A Zr/Sm co-doped Fe₂O₃:NiCo-MOF heterojunction photoanode for augmented photoelectrochemical water splitting†

Periyasamy Anushkkaran^a, Weon-Sik Chae^b, Hyobin Han^c, Tae Woo Kim^c, Sun Hee Choi^d, Richard Dinsdale^e, Bongkyu Kim*^f and Jum Suk Jang*^af
aDepartment of Integrative Environmental Biotechnology, College of Environmental and Bioresource Sciences, Jeonbuk National University, Iksan 54596, Republic of Korea. E-mail: jangjs75@jbnu.ac.kr
^bDaegu Center, Korea Basic Science Institute, Daegu 41566, Republic of Korea
^cHydrogen Research Department, Korea Institute of Energy Research, Gajeong-ro, Yuseong-gu, Daejeon 34129, Republic of Korea
^dPohang Accelerator Laboratory (PAL), Pohang University of Science and Technology (POSTECH), Pohang 37673, Republic of Korea
^eSustainable Environment Research Centre, University of South Wales, Pontypridd, CF37 1DL, Wales, UK
^fDivision of Biotechnology, College of Environmental and Bioresource Sciences, Jeonbuk National University, Iksan 54596, Republic of Korea. E-mail: bkim@jbnu.ac.kr

First published on 1st August 2025

---

---

Introduction

Currently, the global imperative to reduce reliance on fossil fuels necessitates the immediate adoption of green and clean energy derived from renewable sources. Photoelectrochemical water splitting (PEC-WS) is a promising technology that can convert solar energy into chemical fuels using semiconductor photoelectrodes. This process offers an efficient, environmentally friendly and carbon-free path to produce hydrogen, which can help address the energy crisis.^1,2 Hematite (Fe₂O₃; HT) has been demonstrated to be one of the most promising contenders for solar water splitting on account of its sufficient bandgap of ∼2.1 eV, high chemical stability in alkaline electrolyte, abundance in nature, non-toxicity and cost-effectiveness. Nevertheless, photoelectrochemical (PEC) efficiency of the HT photoanode is sternly constrained due to many factors, such as poor bulk electronic conductivity, substantial bulk and surface charge recombination and sluggish water oxidation kinetics at the surface.^3–5 In general, the poor inherent conductivity of HT leads to an increase in bulk charge recombination, resulting in a decrease in the lifespan of charge carriers. Moreover, the presence of surface-trapping states causes the Fermi level pinning (FLP) effect, leading to a lethargic rate of water oxidation.

These aforementioned issues have impeded the effective application of HT as a photoanode material for PEC-WS. One potential strategy for surmounting these limitations is to augment the bulk electronic conductivity of HT via doping it with high-valence ions, which can mitigate bulk charge recombination. Additionally, surface engineering techniques, including passivation overlayer or cocatalyst loading, can help alleviate the FLP effect and enhance water oxidation kinetics.^6–8 Mono-ion doping can create impurity states within the bandgap, which can act as recombination centers. Therefore, co-doping is an effective strategy to reduce the defect concentration and modify the electronic properties of HT nanostructures, thereby minimizing bulk electron–hole recombination.^9–11

Several studies have demonstrated that the co-doping approach has been conducted using various tetravalent dopants.^12–14 In addition to tetra and pentavalent dopants, researchers have focused on rare-earth metal-doped Fe₂O₃ photoanodes. For instance, a simple in situ hydrothermal method was utilized by Bai et al. to fabricate a Ce-doped Fe₂O₃ photoanode with an enhanced photocurrent density (PD) of 0.91 mA cm⁻² at 1.23 V_RHE.¹⁵ Further on, Jin and co-workers were able to enhance the performance of the Fe₂O₃ photoanode for PEC-WS by introducing Gd-doping and attaining a PD of 1.37 mA cm⁻² at 1.23 V_RHE.^16,17 In addition to these dopants, samarium (Sm) has been shown to have the ability to augment ionic conductivity, carrier mobility and catalytic activity of a photoanode.¹⁸ Besides, researchers are currently directing their attention toward rare-earth metal co-doped metal oxide photoanodes.^19,20 Geng et al. recently developed a Co/La co-doped porous BiVO₄ photoanode for PEC-WS.²¹ However, there has been no report of lanthanides co-doped in Fe₂O₃ photoanodes for PEC-WS. Thus, we focused on Zr/Sm co-doped Fe₂O₃ photoanodes for the first time with the purpose of PEC-WS.

Metal–organic frameworks (MOFs) are a new type of porous inorganic–organic hybrid material that are constructed by arranging metal clusters and organic ligands in an organized manner.^22,23 In recent years, MOFs have been utilized as effective cocatalysts for modifying hematite in the process of PEC-WS.^24,25 MOFs serve as scaffolds that are capable of easily accommodating the electronic demands of metal centers in high oxidation states. As a consequence, the formation of O–O bonds for the oxygen evolution reaction (OER) is favourable in an entropic manner during the water oxidation reaction.²⁶ Past studies have demonstrated that bimetallic organic frameworks have exhibited exceptional results in terms of PEC properties.^27,28 For instance, Wang et al. deposited NiFe-MOFs on a Zr-doped Fe₂O₃ photoanode using an indirect ligand-assisted transformation method and achieved a PD of 1.81 mA cm⁻² at 1.23 V_RHE.²⁹ Co-containing MOF materials have garnered significant research attention due to their exceptional performance in the OER, which can be attributed to the favourable physical and chemical characteristics, such as tiny particle size and an abundance of metal sites inside the MOF materials.³⁰ Chen and colleagues fabricated Fe₂O₃ photoanodes covered with FeCo-MOFs to enhance PEC-WS.³¹ The FeCo-MOF not only performed as an OER cocatalyst, but also formed a p–n heterojunction with the Fe₂O₃ photoanode. In addition, Yang et al. reported that the HT photoanode formed a p–n heterojunction with the CoFe metal–thiolate framework.³² On the other hand, as far as our understanding goes, NiCo-MOF-modified Fe₂O₃ photoanodes have not yet been reported.

