Hybridization of covalent organic frameworks and photosensitive metal‐organic rings: A new strategy for constructing supramolecular Z‐scheme heterostructures for ultrahigh photocatalytic hydrogen evolution

The rational design of Z‐scheme heterojunction photosystems based on covalent organic frameworks (COFs) is a promising strategy for harnessing solar energy for hydrogen conversion. Herein, a direct Z‐scheme single‐atom photocatalyst based on COF and metal‐organic ring has been constructed through the supramolecular interactions of coral‐like COF (S‐COF) and photosensitized Pd2L2 type metal‐organic ring (MAC‐FA1). The MAC‐FA1/S‐COF heterojunction exhibits good light absorption, efficient charge separation and transfer, slow electron‐hole recombination, and highly dispersed Pd active sites, enabling an efficient and stable H2 evolution reaction. The optimized 4% MAC‐FA1/S‐COF achieves an H2 evolution rate of 100 mmol g−1 h−1 within 5 h and obtains a total accumulated turn‐over number relative to Pd (TONPd) of 437,685 within 20 h, far superior to S‐COF, MAC‐FA1, M‐5/S‐COF, Pd/S‐COF, and M‐5/Pd/S‐COF, which is one of the highest records among COF‐based photocatalysts for solar‐driven H2 evolution. This is the first work to incorporate photosensitized metal‐organic rings/cages into porous crystalline COFs to form a supramolecular Z‐scheme heterojunction, which has significant potential as a high‐performance photocatalyst for solar‐driven H2 production.


INTRODUCTION
[3] Despite ongoing efforts, the development of robust photosystems to efficiently convert solar energy into hydrogen energy sources remains critical. [4,5][8][9] Among them, covalent organic frameworks (COFs), highly ordered materials with controllable porosity formed by the covalent bonding of organic building blocks, [10][11][12] are attracting considerable attention due to their programmability and flexibility, high carrier mobility, good stability, and recyclability, [13] with great potential in the field of photocatalysis.[16] However, challenges such as rapid recombination of photogenerated electrons and holes and limited light absorption capability still restrict the further application of COFs in photocatalysis, so various strategies have been explored. [17]The introduction of functional groups or the doping of different atoms into the COF framework is a dominant strategy, which is beneficial for photoinduced charge transfer and light absorption. [18,19]Furthermore, modulating the morphology of the COF can enhance the visible photocatalytic activity.Various microstructures of COFs, including but not limited to flower-like, [20,21] spherical, [22,23] and rod-like shapes, [24,25] have been synthesized through template effects, direct synthesis methods, post-modification, and so on.For example, Lotsch et al. investigated the relationship between their extrinsic morphology, intrinsic structure, and photocatalytic hydrogen evolution performance by preparing a series of azine-linked Nx-COFs (x = 0-3) with varying nitrogen content in the central aryl ring.28] In addition, the construction of COF-based heterojunctions is also an important way to reduce photogenerated electron-hole recombination within bare COF photocatalysts.[31] In various types of heterojunctions, such as type-II, p-n type, and Z-scheme, a Z-scheme system has also been recognized to achieve strengthened light-harvesting ability coupled with high charge-transfer efficiency and enhanced redox property. [32,33]For example, Liu et al. [34] achieved Z-scheme heterojunction T-COF@CdS through direct growth by synthesizing COF on CdS, where T-COF shell protected CdS catalytic centers from deactivation, and also acted as oxidation sites to avoid CdS photocorrosion.As a result, the T-COF@CdS exhibited a hydrogen production rate of 12.5 mmol g −1 h −1 under full-spectrum irradiation, 10 times higher than that of bare COF, due to efficient photogenerated carrier generation and separation facilitated by the Z-scheme heterojunction.
