Two-dimensional p-π conjugated amide-linked polymeric photocatalysts with enhanced carrier separation Authors: Chang He, Weixu Liu, Xing Cao, Xin Tan, Junguo Ma, Yu Chen, Haijun Yang, Yongfa Zhu, Bo Wang, Chen Chen Published: 2026 ABSTRACT AbstractTwo-dimensional (2D) polymers have recently emerged as a hot topic in photocatalysis because of their rich active sites, tunable topology, and electronic structures. However, they typically suffer from severe photogenerated charge recombination, which hinders further elevation in photocatalytic performance. Here, we develop a method of direct condensation between amino and carboxyl groups to prepare a series of 2D polyaramids (2DPAs). Compared with the conventional fully π-conjugated 2D polyimide (2DPI), one of the 2DPAs (denoted as 2DPA-PPA) has an almost identical molecular skeleton but features a close yet spatially separated HOMO and LUMO distribution (the recombination rate is below 1/3 that for 2DPI, and the charge separation state lifetime is 5.5 times as long). Consequently, the rate of photocatalytic O2 activation over 2DPA-PPA is at least 5 times higher than that over 2DPI. Our findings on the preparation of 2DPAs and catalytic mechanism would be of guidance for designing high-performance photocatalysts.Subject terms: Photocatalysis, Metal-organic frameworks Introduction Two-dimensional (2D) polymers, as an emerging class of 2D materials, have rich catalytic active sites benefiting from the large surface area1; in addition, the organic modules constituting the 2D polymers are highly diversified, which offers immense opportunities for further optimizing their topological and electronic structures2–4. As a result, 2D polymers have been considered promising candidates for high-performance photocatalysts5,6. However, 2D polymer generally suffers from the critical issue of severe photogenerated charge recombination, which constrains their practical application1. The carrier separation efficiency in polymers could be further elevated by designing and constructing donor–acceptor (D–A) systems1,7,8. Nonetheless, most organic 2D polymers are linked via conjugated double bonds9–13. Such fully π-conjugated structures, while boosting the transport of carriers7,14,15, would also facilitate their recombination, and therefore the resulting improvement of carrier efficiency would be rather limited. This issue could be addressed by manipulating the linkages between organic modules in the 2D polymers, as the tuned electronic structure would potentially suppress the charge recombination. In an amide bond, the lone-pair electrons on the nitrogen atom and the π* orbital of the carbonyl group are conjugated16–18, and the 2D polymers linked via amide bonds would feature a planar structure (at least to a certain degree)19–22; in addition, the nitrogen atom is sp3 hybridized, which would endow the resulting 2D polymers with electronic structures different from those based on conventional linkages (such as C = N bond9,10 and C=C bond11–13,15). The chemically stable amide bonds would also confer excellent mechanical strength and catalytic stability23,24. However, the amide bond is generally considered irreversible (because amide bonds could be hydrolyzed only under strongly acidic or alkaline conditions25). The synthesis of 2D polymers based on the irreversible amide bond is rather challenging20–22, as an ideal synthetic method needs to suppress the growth of out-of-plane branches (which could readily occur and result in irreversible 3D structures), and also needs to overcome the high barrier of maintaining the low-entropy in-plane bonding26,27. Recently, Strano et al. made an advance in this area by employing a reaction between trimesoyl chloride and melamine, and thus obtained 2D polymer features with high mechanical strength originating from the chemical stability of amide linkage19. And also, Wan et al. reported a strategy of direct synthesis of amide-linked 2D polymer network via ester-amine exchange reaction28. However, limited precursor accessibility and stringent reaction conditions still restrict the widespread application of 2D polyaramids. Additionally, thus far, investigations on the photophysics and photocatalysis of these 2D polymers have been relatively scarce. Here in this work, we establish a method of direct condensation between amino-group-containing and carboxyl-group-containing modules, which proceeds without the need for prefunctionalized precursors, degassing treatments (e.g., typical three freeze–pump–thaw cycles), or any special pretreatment of the reaction system, and thereby prepared a series of 2D polyaramids (2DPAs, Fig. 1) with high chemical stability under ambient or near-ambient pressure, which could substantially broaden the scope of precursors for the synthesis of 2DPAs. One of the 2DPAs that we obtained (denoted as 2DPA-PPA) has a molecular skeleton almost identical to that of the conventional fully π-conjugated 2D polyimide (2DPI), but features a close yet spatially separated HOMO and LUMO distribution as evidenced by DFT calculation, UV–Vis–NIR diffuse reflectance spectroscopy (DRS), and ultraviolet photoelectron spectroscopy (UPS). Such a distinctive electronic structure of 2DPA-PPA could facilitate the separation of electron-hole pairs and suppress their recombination. As a result, the carrier separation efficiency and the photocatalytic performance in oxygen activation are significantly enhanced, including photocatalytic oxidation of thioanisole to benzyl methyl sulfoxide, photocatalytic oxidative coupling of benzylamine, photocatalytic oxidation of benzylalcohol to benzaldehyde, and hydrogen peroxide photosynthesis. Transient absorption spectroscopy (TAS), in-situ electron paramagnetic resonance (EPR), and in-situ light-irradiated X-ray photoelectron spectroscopy (XPS) were conducted to confirm the existence of 1,3,5-tris(4-aminophenyl)benzene (TAPB) radical cation moieties, which serve as the catalytically active sites. And the TAPB·+ sites generated in our 2DPAs all have longer lifetimes than 2DPI, exhibiting the characteristics of the heterogenization of TAPB molecular catalysts. Our findings here on the preparation of 2D polymers linked via the irreversible amide bond and on the related photocatalytic mechanism not only would remarkably extend the