Here, we designed, for the first time, a Zr/Sm co-doped Fe₂O₃ photoanode with a NiCo-MOF coating for PEC-WS. The bulk characteristics of the hematite photoanode were examined, and the influence of Zr/Sm co-doping was found to be significant. Furthermore, the outcomes of this study revealed that the NiCo-MOF not only effectually passivated the surface-trapping states (recombination surface states or surface defects) by alleviating the FLP effect to overcome surface charge recombination, but it also constructed a p–n heterojunction with Zr/Sm–Fe₂O₃ to accelerate bulk charge separation and transport via an effective internal electric field, and played the role of a cocatalyst to ameliorate hole transfer and thus advance the water oxidation kinetics. The optimized Zr/Sm–HT:NiCo-MOF photoanode attained a photocurrent density of 2.36 mA cm⁻² at 1.23 V_RHE, which is 143% higher than that of Bare-HT. Thus, effective PEC-WS could be significantly enhanced by leveraging the combined benefits of co-doping and bimetal MOF decoration.

Experimental

Synthesis of bare-Fe₂O₃ (Bare-HT) photoanodes

First, a solution was prepared by dissolving 4.5 mM FeCl₃·6H₂O and 1 M NaNO₃ in 10 mL deionized (DI) water. The pH of the solution was adjusted to 1.5 using HCl, and then 1 mL of ethanol was added. Subsequently, F-doped SnO₂ (FTO) glass was subjected to ultrasonic cleaning and then placed in a vial holding the above precursor solution. A hydrothermal reaction was carried out at 100 °C for 6 h. The as-grown bare-FeOOH films were cleansed with DI water, air dried and then quenched at 800 °C for 10 min to convert them into Bare-HT photoanodes.

Synthesis of Zr-doped Fe₂O₃ (Zr–HT) photoanodes

The experimental method was the same as that used for the synthesis of Bare-HT, with the exception that 1 mL of a 0.9 mM ZrO(NO₃)₂–ethanol solution was introduced into the solution prior to the hydrothermal process. The as-grown Zr–FeOOH films were washed with DI water, air dried and subsequently quenched at 800 °C for 10 min to fabricate Zr–HT photoanodes.

Fabrication of Sm-doped Fe₂O₃ (Sm–HT) and Zr/Sm co-doped Fe₂O₃ (Zr/Sm–HT) photoanodes

The hydrothermally prepared bare-FeOOH films were subjected to microwave (MW) heat treatment for 20 s in a 0.15 mM SmCl₃·6H₂O solution. Following MW treatment, the Sm–FeOOH samples were dried naturally and quenched at 800 °C for 10 min to convert them into Sm–HT photoanodes. Similarly, Zr/Sm–HT photoanodes were prepared using microwave heat treatment for Zr–FeOOH films containing a Sm precursor.

Deposition of NiCo-MOF on Zr/Sm–HT photoanodes

Initially, 25 mM Ni(NO₃)₂·6H₂O and 50 mM Co(NO₃)₂·6H₂O were dissolved in methanol. Then, 300 mM 2-methylimidazole (2-MIM) was added to the solution while it was being sonicated in order to achieve a homogeneous solution. Following this, the mixed solution was transferred into a 20 mL vial and a Zr/Sm–HT photoanode was mounted in it. The vial was then maintained at 60 °C for 9 h. Finally, the obtained NiCo-MOF-coated Zr/Sm–HT photoanode (Zr/Sm–HT:NiCo-MOF) was rinsed with ethanol and DI water and then air-dried.

Results and discussion

Structural and PEC characteristics of Zr/Sm co-doped-HT photoanodes

Scheme 1 depicts the detailed synthesis process of the co-doped and NiCo-MOF-coated HT photoanode. Initially, in situ Zr-doped FeOOH was fabricated via a hydrothermal method, followed by ex situ co-doping with Sm. A high-temperature heat treatment was conducted to transform it into a Zr/Sm co-doped Fe₂O₃ photoanode, subsequently followed by a hydrothermal reaction of NiCo-MOF. X-ray diffraction (XRD) patterns were obtained to investigate the crystalline structure of the various photoanodes. As can be seen from Fig. 1a, hematite and FTO-related peaks were observed. Noticeably, no other crystalline impurities were detected, indicating that the dopants were incorporated in the lattice of the hematite photoanode. Further on, Fig. S1† illustrates that the hematite peak intensities of the doped samples were lower than those of the Bare-HT. This can be attributed to lattice distortion, contraction of interplanar spacings, increased defects and crystallization disruption.^33,34 An XPS analysis was performed to assess the elemental compositions and valence states. Fe 2p spectra depicted in Fig. 1b can be split into two major peaks with binding energies (BEs) of 724.1 and 710.9 eV, which correspond to Fe 2p_1/2 and 2p_3/2 spin–orbit split peaks of Fe₂O₃, respectively.⁶ The satellite peaks at 718.9 and 733.4 eV for the Fe³⁺ species were consistent across all samples. The deconvoluted O 1s spectra showed two distinct peaks positioned at 529.9 and 531.5 eV, which can be attributed to lattice oxygen in the hematite and surface-adsorbed hydroxyl groups, respectively (Fig. 1c).³⁵ The XPS spectra of Zr 3d displayed BEs of 181.8 and 184.2 eV, representing Zr 3d_3/2 and Zr 3d_1/2, respectively (Fig. S2a†), arising from typical Zr⁴⁺ values and indicating that Zr had been doped successfully in all the studied photoanodes.³⁶ The absence of Sm peaks in the XPS spectra could be due to the effective diffusion of Sm into the bulk and doping into the hematite lattice during the high-temperature quenching (Fig. S2b†). A similar phenomenon was observed in a previous study for microwave-assisted Pt doping.³⁷ The peak observed at a BE of approximately 486.2 eV belonged to Sn 3d_5/2, revealing that the Sn⁴⁺ ions have successfully permeated the Fe₂O₃ lattice via diffusion from the FTO substrate during the high-temperature quenching (Fig. S2c†).

The local structure was investigated by XANES and EXAFS analyses. As can be seen in Fig. 1d, Sm–HT exhibited almost the same intensity in the absorption edge as Bare-HT and it is distinguishably lower than those of Zr–HT and Zr/Sm–HT. The doping of Sm with an identical oxidation number did not cause any particular oxidation change in the HT lattice, but co-doping with a Zr atom made the electronic state of Fe equivalent to that of Zr only-doped hematite. In the Fourier-transformed EXAFS spectra (Fig. 1e), Zr/Sm–HT showed a decrease in intensity at 1.6–2.0 Å, implying that Fe–O bonds associated with the octahedron (Oh) in the HT lattice became more disordered with the addition of Sm and Zr. This proves that doped Zr/Sm atoms could be securely accommodated in a hematite lattice. In addition, Fig. S3† illustrates the HR FE-SEM images of Sm–HT and Zr–HT. Moreover, the HR FE-SEM image disclosed the nanorod morphology of the Zr/Sm–HT sample (Fig. 1f). Furthermore, Zr-doped samples showed a higher absorbance, indicating the effective generation of photoinduced charge carriers (Fig. S4a†). These samples demonstrated a bandgap of 2.04 eV, consistent with the typical bandgap of HT (Fig. S4b†).

