利用改性g-C3N4光催化還原CO2

關(guān)鍵詞：g-C3N4；光催化；CO2還原；改性；S型異質(zhì)結(jié)

1 Introduction

The rapid population growth and extensive reliance oncarbon-based fuels during the industrialization era have led tothe depletion of nonrenewable fossil fuel resources. Thisoverconsumption of fossil fuels has precipitated an acute energycrisis and the continuous release of greenhouse gases， mostnotably carbon dioxide （CO2）. The escalating levels of CO2 inthe atmosphere， largely driven by industrial activities， haveraised significant concerns due to their substantial contributionto global warming. Consequently， addressing this issue throughthe progression of green and renewable energy technologies hasbecome imperative.

Solar energy， in particular， is considered an inexhaustible andpromising alternative to fossil fuels. The growths in solar energytechnologies are critical for reducing dependence on finiteresources and extenuating the adverse environmental impactslinked with traditional energy sources. By capitalizing in solarenergy， society can progress toward a more sustainable future，fostering a cleaner and healthier planet for future generations.The conversion of solar energy into chemical fuels is widelyrecognized as a sustainable and environmentally friendlyapproach to addressing the dual challenges of energy scarcityand environmental degradation.

Inspired by the natural process of photosynthesis， scientistshave developed environmentally benign technologies that utilizesolar energy to change CO2 into hydrocarbons. In 1978，Halmann et al. introduced the concept of photocatalytic CO2reduction， often referred to as artificial photosynthesis 1.Subsequently， Inoue et al. explored the removal of CO2 via gassolidreactions involving water vapor and semiconductors suchas TiO2， CdS， GaP， ZnO， and SiC 2. Their research demonstratedthe photochemical conversion of CO2 into various compounds，including HCOOH， HCHO， CH4， and CH3OH， and providedinsights into the underlying reaction mechanisms. Thistechnology represents a green and sustainable approach tomitigating CO2 emissions， with the potential to produce valuablehydrocarbons using solar energy. By emulating naturalphotosynthesis， scientists have made significant strides towardcombating climate change and developing innovative solutionsfor a more environmentally sustainable future.

Initial scientific investigations into the photocatalyticprocesses of g-C3N4 primarily focused on examining itsconduction band （CB） and valence band （VB） positions， as wellas the redox processes involved in water splitting 3. In 2009，Wang and colleagues successfully employed g-C3N4 to catalyzewater splitting into hydrogen under solar irradiation， marking asignificant advancement that spurred extensive research into thephotocatalytic properties of g-C3N4 4. Subsequently， Dong et al.reported the efficacy of g-C3N4 in converting CO2 into COthrough gas-solid heterogeneous catalysis under visible lightirradiation 5. Since then， g-C3N4 has gained recognition as apromising material for reducing atmospheric CO2concentrations.

g-C3N4 is composed of carbon and nitrogen atomsinterconnected by strong covalent bonds， forming a robustskeleton structure analogous to the two-dimensional conjugatedplane structure of graphene. The interlayer distance isapproximately 0.326 nm. Chemically， g-C3N4 exhibits highstability， even in acidic or basic solutions 6. It is a non-toxiccompound with a band gap of 2.7 eV and can be easilysynthesized from low-cost raw materials 7. These attributes makeg-C3N4 an attractive candidate for researchers seeking efficientand environmentally friendly catalysts for CO2 reduction 8. Asdepicted in Fig. 1， g-C3N4 entails two basic structural units： thetri-s-triazine ring （C6N7） and the s-triazine ring （C3N3） 9，10. Theformation energy of C6N7 is relatively low， making g-C3N4 withC6N7 as its fundamental unit a relatively stable structure 11.

Photocatalytic CO2 reduction is a complex process involvingmultiple sequential steps. To gain a comprehensiveunderstanding of this process， it can be broadly categorized intothree pivotal stages： （1） the photocatalyst absorbs a photon withenergy greater than its bandgap energy （Eg）， resulting in thegeneration of electron-hole pairs; （2） the separation andmigration of these photogenerated electron-hole pairs; （3） thephotogenerated electrons migrate to the surface of thephotocatalyst， where they reduce CO2 and water into fuels atactive sites （I） associated with the CB edge， while the holespartake in the oxidation of water into O2 at other active sites （II）linked to the VB edge （Fig. 2）.

The products formed during this reaction， such as methane，methanol， carbon monoxide， formic acid， hydrogen， and oxygen，are highly valuable compounds with diverse applications in thechemical industry. However， the process of CO2 reduction isintricate and multifaceted， as illustrated by Eqs. （1）–（8）. Exceptin Eq. （1）， the reduction potentials for CO2 to CO （?0.12 V），CH3OH （0.03 V）， and CH2O （?0.07 V） are quite similar， leadingto the parallel formation of multiple products with lowselectivity for the desired product 12.

This review explores various strategies to enhance theseparation efficiency of photoinduced electron-hole pairs in g-C3N4， which provides a decisive role in CO2 reduction. Thesestrategies include controlling the morphology， doping with ions，creating defects， and forming heterojunctions. The chargetransfer mechanisms of type-I， type-II， and type-IIIheterojunctions， along with their associated limitations， are alsoanalyzed. The primary focus of this review is on the chargetransfer mechanism of g-C3N4-based S-scheme heterojunctionsand their services in photocatalytic CO2 reduction. Finally，considering the significant potential of g-C3N4 and the currentstate of research worldwide， the standing challenges and keyareas for future breakthroughs are identified and discussed.

2 Synthetic methods

As illustrated in Fig. 3， there are several methods available forthe synthesis of g-C3N4. Typically， g-C3N4 is fabricated throughthe chemical reaction of nitrogen-rich precursors in specificproportions using methods such as thermal polymerization，chemical deposition， hydrothermal/solvothermal techniques，and microwave-assisted processes， among others. Commonprecursors include melamine， urea， thiourea， and melaminechloride. This section provides a brief overview of the mostcommonly employed synthesis methods， highlighting theirrespective advantages and limitations. A summary of the variousmethods for preparing g-C3N4 using different precursors ispresented in Table 1.

2.1 Thermal polymerization method

The solid-state method， a subset of thermal polymerization，involves the reaction of nitrogen-rich organic compounds at hightemperatures and pressures， typically without the addition ofsolvents. For example， Zhang et al. utilized melamine as anitrogen source， reacting it with cyanuric chloride to synthesizegraphitic carbon nitride （gCN） at temperatures of 500–600 °Cand pressures of 1–1.5 GPa 13. Guo et al. also employed asolvent-free method to successfully prepare carbon nitride witha graphite-like layered structure 14. This strategy presents thepossibility for synthesizing superhard carbon nitride. Venkateshet al. applied a straightforward solid-state reaction to fabricate aperovskite-type CaSnO3/g-C3N4 （GCN） heterojunctionstructure 15. The CaSnO3 nanoparticles were anchored onto gCNnanosheets， and their photocatalytic properties were explored.Study of photocatalytic mechanisms revealed a reduction in theoptical band gap of the CaSnO3/g-C3N4 composite， alongside asuppressed recombination of excitons， resulting to enlargephotocatalytic performance. Consequently， the GCN exhibitedsuperior degradation proficiency of methylene blue （MB） dye，thereby establishing a foundation for the design of 2Dheterojunction structures aimed at improving photocatalyticperformance.

However， the solid-state method， which requires hightemperatures and pressures， results in highly crystalline g-C3N4，but the nitrogen content in the products is relatively low. Theprocess is difficult to control， and the introduction of impuritiesduring synthesis often leads to low yields. Furthermore， there aresignificant safety concerns associated with this method， whichlimits its broader application.

