Introduction

Splitting hydroiodic acid (HI) has significant research value in the field of energy science and technology1,2,3,4,5. The traditional decomposition of HI at high temperature of 500 °C has been demonstrated to be an effective approach to produce hydrogen2,5, however, it is unsustainable, dangerous, and not cost-effective. Solar-driven splitting of HI, as a promising low-cost technique, has recently attracted more research interest because it can achieve the co-production of zero-emission hydrogen (H2) fuel and value-added chemicals (I2/I3−) only under light condition at room temperature (RT)6,7,8. The development of advanced photocatalysts is very necessary for achieving the high efficiency of photocatalytic HI splitting, however, unfortunately, the reported photocatalysts cannot stably work in HI solution with strong acid property, which poses a high demand on new material selection and photocatalyst design. Recently, the organic–inorganic lead halide perovskites (OLHPs), such as MAPbI3 (MA = CH3NH3), with the advantages of facile synthesis, low cost and superior optoelectronic characteristics9,10,11,12,13, have been proved to be a promising photocatalyst for hydrogen production. Nevertheless, the unstable organic component in those MAPbI3 photocatalytic materials easily suffers from the serious photo corrosion in HI solution14,15,16,17,18,19,20,21,22,23,24,25, which severely limit the applications of OLHPs in the photocatalytic HI splitting into hydrogen19,21. Besides, the lead toxicity of MAPbI3 also inhibits its practical application. In the seek for the lead-free perovskite materials, Sn-based perovskites have been demonstrated to have a narrower optical bandgap than that of the Pb-based perovskites26,27,28,29,30, indicating their larger light absorption range. Especially, all-inorganic Cs2SnI6 perovskite is more preferred material system in view of its good stability, superior conductivity, and appropriate energy band levels31,32,33,34,35.

Apart from above obstacles, the reported OLHPs-based photocatalysts still suffer from a very low photocatalytic activity caused by the serious photogenerated electron–hole recombination15,17, restricting their further development. Decorating cocatalyst onto semiconductors to form the hetero-structured photocatalysts has been regarded as one of the simplest and most effective strategies for inhibiting the charge recombination, and hence improving the photocatalytic performance in the H2 evolution reaction36,37,38,39,40. Considering the fact that halide perovskites possess a large number of defects caused by its inherent property of low temperature crystallization41, they might be ideal scaffolds to anchor metal atoms and stabilize single atoms for achieving the greatly enhanced photocatalysis. However, developing new procedures for achieving highly efficient photocatalysts of halide perovskites anchored by a new class of metal single atoms for the hydrogen production, to the best of our knowledge, is still a great challenge in the field of photocatalysis.

Herein, we demonstrate the first example on making a new class of single-atom Pt–I3 sites anchored on all-inorganic Cs2SnI6 perovskite (PtSA/Cs2SnI6) for efficient H2 evolution photocatalysis from HI splitting at RT via a facile and cost-effective strategy. Turnover frequency (TOF) of as-prepared PtSA/Cs2SnI6 catalyst exhibits 176.5-fold enhancements compared with Pt nanoparticle anchored on Cs2SnI6 (PtNP/Cs2SnI6) catalyst, outperforming all of reported Pt-loaded halide perovskites photocatalysts, along with excellent catalytic stability. Combining charge-carrier dynamics studies with theory calculations reveals that both unique coordination structure and electronic property of Pt–I3 sites and strong metal–support interaction (SMSI) effect are the main reasons in achieving the greatly enhanced photocatalytic performance in hydrogen production from HI aqueous solution.

