Boosting hydrogen peroxide production via establishment and reconstruction of single‐metal sites in covalent organic frameworks

Covalent organic frameworks (COFs) have been well developed in electrocatalytic systems owing to their controllable skeletons, porosities, and functions. However, the catalytic process in COFs remains underexplored, hindering an in‐depth understanding of the catalytic mechanism. In this work, uniform Pt–N1O1Cl4 sites chelated via C–N and C=O bonds along the one‐dimensional and open channels of TP–TTA–COF were established. Different from conventional single‐metal sites constructed for the near‐free platinum for hydrogen evolution, the as‐constructed PtCl–COF showed 2e− oxygen reduction for H2O2 production. We tracked the dynamic evolution process of atomic Pt sites in which Pt–N1O1Cl4 was transformed into Pt–N1O1(OH)2 using in situ X‐ray adsorption. The theoretical calculations revealed that the strong Pt–support interaction in Pt–N1O1(OH)2 facilitated *OOH formation and thus led to higher selectivity and activity for the oxygen reduction reaction in the 2e− pathway. This work can expand the applications of COFs through the regulation of their local electronic states for the manipulation of the metal center.

The ORR in the 2e − pathway is an important method for the electrosynthesis of H 2 O 2 , which was considered a substitute for the energy-intensive anthraquinone process. [49][50][51][52][53][54][55][56][57] However, achieving a high proportion of the 2e − ORR is difficult because of the competing 4e − pathway in the ORR process. Many single-atom catalysts supported by different templates (CuS x , TiN, and doped carbon) have been adopted to catalyze the ORR in the 2e − pathway. [57][58][59][60] However, tuning the coordination environments of these atoms is still a challenge. Thus, we used COFs as a template to construct tunable SACs for ORR. We developed a COF (TP-TTA-COF) with stable frameworks and high porosity and then introduced atomic Pt δ+ along the one-dimensional and open channels. Different from the constructed catalytic sites in the skeletons, the atomic sites immobilized along the porous intersurface enhance access to electrolytes, resulting in high atomic utilization efficiencies. Based on the highly efficient catalytic sites, in situ X-ray absorption fine structure (XAFS) spectroscopy was adopted to probe the catalytic process. The prepared catalyst enabled the catalysis of the ORR with high selectivity (close to 90%) and considerable activity in the alkaline condition. This work provides valuable insights into the catalytic process of COF-based catalysts.
Powder X-ray diffraction (PXRD) was used to evaluate the crystallinity of TP-TTA-COF and PtCl-COF. The identified peaks at 2θ = 4.71 • , 7.74 • , 8.83 • , 11.68 • , and 25.3 • were attributed to the (1 0 0), (1 1 0), (0 2 0), (2 1 0), and (1 0 1) facets, respectively ( Figure 1B). Pawley refinements were applied to determine the theoretical structures using the self-consistent charge density functional tight binding method. The experimental results agreed well with the simulated results (R wp = 4.33% and R p = 3.42%, respectively). Both the eclipsed (AA) ( Figure 1A and Figure  S1) and staggered (AB) stacking ( Figure S2) structures were simulated for the COFs. Furthermore, the simulated PXRD patterns of the eclipsed structures exhibited a much better agreement with the experimental results ( Figure 1B). The COF structure adopts eclipsed stacking in the P1 space group with refined cell parameters of a = 23.18 Å, b = 23.67 Å, c = 4.04 Å, α = 92.90 • , β = 104.31 • , and γ = 113.73 • (Tables S1 and S2). Extended structures were constructed using space-filling models. Slight shifts in the peak positions of PtCl-COF were observed upon the introduction of Pt ions into the interporous channels ( Figure 1C). However, the peak intensities remarkably decreased, indicating that the ordered COF skeletons were retained, and the Pt ions were loaded into the pores. N 2 physisorption (at 77 K) was performed to evaluate the permanent porosity ( Figure 1D and Figure S3). The calculated Brunauer-Emmett-Teller surface areas of TP-TTA-COF and PtCl-COF were 1052.6 and 650.5 m 2 g −1 , with pore volumes of 0.71 and 0.57 cm 3 g −1 , respectively. The pore size distribution curves revealed that TP-TTA-COF exhibited a sharp peak centered at 1.2 nm, which is similar to the theoretical value ( Figure S3). The micropore distribution of PtCl-COF was lower than that of TP-TTA-COF. The decreased surface areas and pore volumes of PtCl-COF further verified the dispersed distribution of Pt ions in the pores.
Subsequently, 13 C solid-state cross-polarization magicangle-spinning nuclear magnetic resonance analysis was used to characterize the skeletons of TP-TTA-COF and PtCl-COF ( Figure 1E, black curve). As shown in Figure 1E Fourier transform (FT) infrared spectroscopy was used to verify the formation of the predesigned chemical structures. The formation of β-ketoenamine is essential to verifying the polymer structure and is indicated by the C-N vibration at ∼1253 cm −1 (Line E, Figure S4). Peaks attributed to other functional groups such as C≡C, C=O, and aromatic C=C were observed at ∼2205, ∼1618, and ∼1451 cm −1 , respectively ( Figure S4). These vibrations were also identified for PtCl-COF. Additionally, the peaks associated with the N-H bond exhibited a blueshift from 3347 to 3405 cm −1 owing to the formation of Pt-N bonds.
Moreover, we also assessed the chemical stability of TP-TTA-COF and thermal stability PtCl-COF, respectively. TP-TTA-COF was observed to be chemically stable in harsh conditions such as a strong base since that TP-TTA-COF retained its main crystallinity and porous structure after being stored in the strong base for 5 days ( Figure  S5). According to the thermogravimetric analysis curve, the weight left was about 60 wt.% at 1000 • C under N 2 atmosphere ( Figure S6). PtCl-COF was unstable at all ranges, which may be caused by the continuous behavior of the H 2 O loss, the evaporation of the Cl elements, and carbonized of the frameworks.
The morphologies of TP-TTA-COF and PtCl-COF were characterized by scanning electron microscopy (SEM) and transmission electron microscopy (Figures S7-S10). The SEM images indicate that TP-TTA-COF exhibited an intertwined threadlike morphology with a diameter of hundreds of nanometers. From a larger perspective, TP-TTA-COF comprised coils and sponge-like structures. Furthermore, the TP-TTA-COF morphology had a twig and layered structure, which comprised particles with a diameter of dozens of nanometers. PtCl-COF exhibited morphology similar to that of TP-TTA-COF; however, metal particles were not observed.