The selection of precursors and preparation methods for gCNis critical in determining its catalytic performance. The mostcommon approach to fabricating gCN is thermal polymerization，using cost-effective nitrogen-rich materials with C-N corestructures， such as melamine 16， cyanamide 17， dicyandiamide 18，urea 19， thiourea 20， or a combination of these 21. As depicted inFig. 4， several researchers have summarized the preparation ofg-C3N4 using nitrogen-rich precursors at varioustemperatures 22–26. Calcination temperature is a key factor indetermining the band structure of gCN， with changes in thecalcination temperature of melamine altering the C/N ratio andband gap of the final product 27. The precursor sources andcalcination temperatures influence the electronic band structureof gCN， ultimately affecting its photocatalytic performance.Understanding these factors and how they control themorphology of gCN is essential for designing efficientphotocatalysts for environmental and energy applications. Forexample， gCN derived from urea typically exhibits a higherspecific surface area than that prepared from melamine， whichsubstantially influences photocatalytic performance 28.Additionally， g-C3N4 obtained from urea has a larger band gapcompared to that produced from thiourea 29. The band gap iscrucial for determining the material's ability to absorb light andfacilitate photocatalytic reactions. Upon exposure to visiblelight， g-C3N4 synthesized from urea demonstrates superiorhydrogen production performance compared to that obtainedfrom thiourea and dicyandiamide because of its porous structureand high specific surface area 30，31. This underscores theimportance of selecting suitable precursors and optimizing thesynthesis process to enlarge the activity of g-C3N4. Byunderstanding how different precursors and polymerizationconditions affect the properties of g-C3N4， researchers can tailorthe synthesis process to achieve the desired photocatalyticactivity. Factors such as specific surface area and band structureare critical during the synthetic process of gCN-basedcatalysts 32. These insights are vital for refining the efficiency ofphotocatalytic applications， including hydrogen production andCO2 reduction 33， and provide valuable guidance for furtherenhancing gCN’s photocatalytic performance.

Yang et al. directed a systematic study on the impact ofpyrolysis temperature， holding time， and heating rate on theshape and properties of gCN 34. Controlling the heating rateduring gCN synthesis can tailor its properties for variousapplications， including photocatalysis， adsorption， andlubrication. For instance， the heating rate significantly affects thestructural properties of gCN， revealing a distinct structureactivityrelationship. Rapid heating results in gCN with thinnerlayers， larger surface areas， more amino defects， and moreconjugated structures， all of which enhance its photocatalyticproperties. In contrast， slower heating produces gCN with ahigher degree of polymerization， a well-defined layeredstructure， and reduced interlayer spacing， making it moresuitable for lubrication applications. Higher pyrolysistemperatures and prolonged heating times can further improvegCN’s performance. Lu et al. synthesized g-C3N4 catalysts usingsix different precursors and investigated their photocatalyticapplications for activating hydrogen peroxide，peroxymonosulfate （PMS）， and peroxydisulfate （PDS） undervisible light irradiation 35. Fig. 5a illustrates the molecularstructures of the different precursors under the same synthesisconditions， and Fig. 5b shows the estimated band structures ofthe resulting g-C3N4 samples. Among the six catalysts， g-C3N4prepared from trichlorocyanuric acid （TCA-CN） exhibited themost favorable characteristics， including the narrowest band gap，the largest specific surface area， the lowest carrier recombinationrate， and excellent performance in degrading bisphenol A （BPA）.On the other hand， gCN synthesized from urea （U-CN）demonstrated the highest photoelectron density， the baggiestband gap， the most positive VB potential， and the highestactivation efficiency. These findings indicate that the choice ofprecursors significantly influences gCN’s performance， withTCA-CN and U-CN emerging as particularly promising optionsfor water treatment applications.

The thermal polymerization method is a simple and efficientapproach for fabricating g-C3N4. However， direct thermalpolymerization often leads to aggregation due to strongintermolecular forces， resulting in a reduced specific surfacearea. Additionally， traditional thermal polymerization methodsface challenges in real-world applications. Consequently，numerous practices have been planned to develop the efficiencyof carbon nitride production.

2.2 Chemical deposition method

Chemical deposition methods encompass bothelectrochemical deposition and vapor deposition. In theelectrochemical deposition method， substances in a solution aredeposited onto an electrode surface using an external electricfield. Under the influence of this electric field， precursorscontaining nitrogen-rich organic compounds in the electrolytemove as positive and negative ions， eventually undergoing aredox reaction on the electrode surface to form g-C3N4. Thistechnique is widely employed to synthesize g-C3N4 with uniquenanostructures for various applications. On the other hand， vapordeposition involves decomposing hydrocarbons into highlyreactive small molecules through heating or other auxiliaryprocesses， which then polymerize under suitable conditions toform g-C3N4 macromolecules.

Bai et al. successfully synthesized hollow g-C3N4microspheres using a straightforward electrodepositiontechnique with silicon nanospheres as templates 36. Thesemicrospheres consisted of numerous nanoparticles， ranging insize from 5 to 30 nm， with overall diameters between 800 and1100 nm and shell thicknesses of approximately 80 to 250 nm.This innovative method of fabricating carbon nitride withspecific nanostructures via electrodeposition demonstrated thepotential for creating these structures under ambient conditionsand at low temperatures. The findings open new pathways forthe production of a wide variety of carbon nitride nanostructureswith diverse morphologies and architectures， using differentprecursors and templates.

Zhou et al. used a simple electrochemically-aided process toconstruct g-C3N4 nanosheets at a temperature of 25 °C 37. In thisstudy， ultrathin g-C3N4 nanosheets （ECN nanosheets） werequickly synthesized within 30 min by applying a direct current（DC） power supply， using melamine as the precursor at roomtemperature （Fig. 6a）. Through the electrophoretic deposition（EPD） method， ECN nanosheets were deposited onto an anodicAl2O3 （AAO） substrate， leading to the rapid formation of ahomogeneous g-C3N4 nanosheet film in just a few minutes， atleast seven times faster than conventional methods， while alsosimplifying the synthesis steps. The effect of voltage on thefabrication of ECN nanosheets was also investigated. When theapplied voltage was 3 V， the ECN formed in filament shapes（Fig. 6b）. As the voltage increased， the sample expanded fromfilament-like forms into larger sheets （Fig. 6c–e）. Additionally，the formation rate of the ECN nanosheets increasedprogressively as the voltage was raised from 3 V to 7 V （Fig. 6f）.However， at 9 V， the formation rate decreased despite theincreased charge density and ion movement， likely due to a sidereaction involving the electrooxidation of the amino C―N bondwith the nitrile C≡N bond at higher voltages. This was confirmedby FTIR spectroscopy results （Fig. 6g）， where the intensity ofthe ―C≡N peak （2150 cm?1） increased with voltage.

The electrochemical deposition method offers advantagessuch as fast production rates， ease of operation， and the ability tocover large deposition areas. Moreover， the morphology andproperties of the resulting products can be finely controlled byadjusting variables such as the concentration of raw materials，electrode potential， and deposition time within theelectrochemical system. However， there are limitations to thistechnology， including the requirement for electrolytes of highpurity and stability， which can restrict its broader applicability.

