1 Introduction
The increasing global demand for energy has raised concerns about the depletion and environmental pollution associated with primary energy sources [
1–
4]. One promising solution is the conversion of CO
2 into valuable chemical products [
5,
6]. Photocatalytic CO
2 reduction stands out as an innovative technology that offers a sustainable way to capture and reuse CO
2 by leveraging solar energy, reducing carbon emissions, and converting CO
2 into valuable chemicals [
7–
9]. As a central component of photocatalysis, numerous photocatalysts have been developed for photocatalytic CO
2 reduction [
5,
10,
11]. However, these materials still face limitations in light absorption, charge separation, CO
2 adsorption, and the high energy barrier required to break the C=O bond (~750 kJ/mol) [
12–
14]. Consequently, further research is essential to optimize the performance of photocatalysts to promote their practical application.
Metal halide perovskites (ABX
3) have emerged as promising candidates for photocatalysis due to their excellent light absorption, charge transport properties, and the ability to tune their band structures via compositional adjustments [
15]. However, lead-containing ABX
3 materials still face significant challenges, such as degradation susceptibility and environmental toxicity [
16,
17], which severely limit their practical application in photocatalysis. Therefore, researchers have begun to explore lead-free alternatives using elements such as bismuth (Bi), antimony (Sb), and tin (Sn) [
18–
21]. Among these, Bi
3+ is particularly appealing due to its similar electronic configuration and ionic radius to Pb
2+, which helps retain the high optoelectronic performance of halide perovskites [
22]. The dimer phase Cs
3Bi
2I
9, in particular, has been developed for a wide range of photocatalytic applications, including CO
2 reduction [
12], hydrogen evolution [
23], and pollutant degradation [
24]. However, its performance in photocatalysis is still limited by the relatively lower charge carrier mobility, the incapability to oxidize water, and a tendency to aggregate during the photocatalytic reactions [
25,
26].
To enhance photocatalyst performance, the Z-scheme strategy, which integrates oxidation photocatalyst and reduction photocatalyst, has proven the most effective due to its intrinsic advantages combining broad solar spectrum absorption, enhanced charge separation, and sufficient redox potentials [
24,
27–
29]. For example, Jiang et al. [
30] reported that a Cs
3Bi
2I
9/BiVO
4 Z-scheme heterojunction enhanced photocatalytic CO
2 reduction performance as compared to the pristine Cs
3Bi
2I
9 due to the enhanced charge separation induced by the Z-scheme band alignment. Moreover, the tunability of the dimensionality of the two counterparts in a Z-scheme heterojunction can further promote photocatalytic performance. For example, the heterojunction with a point-on-line structure formed by 1D nanowires and 0D nanoparticles not only enhances charge separation but also alleviates aggregation, ensuring better exposure of active sites. Previous studies have shown that Z-scheme heterojunctions, created from materials with different dimensionalities, have exhibited excellent performance in photocatalytic CO
2 reduction and hydrogen production. A typical example is the CdS/WO
3 Z-scheme photocatalyst, where 0D CdS nanoparticles were loaded onto the 1D WO
3 nanorods, promoting the photocatalytic hydrogen evolution by virtue of the 0D/1D structure with Z-scheme band alignment [
31]. Interestingly, 1D WO
3 nanorod is an ideal support and oxidation photocatalyst for Z-scheme heterojunction due to its low cost, ease of access, good stability, and strong water oxidizing properties [
31–
33]. Loading 0D Cs
3Bi
2I
9 onto 1D WO
3 therefore may enable the full exploitation of the reduction capability and light-responsive activity, while simultaneously preventing self-aggregation of Cs
3Bi
2I
9, thereby exposing more active sites. Moreover, if a Z-scheme heterojunction can be formed between 0D Cs
3Bi
2I
9 and 1D WO
3, the superior reducing ability of Cs
3Bi
2I
9 and the oxidizing power of WO
3 can be combined for completing the photocatalytic redox reactions of CO
2 reduction and water oxidation [
34]. However, to the best of the authors’ knowledge, no previous study has attempted to integrate 0D Cs
3Bi
2I
9 nanoparticles with 1D WO
3 nanorods in a Z-scheme band alignment to enhance photocatalytic CO
2 reduction reaction and water oxidation.
