原位合成S型氮缺陷ZnWO4/g-C3N4異質結及CO2光還原性能

關鍵詞：S型異質結；ZnWO4；g-C3N4；氮空位；CO2光還原

1 Introduction

The continued emission of excess CO2 into the atmosphere isleading to unprecedented challenges in global warming andenergy supply 1–3. Photo-induced CO2 reduction throughartificial photosynthesis is among the key strategies forproducing renewable fuels and environmentally friendlychemicals 4，5. Unfortunately， the dissociation energy of the C＝Obond in the CO2 molecule reaches 750 kJ?mol?1. Theeffectiveness of CO2 photoreduction is limited due to thedilemma of sorption and revivification of CO2 molecules onphotocatalyst surface 6–9. Kinetically， the photoreductionprocess， which requires multi-electron coupled protons， leads topoor efficiency and low selectivity for carbon products 10–13.Therefore， the development of photocatalytic systems capable ofeffective CO2 adsorption and activation， improved carrierseparation and transfer， and enhanced redox capacity is highlydesired.

Tungstate semiconductors have received much attentionbecause of their unique physicochemical properties andstructures. Among them， ZnWO4 is favored by researchersbecause of its high excitation energy， strong ultraviolet response，non-depositional nature， and strong resistance to light andradiation 14. Liu et al. 15 prepared ZnWO4/ZnO compositecatalysts for the primary compositions of glycerol carbonatefrom CO2 and glycerol. Liang et al. 16 prepared ZnSe/ZnWO4 Sschemeheterojunctions with W， O double vacancies for theefficient reduction of CO2， achieving a CO generation rate ashigh as 96.9 μmol?g?1?h?1， about 21.0 times higher than that ofZnSe. However， ZnWO4 has a large band gap （3.3 eV） and slowelectron-hole pair separation efficiency， leading to lowphotocatalytic performance 17，18. The development of compositeheterojunctions containing ZnWO4 represents an excellentmethod to overcome these drawbacks and improve theirphotoresponsive performance.

Constructing heterojunctions is an effective approach totackling energy shortages and overcoming the disadvantages ofsingle-component catalysts. Jin et al. 19，20 demonstrated thatCo3O4 quantum dots/2D graphdiyne （GDY） and GDY/CuBrheterojunctions significantly impacted the field of photocatalytichydrogen evolution. Recently， constructing S-schemeheterojunctions has emerged as a vital research 21–24. S-schemeelectron transport mode achieves the separation ofphotogenerated carriers and preserves the strong redox capacityof whole system， making it a hopeful candidate of highperformanceCO2 reduction 25–27. Typically， S-schemeheterojunctions include both oxidative and reductive photocatalystswith diverse work functions and interleaved energyband structures. Discrepancies in Fermi energy levels can resultin energy band curving and the establishment of a built-inelectric field （IEF）， both of which act as driving forces for carriermigration. Consequently， light-generated electrons with intensereducing capability in reduced photocatalysts and holes withintense oxidizing capability in oxidative photocatalysts areretained to participate in the photoreaction. For example， Xu etal. 28 designed SnO2/Cs3Bi2Br9 heterojunctions， which exhibitedexcellent CO2 reduction activity on account of the S-schemecharge transport route and built-in electric field， with a CH4 yieldof 21.4 μmol?g?1?h?1 and selectivity of more than 70%. Wang etal. 29 constructed a NiS@Ta2O5 heterojunction with a unique Sschemecharge-transfer mode that promoted efficient electronholepair separation， achieving a CO rate of 43.3 μmol?g?1?h?1.