Results

Energy band structure of Cs2SnI6 and its stability in aqueous HI solution system

The synthetic procedure of PtSA/Cs2SnI6 is schematically illustrated in Fig. 1a. Briefly, the Cs2SnI6 was firstly synthesized by a one-pot hydrothermal treatment of cesium acetate and tin (II) acetate in presence of the excess hydriodic acid (HI) solution, followed by the impregnation of the platinum complex. Subsequently, the PtSA/Cs2SnI6 was obtained after activation at 160 °C for 1 h in H2/Ar atmosphere. Field-emission scanning electron microscopy image of the as-prepared Cs2SnI6 demonstrates that the Cs2SnI6 mainly adopts the octahedral morphology (Fig. 1b). Powder X-ray diffraction (PXRD) pattern (Fig. 1c) of the obtained Cs2SnI6 can be well indexed to the cubic Cs2SnI6 phase (JCPDS card NO. 51-0466), confirming the successful synthesis of Cs2SnI6. Furthermore, the as-synthesized Cs2SnI6 with a high yield of 93% exhibits the high-temperature stability with a decomposition temperature up to 350 °C in air atmosphere (Supplementary Fig. 1). Furthermore, Cs2SnI6 is insoluble in aqueous HI solution at RT, confirmed by the fact that the solubility of Cs2SnI6 was zero at 25 °C, and slightly increased to 10.0 × 10−6 mol L−1 as the temperature was improved to 100 °C, still far lower than that of reported MAPbI3 (0.645 mol L−1 at 20 °C) (Fig. 1d)25. Moreover, there is no change in the color of aqueous HI solution with or without Cs2SnI6 powder, which further verifies the insolubility for Cs2SnI6 (inset of Fig. 1d).

Fig. 1: Energy band structure of Cs2SnI6 and its stability in aqueous HI solution system.
figure 1

a Schematic diagram of preparation process for the PtSA/Cs2SnI6 catalyst. b SEM image and c PXRD pattern of Cs2SnI6. d Solubility of Cs2SnI6 in aqueous HI solution at different temperature. The inset shows the photograph of the 6 M HI aqueous solution with and without Cs2SnI6 powder at 25 °C. e PXRD patterns of precipitates for Cs2SnI6 powder in aqueous HI solution with various concentrations. f UV–visible absorption spectrum of Cs2SnI6 powder. The inset is the photograph of the Cs2SnI6. g Schematic energy band diagram of the Cs2SnI6, charge generation and charge transfer process over the Cs2SnI6 under visible-light irradiation.

Full size image

Furthermore, the stability of Cs2SnI6 in aqueous HI solution system was further investigated via PXRD (Fig. 1e). We find that the as-prepared Cs2SnI6 powder exhibits the excellent stability in aqueous HI solution system (decompose when HI concentration is <1.0 M), better than that of reported MAPbI3 (3.16 M)25. These results demonstrate that the Cs2SnI6-based aqueous HI splitting system is superior to the MAPbI3-based one, suggesting the great potential in photocatalytic HI splitting. The ultraviolet–visible (UV–vis) absorption spectrum and X-ray photoelectron spectroscopy (XPS) valence spectrum were employed to determine the energy band structure of as-prepared Cs2SnI6. As shown in Fig. 1f, the Cs2SnI6 powder has a broad optical absorption range, and its bandgap energy (Eg) is 1.22 eV, by calculating from the absorbance data on the basis of the Kubelka–Munk equation, less than that of reported MAPbI3 (1.53 eV) (Supplementary Fig. 2)25. The result indicates that Cs2SnI6 possesses a larger light absorption range and higher conductivity than MAPbI3. The valence band (VB) position of the Cs2SnI6 is −5.46 eV with respect to the vacuum level (corresponding to 0.96 eV versus the normal hydrogen electrode (NHE)), as obtained from XPS VB spectroscopy (Supplementary Fig. 3). Accordingly, the corresponding conduction band (CB) position of Cs2SnI6, obtained by coupling the VB positions and Eg, is calculated to be −4.24 eV with respect to the vacuum level (corresponding to −0.26 eV versus NHE). It should be noted that the calculated values of CB and VB are an approximation since the complicated electrolyte–perovskite interface effects were not considered herein. The suitable energy band levels of Cs2SnI6 (inset of Fig. 1g) provides new opportunity for straddling the redox potentials of HI to split HI into H2 and I3−.