Single-atom catalysts (SACs), composed of single atoms and solid carriers, can effectively enhance photocatalytic performance and stability owing to their distinct active sites and unique coordination environment. [35,36][39] However, in terms of recovery and the vulnerability of metal-organic rings/cages to deactivation, there is a need to anchor them onto other solid materials with multiple binding motifs to improve catalytic performance and address their instability issues while allowing recyclability.Therefore, we have designed and synthesized several photosensitized metal-organic rings/cages, including metal-organic cages MOC-Q1, [40] MOC-Q2, [41] MOC-Py-Zn, [42] MOC-16, [43,44] and metal-organic ring MAC-1, [45] immobilized on semiconductors, such as g-C 3 N 4 or TiO 2 , to construct versatile heterojunctions with superior photocatalytic activity and stability compared to the homogeneous metal-organic cages/rings.This improvement was due to the synergistic effects of highly efficient electron transfer, an extended visible light response range, and highly monodisperse Pd active sites.Thanks to these attractive features, supramolecular photosystems based on metal-organic rings/cages are promising candidates for photocatalysis.Moreover, compared to the covalent binding strategy, the noncovalent binding between semiconductor and metal-organic rings/cages offers superiority in the synthesis of integrated photosystems by exploiting the large number of available metal-organic rings/cages and the diversity of semiconductors.However, there are few reports on the precise fabrication of the well-defined heterostructures of COFs and photosensitized metal-organic rings/cages with single metal atoms.
In view of these factors, we designed and synthesized a coral-like COF (S-COF) with enhanced visible light harvesting ability and a novel M 2 L 2 -type metal-organic ring (MAC-FA1) self-assembled from two photosensitive 4-nitro-N,N-bis(4-(5-(pyridin-4-yl)thiophen-2-yl)phenyl)aniline ligands (M-5) and two catalytic Pd 2+ ions.Then, a unique supramolecular Z-scheme photosystem consisting of S-COF and MAC-FA1 was obtained for the first time by solution impregnation through their hydrogen bonding and π-π stacking effect.In this photosystem, S-COF could collect light and generate hot electrons, while MAC-FA1 on the COF surface acted both as a catalytic site and as a light absorber.The optimized 4% MAC-FA1/S-COF showed excellent hydrogen production performance (H 2 evolution rate of 100 mmol g −1 h −1 and TON Pd of 437,685), significantly outperforming S-COF, MAC-FA1, M-5/S-COF, Pd/S-COF, and M-5/Pd/S-COF under the same conditions.Furthermore, the photocatalytic mechanism of this photosystem was studied in detail and it was confirmed that the charge transfer at the MAC-FA1/S-COF interface took place via a "Z-scheme" pathway.

Sample synthesis and characterization
MAC-FA1 was obtained by self-assembly of the M-5 ligand and Pd metal ions (Figure 1A), while the detailed synthetic procedures of ligands M-1−M-5 and MAC-FA1 could be found in the experimental section of the Supporting Information.The specific synthesis route of MAC-FA1 is illustrated in Figure S1.The 1 H NMR spectra of ligands M-1−M-5 and MAC-FA1 are shown in Figures S2 and S3, respectively.The shift of the resonance signals in the 1 H NMR spectra compared to the ligand M-5 confirmed the formation of MAC-FA1 (Figure S4).To further validate the MAC-FA1 structure, [Pd 2 (M-5) 2 ](NO 3 ) 4 was measured by electrospray ionization mass spectrometry (ESI-MS) (Figure S5A).An observed peak at m/z 415.5847 with a charge state of 4+ indicated the loss of four nitrate ions (NO 3 − ) from MAC-FA1 (Figure S5B).In addition, the energy-optimized structure of MAC-FA1 was revealed on the basis of theoretical calculations (Figure 1A and Table S1), and the stabilization energy of MAC-FA1 was calculated to be −952.5 kcal/mol (Table S2).The measured dimension of MAC-FA1 was 27.32 Å × 32.16 Å.The diameter for MAC-FA1 based on the ligand N-N distance was approximately 15.07 Å, while the diameter based on the Pd-Pd distance was approximately 26.54 Å.