In order to examine the impact of co-doping on hematite bulk properties, we assessed the PEC characteristics of the as-fabricated Bare-HT, Sm–HT, Zr–HT and Zr/Sm–HT photoanodes. As shown by photocurrent density–potential (J–V) curves in Fig. 2a, the Zr–HT photoanode had a substantial increase in PD at 1.23 V_RHE. The Zr–HT photoanode exhibited a shift in the anodic onset potential due to the presence of surface-trapping states. This behavior has been commonly reported for the hematite photoanodes doped with tetravalent dopants.¹⁰ Conversely, the Sm–HT photoanode exhibited inferior performance as compared to Bare-HT. It is possible that the isovalent substitution of Sm³⁺ for Fe³⁺ is the cause of the reduced photocurrent in the Sm–HT photoanode, as it did not enhance the n-type semiconductor properties. A similar phenomenon was found in a prior investigation.³⁷ However, when Sm was co-doped with Zr, it resulted in a significant increase in the photoelectrochemical performance. Specifically, the PD reached 1.99 mA cm⁻² at 1.23 V_RHE, which is 2.05-fold greater than that of the Bare-HT. This could be attributed to the Sm co-dopant, which reduced the number of bulk charge trapping states by modifying the electronic properties. As a result, the bulk electrical conductivity was increased through the use of co-dopants. Nevertheless, there is still a significant occurrence of charge recombination on the surface. The optimizations of Sm co-dopant concentration and duration of microwave attachment are depicted in Fig. S5.† Moreover, the EIS data (Fig. 2b) clearly demonstrated that Sm–HT had the largest radius, suggesting significant charge recombination at both bulk and the surface (Table S1†). One possible explanation for this phenomenon is that the process of Sm doping led to the development of electronic trapping energy levels in the bulk, which may then result in bulk recombination. In addition, Zr–HT displayed a smaller radius in comparison to Bare-HT, implying that the Zr-dopant decreased the bulk charge transfer resistance (R₁), while the Sm co-dopant further lowered the charge transfer resistance, indicating that Sm has an influence on bulk characteristics.

A Mott–Schottky (MS) analysis was performed for all-fabricated photoanodes under dark conditions (Fig. 2c). It is apparent that all the photoanodes exhibited positive slopes in the full potential region, confirming the n-type semiconductor properties of the HT photoanode. Due to the fact that Sm³⁺ does not produce Fe²⁺ ions in the lattice as a result of the isovalent substitution,³⁸ as attested by the EXAFS analysis, there is no enhancement observed in the donor density (N_D) when compared to that of the Bare-HT photoanode. On the other hand, the Zr dopant played the role of an additional electron donor, resulting in the introduction of additional electrons near the Fe³⁺ site, which led to the formation of Fe²⁺ sites owing to the replacement of Fe³⁺ by Zr⁴⁺ in the HT lattice.³⁹ As a consequence, the N_D of the Zr–HT sample was increased by 50%. Further enhancement in the N_D was seen in the Zr/Sm–HT photoanode, which provided evidence that the Sm was co-doped into the HT lattice. Additionally, bulk charge separation efficiencies (η_bulk) were estimated using an electrolyte consisting of 1 M NaOH and 0.5 M H₂O₂ (Fig. S6†). As depicted in Fig. 2d, it is evident that Sm–HT displayed a lower η_bulk compared to Bare-HT, as expected. Whereas, the Zr dopant increased η_bulk to 25.1% and Sm co-dopant further enhanced it to 27.3% at 1.23 V_RHE, which is 64.5% higher than that of Bare-HT, vividly demonstrating that the Zr/Sm co-dopants considerably improved bulk charge separation, which in turn augmented bulk electrical conductivity. In contrast, Zr/Sm–HT did not show significant enhancement in the surface charge separation efficiency (η_surface) (Fig. S7†). Moreover, we investigated the impact of ex situ Sm-co-doping via synthesizing in situ Sm co-doped Fe₂O₃ photoanodes (Fig. S8†).

Structural and morphological analysis of the Zr/Sm–HT:NiCo-MOF photoanode

In order to mitigate the recombination of surface charges, we applied a NiCo-MOF cocatalyst on the Zr/Sm–HT photoanode and assessed its effect. The optimal deposition period for the NiCo-MOF was determined to be 9 h, as shown in Fig. S9.† Furthermore, Fig. S10† illustrates the optimization of the PEC activity of the Zr/Sm–HT:NiCo-MOF photoanode with varying concentrations of Ni and Co. The crystal and morphology did not undergo any significant changes following the addition of NiCo-MOF, as indicated by XRD and HR FE-SEM images (Fig. S11†). A transmission electron microscopy (TEM) examination was carried out in order to gain a better understanding of the morphology and microstructure of the Zr/Sm–HT:NiCo-MOF photoanode (Fig. 3a,b and S12a†). The HR-TEM images demonstrated that NiCo-MOF nanolayers, with a thickness of around 5–10 nm, were evenly dispersed and firmly attached to hematite through discrete interfaces, representing the successful establishment of the Zr/Sm–HT and NiCo-MOF heterojunction (Fig. 3c, d and S12b†). It was possible to identify the different lattice fringes with d-spacings of 0.227 and 0.271 nm, which coincided with the (006) and (104) crystal planes of HT, respectively. Despite this, there was no crystal lattice of NiCo-MOF observed because of the low crystallinity. The energy dispersive X-ray spectroscopy (EDS) elemental mapping images provide additional evidence that the coating of NiCo-MOF on the surface of the HT was successful, with a uniform distribution. This was demonstrated by the presence of elements such as Fe, O, Ni, Co and N that were detected (Fig. 3e and S13†). In addition, the presence of Zr and Sm was also identified, affording confirmation of their incorporation into the bulk of the HT. Further on, XPS spectra verified the presence of Ni and Co (Fig. S14†).

