Precise fabrication of single-atom alloy co-catalyst with optimal charge state for enhanced photocatalysis

Abstract While the surface charge state of co-catalysts plays a critical role for boosting photocatalysis, studies on surface charge regulation via their precise structure control remain extremely rare. Herein, metal-organic framework (MOF) stabilized bimetallic Pd@Pt nanoparticles, which feature adjustable Pt coordination environment and a controlled structure from core-shell to single-atom alloy (SAA), have been fabricated. Significantly, apart from the formation of a Mott-Schottky junction in a conventional way, we elucidate that Pt surface charge regulation can be alternatively achieved by changing its coordination environment and the structure of the Pd@Pt co-catalyst, where the charge between Pd and Pt is redistributed. As a result, the optimized Pd10@Pt1/MOF composite, which involves an unprecedented SAA co-catalyst, exhibits exceptionally high photocatalytic hydrogen production activity, far surpassing its corresponding counterparts.


INTRODUCTION
Photocatalysis, converting solar to chemical energy, represents a very promising solution to current energy and environmental issues [1][2][3][4]. Photocatalytic performance depends largely on the surface charge state of a catalyst, as it directly correlates with charge transfer from the catalyst surface to reactant molecules as well as adsorption and activation of the latter [5][6][7]. Various strategies have been developed to control the surface charge states of photocatalytic materials [6,7]. Amongst them, the construction of a Schottky junction by introducing co-catalyst is one of the most common and effective strategies via increasing the active sites, promoting charge separation, accelerating charge transfer and minimizing reaction over-potentials [8][9][10][11][12]. Recent studies on co-catalysts are mostly related to the development of non-noble metal co-catalysts, the control of particle sizes, particle distribution, exposed crystal facets as well as their interface contact with photocatalysts, and even synergistic effect among various co-catalysts [6,[8][9][10][11][12]. Nonetheless, specialized regulation on the surface charge state of co-catalysts, especially by changing their microstructure, remains largely unexplored, though this would be highly desired to boost the photocatalysis [6,10,12].
To achieve surface charge regulation, the introduction of bimetallic nanoparticles (NPs) as cocatalysts, in which one metal behaves as the active site and the other as charge regulator, would be a promising solution. Of the common alloyed or coreshell bimetallic NPs, the core-shell architecture featuring active metal shell would be preferred, for ease of access to active sites. The gradually decreased content of active metal would cause its varying coordination environment and surface charge. Particularly, atomically dispersed active metal atoms will be formed when their content goes down to a certain amount. The microstructure of active metal in its atomically dispersed form on the other metal support is called single-atom alloy (SAA) [13][14][15][16][17]. Compared with other single-atom catalysts (SACs) [18,19], SAA catalysts present geometric/strain, electronic and synergistic effects originating from bimetallic components [13][14][15][16][17]20,21], causing the d-band center change of the surface active metals Figure 1. Schematic illustration. Illustration showing the synthetic strategy for Pd 10 @Pt x /UiO-66-NH 2 toward photocatalytic hydrogen production, highlighting the much higher activity of Pd 10 @Pt 1 SAA than Pd 10 @Pt 10 core-shell NPs. and thereby affecting their catalytic performance [13][14][15][16][17]. Surprisingly, the related exploration on SAA catalysts toward their photocatalysis has not yet been reported. In this context, it is imperative to investigate the structural evolution of bimetallic NPs from core-shell to SAA, which might be able to systematically regulate the charge state of surface active metal atoms, for boosting photocatalysis.