Graphite-phase carbon nitride （g-C3N4）， a valuable newphoto-catalysts in the past decade， offers renewed possibilitiesfor the invention of nanomaterials with unique properties 30–33.g-C3N4 has a favorable energy band structure suitable for CO2photoreduction 34–36. Zhang et al. 37 successfully loaded singleatomAu onto g-C3N4， achieving a CO precipitation rate of 0.6μmol?g?1?h?1 with 94% CO selectivity. Shen et al. 38 constructeda dendritic structure α-Fe2O3/g-C3N4 for CO2 photoreduction，which demonstrates a 3-fold enhancement of CO production ratewith respect to g-C3N4. However， the π-electron conjugatedplanar structure of its triazine ring （C3N3） and 3-s-triazine ring（C6N7） allows for the random transfer of localizedphotogenerated charges in the g-C3N4 planes. The weak van derWaals forces among layers result in unsatisfactory interlayercharge transport and low photogenerated charge separationefficiency， leading to suboptimal photocatalytic CO2 reductionactivity 39. Moreover， defect engineering can achieve effectivecharge separation and migration by guiding internal electronarrangement， changing the surface microstructure， or generatingnew electron excitation orbital directions 40–42. Taken together，constructing S-scheme heterojunctions with defects is expectedto enhance carrier separation efficiency and increase CO2convertibility.

In this work， we have constructed ZnWO4/g-C3N4 S-schemeheterojunctions with nitrogen vacancies via an in situ growthmethod. As a result， the generation yield of CO2 to CO is 232.4μmol?g?1?h?1， which is 11.6 and 8.5 times higher than those ofpristine ZnWO4 and g-C3N4， respectively. This result is superiorto most heterojunction catalysts published in recent years （TableS1）. The S-scheme heterojunction composed of ZnWO4 and g-C3N4 effectively realizes the spatial separation ofphotogenerated carriers， preserving the most potent redox abilityof the photocatalyst. Additionally， nitrogen vacancies （Nv） notonly inhibit electron-hole pair recombination and extend carrierlifetime but also significantly enhance CO2 chemisorption andactivation. The S-scheme charge flow patterns and Nvsynergistically contribute to high CO2 reduction performance.This research provides a guideline for designing S-schemeheterojunctions with defects to enhance photocatalyticperformance.

2 Experimental section

Synthesis of pristine g-C3N4： 10 g of urea as precursor wascontained in a porcelain crucible and calcined in a muffle furnaceat 550 °C for 3.5 h.

Synthesis of ZnWO4/g-C3N4： 1 mmol of Zn（NO3）2?6H2O waspartitioned into 50 mL of deionized water and ultra-sonicated for10 min. Then， 1 mmol Na2WO4?2H2O and 0.03 g of hexadecyltrimethyl ammonium Bromide were added sequentially. Finally，2 mmol of g-C3N4 was mixed with ultrasonication for 20 minand agitation for 1 h. The mixed liquid was diverted to anautoclave and reacted at 160 °C for 6 h. It was cooled to 25 °C，washed and dried. The catalysts of ZnWO4/g-C3N4 with variousmolar ratios （ZnWO4 ： g-C3N4 = 30% ： 100%， 40% ： 100% and50% ： 100%） were named as ZCN-30， ZCN-40 and ZCN-50，respectively.

Synthesis of pristine ZnWO4： The preparation method issimilar to that of ZnWO4/g-C3N4 without adding g-C3N4 powder.

Synthesis of physical mixtures of ZnWO4 and g-C3N4： 0.8mmol of ZnWO4 and 2 mmol of g-C3N4 were preparedseparately and fully milled together. The obtained sample wasnoted as 40ZnWO4 + g-C3N4.

The characterization method and photocatalytic activitymeasurement were shown in the Supporting Information.

3 Results and discussion

3.1 Characterizations

The crystal structure and purity of photocatalysts areillustrated in the XRD plots （Fig. 1a）. The prominent derivationpeak at 27.5° responds to the （002） crystal plane of g-C3N4，which is attributed to the periodic stacking between layers of theconjugated aromatic structure （JCPDS： 33-0664） 43. The weakpeak at 13.2° corresponds to the （100） crystal plane， whichbelongs to the stacking of repeating units of the 3-s-triazine ring.The diffraction peaks of ZnWO4 at 15.5°， 18.8°， 23.8°， 24.5°，30.5°， 36.3°， 38.1°， 41.2°， and 53.7° are attributed to the （010），（100）， （011）， （110）， （?111）， （021）， （200）， （?121） and （202）crystal planes， respectively. It closely matches the standard card（JCPDS No. 73-0554） for monoclinic phase ZnWO4 44. Alldiffraction peaks of the composite catalyst belong to g-C3N4 andZnWO4， indicating that the ZnWO4/g-C3N4 heterojunction withhigh purity is successfully synthesised.

