Oleic acid (OAc, 90%, Sigma-Aldrich), n-octylamine (99.0%, Aladdin), N,N-dimethyl-formamide (DMF, 99.9%, Aladdin), cyclohexane (99.0%, Aladdin), methyl acetate (99.9%, Aladdin), n-Hexane (99.0%, Aladdin), styrene (99.9%, Aladdin), cesium bromide (CsBr, 99.9%, Sigma-Aldrich), cesium chloride (CsCl, 99.9%, Sigma-Aldrich), lead(II) bromide (PbBr₂, 99.9%, Sigma-Aldrich), and zirconium (IV) chloride (ZrCl₄, 99.9%, Sigma-Aldrich). All chemical reagents are used as received.

The synthesis employs the ligand-assisted precipitation (LARP) method as illustrated in Figure 1. Three precursor solutions were prepared by dissolving 2.0 mmol CsBr in 1.0 ml of H₂O, 2.0 mmol PbBr₂ in 2.5 ml of DMF, and 0.25 mmol ZrCl₄ in 0.2 ml of DMF, respectively. Then, 4 ml of oleic acid and 2 ml of n-octylamine were added to a vigorously stirred cyclohexane (200 ml). The PbBr₂ solution, ZrCl₄ solution, and CsBr solution were added. A microemulsion started to form and the solution turned from transparent into light yellow. Afterwards, methyl acetate (100 ml) was added to break up the emulsion. The as-synthesized CsPbBr₃/Cs₄PbBr₆ nanosheets (named as CPB-Zr-0.25) were separated by centrifugation at 10,000 rpm for 10 min, and washed twice with toluene, dried at 80°C, and collected for storage. By changing the amount of ZrCl₄ solution added, CPB-Zr-x (x = 0.25, 0.5, 0.75, 1) were obtained. The organic–inorganic hybrid lead halide perovskite materials MAPbBr₃ and MAPbBr₃-Zr-x nanosheets, named MAPB-Zr-x (MA = CH₃NH3+, x = 0.25, 0.5, 0.75, 1), were obtained with the same protocol by replacing CsBr with MABr. The unmodified CsPbBr₃/Cs₄PbBr₆ nanosheets were synthesized with the same procedure except that the ZrCl₄ precursor was not added. ZrO_xCl_y was prepared with the above procedure except that CsBr and PbBr₂ were not added, as the comparison for photocatalytic evaluation.

Photocatalytic reactions were conducted in a 20-ml quartz flask matching the Pcx-50c multichannel photochemical reaction device with magnetic stirring at a rate of 500 rpm. The specific procedure was as follows: 20 mg of photocatalyst was dispersed into 9 ml of n-Hexane saturated with molecular oxygen and mixed with 1 ml of styrene. The photocatalytic reaction system was irradiated with a white light-emitting diode (LED) illumination (100 mW cm⁻²). After reaction for 5 h, the suspension was centrifuged at 10,000 rpm for 8 min and the supernatant was analyzed on Shimadzu GC-2014.

X-ray diffraction (XRD) analysis was performed on a Rigaku D/MAX2500 VB2+/PCX-ray diffractometer using Ni-filtered Cu Kα radiation (λ = 1.5406 Å) with a scanning angle (2θ) ranging 10°-60° and the scanning speed at 5°/min. UV-vis diffuse reflectance spectra (DRS) of the samples were studied by a UV-vis spectrophotometer (Perkin Elmer Lambda 950) from 300 to 800 nm. FEI Tecnai G2 F20 was used for high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) imaging, and electron energy spectrum (EDX) is obtained from Super-EDX equipped on the TEM FEI G2 (60-300) TEM. The photoluminescence (PL) spectra were acquired by an FLS980 fluorescence spectrometer (steady state, lifetime, phosphorescence, Edinburgh Instruments Ltd.). The steady-state PL was irradiated with 370-nm excitation light. The time-resolved photoluminescence (TRPL) spectra were performed using a 375-nm laser emitter to determine the lifetime of photogenerated carriers. X-ray photoelectron spectroscopy (XPS) was used to study the chemical compositions and the chemical states by a Thermo escalab 250Xi photoelectron spectrometer. Ultraviolet photoelectron spectra (UPS) analysis was performed on British VG Scienta R4000, using He I line as ultraviolet light source with photon energy 21.2 eV, spectrum acquisition range (kinetic energy) 0–22 eV, and step length 0.02 eV. Surface photovoltage (SPV) spectra were conducted in a vacuum chamber with a quartz window. As-prepared samples were placed into a vacuum chamber and illuminated by monochromatic light from a 150-W Xe lamp filtered through an Oriel Cornerstone 130 monochromator (1–10 mW cm⁻²) within a range of 300–600 nm.