Confirmation of the isolated Pt sites
X-ray photoelectron spectroscopy was employed to study the electronic states of PtCl-COF. The elemental contents were evaluated by the survey spectra ( Figure S11 and Table  S3). The fitting results of the high-resolution Pt 4f spectrum for PtCl-COF showed only the oxidated state Pt δ+ ( Figure 2A). At the same time, the Pt content in the surface of the COFs was also estimated as 9.1 wt.%, whereas the inductively coupled plasma optical emission spectrometry record the average value as 4.2 wt.% (Table S3). The high-resolution N 1s spectrum was fitted to three peaks attributed to the -NH, C=N, and Pt-N bonds ( Figure S12). Pt L 3 -edge X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine structure (EXAFS) were employed to evaluate single-atom sites, the coordination environment, and electronic states. The white-line peak was generated from the migration of the excited 2p 3/2 core electron to the unoccupied 5d orbital ( Figure 2B). The fingerprint identification indicated that the coordination environment of PtCl-COF was similar to that of H 2 PtCl 6 . The white-line peak intensity corresponded to the vacant 5d-band state of Pt, and the oxidation state of PtCl-COF is between Pt (0) and Pt (4+). To obtain accurate quantitative information regarding the electronic structure, the white-line peak area and the oxidation state were used to plot standard curves ( Figure 2C). The selected Pt standards were only coordinated with Cl − , and the chemical valence of PtCl-COF was estimated to be +2.70. It can be attributed to the octahedron Pt-Cl 6 configuration, which was partially substituted by functional groups from the pore of TP-TTA-COF. The scattering amplitude gaps between N/O and Cl atoms also induced the valence state error, and the real valence was slightly higher than the estimated value.
FT k 2 -weighted EXAFS spectra for the object samples and contrast substances are shown in Figure 2D. The main peak of PtCl-COF was observed at 1.90 Å, which is slightly lower but similar to that of H 2 PtCl 6 at 1.92 Å. Wavelettransformed X-ray absorption fine structure (wt-XAFS) is a powerful technique used to elucidate the coordination bond type and atomic distance of the central metal atom ( Figure 2E). Despite the similar k responses of the samples, PtCl-COF exhibited a downshift in contrast to H 2 PtCl 6 , indicating that PtCl and Pt-N/O involved the first shell of Pt atoms. Based on the least-squares EXAFS curve-fitting analysis, the main peak of PtCl-COF was attributed to the Pt-Cl pathway with a coordination number of 3.8 and a radial distance of 2.29 Å, accompanied by the Pt-N/O pathway with a coordination number of 1.4 and an average radial distance of 2.00 Å ( Figure S13). Thus, the Pt configuration was confirmed to be Pt-N 1 O 1 Cl 4 in PtCl-COF.
Scanning transmission electron microscopy (STEM) using the high-angle annular dark field (HAADF) mode was performed to characterize the atomic states of Pt. The HAADF-STEM images of PtCl-COF exhibited a highdensity bright spot corresponding to isolated Pt atoms uniformly dispersed on the substrate ( Figure 2F). Additionally, energy-dispersive X-ray spectroscopy elemental mapping ( Figure 2G) revealed that all elements (C, N, O, Cl, and Pt) were homogeneously distributed in the matrices.