FTIR spectra （Fig. 1b） further elucidate the structural profileof the catalysts. The peak at 463 cm?1 in the FTIR spectrum ofZnWO4 corresponds to the bending mode of Zn―O 45. Thepeaks at 589 and 938 cm–1 are ascribed to the stretching mode ofthe W―O bond， while the sharp peak at 884 cm?1 correspondsto the bending mode of Zn―O―W. The triple-s-triazine unit ofg-C3N4 with the vibrationally crosslinked heptazine deformationmode corresponds to the peaks at 815 and 895 cm?1， and thepeaks within 1246–1640 cm?1 can be classified as C―N and C＝N groups in heterocycle 46. The characteristic peaks of FTIRfor the composite catalysts are common to ZnWO4 and g-C3N4，further demonstrating the successful preparation of theheterojunction.

The morphology and microscopic surface features of thecatalysts were examined by scanning electron microscopy（SEM）， transmission electron microscopy （TEM） and highresolutiontransmission electron microscopy （HRTEM）. ZnWO4exhibits aggregates of nanocrystals （Fig. S1a，b）， while g-C3N4shows a thin-layered porous structure （Fig. S1c，d）. Fig. S1e，fshows the morphology of ZCN-40， indicating a closecombination of ZnWO4 and g-C3N4. TEM pictures for ZCN-40are shown in Fig. 1c，d. Selected area electron diffraction（SAED） of ZCN-40 indicates its polycrystalline structure.Zooming in on the TEM image of ZCN-40 is shown in Fig. 1e.It is distinctly seen from the HRTEM that the lattice stripe withd value of 2.36 nm， corresponding to the （200） crystallographicplane for ZnWO4. In addition， the elemental distribution ofZCN-40 is also analysed by energy dispersive spectroscopymapping （EDS mapping） （Fig. 1f）， which shows that theelements C， N， Zn， W， O are uniformly dispersed in ZCN-40.

3.2 Charge transfer and formation of S-schemejunction

High-resolution X-ray photoelectron spectroscopy （XPS） wasdeployed to characterize the surface chemical composition andelemental valence states of the catalysts. Characteristic peaks ofelements such as C， N， Zn， O and W are detected in the full XPSspectrum （Fig. 2a） of ZCN-40， indicating that the composite iscomposed of ZnWO4 and g-C3N4. For g-C3N4， the characteristiccrest of the C 1s spectrum （Fig. 2b） positioned at 284.80， 285.98，and 288.22 eV attribute to C―C， C―NHX， and N＝C―N 47.The characteristic peaks in the N 1s spectra （Fig. 2c） at 398.18，399.38， and 400.62 eV caused by C―N ＝ C， N―C3， andC―NH 48. The binding energies of C 1s and N 1s for ZCN-40are shifted towards higher binding energies， indicating that g-C3N4 in ZCN-40 loses electrons compared to bare g-C3N4. Thepeaks positioned at 530.30， 532.17， and 533.61 eV in the O 1soptical spectrum （Fig. 2d） of ZnWO4 correspond to latticeoxygen， chemisorbed oxygen， and water molecules 49. The twocharacteristic peaks at 35.08 and 37.22 eV in the W 4f spectrum（Fig. 2e） of the ZnWO4 catalyst are respectively correlated withto W 4f7/2 and W 4f5/2 of W6+ ions 50. The two characteristic peaksat 1025.08， 1048.48 eV in the Zn 2p spectra （Fig. 2f） attribute toZn 2p3/2 and Zn 2p1/2， respectively 16. In contrast to bare ZnWO4，the binding energies of Zn， O， W in ZCN-40 are negativelyshifted， which proves that ZnWO4 gains electrons during theformation of heterojunction. The results show that the chargedistribution is rebalanced during the heterojunction formation 51，and the electrons in g-C3N4 are transported to ZnWO4， leadingto the energy band bending and the generation of IEF.

