Introducing foreign metals into Ir-based oxides is a versatile strategy for acidic OER. It can tune the adsorption energies of *OH, *O, and *OOH intermediates and at the same time improve resistance against oxidative corrosion [3]. Dopants enter either by lattice substitution or by interstitial incorporation. Charge is then compensated through Ir valence change, oxygen vacancy formation, or protonation of lattice oxygen. In practice, commonly used dopants include niobium (Nb) [57,58], tungsten (W) [59,60], molybdenum (Mo) [61,62], nickel (Ni) [63,64], cobalt (Co) [63], and iron (Fe) [2,65]. Depending on how they are introduced, these elements can influence Ir in several ways [66]. The choice of dopant is guided by descriptors such as dband center, metal oxygen covalency, and electronegativity as well as by the formal charge and ionic radius, which together determine how Ir-O bonds reorganize under bias [67–69].

Li et al. reported that introducing a small amount of Co into strontium iridate (SrIrO₃) markedly influences surface reconstruction and OER performance (Fig. 2a) [70]. In a PEMWE, the performance data in Fig. 2b show that Sr₇Ir₆CoO_x exhibits much higher catalytic activity than SrIrO_x and IrO₂. Operando Raman and X-ray Absorption Spectroscopy (XAS) analyses revealed that Co plays a dual role. It activates surface lattice oxygen, which accelerates the exposure of IrO_x active sites, and it also adjusts the adsorption energy of OOH species at the surface layer (Fig. 2c). As a result, the Co modified catalyst outperformed commercial IrO₂ in membrane electrode assembly (MEA) tests, reflecting more efficient mass and charge transport. Building on this approach, double perovskites such as Sr₂MIrO₆ were further employed to examine the impact of alkaline cations. Variants containing calcium (Ca), magnesium (Mg), or zinc (Zn) were investigated, with Sr₂CaIrO₆ showing rapid reconstruction in acidic electrolyte during the initial electrochemical cycles. This transformation produced a stable surface composed of short-range edge sharing IrO₆ octahedra with an open framework, which was linked to both high activity and long term durability. In practical tests, Sr₂CaIrO₆ delivered 6 A cm^-2 at 2.4 V and sustained 2.0 A cm^-2 for 400 h.

Jin and co-workers reported torsion strained, grain boundary rich tantalum (Ta) and thulium (Tm) codoped IrO_2-δ prepared by a rapid pyrolysis route (Fig. 2d) [71]. Metal precursors of Ir, Ta, and Tm were coprecipitated with ammonia and citric acid and then introduced into a pre-heated 450^oC furnace within seconds, which produced massive supersaturation, burst nucleation, and mesocrystal growth through particle collision and coalescence. This process created dense twins, stacking faults, and trigeminal grain boundaries (GBs) that store torsional shear strain as shown in transmission electron microscopy (TEM) images (Fig. 2e). Ta and Tm co-doping works with the torsion strain at GBs. The strain lengthens Ir-O bonds and introduces shear, while the dopants shift the O 2p and Ir 5d-band centers toward the Fermi level. Together these effects reduce the difference between ∆G_O and ∆G_OH and place oxygen intermediate binding near the optimum. Spectroscopic analyses, including Ir 4f X-ray photoelectron spectroscopy (XPS), L-edge X-ray absorption near edge structure (XANES), and extended X-ray absorption fine structure (EXAFS), revealed that co-doping lowers the average Ir valence. They also show that Ir-O bonds are elongated relative to undoped controls, indicating that the dopants tune Ir-O covalency. Density of states calculations further showed that Ta and Tm shift the p and d-band centers toward the Fermi level, which reduces the ∆G_O − ∆G_OH descriptor and facilitates the potential determining step. The mass activity was evaluated at an overpotential of 266 mV in 0.5 M H₂SO₄ (Fig. 2f). Both GB engineering and Ta/Tm co-doping improved the activity of IrO₂ compared to commercial IrO₂, and their combination produced the highest mass activity, highlighting the strong synergy between strain and doping. Electrochemically, GB rich Ta_0.1Tm_0.1Ir_0.8O_2-δ achieved a low overpotential of 198 mV at 10 mA cm^-2 in 0.5 M H₂SO₄ and sustained 1.5 A cm^-2 for 500 h at only 0.2 mg_Ir cm^-2 in PEMWE with high Faradaic efficiency (Fig. 2g). Inductively coupled plasma–mass spectrometry (ICP-MS) analysis further showed that Tm leaches more readily than Ir or Ta under long term operation (Fig. 2h). Dopant leaching is critical because dopants are initially introduced to tune the electronic structure of catalyst and reaction activity. Once they dissolve, the reaction activity may change, causing the catalyst to deviate from its intended design and reported behavior.