COF was synthesized via solvothermal method by the combination of monomers 2,4,6-trihydroxy-1,3,5-benzenetricarbaldehyde (TFG) and 5,5-dioxodibenzothiophene-3,7diamine (SA) (Figure 1B), and the detailed synthesis process is described in the experimental section.The chemical structures of the samples were characterized by the Fourier transform infrared (FT-IR) spectrum.In the FTIR spectra, the C═O and C═C stretching vibration peaks appeared at 1619 and 1596 cm −1 in the resulting S-COF, while two NH 2 stretching vibration peaks at 3467 and 3373 cm −1 for SA and a C═O peak belonging to ─CHO group at 1643 cm −1 for TFG decreased significantly (Figure S6A).The 13 C NMR spectra of S-COF (Figure S6B,C) revealed the presence of carbonyl C(a) resonating at 185 ppm and secondary amine carbon C(b) resonating at 147 ppm.Additionally, a characteristic signal corresponding to the carbon C(c) of the ─NH─C═C group was identified at 160 ppm. [46,47]The crystal phases of the samples were validated by powder X-ray diffraction (PXRD) in combination with theoretical structure simulations using Materials Studio version 2020.The PXRD pattern showed an intense peak at 3.47 • and a weak peak at 6.87 • , which were assigned to the (1 0 0) and (2 0 0) facets of the S-COF, respectively (Figure S7A), in agreement with the results of the S-COF prepared in Wang's literature. [48]By comparing the experimental XRD pattern with the simulated one corresponding to two different stacking modes, it was found that the synthesized S-COF exhibited an AA stacking arrangement (Figures S7B  and S8).Contrary to the literature, [48] the solution was not sonicated after mixing the ligands, and the catalyst was added at the end of degassing to avoid premature reaction of the ligands at room temperature.The coral-like S-COF was obtained by optimizing the amount of catalyst (0.05 mmol of TFG and 0.075 mmol of SA using 0.4 mL of catalyst containing glacial acetic acid and water in a 1:3 volume ratio) and its morphology was characterized by scanning electron microscope (SEM) and high-resolution transmission electron microscopy (HRTEM) in Figures S9A and S9C, which grew orderly from the center outward into a 3D coral flower based on a 1D stripe shape.The optimized catalyst amount experiments showed that as the catalyst amount increased from 0.1, 0.2, 0.4 to 0.8 mL, the morphology of the S-COF samples changed from leaf (Figure S10A) to rod shape (Figure S10B), followed by coral-like (Figure S10C) and then flower-like morphology (Figure S10D).For comparison, S-COF was also prepared by Wang's method, which presented a lamellar structure displayed in Figures S9B and  S9D, in accordance with the literature. [48]It is well known that the morphology of the COFs has an impact on the photocatalytic reduction processes due to changes in dispersibility, aggregation behavior, transfer of photogenerated charges to the surface, and light-harvesting capacity in water. [49]OF with one-dimensional (1D) channel micropores was found to possess the ultrashort transfer path that could reduce charge loss and facilitate the catalytic reaction. [50]Therefore, PXRD patterns and solid UV-Vis absorption spectra of coral-like and sheet-form S-COF were measured to compare the crystallinity and visible light absorption properties of the two COFs (Figure S11A,B).It could be seen that the intensity of the diffraction peak at 3.46 • and the absorbance intensity in the range of 550-750 nm of coral-like S-COF were significantly increased compared to sheet form S-COF, which could facilitate its photocatalytic hydrogen production.The Brunauer−Emmett−Teller (BET) surface areas (S BET ) of the two COFs were also measured for comparison (Figure S11C).It was found that coral-like S-COF had higher surface areas (872 m 2 /g) than sheet-form S-COF (502 m 2 /g), indicating that coral-like S-COF had better light-harvesting ability and more abundant photocatalytic active sites.The comparison of the hydrogen evolution performance of MAC-FA1/S-COF in Figure S11D and Pt/S-COF with Pt as a co-catalyst in Figure S11E supported this conclusion.In addition, when the coral-like catalyst was used for photocatalytic hydrogen production, the H 2 gas could easily escape from the surface without blocking the reaction sites, which also enhanced its photocatalytic performance.For these reasons, the coral-like S-COF was selected for further research.
The 1−5% MAC-FA1/S-COF composites, with MAC-FA1 contents calculated as 1−5 wt%, were prepared by solution impregnation and the detailed synthesis procedures are described in the experimental section.The actual MAC-FA1 content in the composites was determined by inductively coupled plasma mass spectrometry (ICP-MS) and is given in Table S3.The simulated structure of MAC-FA1/S-COF is shown in Figure 1C, and MAC-FA1 was loaded on the S-COF surface through the hydrogen bonds (X⋅⋅⋅C─H, X═S, N, O) and π-π interactions.