scope of available precursor modules, but also could offer new perspectives for the design of advanced 2D polymer photocatalysts and the heterogenization of small-molecule catalysts. Schematic illustration of the synthesis of 2D polyaramid materials, named 2DPA-PPA, 2DPA-Ala, 2DPA-Gly, and the presence of internal D–A interactions. D and A correspond to donor and acceptor. Results Synthesis and characterization of 2DPAs Herein, we report the synthesis of a series of 2DPAs (Fig. 1). Our synthesis strategy is relatively simple and direct, which involves an amide condensation of C3-symmetric TAPB and C2-symmetric terephthalic acid (PPA). The 2D polymer based on the irreversible amide bonds was synthesized by refluxing under near-ambient pressure in the double solvent of THF and toluene, with HOBT and HBTU serving as condensing agents. For the C2-symmetric organic building block, apart from the pristine PPA, we also used bis(glycine) terephthalic acid (Gly-PPA-Gly) and bis(alanine) terephthalic acid (Ala-PPA-Ala), each of which features two amino-acid moieties linked to the PPA backbone to adjust the length of the amide bonds. We found that this synthesis strategy indeed enables 2D condensation in the solution phase and produces a series of 2D polyaramids, denoted as 2DPA-X (2DPA-PPA, 2DPA-Ala, and 2DPA-Gly). Here, we take 2DPA-PPA as an example to illustrate the synthesis of 2DPA. FT-IR study confirms the formation of amide linkages by both amide I (1600–1700 cm−1) and amide II (1500–1600 cm−1) regions29 (Fig. 2a). In 13C ssNMR spectra (Fig. 2b), the presence of TAPB and PPA skeleton peaks indicates the chemical structure of polyaramid. Thermogravimetric analysis (TGA) shows that the thermal stability of 2DPA-PPA has been enhanced up to 200 °C (Fig. S6), corresponding to the formation of chemically stable amide bonds. The inert amide linkage allows harsh acid and alkaline treatment on 2DPA-PPA, whose structure is still maintained, as demonstrated by the FT-IR (Fig. S7). Characterization of 2D polyaramid materials. a Fourier transform infrared (FT-IR) spectra of as-synthesized 2DPA-PPA (purple), PPA (cyan), and TAPB (yellow) powders; b 13C Solid-state NMR (ssNMR) spectrum of as-synthesized 2DPA-PPA powder. Inset: the carbon atoms identification results of 2DPA-PPA. Note that the peak marked with an asterisk is attributed to the methyl of the small amount of residual DMF. c Powder X-ray diffraction (PXRD) patterns of 2DPA-PPA (purple), 2DPA-Ala (green), 2DPA-Gly (orange), and substrate (gray). d UV–vis–NIR diffuse reflection spectra (DRS) of TAPB (yellow), 2DPA-Ala (orange), 2DPA-Gly (green), 2DPA-PPA (purple), and 2DPI (red) powders; e CO2 sorption isotherm profiles of 2DPA-PPA (purple), 2DPA-Ala (green), 2DPA-Gly (orange), and the mixture of monomers (gray). f Low-electron-dose cryo-EM image of 2DPA-PPA. Inset (left): the locally enlarged image of 2DPA-PPA. Inset (right): scanning electron microscopy (SEM) image of 2DPA-PPA. SEM images (Fig. 2f inset and S3) and TEM images (Fig. S4) reveal the morphology of stacked nanosheets for 2DPA-PPA. In the PXRD patterns, the broad peak centered at 20.27° corresponds to an average interlayer spacing of 4.376 Å and a mean ordered domain size of 13 Å. To confirm the 2D characteristics of the 2DPA-PPAs, we initially performed HRAFM testing after conventional ultrasonic exfoliation (Fig. S11). However, owing to its strong aggregation tendency induced by interlayer hydrogen bonding, as evidenced by the dashed circle in Fig. 2a, individual molecules could hardly be exfoliated. Therefore, we used the method of chemical exfoliation19 by silylation of amide bonds (Fig. S12), to obtain a high ratio of individual molecules with an average thickness of 0.433 ± 0.040 nm (Figs. 3a, d, g and S15). This monomer-scale thickness has been generally recognized as an unambiguous signature of 2D polymerization6. The 2D characteristics can also be proved by grazing-incidence wide-angle X-ray scattering (GIWAXS, Fig. S16). 2DPA-Ala and 2DPA-Gly are also proven to feature 2D structure, with an average thickness of 0.416 ± 0.033 nm (Figs. 3b, e, h and S13) and 0.431 ± 0.038 nm (Figs. 3c, f, i and S14), respectively (detailed data shown in Supplementary Information). In the low-electron-dose cryo-EM image, disordered pores can be observed for 2DPA-PPA (Fig. 2f). In addition, 2DPAs all showed CO2 uptake ability at 273.15 K (Fig. 2e), whereas the physical mixture of the monomers (TAPB + PPA) did not show apparent adsorption. These results confirmed that polymeric layers of 2DPAs with conformational flexibility can still create nanosized apertures. We also found the bandgap of as-synthesized 2DPA-PPA is 1.93 eV, significantly lower than the 2.40 eV of structurally similar 2DPI, indicating the influence of amide bonds on the electronic structure of the semiconductor (Figs. 2d and S18). Height characterization of 2D polyaramid. High-resolution atomic force microscopy (HRAFM) images of TMS-2DPA-PPA (a), TMS-2DPA-Ala (b), and TMS-2DPA-Gly (c). d–f, Height profiles along the light green lines from top to bottom indicated in (a–c) (from top to bottom: red, green, cyan, purple, and orange). Height distribution of observed TMS-2DPA-PPA and 2DPA-PPA (g), TMS-2DPA-Ala and 2DPA-Ala (h), TMS-2DPA-Gly and 2DPA-Gly (i). Photocatalytic oxygen activation and the corresponding mechanism Arylamines and their analogs (such as TAPB)30,31 can produce cation radicals (generated from electrochemistry and irradiation via electron transfer) for catalysis or oxidation. Thus, 2DPAs could also be considered as metal-free heterogeneous photocatalysts for photocatalytic oxidation of tetrahydroquinoline under visible light (Fig. 4a). The pathway of photocatalytic oxygen activation was easily verified by exceedingly low conversions in the absence of light or oxygen (Table S2). 2DPA-PPA outperformed 2DPA-Ala, 2DPA-Gly, and the small-molecule catalyst, TAPB. Furthermore, 2DPA-PPA could be easily isolated by filtration and reused for the next cycle, showing no discernible decline in photocatalytic activity after five runs (Fig. 4b). The structure and morphology also show negligible changes after reactions (Fig. S20–S22), demonstrating the high stability of polymeric photocatalysts conferred by highly chemically stable amide bonds. We carried out a series of comparative experiments as well as quenching experiments on the possible active species (Fig. S24). 