CoSe2 supported single Pt site catalysts for hydrogen peroxide generation via two‐electron oxygen reduction

Electrocatalytic oxygen reduction reaction (ORR) to prepare H2O2 in acidic medium has the advantages of green, safety, and portability, which shows broad development prospects. However, it still suffers from low catalyst activity, insufficient selectivity, and high cost. Herein, Pt1/CoSe2 with ultralow 0.01 wt.% Pt atomic distribution was synthesized by a simple hydrothermal method. The Pt1/CoSe2 with ultralow Pt content exhibits high activity, high selectivity, and long‐term stability for ORR to H2O2 in O2‐saturated 0.1 M HClO4. The onset potential is as low as 0.75 V versus reversible hydrogen electrode (RHE), H2O2 selectivity is as high as 84% (0.4 V vs. RHE), and the electron transfer number is 2.3 (0.4 V vs. RHE). Moreover, the hydrogen peroxide yield using the flow cell testing is 110.02 mmol gcat.−1 h−1 with high Faradaic efficiency of 78% (0 V vs. RHE) at 0.1 M HClO4, and the catalyst did not deactivate significantly after 60 h stability testing. Mechanistic studies and in situ X‐ray photoelectron spectroscopy characterization confirm that the ultralow Pt content on CoSe2 can effectively regulate the electronic structure of Co as the real active site around the Pt site, which gives a suitable ∆dp value (the difference between the d‐band center of the active metal site and the p‐band center of the terminal oxygen in *OOH), provides an ideal *OOH binding energy, and inhibits the O–O bond breakage. This work successfully improves the intrinsic activity of the Co active sites around Pt in Pt1/CoSe2 for acidic ORR to H2O2 by constructing ultralow‐content Pt single atom.


INTRODUCTION
Hydrogen peroxide (H 2 O 2 ), as an essential chemical, has a wide range of applications. The global demand for H 2 O 2 will exceed 6 × 10 6 t in 2024. [1][2][3][4][5][6] The traditional anthraquinone self-oxidation method or the direct synthesis of H 2 O 2 from H 2 and O 2 has problems, such as complex production process, high energy consumption, and low safety. [7][8][9][10][11][12] In recent years, electrocatalytic oxygen reduction reaction (ORR) to H 2 O 2 in acidic media has attracted much attention (O 2 + 2H + + 2e − → H 2 O 2 E • = 0.70 V vs. reversible hydrogen electrode [RHE]). 13,14 Compared with alkaline electrolytes, acidic electrolytes can effectively inhibit the self-decomposition of H 2 O 2 and have good compatibility with commercial proton exchange membrane electrolysis devices, which can significantly reduce production costs. [15][16][17][18][19] However, there are still some problems in the preparation of H 2 O 2 from ORR in acidic media, such as slow reaction kinetics, low efficiency, and high cost. It is urgent to develop efficient catalysts for electrocatalytic acidic ORR to H 2 O 2 .
Recently, the application of cobalt selenide (CoSe 2 ) in the preparation of H 2 O 2 from acidic ORR has attracted much attention due to its unique atomic and electronic structure. 20,21 Schmidt et al. found that CoSe 2 exhibited high ORR activity and selectivity for the preparation of H 2 O 2 in an acidic medium and can provide higher kinetic current density than noble metals or single-atom catalysts. 16 Gao et al. used the ion exchange method to reduce the interlayer spacing of layered CoSe 2 nanobelts to synthesize a strongly coupled catalyst, which showed good ORR activity and stability for H 2 O 2 preparation in acidic electrolyte. 22 The strong coupling between the atomic layers of CoSe 2 enables the essential reaction intermediate *OOH to have suitable binding energy. However, the existing CoSe 2 catalysts still have some problems such as insufficient active sites and too strong or too weak binding energy of the critical reaction intermediate *OOH, resulting in the H 2 O 2 synthesis still not meeting the requirements of large-scale production. Therefore, precisely controlling the structure of CoSe 2 to achieve the exposure of more active sites and ensure that *OOH has a suitable binding energy becomes the key to realizing the efficient reaction of acidic ORR to H 2 O 2 .
