For more information, please visit our website.
Proton exchange membrane water electrolysis (PEMWE) is the technology of choice for the large-scale production of green hydrogen from renewable energy. Current PEMWEs utilize large amounts of critical raw materials such as iridium and platinum in the anode and cathode electrodes, respectively. In addition to its high cost, the use of Ir-based catalysts may represent a critical bottleneck for the large-scale production of PEM electrolyzers since iridium is a very expensive, scarce, and ill-distributed element. Replacing iridium from PEM anodes is a challenging matter since Ir-oxides are the only materials with sufficient stability under the highly oxidant environment of the anode reaction. One of the current strategies aiming to reduce Ir content is the design of advanced Ir-mixed oxides, in which the introduction of cations in different crystallographic sites can help to engineer the Ir active sites with certain characteristics, that is, environment, coordination, distances, oxidation state, etc. This strategy comes with its own problems, since most mixed oxides lack stability during the OER in acidic electrolyte, suffering severe structural reconstruction, which may lead to surfaces with catalytic activity and durability different from that of the original mixed oxide. Only after understanding such a reconstruction process would it be possible to design durable and stable Ir-based catalysts for the OER. In this Perspective, we highlight the most successful strategies to design Ir mixed oxides for the OER in acidic electrolyte and discuss the most promising lines of evolution in the field.
Keywords:
Water Electrolysis, Green-Hydrogen, Iridium, Mixed-Oxide, Oxygen Evolution Reaction, Connectivity, PEMWE
Despite the increasing number of government initiatives, plans, and actions to limit the temperature rise to 1.5 °C,1−3 global CO2 emissions have increased every year since 2015, with the exception of 2020. Electrification of energy and production sectors with renewable energy and the use of green hydrogen and sustainable biomass in sectors where direct electrification is not feasible are the cornerstones for meeting the 1.5 °C target.4 In addition, and triggered by recent geopolitical tensions, many countries have taken decisive actions to bolster their energy security mostly by promoting the use of domestically produced renewable energy. Within this context of energy transition, (green) hydrogen is gaining momentum in both academia and industry.5 The term green hydrogen was coined to identify the hydrogen obtained from carbon-decoupled sources, ideally by using renewable energy.6 It can be produced by several technologies, including water electrolysis, biomass gasification, and photochemical and thermochemical water splitting.7
Water electrolysis is the most developed technology for producing green hydrogen in the GW to TW scale needed to realize the climate neutral targets.6 There are several types of electrolyzers, namely, alkaline electrolyzers (AWE), Proton Exchange Membrane Water Electrolyzer (PEMWE), Anion Exchange Membrane Water electrolyzer (AEMWE), oxygen ion-conducting Solid Oxide Electrolyzer (SOEC), and proton-conducting SOEC, also called proton-conducting ceramic electrolyzer (PCCEL), with the first two technologies being the most developed ones.8
Electrolyzers can be classified according to different operational or composition parameters. Thus, AWE, AEMWE, and PEMWE operate at low temperatures (40–90 °C) while solid oxide technologies work above 300 °C in the case of PCCEL and around 700 °C for SOEC.5 On the other hand, in AWE and AEMWE, hydroxyl ions (OH–) are transported through the electrolyte, namely, a diaphragm in AWE or a membrane in AEMWE, whereas protons (H+) are transported in PEMWE and PCCEL. AWE and PEM electrolyzers are already commercially available and in fact alkaline electrolyzers are widely used in the chlori-alkaline industry to produce chlorine and sodium hydroxide, but also hydrogen.5,8 The components of AEM electrolyzers are based upon Ni-based materials, which is a clear benefit to their scaling-up. The main drawback of AWE is the intermixing between the H2 and O2 generated during operation, and work at low operating pressures. In addition, they are optimized for continuous operation, a feature that conflicts with the intermittent production of renewable electricity. These features can be avoided with AEMWE, since they can operate at high pressure, while obtaining high purity hydrogen without using PGMs in their components. However, the membranes still need to be improved with mechanical and thermal stability in the electrolysis conditions, efficient to OH– transfer, low permeability of gases, etc.5,7,9 PEMWE technology is the most effective technology for the production of green hydrogen from renewable electricity. PEMWE uses solid electrolyte membranes based upon perfluorinated sulfonic acid copolymer membranes, which display high ionic conductivity (low resistivity 5–6 Ω cm at 80–90 °C), resulting in high efficiencies and allowing the possibility of working at high current densities (4–6 A cm–2 or even higher).5,10 PEMWE can work in a dynamic mode, with a very rapid transition between idle and full load periods (in less than one s at nominal temperature). As a consequence, they are suitable for work under the dynamic operation conditions imposed by the intermittency of renewable electricity generation. Moreover, due to the compact design, lower thermal capacity and operating temperature cold start-up in scale-up PEMWE systems have been reported to take 5–10 min.11 Due to the low gas crossover, high-purity hydrogen (99.99–99.9999%) can be obtained, which furthermore can be pressurized during electrolysis, reducing the costs of subsequent pressurization. Finally, they have a very compact and easy-to-stack design, which allows them to be easily scaled.12,13 The most critical feature of PEMWE is the use of costly materials in the electrodes: Pt and Ir in the cathode and anode electrodes, respectively, and Ti in the bipolar plates and porous transport layers and the degradation of the polymeric membrane. SOEC technologies can operate at high current densities and high efficiency, and their electrochemical processes are reversible, allowing them to work in fuel cell or electrolysis mode. However, the electrochemical degradation and thermomechanical stability of their components still need to be improved.14 PCCELs are also very efficient, but they still have difficulties in the configuration of the electrolyte/porous electrode support and also poor thermomechanical properties.15