2.3 Hydrothermal/solvothermal method

Hydrothermal/solvothermal methods are widely utilized forthe synthesis of g-C3N4 because of their low cost， simpleoperation， and minimal environmental pollution. Wang et al.prepared hollow g-C3N4 microspheres （HCNMS） exhaustingmelamine and dicyandiamide as precursors through a templatefreehydrothermal method 38. Subsequently， the HCNMS werefunctionalized with hydroxyl groups to form hydroxylfunctionalizedHCNMS （OH-HCNMS） by replacinghydrochloric acid with sodium hydroxide， as depicted in Fig. 7.The OH-HCNMS demonstrated considerably improvedphotocatalytic activity， with a degradation efficiency of phydroxybenzoicacid （HBA） that was 4.3 times larger than theoriginal g-C3N4. Additionally， the apparent quantum efficiency（AQE） of OH-HCNMS at a wavelength of 420 nm was found tobe 3.7%.

Wang et al. also synthesized g-C3N4 microspheres （CNMS）via a simple solvothermal route with dicyandiamide andmelamine as starting materials 39. They investigated the effect ofreaction temperature on the structural， energy band， andphotocatalytic activities of CNMS. It was observed that samplesprepared at 180 °C and 200 °C exhibited spherical morphologies，while the sample fabricated at 160 °C had an asymmetricalshape. Moreover， lowering the temperature led to increased bandgap. Therefore， CNMS synthesized at 180 °C demonstratedsuperior photocatalytic activity compared to those prepared atother temperatures.

Cui et al. synthesized g-C3N4 hollow spheres using a one-stepsolvothermal approach with acetonitrile as the organic solvent，without employing any templates 40. The resulting materialsexhibited a typical graphite-like structure， and a π-conjugatedskeleton based on heptazine units. The higher condensationtemperature increased the surface area of samples， broadened thevisible light absorption range， and reduced the recombinationrate of photoinduced charge carriers. This solvothermalapproach， conducted at relatively low temperatures， isparticularly advantageous for designing and fabricating polymerphotocatalysts with advanced morphologies andmicrostructures.

The solvothermal method is favored for the synthesis of g-C3N4 because of its relatively mild reaction conditions and thehigh purity of the products. g-C3N4 prepared through this methodgenerally exhibits a well-defined structure， reducedagglomeration， minimal incomplete polymerization， and goodphotoelectrochemical properties. However， the solvothermalmethod has certain drawbacks， including a lengthy reactioncycle， low yield， and specific equipment requirements.Consequently， despite its potential， the method’s longpreparation time and limited production scale restrict its broaderapplication.

2.4 Microwave-assisted method

Compared to the hydrothermal and solvothermal methods， themicrowave-assisted approach offers a more efficient and timesavingalternative， making it an invaluable tool for researchersin various fields. This method， which employs microwaves as aheating source， is a rapid polymerization technique with thepotential for large-scale preparation of g-C3N4. In 2014， Yuan etal. pioneered a microwave-assisted technique for synthesizinghighly crystalline g-C3N4 photocatalysts in a short amount oftime 41. The resulting g-C3N4 exhibited excellent hydrogen （H2）production performance， demonstrating the effectiveness of thismethod for photocatalytic applications. Additionally， theflexibility of this approach allows for the use of various nitrogenrichcompounds， such as melamine， cyanamide， anthiourea，expanding the range of possible precursor materials.

Compared to traditional methods of g-C3N4 fabrication， themicrowave-assisted heating synthesis offers several advantages.First， it is both time- and energy-efficient， as the reaction can becompleted within minutes. Second， this method is highlyversatile， accommodating a variety of nitrogen-rich organicprecursors for the production of g-C3N4. Third， the microwaveassistedapproach allows for large-scale production， yieldingseveral grams of the photocatalyst in a single batch. Lastly， theprocess is easy to operate and reproducible， making it anattractive option for practical applications.

Guo et al. successfully synthesized high-quality g-C3N4 withminimal structural defects within just 16 min using microwaveassistedpyrolysis of a melamine-cyanuric acid supramolecularaggregate （MCA） 42. The hydrogen production rate of themicrowave-prepared samples was double that of those preparedusing conventional heating at 540 °C for 2 h in a muffle furnace.This cost-effective method enables the pyrolysis of precursorsand supramolecular aggregates， facilitating the customizeddesign of efficient photocatalysts for various applications

2.5 Other methods

With the exception of the methods discussed above， othertechniques including the sol-gel method， precursor preassemblymethod， and template method are also employed for thesynthesis of g-C3N4. The sol-gel method is particularlypromising for the fabrication of nanomaterials， as it effectivelydisperses precursors in solvents to form a sol suspension atrelatively low temperatures. Through hydrolysis andcondensation， colloidal particles in the sol form nanomaterials.This technique permits a precise control over the size and shapeof the resulting nanomaterials， making it exceedingly fit fornumerous services in material science and nanotechnology.

Kailasam et al. used the sol-gel method to fabricatemesoporous carbon nitride using cyanamide and tetraethylorthosilicate as starting materials， with acidic ethanol as thesolvent 43. Through subsequent thermal treatment andcondensation， the resulting mesoporous carbon nitride displayedboosted hydrogen generation performance compared to bulk g-C3N4 and other mesoporous carbon nitrides prepared using therigid template method. This method’s ability to finely tunematerial properties makes it a valuable tool for enlarging theefficiency of photocatalysts and other nanomaterials. As shownin Table 1， the present situation of g-C3N4 synthesized bydifferent precursors in the field of photocatalytic reduction ofCO2 is summarized 44–54.

3 Modified g-C3N4

Despite its many advantages， the photocatalytic performanceof g-C3N4 remains suboptimal for practical applications，necessitating further development. To enhance its photocatalyticactivity， many modification strategies have been employed.These include morphology control， doping， single-atommodification， and the formation of heterojunctions. Each ofthese strategies is aimed at improving specific aspects of g-C3N4’s performance， such as increasing its surface area， tuningits electronic properties， or improving charge separationefficiency 55–74. An overview of these modification approachesis presented in Table 2.

3.1 Morphology control

The photocatalytic activity of bulk g-C3N4 for CO2 reductionremains under shadow due to its inadequate light absorption， lowspecific surface area， and rapid recombination of photogeneratedcarriers. To overcome these challenges， researchers havedeveloped various morphologies of g-C3N4 including 0-dimensional quantum dots 75， 1-dimensional nanorods 76， 2-dimensional nanosheets 77， 3-dimensional porous materials 78，hollow nanospheres 79 and other morphologies 80 （Fig. 8）. Thesemorphological modifications significantly enhance the catalyticefficiency of g-C3N4 by altering its physicochemical properties.

At the macroscopic level， tuning the structure of g-C3N4improves its optical and physicochemical properties byincreasing light absorption and providing more active sites forphotocatalytic reactions. Additionally， these structural changesoptimize the diffusion paths and distances of charge carriers，causing more efficient charge separation and reducedrecombination. Indeed， the photocatalytic performance of g-C3N4 for CO2 reduction varies considerably depending on itsmorphology.

Previous studies have extensively highlighted the significanceof carefully preparing and designing g-C3N4 photocatalysts withvaried structures， ranging from quantum dots to nanospheres 81.It has been consistently observed that the morphology of g-C3N4plays a critical role in determining its efficiency in photocatalyticreactions. Factors such as the size and dimensionality of g-C3N4structures are identified as key determinants of theirphotocatalytic performance. For instance， smaller， lowerdimensionalstructures tend to exhibit enhanced quantum effects，whereas larger， three-dimensional structures offer increasedsurface area， contributing to higher catalytic activity.