PEC water splitting

The PEC performances of as-prepared photoanodes were assessed in 1 M NaOH electrolyte under AM 1.5 G illumination. As can be seen from Fig. 4a, the Zr/Sm–HT:NiCo-MOF photoanode displayed an improved PD of 2.36 mA cm⁻² at 1.23 V_RHE, which is 143% greater than that of Bare-HT, and it exhibited cathodic onset shifting (Fig. S15a†). This shifting of the cathodic onset indicates that NiCo-MOF is a remarkable cocatalyst for water oxidation, which in turn reduces the energy barrier for water oxidation. A further explanation for this result can be observed in the dark current, which specifies that NiCo-MOF possessed a greater capability for water oxidation in comparison to other photoanodes (Fig. S15b†). Besides, there is a noticeable oxidation peak observed at 1.2 V_RHE for the Zr/Sm–HT:NiCo-MOF photoanode, signifying the oxidation of Co²⁺ to Co³⁺.⁴⁰ It is obvious that the Zr/Sm–HT:NiCo-MOF photoanode achieved the highest ABPE value of 0.247%, which is 154 and 72% higher than that of Bare-HT and Zr/Sm–HT photoanodes, respectively (Fig. S16†). These findings suggest that NiCo-MOF has the potential to significantly enhance the water oxidation performance of Zr/Sm–HT. Additionally, the electrochemical active surface area (ECSA) can be obtained by analyzing the cyclic voltammetry curves in the non-Faradaic region (Fig. S17†). As depicted in Fig. 4b, the apparent highest double layer capacitance (C_dl) value for Zr/Sm–HT:NiCo-MOF demonstrates that hole extraction and transfer were caused by the presence of the NiCo-MOF cocatalyst.

In order to verify the occurrence of a p–n heterojunction in the Zr/Sm–HT:NiCo-MOF photoanode, the MS plots were performed. As illustrated in Fig. 4c, the MS plot of Zr/Sm–HT:NiCo-MOF unveiled an inverted “V-shaped” pattern. This pattern featured clear positive and negative slopes in certain potential ranges, which corresponded to the characteristics of n-type Fe₂O₃ and p-type NiCo-MOF (Fig. S18†), respectively. The results provided substantial validation of the successful generation of a p–n NiCo-MOF@Zr/Sm–Fe₂O₃ heterojunction.¹⁷ In addition, Zr/Sm–HT:NiCo-MOF demonstrated a negative shift in flat band potentials (E_FB) when contrasted with Zr/Sm–HT, as deduced from the MS plots extrapolated to the x-axis. This implies that the modification of NiCo-MOF resulted in a reduction of the FLP effect that was triggered by surface-trapping states.⁴¹ Additionally, the heterostructure photoanode displayed an increase in donor density in comparison with the Zr/Sm–HT photoelectrode, signifying that the surface decoration of MOFs has the potential to increase the conductivity of Zr/Sm–HT photoanodes by facilitating an effective internal built-in electric field, which could enable efficient bulk charge separation. Furthermore, incident photon-to-current efficiency (IPCE) measurements were conducted at 1.23 V_RHE, as depicted in Fig. 4d. It is manifest that Zr/Sm–HT was able to attain an IPCE value of 17.5% at 400 nm, which is nearly 84% higher than that of Bare-HT. This might be attributable to the fact that co-doping considerably improved the efficiency of photon conversion. In addition, the heterostructure photoelectrode reached the highest IPCE of 21.5% (at 400 nm), 2.26-fold higher than that of Bare-HT.

To elucidate charge carrier recombination characteristics and dynamics in photoexcited states, we performed steady-state and time-resolved photoluminescence (PL) spectroscopy. The observed PL spectrum showed a range of emissions, including the intense PL near the absorption edge (∼600 nm) of hematite and additional extended emissions beyond 650 nm (Fig. 4e). These absorption band-edge emissions resulted from the direct transition between the valence and conduction bands of hematite, and the Stokes-shifted extended emissions originated from trapped electronic states due to internal crystal and surface defects.^42,43 A notable point was that the Stokes-shifted emissions from the trapped states were distinctively suppressed by introducing Zr/Sm and further with the NiCo-MOF, whereas the intensities of the band-edge emissions were enhanced. The detected PL decays showed typical multiple-route recombination pathways in the bare and doped hematite materials, consisting of at least three lifetime components (Fig. 4f and Table S2†). In principle, the PL recombination rate (k_PL) is the sum of radiative (k_r) and nonradiative (k_nr) recombination rates; k_PL = k_r + k_nr.⁴⁴ In this study, the average PL lifetime (〈τ〉) was quite decreased (increased PL recombination rate because of k_PL = 1/τ_PL) in the composited hematite materials with Zr/Sm and the NiCo-MOF in the following order: Bare-HT (0.31 ns) > Zr/Sm–HT (0.20 ns) > Zr/Sm–HT:NiCo-MOF (0.15 ns). From the PL lifetime images, it is clearly confirmed that the localized colour centers with a longer lifetime were largely suppressed with Zr/Sm, and further quite disappeared with the NiCo-MOF (Fig. S19a–c†). The Zr/Sm–HT:NiCo-MOF with the best PEC performance had the shortest PL lifetime, nonradiative charge transfer processes might be effectively worked in this composited photoanode.^42,45 It is noteworthy that the co-doping resulted in an extension of electron lifetime (τ_el) from 1.24 ms (Bare-HT) to 1.82 ms, as evidenced by the Bode plot (Fig. 4g). Moreover, the heterostructure composite further enhanced the τ_el (2.75 ms), substantiating the heterojunction intensified bulk charge separation and transfer by generating an intrinsic electric field. The resistance profiles of the as-fabricated samples at diverse frequencies are depicted in Fig. S20.† The Zr/Sm–HT:NiCo-MOF photoelectrode exhibited the lowest resistance, indicating efficient and fast charge transfer. Fig. 4h depicts the comparison studies of PD of the Zr/Sm–HT:NiCo-MOF photoanode with other MOF-decorated Fe₂O₃ and Fe₂O₃-based heterojunction photoanodes at 1.23 V_RHE, as stated in the literature (Tables S3 and S4†).