The basic operation of a PEMWE is as follows: at the anode electrode, the Oxygen Evolution Reaction (OER; H2O → O2 + 4H+ + 4e–) takes place in the presence of an Ir-based catalyst. At the cathode electrode, the Hydrogen Evolution Reaction (HER; 4H+ + 4e– → 2H2) takes place. Although H2 is the desired product, the overall water electrolysis reaction is limited by the sluggish kinetics for the OER. As a result, most of the research concerning electrocatalyst development for water electrolysis focuses on the OER.16−20 Note also that the OER is a key process in other applications such as photoelectrochemical devices, regenerative cells and metal-air batteries, so the improvement of this reaction is essential in the whole scenario of sustainable energy.21,22
Although water electrolysis is a commercially available technology, the cost of green hydrogen is higher than that of hydrogen produced from fossil-fuels, typically steam methane reforming.6,23 This accounts to several features, including the high price of renewable electricity and to the high price of today’s PEMWE.8 This scenario is expected to change in the near term, with projected prices of renewable hydrogen close to, or below, 1 USD/kgH2.24 On the one hand, the price of renewable electricity is decreasing rapidly, mostly due to the decreasing price of photovoltaic panels. On the other hand, the high cost of today’s commercially available PEMWEs stacks accounts to the use of nonoptimized materials and components, and low-scale production.8 Platinized titanium bipolar plates (BPs) and porous transport layers (PTLs) are major cost components, with the cost of the catalyst coated membrane (CCM) representing around 24% of the total cost of the stack for a 1 MW PEMWE stack electrolyzer, with the contribution of PGMs (Ir and Pt) used in the electrodes representing around 5–8% of the total price.8 The novel generation of PEMWE are expected to contain a significantly lower amount of Platinum Group Metal (PGM) elements (Ir and Pt), while replacing Ti-based components (PTLs and BPs) by affordable stainless-steel based ones.25
Due to the very fast kinetics of the HER in acidic electrolyte, Pt loading can be reduced without losing performance, the typical Pt loading in the cathode being ≈0.5–1.0 mgPt cm–2.26 Ir-based electrocatalysts are the only ones that can withstand the harsh oxidizing conditions in the anode electrodes of PEMWEs. Due to the sluggish kinetics of the OER a high iridium loading (between 1 and 3 mgIr cm–2) is used in the anode electrode of today’s PEMWEs.26 Ir is one of the rarest and ill-distributed elements on earth’s crust. As a consequence, Ir supply is currently dominated by South Africa (ca. 87% of global Ir production), followed by Zimbabwe (8%), Russia and Canada (3% each).27 Therefore, Ir is extremely expensive (4600 $/troy oz as of May 2023).28 In addition, the very small market results in Ir prices being strongly subjected to speculative purchasing, hence posing a serious risk for the large scale (GW to TW) deployment of electrolyzers. It is therefore needed to develop novel anode electrodes that can display similar performances with lower Ir loadings of ca. 0.5–0.2 mgIr cm–2 or lower, that can deliver high current density at low potentials while displaying high durability in the range of tens of thousands of hours.
In this Perspective, we will first identify the state-of-the-art electrocatalysts for the OER in acidic media. Next, we will introduce the Ir-mixed oxides reported in the literature and analyze features such as their crystal structure and morphology and their relationship with their OER activity. Finally, we will discuss the evolution of the Ir-mixed oxide structure and composition during the OER and indicate good practices and recommendations to assess their OER performance.
Through this paper, we will report and discuss the OER activity and durability data obtained in aqueous acidic electrolyte using rotating disk electrodes (RDEs) or, when available, in lab-scale PEMWE. Although the performance of promising newly developed catalysts should be tested in PEMWE, RDE is a suitable approach for the fast screening of a large number of electrocatalysts, usually prepared in the milligram scale. In fact, most scientific studies dealing with the development of novel electrocatalysts report only activity and durability data from RDE experiments. Sadly, and contrary to the oxygen reduction reaction (ORR), a benchmark protocol for testing OER performance of an oxygen ion in RDE (and in PEMWE) has not been universally implemented and followed by the scientific community yet, so activity data obtained in a wide range of reaction conditions are usually reported. As a consequence, it is very difficult, if not simply impossible, to establish proper comparisons between the OER performances of novel electrocatalysts measured in different laboratories. The final section of this perspective will discuss both approaches and indicate recommendations to perform RDE testing.
The two most accepted mechanisms by the scientific community are the adsorbate evolution mechanism (AEM) and the lattice-oxygen mechanism (LOM). The AEM mechanism was proposed by Rossmeisl and co-workers through DFT calculations.29 The mechanism is schematized in a, where the asterisk (*) represents a surface-active site. The overpotential of the OER is strongly related to O* surface binding energies of the O*, and therefore, the electrochemical activity is limited by the O* and the OOH* formation steps. For materials that bind O* too strongly, the activity is limited by the formation of OOH*, whereas for surfaces on which O* binds too weakly, it is limited by the O* formation. For example, RuO2 binds O* a little too weakly, whereas IrO2 binds it too strongly.29
Open in a separate windowThe first two steps of the LOM mechanism ( b) are the same as those for the AEM. The difference in the following steps is that in the LOM, the O* together with an oxygen from the catalyst forms O2(g), in such a way that an oxygen vacancy (□) is generated on the surface. Finally, this vacancy is replaced with oxygen from a water molecule. Clearly, the presence of oxygen vacancies is a key feature in the LOM.
