Construction of organic heterojunctions as metal-free photocatalysts for enhancing water splitting and phenol degradation by regulating charge flow

Abstract

Metal-free photocatalysts derived from earth-abundant elements have drawn significant attention owing to their ample supply for potential large-scale applications. However, it is still challenging to achieve highly efficient photocatalytic performance owing to their sluggish charge separation and lack of active catalytic sites. Herein, we designed and constructed a series of covalently bonded organic semiconductors to enhance water splitting and phenol degradation. Experimental and theoretical results revealed that the charge transfer mechanism transformed from type II in the physical mixture to a Z-scheme in the covalently bonded composite, resulting from the interfacial electric field formed at the interface between a β-ketoenamine-linked covalent organic framework (TP-COF) and a urea linked perylene diimide (PDI) semiconductor (UP) linked by amide bonds. The Z-scheme charge transfer route not only improved charge separation but also preserved the high redox ability of both semiconductors. Moreover, more active catalytic sites were created owing to the net charge transfer from the UP to TP-COFs with the amide bonds, contributing to improved photocatalytic performance. As a result, high HER, OER and phenol degradation rates of 613.30 μmol g⁻¹ h⁻¹, 1169.36 μmol g⁻¹ h⁻¹, and 0.81 h⁻¹ were achieved, respectively. This work provides a new strategy to develop metal-free photocatalysts with simultaneously improved charge separation efficiency and catalytic site activity.

---

New concepts

In this work, we introduce a strategy to develop efficient metal-free photocatalysts with simultaneously improved charge separation, high redox driving force and active sites with enhanced activity. Experimental results and theoretical calculations reveal that the charge transfer mechanism transforms from type II in the physical mixture to a Z-scheme in the covalently bonded composite, resulting from the interfacial electric field formed at the interface between a β-ketoenamine-linked COF (TP-COF) and a urea linked perylene diimide (PDI) semiconductor (UP) linked by amine bonds. The Z-scheme charge transfer route not only improves charge separation but also preserves the high redox ability of both semiconductors. Moreover, more active catalytic sites are created owing to the net charge transfer from the UP to TP-COFs with the amine bonds, contributing to improved photocatalytic performance. As a result, the HER, OER and phenol degradation rates of the covalently bonded sample increase by 20, 14, and 3 times than those of the physical mixture. This work provides a new strategy to develop metal-free photocatalysts with simultaneously improved charge separation efficiency and catalytic site activity.

Introduction

Photocatalysis, which converts solar energy into chemical energy, has long been considered a sustainable pathway to solve the energy and environmental problems faced by humankind. Recently, metal-free photocatalysts derived from natural abundant elements have been paid particular attention as an alternative class of photocatalysts owing to their low cost. Covalent organic frameworks (COFs), as an emerging class of porous organic material composed of lightweight elements, have been employed in various areas such as gas storage,¹ catalysis,^2,3 and sensing.^4,5 Particularly, two dimensional (2D) COFs obtained through the Schiff base reaction usually exhibit broad light harvesting ability in the visible-light region, which is attributed to the absorbance of characteristic groups and large conjugation systems.⁶ However, the sluggish kinetics of these metal-free photocatalysts due to lack of active sites severely limit their catalytic efficiencies. Therefore, noble metal-based (Pt, Pd, etc.) and other low natural-abundance transition metal-based (Co, Ni, etc.) co-catalysts are usually required to accelerate surface catalytic reactions, which leads to increased cost. Meanwhile, severe recombination of photogenerated electron–hole pairs limits photocatalytic efficiency. In this regard, construction of semiconductor heterostructures with staggered band positions provides a route to improve charge separation^7,8 by forming type II or Z-scheme heterojunctions at the interface. Photogenerated charge carriers are directed to migrate to different semiconductors in both type II or Z-scheme heterojunctions, which is expected to enhance charge separation. However, electrons and holes are transferred to semiconductors with lower reductive and oxidative abilities in type II heterojunctions. In contrast, the high redox ability of semiconductors can be preserved in Z-scheme heterojunctions, which is preferable for many photocatalytic reactions (especially those without selectivity issues), e.g. water splitting and pollutant treatment, by providing a larger driving force.^7,9 Therefore, constructing a Z-scheme provides a solution to enhance charge separation and preserve redox ability simultaneously. Besides, electrons and holes can only participate in redox reactions when they migrate to the corresponding reaction sites. In this sense, it is also critical to direct the photogenerated charge carriers to transport to the corresponding reaction sites. Therefore, it can simultaneously enhance charge separation and preserve high redox ability by constructing a Z-scheme junction with directed charge migration.

A direct Z-scheme photocatalyst built by intimate contact of two semiconductors does not require electron or hole mediators, drawing increasing attention recently.^10,11 Until now, it is still challenging to intentionally construct type II heterojunction and/or direct Z-scheme catalysts by the contact of two semiconductors. Moreover, the reductive and oxidative semiconductors are reversed in type II and in direct Z-scheme heterojunction due to their opposite charge flow direction. Therefore, it is critical to control the charge flow direction in order to consciously design reductive/oxidative sites on the semiconductors with accumulated electrons/holes. Instead of the simple contact of two semiconductors, covalent bonding provides a liable intermolecular charge transfer pathway, which has been widely applied in organic materials with donor and acceptor pairs such as donor–acceptor covalent organic frameworks (COFs).¹² Lan's Group has covalently bonded TpPa-1-COF with amino-functionalized BiFeO₃ nanosheets to construct a BiFeO₃@TpPa-1-COF core–shell hybrid material.¹³ For the first time, they combined a COF and piezoelectric material with covalent bonds to form a highly efficient Z-scheme heterostructure piezo-photocatalyst for overall water splitting.¹³