This work proposes the construction of a 0D/1D Z-scheme heterojunction photocatalyst by in situ loading of Cs3Bi2I9 onto WO3 nanorods for photocatalytic CO2 reduction. The resulting Z-scheme heterojunction, Cs3Bi2I9/WO3 (CBI/WO3), achieved a CO production rate of 16.5 μmol/(g·h) under visible light irradiation (λ≥ 420 nm), demonstrating a threefold enhancement as compared to the pristine Cs3Bi2I9. Additionally, the CO selectivity reached up to 98.7%, with only a trace amount of CH4 detected. This high CO2-to-CO rate and CO selectivity stands among the best photocatalysts composed of lead-free halide perovskites (Table S1). The highly enhanced performance is attributed to the formation of 0D/1D Z-scheme heterojunction, which facilitates charge transfer, reduces charge recombination, and preserves and the active sites of 0D Cs3Bi2I9 for CO2 reduction along with the active sites of 1D WO3 for water oxidation.
2 Results and discussion
2.1 Morphological analysis of CBI/WO3
As illustrated in Fig.1(a), WO
3 nanorods were prepared via a hydrothermal method [
31], while Cs
3Bi
2I
9 nanoparticles were prepared via a hot-injection method [
35]. The CBI/WO
3 heterojunction photocatalysts were then synthesized by
in situ growth of Cs
3Bi
2I
9 onto WO
3 (for experimental details, see the Electronic Supplementary Material). In this work, a series of CBI/WO
3 heterojunctions with varying CBI to WO
3 mass ratios were synthesized, including CBI/WO
3-5%, CBI/WO
3-8%, CBI/WO
3-10%, CBI/WO
3-15%, and CBI/WO
3-20%. Digital photographs of the gray WO
3, bright light Cs
3Bi
2I
9, and dark brown CBI/WO
3-15% powder samples are shown in Figure S1. For comparison, a CBI/WO
3 mixture (CBI/WO
3-mix) was prepared by physically mixing Cs
3Bi
2I
9 and WO
3.
The scanning electron microscopy (SEM) image in Fig.1(b) demonstrates that the 0D Cs
3Bi
2I
9 nanoparticle tends to aggregate. In contrast, Fig.1(c) shows that the as-synthesized WO
3 forms 1D rod-like structures. The SEM image of CBI/WO
3-15% in Fig.1(d) reveals that 0D Cs
3Bi
2I
9 nanoparticles are uniformly dispersed on the 1D rod-like WO
3, forming a 0D/1D structure. Energy-dispersive X-ray spectroscopy (EDX) mapping of CBI/WO
3-15% confirms the uniform distribution of all elements, validating the even dispersion of 0D Cs
3Bi
2I
9 on the 1D WO
3 support. The Zeta potentials of Cs
3Bi
2I
9 and WO
3 were measured to support the formation of the 0D/1D structure. As shown in Fig. S2, Cs
3Bi
2I
9 has a Zeta potential of −47.06 mV, while WO
3 has a Zeta potential of 23.26 mV. These opposite polarities create an interfacial potential difference, which facilitates electrostatic attraction between Cs
3Bi
2I
9 and WO
3 [
36], favoring the formation of the heterojunction.
2.2 Structural and chemical states analysis of CBI/WO3
X-ray diffraction (XRD), high-resolution transmission electron microscopy (HRTEM), and Fourier transform infrared spectroscopy (FTIR) were conducted to comprehensive analyze the structure and chemical states of the CBI/WO
3-15% heterojunction. As illustrated in Fig.2(a), the characteristic XRD peaks of WO
3 align with those of the monoclinic WO
3 (JCPDS No. 33-1387) [
31], while the characteristic peaks of Cs
3Bi
2I
9 correspond to the hexagonal phase of Cs
3Bi
2I
9 (JCPDS No. 23-0847) [
37]. Notably, both XRD peaks of WO
3 and Cs
3Bi
2I
9 can be found in the XRD pattern of CBI/WO
3-15%, providing evidence for the co-existence of WO
3 and Cs
3Bi
2I
9 in CBI/WO
3-15%.