Doping layered oxides is challenging because foreign cations can distort or even collapse the lamellae. Lu and co-workers introduced acid resistant Nb into two-dimensional layered edge-shared IrO₂ (1T IrO₂) while preserving the stacking [57]. They employed an alkali-assisted mechano-thermal route, in which the alkali flux generates a transient melt that promotes solid-state diffusion of Nb. This process produced lamellar Nb_0.05Ir_0.95O₂ with uniform Nb distribution. Spectroscopic analysis with Ir 4f XPS and Ir L₃-edge XANES revealed charge redistribution between Nb and Ir. Despite this redistribution, the Ir-O coordination remained 1T-like. The partial density of states further showed that the Ir d-band center shifted toward the Fermi level, which indicates tuned Ir-O covalency and optimized intermediate binding. Density functional theory (DFT) confirmed that Nb functions as modulation doping. Subsurface Nb perturbs the electronic structure of surface Ir sites, leading to new scaling relations. As a result, the potential determining step shifts from *OOH deprotonation on pristine 1T IrO₂ to *O-OH coupling on Nb doped surfaces, and the free energy for Ir dissolution increases from 3.67 to 3.86 eV. Consistent with these effects, Nb_0.05Ir_0.95O₂ lowers overpotential at 10 mA cm^-2 (η₁₀) by 56 mV compared to 1T IrO₂ and achieves a stability number (S-number) of about 5.9 × 10⁶ . It also operates stably in single cell at about 270 mA cm^-2 under 1.4 V for 1100 h with negligible decay. These results demonstrate that the size and valence compatibility of Nb, together with its acid stability, enables structure preserving doping that optimizes intermediate energetics and suppresses Ir loss.

Wang et al. and co-workers reported that doping can regulate the OER mechanism [38]. In Sr-doped IrO_x with preferentially exposed (200) facets, Sr decreases Ir-O covalency and perturbs the connectivity of IrO₆ octahedra. This modification generates adjacent Ir sites with favorable separation and orientation, enabling direct O*-O* coupling from neighboring M-O* moieties. Operando Fourier transform infrared spectroscopy (FTIR) and Raman spectroscopy reveal vibrational features associated with O*-O* coupling within the OER potential window. In addition, isotope-labeled DEMS with H₂ ¹⁸O detects ³⁶O₂ and ³⁴O₂ together with ³²O₂. These findings are consistent with adjacent site coupling rather than a conventional AEM or LOM. Sr-doping induces elongation of the Ir–O bond, with EXAFS confirming an average bond length of ~2.03 Å compared with 1.98 Å in rutile IrO₂. This effect originates from the large ionic radius and lower valence of Sr²⁺, which laterally displace neighboring oxygen atoms and weaken orbital overlap between Ir d-state and O p-state. Consistently, XANES places the average Ir valence at ~3.75. The fingerprints of weakened covalency are also evident in projected crystal orbital Hamilton population (pCOHP) analysis and in a downward shift of the Ir d-band center relative to the Fermi level. On the (200) facet, these electronic and structural relaxations suppress bridging oxygen connectivity and expose adjacent Ir site pairs with a spacing that falls within the range favorable for OPM. DFT calculations on Sr-IrO_x(200) confirm that O*-O* bond formation is the RDS with a barrier of ~1.35 eV, substantially lower than the barriers for the AEM on undoped IrO₂. When the active Ir pairs are positioned farther from Sr, the O*-O* barrier increases to ~1.71 eV, underscoring the spatial requirement for adjacent site coupling. In device tests, a PEMWE cell with Sr-IrO_x as the anode operates at 1.72 V and 1 A cm^-2 for more than 500 h with stable performance, demonstrating that Sr-driven electronic and structural modulation can enhance intrinsic activity while maintaining durability.