Unfortunately, and to the best of our knowledge, it remains unknown which structural or compositional factors make a catalyst proceed through a specific mechanism. As observed from , the active sites for the AEM are coordinating unsaturated metal ions, whereas the active sites for the LOM are coordinating unsaturated oxygen ions. The LOM mechanism is favored when the Fermi level decreases to the oxygen 2p band induced by the strong overlap of metal 3d with oxygen 2p.30 Moreover, the LOM pathway is easier to occur when the catalyst’s lattice parameters are small and the separation between surface oxygen is short, features than can be controlled by tuning the bending of the metal–O–metal bond by properly choosing the nature of the metal.30 Moreover, the presence of oxygen vacancies seems to be an essential factor for the occurrence of the LOM.17,19,31 Thus, it is more likely to observe the LOM process for amorphous catalysts since it is easier to add or remove oxygen in the lattice. For more information about the OER mechanisms, please refer to the following references: refs (32) and (33).
The benchmark electrocatalyst for the OER in acidic media is rutile IrO2. However, some papers report activity values obtained with homemade IrO2,34,35 while other researchers use commercial IrO2 samples obtained from different suppliers such as Umicore, Haereus, Sigma-Aldrich Corporation, or Alfa Aesar.36,37 For instance, Ir mass-normalized activities of ca. 10 and 42 A gIr–1 at 1.525 and 1.6 V, respectively, were obtained with IrO2 purchased from Sigma-Aldrich (IrO2, 99.9%) in 0.5 M H2SO4.36 However, mass-normalized activities of 50 and 475 A gIr–1 at 1.525 and 1.6 V, respectively, were obtained with an IrO2 catalysts purchased from Alfa Aesar (99% Ir).36 In agreement with the former value, a similar value of ca. 30 A gIr–1 at 1.51 V was reported by Daiane Ferreira da Silva et al.38 for an IrO2 catalyst purchased from Alfa Aesar (99.99% trace metals basis) in 0.05 M H2SO4. Moreover, they tested several supported and unsupported Ir-based catalysts for the OER and concluded that whereas supported samples provided higher activity than unsupported ones, they also showed less stability. As shown above, the activity of the Ir-based electrocatalysts strongly depends on the nature of the initial Ir species, the synthesis procedure and the protocol for the OER measurement.36,39 Therefore, it is highly recommended to compare the activity results to those obtained by using commercially available electrocatalysts.
In addition, the activity of IrO2 measured in PEMWE depends on a number of factors,40 including, but not only, the thickness of the membrane, the structure of the PTL and its coating, the cell temperature, and the catalyst loading on the electrode. A catalyst coated membrane (CCM) produced with Nafion 212 and an Ir loading of 2 mgIr cm–2 on the anode at a temperature of 80 °C can reach a current density of 4 A cm–2 at 1.79 V (70% efficiency vs LHV).41 When measured in PEMWE, Ir-mass normalized activities of 12 and 420 A gIr–1 have been reported with two different catalysts, namely, c-IrO2/TiO2 purchased from Umicore and a-IrO(OH)x/TiO2 from Heraeus Deutschland, respectively.42 Contrary to RDE measurements, when measured in PEMWE cell, Ir-based electrocatalyst are usually supported/dispersed onto a (more or less) conducting oxide, typically TiO2, or due to its poor conductivity, onto ATO (antimony tin oxide),12,26,43,44 Due to the low conductivity of these oxides, a high loading of Ir, between 40 and 70 wt %, is usually needed to produce a catalyst layer with sufficient conductivity.
In order to succeed in the realization of a hydrogen economy based on water electrolyzers in the GW to TW scale, it is imperative to decrease Ir loading on the electrodes of PEMWEs from today’s 1–3 mgIr cm–2 to more feasible values of around 0.2–0.4 mgIr cm–2, or lower. However, several studies concluded that decreasing Ir loading to 0.5 mgIr cm–2 using the same IrO2/TiO2 catalysts that are used in state-of-the-art CCMs with 2 mgIr cm–2, will result in very thin and inhomogeneous catalyst layers resulting in high performance loss.16 Therefore, novel catalysts should be designed to be used in low loaded CCMs. These catalysts should display high mass-normalized OER and durability but also display high conductivity and optimized packing density.