Additionally, X-ray photoelectron spectroscopy (XPS) analysis shows peaks corresponding to Cs 3d, Bi 4f, I 3d, W 4f, and O 1s in the full XPS spectrum of the CBI/WO3-15% heterojunction (Fig. S3), further supporting the co-existence of Cs3Bi2I9 and WO3 in the CBI/WO3-15% heterojunction. The HRTEM images of CBI/WO3-15% presented in Fig.2(b)-Fig.2(d) reveal the formation of CBI/WO3 heterojunction, with lattice spacings of 0.364 nm belonging to the (200) crystal plane of WO3, and lattice spacing of 0.42 nm belonging to the (110) plane of Cs3Bi2I9 being observed in the same CBI/WO3-15% sample. Those observations confirm that Cs3Bi2I9 has been successfully grown on the surface of WO3, aligning with the SEM results (Fig.1(d) and Fig.1(e)).
Furthermore, the FTIR spectrum in Fig. S4 shows that the structure of WO
3 is preserved in the CBI/WO
3-15% heterojunction, with prominent transmittance peaks in the range of 650-1000 cm
−1 and around 3400 cm
−1, which can be attributed to the O-W-O and W-OH stretching vibrations [
38], respectively.
2.3 Photocatalytic CO2 reduction performance
A 300 W Xe lamp equipped with a 420 nm cutoff filter (light intensity of 100 mW/cm
2) was used as the visible light source to evaluate the photocatalytic CO
2 reduction performance in a semi-closed gas cycle reactor. As shown in Fig.3(a), no CO
2 reduction product was observed in the presence of pristine WO
3, and the rate of CO
2-to-CO production for pristine Cs
3Bi
2I
9 was only 5.3 μmol/(g·h). Detailed CO
2-to-CO yield data, along with their error values from three independent tests, are provided in Table S2. It is speculated that the conduction band potential of a single WO
3 photocatalyst is not negative enough to drive the CO
2 reduction reaction [
39]. While a single Cs
3Bi
2I
9 photocatalyst can reduce CO
2, its efficiency is limited due to its inability to oxidize water, charge carrier recombination, and self-aggregation.
In contrast, the 0D/1D heterostructure of CBI/WO3 demonstrated significantly enhanced CO2 reduction performance, with the highest CO2-to-CO rate reaching up to 16.5 μmol/(g·h) in the presence of the CBI/WO3-15% photocatalyst, approximately 3 times that of the pristine Cs3Bi2I9. Additionally, the CO selectivity was as high as 98.7%, with only a trace amount of CH4 detected (Fig.3(b)). The CBI/WO3-15% photocatalyst, with its high CO2-to-CO rate and CO selectivity, ranks among the best photocatalysts composed of lead-free halide perovskites using environmentally friendly materials (Table S1). For comparison, a physical mixture of Cs3Bi2I9 and WO3 (CBI/WO3-mix) was also tested, yielding a CO production rate of 7.2 μmol/(g·h), which was significantly lower than that of CBI/WO3-15%, highlighting the importance of interfacial charge transfer in the intimate heterojunction formed by in situ growth of 0D Cs3Bi2I9 on 1D WO3.
Moreover, as shown in Fig.3(c), no CO or other products were detected in the absence of photocatalyst, light illumination, or CO2, confirming that CO2 was reduced through a photocatalytic process. A small amount of O2 was also detected (Table S3). The 0D/1D Cs3Bi2I9/WO3 Z-scheme heterojunction photocatalyst exhibited a visible-light-driven photocatalytic H2O oxidation performance for O2 production, with a rate of approximately 8.0 μmol/(g·h).