Rational control of catalyst morphology is a direct way to improve Ir utilization and to unlock high current density PEMWE. Engineering 1D, 2D, and 3D nanostructures increases accessible surface sites and electrochemically active surface area (ECSA) [72]. It also shortens ion and electron pathways and improves contact with the ionomer in the anode layer [73]. Reported morphologies, ranging from nanoparticles [74–76] to nanowires [59,77], nanosheets [78] and even single atom motifs [73], illustrate how structural engineering can raise intrinsic activity at fixed Ir loading and enable better percolated catalyst layers. At device-relevant currents, transport through the catalyst–ionomer architecture becomes rate limiting. A dense ionomer skin at the catalyst surface impedes O₂ release and water access. In addition, an unfavorable pore spectrum either traps gas in micropores or penalizes proton conduction through excess macropores [79]. A mesoporous structure or a meso-macro gradient therefore provides the most balanced pathway for simultaneous gas and ion transport [80,81].

A chemical dealloying strategy has been applied to fabricate porous Ir metallic aerogels (MAs) (Fig. 3a) [82]. Liang et al. utilized an IrCu3 alloy prepared through a wet-chemical route as the precursor, followed by extensive acid leaching in 0.1 M HClO₄ at 85^oC. In this process, the Cu component acts as a sacrificial template, driving the reduction of Ir via galvanic replacement. Substituting Cu with other transition metals such as Co or Ni leads to much lower yields of Ir nanoparticles. The resulting Ir MA displays a three-dimensional interconnected framework with hierarchical porosity, where most pores are around 1–3 nm and additional microchannels fall within the 3–8 nm range (Fig. 3b). This architecture increases the ECSA by nearly 70% relative to Ir black (Fig. 3c). Consequently, the catalyst delivers substantially higher OER activity and device performance, reaching 2.0 A cm^-2 at 1.79 V with an Ir loading of only 0.5 mg_Ir cm^-2, surpassing the Ir black benchmark that required four times more Ir loading to approach similar current densities. The MEA assembled with this aerogel catalyst also exhibited excellent stability, with a degradation rate of about 40 μV h^-1. This performance was maintained during long term testing at 0.5 and 2.0 A cm^-2 for 1000 and 400 h, respectively (Fig. 3d). The fine porous structure also facilitates the conversion of metallic Ir to IrO_x during operation, which suppresses Ir³⁺ dissolution and improves durability. In related studies, dealloyed Ir-Ni or Ir-Co nanowires were also reported as highly efficient OER catalysts. They showed activities several times higher than conventional Ir nanoparticles and underscored the potential of dealloying approaches to reduce Ir usage without sacrificing performance.

Zhang et al also reported a three-dimensional mesoporous IrO₂ skeleton created by dynamically loading Ir into a La₂O₃ sacrificial template that reconstructs to LaOCl under Cl^- assistance, followed by in-situ oxidation to IrO₂.23 The sequence involves Ir/Cl co-adsorption on La₂O₃, LaOCl formation, rapid LaOCl hydrolysis, and dense Ir nucleation within the evolving template. Together, these steps drive uniform Ir incorporation throughout the bulk and surface. Operando/ex-situ spectroscopy captures this reconstruction: the Cl signal evolves into an ν(Ir–Cl) vibration while La₂O₃ features shift consistently with LaOCl formation. X-ray diffraction (XRD)/XPS follow the transient LaOCl lattice and its collapse during growth. After template removal and oxidation, the product is a robust IrO₂ framework with abundant ~5 nm mesopores that lower charge transfer and mass transfer resistances versus a commercial Ir benchmark. Consistent with these transport gains, the mesoporous IrO₂ shows a lower cell voltage across the polarization curve. It also sustains 5 A cm^-2 at ~2.0 V for 2,700 h at 80^oC with a minute decay rate, demonstrating that properly tuned porosity can decisively mitigate bottlenecks under industrial loads. While compelling, the optimum pore spectrum and the mesoscale gradient across the anode remain open questions.

Furthermore, controlling the exposed crystal facets can also tune the OER pathway and energetics on IrO₂ [25]. Liu and co-workers reported that DFT reveals distinct rate determining barriers on rutile facets under the associative AEM, with ~0.72 eV on (110), ~0.64 eV on (200), and ~0.53 eV on (101), identifying (101) as the most active surface (Fig. 3e). To isolate this facet in a practical catalyst, an ammonia-assisted facet engineering route was developed. In this process, NH₃ adsorbs most strongly on (101), passivating its out-of-plane growth and biasing grains to expand laterally while maintaining monolayer thickness. Calculations confirm the facet dependent adsorption strength of ammonia, with absolute adsorption energy on (101) greater than on (200) and (110) (Fig. 3f), thereby underpinning preferential (101) exposure. The resulting two-dimensional IrO₂ monolayer displays clear (101) lattice fringes in high resolution-TEM (HR-TEM) and fast Fourier transform (FFT) images (Fig. 3g), translating the facet advantage into catalytic performance. Unlike commercial rutile IrO₂ nanoparticles that expose multiple facets, this single-facet monolayer exposes only the most active (101) surface. Also, it exhibits lower overpotential in half-cell tests and, more importantly, sustains over 10,000 h at 1.5 A cm^-2 and over 2,000 h at 2 A cm^-2 even at an anode loading of only 0.2 mg_Ir cm^-2 (Fig. 3h). The facet trend persists under alternative OER pathways, with (101) consistently showing the lowest d-band center among the common facets, consistent with weaker *O and *OOH binding and accelerated turnover.