Structure characterization of PtSA/Cs2SnI6

The aberration-corrected high-angle annular dark field-scanning transmission electron microscopy (HAADF-STEM) was performed to confirm the distribution and configuration of Pt single-atom in Cs2SnI6. No Pt clusters or large nanoparticles were observed in the low-magnification HAADF-STEM image (Fig. 2a), further confirmed by PXRD analysis (Supplementary Fig. 4), in which there are no characteristic peaks of crystalline Pt species. The atomic-resolution HAADF-STEM image (Fig. 2b) clearly depicts the individual dispersion of Pt atoms on the surface of as-prepared Cs2SnI6. The corresponding energy dispersive X-ray spectroscopy (EDS) mapping images confirm the uniform dispersion of Sn, I, and Pt atoms in PtSA/Cs2SnI6 (Fig. 2c), indicating that Pt atoms are uniformly distributed over Cs2SnI6. The Pt content on PtSA/Cs2SnI6 is determined to be 0.12 wt% by inductively coupled plasma-atomic emission spectrometry (ICP-AES).

Fig. 2: Structure characterization of PtSA/Cs2SnI6.
figure 2

a Low-magnification and b high-magnification HAADF-STEM images, and c the corresponding STEM-EDS elemental mapping of PtSA/Cs2SnI6. d Pt L3-edge XANES spectra and corresponding K3-weighted FT spectra at e R space and g k-space of PtSA/Cs2SnI6, PbI2 and Pt foil. XANES f R space and h k-space fitting curves of PtSA/Cs2SnI6. i High-resolution XPS Pt 4f spectrum of PtSA/Cs2SnI6.

Full size image

The oxidation state and coordination environment of the Pt species in PtSA/Cs2SnI6 were further investigated by X-ray absorption fine structure (XAFS) spectroscopy. Figure 2d depicts the normalized Pt L3-edge X-ray absorption near-edge structure (XANES) spectra of PtSA/Cs2SnI6, PtI2 and Pt foil. The white-line intensity of Pt L3-edge XANES of PtSA/Cs2SnI6 is higher than those of PtI2 and Pt foil, indicating that the oxidation state of Pt in PtSA/Cs2SnI6 is more than +2. The Fourier-transformed (FT) k3-weighted extended X-ray absorption fine structure (EXAFS) spectra (Fig. 2e) exhibits a prominent peak at ∼2.49 Å in PtSA/Cs2SnI6, similar to that of the Pt–I bond in PtI2, and shorter than that of the Pt–Pt bond in Pt foil, implying that the Pt atoms are anchored onto the surface of PtSA/Cs2SnI6 by the Pt–I bond. According to the EXAFS fitting results (Fig. 2f, Supplementary Figs. 5, 6, and Table 1), one Pt atom in PtSA/Cs2SnI6 is coordinated with approximately three I atoms (labeled as Pt–I3). To further confirm the Pt single atoms in the PtSA/Cs2SnI6, the wavelet transform (WT) EXAFS of the PtSA/Cs2SnI6 and the reference systems (PtI2 and Pt foil) was performed (Fig. 2g, h and Supplementary Figs. 7, 8). The WT-EXAFS maximum is observed at 9.38 Å−1 in Pt foil, assigned to the Pt–Pt bond. In contrast, the PtSA/Cs2SnI6 presents an intensity maximum value at 9.89 Å−1, similar to that of Pt–I bond in PtI2 (9.82 Å−1), further indicating the atomic dispersion of Pt coordinated with I atoms on the surface of PtSA/Cs2SnI6.

