Photo-coupled electrocatalytic oxygen reduction to hydrogen peroxide using metal-free CNT-threaded oxidized g-C 3 N 4

Hydrogen peroxide (H 2 O 2 ) has been widely used in environmental cleaning, hospital disinfecting and chemical engineering. Compared to the traditional anthraquinone oxidation method, the electrocatalytic two-electron oxygen reduction reaction (2e - ORR) to produce H 2 O 2 has become a promising alternative due to its green, safety and reliability. However, its industrial application is still limited by the slow reaction kinetics and low selectivity due to the competitive reaction of the 4e - ORR to H 2 O. Herein, we prepare a novel photoresponsive metal-free electrocatalyst based on oxidized g-C 3 N 4 /carbon nanotubes (OCN/CNTs) and introduce an external light field to realize the high-performance electrocatalytic 2e - ORR to produce H 2 O 2 . Impressively, the OCN/CNT electrocatalyst exhibits an outstanding H 2 O 2 productivity of 30.7 mmol/g cat /h with a high faradaic H 2 O 2 efficiency of 95%. The oxygen-containing groups of the OCN/CNTs promote the adsorption of oxygen intermediates and the photocoupled electrocatalysis simultaneously improves the electron transport efficiency and enhances the electrocatalytic selectivity.


INTRODUCTION
Hydrogen peroxide (H 2 O 2 ) is a multifunctional strong oxidant that has been widely used in major industrial production, including chemical synthesis [1,2] , sterilization disinfection [3,4] and environmental governance [5,6] . H 2 O 2 is also regarded as a new clean fuel energy due to its extremely high transportation safety [7][8][9][10] . The traditional method of anthraquinone oxidation to produce H 2 O 2 requires high energy consumption and also causes serious pollution [11,12] . Recently, the electrocatalytic production of H 2 O 2 through the two-electron oxygen reduction reaction (2e -ORR) has arisen significant attention due to its safety, portability and environmental friendliness, which can effectively avoid the large-scale transportation of H 2 O 2 and realize its in-time production and use [13][14][15] . High-efficiency, low-cost electrocatalysts with excellent selectivity are pivotal for a high-performance electrocatalytic 2e -ORR to produce H 2 O 2 . Although traditional noble metalbased catalysts, including Pt, Au and their alloys, exhibit good electrocatalytic activity and 2e -ORR selectivity, their scarcity and high cost and the complexity of electrode forming strongly impede their practical applications [16,17] . Therefore, transitional metal-based and metal-free catalysts are constantly being designed and developed toward the electrocatalytic 2e -ORR and are close to the electrocatalytic performance of noble metal-based catalysts [18][19][20][21] .
Due to their high conductivity and structural tunability as metal-free catalysts, carbon-based 2e -ORR catalysts, including carbon nanotube (CNTs), graphene and carbon nanofibers, are regarded as the most promising potential alternatives to the abovementioned noble metal-based catalysts [22][23][24][25] . However, for pure carbon-based catalysts, their poor intrinsic catalytic activity cannot satisfy the demand for H 2 O 2 production. Furthermore, although the electrocatalytic activity and selectivity of carbon-based catalysts can be significantly improved by constructing structural defects [24,26] and doping with heteroatoms, such as N, B and O [22,27,28] , their catalytic performance is still relatively moderate with significant potential for improvement.
It is noteworthy that in addition to the modification of electrocatalysts, the introduction of external fields, including light, thermal, magnetic and piezoelectric fields, could further improve the catalytic performance [29][30][31][32] . Innovatively, Shen et al. found that Co 3 O 4 nanotube array-supported copper nanoparticles (NPs) showed a higher CO 2 conversion rate under illumination and demonstrated that the light field helped to reduce the potential barrier of electrocatalysis and promoted the electrode kinetics [33] . In general, when external light fields are introduced into traditional electrocatalysis, the specific electronic properties of the electrocatalyst, such as electron transfer, band bending, Fermi energy levels and intermediate desorption energy, are altered, which significantly changes the intrinsic path and performance of the catalysts [34][35][36] . More importantly, under the synergistic effect of photocatalysis and electrocatalysis, the use of a photocoupled electrocatalytic system to enhance the 2e -ORR performance could be a promising strategy for H 2 O 2 production.