Figure 2A shows the XRD pattern of CPB and CPB-Zr-x (x = 0.25, 0.5, 0.75, 1). The naked CPB shows a series of peak at 25.3°, 27.1°, 37.8°, and 48.0° (2θ), corresponding to the (001), (110), (002), and (220) crystallographic plane of CsPbBr₃ (JCPDS card No. 18-0364), respectively, whereas the peak at 12.55°, 20.25°, 22.56°, and 25.55° (2θ) corresponds to the (012), (113), (300), and (024) crystallographic plane of Cs₄PbBr₆ (JCPDS card No. 73-2478), indicating that the CPB prepared by LARP method has a mixed structure of CsPbBr₃ and Cs₄PbBr₆. With the increased amount of ZrCl₄, the peak at peaks of CPB-Zr shifts to the higher values with decreased intensity of the diffraction peak. The positive shift of diffraction peaks may largely originate from the inclusion of Cl⁻ (from ZrCl₄) into the lattice, whereby the Cl⁻ has a smaller radius than that of Br⁻. When the amount of ZrCl₄ further increases, e.g., CPB-Zr-0.75 and CPB-Zr-1, new diffraction peaks start to appear. A thorough search of PDF database turns out that the newly appearing peaks are consistent with the characteristic diffraction peaks of zirconium oxide (JCPDS card No. 1-750). Figures S1a,b compare the core-level Cl 2p and Zr 3d XPS spectra from CPB and CPB-Zr-0.75. It can be clearly found the CPB-Zr-0.75 sample shows obvious Cl 2p and Zr 3d peaks, whereas the CPB shows no signal for either spectrum, confirming that the CPB-Zr-0.75 contains Zr and Cl. It can be referred that during the synthesis, part of Cl⁻ from ZrCl₄ dopes into the CPB to replace Br⁻, while the rest forms ZrO₂ on the surface of CsPbBr₃/Cs₄PbBr₆ nanosheets (Xu et al., 2016).

The UV-vis diffuse reflectance spectra of as-prepared CPB and CPB-Zr-x (x = 0.25, 0.5, 0.75, 1) samples are shown in Figure 2B. The spectrum of CPB sample displays the absorption in the visible-light region exhibiting an absorption edge at 545 nm, while the CPB-Zr-x samples show blue shift of absorption edge. The ZrCl₄ sample cannot absorb light as shown in Figure S2a. It is obvious that the chemical functionalization with ZrCl₄ can significantly affect the optical absorption feature of CPB. By increasing the amount of ZrCl₄, the absorption edge of the samples shows a further blue shift. In particular, the absorption edge of CPB-Zr-1 is at 473 nm. The optical bandgap value can be calculated via the Kubelka–Munk plot shown in Figure S2b. The bandgaps are 2.27, 2.36, 2.46, 2.56, and 2.67 eV for CPB, CPB-Zr-0.25, CPB-Zr-0.5, CPB-Zr-0.75, and CPB-Zr-1, respectively. Figure 2C displays the PL spectra of all samples. The CPB sample shows a PL peak centered at 521 nm, consistent with the reported values (Shamsi et al., 2016). The trend of PL result is in good agreement with that of the UV-vis diffuse reflectance spectra (Figure S3).

The TRPL spectra in Figure 2C show the lifetime of photogenerated carriers and the carrier separation kinetics. The fitting results for TRPL spectra of different samples are listed in Table S1. It is found that the carrier lifetime increases and then decreases with the increased adding amount of ZrCl₄. Compared with the carrier lifetime (24.46 ns) of pure CPB, the CPB-Zr-0.25 shows a longer carrier lifetime at 49.38 ns, but declined gradually to 6.81 ns of CPB-Zr-1. With the small amount of ZrCl₄ added in precursor, the Cl⁻ doped into CPB may result in the elongated carrier lifetime (Jin and Oleg, 2015). With the further increase of added ZrCl₄, ZrO₂ starts to form on the surface of the CPB and promotes the transfer of carriers and shortens the carrier lifetime.

The SPV spectra shown in Figure 2D, employed to evaluate the photochemical charge transfer in the catalyst, reveal that the CPB has very low photovoltage, while the photovoltage increases first with the addition of ZrCl₄ and decreases thereafter. It is probably due to the fact that when a small amount of ZrCl₄ is added, the surface of CPB is modified and passivated by Zr species to reduce surface defects and increase the photovoltage, while the enlarged bandgap induced by the Cl doping leads to the blue shift of the absorption edge, consistent with the UV-vis results (Zhao et al., 2016). With further increase of ZrCl₄, the bandgap of CPB further expands and drastically reduces the absorption and leads to the decrease of surface photovoltage.

Figure 3 shows the TEM images of CPB and CPB-Zr-x. It can be seen that all samples consist of rectangular nanosheets with a size range of 100–400 nm. As exhibited in Figures 3d,e, small particles appear on the surface of CsPbBr₃/Cs₄PbBr₆ nanosheets with the increase of ZrCl₄ molar proportion. The high-resolution TEM in Figure 3f shows a lattice spacing of 0.589 nm determined by fast Fourier transform (FFT), which corresponds to the spacing between the (100) plane. The lattice spacing of 0.566 nm for CPB-Zr-0.75 corresponds to the distances between the (100) plane. The size of CPB-Zr-x nanosheet decreases with the increased amount of ZrCl₄.