To further confirm the chemical state of Pt and chemical environment of Cs, Sn, and I in the catalysts, XPS measurements were conducted. As shown in Fig. 2i, the binding energy of Pt 4f in the high-resolution XPS spectrum of PtSA/Cs2SnI6 can be deconvoluted into two chemical states, assigned to Pt2+ at 73.1 and 75.6 eV, and Pt4+ at 75.1 and 77.5 eV, respectively, suggesting that Pt in the PtSA/Cs2SnI6 is Ptδ+ (2 < δ < 4), in agreement with the aforementioned Pt L3-edge XANES results. Noticeably, a positive shift in the I 3d XPS spectra (Supplementary Fig. 9a) is observed upon Pt loading, implying the strong interaction between the Pt and I atoms in PtSA/Cs2SnI6. It means that those I species can serve as the anchor sites, and coordinate with the isolated Pt atoms. Moreover, the binding energy of Sn 3d in PtSA/Cs2SnI6 has an obviously positive shift, whereas that of Cs 3d was unchanged relative to that in pure Cs2SnI6 (Supplementary Fig. 9b, c), implying that the electron transfer from Cs2SnI6 to PtSA depends on the Sn–I–Pt bonds rather than the Cs–I–Pt bonds.

Superior photocatalytic activity and stability of PtSA/Cs2SnI6 catalyst

The activities of photocatalytic H2 evolution over PtSA/Cs2SnI6, PtNP/Cs2SnI6, and Cs2SnI6 catalysts were evaluated in aqueous HI solution system under visible-light (λ ≥ 420 nm, 100 mW cm−2) irradiation by a 300 W Xe lamp and a homemade double-layered Pyrex vessel. The results show that pristine Cs2SnI6 owns a poor photocatalytic performance with a H2 production rate of 25 μmol h−1 g−1 (Fig. 3a and Supplementary Fig. 10). We find that the photocatalytic performance of Cs2SnI6 can be dramatically increased by anchoring Pt single atoms onto the surface of Cs2SnI6, reaching its maximum upon Pt loading up to 0.12 wt% (Fig. 3a and Supplementary Fig. 11). Further increasing or decreasing the loading content of Pt on Cs2SnI6 surface leads to the decreased photocatalytic activities. This is because the excessive Pt species can reduce the light absorption of Cs2SnI6 due to the shading effect whereas the insufficient Pt species cannot provide the rich H2-releasing active sites. Considering that the interfacial contact can affect the charge transfer from Cs2SnI6 to PtNP, we also prepared the PtNP on Cs2SnI6 (noted as PtNPphoto/Cs2SnI6) by the direct photo-deposition method, and studied its photocatalytic activity. The result shows that the rate of H2 evolution (101 μmol g−1 h−1) over the optimal PtNPphoto/Cs2SnI6 sample is only slightly better than that of PtNP/Cs2SnI6 (Supplementary Fig. 12), still much lower than that of PtSA/Cs2SnI6. The 0.12wt% PtSA/Cs2SnI6 shows the champion activity for the photocatalytic H2 production with a H2 production rate of 430 μmol h−1 g−1, 17.2 and 5.8 times higher than pure Cs2SnI6 and optimized 3.88wt% PtNP/Cs2SnI6 (Fig. 3a and Supplementary Figs. 13, 14), and also achieves a TOF of 70.6 h−1 per Pt, 176.5 times higher than that of PtNP/Cs2SnI6 catalyst (0.4 h−1) (Fig. 3b). Moreover, the TOF of PtSA/Cs2SnI6 catalyst is superior to all of the reported Pt-loaded halide perovskite photocatalysts (Fig. 3c and Supplementary Table 2).

Fig. 3: Superior photocatalytic activity and stability of PtSA/Cs2SnI6 catalyst.
figure 3

a The rate of photocatalytic H2 evolution over PtSA/Cs2SnI6, PtNP/Cs2SnI6, and Cs2SnI6 catalysts. b TOF of PtSA/Cs2SnI6 and PtNP/Cs2SnI6 catalysts. c TOF comparisons of PtSA/Cs2SnI6 catalyst and other reported Pt-loaded halide perovskite photocatalysts. d Cyclic stability of the PtSA/Cs2SnI6 catalyst.