The ORR evaluation and in situ tracking the active sites
The dispersed single Pt atoms would catalyze the ORR in the 2e − pathway. To evaluate the activity and selectivity of the COFs, a three-electrode system equipped with a rotation ring disk electrode was used. TP-TTA-COF exhibited an onset potential (E o ) of 0.662 V and a limited current density (J lim ) of 1.32 mA cm −2 ( Figure 3A, black curve). By loading Pt atoms in the pore channels, PtCl-COF exhibited a positive E o value of 0.675 V and a higher J lim of 1.83 mA cm −2 (red curve). According to the ORR and ring currents, the electron transfer number and hydrogen per-oxide selectivity were calculated, respectively. In the wide potential range of ∼0.2-0.6 V, the electron transfer number of TP-TTA-COF ranged from 2.68 to 2.58 ( Figure S14, black curve), corresponding to H 2 O 2 selectivities of 66.0% and 70.9% ( Figure 3B, black curve), respectively. After the Pt atoms were embedded, PtCl-COF exhibited higher H 2 O 2 selectivity, which increased from 81.6% to 87.2% ( Figure 3B, red curve), and lower electron transfer numbers, which decreased from 2.37 to 2.26 ( Figure S14, red curve). This single-metal immobilization strategy increased the selectivity up to 17.92% at the same potential. The Tafel slope of PtCl-COF was 77.1 mV dec −1 , which was very close to that of COF (78.8 mV dec −1 ), indicating the similar rapid kinetics of COFs ( Figure 3C). The 10-h durability tests demonstrated that PtCl-COF maintained a H 2 O 2 selectivity of over 80% with a final current density of 1.42 mA cm −2 ( Figures S15 and S16).
Considering that the metal-chlorine bond was unstable in an aqueous electrolyte, in situ XAFS was used to clarify the Pt center behavior in PtCl-COF and identify the configuration of the actual active species and the d-band hole count. The variation of curves in XANES between the dry carbon paper sample and open-circuit potential (OCP) showed that the white-line peaks improved in intensity and broadened ( Figure 3D). The above-described variations can be described more clearly from differential XANES. This phenomenon may be attributed to the first shell atomic substitution ( Figure 3E). The potentialinduced XANES compared the curves of the non-Faraday (0.8 V) and Faraday (0.2 V) conditions, indicating that the white-line peak was elevated. The formal d-band hole count was estimated using the Pt-O bond standards (from Pt foil (5d 9 s 1 ) to Pt IV O 2 (5d 6 s 0 )). At a slope of 1.221 unit area per d-band hole standard curve, we confirmed the d-band hole value as OCP-2.18, 0.8 V-2.28, and 0.2 V-2.55. The increase in d-band holes under the ORR process facilitated oxygen adsorption and reduction.
The FT-EXAFS and wt-XAFS spectra corresponding to Figure 3D are shown in Figure S17 and Figure 3J, respectively. As shown in Figure 1A, we tracked the evolution of single Pt sites, displayed as the Pt-Cl bonds in the COFs changed to the unitary Pt-N 1 O 1 (OH) 2 . Compared to the state of the OCP, the main peak position of the dry electrode sample decreased from 1.69 to 1.59 Å ( Figure S17). Moreover, the center of the wt-XAFS spectrum shifted from 4.25 to 3.75 Å −1 (Figure 3J), indicating that the coordinated atoms were substituted by low-atomic-number elements. The least-squares EXAFS curve-fitting analysis ( Figure 3G Figure S17). The black dots are experimental results, and the redlines are best-fit curves. In addition, insert pattern was corresponding k 2 χ(k) oscillations of PtCl-COF in OCP condition. COF, covalent organic framework; TP, 1,3,5-triformylphloroglucinol; TTA, 4,4′,4″-(benzene-1,3,5-triyltris(ethyne-2,1-diyl))trianiline. non-Faradaic potential, the main peak position corresponded with that of the OCP. The fitting results indicated that Pt-N/O had an average coordination number of 4.3 and maintained a radial space distance of 2.04 Å (Table  S4). The increased coordination number is attributed to peroxide or oxygen molecule adsorption. At the operating potential, the main peak was broadened because of the increased structural disorder. The fitting parameters indicated that the coordination number of Pt-N/O decreased to 3.8 with an average bond length of 2.04 Å. Under the OH − -rich ORR process, this phenomenon corresponds to the stable structure of Pt-N 1 O 1 (OH) 2 and rapid molecular exchange. Thus, the in situ experiment identified that the Pt-N 1 O 1 (OH) 2 sites changed to the Pt-N 1 O 1 (OH) 2 structure, and the potential induced the valence increase and maintained the structure of Pt-N 1 O 1 (OH) 2 .

