Modulating hydroxyl adsorption on Pd–Rh heterostructures through interfacial electron redistribution: a pathway to high-efficiency alkaline HOR catalysis

Abstract

The practical implementation of anion exchange membrane fuel cells (AEMFCs) using cost-effective reformate hydrogen is severely hindered by the trade-off between the catalytic activity and CO tolerance of electrocatalysts in alkaline hydrogen oxidation reaction (HOR). Here, we report a rational design of Pd–Rh bimetallic interfaces with tailored electronic gradients to tackle this dilemma. The construction of Pd–Rh heterostructures enables the optimal PdRh_0.05/C catalyst to exhibit an exceptional balance between HOR activity and CO resistance. At an overpotential of 50 mV, PdRh_0.05/C shows a 30.7 times enhancement in specific activity and a 35 times enhancement in mass activity, compared to Pd/C. PdRh_0.05/C also exhibits exceptional endurance with only 13% current decay after 10 000 s of operation, compared to >40% degradation recorded for Pd/C. Furthermore, PdRh_0.05/C delivers improved CO tolerance and can preserve 83% of its performance under 1000 ppm CO/H₂ after 1500 s, while Pd/C loses 78% of its performance. DFT studies demonstrate that the Pd–Rh interface promotes valence electron redistribution, greatly improving Rh–O orbital hybridization, reducing the OH* adsorption barrier by 326%, and thus accelerating the rate-determining Volmer step and increasing overall HOR performance. This study presents an exceptional Pd–Rh bimetallic electrocatalyst exhibiting both elevated hydrogen oxidation reaction activity and carbon monoxide tolerance, while also introducing a comprehensive technique for regulating electronic structures in high-efficiency electrocatalysts.

---

Bang-An Lu

Bang-An Lu obtained his PhD degree from Xiamen University, China, in 2019 under the supervision of Prof. Dr Shi-Gang Sun. After serving as an Assistant Research Scientist at Xiamen University (2019–2020), he held the position of Associate Professor at Zhengzhou University (2020–2023). Currently, Dr Lu is a Professor in the College of Materials Science and Engineering at Zhengzhou University. His research focuses on the design and synthesis of highly efficient electrocatalysts for fuel cells and water splitting, the kinetics of electrode/electrolyte interface processes, and electrochemical in situ spectroscopic characterization techniques. Dr Lu has authored over 50 publications in leading international journals, including Angew. Chem. Int. Ed., Joule, ACS Catal., ACS Nano, and J. Catal.

---

1. Introduction

Anion exchange membrane fuel cells (AEMFCs) have emerged as a cost-effective alternative to proton exchange membrane fuel cells (PEMFCs), leveraging their compatibility with non-precious metal catalysts in alkaline media.^1–6 However, their widespread adoption is hindered by two major challenges: (1) sluggish hydrogen oxidation reaction (HOR) kinetics in alkaline environments—2–3 orders of magnitude slower than in acidic media^7–9—and (2) severe anode catalyst poisoning by trace CO (≤10 ppm), which precludes the use of reforming gas (the most economically viable hydrogen source).^10,11 Reforming-derived hydrogen dominates the current market due to its low cost of production (∼US$1.5 per kg, one-third that of electrolytic hydrogen at ∼US$4.2 per kg).¹² Therefore, enhancing CO tolerance could thus significantly reduce system costs by enabling the direct utilization of reforming-derived hydrogen without further purification steps.¹³ To unlock this potential, the development of cost-efficient HOR catalysts with high activity, durability, and great CO tolerance in alkaline remains an urgent priority for AEMFC commercialization.^14,15

Palladium (Pd) emerges as a compelling alternative to Pt in terms of both greater global production (210 vs. 160 tons) and substantially lower cost (∼1195 vs. ∼1454.5 USD/Oz).^16,17 However, its alkaline HOR activity remains considerably lower than that of Pt due to unfavorable adsorption energetics.¹⁸ The excessively strong hydrogen binding energy (HBE) and the weak hydroxide adsorption jointly lead to a kinetic bottleneck in the Volmer step (H* + OH⁻ → H₂O + e⁻).^16,19,20 More critically, CO adsorption is notably stronger on Pd than on Pt, as demonstrated by the lower CO stretching frequency on the Pd surface (∼1950–2050 cm⁻¹versus 2050–2100 cm⁻¹ for Pt).^21,22 This reflects enhanced back-donation from Pd to the CO antibonding orbitals, which presents a fundamental trade-off: while efforts to enhance Pd's HOR activity often focus on modifying its electronic structure, such modifications typically intensify CO poisoning—a critical limitation for practical implementation. These intrinsic challenges underscore the urgent need for innovative catalyst design approaches that can simultaneously optimize Pd's electronic structure, adsorption energetics, and CO tolerance to realize its full potential as a Pt-alternative catalyst.

An ideal catalyst for practical HOR applications must achieve a delicate balance in the adsorption strengths of *H, *OH, and CO—a formidable challenge in electrocatalyst design.^23,24 As proposed by Markovic et al. via the well-known bifunctional mechanism, alkaline HOR kinetics are primarily limited by the sluggish Volmer step, where reactive hydroxy species (*OH) play a critical role in facilitating proton dissociation²⁵. Introducing oxophilic surface sites can effectively stabilize *OH intermediates, thereby enhancing HOR activity in alkaline media.²⁶ Interestingly, a promising strategy to mitigate CO poisoning involves incorporating heteroatoms that either weaken CO binding strength or promote *CO oxidation to CO₂.²⁷ A well-established approach is the bifunctional CO oxidation mechanism, where alloying with oxophilic metals (e.g., Ru and Mo) accelerates CO removal.^28–30 In such systems, oxygen-containing species readily form on the oxophilic sites, facilitating oxidative stripping of adsorbed CO from adjacent Pt active centers.^31–33 These insights suggest that constructing heterostructure catalysts with carefully integrated oxophilic components can simultaneously optimize both HOR activity and CO tolerance—offering a viable pathway to develop high-performance, poisoning-resistant Pd-based electrocatalysts.