Full size image

Furthermore, the photocatalytic activity of PtSA/Cs2SnI6 can be stable without obvious decrease over four cycles and even at a successive 180-h tracking (Fig. 3d and Supplementary Fig. 15). After stability test, the Pt species are still atomically dispersed on the surface of Cs2SnI6, confirmed by the diverse results of PXRD, HAADF-STEM, and STEM-EDS elemental mappings (Supplementary Fig. 16–18). In addition, the XPS spectrum of Pt 4f in PtSA/Cs2SnI6 after photocatalytic test reveals that the valence of Pt species has no obvious change (Supplementary Fig. 19), revealing the excellent chemical stability of PtSA on Cs2SnI6.

Charge-carrier dynamics

Photoluminescence (PL) technique was employed to evaluate the effect of Pt single atoms on the kinetics of charge carriers transfer and recombination over the Cs2SnI6. As depicted in Fig. 4a, the PL quenching of PtSA/Cs2SnI6 is more efficient than those of PtNP/Cs2SnI6 and Cs2SnI6, indicating the enhanced extraction and reduced recombination of charge carriers. To quantify the charge-carrier dynamic, TRPL spectroscopy was performed (Fig. 4b). The average decay lifetime of PtSA/Cs2SnI6 is ca. 61 ns (inset of Fig. 4b), smaller than those of PtNP/Cs2SnI6 (ca. 98 ns) and Cs2SnI6 (ca. 109 ns), further confirming that the atomically dispersed Pt species can enhance the separation of charge carriers. Besides, the photocurrent response spectra reveal that the PtSA/Cs2SnI6 exhibits a significantly enhanced photocurrent than those of PtNP/Cs2SnI6 and Cs2SnI6 (Fig. 4c), indicating its more efficient separation and transfer of photogenerated electron–hole pairs. Similarly, the linear-sweep voltammogram curves demonstrate that PtSA/Cs2SnI6 presents an overpotential of −0.38 V (versus RHE) at current density of 10 mA cm−2, much lower than those of PtNP/Cs2SnI6 and Cs2SnI6 (Supplementary Fig. 20), implying its higher efficiency during the surface catalytic reaction. Furthermore, the PtSA/Cs2SnI6 displays the smallest semicircle in Nyquist plots obtained from the electrochemical impedance spectroscopy (Fig. 4d), also suggesting that the atomically dispersed Pt–I3 configuration makes a great contribution to improving the efficiency of charge separation and transfer over Cs2SnI6.

Fig. 4: Charge-carrier dynamics.
figure 4

a Steady-state PL spectra, b time-resolved transient PL decay, c photocurrent responses spectra, and d electrochemical impedance spectroscopy of Cs2SnI6, PtNP/Cs2SnI6, and PtSA/Cs2SnI6 catalysts.

Full size image

Theoretical calculations

The photogenerated charge transfer on PtNP/Cs2SnI6 and PtSA/Cs2SnI6 was investigated by the DFT calculations. An extra electron was introduced to simulate the photogenerated electron in PtNP/Cs2SnI6 and PtSA/Cs2SnI6. The obtained charge density difference before and after photoexcitation reveals that the introduced photogenerated electrons tend to be distributed in the whole PtNP in PtNP-Cs2SnI6 (Fig. 5a), which undesirably decreases the electron density per Pt atom in PtNP. Instead, the electron in PtSA/Cs2SnI6 is only located between the PtSA and neighboring three I atoms (Fig. 5b), implying the high electron density on the Pt–I3 site. This is attributed to the SMSI effect42. Thus, the electron density per Pt atom in PtNP is further lower than that of PtSA, which is considered to lead to the lower HER activity of PtNP. To exclude the background charge effect, one donor hydrogen atom was introduced to calculate the charge density difference43. The obtained results also showed the localized distribution of electrons in the Pt–I3 region (Supplementary Fig. 21). This SMSI effect is beneficial to the photogenerated charge transfer between the photocatalyst and cocatalyst42, also confirmed by the calculated PDOS of PtNP and PtSA on Cs2SnI6 (Fig. 5c). The most 5d states of PtSA are below the Fermi level, indicating its electron-rich state. The integrated PDOS areas of uncaptured Pt 5d states above the Fermi level in PtNP and PtSA are calculated to be 1.19 and 0.71, respectively, which quantifies the different electron-saturation levels of PtNP and PtSA. This implies the relative electron-deficient property of PtNP. This means that the PtSA species owns a stronger ability for capturing electrons from the Cs2SnI6, leading to the higher hydrogen production activity of PtSA/Cs2SnI6. Thus the different electronic properties of PtNP and PtSA on Cs2SnI6 lead to their different catalytic dynamics in the H2 evolution process, in which the PtSA owns a remarkably lower energy barrier (0.11 eV) than PtNP (0.92 eV) (Fig. 5d). This well explains the observably higher photocatalytic H2 production activity of PtSA/Cs2SnI6.