To analyze the crystal structures of the composites, XRD patterns of MAC-FA1, S-COF, and 4% MAC-FA1/S-COF were recorded.As shown in Figure 1D, no obvious peak was observed for MAC-FA1, indicating that it was in an amorphous state.The pure S-COF samples showed a sharp diffraction peak at 3.46 • , while the composite displayed a slightly weaker peak at the same position, suggesting that the addition of MAC-FA1 did not affect the main structure of S-COF, although it slightly reduced the S-COF crystalline phase.In addition, their porosity was assessed by N 2 adsorption-desorption measurements at 77 K and the isotherms are shown in Figure S12.The S BET of S-COF and MAC-FA1/S-COF were found to be 872 and 623 m 2 /g, respectively.And their pore diameters, derived by fitting nonlocal density functional theory models to the N 2 isotherms, were 2.11 nm.This meant that MAC-FA1 evenly covered the surface of the S-COF, resulting in a reduction in S BET without destroying the crystal structure.
The solid UV-Vis absorption spectra of MAC-FA1, S-COF, and 4% MAC-FA1/S-COF were measured (Figure 1E).MAC-FA1 had an intense absorption band at 400-600 nm centered at 425 nm.S-COF showed a broad absorption band at 350-550 nm with an edge at approx.650 nm, corresponding to its intrinsic optical band gap of 1.75 eV.MAC-FA1/S-COF displayed clear superimposed peaks of MAC-FA1 and S-COF, indicating that MAC-FA1 was successfully immobilized on the surface of S-COF.To further substantiate the results, their FTIR spectra were also performed, showing that MAC-FA1/S-COF exhibited the overlapping IR features of both S-COF and MAC-FA1 (Figure S13).
The SEM images of MAC-FA1/S-COF are shown in Figure 2A, where the distinct coral-like structure could still be observed, demonstrating that the loading of MAC-FA1 did not disrupt or affect the morphology of S-COF.The HRTEM images of MAC-FA1/S-COF, as shown in Figure 2B, revealed clear details of each branch of the corallike structure.SEM and TEM elemental mapping results were obtained to verify the uniform distribution of C, N, O, S, and Pd atoms in the 4% MAC-FA1/S-COF (Figures 2D  and S14, and Table S4).In addition, aberration-corrected high-angle annular dark-field scanning transmission electron microscopy (AC-HAADF-STEM) was performed to observe that the single bright dots attributed to isolated Pd atoms were uniformly distributed on the surface of S-COF (Figure 2C).The oxidation state of Pd on the S-COF surface was found to be +2 from the Pd 3d X-ray photoelectron spectroscopy (XPS) patterns (Figure S15).Notably, compared to the pure MAC-FA1, the Pd 3d XPS peaks of the composites were shifted toward a lower binding energy because the electronrich environment caused the Pd electron binding energy to become smaller, further suggesting that MAC-FA1 was successfully combined with S-COF through supramolecular interactions.

Photocatalytic H 2 production
The activity of MAC-FA1/S-COF with 1−5 wt% MAC-FA1 content for photocatalytic water reduction using ascorbic acid (AA) as sacrificial electron donor was first evaluated under visible light in Figure 3A and the photocatalytic equipment is shown in Figure S16.It was found that in the range of 1−4 wt%, the photoactivity increased with increasing of MAC-FA1 content, reaching a maximum at 4 wt%.However, the photocatalytic performance of the composite decreased from 4 to 5 wt% MAC-FA1, which could be attributed to the reduced actual MAC-FA1 content in the composites as determined by Inductively coupled plasma-atomic emission spectrometry (ICP-AES) (Table S3).5% MAC-FA1/S-COF has a lower Pd loading compared to the 4 wt% sample, suggesting that a higher MAC-FA1 cast is rather unfavorable for MAC-FA1 uptake when preparing the composite.It was deduced that a higher content of MAC-FA1 made them easier to stack on top of each other, and the π-π supramolecular interactions between MAC-FA1 nanorings weakened the interactions between MAC-FA1 and S-COF, making it easier for MAC-FA1 to fall off the surface of S-COF.Thus, the H 2 production rate for the samples was in the order of 1% < 2% < 3% < 5% < 4% MAC-FA1/S-COF, among which the highest H 2 generation rate of 100 mmol g −1 h −1 was achieved by the optimized 4% MAC-FA1/S-COF SAC within 5 h.In addition, the apparent quantum yields (AQYs) of 4% MAC-FA1/S-COF were measured under irradiation with monochromatic light at 425, 450, 470, 515, 550, and 590 nm.