OER kinetics

The charge separation efficiencies were evaluated based on PD using a hole scavenger electrolyte (Fig. S21†). As shown in Fig. S22,† the Zr/Sm–HT:NiCo-MOF photoanode achieved an enhanced η_bulk of 29.7% at 1.23 V_RHE. This result explicitly corroborates that the bulk charge carrier separation was facilitated by a generated inherent built-in electric field at the p–n heterojunction interface and thereby accelerated the transport of photogenerated electrons and holes in opposing directions. Besides, η_surface substantially rose to 95.3% at 1.23 V_RHE for the Zr/Sm–HT:NiCo-MOF photoanode (Fig. 5a), indubitably demonstrating that the NiCo-MOF passivated surface-trapping states, thus impeding surface charge recombination. These surface-trapping states are formed by dangling bonds on the surface of the photoanode, which induce the FLP effect and are responsible for the surface charge recombination.^17,46 To investigate the band bending and passivation of surface-trapping states, the open-circuit potential (OCP) and EIS were utilized. Under illumination, the Fermi level splits into the quasi-Fermi levels of holes and electrons (E_Fp and E_Fn, respectively). The splitting of the Fermi level causes an open-circuit photovoltage (V_ph) to be generated, which is equal to the difference between the two quasi-Fermi levels.⁴⁷ Moreover, a photo-generated interfacial band bending effect can be indicated by V_ph.⁴⁸ As illustrated in Fig. 5b, the fact that the Zr/Sm–HT photoanode did not show significant changes in the V_ph value when compared to the Bare-HT demonstrates that band bending was impeded due to an intense Fermi pegging effect. This Fermi level pinning triggered charge recombination at recombination surface states (r-SS). However, the coating of NiCo-MOF resulted in a substantial 128% increase in V_ph relative to Zr/Sm–HT, indisputably denoting that the NiCo-MOF cocatalyst enlarged band bending and hastened the charge transfer at the photoanode/electrolyte interface. Besides, this intensified V_ph disclosed the presence of an intrinsic built-in electric field that promoted the driving force for bulk charge separation and boosted hole migration from the Zr/Sm–HT to NiCo-MOF. Further on, an anodic OCP dark shifting for the Zr/Sm–HT:NiCo-MOF photoanode indicates that the NiCo-MOF passivated r-SS, leading to the Fermi level de-pinning effect. As a consequence, the potential drop in the Helmholtz layer lessened and hence expedited hole transfer via intermediate surface states (i-SS) for the effective OER.^10,43

In a simplified PEC system, it is generally perceived that modulating the surface state of the photoanode can exert considerable impacts on the augmentation of the water oxidation and the suppression of the recombination of photogenerated carriers. Therefore, photoelectrochemical impedance spectroscopy was utilized to scrutinize the kinetics of the charge transfer process via i-SS and surface carrier recombination at surface-trapping states/r-SS. Fig. S23† displays the EIS responses of as-fabricated photoanodes tested at applied potentials ranging from 0.85 to 1.4 V_RHE. The Nyquist plots were fitted using the equivalent circuit depicted in Fig. S23a.† The series resistances (R_s) associated with the electrolyte and external circuit did not exhibit any considerable variations among the samples (Fig. S24a†), representing that the series resistance had a limited impact on the charge transfer process and the PEC behavior. Fig. S24b† demonstrates that the Zr/Sm–HT photoanode displayed a lower bulk charge transfer resistance (R₁) compared to the Bare-HT photoanode across the entire measured potential, signifying that the co-doping of Zr/Sm effectively boosted the bulk conductivity of the photoanode. Notwithstanding this, the R₁ values of the Zr/Sm–HT:NiCo-MOF photoanode decreased above 1.05 V_RHE. This could be ascribed to the heterojunction formation. Intriguingly, the charge transfer resistance values at the photoanode/electrolyte interface (R₂) were found to be the lowest for Zr/Sm–HT:NiCo-MOF, as shown in Fig. 5c. This unambiguously demonstrates that the photoinduced charge carriers had scarcer possibilities for recombination at surface recombination sites than Bare-HT. Specifically, the R₂ value of the Zr/Sm–HT:NiCo-MOF photoanode unveiled a reduction of 91.1 and 69.5% when compared to that of Bare-HT and Zr/Sm–HT photoanodes, respectively, at a water oxidation potential of 1.23 V_RHE. Further on, we computed an effective capacitance (CPE_eff) of the photoanode according to the following equation.⁴⁹ The CPE component was fit with CPE-P and CPE-T values.⁵⁰

	(1)

where α and Q denote CPE-P and CPE-T, respectively. The corresponding CPE-P and CPE-T values are illustrated in Fig. S25.† Remarkably, the CPE_1(eff) of Zr/Sm–HT:NiCo-MOF was notably superior to those of both photoanodes at potentials greater than 0.9 V_RHE (Fig. S26a†). The enhanced CPE_1(eff) specifies the amplified concentration of photogenerated carriers in the bulk owing to the effective charge separation as a result of the heterojunction.⁵¹ Besides, NiCo-MOF decorated samples exhibited markedly greater values of CPE_2(eff) compared to the other photoanodes (Fig. S26b†). Effective PEC characteristics of the Zr/Sm–HT:NiCo-MOF photoanode were accounted for by a higher CPE_2(eff) and lower R₂.⁵² In order to emphasize the significance of surface states in the charge transfer process at the SEI, we derived the density of surface states (DOSS; N_SS = CPE_2(eff)/q) based on CPE_2(eff) values. As displayed in Fig. S27,† the highest DOSS for the Zr/Sm–HT:NiCo-MOF photoanode, suggests its efficient extraction of holes for water oxidation, particularly at a higher oxidative potential.

The surface charge recombination (k_rec) and transfer rate constants (k_ct) were estimated in accordance with the literature.⁵³ As illustrated in Fig. 5d, it is noteworthy that the k_rec of Bare-HT has been shown to be lower than that of Zr/Sm–HT below 1.1 V_RHE. This could be ascribed to the fact that the tetravalent dopants are responsible for the severe effect of surface-trapping states. Conversely, the k_rec of the Zr/Sm–HT:NiCo-MOF photoanode was the lowest throughout the whole potential region, denoting that the FLP effect can be mitigated by the presence of the NiCo-MOF cocatalyst through passivation of surface defects.⁵⁴ To be more specific, Zr/Sm–HT:NiCo-MOF showed a 70.5% decrease in k_rec relative to Bare-HT (8.8 to 2.6 s⁻¹) at 1.23 V_RHE. Additionally, it should also be noted that the NiCo-MOF cocatalyst coating had the highest k_ct after a potential of >1.10 V_RHE (Fig. 5e). Zr/Sm co-doping increased the k_ct from 15.6 to 18.3 s⁻¹, and the value further ameliorated to 28.8 s⁻¹ for the Zr/Sm–HT:NiCo-MOF photoelectrode at 1.23 V_RHE, explicitly indicating that the NiCo-MOF cocatalyst adjusted the i-SS to expedite hole transfer for potent water oxidation. The declined k_rec and increased k_ct values observed for the Zr/Sm–HT:NiCo-MOF photoanode demonstrate accelerated charge separation and transfer dynamics. Further on, it is evident from Fig. 5f that the Zr/Sm–HT:NiCo-MOF photoanode attained a charge transfer efficiency (η_CT) of 91.6% at water oxidation potential, which is 43.1% advanced than that of the Bare-HT. Hence, it has been proven that the exceptional η_CT of Zr/Sm–HT:NiCo-MOF can be attributed to the effective bulk charge transfer, an increase in surface reaction kinetics and the inhibition of surface charge recombination through the combined effect of a heterojunction, a cocatalyst and passivation.