1,4-benzoquinone (BQ) or 1,4-Diazabicyclo[2.2.2]octane (DABCO) was added as the scavenging agent for ·O2− or 1O2, respectively. The results showed that with the scavenging agent added, the conversion declined to 20.9% (for ·O2−) or 76.1% (for 1O2), indicating that ·O2− is the main active species. The signals of ·O2− and 1O2 were also detected in the in-situ light-induced EPR tests (Fig. S25). In addition, we added the scavenging agent for holes (h+), and the performance declined to 30.89%, indicating that holes are also the main active species. The energy band structure (Fig. S32) was determined via UV–vis–NIR DRS (for optical band gap, Figs. 2d and S18), UPS (Fig. S30), and XPS valence spectra (for Fermi level and valence band position, Fig. S31), indicating that it can thermodynamically activate oxygen to superoxide radicals. These results suggest that both the ·O2− and holes function as the crucial active species to participate in catalytic oxidation. Moreover, 2DPA-PPA has also shown the capability of catalyzing the oxidative dehydrogenation of a number of other tetrahydroquinoline derivatives (Table S3). And more, the production rate of H2O2 over 2DPA-PPA can reach up to 37.1 μmol h–1 or 3.71 mmol g–1 h–1 (Figs. 4c and S33). Photocatalytic performance and mechanism in the oxidation of tetrahydroquinoline to quinoline. a Conversions of tetrahydroquinoline over different photocatalysts, the number of independent measurements is four. Conditions: 1 atm oxygen, 298 K, 0.1 mmol tetrahydroquinoline, 10 mg catalyst, 3 mL DMF, 8 h, visible light (>400 nm) from Xe Lamp (148 mW cm−2). It should be noted that the selectivity is over 99.9% unless otherwise stated. b Catalytic durability of 2DPA-PPA, the number of independent measurements is three. c Production of quinoline and hydrogen peroxide over 2DPA-PPA. d Hydrogen peroxide photosynthesis over 2DPA-PPA and 2DPI utilizing the two-phase system, the number of independent measurements is two. Conditions: 1 atm oxygen, 298 K, 5 mg catalyst, 2 h, visible light (>400 nm) from Xe Lamp (148 mW cm−2), and the two-phase systems composed of an aqueous phase (1 mL) and a p-methoxybenzyl alcohol phase (2 mL), in which the catalyst is primarily dispersed in the organic phase. e Schematic diagram of the mechanism for photocatalytic oxidation of tetrahydroquinoline to quinoline. TAS, in-situ XPS, in-situ EPR, and ESEEM were conducted to study the catalytically active sites of 2DPA-PPA and the difference in catalytic mechanisms between 2DPA-PPA and the small-molecule catalyst, TAPB. Considering that the N-centered radical cations are readily produced from arylamine analogs after irradiation, we characterized the generation and annihilation of the TAPB·+ in transient absorption spectra, shown in Fig. 5a, b, d, e. Femtosecond-TAS (fs-TAS) of photocatalysts dispersed in DMF was recorded with irradiation by a 400 nm fs-laser. The intense excited-state absorption (ESA) peaks from 550 to 1350 nm indicate the formation of TAPB·+, which we have confirmed by electrochemical/chemical oxidation, UV–Vis–NIR absorption spectroscopy (Figs. 5c and S37). Additionally, as shown in Fig. 5g, the signal of TAPB·+ was also observed in the solid-state EPR spectra of the four materials32,33. To prove it, we conduct the continuous wave EPR (CWEPR) under modulated temperature, atmosphere, and photoillumination, and, moreover, we conduct the pulse EPR, providing evidence for the formation of the TAPB·+. In-situ EPR is a good tool to explore the charge transfer process in photocatalytic reactions. Thus, we conducted in-situ EPR on 2DPA-PPAs at 298 K under dark and light irradiation conditions, then under oxygen atmosphere, and then with the addition of methanol, respectively. In Fig. 5h, the results showed that the EPR signal of 2DPA-PPA increased after light irradiation compared with that under dark conditions, indicating the formation of more radicals. Even in the absence of LED light irradiation, 2DPA-PPA still exhibits EPR signals of the TAPB·+, possibly induced by ambient indoor light and sunlight. When injecting the oxygen, the EPR signal increases significantly, indicating that the formed radicals are organic radicals, rather than the electrons in CB, which will display a decrease of the signal in the presence of oxygen34,35. And more, we inject the methanol into the tube, which plays the role of the quencher to quench the radical cation. The significant decrease of the EPR signal proved the formation of the radical cations. We also conducted the EPR of the I2-doped 2DPA-PPA (Fig. S34d), and the enhanced peaks observed at g = 2.0038 clearly confirmed the enhanced concentration of TAPB·+. And subsequently, we simulated the EPR signal of the TAPB radical cations, obtaining the simulated spectrum shown in Fig. 5i, which shows an excellent match with the measured EPR spectra. In-situ light-irradiated XPS experiments further confirmed the existence of TAPB·+. As shown in Figs. 5f and S36, the N 1s peaks can be deconvoluted into two peaks. It is well known that the binding energy of the inner electron is enhanced once the outer electron density decreases. After being illuminated, the relative proportion of the higher binding energy peak increases, indicating the generation of an electron-deficient N atom. Therefore, the two peaks convoluted from the N 1s peaks are identified as C(=O)–N–C and its corresponding excited state C(=O)–N·+–C36,37, which further confirms the generation of TAPB·+. For the samples of the 2DPAs and 2DPI, the area ratios of the excited state (higher binding energy) of N1s after illumination increase with respect to the ones in the dark, which indeed confirms the formation of an electron-deficient N atom belonging to TAPB·+. Since N cation radicals have a strong oxidation ability and TAPB small-molecules can catalyze the dehydrogenation of tetrahydroquinoline by themselves, we confirmed the heterogenization of small-molecule catalyst TAPB. The emergence of TAPB·+ signals also indicates that TAPB serves as the donor to constitute the D–A system with PPA. The presence of TAPB·+. Fs-TAS 2D plots of 2DPI (a),2DPA-PPA (b), 2DPA-Ala (d), and 2DPA-Gly (e) dispersed in DMF following excitation