The corresponding AQYs of the samples were 2.93%, 3.00%, 2.83%, 1.57%, 0.72%, and 0.48%, respectively, in agreement with the light absorbance results in the UV-Vis absorption spectra of 4% MAC-FA1/S-COF (Figure 1D).H 2 generation of pure MAC-FA1, bare S-COF, M-5/S-COF, Pd/S-COF, and M-5/Pd/S-COF was performed as controlled experiments (Figure 3B) and comparative data are shown in Figure 3C.S-COF and M-5/S-COF showed low photocatalytic performance under visible light irradiation because no Pd catalyst was used and most of the photogenerated electron-hole pairs were recombined.Pd/S-COF exhibited better photocatalytic performance compared to S-COF, suggesting that Pd played a role as an active site on the S-COF surface.The photocatalytic activity of M-5/Pd/S-COF was then further enhanced in comparison to that of Pd/S-COF due to the sensitizing effect of M-5.In addition, MAC-FA1 alone was able to produce a small amount of H 2 .However, their photocatalytic efficiencies were much lower than that of 4% MAC-FA1/S-COF, suggesting that the superior activity of the composite was due to the sophisticated combination and individual atomic distribution of Pd sites in the surface of MAC-FA1/S-COF.
The stability of the photocatalyst is a key indicator for the evaluation of the quality of the catalyst, so the longterm continuous photocatalytic H 2 production experiments of 4% MAC-FA1/S-COF were carried out.As shown in Figure 3D, the SAC exhibited excellent photocatalytic activity and stability, achieving a total H 2 yield of 1830 mmol/g within 20 h.The accumulated turnover number based on the Pd amount (TON Pd ) was calculated to be 437,685, while the turnover frequency (TOF) was basically stable at around 22,000 h −1 , which was one of the highest H 2 production levels achieved for the COF-based photocatalysts (Table S5).To investigate the photocatalytic stability of MAC-FA1/S-COF, the chemical states of the Pd element in the composite were determined by XPS measurement before and after catalysis (Figure S17A).The results showed that for the spent MAC-FA1/S-COF, the Pd element retained a valence of +2, the same as that in the original fresh sample.The used MAC-FA1/S-COF was also characterized by PXRD and the result was compared with that of the original sample (Figure S17B).It was observed that the diffraction peak at 3.46 • became weaker than the initial one, indicating that the photoreduction affected the crystalline phase of the SAC samples.In addition, solid UV-Vis absorption spectra (Figure S17C) and FTIR spectra (Figure S17D) of two samples were measured.No significant changes were observed in their spectra, implying that MAC-FA1/S-COF retained the initial molecular structure after photocatalysis.
The photocatalytic performance of 4% MAC-FA1/S-COF in the laboratory using an Xe lamp and outdoors under natural sunlight using the same equipment was compared (Figure 3E) and the photocatalytic equipment under natural sunlight is shown in Figure S18.The performance of the SAC under both Xe lamp and sunlight was almost identical, showing that the above tests provided a direction for the practical application of this photocatalyst.A photocatalytic test was also carried out under a Xenon lamp using 15 mg of MAC-FA1/S-COF on a rectangular glass substrate measuring 5 cm × 7 cm (Figure S19 and Video S1).It was noteworthy that a significant abundance of hydrogen gas bubbled, which escaped from the surface of the film, could be visually observed under illumination.This phenomenon clearly demonstrated the remarkable capabilities of MAC-FA1/S-COF in photocatalytic hydrogen production.