The noble metal Pt can efficiently catalyze ORR. [23][24][25][26][27] However, most Pt catalysts have low selectivity for H 2 O 2 due to the over-activation of *O 2 by the adjacent "Pt-Pt" atoms, which directly reduces O 2 to the undesired H 2 O via the 4e − pathway. 28,29 In addition, the low utilization of Pt atoms limited its application. 30 Compared with traditional nanoparticle and cluster catalysts, single-atom catalysts with unsaturated coordination sites and unique electronic structures have nearly 100% atomic utilization, which can effectively improve the intrinsic catalytic activity and selectivity of active sites, and regulate the interaction of catalyst active sites and reaction intermediates to improve catalytic reaction kinetics. [31][32][33][34] Studies have shown that when the size of Pt is changed from nanoparticles and atomic clusters to single-atomic dispersion, the ORR reaction changes from the 4e − pathway to the 2e − pathway, which significantly improves the selectivity and reactivity of ORR to H 2 O 2 . 25,35 Herein, using a simple hydrothermal method, Pt 1 /CoSe 2 with ultralow 0.01 wt.% Pt atomic distribution was synthesized. The catalytic mechanism of Pt 1 /CoSe 2 is clarified by using DFT calculation and experiment, and the relationship between Pt 1 /CoSe 2 and the performance of acidic ORR to H 2 O 2 is discussed. The ultralow Pt content on CoSe 2 can effectively regulate Co's electronic structure around the Pt site with a suitable *OOH binding energy, and the Co around the Pt monoatomic site acts as the real active site for H 2 O 2 production. The H 2 O 2 yield using the flow cell is 110.02 mmol g cat.
−1 h −1 with a high Faradaic efficiency of 78% (0 V vs. RHE) at 0.1 M HClO 4 , and the catalyst did not deactivate significantly after 60 h stability testing. Importantly, we provide an essential activity descriptor of ∆ dp (defined as the difference between the d-band center of the active metal site and the p-band center of terminal oxygen in *OOH) to evaluate the activity of ORR to H 2 O 2 . This work helps to understand the nature of Pt 1 /CoSe 2 active sites during the preparation of H 2 O 2 from acidic ORR and provides an efficient catalyst with potential applications.

Synthesis and characterization of the Pt 1 /CoSe 2 with ultralow Pt content
To synthesize Pt 1 /CoSe 2 with ultralow Pt content, a hydrothermal followed by reduction calcination method was developed. A schematic diagram of the synthesis process is shown in Figure 1A. As shown in Figure 1B, the Pt 1 /CoSe 2 exhibits nanorod morphology with a size of 150 nm. The precise lattice fringe spacing of 0.26 nm in Figure 1C is assigned to the (1 1 1) plane of CoSe 2 , consistency with the X-ray diffraction (XRD) characterization. Inductively coupled plasma mass spectrometry (ICP-MS) confirmed that the Pt content in Pt 1 /CoSe 2 was 0.01 wt.%. The spherical aberration electron microscope characterization shows that the ultralow content of Pt is atomically distributed on the CoSe 2 surface ( Figure 1D). From the energy-dispersive X-ray spectroscopy (EDS) mapping characterization in Figure 1e-h, all Co, Pt, and Se elements are evenly distributed. Co and Se atomic proportions are 37% and 63%, respectively ( Figure S1). XRD characterization in Figure 2B confirms the successful synthesis of orthorhombic CoSe 2 (JCPDS no. 00-053-0449), and the prominent diffraction peaks representing different crystal planes are marked. 36 Due to the extremely low content of Pt, no new diffraction peaks appear in Pt 1 /CoSe 2 sample. From Raman testing in Figure 2B, there are two typical characteristic peaks at 168 and 668 cm −1 , which are attributed to the A g and A 1g vibration models of Se-Se in CoSe 2 . 