Inspired by this, we have constructed a photocatalytic system by covalently bonding a β-ketoenamine-linked COF with a urea-linked perylene diimide (PDI) semiconductor (UP), in which an interfacial electric field (IEF) forms at the interface between the COF and the UP. The charge transfer follows a Z-scheme mechanism due to the formation of the IEF, which boosts the charge separation efficiency and reserves the high redox potential of both the COF and UP. Moreover, a net charge transfer from UP to the COF occurs after covalent bonding, forming more active sites for hydrogen evolution and oxygen evolution on the COF and UP, respectively. In contrast, a type II heterostructures are formed without covalent bonds, in which the electrons and holes migrate to the UP and COF with lower redox potential and activity. Therefore, not only are the merits of a Z-scheme photocatalytic system, including efficient charge separation and high redox potential, inherited in this covalently bonded hybrid material, but also the electrons and holes are directed to the corresponding reduction and oxidation sites with higher activity. As an internal result, the photocatalytic hydrogen and oxygen evolution rate reaches 613.30 and 1169.36 μmol g⁻¹ h⁻¹ after optimizing the ratio between the COF and UP, which is 20 and 14 times that of the physical mixture. Additionally, this hybrid material also demonstrates excellent wastewater treatment ability with a degradation rate of 95% for phenol, which is 3 times that of the physical mixture. This work provides a way to deliberately construct Z-scheme photocatalytic systems and direct the charge carriers to transfer to their corresponding reaction centers with enhanced activity. By choosing proper semiconductors, this strategy can be simply applied to other photocatalytic reactions and provides a universal way to construct photocatalysts with enhanced charge separation and active sites simultaneously.

Results and discussions

Synthesis and characterization

The schematic diagram in Fig. 1A illustrates the construction of the UP@TP-COF heterojunctions. Briefly, Urea-PDI polymer (UP) with the chemical structure shown in Fig. 1A was first synthesized via a simple wet-chemical method. Subsequently, the covalently bonded UP and TP-COF hybrids have been obtained by synthesizing TP-COF in the dispersion of the as-synthesized UP and the TP-COF precursors. By varying the loading mass ratio between UP and TP-COF precursors, a series of composites, namely UP@TP-COF-x (x = 0.1, 0.2, 0.3, and 0.4, representing the mass ratio between UP and TP-COF precursors), have been obtained. A clear morphology change can be identified after TP-COF is loaded on UP, as revealed by scanning electron microscopy (SEM). The as-synthesized UP demonstrates smooth micro-rod morphology with an average size of 5 × 4 μm (Fig. 1C), while the TP-COF displays a fluffy morphology (Fig. 1B). In comparison, micro-rod-shaped cores covered by fluffy sheets are present in UP@TP-COF-0.2 (Fig. 1D), indicating the TP-COF is loaded outside the rods of UP. Similar morphologies of other UP@TP-COF-x samples are observed, as displayed in Fig. S1 (ESI†). The successful synthesis of UP and TP-COF composites is further supported by powder X-ray diffraction (PXRD), as shown in Fig. 1E. An intense peak at 2.8° corresponding to the (100) facet of TP-COF can be identified for all UP@TP-COF-x samples,¹⁴ while the typical diffraction peaks at 10.1, 12.0, 25.0, and 27.2° related to UP are also clearly recognized in the PXRD of these samples,⁹ which suggests the formation of both TP-COF and UP in the composites. Notably, the relative peak intensity corresponding to UP rises progressively as the mass ratio of UP used during synthesis increases, indicating the gradual increase of the UP content. The thermal stability of the individual components and the composite has also been evaluated via thermogravimetric analysis (TGA), as displayed in Fig. S2 (ESI†). As shown, the UP starts to decompose at ∼490 °C, while the onset temperature of decomposition for TP-COF is around 400 °C. The onset temperature of decomposition for the composite UP@TP-COF-0.2 is very close to that for TP-COF, and the TGA curve of UP@TP-COF-0.2 shows the shape of a linear combination of the TGA curves of UP and TP-COF. These results indicate a slight decrease in the thermal stability of each component after the composite is formed.

Fig. 1 (A) Schematic illustrating the synthesis procedure for UP@TP-COF-x and its chemical structure at the interface of TP-COF and UP. (B)–(D) SEM images of TP-COF, UP, and UP@TP-COF-0.2. The scale bar = 2 μm. (E) PXRD patterns of individual UP, TP-COF, and UP@TP-COF-x samples. (F) FT-IR spectra of UP, TP-COF, and UP@TP-COF-0.2. (G) Enlarged FT-IP spectra of (F) in the range of 1200–1900 cm⁻¹. (H) C 1s and (I) N 1s XPS spectra of UP, TP-COFs, UP@TP-COF-0.2 and physical mixture of UP and TP-COFs.