With the intention of further confirm the electron transportbetween g-C3N4 and ZnWO4， the energy difference between thevacuum and Fermi energy levels was calculated using DFT 52–54.And work function （Figs. 3a，b， S2 and S3） of g-C3N4 and ZnWO4are computed to be 5.33 eV and 5.58 eV， respectively. As aconsequence， the Ef of g-C3N4 is above those of ZnWO4 55，56.When the two are in contact， the free electrons in g-C3N4automatically flow towards ZnWO4 until the Ef bend at theinterface. Meanwhile， a charge depletion region is formed on theg-C3N4 side and a charge accumulation region is formed on theZnWO4 side， which creates an IEF directing from g-C3N4 toZnWO4， which is correspond to the XPS results.

Under light irradiation （Fig. 3c）， electrons in g-C3N4 andZnWO4 are energised from VB to CB， respectively. Due to thecombined effect of energy band bending， Coulomb attractionand IEF， the light-generated electrons on the ZnWO4 CB tend tomigrate to VB of g-C3N4， and the electrons with the strongestreducing and oxidising capacity are retained to participate in thephotoreaction， which is thermodynamically favourable for CO2reduction and implies the initiation of an S-schemeheterojunction between ZnWO4 and g-C3N4. For the purpose ofverifying S-scheme charge transport route in ZnWO4/g-C3N4heterojunctions， we performed in situ irradiation XPS tests.When irradiated with light， the peaks of C 1s and N 1s in ZCN-40 are mobilized towards lower binding energies （Fig. 2b，c）. Incontrast， the binding energies of O 1s， W 4f， and Zn 2p undergosignificant positive shifts （Fig. 2d–f）. Both positive and negativeshifts in binding energies represent decreasing and increasingelectron densities for ZnWO4 and g-C3N4， respectively. Namely，the light-generated electrons in ZnWO4 CB are transferred to theVB of g-C3N4 to recombine with holes. In situ XPS providesconvincing evidence to support the inference of S-schemeheterojunctions.

3.3 Photocatalytic CO2 conversion

To investigate the CO2 photoreduction capabilities of g-C3N4，ZnWO4 and ZnWO4/g-C3N4 heterojunctions， CO?photocatalytic experiments were conducted in a gas-solid reactorfor 3 h without the presence of any molecular co-catalysts orscavengers under irradiation of simulated visible light with axenon lamp. The results （Fig. 4a and Table S2） indicate that theproduction rates of CO for ZnWO4， ZCN-30， ZCN-40， ZCN-50，g-C3N4 are 20.0， 206.7， 232.4， 170.5， and 27.3 μmol?g?1?h?1，respectively. The corresponding CH4 yields were 0.1， 2.3， 3.4，2.1， and 0.2 μmol?g?1?h?1， respectively. The conversion rates（Fig. S4） of O2 were 6.2， 73.9， 81.4， 56.3， and 10.4 μmol?g?1?h?1，respectively. CO and CH4 were the main reduction products andO2 was the main oxidation product in the light-conversion ofCO2， indicating that the catalyst had sufficient redox capacity ofCO2 reduction and H2O oxidation. Bare ZnWO4 and g-C3N4exhibited low CO2 reduction activity owing to the speedycomplexation of light-generated carriers in a singlephotocatalyst. The best CO2 reduction rates for ZCN-40heterojunction is 11.6 and 8.5 times higher than those of pureZnWO4 and g-C3N4， respectively， with a selectivity close to100%. This result could be ascribed to the building ofheterojunctions， which promote charge separation， thusenhancing the catalytic rate. Then， the photocatalytic propertiesof CO2 reduction for the heterojunction ZCN-40 and the physicalmixed material （40ZnWO4+g-C3N4） are compared （Fig. 4b）. Therate of CO generation for ZCN-40 is much greater than that ofphysical mixture （40ZnWO4+g-C3N4）. This comparisonemphasizes that a close interfacial contact between the twophases is essential to facilitate interfacial electron transfer. Blankcontrol experiments （Fig. 4c） confirm that the simultaneouspresence of CO2， H2O， photocatalyst and light irradiation isessential to initiate the photoreaction. In addition， we furthertraced the carbon source through the isotope 13CO2. The massspectra exhibited two distinct fronts at m/z = 29 and m/z= 17 （Fig.S5）， which were assigned to 13CO and 13CH4， respectively，confirming that the products of CO and CH4 were originatedfrom the input CO2. The yields of CO and CH4 increase nearlylinearly with the increase of illumination time （Fig. 4d）. Stabilityis an important measure of a good catalyst. After the catalyst istested for 6 times of CO2 conversion reaction （Fig. 4e）， it stillmaintainsd 92% activity. XRD spectra （Fig. 4f） and XPS spectra（Fig. S6） of N-elemental for ZCN-40 are insignificantly changedafter the CO2 conversion reaction， suggesting that itsmicrostructure is not destroyed， which suggests that its stabilityduring the photocatalytic CO2 conversion process.

