹⁶ These oxidizing species attack the ionomer, leading to chain scission, unzipping, and loss of functional groups. Stucki et al. have observed significant membrane thinning after 15,000 hours of PEWE stack experiments with varying electric input, which eventually led to a short circuit in one of the cells of the stack.⁹³ The degradation was attributed to cathode-side chemical decomposition of the ionomer with eventual scission of the polymer chains.

The fluoride release rate (FRR) is a commonly used metric for quantifying the chemical degradation rate of PFSA membranes and predict the lifetime.^88,90,97 LaConti et al. showed that the FRR increases approximately by two orders of magnitude with a temperature increase from 55 to 150°C.⁹⁰ Based on the literature reports, the cathode side degradation of the membrane is more pronounced as a result of radical formation triggered by O₂ crossover and interaction with hydrogen adsorbed on the Pt catalyst of the cathode.^90,93,98 Also, the electroosmotic drag creates higher concentration of H₂O₂ on the H₂ side. Additionally, the anode catalyst on the O₂ side in the PEWEs is at high potential. Hence, the surface of the Ir-based anode catalyst is in the oxide form, which can suppress the direct combination reaction of H₂ and O₂.⁹⁵

In terms of structural integrity of the membrane, its mechanical properties are of importance, i.e., the yield, tensile strength and elongation at break. Moreover, time-dependent behavior is critical for long-term operation in the PEWE cell, and creep of the ionomer can be the factor limiting the lifetime of the membrane.^30,59 Evidently, mechanical properties of the membrane strongly depend on temperature and humidity. Measuring the stress-strain relationship of immersed Nafion 117 shows that the initial slope of the stress-strain curves, and thus the Young modulus, decreases with increasing temperature.⁹⁹ Softer membrane material could prove less reliable in high-pressure or differential pressure operation modes.

Excessive compression of the cell can lead to damage of the PTL and CCM material. Microscopy techniques are commonly used to visualize morphology changes before and after electrolysis operation. For instance, using scanning electron microscopy (SEM) the morphology change of a catalyst coated membrane (CCM) during electrolyzer operation is illustrated in Figure 8. Clearly the CL is deformed by intrusion into the PTL structure. A similar example (top view) of the anodic side of a CCM after 100 h of electrolyzer operation including cracks caused by excessive cell compression is shown by Millet et al.¹⁰⁰

Zoom In Zoom Out Reset image size

Performance deterioration is mostly related to problems with the purity of feed water, and is a reversible degradation mechanism.¹⁰¹ Metallic cations are dissolved in the de-ionized water that is circulated through the system components made of stainless steel. The cations contaminate the membrane by exchanging with protons in the PFSA material, which leads to decreased proton conductivity. Sun et al.⁹² have confirmed the reversibility of the ion-poisoning mechanism when they measured Cu, Fe and Ca contamination of the MEAs before and after 7800 hours of operation in a 9-cell electrolyzer stack using electron probe microanalysis. The increase of the content of these elements is more pronounced in the anode and membrane region compared to the cathode. After treating the MEAs with 0.5 M sulfuric acid, the content was at a similar level to the one before the electrolysis test, showing the reversibility of the performance loss by ion-poisoning. When the membrane is contaminated by transition metal cations, they can promote the chemical decomposition of hydrogen peroxide and generate radicals, which accelerates membrane thinning.^95,102 Iron and copper ions drastically increase membrane degradation, while cobalt and chromium ions do not appear to play a significant role.¹⁰³ One way to counter the cationic contamination of the membrane and the catalyst is to substitute metal system components (e.g. fittings, piping) with polymer materials. However, this is only possible for low-pressure electrolysis, as the high-pressure rated system components are almost exclusively made of stainless steel. Maintaining high-purity water could extend the CCM lifetime significantly, but current ion-exchangers on the market are not rated for high pressure and temperature operation, and the water purification is mainly done outside of the loop or at low operating temperatures and pressures.

High potentials limit the use of stainless steel cell components due to corrosion.¹⁰⁴ Titanium is the state-of-the-art material for (anodic) PTLs and BPPs of the PEWE cell due to its stability and intrinsic electrical conductivity. Ti can however develop an oxide layer over time, thus reducing the effective electrical conductivity and cell performance. When the reversible cationic contamination of the ion-exchange sites is mitigated, the Ti-PTL passivation is the main contribution to the performance deterioration.¹⁰⁵ Additionally, on the high pressure H₂ side of the system (cathode), Ti-hydride formation can be significant, which in turn weakens the Ti.¹⁰⁶ Coating the Ti parts is often mentioned as a way to protect against oxidation. Coatings of Ti, TiN, Pt or Au on stainless steel or carbon-based composites are possible candidates for the cathode BPPs and PTLs.^12,26

Unlike in the case of fuel cells, there are no standardized AST protocols for PEWE components, which makes the experimental results across the literature difficult to compare. The aging of components is on the other hand an important attribute, considering the DOE target for component replacement intervals being as high as 10 years.¹⁰⁷ There are no standardized sets of stressors and testing protocols to trigger the degradation of specific components. Research efforts in the scientific community are mainly targeted at demonstrating the long term stability of materials, and give a degradation rate in μV/h as a degradation metric. However, with variations in hardware and operating conditions, it is difficult to correlate the testing parameters with changes in degradation rate.