Synthesis and characterization of Pt 2 /mpg-C 3 N 4 samples. The Pt2/mpg-C3N4 sample was synthesized using the wet-chemical strategy. (Ethylenediamine)iodoplatinum(II) dimer dinitrate and mesoporous graphitic carbon nitride (mpg-C3N4) were selected as the dual-atomic Pt precursor and the substrate. They were mixed and further pyrolyzed to remove the ligands from the dual-atom Pt precursor. The Pt1/mpg-C3N4 and the Pt nanoparticle/mpg-C3N4 samples were synthesized using the same method, except the Pt species being replaced by H2PtCl6 and Pt nanoparticles, respectively.
The X-ray diffraction (XRD) pattern (Supplementary Fig. S1) demonstrated that the synthesized mpg-C3N4 sample has a graphitic packing structure,45,46 and the disordered spherical pores of mpg-C3N4 were captured by the transmission electron microscopy (TEM) image (Supplementary Fig. S2). Upon the loading of the dual-atom Pt precursor, neither Pt nanoparticles nor nanoclusters were observed in the TEM (Supplementary Fig. S3) and HAADF-STEM images (Fig. 1a). Moreover, no additional diffraction peaks of Pt lattices were found in the XRD pattern (Supplementary Fig. S1). The energy-dispersive X-ray (EDX) spectroscopy further demonstrated a homogeneous distribution of the Pt species (Fig. 1b). The inductively coupled plasma optical emission spectrometry indicated that the content of Pt is 0.15 wt%. All the above results indicated that dual-atom Pt had been homogeneously dispersed on the mpg-C3N4 substrate. After the pyrolysis procedure, no infrared spectroscopy absorption peaks that correspond to the ligands of the precursor were observed in the Pt2/mpg-C3N4 sample, which supported a complete removal of the ligand molecules (Supplementary Fig. S4). To further confirm the dual-atomic feature of the Pt species, aberration-corrected (AC) HAADF-STEM was applied to characterize the Pt2/mpg-C3N4 sample. Many paired bright dots (marked with white circles) were observed in the AC HAADF-STEM image, which is consistent with the feature of two Pt atoms (Supplementary Fig. 1c). Besides, a few isolated bright dots (marked with green circles) were also observed, which we attributed to an overlap of paired Pt dots in the incident beam direction or an incomplete imaging of the dual-atom Pt species due to incomplete focusing (Supplementary Fig. 1c). The detailed features of Pt2 dual-atom are different from each other depending on their orientations in three dimensions because the AC STEM image just showed a two-dimensional projection from the three-dimensional Pt2/mpg-C3N4 samples.47 For comparison, only isolated bright dots appeared in the corresponding Pt1/mpg-C3N4 image, further confirming the sharply difference between dual-atom and single-atom Pt species. (Fig. 1d and Supplementary Fig. S5-7).
X-ray absorption fine structure spectroscopy, which is a powerful technique for determining the chemical state and coordinated environment, was also applied to characterize the Pt species. The Pt L3-edge X-ray absorption near-edge structure (XANES) spectra of the Pt2/mpg-C3N4 and Pt1/mpg-C3N4 samples, as well as the corresponding references, are shown in Fig. 2a. Here, the white line intensity peak of dual-atom Pt2 is located between those of Pt foil and PtO2, indicating that the two Pt atoms possess positive charges. This can be attributed to the strong interaction between the dual-atom Pt species and the mpg-C3N4 substrate or partial oxidation of dual-atom Pt by the O2, which is similar to the case of Pt1/mpg-C3N4. The Fourier-transformed (FT) k3-weighted extended X-ray absorption fine structure (EXAFS) spectra of Pt2/mpg-C3N4 showed a sharp peak located at 1.62 Å, which is also similar to the result of Pt1/mpg-C3N4 and can be assigned to the Pt-N/O contributions (Fig. 2b). For Pt2/mpg-C3N4, another distinct peak at 2.44 Å was found, similar to the Pt-Pt path of Pt foil but not observed in the spectrum of the Pt1/mpg-C3N4 sample. It reveals that the Pt-Pt path should also be taken into account in the spectrum of Pt2/mpg-C3N4. Wavelet Transforms (WT) EXAFS analysis is powerful method to discern the scattering atoms because it can provide both R-space and k-space resolutions. WT EXAFS analysis results show that the WT EXAFS spectrum of dual-atom Pt2 shows maximum at 4.6 Å−1 in k-space and 1.6 Å in R-space, corresponding to the Pt-O and Pt-N bond. Besides, two distinct peaks are observed at ~ 2.5 Å in R-space, which refers to the Pt-O (oxygen at second shell, k = 3.7 Å−1) and Pt-Pt (k = 8 Å−1). Therefore, the structure of Pt2 consists of a Pt-Pt bond with surrounding O attached (Supplementary Fig. 2e). The EXAFS fitting results further showed that the first peak at 1.62 Å comes from the Pt–N/O contributions and the second one at 2.44 Å is from the Pt-Pt path (Fig. 2c-2d, Supplementary Fig. S8-9, and Supplementary Table 1). The coordination number of the Pt–N/O path was estimated to be 2.4 at the distances of 2.02 Å, and the second coordination sphere by the Pt-Pt path was assessed to 1.1 with the corresponding distance to be 2.61 Å.