3.4 In situ ESR spectra

Fig. 5a–c displays the in situ ESR spectra of ZnWO4， g-C3N4and ZCN-40 in both air and CO2 atmospheres under dark andlight conditions separately. ZnWO4 has no characteristic signalsof defects in any condition. In contrast， under dark conditions， g-C3N4 and ZCN-40 exhibit defect signals belonging to Nvacancies and the defect signals of ZCN-40 are stronger thanthose of g-C3N4， signifying an increase in the Nv-inducedconcentration of unpaired electrons. Under CO2 atmosphere， thedefective signal peaks of g-C3N4 and ZCN-40 are weakened，indicating that CO2 is sorbed on the catalyst surface. Under lightconditions， the intensity of the defect signals increases in eitherair or CO2 atmosphere， suggesting that the role of Nv as an"electron receptor and the successive flow of light-generatedelectrons from CB to Nv drastically facilitate the separation ofcarriers in favour of the CO2 reduction performance 57，58.

Figs. S7 and 5d illustrate the nitrogen （N2） physicaladsorption-desorption isotherms and pore size distributions ofZnWO4， g-C3N4 and different ratios of ZnWO4/g-C3N4heterojunctions （Table S3）. All samples behaved with typical IVisotherms and H3-type hysteresis loops 59，60， indicating that thesesamples are mesoporous structures with slit-like pores， whichcan provide more surface adsorption sites for the CO2 reductionprocess and promote the sorption and revitalization of CO2.

3.5 DFT calculation

Density functional theory （DFT） simulation provides animportant theoretical support to study the energy band structureand electronic properties of g-C3N4 and ZnWO4. The zone gapsof ZnWO4 and g-C3N4 are computed to be 0.25 and 1.16 eV （Fig.6a，b）， respectively. The values are smaller than the experimentalvalues， which is due to the limitations of DFT itself and the lackof computational accuracy in optimally selecting the lattice inthe process of calculating the energy bands. The differencebetween the theoretically calculated and experimental valuesdoes not affect the qualitative analysis of the experimentalresults 61. The calculated results show that ZnWO4 and g-C3N4belong to semiconductors. The DOS plots （Fig. 6c，d） of ZnWO4and g-C3N4 show that the TDOS of ZnWO4 is primarily suppliedby the p- and d-orbitals， and the VB hybridisation is provided bythe p- and d-orbitals together， and CB hybridisation isprincipally provided by the p-orbitals. The TDOS of g-C3N4 isdominantly supplied by the p-orbital and both CB and VB areaffected by the p-orbital 62. DFT results help us to understand theband structure of ZnWO4 and g-C3N4 much better.