Mechanism of photocatalysis
To investigate the photocatalytic mechanism of MAC-FA1/S-COF, cyclic voltammetry (CV) and optical tests were performed on MAC-FA1, while impedance potential measurements were conducted on S-COF.The highest occupied molecular orbital (HOMO) value of MAC-FA1 was 0.56 V versus normal hydrogen electrode (NHE) according to the CV results, and its lowest unoccupied molecular orbital (LUMO) value was determined to be −2.07V versus NHE from the HOMO value and the intersection of its UV-V is absorption and fluorescence spectra at 2.63 eV (Figure S20 and Table S6).Thus, the H + reduction potential was much more positive than the LUMO value of MAC-FA1, indicating that MAC-FA1 could effectively act as a PMD for light harvesting and photo conversion in the H 2 reduction reaction using its active Pd 2+ sites.On the other hand, the flat band potential of S-COF (−0.74 V vs. NHE), corresponding to its Fermi level (E f ), was obtained by plotting the Mott−Schottky curve of S-COF in Figure S21A.The difference between E f and S-COF valance band (VB) (Δ(E f /VB)) was 1.59 eV, confirmed by the VB-XPS spectrum of S-COF (Figure S21B), and the VB potential was, therefore, estimated at 0.85 V versus NHE.On the basis of the VB potential of S-COF and the band gap of 2.37 eV (Figure S21C), the CB potential of S-COF was estimated to be −0.95V versus NHE (Figure S21D).Based on the above results, an energy level diagram of MAC-FA1/S-COF was created as shown in Figure 4A.It could be seen that the CB edge of the S-COF had a more negative potential when compared to the HOMO value of the MAC-FA1.The results indicated that the Z-scheme electron transfer from the CB of S-COF to the HOMO of MAC-FA1 was thermodynamically possible, and thus MAC-FA1 could also accept the transferred electrons and use them for H 2 production under visible light irradiation.In addition, the holes on S-COF were favorably quenched by AA because the redox potential of AA (0.47 V vs. NHE) [51] was more negative than the VB edge of S-COF.composite is displayed in Figure 4B, and the other boundary orbitals are shown in Figure S22.For the MAC-FA1/S-COF, the highest VB was located at −5.51 eV, which was mainly distributed on the MAC-FA1 molecule.Another VB, which was 0.13 eV lower than the first VB, was delocalized on the S-COF.The orbital of the CB was also located on the S-COF, while the CB on the MAC-FA1 molecule was 0.2 eV higher than that on the S-COF.In general, the calculated frontier energy levels of the S-COF were slightly lower than those of the MAC-FA1 molecule, which was consistent with the above experimental results.
To validate the electron transfer mechanism of the aforementioned Z-scheme mechanism and determine the direction of electron transfer between components in the composite under visible light excitation, •OH radical trapping experiments were performed and details are described in the Supporting Information.The VB potential of S-COF was measured at −0.85 V versus NHE, more negative than Eox of •OH/OH (+2.70 V vs. NHE). [52]Therefore, it was not possible for •OH to be generated by the holes (h + ) in the VB of S-COF.Instead, the electrons from the CB of the S-COF combined with oxygen to form •OH via a two-electron oxidation pathway, where the specific processes are expressed by the following Equations ( 1)-( 5) (Figure S23) [53] : Terephthalic acid (TA), widely known for its ability to rapidly trap •OH radicals to form the highly fluorescent compound TA-OH, was used in the -OH radical trapping experiments (Figure 5A).When the reaction mixture of S-COF and TA was excited at 315 nm, a robust fluorescence signal was observed at 440 nm.Conversely, the samples containing MAC-FA1 and TA showed negligible fluorescence intensity.This observation indicated that the production of •OH took place exclusively in reactions involving electrons from the CB of the S-COF and not from the MAC-FA1.It was noteworthy that the mixture of 4% MAC-FA1/S-COF and TA showed a weaker fluorescence at 440 nm than the mixture of S-COF and TA.The decrease in fluorescence intensity was ascribed to the incorporation of MAC-1, which took some of the electrons from the CB of S-COF and consequently reduced the •OH levels.Thus, this experiment served as one of the compelling evidences supporting the use of the "Zscheme" electron transfer mechanism in MAC-FA1/S-COF photocatalysis.