The success of linking TP-COF and UP via amide bonds has been confirmed via Fourier transform infrared reflection (FT-IR) spectra by comparing the spectra of TP-COF and UP with UP@TP-COF-0.2 as an example of the composites. In Fig. 1F and G, peaks at 1752 and 1688 cm⁻¹ are clearly recognized in both UP and UP@TP-COF-0.2, which corresponds to the stretching vibration of C=O in the urea linker and the PDA motif after polymerization of urea and PDA, respectively.⁹ In the meantime, the characteristic peaks at 1618 cm⁻¹ and 1578 cm⁻¹ related to the C=O originating from the condensation between aldehydes and amines and the C=C due to isomorphism from enols to ketones, respectively, during the formation of TP-COP can be identified in both TP-COF and UP-TP. These results indicate that the UP structure is well maintained during the synthesis of TP-COF. More importantly, an additional peak at 1600 cm⁻¹ originating from the deformation/bending vibration of the –CONH– groups at the junction of the TP-COF skeleton and UP emerges in UP@TP-COF-0.2, indicating that the TP-COF and UP are covalently bonded via amide bonds.¹⁵ Solid state ¹³C nuclear magnetic resonance (NMR) spectra have also been obtained, which further supports the formation of amide bonds between UP and TP-COF in the covalently bonded sample. In comparison to the solid state ¹³C NMR spectra of individual TP and UP, a new peak is clearly identified with a chemical shift of 177.5 ppm in the UP@TP-COF-0.2 spectrum corresponding to the signal of the -CONH-, as shown in Fig. S3 (ESI†). The chemical bonding of TP-COF and UP through amide bonds is further confirmed via the narrow scan C 1s and N 1s XPS spectra, as shown in Fig. 1H and I, respectively. In the C 1s spectrum of UP, three peaks with binding energies of 284.7 eV, 285.4 eV, and 288.0 eV are identified, which is related to perylene (position 1), carbonyl carbon in urea (position 2) and in PDI (position 3), respectively.⁹ Meanwhile, three peaks corresponding to the carbon of position 1′, 2′ and 3′ in TP-COF are resolved in the C 1s spectrum of TP at 284.7 eV, 286.3, and 287.2 eV, respectively. As expected, the C 1s spectrum of the physical mixture shows a pattern of simple addition of the C 1s spectra of UP and TP-COF, in which the peak positions remain nearly unchanged as in the corresponding individual spectrum. Similarly, the N 1s spectrum of the physical mixture displays an overlapped pattern of individual UP and TP-COF, in which the two peaks at 400.4 and 400.0 eV are corresponding to the N in UP (at 400.4 eV) and TP-COF (at 400.0 eV), respectively. In contrast, peak shifts are clearly present in the spectra of the covalently bonded sample. Taking UP@TP-COF-0.2 as an example, the binding energies of the carbonyl carbons in urea and in PDI of UP increase, while the peaks corresponding to position 2′ in TP-COF are shifted to a lower binding energy. Consistently, the peak corresponding to the nitrogen atoms in UP shifts to a larger binding energy, while the peak corresponding to the nitrogen atoms in TP-COF shifts negatively. These results indicate that the electron density of the UP motif decreases while that of the TP-COF motif increases, i.e., net electron transfers from UP to TP-COF after they are covalently bonded. Additionally, a new peak at 286.6 eV corresponding to amide bonds emerges in the C 1s spectrum of UP@TP-COF-0.2, further supporting that UP and TP-COF are covalently bonded via amide bonds. Since no additional signal can be detected in the physical mixture, physical contact between UP and TP-COF in the UP@TP-COF-x samples cannot be excluded. Therefore, there may be physical contact besides amide bonds at the interface between UP and TP-COF in UP@TP-COF-x.

Optical properties and band structures

The optical properties and band structures of the two semiconductors composing UP@TP-COF-x are further investigated using ultraviolet–visible diffuse reflectance spectroscopy (UV-vis DRS), Mott–Schottky (MS), and valence band X-ray photoelectron spectroscopy (VB-XPS) techniques. As shown in Fig. 2A and B, the adsorption edges of TP-COF and UP are measured to be ∼561 and 750 nm based on UV-vis DRS, corresponding to the band gap of 2.23 and 1.89 eV, respectively. The valence band maximum (VBM) of TP-COF and UP is measured to be 1.00 and 1.44 V vs. normal hydrogen electrode (NHE) based on VB-XPS shown in Fig. 2C, respectively. Based on these data, the conductive band minimum of TP-COF and UP is calculated to be located at −1.23 and −0.45 V vs. NHE. To further confirm their band positions, Mott–Schottky (MS) spectra were also obtained. The positive slopes of the tangent of TP-COF and UP in the Mott–Schottky spectra illustrate that both of them are n-type semiconductors,^7,8,15 which is consistent with previous reports.^9,16 The flat band potentials of TP-COF and UP are estimated to be −1.05 and −0.23 V (vs. NHE), respectively, based on the intercepts with the x-axis. In general, the value of flat potentials is approximately equal to that of the Fermi level, which is about 0.20 V more positive than that of the CBM of an n-type semiconductor.^7,8,15 Therefore, the CBM of TP-COF and UP is estimated to be −1.25 and −0.43 V vs. NHE, respectively, is in excellent agreement with that calculated based on UV-vis DRS and VB-XPS. Based on the above experimental results, the band structures of TP-COF and UP are illustrated in Fig. 2F, which clearly shows that these n-type semiconductors display staggered band structures known as the prerequisite for the construction of a direct Z-scheme heterojunction.¹¹

Fig. 2 (A) and (B) MS spectra of UP and TP-COF. (C)–(E) UV-vis DRS, plots of the transformed Kubelka–Munk function vs. photon energy, VB-XPS of UP and TP-COF. (F) Band structure alignments of UP and TP-COFs together with the potential of key half-reactions at pH = 0.

Interfacial electric field directed charge migration

It is expected that charge separation can be enhanced by forming heterojunctions of semiconductors with staggered band structures by forming either Z-scheme or type II heterojunctions.^7,9 To evaluate the charge separation efficiency in these heterostructures, photoluminescence spectroscopy and linear sweep voltammetry (LSV) were conducted under chopped light illumination. As shown in Fig. S5A (ESI†), an emission peak at ∼630 nm corresponding to the charge recombination of the electrons on CB with holes on VB in TP-COF can be observed in the photoluminescence (PL) spectra of TP-COF, covalently bonded and the physically mixed composites. Compared with TP-COF, the PL intensity of covalently bonded and physically mixed composites is only 45% and 77% of that of TP-COF, indicating dramatically depressed charge recombination in the composite materials.¹⁷ Consistently, linear sweep voltammetry (LSV) measured under chopped light illumination further confirms the promoted charge separation abilities of the composites (Fig. S5B and C, ESI†). With almost identical onset potential, the photocurrent of covalently bonded and physically mixed composites is 167 and 140 μA, respectively, which is 1.72 and 1.44 times that of TP-COF under a bias of 1.74 V vs. Ag/AgCl, respectively. Meanwhile, the photocurrent under a bias of −1.56 V vs. Ag/AgCl, the photocurrent of covalently bonded and physically mixed composites is 2.23 and 1.56 times that of UP, respectively. The enlarged photocurrents in the composites further confirm their enhanced charge separation.^7,17–19