As a metal-free conjugated polymer, graphitic carbon nitride (g-C 3 N 4 ) NPs have attracted significant attention in the application of the electrocatalytic ORR because of their unique three-dimensional morphology, tunable electronic structure, superior chemical stability, abundant pyridine nitrogen content, high durability and low cost [37][38][39][40] . More importantly, g-C 3 N 4 possesses a good adsorption capacity for oxygen due to its rich nitrogen content and has been successfully applied in the photocatalytic 2e -ORR to produce H 2 O 2 [41][42][43] . Nevertheless, the poor H 2 O 2 selectivity and high electrical resistance result in inferior electrocatalytic H 2 O 2 production of the pristine g-C 3 N 4 . Various studies have shown that introducing a co-catalyst and constructing a heterojunction system can significantly improve the charge migration and ORR performance of g-C 3 N 4 [44][45][46] .
Herein, we propose a light-field enhancement route for improving the electrocatalytic 2e -ORR to produce H 2 O 2 by designing a novel photocoupled electrocatalyst based on oxidized g-C 3 N 4 /CNTs. Compared to standard electrocatalysis, photo-coupled electrocatalysis can reduce the electrode overpotentials, improve reaction kinetics and promote the electrode stability to facilitate catalytic performance. The CNTs can greatly promote the charge separation and light absorption of g-C 3 N 4 under the introduction of illumination, exposing more electrons for O 2 reduction. Furthermore, the g-C 3 N 4 /CNT hybrid is further oxidized to add more oxygen-containing groups. Rotating disk electrode (RDE) measurements and density functional theory (DFT) calculations illustrate that the combination of OCN and CNTs effectively improve the H 2 O 2 output and enhance the 2e -ORR selectivity.

Preparation of OCN/CNTs
In a typical synthesis process, the GCN was obtained by 15 g of melamine calcined in a muffle furnace at 500 °C for 4 h at a heating rate of 2.19 °C/min. The CNT-threaded GCN-NPs were prepared by nitric acid oxidation. Firstly, the GCN nanorods (GCN-NRs) were synthesized by a hydrothermal treatment [47] . Specifically, 1.5 g of GCN and 1.5 g of urea were dissolved in 150 mL of deionized water and sonicated for 3 h until the solid matter was completely dispersed. Afterward, the solution mixture was transferred into a Teflon-stainless autoclave, followed by the synthetic reactions in an oven at 170 °C for 24 h. Subsequently, a certain amount of GCN-NRs (x = 100, 200, 300 or 400 mg) and 1 mg of CNTs were dissolved in 200 mL of 12 M nitric acid in a three-neck round-bottom flask to synthesize the OCN-x/CNTs (x = 100, 200, 300 or 400 mg) [22] . The effect of different acidification times of the sample was subsequently investigated on the OCN-200/CNTs-y (y = 0, 6, 12, 18 or 24 h). The OCN/CNTs were collected by centrifuging in deionized water and ethanol several times until the pH of the solution reached neutral and dried overnight in a vacuum at 60 °C. The complementary information regarding the electrode preparation, characterization and photoelectric performance measurements of the electrodes, the detection of H 2 O 2 concentration, the RDE measurements and DFT calculation methods are given in the Supporting Information.