In addition to size and dimensionality， the composition andsurface characteristics of g-C3N4 are crucial in influencing itsphotocatalytic activity 82. Modifications to the surface， such asthe incorporation of heteroatoms or the introduction of defects，have been shown to meaningfully extend photocatalysis withg-C3N4. These surface modifications enhance light absorption，charge transfer， and overall catalytic efficiency. Therefore， acomprehensive approach to the design of g-C3N4 photocatalyststructures must account for not only their physical dimensionsbut also their surface composition.

In summary， the effectiveness of g-C3N4 in photocatalyticapplications can be greatly improved through careful control ofits morphology， composition， and surface properties. An indepthanalysis of these factors and their practical implicationshas been explored in the literature， offering valuable insights foroptimizing g-C3N4 photocatalysts for enhanced performance invarious photocatalytic processes 83，84.

3.1.1 g-C3N4 with 0D structure （quantum dots）

g-C3N4 quantum dots （CNQDs） are nanomaterials withdimensions of less than 10 nm in all three axes， classifying themas zero-dimensional quantum dots （QDs） 85. Due to theirextremely small size， CNQDs significantly reduce the averagediffusion distance of photo-excited carriers， enhancing theseparation efficiency of electron-hole pairs in catalysts. Thisleads to improved electron transport on the catalyst surface andnotably higher quantum efficiency across the system.

Since their initial development in 2014， CNQDs havegarnered substantial interest because of their sole physical andchemical properties， including their structure， morphology，electrochemical behavior， and photoelectric performance 86. Thefabrication process of CNQDs generally involves multiple steps（Fig. 9a）. For example， one method starts with the thermaloxidation and engraving of bulk g-C3N4 into nanoplates，followed by acid cutting to obtain nanobelts， and finally，hydrothermal cutting at 200 °C to produce quantum dots. TEManalysis has shown that these CNQDs typically have diametersranging from 5 to 9 nm （Fig. 9c，d）， with a crystal lattice spacingof 0.336 nm （Fig. 9e）. Another approach， developed by Zhan etal.， employs a one-step hydrothermal process to prepareCNQDs. In this method， bulk g-C3N4 is suspended in apotassium hydroxide and ethanol solution， followed by heatinghydrothermally at 180 °C for 16 h 87. The presence of potassiumhydroxide facilitates the stripping process， yielding CNQDs withan average diameter of 3.3 nm （Fig. 9b）.

Both of these routes follow a top-down method， whichtypically involves reducing the size of larger g-C3N4nanostructures to form nanoscale CNQDs. While the top-downmethod is intricate and often involves harsh reagents like strongacids and bases， it allows for better control over the uniformityof CNQDs and is more scalable for mass production.

Alternatively， the bottom-up method offers a morestreamlined synthesis process by utilizing organic molecules asprecursors in a single step. In this method， organic moleculesundergo condensation reactions to form macromolecularintermediates， which are then broken down by controllingexternal pressure and temperature. This process generatescarbon-based fragments and free radicals that undergocarbonization， resulting in the formation of nanoscale quantumdots. Typically， the bottom-up method employs compounds likeformamide， N，N-dimethylformamide （DMF）， melamine，guanidine hydrochloride， urea， dicyanamide， and organic aminesas building blocks. These compounds facilitate the synthesis ofnon-uniformly doped quantum dots. The method also allows forprecise control over the nanostructure of CNQDs. For instance，Liu et al. used microwave irradiation to break chemical bonds incarbon （CCl4） and nitrogen （1，2-ethylenediamine， EDA）precursors. This rapidly triggered the polymerization andcarbonization processes， producing CNQDs （Fig. 9f）.Microwave radiation provided uniform heat distribution，ensuring a consistent size for the photoluminescent CNQDs，which ranged from 2 to 4 nm in diameter， with well-dispersedparticles averaging between 1 to 5 nm in size 88.

One of the most distinctive properties of quantum dots is theirup-conversion effect， which leads to the initiation of the anti-Stokes shift 89. In typical photocatalytic reactions， much of theavailable infrared and near-infrared light remains unusable bymost semiconductor materials， which can only harness visible orultraviolet light. The introduction of quantum dots into thecatalytic material allows for the conversion of infrared or nearinfraredlight into visible light that the semiconductor canabsorb. This expands the material’s light response range andsignificantly increases photocatalytic efficiency 90.

Tao et al. developed a composite material by depositingCNQDs onto nickel monoxide （NiO） microtubes via ahydrothermal reaction 91. The resulting NiO/g-C3N4 quantumdots exhibited a larger surface area and boosted light absorptionbecause of the anti-Stokes shift. As a result， this compositedemonstrated superior performance in photocatalytic CO2reduction， converting CO2 into CO and methane at significantlyhigher rates than pure NiO microtubes. Under simulatedsunlight， the optimized NiO/g-C3N4 quantum dot composite（NCNQD-3） achieved average production rates of 3.78μmol·g?1·h?1 for CO and 1.74 μmol·g?1·h?1 for methane， whichrepresented 2.1 and 4.2-fold increases， respectively. Thisimprovement in performance is largely ascribed to the upconversioneffect of CNQDs， which enhances light utilizationacross a broader spectrum. Thus， CNQDs offer remarkablepotential for enhancing the light absorption and catalytic activityof g-C3N4， making them a promising option for variousphotocatalytic applications.

3.1.2 g-C3N4 with 1D structure

g-C3N4 with a one-dimensional （1D） morphology offersdistinct advantages over its bulk counterpart， including a shortercharge diffusion distance and more exposed active sites.Common 1D g-C3N4 structures include nanotubes 92， nanofibers 93and nanorods 94. As the axial dimension increases， the material’slight absorption capability improves while lowering the highestoccupied molecular orbital （HOMO） position. This leads to anoptimal adjustment of the VB position， ultimately enhancingoxygen generation activity. Among these structures， nanotubesstand out because of their unique hollow structure， offering alarge surface area， high aspect ratio， axial electron mobility， andquantum confinement effects 95.

Nanorods of g-C3N4 can be synthesized via a solvothermalreaction. Typically， cyanuric chloride and melamine are used asprecursors， with subcritical acetonitrile serving as a solvent. Thisprocess occurs at a temperature of 180 °C， significantly lowerthan the temperatures required for calcination 96. As shown inFig. 10a，b， the resulting g-C3N4 nanorods have a diameter of50–60 nm and a length extending to several microns. Foreffective crosslinking and assembly of precursor molecules， areaction time of 96 h is essential. Shorter reaction times mayresult in insufficient crosslinking， affecting the finalnanostructure. The use of a strongly polar solvent， such asacetonitrile， enhances the diffusion of reactant molecules，thereby accelerating the polymerization process and promotingcrystallization. Although the crystallinity of the resulting g-C3N4nanorods may not match that of calcined g-C3N4， their enhancedlight absorption improves their photocatalytic performance，underscoring the importance of optimizing reaction conditionsand solvent choices.

Wang et al. developed a method for fabricating porous g-C3N4nanotubes through the thermal treatment of melamine and urea，or a mixture of melamine-cyanuric acid in specific ratios 97.Interesting， the formation of supramolecular melamine-cyanuricacid complex occurs around 400 °C， serving as a precursor forthe growth of porous g-C3N4 nanotubes at higher temperatures，the preparation process is shown in the Fig. 10c. At 550 °C， thesenanorods are converted into porous g-C3N4 nanotubes with alength of 2–3 μm， a diameter of about 250 nm， and pore sizesranging from 5 to 30 nm （Fig. 10d，e）. The hydrogen productionrate of these nanotubes， synthesized with a melamine-to-ureamass ratio of 1 ： 10， reaches 1073.6 μmol·h?1·g?1. This is 4.7 and3.1 times higher than conventional g-C3N4 photocatalystsprepared via direct polymerization of melamine and urea，respectively.