In order to further understand the modulation in their electronic structures, theoretical calculations have been performed. As indicated by the XPS results (Fig. 2), a net electron transfer from UP to TP-COF occurs after they are covalently bonded. Consistently, a 0.29 e net charge transfer from UP to TP-COF is suggested by calculating the electron–hole distribution at the interface (Fig. 3A). Furthermore, at the interface between UP and TP-COF, the electron potential of the UP motif side is obviously lower than that of the TP-COF motif side, suggesting an IEF directing from TP-COF to UP is formed (Fig. 3B). Therefore, an IEF-directed charge migration is expected, which is further proven by in situ XPS and Kelvin probe force microscopy (KPFM). As shown in Fig. 3C and D, the binding energy of carbon and nitrogen atoms in UP shifts positively under illumination, indicating hole accumulation on the UP motif. In contrast, the binding energy of carbon and nitrogen atoms in TP-COF decreases after exposure to light, implying electron gathering on the TP-COF motif. These results suggest that the photogenerated electrons and holes are directed to migrate to TP-COF and UP in the covalently bonded composite, respectively, which is further supported by in situ KPFM measurements. Fig. 3E and F display the topographic images of UP@TP-COF-0.2 in the dark and under light, respectively. As shown, negligible morphological change can be identified after the sample is exposed to light. In contrast, the surface potential of the sample clearly drops from 60 mV to 35 mV after light on, indicating positive charge accumulation on the surface. Since TP-COF covers UP in UP@TP-COF-0.2, this result further confirms that positively charged holes are gathered on TP-COF under light after being covalently bonded with UP. Interestingly, the charge transfer follows an opposite direction in the physical mixture compared with the covalently bonded composite. As shown in Fig. 3E and F, the binding energies of carbon and nitrogen atoms in TP-COF shift positively under illumination, while those binding energies corresponding to UP shift negatively under light, indicating that holes and electrons are gathered on TP-COF and UP, respectively. Consistently, the surface potential measured via KPFM increases instead of decreasing after light on (Fig. 3O to V), implying electron accumulation rather than hole accumulation on the surface of the physical mixture. Considering the surface of the physical mixture is also covered by TP-COF (Fig. S4, ESI†), this result, together with the in situ XPS measurements, supports that the charge migration follows a type II mechanism, which is opposite to that in the covalently bonded composites.

Fig. 3 (A) Bader charge distribution and (B) surface potential distribution calculated at the interface between UP and TP-COF with amide bonds. (C) and (D) In situ XPS of C 1s and N 1s spectra of UP@TP-COF-0.2. (E) and (F) In situ XPS of C 1s and N 1s spectra of the physical mixture. (G) and (H) Topographic images of UP@TP-COF-0.2 taken in the dark and under light. (I) and (J) Potential images of UP@TP-COF-0.2 taken simultaneously with (G) and (H). (K) and (L) 3D images corresponding to (I) and (J). (M) and (N) Surface potential cursor profiles as indicated by the lines in (I) and (J). (O) and (P) Topographic images of UP@TP-COF-0.2 taken in the dark and under light. (Q) and (R) Potential images of UP@TP-COF-0.2 taken simultaneously with (O) and (P). (S) and (T) 3D images corresponding to (Q) and (R). (U) and (V) are the surface potential cursor profiles as indicated by the lines in (Q) and (R).

Based on the above analysis, it can be proposed that the charge migration follows a Z-scheme mechanism for the covalently bonded composite, while it follows a type II way in the physical mixture, as shown in Fig. S6A–C (ESI†). For the covalently bonded composite, a potential barrier forms at the interface due to the band bending as the IEF directs from TP-COF to UP. Therefore, the migration of the photogenerated electrons on TP-COF and holes on UP across the interface is blocked. Meanwhile, the photogenerated holes on TP-COF and electrons on UP are recombined at the interface, leaving electrons and holes accumulated on TP-COF and UP, respectively.⁷ In order to further understand the change in surface potential, the electrostatic potential and charge distribution under light have been calculated (Fig. S7, ESI†). Compared to that in the dark, the region of the interfacial electric field (IEF) directing from TP to UP enlarges under light irradiation, indicating an increase in the IEF in the excited state. The IEF provides a driving force to recombine the electrons generated on UP with holes generated on TP-COF at the interface. Meanwhile, the holes generated on UP and electrons generated on TP-COF are forced to separate and gather on UP and TP-COF, respectively. Therefore, the enlarged IEF enhances the separation of photogenerated charge carriers and the accumulation of electrons and holes on TP-COF and UP, respectively. In contrast, the photogenerated electrons are accumulated on UP with lower CBM, while the holes are gathered on TP-COF with higher VBM in the physical mixture, resulting from a type II charge migration mechanism. Notably, the Z-scheme charge transfer route not only improves charge separation but also preserves the high reductive and oxidative abilities of TP-COF and UP, respectively, both of which are beneficial for photocatalytic redox reactions.⁷ However, the redox ability in the type II heterojunction is sacrificed for improving charge separation.⁹ Moreover, TP-COF and UP are expected to be more active towards HER and OER, respectively, than in a reverse fashion.⁹ Thus, the Z-scheme charge migration leads the photogenerated electrons and holes gathering to their corresponding desired reaction centers. In contrast, the type II pathway results in charge carrier transfer to the unwanted reaction centers, which is not favorable and may lead to lower photocatalytic rates.