The most traditional approach to increase Ir mass-normalized activity is to maximize catalyst dispersion, i.e., to increase the fraction of surface atoms vs bulk atoms. One obvious strategy to maximize catalyst dispersion is to reduce the size of Ir particles to the nanometer range. Reier et al.45 reported that Ir nanoparticles have comparable OER activity and durability to bulk Ir, indicating their potential as nanoscaled OER catalyst. However, the electrochemical surface characteristics of Ir nanoparticles differed from those of bulk Ir material, with Ir nanoparticles losing their voltammetric metallic features during voltage cycling, indicating changes in oxidation chemistry. Overall, the results suggest that Ir nanoparticles could be a promising option for the OER catalysis. A number of studies have focused on synthesizing nanoscale IrOx materials to improve their mass activity compared to bulk IrOx.46 These studies include rutile nanoparticles, Ir nanodendrites47 or Ir single atoms.48 Results showed potentials of around 1.48 V for Ir single atoms and 1.65 V for Ir nanoparticles in the RDE at 10 mA cm–2. Ir nanoparticles were also studied in a MEA type electrolyzer49 and displayed a 10-fold higher OER activity compared to that of flake-like structured Ir-black, not only because of the increase in the surface area but also due to the nanoporous structure of the Ir-nano catalyst. These findings suggest that nanoscale IrOx materials have the potential to improve the efficiency of the OER catalysis. Recently, the formation of IrO2 nanoribbons, with the monoclinic phase, has been reported to result in very active Ir sites for the OER with a low potential of 1.435 V to achieve 10 mA cm–2 in RDE.50 Other approaches for increasing the fraction of Ir atoms at the surface include the formation of core@shell structures, in which core elements, which are usually inactive particles such as Au,51 Co,52 or Pd.53
Theoretical studies suggest that the energy of oxygen bonding onto the surface of the catalyst determines its OER activity, with rutile RuO2 and IrO2 displaying the most favorable energies. However, neither of them is located at the apex of the volcano curve; RuO2 binds oxygen too weak and IrO2 too strong.29 In principle, it would be possible to tune the energy of oxygen adsorption on the catalyst surface by doping RuO2 or IrO2 with other metals. The number of doping agents is very high, including transition metals such as Ti, Mn, Mo, etc., or noble metals such as Pd, Au, or Ru.54−56 For instance, RuIrOx improved the activity of IrO2, yielding 1225 A g–1noble-metal.57 One of the most active catalysts was obtained by doping IrO2 with Ta and Tm by using a fast pyrolysis, resulting in an mass-normalized activity of 3126 A g–1Ir at 1.5 V for Ta0.1Tm0.1Ir0.8Ox.58 Other catalysts based on doped IrOx showing high mass-normalized activity have been reported: 100 and 900 A gIr–1 for Li–IrOx and W0.8Ir0.2Oy, respectively, at 1.525 V.59−61 The nature of the doping agent can affect parameters such as Ir’s oxidation state, Ir–Ir and Ir–O–Ir bond lengths, and orbital hybridization. The increased activity of Ta0.1Tm0.1Ir0.8Ox was ascribed to strain and tuned electronic structure leading to an optimized electronic structure. Lower oxidation state than +4 and longer length of the first shell Ir–O (2.06 Å) compared to IrOx were also observed. The improved activities of Ir0.6Mn0.4Ox and RuIrOx were ascribed to the existence of Ir3+ and a shift of the Fermi level to the d band center. Consequently, the adsorption of oxygen intermediates is enhanced and the energy barrier of the potential-determining step decreased.57,61 Other effect that has been reported to result in higher OER activity is the segregation of phases, giving rise to the formation of iridium oxide on the surface.60 However, when IrOx was doped with Li the faster OER was ascribed to the formation of more flexible, disordered IrO6 octahedra, which are more easily oxidized during OER, along with the shrinkage of the Ir–O bond, acting as more electrophilic centers.59
In doped-IrOx (or Ir black), the dopant element replaces a fraction of Ir cations in the IrOx network, with both atoms occupying the same crystallographic positions randomly. Therefore, it is difficult, if not impossible, to pinpoint the precise location of the dopant atoms in the oxide network, hence to understand or tune the effect of such element. In the following sections we will present Ir-mixed oxides as potential candidates to develop advanced OER catalysts with high activity and durability. Due to their unique characteristics, namely, easy control of their composition and structure, Ir mixed-oxides allow the identification of accurate structure–activity descriptions, that would permit one to rationalize the designing of advanced Ir catalysts.
In order to be suitable for practical applications, Ir-mixed oxides should combine high activity and durability. Whereas obtaining electrocatalysts with high OER activity is possible, see above, increasing the stability of Ir-mixed oxides is a still a grand challenge, since even pure IrO2 dissolves at potentials above 1.6 V.98 Moreover, the dissolution processes may be exacerbated by the presence of non-noble metal elements in Ir-mixed oxides. As a result, almost all, if not all, Ir-mixed oxides are unstable during the OER operation and tend to dissolve and reconstruct during the process.
The quantification of the dissolution of Ir, or the non-noble elements, has been observed and quantified in several papers, mostly by analyzing the content of the element dissolved in the electrolyte and its evolution with time. Recently it has been proposed that it is possible to predict the durability of electrocatalysts during the OER from the ratio between evolved oxygen and the amount of dissolved iridium. This metric, which has been proposed by several groups, is referred to as S-number99 or activity stability factor.100 As shown in a–c, the highest S–numbers were obtained for crystalline IrO2, with Ir-mixed oxides displaying 2 orders of magnitude lower S-number, hence lower durability. However, since the evolution of metal dissolution was conducted during a somehow short period of time, this metric fails to consider the possible stabilization of the phases formed during operation, a feature that has actually been reported for Ir double perovskites.80,84 Also, it is important to remark that S-numbers obtained in aqueous electrolyte with Ir-catalysts are significantly smaller (by ca. 5 orders of magnitude) than the ones obtained in PEMWE configuration.101
Open in a separate windowIn order to achieve a rational design of advanced catalysts, it would be desirable to be able to predict the stability of Ir-mixed oxides under typical OER conditions. In principle, this is possible by constructing Pourbaix diagrams, which allow us to predict the thermodynamic stable phases of a chemical species based on pH and E. However, constructing Pourbaix diagrams based upon experimental results for all possible mixed oxides is not feasible. To address this issue (and others), the Massachusetts Institute of Technology and Lawrence Berkeley National Laboratory collaborated to create the Materials Project. This initiative has resulted in the development of a vast database, which enables the construction of Pourbaix diagrams using computational data for thousands of compounds , including those that have not yet been synthesized.102 Recently, a new approach that quantitatively evaluates the thermodynamic stability of a phase by comparing the difference in Gibbs free energies (ΔGpbx) such as phase vs potential and pH with respect to the stable phase has been proposed. A stable material should display ΔGpbx = 0; however, it is suggested that ΔGpbx below 0.5 eV atom–1 would suffice to suggest that the material would be stable under the studied conditions.103 demonstrates how DFT, ΔGpbx, and the Materials Project database can be used to predict and discover the stability of new acid-stable compounds. The Pourbaix diagrams for monometallic Pd and Ir with respect to the water system are shown in d and e. f displays the Pourbaix diagram for bimetallic Ir–Pd oxides, with the color indicating the value of ΔGpbx for one of the phases with respect to the other phases. The bluer the color is, the closer ΔGpbx is to zero, indicating greater stability. From d–f, it is possible to predict that when Ir and Pd are combined to form Ir0.5Pd0.5O2, the resulting compound is even more stable than the monometallic phases, especially in regions close to 2 V.