The stability of the CBI/WO3-15% heterojunction photocatalyst was evaluated by conducting a cyclic catalytic test, with a duration of 3 h per cycle, as shown in Fig.3(d). The results demonstrated that the CBI/WO3-15% photocatalyst maintained a stable CO production rate after three cycles of 3 h reaction. XRD and FTIR analyses before and after the cyclic catalytic tests showed no significant changes in the structure of the photocatalyst (Figs. S5(a) and S5(b)), indicating its decent stability.
2.4 Charge transfer direction in CBI/WO3 heterojunction
Different directions of charge transfer can lead to different types of heterojunctions, significantly influencing the photocatalytic performance. Previous studies have revealed that when Cs
3Bi
2I
9 is combined with other materials to form heterojunctions, charge transfer primarily occurs through the Bi and I atoms, specifically their associated 3p and 5p orbitals [
40]. This is due to the conduction band minimum of Cs
3Bi
2I
9 being derived from the 3p and 5p orbitals of Bi and I [
40,
41]. In the case of WO
3, charge transfer predominantly occurs through the 4d orbitals of W, as these orbitals contribute to the conduction band [
42].
To investigate the interfacial charge transfer direction between 0D Cs
3Bi
2I
9 and 1D WO
3 in CBI/WO
3-15%,
in situ XPS and electron spin resonance (ESR) measurements were conducted. As shown in Fig.4(a)-Fig.4(c), both pristine Cs
3Bi
2I
9 and CBI/WO
3-15% exhibit distinct XPS peaks for Cs 3d, Bi 4f, and I 3d. Specifically, the XPS peaks at 738.40 and 724.41 eV correspond to the Cs 3d
3/2 and Cs 3d
5/2 of Cs
3Bi
2I
9 [
43] (Fig.4(a)). The peaks at 164.31 and 158.98 eV correspond to Bi 4f
5/2 and Bi 4f
7/2 (Fig.4(b)), while the peaks at 630.14 and 618.68 eV correspond to I 3d
3/2 and I 3d
5/2 of Cs
3Bi
2I
9 [
4], respectively (Fig.4(c)). Notably, under dark conditions, the Cs 3d, Bi 4f, and I 4f peaks of CBI/WO
3-15% shift to higher binding energies compared to those of pure 0D Cs
3Bi
2I
9 (Fig.4(a)-Fig.4(c)).
Furthermore, as depicted in Fig.4(d), both pure WO
3 and CBI/WO
3-15% exhibit W 4f characteristic peaks at 38.01 and 35.84 eV, corresponding to W 4f
5/2 and W 4f
7/2 [
44], respectively. Under dark conditions, the W 4f peaks of CBI/WO
3-15% shift to lower binding energies compared to those of pure WO
3. These findings suggest that electrons transfer from Cs
3Bi
2I
9 to WO
3 in the dark, indicating the formation of an intimate heterojunction between Cs
3Bi
2I
9 and WO
3.
Interestingly, under light illumination, the Cs 3d, Bi 4f, and I 4f peaks of CBI/WO
3-15% shift to lower binding energies, while the W 4f peaks shift to higher binding energies, compared to their pristine counterparts, indicating that electrons transfer from WO
3 to Cs
3Bi
2I
9 under illumination. The different directions of electron transfer under dark and light conditions can be explained by the fact that, in the dark, electrons from Cs
3Bi
2I
9 tend to migrate to WO
3 upon contact to align their Fermi levels. Once the Fermi levels of Cs
3Bi
2I
9 and WO
3 align, the energy bands of Cs
3Bi
2I
9 and WO
3 bend upwards and downwards at the interface [
45,
46], respectively, creating an internal electric filed pointing from Cs
3Bi
2I
9 to WO
3. Under light illumination, photons excite the semiconductors, generating electron-hole pairs and increasing the electron concentration in the 4d orbitals of W in WO
3. Thereafter, as driven by the IEF, the electrons from the 4d orbitals of W will be forced to transfer toward Cs
3Bi
2I
9, where they recombine with the holes of Cs
3Bi
2I
9.