3.6 Electronic band structure

The optical features of the photo-catalysts were investigatedemploying UV-Vis diffuse reflectance spectroscopy and thebandgap of the catalysts was calculated. ZnWO4 shows that poorvisible absorption. g-C3N4 has a large visible absorption range.ZCN-40 exhibits obvious double absorption edgescorresponding to ZnWO4 and g-C3N4 （Fig. 7a）， which provesthat the construction of heterojunction is beneficial to theenhancement of catalyst's light absorption ability. Besides， theband gaps of ZnWO4 and g-C3N4 are computed to be 3.36 and2.92 eV （Fig. 7b）， respectively， derived from the equation αhν =A（hν ? Eg）n/2 63. The Mott Schottky curves （Fig. 7c，d） indicatethat both ZnWO4 and g-C3N4 have positive slopes， suggestingthat both are n-type semiconductors. The flat-band （Efb）potentials of pure g-C3N4 and ZnWO4 are measured to be ?1.01and ?0.45 V for Ag/AgCl （?0.79 and ?0.23 V vs. NHE），respectively. Since the CB potentials of n-type semiconductorsare quite nearly the same as Efb， the CB for g-C3N4 and ZnWO4are calculated to be ?0.79 and ?0.23 V （vs. NHE）， respectively.The VB potentials of g-C3N4 and ZnWO4 are calculated to be2.13 and 3.13 eV， respectively， from the equation EVB = Eg +ECB 64. The spacing of the VB to the Fermi energy level （Ef） canbe ascertained from the VB-XPS of pure g-C3N4 and ZnWO4，indicating that the Ef of bare g-C3N4 and ZnWO4 are 0.63 and1.25 eV （Fig. 7e）， respectively. Fig. 7f clearly shows that thestaggered energy band arrangement between ZnWO4 and g-C3N4contributes to the construction of S-scheme structures.

3.7 Opto electronic characteristics

Photo-electrochemical tests further investigated the isolationand migration of light-generated carriers in the catalysts. Thetransient photocurrent response （Fig. 8a） is related to theeffectiveness of the separation of photogenerated carrier incatalyst 65，66. Under visible light irradiation， ZCN-40 displays astronger photocurrent response over ZnWO4 and g-C3N4，highlighting the existence of efficiently separating the charges.ZCN-40 possesses the smallest Nyquist semicircle diameter（Fig. 8b）， suggesting that its interfacial carrier transfer resistanceis the lowest 67，68. The transfer kinetics of light-generated carriersin the catalysts are investigated using time-resolvedphotoluminescence decay spectroscopy （TRPL）. The averagelifetimes of ZnWO4， g-C3N4 and ZCN-40 are calculated to be6.30， 6.77， and 8.27 ns （Fig. 8c， Table S4）， respectively， usingthe equation： τa = （A1τ1 2 + A2τ2 2 ）/（A1τ1 + A2τ2） 69， and theconstruction of the heterojunction significantly prolong thecarrier longevity and improves the utilization of photogeneratedelectrons. The steady-state photoluminescence （PL） spectra （Fig.8d） show that the PL peak of the ZCN-40 heterojunction isweaker than those of ZnWO4 and g-C3N4， suggesting that the SschemeZnWO4/g-C3N4 significantly inhibits the electron-holerecombination and possesses stronger photo-reduced CO2performance.