The recombination of the accumulated holes in the space-charge region (SCR) is a significant obstacle and a crucial scientific issue in the process of PEC-WS. Transient photocurrent measurements were undertaken to explore the charge recombination rate in the SCR and assess the transient time constant (τ_t). It is reported that the transient anodic photocurrent peaks upon exposure to light, derived from the process where photogenerated holes, which reached the interface or surface of the photoanode, would recombine with the electrons when the photogenerated holes cannot be quickly injected into the electrolyte.^55,56 Thus, these positive photocurrent spikes are associated with the accumulation of holes in the electrode space-charge layer under illumination.⁵⁷ It is obvious that the Zr/Sm–HT:NiCo-MOF photoanode displayed a remarkable reduction in the accumulation of holes (Fig. 5g), which was indicated by the attenuation ratio of PD,¹⁰ affording irrefutable evidence that the NiCo-MOF cocatalyst can hasten hole transfer at the SEI via effective extraction of holes on the surface of the photoanode. The τ_t can be considered a significant metric for indicating the lifespan of the charge carrier (hole).⁵⁸ As can be observed from Fig. 5h, it is apparent that the τ_t value of Zr/Sm–HT:NiCo-MOF increased sharply (4.48 s), revealing that NiCo-MOF possessed the capability to impede charge recombination, hence extending the lifetime of holes. The theoretical maximum photocurrent density (J_abs) plots of the as-synthesized Bare-HT, Zr/Sm–HT and Zr/Sm–HT:NiCo-MOF photoanodes are illustrated in Fig. S28.† The PEC-WS and stability experiment of the Zr/Sm–HT:NiCo-MOF photoanode were conducted at 1.23 V_RHE under 1 sun illumination. As observed from Fig. 5i, the evolved H₂ and O₂ gases were 425.6 and 208.4 μmol, respectively, following 10 h of illumination. It was discovered that the molar ratio of the released H₂ and O₂ was almost exactly 2:1, demonstrating that H₂ and O₂ were generated simultaneously from PEC-WS. The Faradaic efficiencies of the evolved gases surpassed 95% (Fig. S29†). Besides, the Zr/Sm–HT:NiCo-MOF photoelectrode retained 99% of its initial stability after being exposed to continuous irradiation for a period of 10 h.

Mechanisms of charge transfer dynamics

Through investigations of the energy band positions of Zr/Sm–HT and NiCo-MOF, we suggest a potential mechanism for the PEC water oxidation of Bare-HT, Zr/Sm–HT and Zr/Sm–HT:NiCo-MOF photoanodes. Typically, the conduction band (CB) of the n-type semiconductor is 0.1 V more negative than E_FB, while the valence band (VB) of the p-type semiconductor is 0.1 V more positive than E_FB.⁵⁹ The bandgap of NiCo-MOF is 2.24 eV (Fig. S30†). At the outset, electrons and holes are typically produced in response to illumination. Bare-HT exhibited severe bulk/surface charge recombination as a result of its abundant recombination sites both in the bulk and at the surface (Scheme 2a). Besides, the presence of a large number of accumulated holes in the SCR leads to increased recombination, thereby decreasing the rate of charge carrier transfer. Also, the presence of several recombination surface states caused the FLP effect, which in turn reduced the V_ph value, therefore preventing the quasi-Fermi level of holes from achieving water oxidation potential. The PEC efficacy of Bare-HT was subpar as a result of all of these factors. Zr/Sm–HT improved bulk charge separation efficiency by reducing bulk charge transfer resistance, thus augmenting electron lifetime (Scheme 2b). Despite this, the FLP effect continues to exist because of the r-SS. Moreover, the recombination of surface charges and the accumulation of holes in the SCR impeded the efficient transmission of holes for water oxidation. In contrast, the Zr/Sm–HT:NiCo-MOF heterostructure photoanode enabled the transfer of electrons from the CB of NiCo-MOF to the CB of Zr/Sm–HT, while the VB holes of Zr/Sm–HT are promptly transferred to the VB of NiCo-MOF within the Zr/Sm–HT:NiCo-MOF heterojunction under a strong built-in electric field (Scheme 2c). This resulted in a more efficient charge separation and transfer, which in turn increased the electron lifetime. In addition, passivation of r-SS resulted in a Fermi level de-pinning, hence remarkably suppressing the surface charge recombination. Furthermore, the NiCo-MOF cocatalyst was able to expedite hole transfer from the SCR to the electrolyte through i-SS by prolonging the hole lifespan. Additionally, the holes can be trapped by low valence Co²⁺/Ni²⁺ in NiCo-MOF and subsequently oxidized to high valence Co³⁺/Co⁴⁺ and Ni³⁺. Then high valence metal ions served as active sites to oxidize water, which resulted in the production of O₂ and eventual return to their initial states. Each of these functionalities of NiCo-MOF contributed to the enhanced kinetics of surface charge transfer for effective water oxidation.

Conclusions

The purpose of this work was to explore the effect of Zr/Sm co-doping and bimetallic NiCo-MOF coating on the PEC properties of a hematite photoanode. The EIS and MS analyses revealed that the Zr/Sm co-dopants augmented the bulk characteristics of the Fe₂O₃ photoanode. Additionally, the findings of this study demonstrated that the NiCo-MOF served numerous purposes. Firstly, it could establish a p–n heterojunction with Zr/Sm–Fe₂O₃, which enables effective separation and transport of charges within the bulk by means of an internal electric field. Secondly, it successfully passivated surface-trapping states, resulting in the Fermi level de-pinning effect, which in turn reduced surface charge recombination. Lastly, it acted as a cocatalyst, enhancing hole transfer by prolonging hole lifespan and consequently accelerating the kinetics of water oxidation. Thereby, the optimized Zr/Sm–HT:NiCo-MOF photoanode, which was tuned, showed a boosted PD of 2.36 mA cm⁻² at 1.23 V_RHE, which is 2.43 times higher than that of Bare-HT. Besides, it attained the maximum IPCE of 21.5% at 1.23 V_RHE at 400 nm. The combined implementation of co-doping and bimetallic multifunctional MOF decoration affords a novel approach to steer the dynamics of charge carriers in PEC-based photocatalysis.