Porous g-C3N4 nanotubes express prodigious potential inphotocatalysis including photocatalytic hydrogen production，CO2 reduction， dye-sensitized solar cells， andphotoelectrochemical sensors. For instance， Li et al. preparedone-dimensional g-C3N4 nanotubes via supramolecularself-assembly and demonstrated their superior performance inCO2 reduction under visible light 98. These nanotubes showed aphotocatalytic CO2-to-CO conversion rate of 12.58μmol·g?1·h?1， which was 3.37 times higher than that of originalg-C3N4. The enhanced performance can be accredited to thepresence of oxygen and nitrogen defects in the tubular structure，which generate additional active sites and improve theseparation and relocation of photogenerated carriers.

Cao et al. further developed a heterojunction system usingCo3O4 nanoparticles supported on tubular g-C3N4 （TCN）structures using hydrothermal pretreatment and calcination 99.The Co3O4 nanoparticles were homogeneously disseminated onthe TCN microtubes， exposing more active Co sites andsignificantly enhancing the photocatalytic CO2 reductionperformance. A p-n junction between the p-type Co3O4 and ntypeg-C3N4 further facilitated the transfer of excited carriers.This composite achieved a turnover number （TON） of 24.72，which was 24 times higher than pure Co3O4， demonstratingsuperior photocatalytic performance.

Overall， the use of 1D g-C3N4 structures， such as nanorods andnanotubes， provides valuable opportunities for improvingphotocatalytic efficiency， especially in applications involvingCO2 reduction and hydrogen production. Their unique structuraland physicochemical properties， combined with advancedfabrication techniques， offer significant potential for optimizinghybrid photocatalysis systems.

3.1.3 g-C3N4 with 2D structure

The two-dimensional （2D） form of g-C3N4 is fundamentallydifferent from its bulk counterpart， with nanosheets being themost common morphology. In recent decades， there has been anexponential increase in reports on the preparation of variousnanostructured materials and their potential applications inphotocatalysis 100，101. Two-dimensional structures have gainedsignificant attention due to several key advantages： （I） 2Dnanostructures typically possess a larger specific surface area（SSA）， providing more active sites for reactions; （II） they exhibitlonger lifetimes for photogenerated carriers due to surfacedefects and pronounced non-stoichiometry; and （III） it isrelatively straightforward to control the surface characteristics ofthese nanostructures， allowing for optimized exposure of activesites.

Despite the promising advantages of 2D g-C3N4 nanosheets，their broader application remains limited due to the challengesassociated with material preparation. Developing a green，efficient， simple， and universal method for fabricating ultra-thin2D g-C3N4 remains an urgent priority. Inspired by the synthesisof graphene-based materials， researchers have adopted severalsimilar methods to produce other ultra-thin 2D materials 102. Themost prominent techniques include mechanical exfoliation 103，the Hummers method 104， and liquid-phase exfoliation 105. Eachof these methods produces g-C3N4 with varying thicknesses.

For example， Xu et al. developed a chemical exfoliationmethod for obtaining single-layer g-C3N4 nanosheets by simplechemical stripping 106. In this method， bulk g-C3N4 was mixedwith H2SO4 （98 wt%）， allowing the acid to intercalate betweeng-C3N4 layers （Fig. 11a）. As shown in Fig. 11b，c， atomic forcemicroscopy （AFM） and TEM images revealed that the flakesproduced range from tens of nanometers to several microns insize. The average thickness， as measured by AFM， wasapproximately 0.4 nm， close to the theoretical thickness of asingle-layer g-C3N4 sheet （about 0.325 nm）. The darker regionsin the TEM images can be accredited to overlapping g-C3N4nanoplates or multilayered structures.

Similarly， Ji et al. developed a simple mechanical grindingmethod， in which aromatic molecules were introduced during thegrinding process of bulk g-C3N4 （Fig. 11d） 107. This techniqueuses shear forces arising from π–π stacking interactions toexfoliate and simultaneously modify g-C3N4. This approacheffectively addresses the challenges of poor dispersion and theprocessing difficulty of g-C3N4 in most solvents. The uniquefeatures of this 2D structure， such as its larger specific surfacearea and optimized pore size， enhance CO2 adsorption，significantly improving the photocatalytic activity for CO2reduction.

3.1.4 g-C3N4 with 3D structure

In comparison to the previously discussed morphologies， thethree-dimensional （3D） nanostructures of g-C3N4， includingporous structures， spheres， aerogels， foams， and core-shellspheres， offer several distinct advantages. These 3Darchitectures demonstrate a high specific surface area， multiplereactive sites， and enhanced light absorption capacity， whichcollectively prevent the agglomeration of nano-units andmaximize the utilization of incident light via multiple lightscattering effects. These features make 3D g-C3N4 structureshighly suitable for photocatalysis， as they augment the overallproficiency of the process.

3D g-C3N4 structures are typically fabricated by treating bulkg-C3N4 through methods such as mixing， evaporation，mechanical stirring， and hydrothermal treatment. For instance，researchers have successfully synthesized macroscopic 3D g-C3N4 with enhanced photocatalytic hydrogen production usingmelamine and urea compounds as precursors. Additionally，photocatalysts with hollow spherical structures are particularlyeffective at capturing more photons due to continuous multiplelight reflections within their cavities. This phenomenonincreases the number of photogenerated electron-hole pairsavailable for photocatalytic reactions， enhancing overallperformance.

Traditionally， the synthesis of g-C3N4 with hollow spheres hasbeen challenging due to the structural instability of g-C3N4during synthesis. However， recent advancements have allowedfor the effective preparation of g-C3N4 hollow spheres， whichgrasp inordinate potential for future applications in addressingenvironmental challenges and meeting societal energy demands.Several synthesis techniques are used to obtain 3D g-C3N4structures， including sol-gel methods， chemical vapordeposition， soft and hard template methods， and template-freeapproaches.

For example， Sun et al. developed hollow g-C3N4 spheres（HCNS） using core-shell structured materials. In this process， athin layer of mesoporous silica is coated onto monodispersedsilica nanoparticles to form a template （Fig. 12a，b） 108.Cyanamide （CY） molecules are injected into the mesoporoussilica shell， and subsequent calcination and polycondensation，followed by removing the silica template， yield uniform g-C3N4hollow nanospheres. By adjusting the thickness of themesoporous silica shell， the wall thickness of the g-C3N4 hollownanospheres can be precisely controlled， ranging from 56 to 85nm.

Liu et al. employed cyanuric acid-melamine supramolecularliquid and ionic liquid as precursors and templates to synthesizethree-dimensional porous g-C3N4 with ultra-thin nanosheets（Fig. 12c） 109. Lin et al. designed a highly efficient threedimensionalordered macroporous （3DOM） g-C3N4 using asimple thermal polycondensation method， with the assistance ofa three-dimensional ordered mesoporous core-shell silicatemplate （Fig. 12d） 110.

Recent advances in supramolecular chemistry have enabledthe successful preparation of hollow g-C3N4 structures usingtriazine molecules. In the initial stages， a network structure isformed through hydrogen bonding between different triazineprecursors 111–113. The supramolecular network exhibits variousmorphologies depending on the solvent used： a 3D assembly orflower-like structure in dimethyl sulfoxide （DMSO）， and anordered pie-like structure in alcohol. During subsequentcalcination， these network structures may retain some of theiroriginal morphological traits， leading to the formation of 3D g-C3N4 structures， such as 3D hollow assemblies， mesoporoushollow spheres， and hollow boxes.