Theoretical calculation
The atomic mechanism of the proton and electron transfer principle on introduced single Pt sites was revealed by density functional theory (DFT) calculations. The local structures that contained suspected active sites in PtCl-COF are shown in Figure 4A, which were optimized by the identified atomic configuration determined by in situ X-ray adsorption measurements. The configuration of Pt sites was confirmed as Pt-N 1 O 1 (OH) 2 , whereas metal-free sites were identified as the adjacent carbons of polar atoms named site 1 to site 3. Based on the 2e − ORR pathway ( Figures S18 and S19), the free-energy diagram collected for the optimized adsorption state of the corresponding active site model is shown in Figure 4B. The adsorption energy of Pt-N 1 O 1 (OH) 2 was 0.20 eV, which was much lower than those of metal-free sites (site 1: 1.44 eV, site 2: 0.56 eV, and site 3: 1.35 eV), indicating that the Pt-N 1 O 1 (OH) 2 site more easily forms *OOH and thus results in higher activity for the ORR in the 2e − pathway. We also derived the free-energy diagram of each site that followed the 4e − oxygen reduction pathway ( Figure 4C regulation of the activity and selectivity in the 2e − ORR.

An extensive single-metal sites platform
Notably, other single-metal catalysts (Cu and Ni) were successfully fabricated via this platform by simply using nickel(II) acetate tetrahydrate and copper(II) acetate monohydrate ion-exchange methods, respectively. The metal K-edge experimental XANES of the as-prepared M-COFs showed oxidation states higher than +2 ( Figure 5A,B). The outline of XANES displayed vibrational widening compared to metal oxide, which may be caused by the single dispersed structure. FT-EXAFS identified the main peaks corresponding to M-N/O paths in M-COFs, and no obvious high shell was observed ( Figure 5C,D). The fitting results confirmed the highly dispersed metal in COFs was constructed by M-N/O ( Figure 5E,F; Table S5 and S6). This work demonstrates that TP-TTA-COF can serve as a multifunctional plat-form for the construction of various single-metal sites, which can be broadly applied in homogeneous and heterogeneous catalysis. 61-63

CONCLUSION
In conclusion, we designed a novel COF, TP-TTA-COF, which consisted of alkynyl and chelate groups. By immobilizing the highly dispersed Pt in 1D porous channels, both the activity and selectivity toward the 2e − ORR were improved. In situ XAFS tracked the Pt atomic dynamic changes in the transformation from

Synthesis of TP-TTA-COF
A mixture of TP (21.0 mg), TTA (42.3 mg), 0.5 mL of 1,4dioxane, 0.5 mL 1,3,5-trimethylbenzene, and 0.1 mL 6 M HAc was added and then degassed in a Pyrex tube (10 mL) by three freeze-pump-thaw cycles. The tube was sealed off and heated at 120 • C for 3 days. The precipitate was collected by centrifugation and washed several times with water (18 MΩ), THF. The material was further purified using a Soxhlet extractor with THF for 1 day and then dried for 24 h in a vacuum (20 mTorr) oven at 65 • C.

Synthesis of M-COF
M-COF was prepared by stirring the as-synthesized TP-TTA-COF powder in metal acetate (Ni or Cu, 50 mg) aqueous solution at room temperature for overnight and washed with plenty of water. The precipitate was collected by centrifugation, washed several times with water and THF, and then dried for 24 h in a vacuum (20 mTorr) oven at 65 • C.

C O N F L I C T O F I N T E R E S T S TAT E M E N T
The authors declare no conflicts of interest.