To determine the band positions of Cs
3Bi
2I
9 and WO
3, ultraviolet visible (UV-Vis) absorption, Mott-Schottky, and valance band-XPS (VB-XPS) measurements [
47] were performed. UV-Vis combined with Tauc plots were used to determine the band gaps (
Eg), which were found to be 1.98 eV for Cs
3Bi
2I
9 [
48,
49] and 2.83 eV for WO
3 [
50,
51], both indicating indirect band gap semiconductors (Figs. S6, S7(a), and S7(d). The Mott-Schottky measurement determined the flat band potentials (
Efb) of Cs
3Bi
2I
9 and WO
3, which were −0.69 and −0.46 eV, respectively (Figs. S7(b) and S7(e)). Additionally, VB-XPS measurements provided the valence band positions (
EVB) for Cs
3Bi
2I
9 and WO
3, yielding values of 1.14 and 2.32 eV, respectively (Figs. S7(c) and S7(f)). Using the relationship of
ECB =
Eg−
EVB, the conduction band positions (
ECB) were calculated to be −0.84 eV for Cs
3Bi
2I
9 and −0.51 eV for WO
3, which were in close agreement with the
Efb values obtained from the Mott-Schottky measurement, confirming the accuracy of the measurement of the conduction band positions. Hence, Cs
3Bi
2I
9 and WO
3 form a staggered band position relationship, as depicted in Fig. S8.
ESR measurements were also conducted to further validate the Z-scheme charge transfer path of CBI/WO3. Under dark conditions, no ESR peak was observed in the presence of the photocatalysts (Fig. S9). However, under light illumination, distinct DMPO-·O2– characteristic peaks were observed for both CBI/WO3-15% and Cs3Bi2I9 displayed (Fig.4(e)), indicating that CBI/WO3-15% preserved the reduction ability of Cs3Bi2I9 to reduce oxygen into superoxide radicals. Furthermore, as shown in Fig.4(f), the appearance of DMPO-OH characteristic peaks with an equidistant pattern of 1:2:2:1 quadruplet in the presence of both CBI/WO3-15% and WO3 confirmed that CBI/WO3-15% maintained the oxidation ability of WO3. These preserved reduction ability of Cs3Bi2I9 and oxidation ability of WO3 are typical features of a Z-scheme heterojunction. Hence, based on the preserved redox capabilities and the staggered bandgap positions, it can be concluded that the CBI/WO3 heterojunction follows a typical Z-scheme charge transfer pathway.
To determine the source of the product from the photocatalytic experiment, CO2 reduction was performed using 18H2O and 13CO2. As shown in Fig.5(a), nearly all the O2 detected originated from 18H2O, with the content of 16O2 content being negligible. The reduction products, 13CO and a trace amount of 13CH4 were identified to be originated from 13CO2 (Fig.5(b)). These results confirm that CO2 was reduced to CO and a trace amount of CH4, while water was oxidized to O2 in the presence of the CBI/WO3 heterojunction photocatalyst.
Based on these findings, the photocatalytic CO2 reduction reaction mechanism for the CBI/WO3 Z-scheme heterojunction is proposed, as depicted in Fig.5(c) and Fig.5(d). Upon light illumination, charge carriers are generated both in Cs3Bi2I9 and WO3. Following the Z-scheme charge transfer pathway, the low-energy holes from Cs3Bi2I9 recombine with the low-energy electrons from WO3, leaving the high-energy electron in the Cs3Bi2I9 to reduce CO2 adsorbed on the catalyst, forming CO and trace amounts of CH4, while the holes generated in WO3 oxidize H2O to produce oxygen. The design of 0D/1D Z-scheme heterojunction allows the high-energy electrons to be retained in 0D Cs3Bi2I9 to participate in the reduction of CO2, producing CO and trace amounts of CH4. Meanwhile, the 1D WO3 nanorod provides a 1D charge carriers transfer channel, efficiently transferring photogenerated holes to react with H2O to produce O2 and photogenerated electrons to Cs3Bi2I9, where they recombine with the low-energy holes in Cs3Bi2I9.