This approach improves the accuracy of predicting thermodynamic stability, especially for catalysts that are not theoretically in a region of high stability. However, it is important to note that kinetic factors are not considered when Pourbaix diagrams are used to study stability. Just because a material is stable in a particular phase or has a low ΔGpbx does not guarantee that it will be kinetically active for the OER. However, the complexity of mixed oxides makes it challenging to identify which phases these materials can form, and interpreting diagrams with many phases can be difficult. Therefore, while Pourbaix diagrams are useful, their limitations should be considered, especially when dealing with mixed oxides. Another way to achieve a rational design of stable and active catalysts for the OER in acid is through the use of machine learning. However, it requires a precise predictive model, accurate input information, and high-quality data obtained via DFT or experimental methods. Furthermore, a large amount of data is necessary for training. To ensure accurate predictions, it is essential to validate the results experimentally and continue training the system with the obtained values.
Regardless of the actual structure and composition, metal-based mixed oxides are prone to dissolution in an acidic electrolyte. Several strategies are being considered to prevent and delay the dissolution of mixed oxides. Recent studies suggest that is possible to increases Ru’s stability during OER by doping with monovalent elements in A sites of the perovskite.105,106 To our knowledge, this approach has not been proved with Ir-mixed oxides.
The dissolution of the non-noble metal and Ir from mixed oxides has profound implications for the OER activity. On the one hand, since Ir is lost during the OER, the durability of the catalysts becomes compromised. On the other hand, the dissolution leads to the reconstruction of the mixed oxide,99,105,107 so the actual structure and composition of the reconstructed solid, especially at the surface level, may be completely different to that of the as synthesized material. Although in general after reconstruction, a more or less amorphous surface layer of Ir-oxides is formed, there is no universal reconstruction pathway. The extension of the reconstruction and the nature of the reconstructed material depend on the structure, composition, and morphology of the as prepared Ir-mixed oxide.
Among the Ir-mixed oxides reported in the literature, hexagonal perovskites with strong IrO6 connectivity between columns exhibit the highest level of OER stability. According to Yang et al.,66 the presence of face-sharing octahedra in 6H-SrIrO3 prevents Sr dissolution, and only ∼1% of total Sr content is leached after 30 h of OER, whereas ∼24% of total Sr content is leached from 3C-SrIrO3. In this line, it has been reported that Co-doped 6H-SrIrO3 is stable during the OER, without reporting surface reconstruction.68 The high stability of 9R-BaIrO3 has also been attributed to the IrO6 connections, with 9R-BaIrO3 being more stable than 3C-SrIrO3, and even more stable than iridates with lower connectivity.69
Other works, however, report the formation of IrOx nanoparticles at the surface of 9R-BaIrO3 during the OER cycles. After further OER cycling, the surface evolves into amorphous Ir4+OxHy/IrO6 octahedra and then to amorphous Ir5+Ox/IrO6 octahedra on the surface.67 Mn-doping in 9R-BaIrO3 promotes the dissolution of Ir, so there is a good balance between Ba- and Ir-dissolution, and a robust and thin surface layer of BaIr1–xMnxO3 with strong IrO6 connectivity is formed at the surface.69
Concerning the perovskite catalysts; SrTi1–xIrxO3 is the only catalyst reported to display highly crystallinity after OER, without surface amorphourization ( a).75 The formation of surface layers of amorphous IrOx during OER has been reported for 3C-SrIrO3 and 3C-BaIrO3,108,109 similar to SrCo0.9Ir0.1O3-δ, with range-ordered amorphous IrOx layers,73 and Sr2CoIrO6, Sr2FeIrO6 and Sr2Fe0.5Ir0.5O4.81,110 SrIr0.8Zn0.2O3 perovskite also forms an Ir-rich amorphous phase, that after OER cycles becomes a resistive material, which is no longer electrochemically accessible.76 In the case of Ba-based double perovskites, the leaching of the non-noble elements leads to the formation of highly active amorphous iridium oxide.99 La2LiIrO6 evolves into IrO2 particles at the surface;19 these particles are small but crystalline (see b).
Open in a separate windowYou will get efficient and thoughtful service from GIANT ANODE.