Preparation of OCN/CNT catalysts
As shown in Figure 1, the metal-free OCN/CNTs catalyst was synthesized via a three-step approach. In the first step, the bulk g-C 3 N 4 (B-GCN) was obtained by the high-temperature pyrolysis of melamine and was mainly composed of flake accumulation [Supplementary Figure 1A and B]. Next, through thermal polymerization, the B-GCN was rolled up and then formed into the GCN-NRs under the interaction between NH 4 + and CO 3 2originating from urea [47] . Finally, the oxidized carbon nitride OCN/CNTs catalysts were prepared by nitric acid oxidation using the reflux condensation method [41] . It is noteworthy that the mass of the CNTs was fixed at 1.0 mg, the samples with different weights of g-C 3 N 4 were denoted as OCNx/CNTs (x = 100, 200, 300 or 400 mg) and the samples with different oxidation times were labeled as OCN-200/CNTs-y (y = 0, 6, 12, 18 or 24 h). In the oxidation process, the GCN-NRs [Supplementary Figure 1C and D] were transformed into smaller-sized oxidized carbon nitride nanoparticles (OCN-NPs) [Supplementary Figure 1E and F] instead of bulk-OCN. The CNTs acted as building blocks for OCN/CNTs assembly. Figure 2A and B show the morphologies of the OCN/CNTs. From the inset of Figure 2B, the mean size of the OCN-NPs was ~400 nm. The OCN-NPs and CNTs were assembled into OCN/CNTs hybrids through π-π stacking. From TEM images [ Figure 2C and D] of the OCN/CNTs, the CNTs were decorated on the surfaces of inner OCN microspheres with a size of ~5 nm. In addition, it could be seen  from the inset of Figure 2C that the composite sample presented a polycrystalline state. As shown in Figure 2E, the CNTs were adsorbed on surfaces of OCN-NPs and in close contact with the CNTs, which could facilitate charge transfer. The CNTs show tube like-structure with distinct inner and outer walls. The d-spacings of 0.328 and 0.305 nm were attributed to the (002) diffraction crystal planes of graphitic carbon of CNTs [ Figure 2F] and g-C 3 N 4 [ Figure 2G], respectively. The high angle annular dark field (HAADF)-STEM and energy-dispersive X-ray spectroscopy (EDS) mapping images [ Figure 2H] verified the presence and distribution of the main elements of C, N and O. Figure 3A shows the X-ray diffraction (XRD) patterns of the OCN-NPs and OCN/CNTs hybrids with different weight ratios of OCN to CNTs. There were two characteristic diffraction peaks at 2θ of 13.1° and 27.6°, which were attributed to the (100) and (002) crystal planes of graphitic carbon nitride, respectively [42,43] . In the XRD patterns of OCN/CNTs hybrids, the (002) peak was slightly shifted accompanied with a decrease in intensity, because the CNTs were inserted between the layers of g-C 3 N 4 and the interlayers were loosened [48] . As the oxidation time increased, the intensity of the (002) diffraction peak decreased with the reduction of layer stacking [Supplementary Figure 2A]. The Raman spectra further confirmed the successful introduction of CNTs, with Figure 3B manifesting the D and G bands of g-C 3 N 4 located at 1316 and 1567 cm -1 , respectively. As for OCN/CNTs, the D-and G-band peaks showed obvious spikes and migrated to 1330 and 1580 cm -1 due to the interaction of g-C 3 N 4 and CNTs [27,49] . Furthermore, the G and D bands of the OCN-200/CNTs-y catalysts were not affected under the acidification, as indicated in Supplementary Figure 2B. In the FTIR spectra of the OCN/CNTs, the vibration peaks of ~3430 and ~800 cm -1 were indexed to the -OH and triazine units of the OCN/CNTs, respectively [50] . It is noteworthy that the intensity of the peak at ~3400-3500 cm -1 increased with increasing acidification time, indicating that the increase in acidification time improved the degree of oxidation [Supplementary Figure 3]. X-ray photoelectron spectroscopy (XPS) was further conducted to characterize the composition of the OCN-200/CNTs-18h. From the full-scan XPS spectra, the C, N and O elements existed in the OCN-NPs and OCN-200/CNTs-18h catalysts [ Figure 3C]. The C 1s spectrum of the OCN-NPs is shown in Figure 3D. There were four peaks at 288.2, 285.87, 289.39 and 284.8 eV for the OCN-NPs, which were ascribed to the bonding of N=C-N, C-O, O-C=O and C-C/C=C, respectively [42,51,52] . After hybridizing with CNTs, the XPS peak of the N=C-N bond shifted to 288.9 eV, implying the π-π conjugation between the CNTs and OCN-NPs. Interestingly, there were obvious increases in the intensity of the peaks for the C-O and O-C=O, indicating that the introduction of CNTs could enrich the oxygen-containing groups. The N 1s spectrum is shown in Figure 3E, with the peaks at 399.59, 400.62 and 401.78 eV attributed to N 2C , N 3C and N-H X , respectively [43,44] . Similarly, from the O 1s XPS spectra [ Figure 3F], the peak areas of C-O, C=O and O-H at 533.1, 532.5, and 531.8 eV became larger after incorporation with CNTs [48,50] . Supplementary Figure 4 and Supplementary Table 1 demonstrated that the increase in the acidification times could enhance the oxygen content of OCN/CNTs.