at 400 nm. c Differential absorption spectra of oxidized TAPB·+ obtained upon chemical oxidation using Magic blue at different equivalents. f In-situ light-irradiated XPS N 1s spectrum of 2DPA-PPA. g Solid-state EPR spectra of 2DPI (red), 2DPA-PPA (purple), 2DPA-Ala (orange), and 2DPA-Gly (green) powders. h In-situ EPR experiments, involving illumination, oxygen introduction, and methanol addition for 2DPA-PPA. i EPR spectrum of 2DPA-PPA and the simulated spectrum using EasySpin. The HOMO and LUMO distribution of 2DPA-PPA matches well with donor–acceptor characteristics of organic building blocks (Fig. 6b). We found that the HOMO is distributed mainly on TAPB, and the LUMO is distributed mainly on PPA. A charge transfer absorption band (650–1000 nm) is also observed in the DRS (Fig. 2d), further demonstrating the D–A charge transfer process in 2DPA-PPA38. Based on the above results, we propose the mechanism shown in Fig. 4e39–41. The photogenerated electron-hole pairs are separated with the assistance of the D–A system, with the holes remaining on the TAPB to generate TAPB·+ and the electrons hopping to the PPA. The TAPB·+ can rapidly oxidize the nearby tetrahydroquinoline molecules to produce the tetrahydroquinoline radical cations, and the photogenerated electrons combine with oxygen to produce ·O2−, which then participates in the dehydrogenation process of tetrahydroquinoline cation radicals to generate the partial dehydrogenation product and H2O2. The dihydroquinoline would be further oxidized to produce quinoline. The mechanism was also confirmed by environment-modulated TAS and TPV. As for environment-modulated fs/ns-TAS (Figs. S40, S41, and S44, the detailed analysis shown in SI), compared to the argon atmosphere, a longer CS state lifetime of up to 56.5 ns was measured in air and oxygen atmospheres (especially oxygen atmosphere), which further indicates the role of oxygen in stabilizing TAPB·+, as analyzed in detail in SI. Meanwhile, we compared the CS state spectrum with TPB·+, TAPB·+, and TAPB-AMI·+ differential absorption spectroscopy (Fig. S37a). By comparison of curves and the report of MacKenzie et al.42, the differences within the 600–800 nm range are assigned as PPA radical anions. For TPV (Fig. S45), after excitation, the photogenerated electrons migrate to the surface, and thus all materials display a negative surface photovoltage signal. As the oxygen concentration increases, the kinetic results demonstrate a change from 124.65 microseconds in argon atmosphere to 101.73 microseconds in air atmosphere, and further down to 34.65 microseconds in oxygen atmosphere. These results proved that the predominant species migrating to the surface of 2DPA-PPA are photogenerated electrons, which can reduce oxygen to superoxide anion radicals. To understand the photocatalytic reaction mechanism insight, we conducted the in-situ diffuse reflectance infrared Fourier transform (DRIFT, Fig. S43) spectrum of 2DPA-PPA under tetrahydroquinoline vapor and O2 feeding, coupled with visible-light irradiation. By prolonging the exposure time, we can observe the occurrence of O2 adsorption (940 cm−1), tetrahydroquinoline adsorption (~3500 cm−1), and the production of quinoline (~3200 cm−1) on the surface of the photocatalyst, which is consistent with the proposed mechanism. The role of the amide bond for photogenerated carrier separation. The HOMOs (green and cyan) and LUMOs (yellow and purple) distribution of 2DPI (a), 2DPA-PPA (b), 2DPA-Ala (c), and 2DPA-Gly (d) (b3lyp/def2TZVPP, isovalue = 0.03). C, N, O, and H atoms are presented in peach, blue, red, and white, respectively. e The charge recombination rate of 2DPI (red), 2DPA-PPA (purple), 2DPA-Ala (orange), and 2DPA-Gly (green) deduced from the Marcus equation. f Time profiles of fs-TAS probed at 1150 nm for 2DPI (red), 2DPA-PPA (purple), 2DPA-Ala (orange), and 2DPA-Gly (green). g TPV and fitting curves of 2DPI (red), 2DPA-PPA (purple), 2DPA-Ala (orange), and 2DPA-Gly (green). Note that the TPV data has been normalized to give a better comparison. h The surface potential measured by the Kelvin probe force microscope (KPFM) overlay on topography for 2DPA-PPA. Enhancement of carrier separation via amide bonds We found the catalytic performance of 2DPA-PPA is much better than that of 2DPI (which has an almost identical molecular skeleton to 2DPA-PPA), in tetrahydroquinoline oxidation (Figs. 4a and S33), thioanisole oxidative (Fig.S46), benzylamine coupling (Fig. S47), benzylalcohol oxidation (Fig. S48), and hydrogen peroxide photosynthesis utilizing the two-phase system (Fig.4d). Because of the electronic structure of the amide bond, we speculate that the as-synthesized 2DPAs are endowed with an electronic structure distinct from that of 2DPI, thereby enhancing carrier separation and catalytic efficiency. Given that TAPB·+ is the catalytic oxidation site, the decay kinetics of TAPB·+ are derived from the time profiles of fs-TAS probed at 1150 nm, as shown in Figs. 6f, S49–S50 and Table S5. Two decay processes were identified for each material, corresponding to the rapid charge recombination and the long-lived charge separation state (the principal active species involved in the photocatalytic process) relaxation and recombination process. Compared with that in 2DPI, the average lifetimes of TAPB·+ in 2DPAs are significantly prolonged. 2DPA-PPA, with the best performance, has the longest average lifetime of 696.2 ps (5.5 times that of 2DPI). 