Photocatalytic water splitting and phenol degradation

The photocatalytic HER and OER were tested under irradiation of a 300 W Xe lamp (≥420 nm) without any co-catalyst. After optimizing the test conditions (Fig. S8A and B, ESI†), the photocatalytic water splitting performance of the composites has been compared with TP-COF and UP. As shown in Fig. 4A, the hydrogen evolution rate of TP-COF is only 24.46 μmol g⁻¹ h⁻¹, while little hydrogen can be generated using UP. In comparison, all the UP@TP-COF-x samples demonstrate improved activities. The hydrogen evolution rate increases first as the mass ratio of UP rises and then drops and reaches the maximum of 613.30 μmol g⁻¹ h⁻¹ with UP@TP-COF-0.2, which is approximately 25 times higher than that of TP-COF alone. Similarly, the maximum oxygen evolution rate is obtained with UP@TP-COF-0.2 (Fig. 4B), with a rate of 1169.36 μmol g⁻¹ h⁻¹, which is 3 times higher than that with pristine UP (391.86 μmol g⁻¹ h⁻¹). As expected, excellent apparent quantum efficiencies (AQEs) have also been achieved with UP@TP-COF-0.2. Based on the hydrogen evolution reaction, the AQEs are measured to be 8.97%, 5.31%, 2.72%, 5.1%, and 4.78% at 420, 520, 660, 390, and 450 nm, respectively (Fig. 4C). UP@TP-COF-0.2 also displays excellent cyclicity and stability. The hydrogen productivity remains stable in 9 continuous 4-hour cycles. Meanwhile, negligible changes in the PXRD patterns and the morphology in scanning electron microscopy (SEM) images can be identified before and after photocatalysis (Fig. S9, ESI†). Compared with the recently reported metal-free photocatalysts, UP@TP-COF-0.2 demonstrates good photocatalytic hydrogen evolution and oxygen evolution efficiencies (Tables S1 and S2, ESI†).^20–39 Notably, the amide bonds between TP-COF and UP play a critical role in photocatalysis. The physical mixture of UP and TP-COF with the same ratio of UP@TP-COF-0.2 only demonstrates HER and OER rates of 38.11 and 421.34 μmol g⁻¹ h⁻¹, which are just 6.2% and 36.0%, respectively, of that achieved with UP@TP-COF-0.2. More interestingly, the HER and OER rates of the physical mixture are between those of TP-COF and UP. As shown by the in-situ XPS and KPFM, the electrons and holes transport to UP and TP-COF, respectively, leading to HER and OER occurring at the lower reductive UP and lower oxidative TP-COF, respectively. Therefore, the photocatalytic performance is poor, although enhancement in charge separation has been achieved in the physical mixture (Fig. S5, ESI†). These results highlight that the charge migration pathway plays a critical role in photocatalysis.

Fig. 4 (A) and (B) Photocatalytic HER and OER performance of all samples. (C) Wavelength-dependent AQE of H₂ generation over UP@TP-COF-0.2 and the corresponding DRS spectrum. (D) Photocatalytic H₂ generation cycling measurements of UP@TP-COF-0.2. All the photocatalytic reactions were taken under a 300 W Xe lamp with a Beijing Perfectlight Labsolar-6A all glass automatic on-line trace gas analysis system.

Besides clean energy generation, pollutant degradation is also an important application of photocatalysis. Therefore, we further investigated the photocatalytic activity of all the samples for pollutant degradation using 10 mg L⁻¹ phenol as an environmental pollutant model. As illustrated in Fig. 5, the pristine TP-COF and UP exhibited low degradation efficiencies, yielding phenol degradation rates of 37% and 12%, respectively, over a 180-minute period. In contrast, UP@TP-COF-0.2 demonstrates superior phenol degradation performance, achieving a phenol removal rate of 95% in 180 min, with a mineralization rate of 60%. Notably, only 36% phenol is degraded using the physical mixture, even lower than that of pure TP-COF, although the charge separation efficiency is boosted in the physical mixture (Fig. S5, ESI†). These results further indicate the importance of charge flow direction for photocatalytic performance. The effect of pH on phenol degradation is further explored. As shown in Fig. 5C and D, the degradation rate of phenol increases when the pH value is between 3 and 11. Phenol exists predominantly in the cationic form at pH < 9.9 since the acid dissociation constant (pK_a) of phenol is 9.9, which is beneficial for phenol adsorption on the electron-accumulated surface under light. In contrast, when pH > 9.9, phenol presents mainly in the anionic form, leading to a lower degradation rate due to weakened adsorption. However, the formation of hydroxyl radicals as a common oxidative active species is easier at higher pH, which is preferable for pollutant degradation. As a balanced result, the reaction rate is the highest at pH = 11. Considering the presence of various ions in real wastewater, we further investigated the effects of chloride, sulfate, nitrate, and bicarbonate ions on the photocatalytic degradation of phenol (Fig. S10, ESI†). It is shown that the degradation efficiency decreases by 21%, 27%, 17%, and 47% after adding chloride, sulfate, nitrate, and bicarbonate ions, respectively. Likely due to the reaction with ˙OH to generate carbonate ions and light adsorption, the degradation drops the most with bicarbonate ions.^40,41 Chloride and sulfate ions also significantly reduced the degradation efficiency by reacting with hydroxyl radicals and generating positive holes.⁴² Sulfate ions display the least impact, likely due to the weak adsorption on the catalyst surface.⁴³

Fig. 5 (A) Phenol degradation and (B) the rate constant (k) obtained by the fitted with first-order kinetics equation using the data in (A) of TP-COF, UP, UP@TP-COF-x, and the physical mixture. (C) Phenol degradation measured at different pH, and (D) the rate constant (k) obtained by fitting the first-order kinetics equation using the data in (C) of UP@TP-COF-0.2. (E) Active species trapping experiments over UP@TP-COF-0.2. (F) The EPR spectra of DMPO-˙O₂⁻ measured with TP-COF, UP, UP@TP-COF-0.2, and the physical mixture.

Further investigation of the active oxidation species during the degradation process has been conducted through scavenger experiments (Fig. 5E). Quinone (BQ), tert-butanol (tBA), and potassium iodide (KI) were used as scavengers to quench ˙O₂⁻, ˙OH radicals, and holes, respectively.^7,44 Compared to the blank, the presence of BQ leads to a sharp decrease of 56% in the degradation efficiency, indicating that ˙O₂⁻ plays the key role in phenol degradation.⁴⁵ The degradation rate decreases gently by 23% with tBA, implying that ˙OH radicals play an auxiliary role.⁴⁶ In contrast, the degradation efficiency only drops by 2.3% in the presence of EDTA-2Na, suggesting that holes may not be the main active oxidation species for phenol degradation.⁴⁷ Electron paramagnetic resonance (EPR) studies were conducted using DMPO as a spin trap, as shown in Fig. 5F. In the dark, no characteristic peak of DMPO-˙O₂⁻ was detected in all samples. Under illumination, TP-COF exhibits a weak characteristic peak of DMPO-˙O₂⁻, indicating that TP-COF has sufficient reducing ability to reduce O₂ to ˙O₂⁻.¹⁵ UP shows a negligible signal of DMPO-˙O₂⁻, which may be attributed to its low reductivity coupled with poor charge separation. In contrast, UP@TP-COF-0.2 exhibited a strong DMPO-˙O₂⁻ signal, attributed to the enhanced charge separation and preserved high reductivity. Notably, even with boosted charge separation, the physical mixture displays a weaker DMPO-˙O₂⁻ signal than TP-COF, indicating less ˙generated O₂⁻. This is likely due to the charge flow direction following which the electrons tend to gather on UP with lower reductivity. Since ˙O₂⁻ is the major oxidative species for phenol degradation, the poor ˙O₂⁻ production ability accounts for the relatively slow phenol degradation rate. Consistent with the HER and OER results, these results further prove that the charge flow direction is critical for photocatalytic activity.