Photo-coupled electrocatalytic oxygen reduction to H 2 O 2
The photo-coupled electrocatalytic ORR installation diagram without any sacrificial agent is shown in Figure 4A. From the UV-vis spectra [Supplementary Figure 5], it could be seen that the GCN could be regarded as an excellent photon absorber under light irradiation and the introduction of the CNTs enhanced the visible-light absorption. Thence, the photo-coupled electrocatalytic 2e -ORR performance of the different catalysts was estimated under visible-light illumination (λ ≥ 400 nm). From Figure 4B, after applying the light field, the yield of H 2 O 2 and the faradaic H 2 O 2 efficiency were significantly improved compared to that of electrocatalysis. Obviously, photo-coupled electrocatalysis is an effective H 2 O 2 production strategy for oxygen reduction. Owing to the excellent photoelectrochemical property, the optimum yield of H 2 O 2 for the OCN/CNTs was 30.7 mmol/gcat/h after normalization, which was ~14 and The H 2 O 2 output of the catalysts is illustrated in Figure 4C, which demonstrates that the H 2 O 2 evolution rate has been significantly improved after incorporation with CNTs. The H 2 O 2 concentration for the OCN-400/CNTs was nearly four times higher than that of the OCN-NPs. The yield of H 2 O 2 also increased with increasing CNTs content of the OCN-100/CNTs. When the weight ratio of OCN-NPs to CNTs was 200:1, the H 2 O 2 concentration achieved the maximum. However, as the weight ratio of OCN-NPs to CNTs reached 100:1, the photo-induced electron used for the ORR would decrease due to the decrease in the relative content of OCN-NPs, resulting in a decrease in H 2 O 2 production. Subsequently, the effect of acidification time on the yield of H 2 O 2 with OCN/CNTs was also investigated in Supplementary Figure 6. The OCN-200/CNTs-18h with 18h of acidification time showed the best ORR activity for H 2 O 2 generation. This was mainly because the suitable amount of oxygen functional groups could improve the selectivity of H 2 O 2 , while too much oxygen content would reduce the conductivity of the catalyst, which was not conducive to producing H 2 O 2 [53] . Simultaneously, the yields of H 2 O 2 by using different catalyst samples were also affected by their different photoelectrochemical properties, which was confirmed in the subsequent photoelectrochemical measurements. The activity of the OCN/CNTs was superior to the OCN-bulk/CNTs and the mixture of OCN/CNTs, demonstrating that the π-π conjugation between the OCN-NPs and CNTs in the OCN/CNTs greatly improved the efficiency of the separation and transmission of photogenerated carriers. Generally, the applied potential could affect the stability of the electrode and there would be a competing HER and 4e -ORR [Equation (2)] in the applied potential [54,55] . Figure 4D shows that the optimal reaction voltage was -0.5 V vs. SCE, indicating that H 2 O 2 could be synthesized efficiently at low bias. Nevertheless, the inferior H 2 O 2 production [Supplementary Figure 9A] and the current [Supplementary Figure 9B] in the electrocatalytic system (-0.7 vs. SCE) were sufficient to reflect that the illumination could provide photogenerated electrons to promote the ORR to synthesize H 2 O 2 for the electrode and the advantages of high efficiency and energy saving in this system.
The decomposition rate of H 2 O 2 in the reaction process also played a very important role in the production of H 2 O 2 . The decomposition experiments of H 2 O 2 with different electrodes are presented in Figure 4E.