37 Further, X-ray photoelectron spectroscopy (XPS) was performed to study the surface chemical state and elemental composition of Pt 1 /CoSe 2 . As presented in Figure 2C, the Co 2p spectra of Pt 1 /CoSe 2 and CoSe 2 have four peaks, including Co-Se, Co-O, and two satellite peaks. The Co 2p peak position of Pt 1 /CoSe 2 shifts to a lower binding energy of about 0.22 eV compared with the CoSe 2 , which means a low surface oxidation state of Pt 1 /CoSe 2 . On the other hand, as shown in Figure 2D, the Se 3d 5/2 and Se 3d 3/2 peak positions of Pt 1 /CoSe 2 are 54.58 and 55.43 eV, respectively, which are lower than those of CoSe 2 (54.78 and 55.63 eV). The Se 3d peak position at about 59.4 eV is from the oxidized Se. 38,39 Therefore, the presence of Pt single atoms facilitates the regulation of the electronic structure of Pt 1 /CoSe 2 . Electron paramagnetic resonance (EPR) spectroscopy was performed further to confirm the structure of Pt 1 /CoSe 2 and CoSe 2 . As shown in Figure 2E, the EPR signal at g = 2.003 is attributed to the unpaired electrons by the Se vacancies, and the vacancies concentration of Pt 1 /CoSe 2 is higher than that of CoSe 2 . 40-42

Electrocatalytic acidic ORR to H 2 O 2 performance
The electrochemical acidic ORR to H 2 O 2 performance of Pt 1 /CoSe 2 as a catalyst was systematically evaluated with a three-electrode (pH ≈ 1.09). The drop coating amount of catalyst on the disc electrode during the test was 0.1 mg cm −2 . The linear sweep voltammetry curves by using the rotating ring disk electrode (RRDE) test in Figure 3A show that Pt 1 /CoSe 2 has higher ring current and disk current than pure CoSe 2 , which facilitates to catalyze 2e − ORR to form H 2 O 2 . Especially, Pt 1 /CoSe 2 has a high ring current density of 0.47 mA cm −2 (0.4 V vs. RHE), much higher than that of pure CoSe 2 (0.11 mA cm −2 ). We also compared the 2e − ORR performance using Pt (0.1 wt.%)/CoSe 2 with relatively high Pt theoretical content. As shown in Figure S2, the LSV curve results show that Pt 1 /CoSe 2 has a very close ring current and disk current to Pt (0.1 wt.%)/CoSe 2 , which further shows the advantage of Pt 1 /CoSe 2 with ultralow Pt content for acidic 2e − ORR to H 2 O 2 . The H 2 O 2 selectivity of Pt 1 /CoSe 2 is as high as 84% at 0.4 V versus RHE, much higher than that of CoSe 2 (43%, Figure 3B). Moreover, the calculated electron transfer number for Pt 1 /CoSe 2 and CoSe 2 are 2.3 and 3.1 at 0.4 V versus RHE, respectively. The ultralow content Pt single atom on CoSe 2 facilitates the 2e − ORR to H 2 O 2 . The Tafel slope of 165 mV dec −1 for Pt 1 /CoSe 2 is much lower than that of CoSe 2 (275 mV dec −1 ), implying a change in the rate-determining step and the formation of *OOH is the critical step for Pt 1 /CoSe 2 ( Figure 3C). 43 Based on the cyclic voltammogram (CV) results in Figure S3, the calculated double-layer capacitances (C dl ) of Pt 1 /CoSe 2 and CoSe 2 are 0.22 and 0.14 mF cm −2 , respectively ( Figure 3D). Therefore, the Pt 1 /CoSe 2 has a higher electrochemical active surface area (ESCA) than the CoSe 2 , and helps to accelerate the 2e − ORR process.
The sinusoidal ac perturbation of 0.56 V (vs. RHE) was used to obtain electrochemical impedance spectroscopy (EIS). As shown in Figure S4, two semicircles appear, which are attributed to the kinetic-diffusion mixed control region. 44 Equivalent circuit fitting plots display that Pt 1 /CoSe 2 has low R 2 (331.6 Ω) and R 3 (415.5 Ω) values. The R 2 and R 3 for CoSe 2 are 430 and 1577 Ω, respectively. Therefore, Pt 1 /CoSe 2 has higher charge-transfer and masstransfer capabilities, which help to accelerate the 2e − ORR kinetics. Besides, Pt 1 /CoSe 2 has good stability, and the activity has no obvious degradation after 60 h chronoamperometry testing ( Figure 3E). We analyzed the structure and morphology of the catalyst after stability testing by using XPS and transmission electron microscopy (TEM) characterization. In the XPS spectrum of Figure S5, we found that the Co-O peak in the Co 2p spectra shifted to a higher binding energy, and Co-O content increases and Co-Se content decreased significantly. This may be due to the oxidation of the catalyst surface caused by H 2 O 2 generated during the long-term stability test. Moreover, the TEM image in Figure S6 shows that the Pt 1 /CoSe 2 catalyst still maintains its rod-like morphology.