The configuration of the Pt2/mpg-C3N4 sample was also explored by extensive first-principles calculations, with the optimized geometry shown in the inset of Fig. 2c. Firstly, we considered various kinds of Pt2/g-C3N4 structures without oxygen atoms (Supplementary Fig. S10) in the simulations, but none of them matched the XAFS information. It is worth noting that, since the measurement of the XAFS spectra were performed in air, the oxygen molecules contained in the atmosphere had interacted with and attached to the Pt species, which is also consistent with the WT EXAFS analysis result (Supplementary Fig. 2e). Such oxygen atoms, however, will be removed by hydrogen molecules at the initial stage of the following hydrogenation reaction. Furthermore, we explored the possibility structures of dual-atom Pt2 with the oxygen atoms (Fig. 2f and Supplementary Fig. S11). Fortunately, the Pt-N/O and the Pt-Pt bond lengths were calculated to be 1.96 ± 0.14 Å and 2.55 Å, with the corresponding coordination numbers being 2.5 and 1.0 based on the model in Fig. 2c. The information agrees well with results from the XAFS data.
Hydrogenation of nitrobenzene to aniline. The catalytic performance of Pt2/mpg-C3N4, Pt1/mpg-C3N4, and Pt nanoparticles (~ 2 nm)/mpg-C3N4 (Supplementary Fig. S12) were then investigated. Under the conditions of 4 MPa H2 pressure and 100℃, a conversion of ˃99% was obtained on the Pt2/mpg-C3N4 catalyst for the hydrogenation of nitrobenzene to aniline, while no by-product was detected (Fig. 3a). Such Pt2/mpg-C3N4 sample can be reused at least five times without any loss of the activity (Fig. 3b). After five cycles, the HAADF-STEM image and EXAFS spectrum of the Pt2/mpg-C3N4 material did not exhibit any change, indicating that the Pt species were still well dispersed as dual-atom Pt pairs (Supplementary Fig. S13-14). It may be worth mentioning that the mpg-C3N4 support is reactively inert under the same conditions. By contrast, the corresponding conversions of Pt1/mpg-C3N4 and Pt nanoparticles/mpg-C3N4 sharply dropped to 23% and 12%, respectively, demonstrating the uniqueness of the dual-atom Pt species in the catalytic properties (Fig. 3a). To investigate whether the outstanding catalytic performance is general for the hydrogenation of nitroarenes, we have explored the hydrogenation of several other nitroarene derivatives, including p-nitrophenol, p-nitrotoluene, tetrachloro-nitrobenzene, and tetrabromonitrobenzene. We found that Pt2/mpg-C3N4 exhibits excellent yields for all corresponding anilines (Fig. 3c).
The superior catalytic performance of Pt2/mpg-C3N4 compared with those of the Pt1/mpg-C3N4 sample and the Pt nanoparticles was, according to our first-principles simulations, attributed to the formation of the Pt-O chemical bonds between the diatomic Pt2 species and the nitro groups, which brings about effective activation of the N-O bonds. In Fig. 4, we present the adsorption configurations of nitrobenzene on Pt2/g-C3N4 and Pt1/g-C3N4 as well as on a Pt(111) surface (to represent the outermost layer of Pt nanoparticles). One can see that on Pt2/g-C3N4, the two oxygen atoms of the nitro group can form chemical bonds with the two Pt atoms, while on Pt1/g-C3N4 and Pt(111), the nitro group becomes far away from the Pt atoms. Only in the former case, the N-O bond lengths exhibit a significant elongation from 1.25 Å (the corresponding value of an isolated nitrobenzene molecule) to 1.35 Å, which makes the N-O bond rupture become much easier in the subsequent hydrogenations.
In the reduction of nitrobenzene to aniline, the detailed processes of the N-O bond cleavage and the N-H bond formation are displayed in Fig. 5. As we have expected based on the elongation of the N-O bonds, the N-O rupture is easy to occur (S2), during which a barrier of only 0.60 eV (TS1) is required to overcome. After that, the first H2 molecule approaches the N and the O atoms (S3) and reacts with them via the Eley–Rideal (ER) mechanism (TS2), producing the first N-H bond and a hydroxyl group (S4). Upon the elimination of the hydroxyl group by a second H2 molecule (not shown in Fig. 5), the second N-O bond is broken in the same way (S6), showing a barrier of 0.61 eV (TS3). As the third H2 molecule participates in the reaction through the same ER mechanism (S7 and TS4), the second N-H bond is formed, corresponding to the production of aniline (S8). After desorption of the aniline product and recombination of the remaining hydroxyl group and H atom, as well as adsorption of another nitrobenzene molecule (S1), a new round of the catalytic cycle will start again.
Hydrogenation of benzaldehyde and epoxidation of alkenes.
The produced Pt2/mpg-C3N4 catalyst is versatile and can be employed in other important reactions besides the selective hydrogenation of nitrobenzene. For example, under the conditions of 4 MPa H2 pressure and 120℃, the Pt2/mpg-C3N4 sample showed optimal catalytic performance toward the hydrogenation of benzaldehyde to benzyl alcohol (Fig. 6a). Specifically, ˃99% conversion and ˃99% selectivity were achieved for 7 hours. It means that the aforementioned activation pattern of Pt2 can be extended to activating substituted C = O bonds near benzene rings (Supplementary Fig. 15). Moreover, the catalyst can be reused at least three times without any loss of the activity (Fig. 6a). As another type of significantly important reactions,48 the epoxidation of alkenes in liquid relies on extensive use of expensive oxidants or co-reagents, which leads to a large increase in the cost. Our prepared Pt2/mpg-C3N4 catalyst exhibited excellent catalytic performance toward the epoxidation of styrene when only O2 molecules were used as the oxidant, which well circumvents the above problem. In Fig. 6b, one can see that Pt2/mpg-C3N4 exhibited a conversion of 93% and a selectivity of 78% after 12 h, and this is one of the best results for the epoxidation of styrene.49–50 Here, the one-coordinated oxygen atom on the Pt2 species (Fig. 2f) plays an important role in the catalytic reaction, which is reminiscent of the diatomic Fe2 system in catalyzing the epoxidation of trans-stilbene.26