Bearing the above in mind, bimetallic coreshell Pd@Pt NPs have been in situ fabricated and stabilized by a representative robust MOF with good light harvesting capability, UiO-66-NH 2 , to afford Pd 10 @Pt x /UiO-66-NH 2 composite (the 10 and x mean the mass permillage, at wt , of Pd and Pt to the MOF in synthesis, respectively). Remarkably, the Pt component on the Pd surface can be precisely controlled from the shell-form to single-atom dispersion. Along with structural evolution of Pd@Pt NPs from core-shell to SAA, the Pt coordination environment changes and the regulation of its surface charge state is accordingly achieved. The obtained Pd 10 @Pt x /UiO-66-NH 2 exhibits excellent photocatalytic activity toward hydrogen production by water splitting (Fig. 1). Particularly, thanks to the charge redistribution, the optimized Pd 10 @Pt 1 /UiO-66-NH 2 with SAA structure possesses superb activity, far surpassing the monometallic NPs and their physical mixture stabilized by the MOF. Though both size and location of Pt NPs have been reported to be crucial for charge separation efficiency and activity [9,11], to the best of our knowledge, there has not yet been explored on regulating the Pt coordination environment and its surface charge state by controlling co-catalyst structure in the photocatalytic system. Moreover, this is the first work investigating the SAA catalyst for photocatalysis.
Powder X-ray diffraction (PXRD) patterns indicate that the high crystallinity of UiO-66-NH 2 is well maintained during the introduction of Pd@Pt NPs ( Supplementary Fig. 1). The actual contents of Pd and Pt are determined by the inductively coupled plasma atomic emission spectrometer (ICP-AES) (Supplementary Table 1). All Pd 10 @Pt x /UiO-66-NH 2 have similar Pd loadings, while the Pt steadily increases as more Pt precursors are introduced. Transmission electron microscopy (TEM) observation shows that Pd NPs in Pd 10 /UiO-66-NH 2  are well dispersed with average sizes of 4.01 nm ( Supplementary Fig. 2). Upon Pt deposition, the Pd@Pt sizes slightly increase with higher Pt contents ( Fig. 2a  To verify whether the Pd@Pt core-shell structure has been successfully constructed, multiple characterizations have been adopted by taking Pd 10 @Pt 10 /UiO-66-NH 2 as a representative. A high-resolution TEM (HRTEM) image reveals the Pd core is covered by a Pt shell, as evidenced by their exposed {111} facets with clear lattice fringes (Fig. 2b). The spatial distribution of the constituent elements of Pd@Pt NPs is detected by synchrotron radiation photoemission spectroscopy (SRPES) spectra for Pd 3d and Pt 4f, at photon energies of 450, 710, 970 and 1253.6 eV, respectively (Supplementary Figs 8-10) [56,57]. The peaks at 340.3 and 335.0 eV can be assigned to Pd 3d 3/2 and Pd 3d 5/2 of metallic Pd 0 , respectively [56], whereas the peaks at 74.3 and 70.9 eV clearly correspond to Pt 0 ( Supplementary Fig. 10) [58,59]. The outside-in atomic fractions of Pd and Pt in Pd@Pt NPs can be obtained from their peak areas calibrated by the beam flux and photoionization cross-section at the corresponding photon energy. As a direct result, a decreased atomic fraction of Pt and an increased atomic fraction of Pd unambiguously suggest that Pt stays on the external surface of Pd, pointing to a Pd 10 @Pt 10 core-shell structure (Fig. 2c). More intuitively, the core-shell structure can also be clearly identified by elemental line-scanning spectra for a single Pd 10 @Pt 10 particle ( Fig. 2d and Supplementary Fig. 11).
Furthermore, the thickness of the Pt shell can be precisely controlled by varying the ratio of Pd/Pt precursors. As the Pt content decreases to give Pd 10 @Pt 1 /UiO-66-NH 2 , strikingly, the high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) image indicates that Pt is atomically dispersed on the Pd surface (Fig. 3a). Fourier transformed extended X-ray absorption fine structure (FT-EXAFS) and X-ray absorption near-edge structure (XANES) analyses provide more detailed structural information on Pd@Pt NPs as well as the Pt coordination environment (Fig. 3b, Supplementary Fig. 12 and Supplementary Tables 3 and 4). The absence of Pt-Pt peak in the Pt L 3 -edge FT-EXAFS spectrum of Pd 10 @Pt 1 /UiO-66-NH 2 further demonstrates that the Pt, in its single-atom form, locates on the Pd surface (Fig. 3b) [58,59]. The weak peak at 1.5Å is ascribed to the Pt-O bonding, which is possibly caused by partial surface oxidation. The slightly oxidized Pt also gives rise to a minor increase in white line intensity in the Pt L 3 -edge XANES spectrum for Pd 10 @Pt 1 /UiO-66-NH 2 (Supplementary Fig. 13) [58,59]. All the HAADF-STEM, FT-EXAFS and XANES results unambiguously confirm that the SAA structure is formed in Pd 10 @Pt 1 /UiO-66-NH 2 . In sharp contrast to Pd 10 @Pt 1 /UiO-66-NH 2 , the Pd 10 @Pt 10 /UiO-66-NH 2 apparently shows the presence of a Pt-Pd/Pt coordination bond. Moreover, the Pt-Pd coordination number in Pd 10 @Pt 10 /UiO-66-NH 2 is much lower than that in Pd 10 @Pt 1 /UiO-66-NH 2 ( Supplementary Tables 3 and 4), indicating that Pt possibly presents a cumulative atomic layer-like distribution on the Pd surface in the former sample. Therefore, the Pd@Pt NPs are successfully stabilized by UiO-66-NH 2 , and the coordination environment of Pt is adjustable along with precise regulation of Pt shells to single Pt atoms on the surface of Pd NPs.