Further, the flow cell was taken to evaluate the H 2 O 2 production from the 2e − ORR process. 22 The H 2 O 2 yield using flow cell device ( Figure S7) was determined by titration with standard Ce(SO 4 ) 2 solution in combination with UV-Vis spectrophotometer testing ( Figure S8). Ce 4+ (yellow) is converted to colorless Ce 3+ according to the reaction of 2Ce 4+ + H 2 O 2 → 2Ce 3+ + 2H + + O 2 . The Pt 1 /CoSe 2 was coated on the surface of hydrophobic carbon paper with a loading of 0.1 mg cm −2 . As shown in Figure 3F, Figures S9 and S10 Table S1, Pt 1 /CoSe 2 with ultralow Pt content (0.01 wt.%) has high acidic 2e − ORR performance. 22,41 In order to explore its application in pollutant degradation, rhodamine B (RhB) was selected to study the degradation ability of Pt 1 /CoSe 2 . As shown in Figure S11 the fitting curves of different concentrations of RhB are given, and the fitting equation is y = 0.1932x, R 2 = 0.9999. 38 Pt 1 /CoSe 2 was coated on the surface of hydrophobic carbon paper with a loading of 0.1 mg cm −2 . The degradation of RhB was tested in acidified 0.5 M Na 2 SO 4 solution containing 20 mg L −1 RhB and 0.5 Mm Fe 2+ (pH = 2.85). As shown in Figure S12 the concentration of RhB dropped rapidly to 50% within the first 10 min, and most of the RhB (90%) was degraded within the first 60 min. However, a small fraction of RhB was not completely degraded, which was caused by a small amount of RhB being adsorbed in the pores of the carbon paper. Overall, Pt 1 /CoSe 2 has potential applications in pollutant degradation.

2.3
The catalytic active sites analysis DFT calculations were performed to understand the active sites for the Pt 1 /CoSe 2 -catalyzed 2e − ORR. 45,46 As shown in Figure S13, the periodic models, including Pt 1 /CoSe 2 (1 1 0), CoSe 2 (1 1 0), and Pt(1 1 1), were constructed. 47,48 The differential charge density of Pt 1 /CoSe 2 (1 1 0) in Figure 4A shows that there is more charge accumulation around the Pt monoatomic site, and the Pt site has significant electronic interaction with its surrounding Co. The calculated density of states (DOS) of Pt 1 /CoSe 2 (1 1 0) and CoSe 2 (1 1 0) exhibit metallic properties due to the continuous electronic DOS at the Fermi level ( Figure 4B). The lower work function of  Figure 4C compared with Pt(1 1 1) means that electrons migrate more easily from the catalyst surface ( Figure S14). In addition, Pt 1 /CoSe 2 (1 1 0) and CoSe 2 (1 1 0) have a similar work function, indicating that the presence of Pt does not affect the mobility of electrons on the catalyst surface. The adsorption and activation of *O 2 is the first critical step for the electrochemical ORR process. 46,49 As shown in Figure S15, the adsorbed *O 2 on the Co-four coordination site of Pt 1 /CoSe 2 (1 1 0) through a "bridge" configuration has a high binding energy of −1.267 eV, whereas *O 2 has a weak binding energy of only −0.07 eV on the Pt site of Pt 1 /CoSe 2 (1 1 0). The *O 2 is adsorbed on the "bridge" site on Pt(1 1 1) with a weak physical interaction, and the calculated binding energy is 0.334 eV ( Figure S16). By contrast, the binding energies of *O 2 on CoSe 2 (1 1 0) are 1.346 eV and 0.334 eV on the Co-four coordination and Co-five coordination sites, respectively ( Figure S17). Subsequently, the adsorbed *O 2 combines with the H proton to generate *OOH. How to accelerate the protonation of *OOH while suppressing the breakage of the O-OH bond is crucial to improve the yield of H 2 O 2 . 