To further study the microstructures of CPB modified by ZrCl₄, EDX-mapping is used to map out a single nanosheet of CPB-Zr-0.75 sample under HAADF-STEM. As shown in Figure 4, the distributions of Cs, Pb, Br, Zr, Cl, and O elements can be visualized. The above elements can almost overlap in the selected region. It implies that Cl⁻ from ZrCl₄ has been doped into CPB. As for the Zr, its distribution overlaps with the nanosheet implying that some Zr⁴⁺ may have been doped into the lattice while at the same time it could form ZrO₂ nanoparticles that attach to the surface of CPB. It remains difficult to determine the exact amount of the Zr⁴⁺ that has been doped into the lattice. Considering that the ZrO₂ diffraction peaks found in XRD pattern for CPB-Zr-0.75 and CPB-Zr-1, the Zr⁴⁺ may exist in both forms.

To evaluate the catalytic performance of the catalyst, the photocatalytic oxidation reaction of styrene was carried out to quantify their photocatalytic activity. The calculated reaction rate is summarized in Figure 5. The production rate for CPB is 372 μmol g⁻¹ h⁻¹. With the increase of added ZrCl₄, it increases first and then declines, whereby the CPB-Zr-0.75 shows the maximum production rate of 1,098 μmol g⁻¹ h⁻¹. The comparison sample, ZrO_xCl_y, prepared with the same protocol by adding ZrCl₄ only, shows the production rate of 52 μmol g⁻¹ h⁻¹, suggesting that the nanosheet is critical in increasing the light absorption. One may consider that the improved photocatalytic activity originates solely from the Cl doping. To verify the role of Cl doping, we prepare the samples with the same doping level by substituting the ZrCl₄ with CsCl for the synthesis. The CPB-Cl-x shows only slight improvement, with the highest production rate 1.09 times that of the pure CPB, as shown in Figure S4. It implies that both the surface functionalization of CPB with Zr species and the Cl doping are necessary in largely improving the catalytic performance toward styrene oxidation, possibly through a synergistic effect between the two methods. It also suggests that the addition of ZrCl₄ during the synthesis for the surface modification can “kill two birds with one stone.”

It is also known that the halide perovskite is prone to be oxidized. To evaluate the structure and phase after the reaction, the catalysts are collected for structural and physicochemical analysis with XRD and PL measurements (Figure S5), showing that the XRD pattern for the catalyst after reaction is almost identical to the fresh catalyst with only a slight blue shift, indicating the chemical rigidity of the PVSK after functionalization. MAPB and MAPB-Zr-x (x = 0.25, 0.5, 0.75, 1) are prepared by the same method as above, with photocatalytic experiments shown in Figure S6. The results suggest that that the chemical modification with ZrCl₄ can also improve the photocatalytic activity of MAPB toward the benzaldehyde production, although the baseline performance is not as good as that of CPB. The color of the MPB totally changes into white after the reaction, suggesting that the MAPB is not a robust photocatalyst as shown in Figure S7.

To investigate the effect of ZrCl₄ on the catalytic performance of CPB, the UPS (Figure 6) determines the valence band positions of CPB and CPB-Zr-0.75. The energy level of the CPB and CPB-Zr-0.75 is determined by using ultraviolet photoelectron spectroscopy (UPS) with a photon energy of 21.20 eV. The work function can be determined from the difference between the photon energy and the binding energy of the secondary cutoff edge (Figure 6A). Therefore, the Fermi level (E_f) of CPB and CPB-Zr-0.75 can be determined to be −3.04 and −4.09 eV relative to the vacuum level (E_vac), respectively. As shown in Figure 6B, the valence band (VB) spectra show that the valence band maxima (VBM) of CPB and CPB-Zr-0.75 are 3.03 and 2.12 eV below E_f, respectively. The bandgap of CPB and CPB-Zr-0.75 are calculated by UV-vis, approximately 2.27 eV and 2.56 eV, respectively. Therefore, the energy levels of the conduction bands of the two can be calculated as −3.8 and −3.65 eV, confirming that the Cl⁻ doped into CPB may result in the widening of bandgap.

Based on the experiment, the possible reaction process is then illustrated as the scheme in Figure 6C; the visible light activates the excitons inside the CPB. Upon the separation of electron–hole pairs, the electrons transport to the surface and react with adsorbed O₂ to generate activated oxygen species. Meanwhile, the styrene is adsorbed on the surface of the CPB and oxidized by the holes to the corresponding cationic radicals. The activated oxygen species then selectively oxidize the cationic radicals, finally leading to the formation of benzaldehyde. Due to the addition of ZrCl₄ precursor, the chemically modified CPB can promote the generation and transfer of excitons and carriers and hence accelerates the photocatalytic production rate of benzaldehyde.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.