These advancements in 3D g-C3N4 design mark a significantstep forward in developing efficient photocatalysts capable ofaddressing contemporary energy and environmental challenges.

3.2 Doping strategy

g-C3N4 is composed of 2D sheets held together by covalentbonds in-plane and weak van der Waals interaction betweenlayers. This unique structure enables the incorporation of foreignatoms， both metallic and non-metallic， into the g-C3N4 lattice.The introduction of dopants into g-C3N4 can create lattice defectsthat inhibit the recombination of excitons， improvingphotocatalytic performance. In this section， we explore howdopant atomic orbitals influence the original molecular orbitalsof g-C3N4 leading to changes in its electronic structure. Thesemodifications can affect the VB and CB energy levels， therebyenhancing the separation of photogenerated charge carriers andimproving light absorption. We will focus on two primary typesof doping-non-metallic and metallic-each with distinct effects onthe electronic properties of g-C3N4.

3.2.1 Non-metal doped g-C3N4

The incorporation of non-metallic dopants into g-C3N4 canextend its light absorption range and improve its redox potentialby regulating the band gap， leading to enhanced photocatalyticperformance. Non-metal heteroatoms typically replace carbon ornitrogen atoms in the lattice， while metal atoms becomeintegrated into the framework. Doping reduces the band gap ofg-C3N4， thereby extending its light absorption into longerwavelengths.

For example， Wang et al. synthesized S-doped g-C3N4 usingsulfur and thiourea as starting materials 114. This modifiedmaterial exhibited significantly enhanced CO2 conversion tomethanol （CH3OH） compared to unmodified g-C3N4. Densityfunctional theory （DFT） calculations revealed that S dopinginduced impurity states within the CB of g-C3N4， leading toelectron spin polarization， band gap reduction， and extendedlight absorption.

Arumugam et al. fabricated g-C3N4 doped with various nonmetallicelements （B， O， P， and S） through solid-statepolycondensation using urea as a precursor 115. S-doped g-C3N4demonstrated the highest efficiency in converting CO2 tomethane （CH4）， outperforming B， O， and P doped samples aswell as undoped g-C3N4. The superior performance of S-dopedg-C3N4 was attributed to the incorporation of S atoms， whichsuppressed the swift recombination of photogenerated electronholepairs.

Fu et al. produced oxygen-doped g-C3N4 （PC3N4-S or PCNS）porous nanotubes by thermal oxidation， continuouslystripping and condensing g-C3N4 116. The inclusion of oxygenatoms reduced the band gap from 2.75 to 2.61 eV， improving thecharge transfer rate at the interfaces and promoting theseparation of electron-hole pairs， finally enhancingphotocatalytic activity.

Qiao et al. synthesized porous P-doped g-C3N4 nanosheets（PCN-S） via phosphorus doping and a thermal exfoliationmethod， using 2-aminoethylphosphonic acid （AEP） andmelamine as precursors 117. The synthesis process is shown inFig. 13a. In this method， melamine （ME） and AEP weredissolved in water， forming an ME-AEP complex through acidbaseand van der Waals interactions. The final product， PCN-S，was obtained through thermal polycondensation and exfoliation.

The resulting PCN-S nanosheets display a macroporousstructure with transverse dimensions of several microns， asrevealed by high-resolution TEM images （Fig. 13b）. Theseimages also showed uniformly dispersed macropores， rangingfrom 90 to 300 nm in size on the surface of the nanosheets. Thispore distribution is attributed to the strong interaction betweenmelamine and AEP. Additionally， AFM images （inset in Fig.13b） inveterate the occurrence of macropores on the surface，with a thickness of about 5–8 nm. The porous structure of thesenanosheets increases their surface area， providing more activesites and enhancing the diffusion of reactants and products. DFTcalculations （Fig. 13c） further revealed that the C1 position inthe heptazine unit was more energetically favorable forphosphorus doping than the C2 position. The incorporation ofphosphorus atoms at the C1 position led to a slight expansion ofthe carbon nitride pores， from 6.95 to 7.04 ?， improving thematerial’s photocatalytic efficiency.

Previous research has shown that exfoliated g-C3N4 nanosheetsoffer several advantages over their bulk counterparts 118， includingenhanced light absorption and photocatalytic activity. However，a key limitation of thermally exfoliated g-C3N4 is the expandedband gap， which reduces its ability to absorb a broad spectrumof solar light. This issue can be addressed through P-doping，which reduces the band gap and enhances tail absorption，making P-doped g-C3N4 more efficient. As shown in Fig. 13d，doping non-metal heteroatoms can create mid-gap levels withinthe band gap of g-C3N4， which lie below the CB minimum dueto the interaction of C 2s2p， N 2s2p， and P 3s3p. These mid-gapstates retain more solar light， increase the production of carriers，and improve overall photocatalytic performance.

3.2.2 Metals doped g-C3N4

Metallic dopants， including transition and noble metals， havebeen extensively used to adjust g-C3N4. These dopants areknown to reduce the band gap of g-C3N4， thereby broadening itslight absorption range and enhancing its photocatalytic activity.

Zhang et al. investigated the effects of alkali metal doping ong-C3N4 through first-principles calculations and experimentalanalysis 119. Their findings revealed that Rb-doped g-C3N4demonstrated enhanced light absorption in the 500 to 1100 nmrange， which was consistent with UV-Vis analysis. Thus， thephotocatalytic reduction of CO2 was significantly improved，with performance levels three times higher than those of pureg-C3N4. This study highlights the potential of Rb-doped g-C3N4as a highly efficient catalyst for CO2 conversion， offering newpossibilities for sustainable energy production andenvironmental remediation.

Tang et al. developed a novel Co-doped material byincorporating cobalt （Co） atoms into mesoporous g-C3N4（CoCN） 120. Unlike conventional doping methods， whichtypically replace carbon or nitrogen atoms with Co， this studyintroduced Co atoms into the interstitial spaces surrounded bythree s-triazine units within the g-C3N4 framework （Fig. 14a）. Asshown in Fig. 14b， the positions of CB and VB of g-C3N4 shiftdownward with increasing Co content. This modificationredistributes the molecular orbitals of g-C3N4， enhancing its lightabsorption in the 450–800 nm range and accelerating chargetransfer. The results demonstrate that Co-doped photocatalystshave significant potential for improving CO2 reductionefficiency by boosting light absorption and promoting electrontransfer.

Wang et al. prepared copper-modified g-C3N4 nanorodbundles （CCNBs） using a chemical vapor deposition （CVD）method， with urea and sodium chlorophyllin as precursors （Fig.14c） 121. The introduction of uniformly dispersed copper into theg-C3N4 framework not only enhanced visible light absorptionbut also increased the number of active sites， promoting thetransfer of charge carriers （Fig. 14d）. Thus， the CCNBs exhibiteda remarkable CO production rate of 9.9 μmol·g?1·h?1， about fivetimes higher than that of pure g-C3N4.

3.3 Monoatomic modification

The introduction of appropriate co-catalysts can significantlyenhance the adsorption and activation of CO2， and facilitate theseparation and transport of exciton carrier pairs， therebypromoting surface redox reactions. While precious metal-basedco-catalysts such as platinum （Pt） 122，123， palladium （Pd） 124， andsilver （Ag） 125 have demonstrated remarkable efficacy inenhancing photocatalytic CO2 reduction， their high cost andlimited availability have prompted the exploration of alternativemodification strategies. These strategies include optimizing sizeand morphology， controlling composition， and fabricatingnanostructured particles 126.