50,51 For the key intermediate of *OOH during 2e − ORR process, the Gibbs free energies change of *OOH at different active sites was determined. As shown in Figure 4D, there are three different types of adsorption sites on the surface of Pt 1 /CoSe 2 (1 1 0), including the surface Pt site (labeled as Pt 1 /CoSe 2 -①), surface five-coordinated Co site (labeled as Pt 1 /CoSe 2 -②), and surface four-coordinated Co site (marked as Pt 1 /CoSe 2 -③). The calculated limiting potentials (defined as the maximum potential) on the above surface active sites are 0.43, 0.63, and 0.54 V, respectively. By contrast, the calculated limiting potentials on the fivecoordinated surface Co site (labeled as CoSe 2 -①) and on the four-coordinated surface Co site (labeled as CoSe 2 -②) are 0.61 and 0.45 V, respectively. The Gibbs free energy change of *OOH at the Pt site (Pt 1 /CoSe 2 -①) is larger, suggesting a weak *OOH binding energy. Compared with CoSe 2 (1 1 0), the presence of Pt single atom on Pt 1 /CoSe 2 (1 1 0) makes the surface four-coordinated Co (Pt 1 /CoSe 2 -②) weaken the adsorption of *OOH, and the surface five-coordinated Co (Pt 1 /CoSe 2 -③) strengthen the adsorption of *OOH, thereby significantly reduce the 2e − ORR theoretical overpotential. Moreover, the O-O bond length decreased in *OOH due to the presence of a Pt single atom ( Figure 4E). In addition, the calculated limiting potential on Pt(1 1 1) is 0.57 V. It is noted that on the left side of the volcano curve (∆G *OOH ˂ 4.23 eV) in Figure 4D, the 2e − ORR process is limited by the protonation of *OOH due to the strong binding energy of *OOH at the active sites. By contrast, on the right side of the volcano curve (∆G *OOH > 4.23 eV), the 2e − ORR process is limited by the protonation of *O 2 to form *OOH due to the weak binding energy of *OOH. 39 Therefore, the ideal 2e − ORR catalyst should have an appropriate *OOH binding energy. The optimized adsorption structures of *OOH on Pt 1 /CoSe 2 (1 1 0), CoSe 2 (1 1 0), and Pt(1 1 1) are given in Figure 4E.
In-situ XPS analysis of the Pt 1 /CoSe 2 at different applied potentials (0.5 V vs. RHE, 0 V vs. RHE, and no bias) was carried out to study the oxidation change of the Co site. Figure 2F shows that the oxidation states of Co gradually increased with the change of applied potential from no bias to 0.5 V versus RHE. The positive shift of Co 2p with the decrease of applied potential (from no bias to 0.5 V vs. RHE) is due to the adsorption of *O 2 and the gradual formation of *OH or *OOH intermediates on the Co site. Under the applied potential of 0 V versus RHE, the binding energy of Co 2p positive shifts slightly more than that under no bias. This is due to the rapid protonation of *OOH intermediate to generate H 2 O 2 and subsequent desorption from the Co-active sites. By contrast, there is no change in the Se 3d of Pt 1 /CoSe 2 under different applied potentials (0.5 V vs. RHE, 0 V vs. RHE, and no bias, Figure S18). Therefore, in -situ XPS confirms the role of Co as an active site in electrocatalytic acidic ORR to H 2 O 2 .