2DPA-Ala and 2DPA-Gly also feature long average lifetimes of 401.2 ps and 454.4 ps, respectively, consistent with their analogous topological structures and comparable catalytic performances. These results indicate that 2DPI is the most prone to charge recombination. At the same time, 2DPA-PPA, which features both planarity and sp3 hybridization characteristics, exhibits the strongest carrier separation ability, thereby presenting the best photocatalytic oxidation performance. The signal of TAPB·+ was also observed in the solid-state EPR spectra of the four materials32,33. The direct comparison of the four materials and their semi-quantitative ratios with the standard 55Mn2+ signal (S/M) indicates that the signal of 2DPA-PPA is the most intense, and the signal of 2DPI is the weakest (Figs. 5g and S51). This trend indicates that 2DPA-PPA can generate the most TAPB·+, which shows that 2DPA-PPA has the highest carrier separation efficiency. From in-situ XPS (Figs. 5f and S36), the most significant increase in the area ratio of the excited state (higher binding energy) of N1s after irradiation with respect to the ones in the dark is observed over 2DPA-PPA, indicating that 2DPA-PPA can generate more TAPB·+ radical cations under the same illumination, confirming the strong charge separation ability of 2DPA-PPA. Moreover, TPV spectra (Fig. 6g) show that 2DPA-PPA features the longest lifetime of photogenerated electrons. The lifetime of photogenerated electrons in 2DPA-PPA reaches 103.02 µs, whereas that in 2DPI is only 11.82 µs; these data match well with the trend of photogenerated hole lifetimes measured via fs-TAS, light-induced EPR, and in-situ light-irradiated XPS. At the same time, we carried out in-situ light-irradiated KPFM measurements to obtain the surface photovoltage at higher spatial resolution for these materials43. As seen in Figs. 6h and S55, the surface potentials of four samples are all reduced under irradiation compared to the dark condition. The surface potential of 2DPA-PPA significantly reduces by 150 mV with the light on and then recovers rapidly after light off. In comparison, 2DPI only shows a 10 mV potential drop, confirming that more photogenerated electrons migrate to the surface of 2DPA-PPA under irradiation, which means the higher carrier separation efficiency of 2DPA-PPA, consistent with the results obtained by TPV. 2DPA-PPA also presents the most intense photocurrent response and remarkably prolonged fluorescence lifetime compared with that of 2DPI (Figs. S52 and S54), which further explains the vast difference in photocatalytic performance. In order to interpret the effect of amide bonds on carrier separation from the perspective of bonding modes, we calculated the HOMO and LUMO distributions for repeated fragments of the four materials, as shown in Fig. 6a–d. The HOMO and LUMO of fully conjugated 2DPI are the closest to each other under the same isosurface. Although this distribution is most conducive to carrier transport, it is also most prone to charge recombination, which would lead to poor catalytic performance. The introduction of the amide bond results in significant alterations in the molecular configuration and the electronic structure. The sp3-hybridized N atom of the amide bond renders the PPA fragment twisted to some extent relative to the TAPB fragment, which ensures that the HOMO and LUMO are separated spatially by the benzene ring, thus remarkably suppressing charge recombination. Moreover, the partial planarity of the amide bond also facilitates carrier transport. Further increasing the linkage length (as in 2DPA-Ala and 2DPA-Gly) makes the HOMO and LUMO farther apart, and the introduction of sp3 carbon atoms also impairs the planarity, thus hindering electron and hole transport. Nevertheless, benefiting from the distinctive HOMO/LUMO distributions (which suppress charge recombination even more), their catalytic performances are still better than that of 2DPI (Fig. 4a). The charge recombination rate (kCR) deduced from the Marcus equation gives a quantitative comparison of carrier separation efficiency, as shown in Fig. 6e. The kCR of 2DPA-Ala with two sp3 carbon atoms in the fragment is the lowest, followed by the 2DPA-Gly with one sp3 carbon atom. By virtue of the distinctive electronic structure of the amide bond, the kCR of 2DPA-PPA is even lower than 1/3 of that for 2DPI. The analysis of the above results demonstrates the crucial role of amide bonds on the electronic structure of 2D polymers. Discussion In summary, we established a method of direct condensation between amino-group-containing and carboxyl-group-containing modules to prepare a series of 2DPAs, showing the potential for gram-scale synthesis. With an almost identical molecular skeleton, compared with 2DPI, 2DPA-PPA shows a much higher carrier separation efficiency, which originates from the close yet spatially separated HOMO and LUMO distribution resulting from the distinctive electronic structure of amide bonds. The photocatalytic O2 activation over 2DPA-PPA is at least 5 times higher than that over 2DPI, respectively, further demonstrating the efficacy of tuning the electronic structure of 2D polymers by introducing amide bonds. The 2DPA-PPA also shows great cyclic stability due to the strong chemical stability of amide bonds. The investigation of substrate scope and stability shows the generality and practical relevance of this photocatalytic system. This polymerization method also offers an effective route to the heterogenization of a wide range of small-molecule catalysts. In addition, the distinctive electronic structure of amide bonds offers a new model for interpreting the structure-activity relationships of photocatalysts, which would promote the application of polymeric photocatalysts in both theoretical and experimental aspects. Methods Materials All reagents were used as received without further purification. O-(Benzotriazol-1-yl)-N,N,N’,N’-tetramethyluronium Hexafluorophosphate (HBTU, 99%, Sigma), 1-Hydroxybenzotriazole (HOBT, 99%, Sigma), diisopropylethylamine (DIEA, 99%, Sigma), 1,3,5-tris(4-aminophenyl) benzene (TAPB, 99%, TCI), L-glycine (Gly, 99%, Adamas), L-alanine (Ala, 99%, Adamas), terephthalic acid (PPA, 99%, TCI), p-phthalaldehyde (PDA, 99%, TCI); acetic acid, tetrahydroquinoline, biphenyl and tetrahydroquinoline derivatives were purchased from Aldrich. Synthesis of Gly-PPA-Gly and Ala-PPA-Ala Ala-PPA-Ala was synthesized based on the previously reported method44. A water solution (30 mL) of L-alanine (5.400 g, 0.02 mol) and NaOH (2.400 g, 0.02 mol) was prepared. In a temperature-controlled ice bath, terephthaloyl chloride and alanine were gradually introduced into a round-bottom flask in a stepwise manner. During this process, a 4 M NaOH solution was simultaneously added to maintain the reaction pH within 8.0–9.0. The addition was completed over approximately 1 h. The mixture was subsequently stirred for an additional 3 h to ensure sufficient reaction. The system was then adjusted to an acidic environment (pH 1–2) using concentrated HCl, after which cooling induced the formation of a white precipitate. The solid was collected by filtration and rinsed with cold water. The crude product was then further washed with a 1:1 mixture of ethanol and water (30 mL), followed by filtration and drying to afford the pure compound. 