Steady-state photoluminescence (PL) spectra were performed on the 4% MAC-FA1/S-COF, MAC-FA1, and bare S-COF samples to further elucidate the Z-scheme electron transfer mechanism (Figure 5B).When excited at 400 nm, bare S-COF showed a prominent fluorescence peak around 650 nm, due to strong recombination of photogenerated electron-hole pairs.Under the same conditions, MAC-FA1 displayed low fluorescence intensity, which was attributed to its ligand-to-metal charge transfer (LMCT) transitions.As expected, when SCOF was combined with MAC-FA1, the fluorescence intensity of the 4% MAC-FA1/S-COF sample was significantly reduced in comparison to S-COF alone.This reduction was attributed to the partial transfer of photo-generated electrons from S-COF to MAC-FA1, which influenced the overall fluorescence performance.It was, therefore, demonstrated that the MAC-FA1/S-COF heterojunction used the "Z scheme" electron transfer mechanism in photocatalysis.
The surface photovoltage (SPV) measurement was then performed to investigate the process of photogenerated charge separation and transfer that occurred at the surface and interface of MAC-FA1/S-COF.From Figure 5C, the bare MAC-FA1 exhibited a minimal SPV response within the recorded light wavelength range, while the S-COF showed an SPV response signal in the UV-Vis region of 300-600 nm due to its inherent transition, indicating that the photogenerated charge carriers could indeed achieve separation at the surface and interface of S-COF. [54]Notably, a significantly stronger SPV signal was observed for the 4% MAC-FA1/S-COF under the same condition, suggesting a more effective separation and transfer of photogenerated charges that occurred spatially, aided by the interfacial electric field, [55] which was an auxiliary evidence for the existence of the Z-scheme mechanism.
Electrochemical impedance spectroscopy (EIS) measurements were performed on the S-COF, MAC-FA1, and 4% MAC-FA1/S-COF electrodes under dark conditions to investigate the recombination resistance of the carriers, and the Nyquist plots are shown in Figure 5D.Based on the equivalent circuit diagram, the arc curvatures represent the charge recombination resistance (R 2 ) at the sample/electrolyte interface, and R 1 is the sum of the electrolyte resistance and other resistances in its system. [56]A lower recombination resistance of the charge carriers is typically indicated by a smaller radius of curvature of the arcs. [57,58]As shown in Figure 5D, the radii of the semicircles for the three materials followed the order of 4% MAC-FA1/S-COF > MAC-FA1 > S-COF, indicating that 4% MAC-FA1/S-COF had the highest recombination resistance of the charge carriers, thus facilitating an efficient charge transfer, which was beneficial for the photocatalytic performance.This confirmed that Z-scheme electron transfer from the CB of S-COF to the HOMO of MAC-FA1 could contribute to the slow carrier recombination in the composite.
Transient absorption spectra (TAS) of S-COF and MAC-FA1/S-COF were also measured to detect the Z-scheme charge transfer mechanism (Figure S24A,B).Two samples showed broad negative bleaching signals in the 450−650 nm region, associated with the generation of excited electrons. [59,60]By fitting the kinetic plots of the TAS, the relationship between the shallow and deep traps could be determined. [61,62]As shown in Figures S24C and S24D, S-COF and MAC-FA1/S-COF were analyzed using a biexponential model to give two lifetimes.The short lifetime τ1 was attributed to electron diffusion across lattices (shal-low trapping) and the other longer τ2 could be ascribed to charge recombination of photogenerated electrons with the trapped holes (deep trapping).The τ1 (0.45 ps) and τ2 (7.9 ps) of MAC-FA1/S-COF were significantly shorter than those (τ1 = 1.28 ps and τ2 = 25.7 ps) of S-COF, indicating the existence of interfacial charge recombination from the CB of SCOF to the HOMO of MAC-FA1, in agreement with the above results of PL, SPV, and EIS measurements.
Therefore, the Z-scheme photocatalytic mechanism of the MAC-FA1/S-COF photosystem was derived as follows (Figure 5E): The HOMO and LUMO values of MAC-FA1 were at 0.56 and −2.07 V versus NHE, while the VB and CB positions for S-COF were at 0.85 and −0.90 V versus NHE, respectively.Both MAC-FA1 and S-COF produced photogenerated electrons and holes under visible light, as evidenced by the suppressed charge recombination and promoted charge separation.The holes tended to remain at the VB of S-COF, while the photoelectrons were transferred from the CB of S-COF to the HOMO position of MAC-FA1, and thus this Z-structure led to an easy and highly efficient spatial separation of the photoelectron-induced charge carriers.Accordingly, the MAC-FA1 functioned both as electron acceptors and as virtual catalysts for H 2 production under visible light.The destination of the photoelectrons from the HOMO to the LUMO position of MAC-FA1 was active Pd metal sites for the H 2 reduction reaction.Meanwhile, the holes left on the VB of S-COF were consumed by the sacrificial agent AA.Therefore, under visible light excitation, the electron-hole pairs could efficiently separate and transfer, facilitating continuous and effective photocatalytic hydrogen production in the presence of a sacrificial agent.