Fig. 5: Charge density distribution and Gibbs energy calculations.
figure 5

The charge density difference maps between before and after photoexcitation: a PtNP/Cs2SnI6 and b PtSA/Cs2SnI6. The isosurface of charge density is 0.001e Å−3. The insets stand for the top view. The yellow region represents the additional electron distribution. An excess electron was added into the models, which was used to describe the photogenerated electron. c The PDOS (5d states) of PtNP/Cs2SnI6 and PtSA/Cs2SnI6. The dashed line stands for the Fermi level. d The calculated energy profile for hydrogen production on PtNP/Cs2SnI6 and PtSA/Cs2SnI6.

Full size image

Discussion

In summary, we report a class of all-inorganic perovskite PtSA/Cs2SnI6 single-atom photocatalyst for achieving highly efficient photocatalytic hydrogen production in aqueous HI solution. The HAADF-STEM, XAFS spectroscopy, and XPS spectroscopy results confirm the atomically dispersed Pt single atoms with the well-defined Pt–I3 structure on Cs2SnI6. By combining charge-carrier dynamics studies and DFT calculations, we discover that the unique coordination structure and electronic property of Pt–I3 species can contribute to the SMSI effect, boosting the photogenerated electron transfer from Cs2SnI6 to Pt single atoms, and simultaneously reducing the Gibbs free energy and accelerating the kinetics for the hydrogen production. Benefiting from these structural advantages, an outstanding TOF of 70.6 h−1 per Pt was achieved over PtSA/Cs2SnI6 catalyst, 176.5-fold higher than that of PtNP/Cs2SnI6, setting a new TOF record reported for Pt-loaded halide perovskite photocatalysts, along with an excellent cycling stability. The achievements herein can significantly stimulate the exploitation of novel metal single atoms-perovskite hetero-structured photocatalysts systems and their further sustainable photocatalytic applications.

Methods

Synthesis of PtSA/Cs2SnI6

As-prepared Cs2SnI6 powder (0.10 g) was dispersed in 50 mL of chloroform containing different amount of Pt(acac)2. Then, the mixture was further ultrasonicated for 30 min and stirred for 12 h. Subsequently, the obtained precipitates were centrifugalized and washed for three times with isopropanol, and further dried in an oven 60 °C for 2 h. Finally, the PtSA/Cs2SnI6 was obtained by the treatment in a tube furnace at 160 °C for 1 h under a 5% H2/Ar atmosphere. The loading content of Pt on PtSA/Cs2SnI6 is determined by ICP-AES.