In Ir-based PEMWE anodes, supports are not passive carriers but key determinants of dispersion, transport and stability [81,83]. An effective support provides a high surface area that anchors Ir and suppresses agglomeration. It also maintains continuous pathways for electrons and protons, either through its own conductivity or by enabling percolating IrO_x networks. In addition, it moderates Ir-O bonding to suppress over-oxidation and dissolution [84]. Acid stable oxides such as TiO₂ [85] and tin oxide (SnO₂) [86], including antimony-doped SnO₂ (ATO) [86], have therefore been widely investigated. They combine chemical robustness with tunable morphology. In addition, they can participate in purposeful metal-support interactions [87,88].

TiO₂ is a representative support material valued for its intrinsic stability and good processability [89–91]. However, its poor miscibility with Ir has traditionally limited its use. Recent advance by Zhang et al. shows that substitutional incorporation of spatially correlated Ir atoms into the TiO₂ surface lattice creates flexible Ir-O-Ir coordination [92]. This structure optimizes intermediate adsorption by shifting the d-band center, enhances intrinsic activity, and suppresses over-oxidation. The result is a low driving voltage of 1.63 V at 2 A cm⁻² and maintains stable performance for over 440 h. Zou et al. introduced bulk-engineered TiO_x@Ti core-shell supports as an alternative strategy (Fig. 4a) [93]. These drive Ir nanoparticles to fully transform into crystalline rutile IrO₂ during operation. In doing so, they prevent the formation of hydrous amorphous phases and resolve the traditional activity-stability trade-off. Jung et al. advanced ATO as a support by introducing excess electron reservoirs (EERs) in the form of charged oxygen species (Fig. 4b) [94]. These species donate electrons to IrO_x, stabilize Ir³⁺ states, and prevent over-oxidation. This design achieves 75 times higher mass activity than commercial Ir catalysts and sustains over 250 h of stable operation at 1 A cm⁻². The Ir-specific power exceeds 70 kW g⁻¹, underscoring its promise for large-scale hydrogen production.

CeO_x support that is not conventional such as TiO₂ and SnO₂ has been used to reinforce the anchoring of Ir-based catalysts. Zhang and co-workers took advantage of the high oxygen storage capacity and abundant oxygen vacancies in CeO_x, which act as electron donors, to design a ripening-induced embedding (RIE) strategy (Fig. 4c).18 In this approach, IrO_x nanoparticles are not only deposited on the surface but also partially embedded into the growing CeO_x matrix. The synthesis begins with an ultrasound-assisted polyol method that produces sub-10 nm CeO_x particles with rough, high-energy facets. During sonication, acoustic cavitation synchronizes CeO_x ripening with Ir nucleation, allowing the support to envelop Ir nanoparticles to a depth of about 1 nm. Cryo-electron tomography and kinetic Monte Carlo simulations confirm this embedded distribution. Vacancy rich CeO_x donates electrons to IrO_x and stabilizes the interface while maintaining sufficient conductivity. When tested in PEMWE, this structure achieved 1.72 V at 3 A cm⁻² with only 0.3 mg_Ir cm⁻² loading (Fig. 4d). It also sustained operation from 1 to 8 A cm⁻² with a very low voltage decay of ~1.33 μV h⁻¹ over 6000 h. This embedding strategy provides a general method to construct robust catalyst-support systems