Charge redistribution effect
Given the different working functions of Pd and Pt ( Supplementary Fig. 14), to balance the Fermi distribution of electrons in Pd@Pt NPs, electrons tend to leave Pd (the lower work function and electrons being less tightly bound) and travel to Pt (the higher work function) at their interface, causing an altered surface charge state of Pt. It is assumed that, as the structure of Pd@Pt NPs evolves from core-shell to SAA, charge redistribution (charge density change between the core and the surface metals) would take place and this affects the Pt surface charge state [12,60]. The work function calculations for Pd@Pt with SAA and core-shell structure support the above point ( Supplementary Fig. 14). To further verify this hypothesis, X-ray photoelectron spectroscopy (XPS) measurements have been conducted (Fig. 4). Compared to the Pt 0 4f 7/2 peak at 70.9 eV for Pt 1 /UiO-66-NH 2 , Pd 10 @Pt 1 /UiO-66-NH 2 presents lower binding energy to varying degrees, reflecting that the Pt surface in Pd@Pt NPs becomes electron-rich and charge redistribution effect between Pd and Pt exists. Moreover, when less Pt x (x value from 10 to 1) is covered, the Pt coordination number increases via surrounding Pd-Pt bonds (Supplementary Tables 2-4) and the Pt binding energy decreases (Fig. 4), further manifesting gradually enhanced charge redistribution effect and the most electron-rich Pt in Pd 10 @Pt 1 NPs. In fact, from the aforementioned SRPES spectra of Pt 4f ( Supplementary Fig. 10), when the probe depth gets close to the Pd core as photon energy increases, the Pt binding energy appears as a negative shift, which further supports the above conclusion.
In addition, the differential charge density of Pd@Pt NPs (from core-shell structure with different Pt thicknesses to SAA) based on densityfunctional theory (DFT) calculations affords consistent results, manifesting the enhanced charge accumulation on the Pt surface along with the decrease of Pt layer numbers (i.e. from core-shell to SAA structure, finally) (Fig. 5, red color). From the differential charge density, the Bader charge of each Pt atom and the binding energy shift of Pt 4f 7/2 show similar slopes along with the increased Pt layer thickness, indicating the consistent charge redistribution trend in simulation and experiment ( Supplementary Fig. 15, Supplementary  Tables 5 and 6).