2.5 Photophysical and photoelectrochemical properties
The photophysical and photoelectrochemical properties of photocatalysts are crucial for their performance. Herein, a series of photophysical and photoelectrochemical characterizations were conducted on Cs
3Bi
2I
9, WO
3, and the Z-scheme photocatalyst CBI/WO
3-15%. The result of the surface photovoltage spectroscopy (SPV) is shown in Fig.6(a). It can be found that the surface photovoltage signals for CBI/WO
3-15%, Cs
3Bi
2I
9, and WO
3 are positive, with CBI/WO
3-15% exhibiting the most positive signal. The positive SPV signals indicate that the CBI/WO
3-15% heterojunction facilitates the most efficient transfer of photogenerated holes to the surface of the semiconductor [
52], considering that both Cs
3Bi
2I
9 and WO
3 are n-type semiconductors as confirmed by Mott-Schottky plots (Figs. S7(b) and S7(e)) [
41].
Electrochemical impedance spectroscopy (EIS) was further employed to assess the charge carrier transfer efficiency. As illustrated in Fig.6(b), CBI/WO
3-15% demonstrated the smallest electrochemical impedance, implying the highest charge transfer efficiency [
53]. Furthermore, the equivalent resistances were fitted by adopting the Randles equivalent circuit model (inset of Fig.6(b)) with the fitting results listed in Table S4. The smaller charge transfer resistance (
Rct) of CBI/WO
3-15% compared to that of pristine Cs
3Bi
2I
9 indicates a rapid transfer of photogenerated charge carriers [
54]. Moreover, the reduced Warburg impedance (
Zw) for CBI/WO
3-15% signifies a more rapid charge diffusion rate [
55].
The photocurrent response is a critical indicator of the photoelectrochemical property of the material. As depicted in Fig.6(c), the current intensity for the Z-scheme heterojunction CBI/WO3-15% was significantly enhanced, providing strong evidence that the charge transfer between Cs3Bi2I9 and WO3 was greatly improved. This improvement in charge transfer efficiency underscores the effectiveness of the Z-scheme configuration in facilitating electron-hole separation, thereby enhancing the overall photocatalytic performance of the heterojunction system. These findings suggest a substantial optimization of the interfacial interactions within the heterojunction, which could be pivotal for advancing the design of efficient photocatalytic materials.
TRPL measurements were conducted to quantitatively assess the charge separation of the Z-scheme heterojunction CBI/WO
3-15%, as shown in Fig.6(d), with the fitting parameters listed in Table S5. The average PL lifetimes for CBI/WO
3-15%, Cs
3Bi
2I
9, and WO
3 were determined to be 61.12, 62.40, and 66.79 ns, respectively. It is speculated that the construction of the Z-scheme heterojunction enhances charge transfer and separation, leading to a reduction in the PL lifetime [
30]. The reduction in PL lifetime can be attributed to the efficient dissociation of photogenerated electron-hole pairs, which minimizes recombination events that would typically prolong the PL emission [
43–
45]. In conclusion, the photophysical and photoelectrochemical characterizations above demonstrate that the Z-scheme heterojunction effectively enhances separation and transfer of photogenerated charge carriers, contributing to improved photocatalytic CO
2 reduction performance.
3 Conclusions
This work introduces a Cs3B2I9/WO3 heterojunction photocatalyst with both 0D/1D structure and Z-scheme band alignment for photocatalytic CO2 reduction in a gas–solid system. The 0D/1D Z-scheme photocatalyst exhibited an efficient photocatalytic CO2 reduction performance under visible light irradiation (λ≥ 420 nm), achieving a CO generation rate of 16.5 μmol/(g·h), which is a threefold increase compared to pristine Cs3Bi2I9, along with a high selectivity for CO of 98.7%. The effective conversion of CO2-to-CO can be attributed to the morphological advantages of the 0D/1D structure, which exposes more active sites, as well as the successful formation of the Z-scheme heterojunction which preserves the strong reduction capability of Cs3Bi2I9 and the robust oxidation ability of WO3. This work provides valuable insights and strategies for the design of heterojunctions with diverse dimensionalities.