Ir⁶⁺-based oxides (IrO₃) are theoretically predicted to offer both high activity and durability (Fig. 4e) in acidic OER but are difficult to stabilize [31,43,95,96]. Recently, Nakamura et al. demonstrated a device relevant strategy using a sacrificial/oxidizing manganese oxide (MnO₂) scaffold that both anchors and oxidizes Ir.⁸³ In this approach, MnO₂ is first electrodeposited on a porous Ti current collector at 95^oC in 0.3 M MnSO₄/0.36 M H₂SO₄, then exposed to K₂IrCl₆ in 0.01 M H₂SO₄ at the same temperature. During this step, ultraviolet-visible (UV-vis) spectrometry shows that oxidative ligand substitution drives quantitative Ir uptake, as evidenced by the disappearance of the 488 nm K₂IrCl₆ band in-situ. Meanwhile, Mn in the scaffold is partially reduced, serving as the oxidant for Ir. Subsequent annealing at 450^oC completes the chlorine to oxygen ligand exchange and yields atomically dispersed Ir⁶⁺ oxide (Ir^VI -ado) embedded within the MnO₂ lattice. Spectroscopic analyses confirm the structure and valence. Ir L₃-edge white-line intensity corresponds to an oxidation state of +5.8 ± 0.1(Fig. 4f), EXAFS detects a shortened Ir-O distance (~1.96 Å) together with distinct Ir-O-Mn coordination. Ir 4f binding energies are also consistent with Ir⁶⁺ references. Operando XAS during PEMWE shows negligible L₃- edge shifts from 1.5 to 2.5 V, indicating that the Ir⁶⁺ state is preserved under load. In single cell tests, the cell delivers 4.0 A cm^-2 at 2.0 V with only 0.08 mg_Ir cm^-2 and sustains 2,700 h of operation near 1.8 A cm^-2 with minimal decay. (Fig. 4g). Together, these results demonstrate that a sacrificial MnO₂ support can act as both oxidant and host, producing strongly anchored, high-valence Ir sites that enable ultralow Ir usage alongside industrial level current densities.

Despite decades of research, the mechanistic picture remains incomplete. AEM predicts a theoretical overpotential of about 0.37 V, yet many reports observe 200–300 mV and still assign activity to the same pathway. This discrepancy may stem from limitations of DFT based scaling relations that neglect interfacial electric fields, explicit solvation, and the formation of hydrous surface layers in acidic media. In addition, it is difficult to clearly link dopant induced changes in catalyst structure to a specific mechanistic pathway. Resolving these issues requires coordinated efforts that combine advanced theory with operando measurements to distinguish between AEM, LOM, OPM, and other proposed mechanisms.

The half cell typically employs acidic electrolytes such as 0.5 M H₂SO₄, which provide high proton conductivity, whereas PEMWE single cell operates with deionized water, where proton transport depends on the membrane and ionomer. This difference results in lower proton accessibility, higher ohmic resistance, and distinct local reaction environments. Moreover, single cell configuration is mechanically compressed by end plates and PTLs, which can deform the catalyst layer, alter porosity, and influence water management. The contact configurations also differ that glassy carbon (GC) substrates are used in half cell measurements, while catalyst coated membrane (CCM) or catalyst coated substrate (CCS) are employed in single cells, leading to variations in electronic and ionic transport pathways as well as overall ohmic behavior.

Most PEMWE studies remain focused on catalysts at the laboratory scale, while investigations on how materials behave in actual stacks are limited. Real progress will depend on integrating catalysts with other essential components, including membranes, ionomers, porous transport layers, current collectors, and flow fields. It will also require attention to system factors such as gas and water management, compression, and thermal regulation. At very low Ir loading, both electronic and protonic percolation can be lost, which lowers device performance even when the intrinsic site activity is high. A useful framework is to separate kinetic, ohmic, and mass transport resistances, and to co-design active site chemistry, conductivity pathways, and pore structures with stack relevant operation in mind.

The community aims for lifetimes approaching 80,000 h, but such tests are not feasible. Standardized accelerated protocols that reflect stack conditions are therefore critical. They should include potential holds, current cycling, start-stop scenarios, and stress from temperature, humidity, and pressure. Degradation must be assessed not only by voltage decay but also by metal dissolution and stability metrics such as S-number and Ir utilization. Validation against short stack tests is essential to ensure predictive value for real devices.

Mechanistic insights need to be translated into structural and electronic descriptors that guide data driven synthesis and single cell level validation. Reporting should be standardized to allow direct comparison, with Ir specific metrics such as power density and durability clearly stated. In parallel, techno economic and life cycle analyses must be incorporated to ensure that advances align with cost and sustainability goals. With coordinated progress in mechanisms, materials, electrode and stack engineering, and measurement protocols, Ir-based PEMWE catalysts can evolve from laboratory concepts to practical technologies for large scale hydrogen production.