1H NMR (400 MHz, DMSO-d6) δ 12.57 (s, 1H), 8.77 (d, J = 7.2 Hz, 1H), 7.93 (s, 2H), 4.39 (s, 1H), 1.37 (d, J = 7.4 Hz, 3H). Gly-PPA-Gly was synthesized by the above method, except that L-glycine replaces L-alanine. 1H NMR (400 MHz, DMSO-d6) δ 12.65 (s, 1H), 8.98 (t, J = 5.9 Hz, 1H), 7.96 (s, 2H), 3.95 (d, J = 5.8 Hz, 2H). Synthesis of 2DPAs The 2DPAs were synthesized by refluxing under near-ambient pressure in the double solvent of THF and toluene, with HOBT and HBTU serving as condensing agents. HOBT and HBTU are widely used condensation agents for peptide synthesis, which can condense carboxylic acids with amines to form amide bonds by activating carboxyl groups. Synthesis of 2DPA-PPA TAPB (28.1 mg, 0.08 mmol) and PPA (19.9 mg, 0.12 mmol) were weighed and dissolved in a mixture of 5 mL tetrahydrofuran and 25 mL toluene in a three-necked flask and then sonicated for 10 min. Then, HBTU (113.8 mg, 0.3 mmol), HOBT (45.9 mg, 0.3 mmol), and DIEA (39.7 mg, 0.3 mmol) were weighed and then added, and the solution was sonicated for 10 min. The mixture was then vigorously stirred for 48 h at 120 °C. Reaction products were centrifuged, washed with DMF and water, and then dried at 60 °C in a vacuum oven, giving the target material in a typical isolated yield of around 60%. And the procedure was found to be amenable to scale-up without notable changes in yield. 2DPA-Gly and 2DPA-Ala were synthesized by the same method as that of 2DPA-PPA, with Gly-PPA-Gly and Ala-PPA-Ala instead of PPA. Synthesis of TMS-2DPAs TMS-2DPAs were synthesized via the method of chemical exfoliation by silylation of amide bonds to obtain a high ratio of individual molecules according to the previously reported method by Strano et al.19. Typically, a highly reactive electrophile (TMSOTf) was used for post-modification (Fig. S12a–c). The consumption of amides is evidenced by the dissolution of 2DPAs in CHCl3 and the disappearance of amide peaks in FT-IR after reaction (Fig. S12d–f). Synthesis of 2DPI The 2DPI was synthesized according to the previously reported method by Zhu et al.45 with some modifications. TAPB (∼21 mg, 0.06 mmol) and PDA (∼12 mg, 0.09 mmol) were combined in a Pyrex tube containing a solvent mixture of dioxane (1.6 mL) and mesitylene (0.4 mL). The suspension was sonicated until a homogeneous solution was obtained. Subsequently, 6 M acetic acid (0.2 mL) was introduced, followed by further sonication for 10 min. The reaction mixture was then subjected to three freeze–pump–thaw cycles to remove dissolved gases, evacuated to approximately 150 mTorr, and flame-sealed under vacuum. After sealing, the tube was allowed to equilibrate to room temperature before being placed in an oven and heated at 120 °C for 72 h. Upon completion of the reaction, a yellow solid was collected, washed thoroughly with THF and ethanol, and subsequently dried under vacuum at 60 °C. The as-synthesized 2DPI was characterized by PXRD (Fig. S23) and FT-IR (Fig. S24). Photocatalyst studies Photocatalytic oxidative dehydrogenation of tetrahydroquinoline In a typical experiment, the photocatalyst (10.0 mg) was dispersed in DMF (3.0 mL) together with tetrahydroquinoline (0.1 mmol) in a 10 mL quartz reaction tube. The mixture was magnetically stirred at 600 rpm and purged with O₂ for 10 min prior to irradiation. Subsequently, the reaction was conducted under a continuous O₂ atmosphere and irradiated with visible light from a 300 W xenon lamp (Microsolar 300, Beijing Perfectlight) equipped with a 405 nm cutoff filter. The reaction temperature was maintained at approximately 298 K using an electric fan. Hydrogen peroxide photosynthesis utilizing the two-phase system The photocatalyst (5.0 mg) was added to a two-phase reaction system composed of water (1.0 mL) and benzylalcohol (2.0 mL). The system was irradiated with visible light (λ > 405 nm) for the photocatalysis. H2O2 concentration in the heterogeneous solution was determined by a potassium titanium oxalate method. Photocatalytic oxidation of thioanisole, benzylamine, and benzylalcohol In a typical experiment, the photocatalyst (5.0 mg) was dispersed in CH₃CN (3 mL) containing the substrate (0.1 mmol) in a 10 mL quartz tube. The mixture was magnetically stirred at 600 rpm and purged with O₂ for 10 min prior to irradiation. The reaction was then performed under visible-light irradiation from a 300 W xenon lamp (Microsolar 300, Beijing Perfectlight) equipped with a 405 nm cutoff filter, without continuous O₂ bubbling. The temperature was maintained at approximately 298 K using an electric fan. Quantification of produced benzaldehyde, benzyl methyl sulfoxide, and N-benzylideneaniline The amount of the desired products was analyzed by gas chromatography. And the conversion and selectivity are calculated by Eqs. (1) and (2), respectively. Note that the light (Microsolar 300; Perfectlight) intensity integrated over 405 nm was measured to be 148 mW cm−2 using a digital radiometer via the five-point method. High-Resolution Atomic Force Microscopy (HRAFM) characterization AFM images were obtained by Oxford Cypher VRS in AC mode, with OMCL-AC160TS-R3 (Olympus) Probe and Arrow-UHFAuD-10 (NanoWorld) probe, and analyzed with Asylum AR16 (Cypher’s built-in analysis module). Controlled experiment: A freshly cleaved mica was measured (Fig. S8) and showed an ultraflat surface with an RMS roughness less than 60 pm over a 1*0.8 μm area (normally 20–40 pm, calculated from Asylum AR16). The mean values of the mica and platelet surfaces are determined by Asylum AR16, using zero-order polynomial fitting (averaging of the curves). Calculation of the interlayer spacing and the mean size of crystal domains According to Bragg’s Law, interlayer spacing can be calculated as below: Where d is the interlayer spacing; θ is the Bragg angle; λ is the X-ray wavelength. Thus, the interlayer spacings of 2DPA-Ala, 2DPA-Gly, and 2DPA-PPA are 0.418 nm, 0.443 nm, and 0.438 nm, respectively. According to the Scherrer equation, the mean size of crystal domains can be determined by the line broadening at half the maximum intensity (FWHM)46: where τ is the mean size of