To further explore why the Co sites around Pt on Pt 1 /CoSe 2 (1 1 0) have efficient H 2 O 2 generation performance, the projected DOS (PDOS) analysis was performed. [52][53][54] As shown in Figure 5, the PDOS of O-p orbital (from the terminal oxygen in *OOH) and Pt-d orbital or Co-d orbital (from the surface metal sites) for different samples are presented. The hybridization degree between the active metal site and oxygenated intermediates is a crucial activity descriptor, 55-57 which can be described by using a d-band center theory of metal and a p-band center theory of oxygenated intermediates. For the adsorbed *OOH on Pt site on Pt 1 /CoSe 2 (1 1 0), Figure 5A shows that the d-band center of Pt-d is −3.57 eV and the p-band center of O-p is −4.13 eV, the ∆ dp is 0.56. By contrast, the value of ∆ dp is 2.11 for *OOH on the four-coordinated Co site on Pt 1 /CoSe 2 (1 1 0), where the d-band center of Co and p-band center of O are −1.96 and −4.08 eV ( Figure 5B). The value of ∆ dp is 2.20 for *OOH on the five-coordinated Co site on Pt 1 /CoSe 2 (1 1 0), and the calculated d-band center of Co and p-band center of O are −2.11 and −4.31 eV, respectively ( Figure 5C).
For the pure CoSe 2 (1 1 0), the values of ∆ dp are 4.10 and 4.27 for *OOH adsorption on the five-coordinated Co site and the four-coordinated Co site on Pt 1 /CoSe 2 (1 1 0) ( Figure 5D,E). In addition, the calculated d-band center of the surface Pt site on Pt(1 1 1) and the p-band center of the terminal oxygen in *OOH are −2.98 and −4.47 eV, and the calculated ∆ dp is 1.49 ( Figure S19). According to the ∆ dp and limiting potential calculated values, as shown in Figure 5F, a volcano-shaped curve is fitted. The closer the ∆ dp value is to the top of the volcano curve, the closer the limiting potential value is to 0.7, which means smaller reaction overpotential. Therefore, the ∆ dp value can be used as an activity descriptor for electrochemical ORR to H 2 O 2 , and the ∆ dp value of a good catalyst is closer to 2.75.

CONCLUSION
In  vacuum drying oven at 60 • C for 24 h. The catalyst was then reduced by high temperature roasting with H 2 /Ar mixture (H 2 /Ar volume ratio is 5%:95%, heating rate of 5 • C min −1 , 300 • C and keeping 2 h) to obtain Pt 1 /CoSe 2 . The Pt content in Pt 1 /CoSe 2 was 0.01 wt.%, which was identified by the ICP-MS. As a comparison, 300 μL of chloroplatinic acid solution was added into the above polytetrafluoroethylene lined autoclave to obtain a theoretical content of 0.1 wt.% Pt/CoSe 2 .

Structure characterization
Tecnai G2 F20 was used for high-resolution TEM (HR-TEM) characterization with an accelerating voltage of 200 kV. The spherical aberration electron microscope characterization was using Titan Themis G2 60-300. XRD characterization was performed on a D/MAX-2500 X-ray diffractometer (0.154 nm Cu K α , scan rate 5 • /min). In order to obtain Raman results, Renishaw inVia reflection Raman spectrometer was used for characterization. EPR spectra data were acquired by a Bruker E500 spectrometer at room temperature of 25 • C. XPS analysis was performed using PHi 5000 X with Al K α radiation, and the binding energy was calibrated using the C 1s peak (284.8 eV) of contaminating carbon. In-situ XPS (Thermo Fisher, ESCALAB 250Xi) was performed on an in situ cell, and it includes three parts: a reaction chamber, an analysis chamber, and a preparation chamber. The catalyst is deposited on the back of the Si 3 N 4 window on the Au/Ti layer, toward the interior of the electrochemical cell. The reference electrode and counter electrode were Ag/AgCl (3 M KCl in gel) and the platinum wire, respectively. Among them, the reference electrode was placed in a Luggin capillary. Prior to testing, to confirm the proper electrical contact between the potentiostat and the catalyst, it is necessary to measure the resistance between the working electrode and the reference electrode. The catalyst is first placed in the reaction chamber in 0.1 M HClO 4 solution (deoxygenated with argon or saturated with oxygen), then the reactor is evacuated to <10 −8 torr and then transferred to the analysis chamber. Notably, at each set potential (0.5 V, 0 V vs. RHE, and no bias), the electrodes were first polarized for 5 min until stable, and then XPS signals were collected at different applied potentials, and the scan rate was set to 1 mV s −1 .