Recently， single-atom catalysts （SACs） have garneredattention due to their efficient utilization of metal atoms as activecenters and their strong interaction with substrates 127，128. SACshave shown significant promise in various catalytic reactions，including photocatalytic CO2 reduction. Numerous studies haveexplored the application of SACs in modifying g-C3N4， resultingin notable improvements in photocatalytic performance.

Li et al. investigated the impact of different forms ofpalladium， including single-atom Pd （Pd-SA）， Pd oxides （PdOx），and Pd nanosheets （Pd-NP）， on the CO2 reduction activity ofg-C3N4 129. Their photocatalytic tests revealed that while allforms of Pd enhanced CH4 formation， the Pd-SA executedlargest activity with a CH4 production rate of 2.25 μmol·g?1，outperforming PdOx/CN （1.08 μmol·g?1）， Pd-NP/CN （0.44μmol·g?1）， and CN （0.104 μmol·g?1）. This superior performanceof Pd-SA is attributed to its favorable CO2 adsorption andactivation， reduced conduction potential， and efficient ·Hoperational capability.

Shi et al. successfully incorporated Pt atoms onto defective g-C3N4 （Pt@Def-CN） 130 （Fig 15a）， resulting in a photocatalystwith enhanced CO2 reduction activity and CH4 selectivitycompared to unmodified carbon nitride. Pt@Def-CN exhibited aremarkable CH4 selectivity of 99%， as illustrated in Fig. 15b.This high selectivity is attributed to the unique characteristics ofPt single atoms， which preferentially bind to -OH groups due totheir instability when binding to H atoms. This interactioninhibits H2 production and provides additional H+ for CH4formation. Additionally， Pt single atoms not only lower theactivation energy barrier for CH4 formation but also increase theformation energy of the CO* intermediate， thereby improvingCH4 selectivity （Fig. 15c）.

Li et al. employed a molten salt and reflux method to fabricatehigh-crystalline g-C3N4 doped with copper atoms （Cu-CCN） 131.The incorporation of Cu atoms enhances CO2 activation，significantly improving CO2 adsorption capacity and overallreaction activity （Fig. 15d）.

3.4 Defect engineering

Defect engineering in g-C3N4 has emerged as a crucialstrategy for enhancing photocatalytic performance by extendingthe light absorption range and prolonging the lifetime of carriers.The presence of defects in g-C3N4 is essential for improvingphotocatalytic reactions 132，133， as these defects can serve asactive sites for catalytic processes and prevent the rapidrecombination of carrier pairs. The inherent defects in g-C3N4include void defects （such as amino defects 134， cyano defects 135，carbon vacancies 136， and nitrogen vacancies 137） and anti-sitedefects （such as carbon 138 and nitrogen 139 self-doping）.

Defect engineering typically involves the selective removal ofspecific atoms from the g-C3N4 framework to create a defectrichstructure. This process generates unsaturated coordinationsites， which enhances CO2 activation. Defects not only facilitatethe separation of charge carriers but also act as active sites forCO2 adsorption， ultimately improving the overall performanceof photocatalytic CO2 reduction 140. For example， g-C3N4 withabundant carbon vacancies demonstrates meaningfully boostedCO2 reduction activity compared to its pristine form 141.Additionally， surface defects in g-C3N4 improve its capacity forwater adsorption and dissociation， thereby increasing the supplyof protons for CO2 reduction and improving the selectivity forhydrocarbon production. Similarly， introducing nitrogenvacancies into g-C3N4 enhances its CO2 photoreductionefficiency by creating defect energy levels within the band gap，reducing the band gap width， and improving light-harvestingproperties 142. These defect energy levels show a key role inenlarging charge carrier separation and transfer processes.

Hou et al. developed a folded structure of g-C3N4 withabundant nitrogen-vacancy defects through an ethanolcontainingvapor process （Fig. 16a） 143. This method allows forprecise tuning of the energy gap of g-C3N4 ranging from 1.82 to2.48 eV （Fig. 16b）. The optimized catalyst， denoted as gCN-10，showed an enhanced photocatalytic CO2 reduction activity（226.1 μmol·g?1）， which is 28.6 times higher than pristine g-C3N4 （7.9 μmol·g?1）. The arrival of carbon and nitrogen-vacancydefects effectively alters the charge density distribution，improving the separation of reduction and oxidation sites （Fig.16c）.

Liu et al. synthesized hollow microtubular g-C3N4 （TCN） withcontrollable nitrogen-vacancy levels （TCN-1） using acombination of hydrothermal self-assembly and hydrogen heattreatment （Fig. 16d） 60. While the band gaps of TCN and TCN-1widen， the redox ability increases （Fig. 16e）. TCN-1demonstrated a photocatalytic activity （7.06 μmol·g?1·h?1） thatwas 2.2 times and 8.8 times greater than that of TCN （3.12μmol·g?1·h?1） and pristine g-C3N4 （0.75 μmol·g?1·h?1），respectively. Studies of mechanisms revealed that nitrogenvacancies induce structural disorder and enhance carrier-pairseparation by acting as electron-trapping states. Additionally， theintroduction of nitrogen vacancies improves CO2 adsorptioncapacity， which is advantageous for photocatalytic CO2reduction.

3.5 Heterostructures

By forming heterojunctions with other semiconductors， issuessuch as low separation efficiency of photogenerated carriers，limited light absorption， and insufficient active sites in g-C3N4can be effectively addressed 144. Various types ofheterojunctions， such as type I， type II， type III， and S-schemeheterojunctions， have been developed and extensively studied inthe context of photocatalytic CO2 conversion involvingg-C3N4 145.

3.5.1 Conventional heterojunction types

The rapid recombination of carriers and the low redox abilityof single semiconductor often result in relatively low overallphotocatalytic efficiency. To address these challenges， differentheterojunction types have been developed. A heterojunction is aninterface structure formed by the contact of two photocatalystswith different band gap positions. Conventional heterojunctionsare typically categorized into three types 146，147： straddle type（Type-I）， stagger type （Type-II）， and broken gap type （Type-III）（Fig. 17）.

In a Type-I heterojunction （Fig. 17a）， both electrons and holesare confined within a single semiconductor. This confinementhampers the efficient separation of electron-hole pairs， limitingits photocatalytic effectiveness 148. In a Type-II heterojunction（Fig. 17b）， the specific energy band structure facilitates thespatial separation of photogenerated charges. Upon lightirradiation， excited electrons migrate from the CB of onesemiconductor （PC I） to the CB of another semiconductor （PCII）， while holes move from the VB of PC II to the VB of PC I.This allows more effective utilization of photogenerated carrierpairs. The Type-III heterojunction （Fig. 17c） features band gapsof each semiconductor that do not overlap， resulting in noeffective transfer or separation of excitons between the twosemiconductors. Consequently， Type-III heterojunctions are notsuitable for improving charge separation.

Among these types， only Type-II heterojunctions facilitateactive interfacial transfer and spatial separation of excitons.However， the enhanced separation of photogenerated electronsand holes in Type-II heterojunctions comes at the expense ofreducing the oxidation ability of the photocatalysts involved（Fig. 18a） 149. The transfer of electrons from a higher CB to alower CB （at the reduction potential side） and the movement ofholes from a lower VB to a higher VB （at the oxidation potentialside） reduce the overall redox capability， which is critical fordriving photocatalytic reactions forward.