In this work, we engineered Pd–Rh heterostructures via chemical vapor deposition (CVD) to construct well-defined interfaces that synergistically enhance HOR kinetics and CO tolerance. By tailoring the Pd surface with rhodium (Rh)—an element with strong OH affinity—we simultaneously optimized the adsorption energies of both H* and OH* intermediates through interfacial electronic modulation. At an overpotential of 50 mV, PdRh_0.05/C gives a 30.7 times enhancement in specific activity (SA) and a 35 times enhancement in mass activity (MA), compared to Pd/C. Remarkably, it maintains robust durability (only 13% current density loss after 10 000 s) and outstanding CO tolerance, exhibiting merely a 17% anode current loss in H₂/1000 ppm CO after 1500 s—a critical metric for practical fuel cell applications. Density functional theory (DFT) calculations unveil that the Pd–Rh interface facilitates valence electron redistribution, significantly enhancing Rh–O orbital hybridization (ICOHP: −2.54 eV vs. −0.06 eV for Pd–O). This electronic synergy reduces the OH* adsorption barrier by 326% (from 0.77 to 0.18 eV), thereby accelerating the rate-determining Volmer step and boosting overall HOR performance.

2. Materials and methods

2.1 Chemicals

Palladium acetylacetone (Pd(acac)₂, AR, Sinopharm), rhodium acetylacetone (Rh(acac)₃, AR, Sinopharm), Nafion solution (D520, 5 wt%, Alfa-Aesar), isopropyl alcohol (IPA, AR, Macklin) and potassium hydroxide (GR, Sinopharm) were used as received. All solutions were prepared with ultrapure water (18.2 MΩ cm). The 20 wt% Pt/C catalyst from Johnson Matthey (JM Pt/C) was used as received.

2.2 Synthesis of catalysts

2.2.1 Synthesis of Pd/C. 5.8 mg Pd(acac)₂ and 20 mg conductive carbon black were dissolved in 50 mL ethanol separately and ultrasonicated for 15 min to make the solutions homogeneous, and then the two solutions were mixed by ultrasonication for 1 hour. Subsequently, the mixture was evaporated using a 70 °C water bath and dried overnight at 60 °C under vacuum. The dried powder was ground, placed in a crucible, heated from room temperature to 500 °C at a rate of 5 °C min⁻¹ under a H₂/Ar atmosphere, and held for 3 hours, and finally Pd/C was obtained with uniformly dispersed Pd metal particles on conductive carbon black.

2.2.2 Synthesis of PdRh_0.05/C. 3.9 mg Rh(acac)₃ was placed at the bottom of a quartz crucible, a quartz fiber membrane was placed in the middle, and 20 mg Pd/C was placed on the membrane. The system was heated from room temperature to 240 °C at a rate of 5 °C min⁻¹ under an Ar atmosphere, held for 1 hour to allow Rh(acac)₃ to decompose into gas and uniformly distribute on Pd/C, and then further heated to 320 °C and held for 1 hour to promote the growth of the heterojunction structure of metallic Rh on Pd/C. The entire process was performed at a heating rate of 5 °C min⁻¹, followed by cooling to room temperature to obtain PdRh_0.05/C. Catalysts with different Rh contents were synthesized using the same method as PdRh_0.05/C, but adjusting the amount of Rh(acac)₃ to 1.6 mg and 7.8 mg, respectively, to obtain PdRh_0.02/C and PdRh_0.1/C.

2.2.3 Synthesis of Rh/C. 7.8 mg Rh(acac)₃ and 20 mg conductive carbon black were dissolved in 50 mL ethanol separately and ultrasonicated for 15 min to make the solutions homogeneous, and then the two solutions were mixed by ultrasonication for 1 hour. Subsequently, the mixture was evaporated using a 70 °C water bath and dried overnight at 60 °C under vacuum. The dried powder was ground, placed in a crucible, heated from room temperature to 500 °C at a rate of 5 °C min⁻¹ under an Ar atmosphere, and held for 3 hours, and finally Rh/C was obtained with uniformly dispersed Rh metal particles on conductive carbon black. The specific catalyst metal loading is shown in Table S1.†

2.3 Physical characterization

X-ray diffraction (XRD) analysis was conducted on a Bruker D8 Advance diffractometer, employing a scan rate of 5° min⁻¹ over the 2θ range of 10–90°. Microstructural characterization and elemental mapping were performed via transmission electron microscopy (TEM) and associated spectroscopy using an FEI Tecnai G220 system operated at 200 kV.³⁴ Atomic-resolution imaging was achieved through aberration-corrected high-angle annular dark-field scanning transmission electron microscopy (AC-HAADF-STEM) on a Titan 80–300 microscope at 300 kV, incorporating a probe-aberration corrector. Surface chemical states were examined by X-ray photoelectron spectroscopy (XPS) on a Thermo Scientific K-Alpha instrument with monochromatic Al Kα radiation (hν = 1486.6 eV). The valence band center position was determined using the first-moment integral: ∫N(ε)εdε/∫N(ε)dε, where N(ε) represents the density of states (DOS), herein approximated by background-subtracted XPS intensity.³⁵ Spectral integration was performed from the Fermi edge (E_F) to 10.0 eV binding energy, with E_F referenced to a sputter-cleaned Ag foil standard. The work function ϕ of the catalyst was derived from ultraviolet photoelectron spectroscopy (UPS) data according to the relationship ϕ = hν − E_cutoff, where hν = 21.2 eV (He I source), with E_cutoff determined from the secondary electron emission threshold relative to the Fermi edge (E_Fermi).

2.4 Electrochemical testing

The electrocatalytic activity toward the HOR was assessed in alkaline electrolyte using a rotating ring-disk electrode (glassy carbon substrate) coupled to a CHI 760E electrochemical workstation. A calibrated saturated calomel electrode served as the reference electrode and a graphite sheet as the counter electrode. All reported potentials were converted to the reversible hydrogen electrode (RHE) scale. The catalyst ink was formulated by dispersing 2.4 mg catalyst in a mixture containing 20 μL Nafion alcohol solution (5 wt%, Aldrich), 150 μL deionized water, and 450 μL isopropyl alcohol, followed by 30 min ultrasonication to achieve homogeneity. Prior to testing, the glassy carbon electrode underwent sequential polishing with Al₂O₃ slurry and deionized water rinsing. A 10 μL aliquot (catalyst loading: ∼0.20 mg cm⁻²) was deposited onto the electrode (geometric area: 0.196 cm²) and air-dried. Commercial Pt/C (20 wt%) served as the benchmark catalyst.

Hydrogen oxidation reaction (HOR) activity was evaluated by linear sweep voltammetry (LSV) at a 5.0 mV per s sweep rate in H₂-saturated 0.1 mol per L KOH, with electrode rotation maintained at 1600 rpm and 90% iR compensation applied. Cyclic voltammograms (CVs) were acquired in N₂-purged 0.1 mol per L KOH from 0.05 to 1.2 V (vs. RHE) at 50 mV s⁻¹. Electrochemical stability assessments employed chronoamperometry on catalyst-coated rotating disk electrodes (RDEs) in either pure H₂ or 1000 ppm CO/H₂-saturated 0.1 mol per L KOH, with constant rotation at 1600 rpm.