The actual restructuring experienced by iridium-mixed oxides during the OER depends not only on their structure but also on their composition. This is clearly observed by following the reconstruction of a family of double perovskites with the same structure, Sr2MIrO6, in which only the cation M is modified. Whereas Sr2FeIrO6 particles display the characteristic crystallographic features of the initial perovskite after 50 OER cycles, the composition, morphology, and structure of Sr2NiIrO6 is strongly affected, and after 50 OER cycles the dissolution of Ni and Sr leads to the collapse of the perovskite particles and the formation of ill-crystalline nanosized IrOx particles.81 Due to the very fast dissolution of Ca and Sr, Sr2CaIrO6 experiences a different restructuring pattern, which commences already upon immersion in aqueous acidic electrolyte (0.1 M HClO4). The rapid dissolution of the cations results in the formation of hollow particles of corner- and edge-sharing IrO6 dimers, in a very open structure;80 see c. However, the morphology of the particles is not strongly affected, as observed with IL-TEM, and the formation of nanoparticles of IrOxHy is not observed. Once formed, this open structure remains very active and stable during OER, as observed from the long-term experiments conducted in RDE and PEMWE configurations; see f and discussion above.84 The restructuration observed on Sr2CaIrO6 is somehow similar to the previously observed for SrIr1–xCrxO3.78 In both cases, the very fast leaching of Ca and Sr cations leads to surfaces rich in edge-shared Ir-oxygen octahedra. In this kind of reconstructions, leading to the formation of hollow surfaces, it has been also reported that the voids formed during the reconstructions are rapidly filled by hydroxonium (H3O+) ions,91,111−113 thereby stabilizing the hollow structure of nanosized (short-range order) clusters of IrO6 octahedra.
Concerning the pyrochlores, Hubert et al.114 reported that cation dissolution during electrochemical testing, resulted in the formation of a surface layer of IrOx on Y2Ir2O7. The Ir oxidation state at the surface is dynamic, with Ir more oxidized than the bulk. In ref (115), the formation of a highly active IrOx surface layer due to leaching of the Y3+ cations into the electrolyte solution was also reported. In Lu2Ir2O7, a surface reconstruction into a metastable [IrO6]–[IrO6] framework occurs due to Lu3+ leaching. The reconstructed IrOx/Lu2Ir2O7 structure contributes to the broadening of the t2g bandwidth, which reflects the downshift of the d-band center. The delocalized feature of Ir 5d conducted the reduced p–d separation and the reduced energy gap, which demonstrated the enhanced Ir–O covalency (enhance the Ir–O hybridization) and the enhanced conductivity of the [IrO6]–[IrO6] framework. After several OER cycles, such an active framework evolves into an amorphous layer with low conductivity, that is the main cause for the decreased of the OER activity.116 In Sm3IrO7, also IrOx is formed at the surface ( d).92
Ruddlesden–Popper Sr2IrO4 reconstructs in corner-shared and under-coordinated IrO6 octahedra, responsible for their high activities.91 Finally, in oxides such as Ca4IrO6 and Sr4IrO6, in which the IrO6 octahedra are completely isolated with no connection between them, a rapid disintegration of the crystal lattice can be anticipated if Ca and Sr leach out in an acid solution.63
To summarize, during the OER, Ir-mixed oxides undergo a reconstruction process triggered by the dissolution of the cations in their structure. As a result, the composition and structure of the mixed oxide are modified in one (or several) of the following ways. The vast majority of mixed oxides develop a surface layer of amorphous IrOx, that in most cases continuously evolves until the materials deactivates, probably due to the complete dissolution of IrOx or due to the formation of a resistive and inert surface. In other oxides, e.g., Sr2FeIrO6 and La2LiIrO6, the dissolution of the non-noble cations leads to the formation of IrO2 nanoparticles, either deposited on the surface of the oxide or as isolated nanoparticles. A third group of materials in which the non-noble metal cations leach very fast, typically Ca, suffers a very fast restructuration into hollow and very active structures with edge-shared IrO6 octahedra. Finally, a small group of oxides, mainly face-shared hexagonal perovskites such as SrTi1–xIrxO3 and Y2Ir2O7, appear to be stable during the OER.
These reconstructions are not compartmentalized, and in many cases, they are interrelated processes taking place more or less simultaneously during the different stages of the OER. In fact, since the actual nature of the outermost layers of the reconstructed catalysts is difficult to be determined accurately, further advanced characterization studies are needed to understand the reconstruction of Ir-mixed oxides. In this sense, in situ/operando techniques, typically XRD, XAS, and XPS, but also IL-TEM, can offer invaluable information about the reconstruction pattern of mixed oxides during the OER, while allowing to identify the nature of the surface (and bulk) species responsible for the OER activity.33,98,117−121
In the previous sections, we have identified a number of discrepancies between activity data reported in the literature for identical catalysts from different laboratories. This is a major hurdle for the proper identification and understanding of the parameters that control the OER activity of mixed oxides and hence for the rational design of advanced OER electrocatalysts. In part, these discrepancies account for the wide range of protocols used for the assessment of the OER activity. RDE is an effective and cost-efficient method for prescreening novel electrocatalysts for the OER. As shown in this perspective, almost all of the studies published about Ir-mixed oxides for the OER report data obtained only from RDE.
Standardized protocols for measuring catalyst activity for OER in RDE in acidic electrolyte have been reported,122 but mixed oxides may undergo significant restructuring during the measurement, altering their crystallographic structure. Some researchers report electrochemical activities and restructuring in acidic media before or after applying a potential program, making it difficult to compare catalysts across studies and identify the active phase and true activity for the OER. As such, protocols must be expanded and adapted for mixed oxides.
We propose a protocol outlined in for studying mixed oxide catalysts. Step 1 involves immersing the synthesized catalyst powder in acidic electrolyte for at least 60 min, washing, and characterizing it to determine the stability of the crystal structure. This crucial step is rarely performed. Note for instance, that non-noble cations can dissolve already during the preparation of the ink. The choice of acid for RDE measurements is a topic of debate,123 with some studies suggesting that HClO4 may be better than H2SO4 due to the strong adsorption of SO42– ions on the surface of the catalyst.124 The counterion effect may be more pronounced in mixed oxides, where the use of H2SO4 can lead to precipitation with leached cations and passivation.125 Therefore, it is recommended to investigate the effect of each acid on the new material.