3.8 Mechanism for CO2 photoreduction

The CO2 photoreduction mechanism of ZnWO4/g-C3N4 wasfurther elucidated using in situ infrared spectroscopy （Fig. 9a）.The characteristic vibrational bands at 1371 and 1453 cm?1 areattributed to bi-dental sorption of CO2 （b-CO2） 70. The uptakepeak at 1472 cm?1 is attested to the monodentate sorption of CO2（m-CO2）. The librational strip at 1747 cm?1 results from thebridging sorption of CO2 （c-CO2）. The peaks at 1627 cm?1 areidentified to H2O. The peaks at 1239， 1255， and 2987 cm–1 thesignal peaks are attributed to COOH* groups 71， which are keyintermediates in CO2 reduction process. The peak of CO isobserved at 2080 cm?1， signifying that CO2 is successfullyconverted to CO in the light. Besides， the peak at 1161 cm?1 isclassified as CH3O*， and the vibrational belts at 2848 and 2911cm–1 are characteristic of CH3OH[v（CH3）] 72， both of which arekey intermediates in the production of CH4 by the reduction ofCO2. The strength of the signature peaks gradually increaseswith the prolongation of the illumination time， signifying that theCO2 light conversion is continued.

Based on the above structural discussion， a proposedmechanism for the light conversion of CO2 to CO on ZnWO4/g-C3N4 S-scheme heterojunction is shown （Fig. 9b）. Whenexposed to visible light， electrons on VB of ZnWO4 and g-C3N4in the ZCN-40 heterojunction are excited to the CB. Actuated bythe IEF， the bending of the energy bands occurs， and the lightgeneratedelectrons on the CB of ZnWO4 are rapidly complexedwith the holes on the VB of g-C3N4， whereas the light-generatedelectrons on the CB of g-C3N4 are retained in large amounts. Onthe other hand， it was found by in situ ESR spectroscopy that Nvalso traps photogenerated electrons， and the photogeneratedelectrons on g-C3N4 CB were continuously flowed to Nv，effectively activating the CO2 molecule and further inhibitingthe electron-hole pair complexation. S-scheme charge transfermode and Nv synergistically facilitate the detachment ofcarriers， prolong the carrier lifetime， and increase the utilizationrate of photogenerated electrons， which enhances CO2 reductionperformance.

The CO2 photoreduction reaction follows the steps shown inFig. 9c. CO2 molecules are first adsorbed on the surface of g-C3N4 in ZnWO4/g-C3N4， and *CO interacts with an H+ and aphotogenerated electron dissociated from H2O to produce theimportant intermediate product *COOH. Then *COOH gainsanother electron and H+ and removes a molecule of H2O toproduce *CO. Finally， *CO desorbs from the ZnWO4/g-C3N4surface to produce CO molecules. In the other part， *COOHundergoes successive multi-step proton-coupled electronreactions to produce *CH3O and CH4.

4 Conclusion

In summary， we have successfully prepared ZnWO4/g-C3N4S-scheme heterojunctions with N defects. High-performance，and hyper-selective diversion of CO2 to CO has been achieved，ZCN-40 heterojunction shows nearly 100% selectivity for CO，with a production rate of up to 232.4 μmol?g?1?h?1， which is 11.6and 8.5 times higher than those of pristine ZnWO4 and g-C3N4，respectively. In this reaction， the unique S-scheme chargetransfer mode significantly promotes the light-generated carriermigration and separation， prolongs the carrier lifetime，maximizes the strong reducing ability of the conserved lightgeneratedelectrons， and improves CO2 reduction activity. Onthe other hand， the presence of N defects further inhibits carrierrecombination， improves the utilization of photogeneratedelectrons， and facilitates CO2 adsorption and activation. The Sschemecharge transfer and N defects synergistically promotehigh-performance， highly selective conversion of CO2. Thiswork presents profound knowledge into the application of Sschemeheterojunctions based on g-C3N4 and ZnWO4 withdefects in photocatalytic CO2 reduction reactions.

Author Contribution： Conceptualization， J. Q. and Y. Z.;Methodology， J. Q. and Y. A.; Validation， J. Q. and Y. A.;Formal Analysis， J. Q.， Y. A. and Y. Z.; Investigation， J. Q.; DataCuration， J. Q.; Writing-Original Draft Preparation， J. Q.;Writing-Review amp; Editing， Y. Z.; Visualization， J. Q.;Supervision， Y. Z.; Project Administration， Y. Z.; FundingAcquisition， Y. Z.

Supporting Information： available free of charge via theinternet at https：//www.whxb.pku.edu.cn.