ordered domains (lateral size of individual 2D molecules); K is the dimensionless shape factor, has a typical value of about 0.9; λ is the X-ray wavelength; β is the line broadening (FWHM), in radians; and θ is the Bragg angle. Pump-probe TA spectroscopy Femtosecond transient absorption (TA) measurements were performed using a home-built spectroscopic system.47 A commercial femtosecond amplifier (Spitfire Ace, Spectra Physics; 800 nm, 35 fs, 1 kHz) served as the laser source. The output beam was divided into two paths. One portion was directed into an optical parametric amplifier (TOPAS Prime with wavelength mixing unit, Spectra Physics) to generate tunable excitation pulses spanning 350–2600 nm, from which 400 nm was selected as the pump wavelength. A white-light probe was generated by directing the attenuated beam onto either a CaF2 plate (4 mm) or a yttrium aluminum garnet (YAG) crystal, enabling measurements in the visible and near-infrared regions. The time delay between excitation and probing was adjusted via a motorized optical delay line. For the visible spectral range, transient signals were recorded with a fiber-coupled spectrometer (AvaSpec-ULS2048CL-EVO, Avantes) synchronized to the laser system. Near-infrared transient signals were recorded using a custom-built spectrometer equipped with a 46-channel lock-in amplifier. During the measurements, the pump fluence at 400 nm was maintained at 4 μJ per pulse. The samples were ultrasonically suspended in DMF at a concentration of 5 × 10−4 M to prepare the corresponding DMF dispersions. Nanosecond TAS were obtained by a laser source of the homemade spectra setup (Spectra Physics Company, American). The output pulse was split into two beams. The first beam was used to generate 400 nm laser pulses. The second beam with weaker energy was focused on a CaF2 plate to generate a white-light continuum as the probe (Energetiq company, EQ-99). The time delay between the pump beam and the probe pulses was controlled by a motorized delay stage. In-situ light-irradiated XPS and UPS The general and in-situ light-irradiated XPS/UPS were measured on Thermo Fisher Scientific ESCALAB-250XI. Powdered samples for XPS measurement were attached to a stainless-steel plate using conductive carbon tape. Films for UPS measurement were prepared by spraying the corresponding ethanol dispersions on the pretreated indium tin oxide (ITO) and dried in a vacuum oven at 60 °C for 4 h. XPS measurements were carried out using monochromatic, micro-focused Al Kα radiation (hν = 1486.6 eV) as the excitation source.48 The emitted photoelectrons were analyzed with a 180° double-focusing hemispherical analyzer equipped with a two-dimensional detector. Charge compensation during the measurements was achieved using a flood gun. Binding energies were calibrated with respect to the C–C component of adventitious carbon in the C 1s spectrum at 284.8 eV to correct for residual surface charging.49 In-situ light-irradiated XPS spectra were measured under a 440 nm laser (LSR-PS-II LASEVER INC.). For UPS measurements, the instrument uses monochromatic and micro-focused He I (hν = 21.22 eV) radiation to eject photoelectrons, which are then analyzed using a 180° double-focusing hemispherical analyzer with a 6-channel electron detector. The Ef positions and analyzer work function were calibrated by measuring the Fermi level (Ef) step-edge of a clean Ar+ Au film. Furthermore, Spectra were recorded in the 4 × 10−9 mbar ultra-high vacuum system with a pass energy of 2 eV and a negative bias of 10 V to observe the secondary electron cutoff. The energetic onsets (HOMO) were determined by an approximate linear extrapolation at the edge of UPS spectra and verified by an approximate linear extrapolation at the edge of XPS valence spectra. All Data was analyzed using the Thermo Avantage package. EPR experiments Electron paramagnetic resonance (EPR) spectra were conducted on a JEOL FA-200 electron paramagnetic resonance spectrometer at room temperature. All measurements were conducted in high-purity quartz tubes (Suprasil, Wilmad; ≤0.5 OD). The g-values were calibrated against a Mn²⁺/MgO standard (JEOL). Light-induced EPR experiments were carried out using a 300 W xenon lamp (PerfectLight) with visible-light irradiation (λ > 420 nm).50 Temperature modulated EPR measurements and three-pulse electron spin echo envelope modulation (3P-ESEEM) were carried out using a CIQTEK EPR100 spectrometer equipped with a liquid helium circulation system for helium gas flow sample cooling, which is in a temperature range between 298 K and 15 K, and at 6 K, respectively. Data processing and analysis were performed using EPR Data Processing (Version 4.1.6, CIQTEK). General characterization The morphologies of the samples were investigated by TEM on Hitachi HT 7700 at an accelerating voltage of 100 kV, and field-emission SEM on a Hitachi SU-8010 at an accelerating voltage of 10 kV. The HRTEM images were collected on a high-resolution transmission electron microscope (JEOL JEM-2100F) at an accelerating voltage of 200 kV. The AFM and KPFM images were obtained by Oxford Cypher VRS, Oxford, with OMCL-AC160TS-R3 (Olympus) Probe and Arrow-UHFAuD-10 (NanoWorld) probe, and analyzed with Cypher. The substrate for KPFM is highly ordered pyrolytic graphite (HOPG) brought from XFNANO. The liquid 1H NMR spectra were recorded on a 400 MHz spectrometer, AVANCE Ⅲ HD 400, Bruker, at 298 K. FT-IR measurements were performed by using a Bruker V70 spectrometer with a reflection diamond ATR module and a resolution of 4 cm−1. The 13C ssNMR spectra were measured on a JEOL JNM-ECZ600R spectrometer at 298 K with the CP MAS-TOSS method. TGA was operated on a Mettler Toledo TGA/DSC1 Star System analyzer under N2 flow at a heating rate of 10 °C min−1 within the temperature range of 30–1000 °C. N2 and CO2 sorption measurements were performed on a Micromeritics ASAP 2460 system and Quantachrome Autosorb-iQ. PXRD patterns were recorded on a Rigaku Smartlab X-ray diffractometer using a Cu Kα1 (λ = 1.54059 Å) at 45 kV and 200 mA, with a scan step of 0.02° and a scan speed of 10° per minute. The small-angle diffraction data were collected on a Rigaku Smartlab X-ray diffractometer using a Cu Kα1 (λ = 1.54059 Å) at 40 kV and 150 mA, with a scan step of 0.01° and a speed of 0.3 s per step. GIWAXS patterns were acquired on the 1W1A beamline at the Beijing Synchrotron Radiation Facility (BSRF). The UV–Vis–NIR DRS was obtained on Agilent Cary 5000, using BaSO4 as the reference. Time-resolved photoluminescence decay spectra were collected on an Edinburgh FLS 1000 fluorescence spectrometer with an excitation wavelength of 442.6 nm and an emission wavelength of 650 nm. The TPV measurements were conducted on CEL-TPV2000 with an excitation wavelength of 355 nm. Electrochemical measurements were performed on an electrochemical system (CHI760E, Shanghai Chenhua Instrument Company). A standard three-electrode cell was employed with a working electrode, a platinum wire as the counter electrode, and a saturated calomel electrode (SCE) as the reference electrode. 