Open in a separate windowAfter assessing the catalyst stability in the electrolyte and determining its nature, step 2 is to deposit the catalyst on the RDE (0.255 mgcat cm–2), immerse the electrode in an O2-saturated electrolyte, and wait for the open circuit potential (OCP) to stabilize. The OCP stability indicates the stability of the electrode–electrolyte interface. The next steps follow a similar protocol proposed previously.122 Step 3 is to measure electrochemical impedance spectroscopy (EIS) at a potential where electrocatalysis takes place. At these values, it is possible to determine Rs and use it to account for the ohmic resistance (iR) during the measurement of the OER activity. We have noticed that throughout the articles cited in this work, different % of iR compensations have been used without any justification. Unless properly justified, 100% of the iR should be used to correct activity curves.126 Additionally, by EIS we can calculate the double layer capacitance (Cdl), which will allow us to determine the electrochemical surface area (ECSA) to later determine the intrinsic activity of the catalysts under study. While it is possible to estimate the ECSA using voltammetric measurements, it can be challenging to do so for mixed oxides. In these cases, it is recommended to use EIS instead of voltammetric measurements.127,128 EIS allows the capacitive component to be considered as a Constant Phase Element (CPE). This approach not only allows for the determination of Cdl but also accounts for the nature of the surface roughness.129 Furthermore, if required, EIS facilitates conducting more comprehensive studies,130 enabling the investigation of the physicochemical properties of the material at the nanoscale, similar to those examined in supercapacitors.131,132 Moreover, measuring the CPE by EIS allows us to study the surface reconstructions and to determine the evolution of the surface area during the OER cycles (see an example in SI Figure S3).
Once the EIS measurement has been carried out, the next step (step 4) is to study the catalytic activity with cyclic voltammetry (CV) at a low scan rate of 10 mV s–1. This technique allows to extract the potential at 10 mA cm–2 and the Ir mass-normalized activity, i.e., the OER activity per gram of Ir on the electrode (expressed in A g–1Ir) at a potential of 1.525 V. These values are important for comparing different catalysts.133 Generally, this last value is given at a potential of 1.525 V, although in some cases, other values are reported. While this value varies strongly every few mV, we recommend always extracting the value at 1.525 V for better comparability among different mixed oxides. Across the papers, obtaining accurate and precise data can be challenging, and it is recommended that numerical values be reported rather than just graphical data. Precise data can be analyzed using machine learning techniques to predict new active catalysts, and this is an area of growing interest.
It is important to collect cyclic voltammograms (CV) instead of applying a linear sweep voltammogram (LSV) program since the former may reveal the hysteresis due to secondary reactions or reconstruction of the surface. Sometimes it is recommended to use steady-state chronoamperometry to measure activity at different overpotentials, as the influence of oxygen intercalation may overestimate the activity.31 Furthermore, performing steady-state measurements allows minimization of the contribution of the capacitive current. Correcting nonfaradaic current should be avoided when side reactions are present, as this correction is only to eliminate the Cdl. With the CV, it is possible to extract the Tafel slope to understand, a priori, the reaction mechanism. Additionally, it is advisible to extract the slope in at least two decades,134 except in cases where there is a transition between one rate determination step and another.135
Once the initial characterization is completed using EIS (step 3) and CV (step 4) at a low scan rate or at steady-state chronoamperometry, step 5 involves conducting stability measurements of the catalyst using CV on the same electrode and examining its degradation during the cycles. For this purpose, a given number of CV cycles should be carried out at a faster scan rate, such as 100 cycles at 100 mV s–1. Following that, steps 3 and 4 are repeated. Conducting step 3 after step 5 allows one to assess the variations of Rct and Cdl, which can indicate structural modifications throughout the cycles. The repetition of step 4 after certain cycles permits evaluation of the evolution of catalytic activity with the OER cycles and identification of possible changes in Tafel slopes. This whole sequence (steps 3–5) should be repeated until several thousands of OER cycles are recorded or until the OER activity declines.
Finally, in step 6, for a better assessment of the durability of a catalyst, after depositing the catalyst again onto the RDE, it is preferable to continue studying its activity degradation using chronopotentiometry at a current of 10 mA cm–2 or chronoamperometry at an equivalent potential to the initial CV of 10 mA cm–2.
It is worth noting that conducting measurements using an RDE provides a quick and relatively effective method for screening a material’s activity and durability. However, measurements taken using an RDE have certain limitations. First, the currents that can be measured are much lower than those in PEMWE. Second, catalysts are much more durable in PEMWE than in RDE due to various factors such as the accumulation of bubbles on the surface of the RDE.136 Although similar formations occur in MEA, it is to a lesser extent.137 As mentioned before, another factor to consider is the effect of counterions, which may affect the catalyst to a varying extent. Lastly, the effect of pH or high potentials on the glassy carbon or the carbon where the catalyst is dispersed can create a passivating/corrode layer, further affecting the measurement.138 Therefore, while measurements taken using RDE are effective for determining a catalyst’s activity, it is highly recommended to measure catalysts using PEMWE for a more accurate representation of their activity and durability in real industrial operating conditions101,139 or in a gas diffusion electrode (GDE) setup as presented recently in reference.140
In summary, in this Perspective, we have shown that Ir-mixed oxides with a wide range of compositions and structures can be synthesized. Regardless of their initial composition and structure, the most active mixed oxides display a potential of around, or slightly lower than, 1.5 V at 10 mA cm–2, in RDE. This value is similar to that reported for state-of-the-art IrOx catalysts. When normalized to the Ir content, however, a number of Ir-mixed oxides display higher mass-normalized OER activities than IrO2 (see Table S1 and Figure S2), making them promising candidates to replace IrOx as electrocatalysts in PEMWEs. Even if an unequivocal conclusion cannot be reached and more iridates need to be studied, the results above distillate a number of structural features relevant for the OER activity of Ir-mixed oxides. Thus, iridates with high OER activity tend to display short Ir–Ir distances, high Ir oxidation state, distorted IrO6 octahedra and a strengthened hybridization between Ir 5d and O 2p orbitals that enhances conductivity. However, establishing proper structure–activity descriptors from the as synthesized Ir-mixed oxides may be misleading. In order to identify the true nature of the active phases, it should be considered that Ir-mixed oxides will undergo a reconstruction process during the EOR triggered by leaching of the non-noble elements. It is therefore imperative to study and understand this reconstruction and identify the nature of the species (surface and bulk) in the reconstructed phase. It is also desirable to unify a protocol for measuring and reporting the OER activity of novel catalysts, including RDE and PEMWE measurements. This will allow construction of accurate and accessible libraries with accurate activity data that can be used to identify accurate structure–activity descriptors that will allow for a rational design of Ir-based catalyst for the OER.