Kinetic current density (j_k) was extracted using the Koutecky–Levich equation:

where j is the measured current and j_d is the diffusion-limited current, which can be obtained through the Levich equation:

j_d = 0.62nFD^3/2ν^−1/6C₀ω^1/2 = BC₀ω^1/2 (2)

where n is the number of electrons involved in the HOR, F is the Faraday constant, D is the diffusion coefficient of the reactant, ν is the viscosity coefficient of the electrolyte, C₀ is the solubility of H₂ in the electrolyte, ω is the rotation speed, and B is the Levich constant. The exchange current density (j₀) is derived from the Butler–Volmer equation:where η is the overpotential, F is the Faraday constant, R is the universal gas constant, T is the Kelvin temperature, and α is the transfer coefficient. In the micropolarization region, this equation can be simplified to:

CO stripping voltammetry was conducted in 0.1 mol per L KOH. The catalyst-modified electrode was conditioned chronopotentiostatically at a fixed potential for 10 min within CO-saturated electrolyte to achieve monolayer CO adsorption. After transferring to N₂-purged 0.1 mol per L KOH, cyclic voltammetry was initiated from 0.05 to 1.2 V (vs. RHE) at 50 mV s⁻¹, with the first anodic scan quantifying the monolayer CO oxidation charge.

Electrochemically active surface area (ECSA) determination employed Cu underpotential deposition (Cu_upd) in 0.05 mol per L H₂SO₄ containing 0.05 mol per L CuSO₄. A background CV was initially recorded in Cu²⁺-free electrolyte. Following rapid electrolyte transfer, Cu_upd was achieved by potentiostatic deposition at 0.4 V (vs. RHE) for 100 s. The stripping profile was then acquired via linear sweep voltammetry at 50 mV s⁻¹ in supporting electrolyte. The ECSA was calculated using eqn (5) and (6).³⁶

where ν is the scan rate, i is the current, and m_PGM is the mass of precious metals on the electrode.

2.5 Theoretical calculation section

The spin-polarized density functional theory (DFT) calculations were carried out with the Vienna ab initio Simulation Package (VASP).^37,38 The projector-augmented plane wave (PAW) method³⁹ was used for electronic structure calculations and the generalized gradient approximation (GGA) with the Perdew–Burke–Ernzerhof (PBE)^40,41 was implemented to treat the exchange and correlation potentials. The long-range van der Waals (vdW) interactions were corrected by using the zero damping DFT-D3 method of Grimme.⁴² A plane wave cutoff energy of 450 eV was employed for all calculations. The geometries were considered to be converged when the forces and total energy were less than 0.02 eV Å⁻¹ and 10⁻⁵ eV, respectively. A 3 × 3 × 1 K-point mesh was sampled in the Brillouin zone⁴³ for structural optimization. The (111) facet with a 4 × 4 supercell is adopted to act as the active surface for the Pd and Rh nanomaterial. The slab has four layers. The upper 2 atomic layers were allowed to relax and the bottom 2 atomic layers were fixed to simulate the bulk system when structural relaxation was performed. The thickness of the vacuum layer is 15 Å. To construct the Pd/Rh interface, we employed an epitaxial growth approach by depositing Rh on the Pd substrate. Given the lattice mismatch, the Rh lattice parameter (3.80 Å) was adjusted to match the Pd lattice constant (3.90 Å). The resulting lattice mismatch of 2.3% falls well below the 5% threshold typically considered acceptable for coherent interface formation.⁴⁴ This rational design ensures that our Pd/Rh interface configuration maintains structural coherence and demonstrates thermodynamic stability, in good agreement with our experimental synthesis conditions.

The Gibbs free energy profiles for the whole reaction pathway from H₂ to H₂O in alkaline are shown, according to the most common Tafel–Volmer mechanism⁴⁵ in the following four elementary steps:

H* + OH⁻ → H* + OH + e⁻ (8)

H* + OH* → H₂O* (9)

H₂O* → H₂O + * (10)

The Gibbs adsorption free energy (ΔG) of each state and reaction free energy are evaluated using

ΔG = ΔE + ΔE_ZPE − TΔS (11)

where ΔE is the adsorption energy of adsorbed species and ΔE_ZPE and TΔS are respectively the differences corresponding to the zero-point energy and entropy at 300 K. The values of E_ZPE and TS for free H₂ and H₂O molecules determine the adsorption free energy (Table S4†). The values of E_ZPE and TS for all the adsorbed species (Table S5†) can be calculated with the vibration frequencies, as shown in previous literature.⁴⁶

The Gibbs free energy changes associated with the four reaction steps are denoted as ΔG₁, ΔG₂, ΔG₃, and ΔG₄, respectively. The theoretical overpotentials (η) from thermodynamics for Pd, Rh and Pd/Rh catalysts are expressed as

η = max(ΔG₁, ΔG₂, ΔG₃, ΔG₄)/e) (12)

3. Results and discussion

3.1 Synthesis and structural characterization

The Pd–Rh heterostructures were fabricated via a stepwise pyrolysis and chemical vapor deposition (CVD) approach⁴⁷ (Fig. 1a). Briefly, palladium acetylacetonate was ultrasonically dispersed in ethanol and uniformly mixed with Vulcan XC-72R carbon, followed by drying and thermal reduction under a H₂/Ar atmosphere to yield Pd/C nanoparticles. Subsequently, the Rh precursor was thermally decomposed into metallic vapor at controlled temperatures under an Ar flow, enabling the growth of heterostructure catalysts. The precise construction of heterojunctions originates from a stepwise controllable synthesis process—first, highly crystalline Pd nanoparticles are formed on the carbon support via pyrolysis at 500 °C; subsequently, the Rh(acac)₃ precursor is selectively deposited onto the pre-synthesized Pd/C surface during a gradient-pyrolysis CVD process. This process drives heterostructure formation through a triple synergistic mechanism: the surface energy difference (the Pd surface free energy of ∼2.5 J m⁻² is significantly lower than that of the carbon support, which is ∼5.0 J m⁻²) promotes preferential Rh adsorption onto Pd crystal facets; directed precursor transport combined with the underpotential deposition effect guides the epitaxial growth of Rh atoms on the Pd surface; simultaneously, pyrolysis at 300 °C facilitates interfacial atomic interdiffusion, forming a transition layer with bimetallic elements, which stabilizes into the heterostructure via lattice strain relaxation. Lattice mismatch between Pd (a = b = c = 3.90 Å) and Rh (a = b = c = 3.80 Å) can induce interface strain and atomic-level distortion, which is critical for tailoring electronic properties.⁴⁸ By modulating the temperature gradient and precursor stoichiometry, a series of catalysts—including monometallic Pd/C, Rh/C, and bimetallic Pd-Rh_x/C (where x denotes Rh/Pd atomic ratios, x = 0.02, 0.05, 0.1)—were synthesized for comparative studies.