Data availability

The data supporting this article have been included as part of the ESI.†

Author contributions

P. A.: conceptualization, methodology, formal analysis, investigation, data curation, validation, writing – original draft, writing – review & editing, validation, visualization. W. S. C.: formal analysis, investigation, data curation. H. H.: formal analysis, data curation. T. W. K.: formal analysis, data curation. S. H. C.: formal analysis, investigation, data curation, funding acquisition. R. D.: formal analysis, funding acquisition, project administration. B. K.: validation, visualization, funding acquisition, project administration. J. S. J.: conceptualization, methodology, validation, formal analysis, resources, writing – review & editing, visualization, supervision, funding acquisition, project administration.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This research was supported by the GRDC (Global Research Development Center) Cooperative Hub Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science and ICT (MSIT) (RS-2023-00258911). Also, this research was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean Government (MSIT) (NRF-2023R1A2C1003088 and RS-2025-00518521).

References

1. D. Gogoi, R. S. Karmur, M. R. Das and N. N. Ghosh, J. Mater. Chem. A, 2023, 11, 23867–23880 RSC .
2. X. S. Xing, X. Ren, X. Zeng, A. Li, Y. Wang, Z. Zhou, Y. Guo, S. Wu and J. Du, Sol. RRL, 2023, 7, 2201041 CrossRef CAS .
3. J. C. N. Fouemina, G. Li, X. She, D. Yan, X. Lv, K. Nie, J. Deng and H. Xu, Small, 2023, 19, 2301114 CrossRef CAS PubMed .
4. J. Kang, K. Y. Yoon, J. E. Lee, J. Park, S. Chaule and J. H. Jang, Nano Energy, 2023, 107, 108090 CrossRef CAS .
5. B. R. Lima, I. Rodriguez-Gutierrez, N. C. Verissimo, Â. Albuquerque, G. T. Santos, J. Bettini and F. L. Souza, Mater. Today Chem., 2023, 34, 101784 CrossRef CAS .
6. W. Li, H. Guo, C. Xu, C. Tang, J. S. Lee and H. Zhang, Appl. Catal., B, 2024, 342, 123465 CrossRef CAS .
7. K. Zhang, Q. Wu, K. Ba, Q. Qiu, Y. Yang, Y. Lin, D. Wang and T. Xie, J. Colloid Interface Sci., 2023, 650, 1022–1031 CrossRef CAS PubMed .
8. J. Rong, Z. Wang, J. Lv, M. Fan, R. Chong and Z. Chang, Chin. J. Catal., 2021, 42, 1999–2009 CrossRef CAS .
9. X. Guo, L. Wang and Y. Tan, Nano Energy, 2015, 16, 320–328 CrossRef CAS .
10. P. Anushkkaran, M. A. Mahadik, W. S. Chae, H. H. Lee, S. H. Choi and J. S. Jang, Chem. Eng. J., 2023, 472, 144998 CrossRef CAS .
11. T. Jiao, C. Lu, K. Feng, J. Deng, D. Long and J. Zhong, J. Colloid Interface Sci., 2021, 585, 660–667 CrossRef CAS PubMed .
12. Q. Zhu, C. Yu and X. Zhang, J. Energy Chem., 2019, 35, 30–36 CrossRef .
13. S. Shao, Y. Xiao, J. Yang, X. Lv, K. Feng, Y. Xia, D. Zhang, H. Xu, J. Zhong and J. Deng, Chem. Eng. J., 2021, 413, 127416 CrossRef CAS .
14. K. Y. Yoon, J. Park, H. Lee, J. H. Seo, M. J. Kwak, J. H. Lee and J. H. Jang, ACS Catal., 2022, 12, 5112–5122 CrossRef CAS .
15. J. Bai, R. T. Gao, X. Guo, J. He, X. Liu, X. Zhang and L. Wang, Chem. Eng. J., 2022, 448, 137602 CrossRef CAS .
16. H. Chai, L. Gao and J. Jin, ChemSusChem, 2022, 15, e202201030 CrossRef CAS PubMed .
17. H. Chai, R. Li, L. Gao, H. Niu and J. Jin, Mater. Today Chem., 2023, 30, 101498 CrossRef CAS .
18. S. Swathi, R. Yuvakkumar, G. Ravi, A. G. Al-Sehemi and D. Velauthapillai, Nanoscale Adv., 2022, 4, 2501–2508 RSC .
19. U. Prasad, J. Prakash and A. M. Kannan, Sustain. Energy Fuels, 2020, 4, 1496–1506 RSC .
20. V. S. Katta, M. Velpandian, S. Challapalli, P. Meduri and S. S. K. Raavi, Sustain. Energy Fuels, 2022, 6, 5539–5556 RSC .
21. H. Geng, S. Huang, D. Kong, E. Chubenko, V. Bondarenko, P. Ying, Y. Sui, Y. Zhao and X. Gu, J. Alloys Compd., 2022, 925, 166667 CrossRef CAS .
22. S. Li, Y. Gao, N. Li, L. Ge, X. Bu and P. Feng, Energy Environ. Sci., 2021, 14, 1897–1927 RSC .
23. B. Zhang, Y. Zheng, T. Ma, C. Yang, Y. Peng, Z. Zhou, M. Zhou, S. Li, Y. Wang and C. Cheng, Adv. Mater., 2021, 33, 2006042 CrossRef CAS PubMed .
24. F. Wu, J. Xie, Y. You, Z. Zhao, L. Wang, X. Chen, P. Yang and Y. Huang, ACS Appl. Energy Mater., 2020, 3, 4867–4876 CrossRef CAS .
25. F. Xiao, R. Guo, X. He, H. Chen, W. Fang, W. Li, H. Wang, Z. Sun, P. Tian and L. Zhao, Int. J. Hydrogen Energy, 2021, 46, 7954–7963 CrossRef CAS .