Mechanistic studies

To further understand the enhanced photocatalytic performance of covalently bonded TP-COF and UP, first-principles calculations were performed to compare the energy barriers to form the critical intermediates during the redox reactions on the composite and the individual components. As is well-known, H* is the key intermediate during HER, and the optimum situation for an active HER catalyst is that the H* adsorbs neither too strongly (G_H* ≪ 0 eV) nor too loosely (G_H* >> 0 eV) on the reaction sites.⁴⁸ Therefore, the adsorption energy of H* on different sites was first compared. Considering that the electron accumulates on TP-COF, and the UP demonstrates negligible HER activity, the active sites for HER are likely on the TP-COF motif. Additionally, the N atoms with lone pairs of electrons are likely to work as active sites for HER, which is further supported by measuring the HER rates in water with cations (Fig. S11, ESI†). A significant decrease in the HER rate is observed after Na⁺ is added to water when no co-catalyst is used. In contrast, the HER rate slightly rises with Na⁺ using Pt as the co-catalyst. The loaded Pt effectively captured the photogenerated electrons and subsequently reacted with the adsorbed H⁺, thereby mitigating the impact of metal cations on the N sites. Moreover, the presence of Na⁺ ions can promote the adsorption of AA on the surface of the photocatalysts,⁴⁹ facilitating the binding of AA to the photogenerated holes of UP, thereby inhibiting charge recombination and promoting the forward reaction. In contrast, cations tend to adsorb on N with lone pairs to block its adsorption of H⁺ for HER, and this result indicates that N atoms may work as the active sites for HER without Pt. Therefore, the G_H* of two N sites are compared, as shown in Fig. 6A. The G_H* at the N atom in the TP-COF (site 1, Fig. 6A and Fig. S12A, ESI†) is 0.021 eV, while it is 0.76 eV on the N atom of the amide bond linking TP-COF and UP (site 2, Fig. 6A and Fig. S12A, ESI†), indicating that site 1 of UP@TP-COF-x is a preferable site for HER. As revealed in Fig. 3B, a net charge of 0.29 e transfers from UP to TP-COF after being covalently bonded. Consequently, the G_H* of site 1 in UP@TP-COF is 0.25 eV lower than that in individual TP-COF, suggesting that the TP-COF motif exhibits higher activity towards HER after covalent bonding with UP. Similarly, the energy barriers during the OER pathway of UP@TP-COF and individual UP are dramatically different. The 4e⁻ OER process is evaluated by comparing the possible intermediates such as *OH, *O, *OOH, and *O–O.⁵⁰ Firstly, the energy barriers of two different OER active sites (sites 1′ and 2′) following the single and dual site pathways have been compared. As shown in Fig. 6B, C, and Fig. S14 and S15 (ESI†), both OER active sites are located on the UP side in UP@TP-COF-x. For the single site pathway, the speed-determining step for both sites 1′ and 2′ is *O + H₂O → *OOH + H⁺ + e⁻. The energy barriers of sites 1′ and 2′ of this step are 2.03 and 1.82 eV, respectively. In comparison, the speed-determining step of the dual site pathway is * + H₂O → *OH + H⁺ + e⁻, for both sites 1′ and 2′. The energy barrier of this step is 1.46 and 1.16 eV for sites 1′ and 2′, respectively. Therefore, the dual site pathway on site 2′ with the lowest energy barrier of the speed-determining step is preferable for UP@TP-COF-x. In order to compare the activity of UP with the covalently bonded sample, the energy barriers on the same site of UP as the site 2′ in UP@TP-COF-x are compared following the dual site pathway. As shown in Fig. 6C, the speed-determining step for UP is the same as that for UP@TP-COF-x but with a higher energy barrier of 1.40 eV, suggesting higher activity of the catalytic sites in UP@TP-COF-x towards OER. As suggested by the active species trapping experiments (Fig. 5E), the main active species for phenol degradation is ˙O²⁻. Therefore, the higher activity of OER catalytic sites also benefits phenol degradation by providing the in situ produced oxygen as the reactant to promote ˙O²⁻ production.

Fig. 6 (A) Free energy diagrams for photocatalytic H₂ production over the covalently bonded composite of different sites and over TP-COFs of site 1. (B) Free energy diagrams for photocatalytic O₂ production over the covalently bonded composite of different sites through a single site pathway. (C) Free energy diagrams for photocatalytic O₂ production over the covalently bonded composite of different sites and over UP of site 2′ through a dual site pathway. (D) and (E) Schematic of the proposed mechanism of the charge flow direction and redox reaction locations over UP@TP-COF-x and physical mixture.

Based on the above analysis, the mechanism for the boosted photocatalytic performance of the covalently bonded TP-COF and UP is schematically illustrated in Fig. 6D. After the covalent bonds are formed between UP and TP-COF, an IEF directing from TP-COF to UP forms, resulting in energy barriers for the transfer of electrons and holes to UP and TP-COF, respectively. Consequently, a Z-scheme charge transfer mechanism dominates at the interface. The holes and electrons produced on TP-COF and UP under light, respectively, recombine at the interface, further boosting the charge separation inside each semiconductor. Meanwhile, the photogenerated electrons and holes accumulate at the more reductive TP-COF and more oxidative UP, respectively. This not only preserves the high redox abilities of both semiconductors but also shortens the charge migration distance to the corresponding catalytic sites. In contrast, the electrons and holes gather on the UP and TP-COF, respectively, in the physical mixture (Fig. 6E). Although the charge separation is enhanced due to the type II charge transfer mechanism, HER and OER occur on UP and TP-COF with lower reductive and oxidative abilities, respectively, which is not favored. Furthermore, a net charge of 0.29 e transfers from UP to TP-COF in UP@TP-COF-x, which leads to the formation of more active catalytic sites to boost the redox reactions. This not only accelerates the surface catalytic reaction but also further improves charge separation by fastening the charge carrier consumption. As an integral result, the photocatalytic performance is significantly improved in the covalently bonded composites.