Fig. 1 (a) Synthesis schematic of PdRh_0.05/C heterostructures. (b) XRD patterns of PdRh_0.05/C, Pd/C and Rh/C. (c) Low-magnification TEM image and particle size statistics of PdRh_0.05/C. (d) Aberration-corrected high-resolution transmission electron microscopy (AC-HRTEM) image of PdRh_0.05/C. (e) Line scan profiles corresponding to the yellow box in (d). (f) EDS elemental mapping of the PdRh_0.05/C catalyst (Pd: pink; Rh: orange), and (g) line scan along the direction of the red arrow in (f), showing the intensities of Pd and Rh.

The crystalline structure of PdRh_0.05/C was confirmed by XRD (Fig. 1b), exhibiting a dominant Pd (111) peak at 40° (PDF#01-088-2335) with a faint Rh (111) signal at 41° (PDF#01-073-6815), indicating successful Rh integration without disrupting the Pd lattice.⁴⁹ The annealing temperature of 320 °C is insufficient to enable significant Rh atom diffusion into the core of Pd nanoparticles. TEM revealed well-dispersed nanoparticles with an average size of about 7.02 nm (Fig. 1c). The high-resolution imaging showed distinct lattice fringes of 0.225 nm and 0.213 nm corresponding to Pd (111) and Rh (111), respectively (Fig. 1c and S3†), consistent with XRD results. The increased average particle size of PdRh_0.05/C (7.02 ± 0.3 nm vs. 5.64 ± 0.2 nm for Pd/C) further corroborates the growth of Rh around Pd particles (Fig. S1 and S2†). The aberration-corrected high-resolution transmission electron microscopy (AC-HRTEM) image (Fig. 1d and e) and EDS mapping (Fig. 1f, S4 and S5†) further unveiled the heterostructure's architecture: Rh nucleates epitaxially on Pd, partially penetrating its lattice to form strained interfaces with local atomic misalignment. EDS line scans (Fig. 1g) further confirmed compositional gradients, with Pd signal attenuation coinciding with Rh signal emergence, demonstrating Rh infiltration during CVD. Limited atomic mixing between Pd and Rh may occur at the nanoscale, particularly in surface and subsurface regions, potentially forming a partial solid solution. This interfacial restructuring, driven by the lattice mismatch between Pd (a = b = c = 3.90 Å) and Rh (a = b = c = 3.80 Å), generates localized strain fields while maintaining the overall crystalline structure of the nanoparticles.⁵⁰

The electronic properties of the PdRh_0.05/C catalyst were comprehensively investigated through X-ray photoelectron spectroscopy (XPS) and ultraviolet photoelectron spectroscopy (UPS) to elucidate the interfacial electronic interactions governing its enhanced catalytic activity. XPS analysis revealed significant electronic modulation upon Rh incorporation into the Pd matrix. For the Pd/C reference, the Pd 3d spectrum exhibited characteristic doublets at 335.7 eV (Pd⁰ 3d_5/2) and 341.1 eV (Pd⁰ 3d_3/2), along with oxidized species at 337.6 eV and 342.2 eV (Pd²⁺) (Fig. 2a).^51–53 The introduction of Rh induced 0.3–0.5 eV negative shifts of Pd binding energy, indicating electron accumulation at Pd sites through interfacial charge redistribution. This electron transfer was further corroborated by the positive shifts (+0.2 eV) observed in the Rh 3d spectrum of PdRh_0.05/C compared to Rh/C, demonstrating the depletion of electrons from Rh (Fig. 2b).^54–56 These shifts are consistent with the work function difference between Pd (4.35 eV) and Rh (4.28 eV), as electrons naturally migrate from Rh to Pd until Fermi level equilibration is achieved (Fig. 2f). XPS survey spectra and high-resolution XPS spectra of O 1s for all catalysts are provided in the ESI (Fig. S6 and S7).†

Fig. 2 (a) High-resolution XPS spectra of Pd 3d for Pd/C and PdRh_0.05/C. (b) High-resolution XPS spectra of Rh 3d for Rh/C and PdRh_0.05/C. (c) Schematic illustration of the direction of electron transfer on the Pd–Rh cluster in PdRh_0.05/C. (d) The high-resolution valence band VB XPS spectra of Pd/C, PdRh_0.05/C, and Rh/C relative to the Fermi level used as simulations of the density of states, with horizontal lines representing the position of the center of the d-band. (e) Secondary electron cutoff edges of Pd/C, PdRh_0.05/C, and Rh/C catalysts; (f) work functions of Pd/C, PdRh_0.05/C, and Rh/C. (g) Valence bands of Pd/C, PdRh_0.05/C, and Rh/C.

To establish the structure–property correlation, the d-band electronic structures of Pd/C, PdRh_0.05/C, and Rh/C were characterized using the XPS high-resolution valence band (VB),^51,57,58 which is proportional to the density of states (DOS) and directly related to the adsorption strength of reaction intermediates (Fig. 2d). In the Pd–Rh heterojunction, electrons directionally migrate from Rh with a lower work function to Pd with a higher work function (Fig. 2c). This charge transfer increases the d-band electron density of Pd, significantly elevating the average d-band energy to −4.46 eV through increased inter-electron repulsion and occupancy of interfacial antibonding states. The downward shift of the d-band center moderately enhances the OH adsorption energy, thereby substantially enhancing HOR kinetics.

Complementary UPS measurements (Fig. 2e) revealed a reduced work function (4.12 eV) for PdRh_0.05/C compared to both Pd/C and Rh/C, indicating a weakened surface electron binding capability that promotes charge transfer during catalytic processes (Fig. 2f). The UPS valence band spectrum (Fig. S8†) further demonstrated the advantages of heterostructures between Pd and Rh: the elevated valence band of Rh (3.15 eV) promotes H₂ dissociation and the lowered valence band of Pd enhances OH⁻ adsorption (Fig. 2g). This synergistic electronic configuration, where the interfacial region exhibits intermediate characteristics, creates an optimal balance between hydrogen activation and oxygen-containing species adsorption, both of which are critical for efficient hydrogen oxidation reactions. The combined XPS and UPS analyses provide compelling evidence that the enhanced catalytic performance of PdRh_0.05/C stems from these interfacial electronic modifications, which collectively lower activation barriers and stabilize key reaction intermediates.