XAFS characterization and data analysis

The Pt L3-edge XAFS spectra were performed at the 1W1B beamline of Beijing Synchrotron Radiation Facility (Beijing), operating at 2.5 GeV with a ring current of 250 mA. The X-ray beam was monochromatized by a Si (111) double crystal monochromator. The ATHENA module of the IFEFFIT software packages was used as the standard procedure to process the acquired EXAFS date. The k3-weighted χ (k) data of Pt K-edge in the k-space (2.0–12 Å−1) were FT to real (R) space by a handing windows (dk = 1.0 Å−1) to separate the EXAFS contribution from different coordination shells. The quantitative curve-fitting was performed by using the ARTEMIS module of IFEFFIT3 to obtain the elaborate structural parameters around Pt central atom in the as-synthesized catalysts. The functions of effective curved-wave backscattering amplitudes F(k) and phase shifts Φ(k) were calculated by the ab initio code FEFF8.0. Based on the fitting of reference samples of metal Pt bulk and PtI2 bulk, S02 (amplitude reduction factor) was fixed to the best-fit value of 0.70. The interatomic distance (R) and the Debye–Waller factor (σ2) were allowed to change during the fitting analysis. The coordination of Pt–I was distinguished from Pt–Pt according to the bond length difference.

Photocatalytic H2 evolution activity

The photocatalytic H2 evolution experiments in aqueous HI solution (containing 20 vol% H3PO2 as a stabilizer) were executed in a homemade double-layered Pyrex vessel. A 300 W Xe lamp with a visible-light illumination (λ ≥ 420 nm, 100 mW cm−2) was employed as a light source for the photocatalytic reaction. In a typical photocatalytic H2 evolution procedure, 10 mg of as-synthesized photocatalyst was introduced into 10 mL aqueous HI solution containing 20 vol% H3PO2 with the constant stirring rate, and then degassed with Ar through the reactor for 30 min to completely remove the dissolved air before irradiation. The reaction temperature was preserved at 25 °C by a circulation cooling water. The amount of evolved H2 was detected every hour in a 6 h test by gas chromatography (GC-7890B, Agilent, America, TCD, with MS-5 Ã… molecular sieve column) with Ar as the carrier gas. The recycling photocatalytic experiment for the stability test of as-prepared photocatalyst was carried out every 6 h as a cycle.

Theoretical calculations

The photocatalytic property of PtSA/Cs2SnI6 was investigated by the Vienna Ab initio Simulation Package. The PAW pseudo-potentials were used to describe the interaction between valence electrons and the ionic core. The energy profile for hydrogen production was calculated by the revised Perdew–Burke–Ernzerh functional of the generalized gradient approximation. The reported standard hydrogen electrode (SHE) model was used in the calculations of Gibbs free energy changes (ΔG) in hydrogen adsorption44. In this model, the chemical potential (µ(H+) + µ(e−)) of a proton–electron pair was equal to half of the chemical potential (µ(H2)) of one gaseous hydrogen at U = 0 V versus SHE at pH = 0. The surface of Cs2SnI6 was simulated by its typical {111} facet, which was described by a 2 × 2 supercell with three Sn layers and six Cs–I layers (equal to three CsI–Sn–CsI layers). The PtSA was simulated by coordinating with three surface I atoms according to the above experimental results. The PtNP was described by a cluster containing 6 Pt atoms. The vacuum thickness was set to 16 Å. The larger supercells and higher vacuum thickness were tested, showing little effect on the calculation of hydrogen adsorption, as shown in Supplementary Table 3. A plane-wave basis with energy cutoff of 400 eV and an energy convergence threshold of 1.0 × 10−5 eV were used to perform the geometry optimization at the gamma point. After geometry optimization, the projected density of states of PtNP/Cs2SnI6 and PtSA/Cs2SnI6 models is calculated with the energy convergence of 1.0 × 10−5 eV and the Monkhorst–Pack k-point mesh of 2 × 1 × 1. To track the transfer of photogenerated electron between the Pt and photocatalyst, an extra electron with a compensating uniform background charge was used to simulate the photogenerated electron43. The energy convergence of 1.0 × 10−5 eV and the Monkhorst–Pack k-point mesh of 2 × 1 × 1 were adopted to calculate the difference of charge density plots between PtNP/PtSA and Cs2SnI6. Besides, to exclude the background charge effect caused by the extra electron, the charge difference density maps between PtNP/PtSA and Cs2SnI6 was also calculated by inserting a donor hydrogen atom into the Cs2SnI6 bulk, which can provide one proton and one electron43.