Conclusions

In conclusion, we have successfully synthesized a series of UP@TP-COF-x composites through covalent linking of TP-COF and UP. The experimental and theoretical results reveal that the charge flow mechanism converts from type II in the physical mixture to Z-scheme in the covalently bonded composite. Further mechanistic studies have revealed that a net charge transfer from UP to TP-COF occurs after covalent bonding, leading to the formation of more active hydrogen and oxygen evolution sites on TP-COF and UP, respectively. Therefore, the photogenerated electrons are directed to migrate to TP-COF with higher reductive ability and hydrogen evolution sites, while holes migrate to UP with higher oxidative ability and oxygen evolution sites in the covalently bonded composite. Thus, not only is the higher redox ability of the TP-COF and UP preserved, but the charge migration distance to the corresponding active sites with enhanced activity is also shortened. As a result of the different charge flow directions, the covalently bonded composite demonstrates significantly higher photocatalytic water splitting and phenol degradation activity compared with the physical mixture. The hydrogen evolution, oxygen evolution and phenol degradation rate of the covalently bonded composite is 20, 14, and 3 times that of the physical mixture, respectively. The strategy of chemically bonded proper semiconductors provides a universal strategy to boost charge separation and create highly active sites simultaneously, which is critical to designing and constructing highly efficient metal-free photocatalysts.

Author contributions

D. B. conceived the idea and supervised the project. Y. Y., X. S., J. Z, and L. H carried out the experiments. L. L. contributes to the theoretical calculations. D. B. wrote the paper and S. H. revised it. All authors analyzed the data and commented on the paper.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Conflicts of interest

The authors declare no conflict of interest.

Acknowledgements

This work was jointly supported by the National Natural Science Foundation of China (No. 21902035, and 51920105004) and the Natural Science Foundation of Guangdong Province (No. 2023A1515030150 and 2024A1515012725). The authors would like to thank Yaping Li from Shiyanjia Lab (www.shiyanjia.com) for the XPS in situ FT-IR analysis. They also would like to thank Analysis and Test Center of Guangdong University of echnology for the TEM, DRS and EPR tests.