3.2 Electrocatalytic performance for alkaline HOR

Initial screening of PdRh_x/C catalysts with varying Rh contents (x = 0.02, 0.05, 0.10) revealed an optimal catalyst of PdRh_0.05/C that exhibited the highest HOR activity (Fig. S9†). This catalyst was therefore selected for detailed electrochemical characterization. All LSV tests are performed with a rigorous 5 mV s⁻¹ low-speed sweep to ensure that the non-faradaic current contribution is <0.5% (a criterion validated by sweep gradient experiments in Fig S10†). The electrochemical behavior of PdRh_0.05/C was investigated using rotating disk electrode (RDE) measurements in N₂-saturated 0.1 M KOH electrolyte (Fig. 3a). Comparative analysis with monometallic Pd/C and Rh/C revealed significant differences in the hydrogen/oxygen adsorption/desorption characteristics.⁵⁹ Compared to that of pure Pd/C, the H_upd peak of PdRh_0.05/C negatively shifts 55 mV, indicating that the introduction of Rh effectively weakens the adsorption of H*. The CV curve of Pd/C exhibited a distinct oxygen region at 0.7 V vs. RHE, while Rh/C showed this feature at a lower potential of 0.4 V vs. RHE. Notably, the oxygen region of PdRh_0.05/C appeared at an intermediate potential between these values, demonstrating a negative shift relative to Pd/C. This potential shift indicates enhanced adsorption strength of oxygen-containing intermediate species (*OH) on the bimetallic surface, which is crucial for maintaining active sites available for the hydrogen oxidation reaction.^49,60 The observed electrochemical behaviour provides further evidence for the electronic interaction between Pd and Rh atoms in the bimetallic system, consistent with previous XPS and UPS analyses. The modified surface electronic structure not only facilitates hydrogen oxidation but also enhances the strong adsorption of oxygen-containing species that could remove poison molecules (e.g. CO) from the catalyst surface.^61,62 These findings collectively demonstrate how the strategic combination of Pd and Rh at the atomic level creates a superior catalytic environment for hydrogen oxidation in alkaline media.

Fig. 3 (a) The cyclic voltammetry (CV) curves of Pd/C, PdRh_0.05/C and Rh/C catalysts recorded on a rotating disk electrode (RDE) in 0.1 M KOH electrolyte saturated with N₂. (b) LSV curves of different catalysts at 1600 rpm. (c) HOR Tafel plots of kinetic current densities (j_k) for different catalysts. (d) Linear fitting diagrams of the micro-polarization regions for different catalysts. (e) Mass activity and specific activity of different catalysts at an overpotential of 50 mV (vs. RHE). (f) Performance comparison with other excellent Pd-based catalysts. (g) Chronoamperometry of Pd/C, PdRh_0.05/C, and Rh/C was performed in H₂-saturated 0.1 M KOH at 0.05 V (vs. RHE).

The electrochemically active surface area of Pd/C and PdRh_0.05/C was measured using the Cu underpotential deposition method (Cu_upd) (Fig. S11†). The corresponding specific active surface areas of Pd/C, PdRh_0.05/C, and Pt/C were 0.471, 0.588, and 0.365 cm² μg_PGM⁻¹, respectively. Linear sweep voltammetry measurements at 1600 rpm rotation speed demonstrated that PdRh_0.05/C achieved the highest current density across the entire potential window (Fig. 3b), significantly outperforming both monometallic counterparts and commercial Pt/C. To elucidate the reaction kinetics, polarization curves were recorded at various rotation speeds (400–3600 rpm) and analyzed using the Koutecký–Levich methodology (Fig. S12†). The near-unity slopes (4.76 cm² mA⁻¹ s^−1/2 for PdRh_0.05/C vs. theoretical 4.87 cm² mA⁻¹ s^−1/2) confirmed the first-order dependence on hydrogen concentration and a complete 2-electron transfer process.^63–65 The exchange current density (j₀) can also be obtained from the micro-polarization region that deviates only several millivolts from the equilibrium potential (i.e., from −5 to 5 mV). Kinetic analysis revealed that the j₀ of PdRh_0.05/C exhibited exceptional activity, 0.34 mA cm⁻², representing a 11.9-fold enhancement over Pd/C (Fig. 3d). Based on intrinsic activity assessed by normalization of the electrochemical surface area derived from Cu_upd (Fig. S11†),⁶⁶ PdRh_0.05/C showed a specific activity of 1.62 mA cm⁻², which is 30.7 times higher than that of Pd/C and 2.5 times greater than that of Pt/C. When normalized with precious metal mass, PdRh_0.05/C demonstrated a remarkable mass activity of 1.16 mA μg_PGM⁻¹ at 50 mV overpotential, which is 35 times higher than that of Pd/C, even surpassing that of Pt/C by 4.8-fold (Fig. 3e). This exceptional performance places PdRh_0.05/C among the most active Pd-based HOR catalysts reported to date (Fig. 3f and Tables S2 and S3†). The stability of PdRh_0.05/C was assessed through chronoamperometry at 0.05 V vs. RHE, revealing excellent durability with only 13% current decay after 10 000 s of operation, compared to >40% degradation observed for Pd/C and Rh/C (Fig. 3g). The PdRh_0.05/C catalyst was characterized by TEM after the long-term stability test, and it was found that the particle size of the catalyst only increased slightly compared with that before the stability test, indicating that PdRh_0.05/C had good structural stability (Fig. S13†). This enhanced stability likely stems from the synergistic electronic effects between Pd and Rh that mitigate surface oxidation and dissolution during prolonged operation. The outstanding HOR performance of PdRh_0.05/C can be attributed to multiple factors: (i) optimized electronic structure facilitating hydrogen adsorption/desorption, (ii) enhanced OH⁻ adsorption at interfacial sites, and (iii) improved resistance to surface oxidation. These results demonstrate that rational design of Pd–Rh heterostructures can overcome the limitations of monometallic Pd catalysts in alkaline media while reducing precious metal requirements compared to Pt-based systems.

3.3 CO-tolerance and stability evaluation

The presence of CO impurities in hydrogen fuel streams, originating from industrial production methods such as natural gas reforming and water–gas shift reactions, presents a critical challenge for AEMFC catalysts.^67–69 Conventional catalysts suffer from severe CO poisoning due to the strong chemisorption of CO molecules on active sites, which significantly deteriorates their hydrogen oxidation reaction (HOR) performance.⁷⁰ In this context, we systematically evaluated the CO tolerance of the PdRh_0.05/C catalyst, which demonstrates remarkable resistance to CO poisoning compared to monometallic and commercial counterparts.