Compared to the high yield of H 2 O 2 , these decomposition amounts were negligible. It could be seen from Equation (2) that the pH value could greatly affect the performance of the ORR for producing H 2 O 2 . Figure 4F demonstrates that the OCN-200/CNTs-18h had extremely high photoelectrochemical activity at pH = 3 due to the high proton concentration. However, when the pH became acidic, the H 2 O 2 yield would be reduced mainly due to the competition of the HER from Equation (3) [48] . Furthermore, the yield of H 2 O 2 decreased as the pH increased from 3 to 7; however, it again increased at pH > 7, giving a peak at pH 10, before falling at a higher pH value. Surprisingly, the electrode achieved H 2 O 2 generation under alkaline conditions, which was mainly due to the peroxygen species produced by Equation (4) at high pH values [55] . The efficient generation of high-value clean energy (H 2 O 2 ) in a wide range of pH values could provide more possibilities for practical applications.
Supplementary Table 2 illustrates a comparison of the H 2 O 2 yield with current studies. The photocatalytic ORR to H 2 O 2 required the consumption of sacrificial agents and its activity was inferior. Compared to the electrocatalytic ORR [56,57] , the photo-coupled electrocatalytic system is relatively low in energy consumption and inexpensive. Our activity was higher than that of photocatalytic technology. For the electrode, the stability of the catalyst during the reaction process was extremely important [40,55] . In the follow-up experiment, we conducted a long-term reaction current test on the OCN-200/CNTs-18h electrode and found that the current remained unchanged after 36 h under the re-addition electrolyte [ Figure 4G].  Figure 10B], the output of H 2 O 2 and the FE nearly remained the same in each cycle. The XRD and SEM analysis for the electrode before and after the reaction, also further demonstrated its excellent stability. XRD pattern analysis was carried out on the electrode after the reaction three times. Compared with pure FTO, the [002] crystal peak of g-C 3 N 4 appeared at 27.6° and the peak intensity remained unchanged compared with that before the reaction [Supplementary Figure 11A]. Moreover, the morphology of the catalyst did not change during the SEM analysis [Supplementary Figure 11B]. These results demonstrated that the designed OCN/CNTs electrode in this work had high activity and excellent stability.

Photoelectrochemical properties of OCN/CNTs catalysts
The high charge separation efficiency and longer electron lifetime are beneficial to the improvement of the photoelectrochemical performance. For further understanding the mechanism for the high activity and stability of OCN/CNTs under illumination, the migration, transfer and recombination processes of the photogenerated electron-hole pairs in the OCN/CNTs electrodes were investigated. As shown in Figure 5A, the photocurrent densities for the OCN/CNTs with different ratios of OCN to CNTs under visible-light irradiation were much higher than that of OCN-NPs, demonstrating that the introduction of CNTs improves the charge migration significantly. More importantly, the OCN-200/CNTs show the highest photocurrent response with excellent carrier separation efficiency. In fact, a few OCN-NPs could reduce the quantity of photo-excited electrons, while redundant OCN-NPs would make electrons and holes easy to recombine [58] . The appropriate content of CNTs could effectively improve the conductivity of the OCN/CNTs and facilitate the transport of photogenerated carriers. From the electrochemical impedance spectroscopy (EIS) [ Figure 5B], the OCN-200/CNTs possessed the smallest arc radius under low-frequency conditions, indicating the faster charge transfer process at the interface between the catalyst and the electrolyte. Similarly, the higher photocurrent signal [ Figure 5C] and smaller arcs [ Figure 5D] of OCN-200/CNTs-18h further provide powerful evidence that a suitable content of oxygen-containing groups could promote the separation of photo-induced electrons and separation in OCN/CNTs and then transfer faster to the electrolyte interface for the 2e -ORR to H 2 O 2 [3] . In addition, photoluminescence (PL) spectra were employed to investigate the carrier separation processes of the photogenerated electron-hole pairs in the OCN/CNTs electrode. As shown in Figure 5E respectively. Generally, the conduction band position is 0.1 eV more negative than the flat band potential [5] .
