Solar-driven chemical reactions based on photoactive materials have attracted massive attention due to the concerns with regard to the current energy and environmental issues (Baharuddin et al., 2021). The fundamental research summarized some key features that were capable of improving the photoactive materials for practical solar-driven chemical reactions. These were broad light absorption range, rapid charge separation, good stability, abundant surface area, and reaction sites. Hematite (α-Fe₂O₃) was considered as an outstanding photocatalyst due to its good ability of light response and light harvesting (absorbs ca. 30% light) (Kuang et al., 2017; Yi et al., 2021). However, its short lifetime of photogenerated charge was attributed to the limited hole diffusion distance (Shen et al., 2016). In order to improve the charge migration rate of α-Fe₂O₃, a previous study focused on constructing heterostructures with different bandgap materials for further improving photogenerated charge separation (Kanwal et al., 2021). For instance, Zhang et al. reported that CdS/α-Fe₂O₃ heterojunction showed great photocatalytic activity and had a prominent charge separation rate due to the fast diffusion of photogenerated charge between Fe₂O₃ and CdS (Zhang et al., 2013). Based on previous studies, we considered an optimization where the role of CdS was to help Fe₂O₃ with a quick charge transfer except for light absorption. Previous research demonstrated that noble metal nanoparticles possessed good chemical stability, large surface area, and fruitful reaction sites. Among the heterogeneous metallic co-catalysts, Pd-based nanoparticles have presented an outstanding performance for promoting photocatalytic reactions under mild conditions (Naina et al., 2021; Peng et al., 2021; Cui et al., 2022). This is due to the intensive electronic interactions between metal nanoparticles and semiconductive photocatalysts (Mott–Schottky effect) (Li and Antonietti, 2013). In comparison to pure Pd, the formation of the PdAg alloy was found to enhance the catalytic performance by applying electronic and geometric effects (Antolini, 2009). Furthermore, the PdAg alloy with different lattice constants of the dual metal atoms (Ag, 408.53pm and Pd, 389.07pm) could subsequently improve the absorption of visible light and reaction rate due to the d-band shifting up (Ghiabi et al., 2018).

The photocatalytic reduction of CO₂ into valuable solar fuels was regarded as an efficient means to address the energy supply and CO₂ emission (Habisreutinger et al., 2013). In addition, using semiconductors such as hematite (α-Fe₂O₃) to achieve this conversion goal had attracted extensive attention, since they were an environmentally friendly low-cost technology (Mu et al., 2021). It is well-known that the main products via photoreduction of CO₂ are CO and CH₄. In addition, controlling the chemical reaction conditions (Qiao et al., 2014) (pH, temperature, sacrificial reagent, et al.) is the common means to manipulate the selectivity, although it might be complex.

However, a dual-metallic co-catalyst could be an alternative strategy to manipulate the selectivity of CO₂ photoreduction, as recent modulating research (Li et al., 2019) demonstrated that the dual-metal sites can dissociate CO production. This is because the C and O atoms from the adsorbed CO₂ species would coordinate with two individual metal atoms, leading to a configuration of M₁-C-O-M₂ (M stands for metal, as shown in Scheme 1). The robust stability of M₁-C-O-M₂ would inhibit the breaking of M-C or M-O bonds, and thus dissociate the yield of CO. More importantly, it would conduct rapid protonation of C atoms in M₁-C-O-M₂ intermediates and provide a reaction path to yield CH₄ (Wang et al., 2020).

Of direct relevance to our work, PdAg NPs have been successfully applied to g-C₃N₄ (Liu et al., 2019), carbon tube (Benipal et al., 2017), and reduced graphene oxide (Martins et al., 2017) for photocatalytic reactions. These studies demonstrated that PdAg was capable of promoting photocatalytic reactions and focused on the role of PdAg under different reaction conditions. However, the Pd:Ag atomic ratio and its influence on the photocatalytic selectivity and activity was still unexplored. Thus, in this work, an α-Fe₂O₃/CdS Z-scheme photocatalyst was employed as a photoactive platform to support a layer of PdAg NPs. The impact of the elemental ratio of PdAg NPs on photoelectrochemical features and photoreduction selectivity over CO₂ was experimentally investigated.