The CO stripping voltammetry measurements revealed a distinct negative shift of 100 mV in the CO oxidation peak potential for PdRh_0.05/C (0.8 V vs. RHE) relative to Pd/C (0.9 V vs. RHE). This significant negative shift indicates enhanced CO electro-oxidation kinetics at lower potentials, suggesting that the incorporation of Rh facilitates the formation of surface hydroxyl species that promote CO removal through the Langmuir–Hinshelwood mechanism (Fig. 4a and S14†).^71,72 The improved CO oxidation capability correlates well with the electronic structure modifications observed in UPS and XPS analyses, where the optimized interfacial electronic environment between Pd and Rh atoms enhances the catalyst's ability to activate water molecules for CO oxidation.

Fig. 4 (a) Comparison of CO adsorption peak positions for Pd/C and PdRh_0.05/C. (b) PdRh_0.05/C and (c) Pd/C were tested by chronopotentiometry in 0.1 M KOH saturated with H₂ (solid line) and in 0.1 M KOH saturated with H₂/1000 ppm CO (dashed line) after immersing the electrode in the H₂/1000 ppm CO saturated electrolyte for 10 min. (d) The electrocatalytic activities of Pd/C and PdRh_0.05/C determined by chronoamperometric measurements in a H₂/1000 ppm CO-saturated 0.1 M KOH solution at 0.05 V (vs. RHE).

Acute poisoning tests conducted in a H₂/1000 ppm CO atmosphere demonstrated the superior CO tolerance of PdRh_0.05/C, which retained 98.23% of its initial current density, while commercial Pt/C and monometallic Pd/C and Rh/C suffered severe activity losses (Fig. 4b, c and S15†). Especially, 80.6% of the performance of Pd/C is lost after being poisoned by CO. This exceptional CO-tolerance was further confirmed by chronopotentiometry stability tests, where PdRh_0.05/C maintained 83% of its initial current after 1500 s of continuous operation in CO-containing electrolyte, representing a 4.5-fold improvement over monometallic catalysts (Fig. 4d). In contrast, Pd/C lost 78% of its performance after 1500 s, and Rh/C even lost 61% of its performance during the first 300 s. The electronic modulation induced by Rh incorporation weakens CO adsorption energy on Pd sites, while the interfacial structure promotes OH adsorption at lower potentials.^28,73 These features collectively contribute to the catalyst's ability to maintain activity in CO-containing environments, addressing one of the most critical challenges in practical fuel cell applications where reformate hydrogen with residual CO must be utilized.

3.4 Mechanism investigation

To elucidate the superior HOR performance of the Pd/Rh bimetallic heterostructure, we conducted systematic first-principles calculations based on DFT using VASP. Three interrelated theoretical models were developed from their crystal structure to simulate the Pd nanomaterial (Fig. S16a†), Rh surface (Fig. S16b†) and Pd/Rh interface (Fig. S16c†) synthesized in experiments. The most stable adsorption sites of the intermediate products (H*, OH* + H* and H₂O*) on these models are systematically studied and the stable adsorption configurations are shown in Fig. S17.† We found that both H* and OH* tend to adsorb at fcc sites, whereas H₂O* preferentially adsorbs at the top sites of metal atoms. In the Pd/Rh bimetallic interface system, H* exhibits preferential adsorption on Pd-terminated domains, while OH* predominantly adsorbs on Rh-rich regions. This behavior aligns with the adsorption energy of H* and OH* on individual Pd and Rh surfaces (Fig. S18† and 5a). To evaluate the HOR activity of the catalyst series, the potential-dependent energy evolution across the alkaline HOR process (H₂ → H₂O) is mapped in Fig. 5b through the widely accepted Tafel–Volmer mechanism.⁷⁴ Notably, the OH* adsorption step is identified as the potential-determining step for all the catalysis systems. The bimetallic Pd/Rh interface exhibits the optimal HOR performance with a low theoretical overpotential of 0.18 V because it offers easier OH* adsorption, when compared to the Pd surface (0.77 V) and Rh surface (0.34 V). The order of theoretical overpotentials from thermodynamics calculated by DFT is consistent with that obtained by experimentally.

Fig. 5 (a) Adsorption energy of OH* on the Pd surface, Rh surface and Pd/Rh interface. (b) Gibbs free energy diagrams on the Pd surface, Rh surface and Pd/Rh interface. (c) The electron density difference shows the rearrangement of electron density resulting from the adsorption of OH*. The blue, orange, red and white balls represent the Pd, Rh, H and O atoms, respectively. COHP for the interaction of (d) the Pd–O bond on the Pd surface and (e and f) the Rh–O bond on the Rh surface and Pd/Rh interface. The gray dashed line represents the Fermi level.

To gain deeper insights into the atomic-scale mechanisms underlying the enhanced OH* adsorption capability and consequent HOR performance improvement in the bimetallic Pd/Rh heterostructure, we systematically conducted a combination of Bader analysis, electron density difference, and Crystal Orbital Hamilton Population (COHP) calculations.⁷⁵ The electron density difference analysis revealed the rearrangement of electron density induced by OH adsorption (Fig. 5c). On Pd/Rh, OH* can get 0.48 electrons from Rh atoms, which is more than that on Pd (0.46e⁻) and Rh (0.47e⁻), indicating that the Pd/Rh interface has a stronger OH* adsorption capacity. The quantitative COHP analysis was conducted to better understand the orbital interactions of valence electrons between the three active center atoms (Pd or Rh) and the oxygen atom in OH*. The filling population of bonding and antibonding orbitals was quantified by integrated COHP (ICOHP). Taking OH* adsorption as an example, the spin-up/spin-down bands of Rh–O bonds for the Pd/Rh interface contribute stronger binding strengths with the ICOHP of −2.54/−2.53 eV than that of Pd–O bonds for the Pd surface (−0.06/−0.06 eV) and Rh–O bonds for the Rh surface (−2.12/−2.12 eV) in Fig. 5d–f, indicating the strengthened interactions of Rh valence-electron orbitals and thus enhanced interaction of OH* on the Pd/Rh interface, which aligns with the electronic structure analysis mentioned above. The electronic structure analysis of the PdRh heterojunction, including Bader charge and charge density difference calculations, reveals substantial charge accumulation at the interface (Fig. S19†).⁷⁶ The interfacial charge transfer (1.35e⁻ from Rh to Pd) revealed by Bader analysis aligns with the work function difference (5.15 eV for Rh vs. 5.26 eV for Pd), suggesting electron flow toward Pd to equilibrate the Fermi levels. This trend is further corroborated by the UPS valence band spectrum, where the sequential shift of spectral features mirrors the theoretical prediction.