To illustrate the mechanism toward the ORR to H 2 O 2 , the electron band structures of the OCN-NP S and OCN-200/CNTs-18h are manifested in Figure 5F. It could be seen that the CB of the OCN-200/CNTs-18h (-0.36 eV) was more negative than that of the OCN-NPs (-0. 31 Figure 13]. The relevant photoelectrochemical characterization under illumination and dark conditions was further used to understand the attribution of solar energy for the photo-coupled electrocatalytic performance. The smaller arcs and more negative current density of OCN-200/CNTs-18h in the EIS and LSV measurements [ Figure 6A and B] verified that the introduction of illumination could improve the electron transfer efficiency and 2e -ORR performance. The corresponding lower Tafel plot slope [ Figure 6C] under light indicated that solar energy could increase the kinetic rate of the reaction to promote the generation of H 2 O 2 . Furthermore, in the CV measurements in an oxygen atmosphere after 1000 cycles, the current density in dark conditions decreased and the half-wave potential decreased by 52 mV, while the current density increased slightly and the half-wave potential increased by 40 mV under illumination [ Figure 6D], which was mainly because of the graphitic carbon nitride continuously producing photogenerated electrons for a supplement under the illumination, making the catalyst in a relatively stable state, while pure electrocatalytic technology was extremely unstable after 1000 cycles. The incident photon-to-current efficiency (IPCE) for the OCN/CNTs electrodes was measured to evaluate the quantum yield as a function For the photo-coupled electrocatalytic system, the electrochemical activity surface area (ECSA) is an important indicator of catalytic performance since it could directly provide the surface area involved in the electrochemical reaction [59] . The double-layer capacitance (C dl ) was measured from the cyclic voltammetry (CV) curves at different scan rates [ Supplementary Figures 17 and 18]. The ECSA of the catalyst samples was calculated by ECSA = C dl /0.0155 mF cm -2 . After normalization, the ECSAs for the OCN-NPs, OCN-100/CNTs, OCN-200/CNTs, OCN-300/CNTs and OCN-400/CNTs were 0.14, 0.43, 0.67, 0.59 and 0.29 m 2 /g cat , respectively [ Figure 7A]. Figure 7B illustrates that the ECSAs of OCN-200/CNTs-18h and OCN-200/CNTs-12h were 0.74 and 0.67 m 2 /g cat , respectively. Surprisingly, as determined from the N 2 adsorption-desorption isotherms [Supplementary Figure 19 and Supplementary As shown in Figure 7C, whether in the Ar (inset) or O 2 atmosphere, the photocurrent density of the OCN-200/CNTs-18h was always higher than that of the OCN-NPs, indicating the excellent carrier separation efficiency of the OCN-200/CNTs-18h catalyst. Interestingly, under the optimal reaction potential conditions (-0.5 V vs. SCE), the photocurrent response and current value of OCN-200/CNTs-18h were significantly  Figure 7D]. The OCN-200/CNTs-18h showed a more negative photocurrent under light and oxygen atmospheres than under dark and Ar atmospheres (inset). It is noteworthy that after illumination the current value of OCN-NPs and OCN/CNTs was lower than that under dark conditions and the Ar atmosphere. The same phenomenon occurred in the O 2 atmosphere of OCN-NPs, which was mainly caused by the increase in resistance after illumination [55] . This phenomenon highlighted that the OCN-200/CNTs-18h electrode was more conducive to electron transfer for the ORRs under the conditions of illumination and O 2 atmosphere. Both the photocurrent and LSV measurements in the Ar and O 2 atmospheres demonstrated that the electrode developed in this work with a superior selectivity for the ORR and could greatly inhibit the HER. Furthermore, the H 2 O 2 production under Ar and O 2 atmosphere is shown in Supplementary Figure 20, which demonstrates that the synthesized H 2 O 2 was derived from O 2 .