References

1. Z. Zhang, X. Chen, H. Zhang, W. Liu, W. Zhu and Y. Zhu, Adv. Mater., 2020, 32, 1907746.
2. P. Sha, L. Huang, J. Zhao, Z. Wu, Q. Wang, L. Li, D. Bu and S. Huang, ACS Catal., 2023, 13, 10474–10486.
3. P. Fu, C. Chen, C. Wu, B. Meng, Q. Yue, T. Chen, W. Yin, X. Chi, X. Yu, R. Li, Y. Wang, Y. Zhang, W. Luo, X. Liu, Y. Han, J. Wang, S. Xi and Y. Zhou, Angew. Chem., Int. Ed., 2024, e202415202,  DOI:10.1002/anie.202415202.
4. Y. Cao, R. Wu, Y.-Y. Gao, Y. Zhou and J.-J. Zhu, Nano-Micro Lett., 2023, 16, 37.
5. X.-H. Liang, A.-X. Yu, X.-J. Bo, D.-Y. Du and Z.-M. Su, Coord. Chem. Rev., 2023, 497, 215427.
6. Y. Wang, H. Liu, Q. Pan, C. Wu, W. Hao, J. Xu, R. Chen, J. Liu, Z. Li and Y. Zhao, J. Am. Chem. Soc., 2020, 142, 5958–5963.
7. Y. You, S. Chen, J. Zhao, J. Lin, D. Wen, P. Sha, L. Li, D. Bu and S. Huang, Adv. Mater., 2024, 36, 2307962.
8. D. Wen, J. Zhao, Y. You, L. Huang, H. Zhu, C. Zhang, D. Bu and S. Huang, Energy Environ. Sci., 2024, 17, 6245–6255.
9. Q. Xu, L. Zhang, B. Cheng, J. Fan and J. Yu, Chem, 2020, 6, 1543–1559.
10. W. Jiang, X. Zong, L. An, S. Hua, X. Miao, S. Luan, Y. Wen, F. F. Tao and Z. Sun, ACS Catal., 2018, 8, 2209–2217.
11. J. Low, C. Jiang, B. Cheng, S. Wageh, A. A. Al-Ghamdi and J. Yu, Small Methods, 2017, 1, 1700080.
12. J. Zhao, J. Ren, G. Zhang, Z. Zhao, S. Liu, W. Zhang and L. Chen, Chem. – Eur. J., 2021, 27, 10781–10797.
13. M.-L. Xu, M. Lu, G.-Y. Qin, X.-M. Wu, T. Yu, L.-N. Zhang, K. Li, X. Cheng and Y.-Q. Lan, Angew. Chem., Int. Ed., 2022, 61, e202210700.
14. K. Xiong, Y. Wang, F. Zhang, X. Li and X. Lang, Appl. Catal., B, 2023, 322, 122135.
15. P. Sha, L. Huang, J. Zhao, Z. Wu, Q. Wang, L. Li, D. Bu and S. Huang, ACS Catal., 2023, 13, 10474–10486.
16. Y. W. Kanghui Xiong, F. Zhang, X. Li and X. Lang, Appl. Catal., B, 2023, 322, 122135.
17. X. Liu, D. Wen, Z. Liu, J. Wei, D. Bu and S. Huang, Chem. Eng. J., 2020, 402, 126178.
18. P. Sha, Y. You, D. Wen, Z. Wu, Q. Wang, D. Bu and S. Huang, Small, 2023, 19, 2205834.
19. S. Wei, G. Zhang and X. Xu, Appl. Catal., B, 2018, 237, 373–381.
20. W. S. Jun Wang, D. Liu, Z. Zhang, Y. Zhu and D. Wang, Appl. Catal., B, 2017, 202, 289–297.
21. Z. Lin, Y. Wang, Z. Peng, Y. C. Huang, F. Meng, J. L. Chen, C. L. Dong, Q. Zhang, R. Wang, D. Zhao, J. Chen, L. Gu and S. Shen, Adv. Energy Mater., 2022, 12, 2200716.
22. Z. A. Lan, Y. Fang, Y. Zhang and X. Wang, Angew. Chem., Int. Ed., 2017, 57, 470–474.
23. J. Zhang, M. Grzelczak, Y. Hou, K. Maeda, K. Domen, X. Fu, M. Antonietti and X. Wang, Chem. Sci., 2012, 3, 443–446.
24. S. Chu, Y. Wang, Y. Guo, P. Zhou, H. Yu, L. Luo, F. Kong and Z. Zou, J. Mater. Chem., 2012, 22, 15519–15521.
25. J. Xie, S. A. Shevlin, Q. Ruan, S. J. A. Moniz, Y. Liu, X. Liu, Y. Li, C. C. Lau, Z. X. Guo and J. Tang, Energy Environ. Sci., 2018, 11, 1617–1624.
26. D. Kong, X. Han, J. Xie, Q. Ruan, C. D. Windle, S. Gadipelli, K. Shen, Z. Bai, Z. Guo and J. Tang, ACS Catal., 2019, 9, 7697–7707.
27. J. Bi, W. Fang, L. Li, J. Wang, S. Liang, Y. He, M. Liu and L. Wu, Macromol. Rapid Commun., 2015, 36, 1799–1805.
28. C. Ye, J.-X. Li, Z.-J. Li, X.-B. Li, X.-B. Fan, L.-P. Zhang, B. Chen, C.-H. Tung and L.-Z. Wu, ACS Catal., 2015, 5, 6973–6979.
29. H. Wang, Y. Wu, M. Feng, W. Tu, T. Xiao, T. Xiong, H. Ang, X. Yuan and J. W. Chew, Water Res., 2018, 144, 215–225.
30. F. Xue, Y. Si, M. Wang, M. Liu and L. Guo, Nano Energy, 2019, 62, 823–831.
31. E. Jin, Z. Lan, Q. Jiang, K. Geng, G. Li, X. Wang and D. Jiang, Chem, 2019, 5, 1632–1647.
32. L. Stegbauer, S. Zech, G. Savasci, T. Banerjee, F. Podjaski, K. Schwinghammer, C. Ochsenfeld and B. V. Lotsch, Adv. Energy Mater., 2018, 8, 1703278.
33. W. Chen, L. Wang, D. Mo, F. He, Z. Wen, X. Wu, H. Xu and L. Chen, Angew. Chem., Int. Ed., 2020, 59, 16902–16909.
34. L. Stegbauer, K. Schwinghammer and B. V. Lotsch, Chem. Sci., 2014, 5, 2789–2793.
35. F. Haase, T. Banerjee, G. Savasci, C. Ochsenfeld and B. V. Lotsch, Faraday Discuss., 2017, 201, 247–264.
36. R. Shi, F. Liu, Z. Wang, Y. Weng and Y. Chen, Chem. Commun., 2019, 55, 12531–12534.
37. P. Pachfule, A. Acharjya, J. Roeser, T. Langenhahn, M. Schwarze, R. Schomäcker, A. Thomas and J. Schmidt, J. Am. Chem. Soc., 2018, 140, 1423–1427.
38. S. Wei, F. Zhang, W. Zhang, P. Qiang, K. Yu, X. Fu, D. Wu, S. Bi and F. Zhang, J. Am. Chem. Soc., 2019, 141, 14272–14279.
39. Y. Yang, N. Luo, S. Lin, H. Yao and Y. Cai, ACS Catal., 2022, 12, 10718–10726.
40. R. Dadigala, B. R. Gangapuram, R. Bandi, A. Dasari and V. Guttena, Acta Metall. Sin. (Engl. Lett.), 2016, 29, 17–27.
41. D. Minakata, D. Kamath and S. Maetzold, Environ. Sci. Technol., 2017, 51, 6918–6926.
42. M. Moradi, Y. Vasseghian, A. Khataee, M. Harati and H. Arfaeinia, Sep. Purif. Technol., 2021, 261, 12.
43. P. M. Martins, C. G. Ferreira, A. R. Silva, B. Magalhaes, M. M. Alves, L. Pereira, P. Marques, M. Melle-Franco and S. Lanceros-Méndez, Composites, Part B, 2018, 145, 39–46.
44. M. Xue, C. Lu, P. Zhang, Z. Lin, D. Bu and S. Huang, Int. J. Hydrogen Energy, 2024, 78, 1214–1223.
45. D. Bu, C. Yang, D. Bu and S. Huang, Int. J. Hydrogen Energy, 2023, 48, 576–585.
46. Y. Xu, J. Liu, M. Xie, L. Jing, H. Xu, X. She, H. Li and J. Xie, Chem. Eng. J., 2019, 357, 487–497.
47. B. Dai, A. Zhang, Z. Liu, T. Wang, C. Li, C. Zhang, H. Li, Z. Liu and X. Zhang, Catal. Commun., 2019, 121, 53–56.
48. D. Bu, D. Bu, W. Chen, C. Huang, L. Li, H. Lei and S. Huang, J. Catal., 2022, 407, 10–18.
49. M. K. T. Chee, B. J. Ng, Y. H. Chew, W. S. Chang and S. P. Chai, Mater, 2022, 15, 13.
50. X. Zeng, Y. Liu, Y. Kang, Q. Li, Y. Xia, Y. Zhu, H. Hou, M. H. Uddin, T. R. Gengenbach, D. Xia, C. Sun, D. T. McCarthy, A. Deletic, J. Yu and X. Zhang, ACS Catal., 2020, 10, 3697–3706.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4mh01596g

‡ These authors contribute equally.

---