Therefore, our DFT calculations reveal that the Pd/Rh interfacial electronic synergy modulates COHP through valence electron redistribution, which strengthens Rh–O orbital hybridization to optimize OH* adsorption. This electronic tailoring lowers the PDS (OH⁻ → OH*) energy barrier by 326% (0.77 → 0.18 eV), ultimately boosting HOR catalytic efficiency by 2–3 orders of magnitude relative to monometallic systems.

4. Conclusions

In summary, we have demonstrated CO-tolerant alkaline HOR catalysts through electrical manipulation of Pd–Rh bimetallic interfaces. A PdRh_0.05/C catalyst with an average size of ∼7 nm is manufactured via a chemical vapor deposition technique. The heterostructure enhances electron transport from Rh sites to Pd sites, which can lower the d-band center of Pd and boost oxygen affinity on Rh. In contrast to Pd/C, PdRh_0.05/C exhibits weakened *H and improved *OH adsorption. PdRh_0.05/C delivers a greater HOR activity with an exchange current density of 0.34 mA cm⁻², outperforming Pd/C by a factor of 11.9. At an overpotential of 50 mV, the SA and MA of PdRh_0.05/C are 1.62 mA cm⁻² and 1.16 mA μg_PGM⁻¹, respectively, which are 30.7 times higher and 35 times higher than those of Pd/C. PdRh_0.05/C has remarkable durability, with merely a 13% reduction in current after 10 000 seconds of operation, in contrast to the over 40% deterioration noted for Pd/C. Furthermore, PdRh_0.05/C delivers improved CO tolerance and can preserve 83% of its performance under 1000 ppm CO/H₂, while Pd/C loses 78% of its performance after 1500 s. Electronic structure investigation showed improved charge transfer (0.48e⁻) and stronger Rh–O orbital hybridization (ICOHP: −5.08 eV) at the interface, enabling OH* binding. This interfacial synergy reduces the potential-determining phase (OH⁻ → OH*) energy barrier by 326%, enhancing HOR efficiency by 2–3 orders of magnitude compared to monometallic systems. These factors synergistically enable the PdRh_0.05/C catalyst to attain an unparalleled level of activity, surpassing the majority of documented Pd-based catalysts. This research addresses the enduring activity–stability trade-off in HOR catalysis and provides inspiration for material design ideas applicable to other electrocatalytic processes sensitive to poisoning.

Data availability

All relevant data are within the manuscript and its additional files.

Author contributions

Si-Yu Rong: conceptualization, methodology, investigation, data curation, writing – original draft, and writing – review & editing. Wei-Dong Li: conceptualization, methodology (optimization), and data curation. Min-Han Li: writing – review & editing. Hao-Ran Wu: validation and formal analysis. Xing Yuan: investigation (supporting) and data analysis. Ning-Ran Kang: investigation (supporting) and discussion. Xiao-Pei Xu: conceptualization, software, and formal analysis (computational simulations). Bang-An Lu: funding acquisition, resources, methodology, conceptualization, and writing – review & editing. Si-Yu Rong and Wei-Dong Li contributed equally to this work.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (No. 22372152, 22409053, and 22202183), the Natural Science Foundation of Henan Province (No. 252300421091 and 242300420544), the China Postdoctoral Science Foundation (No. 2021TQ0295 and 2022M712865), the Talent Introduction Fund of Henan University of Technology (No. 2021BS003) and the National Natural Science Foundation Cultivation Project of Henan University of Technology (No. 2024PYJH058).