Electrochemical test details
The CHI760E electrochemical workstation with the Ivium RRDE (the diameter of the glassy carbon disk electrode is 4 mm and the disk area is 0.1256 cm 2 , and Pt ring area is 0.1885 cm 2 ), using a three-electrode system with 0.1 M HClO 4 as electrolyte, was used to obtain the ORR performance. Ag/AgCl electrode and graphite rod were used as reference electrode and counter electrode, respectively. Using the Nernst equation (E = E Ag/AgCl + 0.059 × pH + 0.197), the measured potential can be converted into a RHE potential. A amount of 5 mg catalyst powder was dispersed with 1 mg of carbon black in 1 mL of isopropanol and ultrapure water (3:1 v/v), and 10 μL of Nafion solution (5 wt.%) was added to form a mixture, which was sonicated for 120 min to get a homogeneous ink. Then, 2.5 μL of catalyst ink was slowly pipetted onto a glassy carbon disk. The coating amount on the ring disk electrode is 0.1 mg cm −2 , and the H 2 O 2 selectivity and the number of electron transfers are determined by the RRDE test at 1600 rpm. The calculation formula of H 2 O 2 selectivity and the number of electron transfers are as follows: Among them, I R and I D are the ring current and disk current, respectively. N is the collection factor of 0.424.
The flow cell device was further used to evaluate the product H 2 O 2 concentration over time ( Figure  S7). The yield of H 2 O 2 was determined by titration with standard Ce(SO 4 ) 2 solution in combination with UV-Vis spectrophotometer test. Ce 4+ (yellow) is converted to colorless Ce 3+ according to the reaction of 2Ce 4+ + H 2 O 2 → 2Ce 3+ + 2H + + O 2 .
The degradation of RhB was performed in the flow cell. Using UV-Vis spectrophotometer, the concentration of RhB can be determined.

DFT calculation details
The VASP software with projector augmented wave method was used to study the ORR mechanism. 58,59 In order to describe the electro-ion interaction and the nonlocal exchange correlation energy, the ultrasoft pseudopotential and the Perdew-Burke-Ernzerhof functional of generalized gradient approximation (GGA-PBE) were performed. 60,61 To eliminate the self-interaction error of Co 3d orbital, the DFT + U method (U = 3.3) was used. 62,63 During structure optimization, the cutoff energy was set to 400 eV, and the K-point was set to 3 × 3 × 1. Moreover, 0.02 eV/Å and 1 × 10 −5 eV were used to set the force and energy convergence criteria, respectively. The 6 × 6 × 1 K-point was used for electronic structure analysis. According to some literature reports, 48,47 CoSe 2 nanorods catalyst mainly expose (1 1 0) crystal facets, thus CoSe 2 (1 1 0) with three layers and p(2 × 2) period model was constructed, and one surface Co atom was replaced by Pt to construct Pt 1 /CoSe 2 catalyst. The vacuum layer of 15 Å was used. The calculation formulae, including the binding energy (E), the Gibbs free energy variation (ΔG *OOH ), and the d-band center or p-band center, were given in the following: Among them, E adsorbate , E substrate , and E adsorbate/substrate are the energy of adsorbate, the energy of substrate, and the energy of substrate adsorbed with adsorbate, respectively. Obviously, a positive value of E indicates a strong interaction.
Among them, ΔE and ΔZPE are the total energy change and the zero point energy change, respectively. ΔZPE can be calculated according to the formula of ∑1/2hV i and V i is the vibration frequency. T and ΔS are the temperature at 298.15 K and the entropy change, respectively.
Δ dp = d − p where E f is the Fermi energy, and the ρ represents the DOS projected onto the surface d-or p-band. University of New York. Xiao-Dong Zhang and Qian Zhang contributed equally to this work.

C O N F L I C T O F I N T E R E S T S S TAT E M E N T
The authors declare no conflict of interest. Dr. Gang Wu is an Associate Editor of SusMat and a coauthor of this article. To minimize bias, he was excluded from all editorial decision-making related to the acceptance of this article for publication.