Photocatalysis and photoelectrochemical measurements
Encouraged by the electron-rich Pt surface in Pd 10 @Pt x /UiO-66-NH 2 , we set out to investigate the influence of charge redistribution for photocatalytic H 2 production (Fig. 6a). Prior to the measurement, UV-Vis absorption spectra of Pd 10 @Pt x /UiO-66-NH 2 have been examined to demonstrate their similar light absorption (Supplementary Fig. 16). As expected, as the Pt loading in Pd@Pt NPs decreases (x from 10 to 1), the Pt-Pd coordination number increases, charge redistribution effect strengthens and Pt becomes more electron-rich. Accordingly, the activity of Pd 10 @Pt x /UiO-66-NH 2 exhibits gradual increase. Strikingly, Pd 10 @Pt 1 /UiO-66-NH 2 featuring SAA structure achieves the highest photocatalytic H 2 production rate, 1200.5 μmol · g −1 · h −1 , among all investigated samples, which is 25, 30 and ∼4.5 times higher than that of Pd 10 /UiO-66-NH 2 , Pt 1 /UiO-66-NH 2 , Pd 10 + Pt 1 /UiO-66-NH 2 with a mixture of monometallic NPs, and Pd 10 Pt 1 /UiO-66-NH 2 , respectively ( Fig. 6a and Supplementary Figs 17-20). When the Pt content continues to decline (x from 1 to 0.3), unexpectedly, the photocatalytic hydrogen production activity slightly decreases, possibly due to the decreased Pt sites in the catalyst.
The calculated Gibbs free energy ( G H * ) of Pt and Pd atoms in the Pd 10 @Pt 1 SAA structure indicates that Pt is the more favorable active site than Pd for hydrogen production (Supplementary Fig. 21). Together with the overwhelmingly higher activity of Pd 10 @Pt 1 /UiO-66-NH 2 than Pd 10 /UiO-66-NH 2 , the H 2 generation taking place on Pd should be negligible and turnover frequencies of Pt (TOF Pt ) are a reasonable descriptor for the activity for the catalysts in SAA structure. The significantly higher TOF Pt for Pd 10 @Pt 0.3 /UiO-66-NH 2 and Pd 10 @Pt 1 /UiO-66-NH 2 than all other counterparts further indicates the superiority of SAA structure in photocatalysis (Supplementary Table 7).
In addition, this optimized Pd 10 @Pt 1 /UiO-66-NH 2 catalyst shows good recyclability and its activity does not present obvious decay in five consecutive runs and a continued test for 25 h (Fig. 6b, Supplementary Fig. 25). Thanks to the good stability, PXRD patterns confirm that the structural integrity and crystallinity of the MOF in all samples are well retained after catalysis ( Supplementary Fig. 22). Especially, both the SAA structure and the dispersion of Pd 10 @Pt 1 NPs are almost maintained, thanks to the excellent stabilization effect by the MOF (Sup- plementary Figs 23 and 24).
The activity promotion by surface charge regulation based on Pd 10 @Pt x /UiO-66-NH 2 for photocatalysis has also been demonstrated by enhanced charge transfer. Photoluminescence (PL) emission spectra for all samples show that Pd 10 @Pt 1 /UiO-66-NH 2 has the weakest fluorescence intensity, indicating its best charge separation efficiency, and the peak intensity order is in perfect reverse correlation with that of photocatalytic activity (Fig. 6c). The photocurrent and electrochemical impedance spectroscopy (EIS) measurements also manifest such a trend ( Fig. 6d and Supplementary Fig. 26), further confirming the discriminative charge separation efficiency and being in good agreement with the results by XPS, SRPES spectra and DFT calculations, and well explaining the activity trend. Moreover, the cathodic polarization curves indicate Pd 10 @Pt 1 / UiO-66-NH 2 has lower over-potential for hydrogen evolution (Supplementary Fig. 27). Therefore, the regulation of Pt surface charge state by charge redistribution at the interface of Pd and Pt plays an important role in optimizing photocatalytic performance. Base on the results above and previous reports [9,48,49], the photocatalytic process of Pd@Pt/UiO-66-NH 2 is clear that the MOF is excited by visible light to produce photogenerated electron-hole pairs. Due to the Schottky junction, the electrons transfer to Pd, and finally to Pt (catalytic center), promoted by the charge redistribution effect between Pd and Pt (Fig. 1). The charge redistribution effect is influenced by the Pt coordination environment and microstructure, which causes distinct charge accumulation on the Pt surface in Pd 10 @Pt x /UiO-66-NH 2 , resulting in different H 2 production activity. The results unambiguously manifest that, in addition to the well-established Schottky junction, the microstructure regulation of a co-catalyst reported herein might be another effective and novel strategy to enhance charge separation and thus photocatalysis.