Mechanism for 2e -ORR on OCN/CNTs
To further study the mechanism of the 2e -ORR on the OCN/CNT-based electrodes, the selectivity of the 2e -ORR was estimated by RDE measurements in 0.05 M Na 2 SO 4 electrolyte under an O 2 atmosphere. Based on the analysis for the effect of pH value on the H 2 O 2 production [ Figure 3F], the pH value of electrolyte was adjusted to 3 for the RDE measurements with 1 M H 2 SO 4 . From the LSV curves of the catalysts under different rotation speeds [ Figure 8A-C], the current density of the OCN-200/CNTs-18h was obviously superior to those of the OCN-NPs and OCN-200/CNTs-12h, suggesting that introducing CNTs and oxygen-containing groups improve the 2e -ORR electrocatalytic activity. The numbers of transferred   (5) and (6)]: j -1 = j k -1 + b -1 ω -1/2 (5) B = 0.2nFv -1/6 CD 2/3 = A*n (6) where j is the measured current density, j k is the kinetic current density, ω is the rotating speed (rpm), F is the Faraday constant, ν is the kinetic viscosity of water, C is the bulk concentration of O 2 in water and D is the diffusion coefficient of O 2 , respectively.
The contribution of all parameters (denoted as A = 0.032 mC S -1/2 ) to Equation (6) were determined in a solution of Na 2 SO 4 [6] . The slope of the curve (B -1 ) calculated by linear regression for j -1 and ω -1/2 at a designated voltage could determine the average number of electrons (n) in the ORR. Figure 8D indicates that the measured n value of the  Figure 8E. The results indicate that the appropriate introduction of CNTs and oxygen content were selective toward two-electron transfer to produce H 2 O 2 . Simultaneously, it also demonstrated that the photoelectrode of OCN-200/CNTs-18h with a superior selectivity of the two-electron pathway for ORR at a lower voltage, which was consistent with the H 2 O 2 production at the different potentials [ Figure 3D]. To further demonstrate that the ORR procedure on the OCN-200/CNTs-18h was more inclined to the two-electron pathway, the temperature-programmed desorption (TPD) measure was used to study the adsorption capacity of O 2 for the catalyst. In Figure 8F,  DFT calculations were carried out to gain insights into the mechanism. The optimized structural models of the OCN-NPs and OCN-200/CNTs-18h are illustrated in Figure 9A and B, respectively. The CNTs interacted with the OCN-NPs via the π-π conjugation. The relative Gibbs free energies of the reaction intermediates of the ORR to H 2 O 2 are shown in Figure 9C and D. O 2 was adsorbed as an adsorption state (O 2 *) on the catalyst surface, which was then converted to an intermediate state (OOH*) with free hydrogen in the aqueous solution. Subsequently, the intermediate state (OOH*) could undergo a proton coupling process with water molecules to form an adsorbed state HOOH*, finally desorbing into H 2 O 2 [48,50] . In particular, the formation of the adsorption state (O 2 *) on OCN-200/CNTs-18h was a spontaneous reaction corresponding to the ΔG ~ -0.67 eV. However, this process of OCN-NPs was a non-spontaneous process due to the endothermic ΔG ~0.11 eV and required a large activation energy barrier. This result indicated that the OCN-200/CNTs-18h electrode could thermodynamically preferentially facilitate the oxygen reduction to H 2 O 2 with a lower Gibbs free energy of adsorbed oxygen in comparison with OCN-NPs. The TPD confirmed the DFT results, as shown in Figure 8F, with OCN-200/CNTs-18h exhibiting excellent chemical adsorption properties of O 2 . Based on the above experimental results, we believe that OCN-200/CNTs-18h is conducive to the two-electron ORR.

CONCLUSIONS
In summary, we developed a novel nanocomposite by threading OCN-NPs with CNTs through a controlled assembly. As an electrode material, it exhibited high efficiency in photo-coupled electrocatalytic H 2 O 2 production through the 2e -ORR with the optimum H 2 O 2 yield and faradaic H 2 O 2 efficiency reaching up to 30.7 mmol/g cat /h and 95%, respectively. Furthermore, it also displayed strong durability in a long reaction period. The high catalytic activity could be attributed to the unique structure of OCN/CNTs, which promoted the exposure of catalytic sites and facilitated the carrier separation and transfer. More importantly, the application of a light field significantly improved electron transport efficiency and enhanced the selectivity toward H 2 O 2 production by the 2e -ORR pathway. This work provides an important platform for developing high-performance carbon-based nanocomposites as metal-free photo-coupled electrocatalysts for the 2e -ORR to produce H 2 O 2 , which may offer more opportunities for industrial applications.