26. N. K. Shrestha, S. A. Patil, S. Cho, Y. Jo, H. Kim and H. Im, J. Mater. Chem. A, 2020, 8, 24408–24418 RSC .
27. S. Zhou, K. Chen, J. Huang, L. Wang, M. Zhang, B. Bai, H. Liu and Q. Wang, Appl. Catal., B, 2020, 266, 118513 CrossRef CAS .
28. Y. Li, Q. Wang, X. Hu, Y. Meng, H. She, L. Wang, J. Huang and G. Zhu, Chem. Eng. J., 2022, 433, 133592 CrossRef CAS .
29. P. Wang, S. Wang, L. Gao, X. Long, H. Chai, F. Li, Q. Wang and J. Jin, J. Catal., 2022, 413, 398–406 CrossRef CAS .
30. X. Li, Z. Wang and L. Wang, Small Sci., 2021, 1, 2000074 CrossRef CAS PubMed .
31. Z. Y. Wang, H. M. Li, S. S. Yi, M. Z. You, H. J. Jing, X. Z. Yue, Z. T. Zhang and D. L. Chen, Appl. Catal., B, 2021, 297, 120406 CrossRef CAS .
32. T. Yang, Z. W. Chen, X. Z. Yue, Q. C. Liu, S. S. Yi and Y. F. Zhu, Adv. Funct. Mater., 2024, 34, 2313767 CrossRef CAS .
33. R. Dhahri, M. Benamara, S. Bouzidi, S. Ben Moussa, A. Y. A. Alzahrani, K. I. Nassar, N. Zahmouli, E. B. Elkenany and A. M. Al-Syadi, J. Sol-Gel Sci. Technol., 2025, 113, 225–242 CrossRef CAS .
34. C. Peng, R. Lu and A. Yu, RSC Adv., 2023, 13, 9168–9179 RSC .
35. Q. Zhao, Y. Li, C. Xu, C. Li, G. Lu, Q. Bu and Q. Liu, Chem. Eng. J., 2024, 479, 147625 CrossRef CAS .
36. T. Wang, L. Gao, P. Wang, X. Long, H. Chai, F. Li and J. Jin, J. Colloid Interface Sci., 2022, 624, 60–69 CrossRef CAS PubMed .
37. P. Anushkkaran, M. A. Mahadik, W. S. Chae, H. H. Lee, S. H. Choi and J. S. Jang, Appl. Surf. Sci., 2023, 623, 157035 CrossRef CAS .
38. A. Tofanello, S. Shen, F. L. de Souza and L. Vayssieres, APL Mater., 2020, 8, 040905 CrossRef CAS .
39. H. Zhang, D. Li, W. J. Byun, X. Wang, T. J. Shin, H. Y. Jeong, H. Han, C. Li and J. S. Lee, Nat. Commun., 2020, 11, 4622 CrossRef PubMed .
40. P. Anushkkaran, T. S. Koh, W. S. Chae, H. H. Lee, S. H. Choi and J. S. Jang, ACS Sustain. Chem. Eng., 2023, 11, 5895–5907 CrossRef CAS .
41. H. Chai, S. Wang, X. Wang, J. Ma and J. Jin, ACS Catal., 2022, 12, 3700–3709 CrossRef CAS .
42. P. Anushkkaran, M. A. Mahadik, W. S. Chae, H. H. Lee, S. H. Choi and J. S. Jang, J. Mater. Chem. A, 2023, 11, 26105–26114 RSC .
43. P. Anushkkaran, W. S. Chae, J. Ryu, S. H. Choi and J. S. Jang, J. Mater. Chem. A, 2024, 12, 4702–4711 RSC .
44. J. R. Lakowicz, Principles of fluorescence spectroscopy, Springer US, Boston, MA, 2006 Search PubMed .
45. S. Kim, M. A. Mahadik, W. S. Chae, J. Ryu, S. H. Choi and J. S. Jang, Appl. Surf. Sci., 2020, 513, 145528 CrossRef CAS .
46. D. C. Moritz, W. Calvet, M. A. Z. Pour, A. Paszuk, T. Mayer, T. Hannappel, J. P. Hofmann and W. Jaegermann, Sol. RRL, 2023, 7, 2201063 CrossRef CAS .
47. Y. Kuang, T. Yamada and K. Domen, Joule, 2017, 1, 290–305 CrossRef CAS .
48. M. Ai, X. Li, L. Pan, X. Xu, J. Yang, J. J. Zou and X. Zhang, Chem. Eng. Sci., 2022, 250, 117397 CrossRef CAS .
49. K. Wang, Y. Liu, K. Kawashima, X. Yang, X. Yin, F. Zhan, M. Liu, X. Qiu, W. Li, C. B. Mullins and J. Li, Adv. Sci., 2020, 7, 2002563 CrossRef CAS PubMed .
50. K. J. Pyper, J. E. Yourey and B. M. Bartlett, J. Phys. Chem. C, 2013, 117, 24726–24732 CrossRef CAS .
51. H. Wang, Y. Xia, N. Wen, Z. Shu, X. Jiao and D. Chen, Appl. Catal., B, 2022, 300, 120717 CrossRef CAS .
52. R. Wang, H. Li, Z. Hao, T. Feng, Y. Li, B. Dong and L. Cao, Chem. Eng. J., 2023, 455, 140861 CrossRef CAS .
53. J. Xiao, X. Jia, B. Du, Z. Zhong, C. Li, J. Sun, Z. Nie, X. Zhang and B. Wang, J. Colloid Interface Sci., 2024, 654, 915–924 CrossRef CAS PubMed .
54. R. T. Gao, J. Zhang, T. Nakajima, J. He, X. Liu, X. Zhang, L. Wang and L. Wu, Nat. Commun., 2023, 14, 2640 CrossRef CAS PubMed .
55. P. Anushkkaran, L. K. Dhandole, W. S. Chae, H. H. Lee, S. H. Choi, J. Ryu and J. S. Jang, Int. J. Hydrogen Energy, 2022, 47, 32015–32030 CrossRef CAS .
56. Q. Bu, S. Li, K. Zhang, Y. Lin, D. Wang, X. Zou and T. Xie, ACS Sustain. Chem. Eng., 2019, 7, 10971–10978 CrossRef CAS .
57. A. G. Tamirat, A. A. Dubale, W. N. Su, H. M. Chen and B. J. Hwang, Phys. Chem. Chem. Phys., 2017, 19, 20881–20890 RSC .
58. C. Hu, C. Xu, X. Li, B. Li, X. Ma, J. Zhu, C. Dong and Y. Ding, ACS Sustain. Chem. Eng., 2023, 11, 7367–7377 CrossRef CAS .
59. H. C. Wang, H. M. Li, T. Yang, J. W. Ji, X. Z. Yue, Q. C. Liu, S. S. Yi and Y. Zhu, Adv. Funct. Mater., 2024, 34, 2406545 CrossRef CAS .

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5ta03908h

This journal is © The Royal Society of Chemistry 2025