Financial support from projects PID2019-103967RJ-I00 and PID2020-116712RB-C21 funded by MCIN/AEI/10.13039/501100011033 is acknowledged. We acknowledge the funding granted to the PROMET-H2 project by the European Union’s Horizon 2020 research and innovation programme under Grant Agreement No. 862253 and the Deputyship for Research & Innovation, Ministry of Education of Saudi Arabia, for the project number 341. We also thank the Consejería de Educación, Juventud y Deporte of the Comunidad de Madrid for the Ayuda Destinada a la Atracción de Talento Investigador (2020-T2/AMB-19927) granted to Á.T.-M.
The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacsau.3c00247.
Tables and figures collating the OER activities of Ir-based mixed oxides in RDE and PEMEW; evolution of the double-layer capacitance during OER cycling ( PDF
† D.G. and Á.T-M. contributed equally to this work. The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. CRediT: Dmitry Galyamin writing-original draft, writing-review & editing; Alvaro Tolosana-Moranchel writing-original draft, writing-review & editing; Maria Retuerto supervision, writing-original draft, writing-review & editing; Sergio Rojas supervision, writing-original draft, writing-review & editing.
The authors declare no competing financial interest.
Product Support and Customer Service
For Further support visit our Contact Page
General
When designing a Membrane Electrode Cell, it is important to select the proper Electrode material. Generally, materials with the following characteristics can be considered for use: readily available; insoluble in a dilute solution of common organic acids and amines; and no components that can foul the ED paint. In cathodic ED paints, another desirable quality is the ability to resist the oxidation that results from hydrolysis of water at the anode’s surface under an elevated voltage. A lifetime of two years or more is desirable. In anodic ED paints, the cathode does not suffer from oxidation and so has a very long life. While the cathode could be made from black iron typically it is made from 304 and 316 stainless steel.
For cathodic ED systems (where the Electrode is an anode), 316L grade stainless steel is the material of choice – any lesser grade will dissolve in a matter of months and create a tremendous amount of iron contamination & iron sludge. 316L stainless is appropriate for most epoxy-based ED paints. Its dissolution by-products form an excellent conductive layer around the Electrode and serve to lessen the overall oxidation process. However, 316L stainless steel is sacrificial and typically needs to be replaced every 2 to 5 years. For the smaller diameter TECTRON Cells Sch 40 wall thickness is common and for the larger diameter TECTRON Cells Sch 10 is used..
For acrylic-based cathodic ED paints, precious metal oxide-coated titanium anodes are an excellent choice. This material does not contain iron and cannot contaminate the ED paint bath with solubilized iron. The Precious Metal Electrode has a hybrid composition. Its substrate is a titanium thin wall tube. The function of the tube is to serve as a form factor for the precious metal oxide coating. The precious metal oxide coating is applied over the titanium tube and then is oven-cured. Whereas precious metal is less soluble than 316L stainless steel, it too can suffer from corrosion & pitting if certain conditions are present in the e-coat tank.
There are two commonly used precious metals: ruthenium & iridium. Ruthenium is about 20% less and is selected by price sensitive clients. Iridium is a much better material and should be used in high painted through put E-coat tanks like appliance and automotive where more durability is required. The chart on the next page provides additional information concerning the choice of Electrode material.
Note: Since anodes are sacrificial and can suffer fast corrosion they are not warranted and no performance guarantees can be offered.
Stainless Alloy Precious Metal Oxides Type 316L Ruthenium or Iriduim Cost Factor 1 > 5-8 times Availability Ok Adequate Fabrication Ok Adequate Substrate Materials None Ti,Ta,Zr,Nb Physical Properties Strong Brittle Quick Failure Yes Yes Dissolution Products Fe -> Fe+2 & Fe +3 TiO2 -> Ti +4 & O2 Dissolution Voltage < 2 > 8-12 Design Basis, Amps/SM (Amps/SF) 55 (5) 55 (5) Measure Wear? Readily No Resistance to X- Factor Poor Good Usage No Corrosion Factor Epoxy ED Paints Low Corrosion Factor Light Colors / Acrylic Only ED Cathodic Paints Application Industrial price sensitive markets High Thru-put systems requiring more durability such as Appliance & Automotive Typical Wear Rates (No Corrosion Factor) Generally < 10-50 mgrams/C Hard to predict pr measure without x-ray equipment Detect Wear Thru Visual Inspection? Yes Limited Life Expectancy (hours) ~20k > 20kBULLETIN 991104
If you are looking for more details, kindly visit iridium oxide coated titanium anodes.