CONCLUSIONS
In
The tube was sealed during evacuation and heated at 120 • C for 3 days.The precipitate was collected by centrifugation and washed with N,N-dimethylformamide and tetrahydrofuran.After drying at 60 • C under a vacuum overnight, the product was obtained as a deep red powder in a yield of 82%.
In addition, a Pd/S-COF sample with the same Pd content as 4% MAC-FA1/S-COF was prepared by the photodeposition method using H 2 PdCl 4 as the Pd source under a Xenon lamp for 2 h.Similarly, a comparative M-5/Pd/S-COF sample with the same Pd and M-5 content as 4% MAC-FA1/S-COF was prepared by loading M-5 on Pd/S-COF using the same procedure as the above-mentioned MAC-FA1/S-COF.

Photocatalytic hydrogen production under Xe lamp
The photocatalytic experiments were performed on an allglass automatic online trace gas analysis system (Labsolar-6A, Beijing Perfectlight Technology Co., Ltd.) connected to an FULI GC9790 PLF-01 gas chromatography.The photocatalyst (2 mg) was dispersed into 0.1 M AA water solution (60 mL) in a quartz reaction cell which was assembled with the online system.After removing the dissolved air in the photocatalytic system, it was degassed and irradiated by a 300 W Xe lamp (Perfect Light PLS-SXE300, light intensity = 1500 mW cm −2 ) equipped with an optical filter (λ > 420 nm) under vacuum.The amount of hydrogen released was measured by online gas chromatography at 1 h intervals.
The procedure for carrying out the photocatalytic experiment on pure S-COF loaded with Pt is analogous to that described above.It consisted of adding an aqueous solution of H 2 PtCl 6 (3 wt%) to the reaction mixture before starting the reaction.

Trapping experiments of hydroxyl radicals
Dissolve TA (5 × 10 −4 mol L −1 ) in an aqueous solution of sodium hydroxide (NaOH) (2 × 10 −3 mol L −1 ) and place in a 40-mL quartz flask.A photocatalyst (5 mg) was then added.The mixture was irradiated with a 300 W Xe lamp (λ > 420 nm) for 1 h, then the supernatant was collected by centrifugation and the fluorescence of the supernatant was detected using a fluorescence spectrometer.

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 conflict of interest.

F
I G U R E 3 H 2 yield-time plots of (A) MAC-FA1/S-COF with different MAC-FA1 contents and (B) other comparison samples under Xe lamp (λ > 420 nm) for 5 h, (C) H 2 yield columns of all samples in 5 h, (D) TON and TOF curves of 4% MAC-FA1/S-COF for 20 h, and (E) H 2 yield-time plots of 4% MAC-FA1/S-COF under sunlight (including sunlight intensity) and Xe lamp.
Computational simulations were performed to validate the accuracy of an energy level diagram of MAC-FA1/S-COF.The simulated energy level of the MAC-FA1/S-COF F I G U R E 4 (A) Energy level diagrams of S-COF and MAC-FA1, and (B) the energy levels of the MAC-FA1/S-COF composite and its decomposed orbitals.All energy levels were realigned to the vacuum layer.

F
I G U R E 5 (A) Liquid PL spectra for TA-OH at 315 nm excitation, (B) steady-state solid PL spectra, (C) SPV spectra, (D) EIS Nyquist plots in the dark of S-COF, 4%MAC-FA1/S-COF, and MAC-FA1, and (E) possible working mechanism of the MAC-FA1/S-COF system.

A
C K N O W L E D G M E N T S This work was supported by financial support from the National Natural Science Foundation Project of China (grant nos.21975291 and 22171211) and NSF of Guangdong Province (grant nos.2022A1515011949 and 2020A1515110474).