The α-Fe₂O₃ film was first grown on the FTO glass substrate, which exhibited the nanorod morphology in SEM images (see supporting information in Supplementary Figure S1). In detail, the as-prepared film presented a light orange color due to the formation of FeOOH after hydrothermal reaction and then turned red after annealing under air at 500°C in a muffle furnace; the thickness of α-Fe₂O₃ was ca. 500 nm (Figure 2A) (Ahn et al., 2016; Yi et al., 2021). Subsequently, CdS was in situ synthesized on α-Fe₂O₃ using the SILAR method (Sankapal et al., 2000). The CdS with a thickness of ca. 270 nm (Figure 2B) was homogeneously coated on the α-Fe₂O₃ nanorod, and the α-Fe₂O₃/CdS heterojunction film presented a crimson color. Furthermore, PdAg alloy NPs were obtained through co-reduction of Pd and Ag precursors in an aqueous solution (Zhou et al., 2020). Then, a thin layer of PdAg 2:1 (the molar ratio between Pd and Ag is 2:1 in precursor) alloy NPs was coated above the CdS layer. The α-Fe₂O₃/CdS/PdAg on FTO was used as an electrode for photoelectrochemistry tests, and then Fe₂O₃/CdS/PdAg were scrapped from the substrate (Figure 1D) for gas chromatography tests. XRD patterns (Supplementary Figure S2A) of α-Fe₂O₃ and CdS were confirmed to be indexed to JCPDS No. 33-0664 and 41-1049, respectively. Moreover, the two components, Fe₂O₃ and CdS, were both in a hexagonal structure with slightly different lattice parameters (Zhang et al., 2013). PdAg NPs with an average particle size of 6.14 nm (calculated from 100 random particles) were found to be uniformly distributed on the α-Fe₂O₃/CdS (Supplementary Figure S2B). Furthermore, the spacing of the lattice fringe related to PdAg NPs was measured to be 2.48 nm between (111) the plane of Pd (2.245 nm) and (004) the plane of Ag (2.425 nm) (Gao et al., 2021) (Supplementary Figure S2C). The lattice fringes of α-Fe₂O₃ and CdS were recorded to be 0.251 and 0.273 nm, corresponding to the (110) and (200) planes, respectively. Meanwhile, it certified the weak diffraction peak of PdAg NPs in XRD patterns. From UV-vis diffusion reflectance spectroscopy (DRS), the light absorption of α-Fe₂O₃/CdS basically showed no differences after coupling with PdAg (Supplementary Figure S3). X-ray photoelectron spectroscopy (XPS) tested the Pd 3 d and Ag 3 d when PdAg was coupled with α-Fe₂O₃/CdS, which presented a slight shift toward higher binding energy (Supplementary Figure S4). Moreover, the pristine PdAg NPs were further qualitatively confirmed by energy-dispersive spectrum (EDS), where the atomic ratio between Pd and Ag in PdAg 2:1 was measured to be 62.1 and 37.9%, respectively (Supplementary Figure S2D). In addition, it is known that the successful connection between metal co-catalysts and photocatalysts would result in improved charge separation (Lan et al., 2021). Therefore, transient resolved photoluminescence (TRPL) was exploited to analyze the lifetime of charge after the decoration of PdAg, from which an enhanced lifetime (from 0.465 to 1.396 ns, Supplementary Figure S5) was observed, suggesting that the charge transfer was built between PdAg NPs and α-Fe₂O₃/CdS.

The photoelectrochemical properties of α-Fe₂O₃, CdS, and α-Fe₂O₃/CdS were tested as an anodic electrode in a standard PEC cell. In accordance with a previous report (Shen et al., 2020), an improvement in the maximum photocurrent density and on-set potential was observed when coupling α-Fe₂O₃ with CdS (Supplementary Figure S6). Next, to evaluate the influence of PdAg on the photoelectrochemical features, α-Fe₂O₃/CdS/PdAg was set as the cathodic electrode in the PEC cell. Among samples, the higher Pd content led to a lower on-set potential to trigger the cathodic photocurrent generation (Figure 2A), due to the higher work function of Pd (5.12) than Ag (4.26). The largest photocurrent density was observed from the sample with PdAg 1:2. This sample also presented smaller radii as revealed by EIS tests (Figures 2B,C) compared to others, suggesting lower electronic resistance in favor of charge transfer and thus enhancing the photocurrent generation. In addition, the stability test was executed by electrochemical characterization. The relative photocurrent displayed a slight decrease (ca. 0.1) after 10 min of reaction in Supplementary Figure S7, which can be attributed to the consumption of sacrificial agents and impeded the photoactivity. Combining with the XRD data shown in Supplementary Figure S8, the position of each peak presented no obvious change before and after the reaction, which certified that the catalyst possessed good stability during the photocatalytic reactions. As for the CO₂ reduction, gas chromatography (GC) confirmed the main products (CH₄ and CO) from the samples decorated with different Pd:Ag ratios. As shown in Figure 3, the gas production rates of the samples were proportional to their photocurrent density in PEC tests. The highest production rate was observed from the sample with PdAg 1:2. Meanwhile, compared to the control with pure Pd (6.62 μmol g⁻¹ h⁻¹), PdAg 1:2 (9.80 μmol g⁻¹ h⁻¹) presented a 48% enhancement in the total conversion rate. More importantly, as the modulating research predicted, by controlling the Pd:Ag ratio, the selectivity toward CH₄ was optimized from 58.7 to 83.2%.

As shown in Scheme 2, an interfacial charge region was constructed via an α-Fe₂O₃/CdS heterojunction, and a CO₂ reduction reaction occurred on the PdAg sites. In detail, the adsorbed CO₂ reacted with electrons accumulated on the PdAg sites to form CO and CH₄, respectively (Ješić et al., 2021). The advantage of this ternary structure was a synergic effect between the Z-scheme α-Fe₂O₃/CdS and co-catalyst PdAg, which helped in prompting the selectivity of CO and CH₄.

Overall, α-Fe₂O₃/CdS decorated by PdAg NPs of different Pd:Ag ratios was successfully prepared. The decoration of PdAg alloy NPs led to an enhanced photoreduction rate of CO₂, but also ended up with an improvement in the selectivity toward CH₄ production. The improved photocatalytic activity was contributed by the accumulation of electrons on the PdAg sides, which resulted in a 48% enhancement in comparison to the pure Pd (6.62 μmol g⁻¹ h⁻¹). On the other hand, the selectivity was attributed to the dual-metallic active sites supplied by PdAg NPs, which conducted continuous protonation of C atoms in M₁-C-O-M₂ intermediates. As a result, the sample with a ratio of PdAg 1:2 was in favor of producing CH₄ instead of CO, which was consistent with the best photocurrent and smallest radii.