0.1 mol L−1 Na2SO4 (aq) was taken as the electrolyte solution. The working electrodes were prepared as follows. 5 mg photocatalyst was dispersed in 1 mL deionized water under ultrasonication to form a 5 mg mL−1 solution. 0.5 ml of the resultant solution was then sprayed on the pretreated indium tin oxide (ITO) and dried for 24 h at room temperature. DFT calculations The fragment geometry optimizations for four materials were performed using the LAMMPS package51 with COMPASS (condensed-phase optimized molecular potentials for atomistic simulation studies) class II force field52. The single-point energy calculations were conducted in the Gaussian 09 D.01 software package53 by using the above-optimized conformation at B3LYP/def2TZVPP and analyzed by VMD v1.9.354. Detailed computational data are available in Supplementary Data 1. The calculation of the charge recombination rate was performed in the Amsterdam Density Functional (ADF) package55. The geometry-optimized conformations of four materials (2DPA-PPA, 2DPA-Ala, 2DPA-Gly, 2DPI, Fig. 6a–d) were the input geometries of ADF. The effective or generalized charge transfer integral (V) for charge recombination was performed using the method in Eq. (5), where the V of each material can be calculated from the spatial overlap (S12), charge transfer integral (J12), and site energies (ε1, ε2). All calculations were carried out within the density functional theory framework using the VWN local density approximation combined with PW91 gradient corrections and a TZ2P basis set, as available in the ADF package. Considering all the materials are n-type semiconductors, the λ was evaluated using the adiabatic potential energy surfaces method56, as shown in Eq. 6. The energy terms were defined based on both optimized and non-equilibrium geometries. Specifically, the energies of the neutral and anionic species at their own relaxed structures are denoted as E0 and E−, respectively. In contrast, E0* and E−* refer to the corresponding states evaluated at geometries associated with the opposite charge configuration. Geometry optimizations for the gas-phase ground-state molecules in both neutral and anionic forms were performed using the B3LYP-D3 functional with the TZ2P basis set, as implemented in the ADF software package.57 Under the hopping model at room temperature, the hopping rate (W) of charge recombination can be given by the Marcus equation58–60 in Eq. 7, where ћ is the reduced Planck constant, kB is the Boltzmann constant, and T is the temperature. It is worth noting that our computational model represents a simplified version of the complex catalytic system under operando conditions, and does not fully capture the effects of the solvent, potential, or surface dynamics. Supplementary information Supplementary Information Description of Additional Supplementary File Supplementary Data 1 Transparent Peer Review file Source data Source Data Acknowledgements This work was supported by This work was supported by Yunnan Provincial Science and Technology Project at Southwest United Graduate School (202302AO370017, C.C), National Key R&D Program of China (2023YFB4005100, C.C), National Natural Science Foundation of China (21925202, U22B2071, C.C), and Tsinghua Xuetang Talents Program. We thank Prof. Wenqing Yao of the Analysis Center, Tsinghua University, for her help on XPS/in-situ-XPS/UPS. We thank Prof. Xingwei Guo for the analysis of EPR and ESSEM. We thank Caiyan Xiao and Zaixing She of Beijing China Education Au-light Technology Co., Ltd. for their help with time-resolved SPV measurement. We thank Dr. Chao Ma of the Analysis Center, Tsinghua University, for his help on HRTEM. We would like to thank Mingxin Zheng from Shiyanjia Lab (www.shiyanjia.com) for the N2 and CO2 sorption measurements. We thank the 1W1A station for GIWAXS measurements at the Beijing Synchrotron Radiation Facility. Author contributions C.H., W.L, B.W., and C.C. conceived and designed the experiments. C.H. and W.L. synthesized the experimental samples and carried out most of the characterization, as well as the photocatalytic performance evaluation. W.L. carried out the DFT calculations. C.H., W.L., X.C., Y.Z., and C.C. cowrote the paper. X.C., J.M., X.T., Y.C., and H.Y. discussed and provided advice on the results. All the authors contributed to the interpretation of the data and preparation of the manuscript. Peer review Peer review information Nature Communications thanks Yuwen Zeng and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available. Data availability The source data generated in this study are provided in the Supplementary Information/ Source Data file. Supplementary Information file and Source Data file are provided with this paper. Source data are provided with this paper. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. These authors contributed equally: Chang He, Weixu Liu. Contributor Information Bo Wang, Email: bowang@bit.edu.cn. Chen Chen, Email: cchen@mail.tsinghua.edu.cn. Supplementary information The online version contains supplementary material available at 10.1038/s41467-026-71621-7. References Associated Data Supplementary Materials Supplementary Information Description of Additional Supplementary File Supplementary Data 1 Transparent Peer Review file Source Data Data Availability Statement The source data generated in this study are provided in the Supplementary Information/ Source Data file. Supplementary Information file and Source Data file are provided with this paper. Source data are provided with this paper.