To overcome these limitations， the concept of the S-schemeheterojunction has been introduced （Fig. 18b）. This newapproach aims to optimize the charge transfer processes toenhance photocatalytic efficiency by balancing the redoxabilities of the involved photocatalysts.

3.5.2 S-scheme heterojunction

In 2019， Yu and colleagues introduced the concept of the stepscheme（S-scheme） heterojunction， which features a staggeredband structure comprising two semiconductors with differingband alignments （Figs. 18b and 19） 150. This heterojunctiontypically involves an oxidation-type semiconductor （OP）， suchas TiO2 or WO3， and a reduction-type semiconductor （RP）， suchas CdS or g-C3N4. In this arrangement， the Fermi energy level（Ef） of the RP is higher than that of the OP. Upon contact，electrons transfer from the RP to the OP until the Fermi energylevels at the interface reach equilibrium. This electron transferdepletes the electrons at the RP interface and enriches theelectrons at the OP interface， leading to a positive bending of theRP band edge and a negative bending of the OP band edge. As aresult， an interfacial electric field （IEF） is established from theRP to the OP.

Under light irradiation， electrons in both the RP and OP areexcited from VB to CB. The combined effects of band bending，Coulomb interactions， and the IEF result in the recombination ofelectrons from the OP with holes from the RP， while the VBholes of the OP and the CB electrons of the RP are preserved.This process effectively separates the charge carriers， retainingthose with high redox abilities for further redox reactions，thereby enhancing the efficiency of CO2 reduction 151–155.

3.5.2.1 Inorganic catalyst/g-C3N4 composite

A variety of inorganic photocatalysts， such as ZnO 156， WO3 157，and TiO2 158， have been combined with g-C3N4 to createS-scheme heterojunctions for CO2 reduction. Such as， Li et al.developed hexagonal g-C3N4 tubes modified with ZnIn2S4nanosheets （ZIS/HCNT） as shown in Fig. 20a 159. SEM images（Fig. 20b–d） confirm the successful fabrication of theZIS/HCNT composite. The charge migration mechanism for thiscomposite is showed in Fig. 20e. Under light irradiation， bothHCNT and ZIS are excited. Electrons in the CB of HCNTrecombine with holes in the VB of ZIS， resulting in effectiveseparation of photogenerated carrier pairs. This process retainselectrons in the CB of ZIS and holes in VB of HCNT with strongredox capabilities. Consequently， the CO2-to-CO conversion rateof ZIS/HCNT （883 μmol·h?1·g?1） is significantly enhanced，surpassing that of HCNT and ZIS by approximately 13 and 2.4times， respectively. As shown in Table 3， the g-C3N4-basedcomposites with S-scheme heterojunctions have beensummarized for photocatalytic reduction of CO2 in recentyears 160–177.

3.5.2.2 MOF/g-C3N4 composites

Metal-organic frameworks （MOFs） are microporous materialswith ultra-high specific surface areas and controllable threedimensionalhollow structures 178. Certain MOFs not onlyenhance efficient charge transfer for redox reactions but alsoexhibit a high capacity for CO2 adsorption 179，180. Consequently，the integration of MOFs with g-C3N4 offers numerous activesites and an enlarged reaction interface， thereby increasing theefficiency of CO2 conversion.

Zhao et al. designed an S-scheme heterojunctionphotocatalyst by incorporating a 3D Fe-MOF into a 2D thinlayerg-C3N4 （CN/Fe-MOF） through an in situ assemblyapproach （Fig. 21a） 181. Under light irradiation， thephotocatalytic redox process occurs as illustrated in Fig. 21b.This configuration retains electrons and holes with high redoxpotential， thereby enhancing the overall activity of thecomposite. Notably， even without the use of sacrificial agents orco-catalysts， the optimized CN/Fe-MOF composite achieved aCO yield that was 10 times higher than the pristine material.

3.5.2.3 COF/g-C3N4 composites

Covalent organic frameworks （COFs） are emerging materialswith widespread applications in catalysis， gas storage，optoelectronics， and lithium-ion batteries， among others. IminelinkedCOFs are particularly noteworthy due to their excellentCO2 adsorption and electron transport properties 182，183. Wang etal. constructed a 2D/2D S-scheme heterojunction photocatalystby mixing defective g-C3N4 nanosheets （C3N4（NH）） with Tp-TtaCOFs using an evaporation-induced self-assembly method （Fig.22a） 184. As depicted in Fig. 22b， the work function ofg-C3N4（NH） （4.425） is lower than that of the Tp-Tta COF（5.026）， leading to a higher Fermi level in g-C3N4（NH）. Whenthe two photocatalysts are in contact， the Fermi level differencedrives electron transfer from gCN （NH） to the Tp-Tta COF，resulting in band bending at the interface and the creation of aninterfacial electric field. Upon irradiation， both Tp-Tta COF andg-C3N4（NH） are excited， generating electron-hole pairs. Theelectric field facilitates the relocation of electrons from the CBof Tp-Tta COF with holes in the VB of g-C3N4（NH）， effectivelyenhancing the charge separation. Consequently， the S-schemeheterojunction demonstrates largely improved photocatalyticactivity. The CO production rate of the g-C3N4（NH）/Tp-Tta COFcomposite is 11.25 μmol·h?1， which is 45 times higher than thatof g-C3N4 and 15 times higher than the g-C3N4（NH）/COF.Moreover， the CO selectivity reaches 90.4%.

4 Conclusion

In photocatalysis， the discovery of g-C3N4 as a non-metallicn-type semiconductor material has revolutionized a welldirectedresearch route for the generation of energy andenvironmental purification. However， the research on g-C3N4 isstill in the early stages， and more research is needed to acquireits large-scale application. Additionally， the practical applicationis limited by the low specific surface area， inadequate visiblelight absorption， and rapid recombination of charge-pairs.Several methods have been employed to expand thephotocatalytic CO2 conversion applications of g-C3N4. However，the actual mechanism of CO2 reduction with gCN-basedphotocatalysts remains unclear， highlighting the ongoingchallenges in developing an effective and cost-efficient g-C3N4-based photocatalyst for large-scale commercial applications.

For an ideal photocatalyst， it is primarily required to utilize alarge portion of solar energy， high photon conversion capability，suitable CB and VB positions， selectivity of products， and longtermstability. Although g-C3N4 has shown excellentperformance in the utilization and conversion of solar energy，there are still many obstacles.

First， the highly crystalline g-C3N4 is effective in CO2reduction， but the existence of defects generated at highsynthesis temperatures may reduce the adsorption capability ofthe target substrate. Thus， it is urgent to develop new methodsfor the synthesis of g-C3N4 with highly crystalline structure andcontrollable defects， allowing for the efficient adsorption of thetargeting substrate and more effective separation ofphotogenerated carrier-pairs.

Second， the lack of consistent evaluation criteria makes itchallenging to estimate the efficiencies of variousphotocatalysts. Different light sources， reaction media，sacrificial reagents， and cocatalysts are used in differentexperiments. Therefore， a standard system is highly required toassess the photocatalytic performance of different g-C3N4 basedphotocatalysts.

Third， the selectivity of CO2 reduction over g-C3N4composites remains a big challenge. Further studies should focuson the selectivity of CO2 reduction.

Fourth， the reaction mechanism of CO2 reduction overg-C3N4-based materials is not clear. More advancedcharacterization techniques are needed to study the reactionmechanisms on the atomic or molecular level.

Author Contributions： Data Curation， X.W.; Investigation，S.D.; Writing， Original Draft Preparation， X.X.; Writing， Reviewamp; Editing， K.Q. and V. P.