Notes and references

1. Y. Yang, P. Li and X. Zheng, et al. , Chem. Soc. Rev., 2022, 51, 9620–9693.
2. J. R. Varcoe, P. Atanassov and D. R. Dekel, et al. , Energy Environ. Sci., 2014, 7, 3135–3191.
3. W. R. Huang, C. X. Yu and Y. R. Lu, et al. , Nano Res., 2019, 12, 1483–1488.
4. H. Lei, X. Yang and Z. Chen, et al. , Adv. Mater., 2025, 37, 2410106.
5. X. Wei, S. Sharma and A. Waeber, et al. , Resour., Conserv. Recycl., 2025, 215, 108134.
6. S. T. Thompson, B. D. James and J. M. Huya-Kouadio, et al. , J. Power Sources, 2018, 399, 304–313.
7. D. R. Dekel, Curr. Opin. Electrochem., 2018, 12, 182–188.
8. J. Zhang, W. Zhu and T. Huang, et al. , Adv. Sci., 2021, 8, 2100284.
9. F. Y. Gao and M. R. Gao, Acc. Chem. Res., 2023, 56, 1445–1457.
10. L. Zhang, C. Jia and F. Bai, et al. , Fuel, 2024, 355, 129455.
11. M. T. Anwar, X. Yan and M. R. Asghar, et al. , Int. J. Energy Res., 2019, 43, 2694–2721.
12. S. T. Thompson, D. Peterson and D. Ho, et al. , J. Electrochem. Soc., 2020, 167, 084514.
13. Y. Li, M. Y. Chen and B. A. Lu, et al. , J. Electrochem., 2023, 29, 2215002.
14. X. Sun, S. Wang and Y. Hou, et al. , J. Mater. Chem. A, 2023, 11, 13089.
15. M. Y. Chen, S. H. Yin and G. Li, et al , ACS Nano, 2024, 18, 26115–26126.
16. B. Zhang, G. Zhao and B. Zhang, et al. , Adv. Mater., 2021, 33, 2105400.
17. L. Su, Y. Zhao and Y. Jin, et al. , Adv. Funct. Mater., 2022, 32, 2113047.
18. X. Song, X. G. Zhang and Y. L. Deng, et al. , J. Am. Chem. Soc., 2023, 145, 12717–12725.
19. S. Henning, J. Herranz and H. A. Gasteiger, J. Electrochem. Soc., 2014, 162, F178.
20. M. Shao, J. Power Sources, 2011, 196, 2433–2444.
21. D. C. Papageorgopoulos, M. Keijzer and J. B. J. Veldhuis, et al. , J. Electrochem. Soc., 2002, 149, A1400.
22. M. Y. Chen, Y. Li and H. R. Wu, et al. , Materials, 2023, 16, 2590.
23. R. Zhao, X. Yue and Q. Li, et al. , Small, 2021, 17, 2100391.
24. L. F. Shen, B. A. Lu and Y. Y. Li, et al. , Angew. Chem., Int. Ed., 2020, 59, 22397.
25. R. Subbaraman, V. Stamenkovic and N.M. Markovic, et al. , Science, 2011, 334, 1256–1260.
26. T. Tang, L. Ding and Z. Yao, et al. , Adv. Funct. Mater., 2022, 32, 2107479.
27. C. A. Campos-Roldán, Energy Rev., 2019, 2, 312–331.
28. D. Strmcnik, M. Uchimura and C. Wang, et al. , Nat. Chem., 2013, 5, 300–306.
29. S. Y. Ma, T. Ma and Q. Hu, et al. , Sci. China Mater., 2022, 65, 3462–3469.
30. H. R. Wu, et al , J. Electrochem. Soc., 2025, 172, 024508.
31. J. Hyun and H. T. Kim, Energy Environ. Sci., 2023, 16, 5633–5662.
32. J. T. Ren, L. Chen and H. Y. Wang, et al. , Energy Environ. Sci., 2024, 17, 3960–4009.
33. B. A. Lu, L. F. Shen and J. Liu, et al. , ACS Catal., 2021, 11, 355–363.
34. H. R. Wu, Y. Li and M. Y. Chen, et al. , J. Electrochem. Soc., 2025, 172, 024508.
35. W. P. Zhou, A. Lewera and R. Larsen, et al. , J. Phys. Chem. B, 2006, 110, 13393–13398.
36. S. Zhang, R. Ren and J. Cao, et al. , Adv. Energy Mater., 2024, 2404266.
37. G. Kresse and J. Furthmüller, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 54, 11169–11186.
38. G. Kresse and D. Joubert, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 1758–1775.
39. P. E. Blöchl, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 50, 17953–17979.
40. J. P. Perdew and Y. Wang, Phys. Rev. B: Condens. Matter Mater. Phys., 1992, 45, 13244–13249.
41. J. P. Perdew, J. A. Chevary and S. H. Vosko, et al. , Phys. Rev. B: Condens. Matter Mater. Phys., 1992, 46, 6671–6687.
42. S. Grimme, J. Antony and S. Ehrlich, et al. , J. Chem. Phys., 2010, 132, 154104.
43. M. Methfessel and A. T. Paxton, Phys. Rev. B: Condens. Matter Mater. Phys., 1989, 40, 3616–3621.
44. L. Yang, H. Liu and Y. Li, et al. , Angew. Chem., Int. Ed., 2025, 64, e202414518.
45. Y. Shen, X. L. Zhang and M. R. Qu, et al. , Nat. Commun., 2024, 15, 7861.
46. J. K. Nørskov, et al , J. Electrochem. Soc., 2005, 152, J23.
47. T. Stracensky, L. Jiao and Q. Sun, et al. , ACS Catal., 2023, 13, 14782–14791.
48. X. Zhang, L. Xia and G. Zhao, et al. , Adv. Mater., 2023, 35, 2208821.
49. F. Tzorbatzoglou, A. Brouzgou and P. Tsiakaras, Appl. Catal., B, 2015, 174–175, 203–211.
50. G. Zhao, Y. Jiang and S. X. Dou, et al. , Sci. Bull., 2021, 66, 85–96.
51. X. Qin, L. Zhang and G. L. Xu, et al. , ACS Catal., 2019, 9, 9614–9621.
52. T. Bhowmik, M. K. Kundu and S. Barman, ACS Catal., 2016, 6, 1929–1941.
53. M. V. Pagliaro, C. Wen and B. Sa, et al. , ACS Catal., 2022, 12, 10894–10904.
54. H. Huang, K. Liu and F. Yang, et al. , Angew. Chem., Int. Ed., 2023, 62, e202315752.
55. L. Su, Y. Zhao and F. Yang, et al. , J. Mater. Chem. A, 2020, 8, 11923–11927.
56. M. K. Kundu, R. Mishra and T. Bhowmik, et al. , J. Mater. Chem. A, 2018, 6, 23531–23541.
57. X. Wu, F. Chen and N. Zhang, et al. , Small, 2017, 13, 1603387.
58. V. R. Stamenkovic, B. Fowler and B. S. Mun, et al. , Science, 2007, 315, 493–497.
59. X. Song, Q. Yang and K. Zou, et al. , Adv. Funct. Mater., 2025, 35, 2414570.
60. Y. Li, H. R. Wu and Y. Yu, et al. , Small, 2025, 21, 2411678.
61. L. Su, H. Wu and S. Zhang, et al. , Adv. Mater., 2025, 37, 2414628.
62. Y. Li, M. Y. Chen and B. A. Lu, et al. , App. Catal. B: Environ., 2024, 342, 123458.
63. X. Zheng, S. Zhang and X. Zheng, et al. , Adv. Mater., 2025, 2502127.
64. Z. Zhuang, S. Giles and J. Zheng, et al. , Nat. Commun., 2016, 7, 10141.
65. H. R. Wu, M. Y. Chen and W. D. Li, et al. , Chem.–Asian J., 2024, 19, e202300862.
66. R. Y. Shao, L. W. Chen and Q. Q. Yan, et al. , ACS Appl. Energy Mater., 2021, 4, 4284–4289.
67. L. Xia and P. Cui, et al. , Nano Lett., 2021, 21, 4845–4852.
68. K. Yang, H. Ma and R. Ren, et al. , Angew. Chem., Int. Ed., 2024, 63, e202318389.
69. M. Y. Chen, J. Le and H. R. Wu, et al. , J. Catal., 2025, 447, 116093.
70. Z. Yang, W. Lai and B. He, et al. , Adv. Energy Mater., 2023, 13, 2300881.
71. B. Zhang, B. Zhang and G. Zhao, et al. , Nat. Commun., 2022, 13, 5894.
72. W. Y. Zhao, M. Y. Chen and H. R. Wu, et al. , Materials, 2024, 17, 2879.
73. D. Strmcnik, M. Uchimura and C. Wang, et al. , Nat. Chem., 2013, 5, 300–306.
74. T. J. Schmidt, P. N. Ross and N. M. Markovic, J. Phys. Chem. B, 2001, 105, 12082–12086.
75. F. Li, S. H. Chan and Z. Tu, Chem. Rec., 2024, 24, e202300067.
76. G. Henkelman, A. Arnaldsson and H. Jónsson, Comput. Mater. Sci., 2006, 36, 354–360.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5ta04787k

‡ Si-Yu Rong and Wei-Dong Li contributed equally to this work.

---