DISCUSSION
In summary, a bimetallic NPs/MOF system, Pd@Pt/UiO-66-NH 2 , has been rationally fabricated to investigate the structural influence of a Pd@Pt co-catalyst on its surface charge state and resulting photocatalytic activity. Remarkably, the Pt shell on Pd surface can be precisely controlled and the structure evolves from core-shell to SAA, so as to change the Pt coordination environment. Consequently, the surface charge on Pd and Pt is redistributed, which regulates the electronic state of Pt active sites and is found to be crucial for the resulting activity. As a result, the photocatalytic hydrogen production activity of Pd 10 @Pt x /UiO-66-NH 2 can be optimized to exponentially increase by precisely fabricating the Pd 10 @Pt 1 SAA co-catalyst, based on the rational control of Pt loading. This work not only presents SAA behaving as a very promising co-catalyst for photocatalysis for the first time, but also opens a novel avenue to the surface charge regulation of co-catalysts for boosting photocatalysis.

Preparation of Pd 10 @Pt x /UiO-66-NH 2
Typically, 100 mg UiO-66-NH 2 was put in a porcelain crucible and a certain amount of 50 mg/mL Pd(NO 3 ) 2 aqueous solution (Pd/UiO-66-NH 2 = 1 wt%) was added. Then, 200 μL MeOH was added into the mixture and rapidly stirred with a glass rod and heated at 80 • C to evaporate the solution. This procedure was repeated twice and the obtained solid was dried in a 85 • C drying oven for 15 min to give Pd 2+ /UiO-66-NH 2 . After that, a 15 mg sample of Pd 2+ /UiO-66-NH 2 was dispersed in 5 mL MeOH by ultrasonication, and 5 mg NH 3 BH 3 was added during stirring. After H 2 bubbling was completed, a certain amount of H 2 PtCl 6 aqueous solution (the mass ratio of Pt/Pd = 0.3, 1, 5, 10/10, respectively) was quickly added and stirred for 1 h. The solid was collected by centrifugation and washed by MeOH several times, and dried at 60 • C under vacuum overnight. The product was named Pd 10 @Pt x /UiO-66-NH 2 (the 10 and x mean the mass permillage, at wt , of Pd and Pt to the MOF in the synthesis, respectively).

Synchrotron radiation photoemission spectroscopy (SRPES)
The SRPES spectra were measured with photon energies of 450 eV, 710 eV, 970 eV and 1253.6 eV at the BL10B beamline in the National Synchrotron Radiation Laboratory (NSRL, Hefei, China) [56,57]. Before the SRPES test, 30 mg Pd 10 @Pt 10 /UiO-66-NH 2 was dispersed in 1 mL H 3 PO 4 and stirred for 2 h at room temperature to dissolve the MOF to improve the electrical conductivity of the sample.

The X-ray absorption spectra
The Pt L 3 -edge X-ray absorption spectra (XAS) data were collected at 1W1B beam line of the Beijing Synchrotron Radiation Facility (BSRF, Beijing) operated at 2.5 GeV at room temperature. The Si (111) double crystal monochromator was calibrated via Pt foil and then the Pt foil XAS data was collected as the reference spectrum. The XAS measurement for the samples was performed in the fluorescence mode using a Lytle detector. Data analysis was performed with Athena and Artemis included in the IF-EFFIT version 0.8.012. A first-shell single scattering path was used in fitting the FT-EXAFS data.

Photocatalytic experiments
Typically, 3 mg photocatalyst was dispersed in 27 mL acetonitrile containing 0.2 mL deionized water and 3 mL triethylamine (as a sacrificial reagent). Then the suspension was transferred into an optical reaction vessel (160 mL) and purged with nitrogen for 15 min to remove air. The reaction solution was irradiated by the 300 W Xe lamp (LX-300F, Japan) equipped with a UV cut-off filter (>380 nm). Hydrogen gas was measured by gas chromatography (Shimadzu GC-2014) using a thermal conductivity detector.