發展二元熔鹽體系用于聚七嗪亞胺薄膜光陽極的制備與水氧化性能研究

關鍵詞：二元熔鹽；聚七嗪亞胺；離子熱制備法；水氧化；光陽極

1 Introduction

The evolution of dioxygen molecules through water oxidationreaction （WOR） has been a subject of sustained research interestrecently 1. Since its discovery， researchers have been intriguedby its potential applications in sustainable energy technologiesand its fundamental role in natural photosynthesis 2–6. Recently，metal-free semiconducting polymers have attracted increasingattention. Notable materials， including carbon nitride 7–9， carbondots 10， linear polymers 11–14， covalent organic frameworks 15，16，black phosphorus 17， conjugated microporous polymers 18， andothers， have demonstrated significant potential for driving theWOR.

The development of polymer-based photoanodes for the WORis a common strategy for utilizing metal-free materials in theoxygen evolution reaction （OER） 19–22. Carbon nitride stands outdue to its ease of synthesis， abundance of precursors， andstructural versatility 23，24. However， the thermodynamicchallenges during its polymerization process often result in a lowpolymerization degree and an amorphous phase 25，26. Thislimitation significantly impacts performance as the poorcrystallinity affects the lifetime of minority carriers 27.

Efforts to synthesize crystalline poly（heptazine imide） （PHI）photoanodes have led to notable improvements in photocurrentdensity 28，29. Typically， potassium thiocyanate （KSCN） acts as amediator salt， melting at high temperatures to address thethermodynamic challenges of the polymerization reaction，specifically ammonia evolution 30. For instance， sulfur fromKSCN reacts with tin on fluorine-doped tin oxide （FTO） to formSnS2 seeds， which facilitate PHI film growth. If excess tin in theFTO glass reacts with sulfur， the substrate becomes asemiconductor， resulting in a significant decrease inconductivity. Additionally， while potassium ions are usuallyincorporated into the PHI framework to improve its crystallinestructure， the molten salt can also degrade the substrateglasses 31. However， achieving individual control overcrystallinity and film quality simultaneously in a stoichiometricK+ and SCN? molten salt environment is challenging.

To elucidate the role of salt in the synthesis of the PHIphotoanode， a system employing binary molten salts wasdeveloped using K2CO3 and NH4SCN in various ratios.Compared to other salts， these two exhibit lower corrosivitytoward FTO substrate. Among the photoanodes， the bestphotocurrent density of 365 μA?cm?2 was attained with 0.3 g ofK2CO3 and 2.0 g of NH4SCN at a voltage bias of 1.23 V versusthe reversible hydrogen electrode （VRHE） under AM 1.5Gillumination， representing about 18 folds to the amorphous PCNphotoanode. In the synthesis strategy， NH4SCN promotes thegrowth of SnS2 as a seeding layer， while K2CO3 enhances thecrystallinity of the films. In situ electrochemicalcharacterizations reveal that this salt combination improvesphotoexcited charge transfer efficiency and minimizes theresistance within the SnS2 layer. This study elucidates the role ofsalts in the synthesis of the PHI photoanode and offers valuableinsights for the design of high-crystallinity carbon nitride-basedfunctional films.

2 Experiment

2.1 Materials

Melamine （analytic grade）， NH4SCN （analytic grade）， andK2CO3 （analytic grade） were procured from SinopharmChemical Reagent Co.， LTD. Fluorine-doped tin oxide glass（FTO， 10 mm × 20 mm × 2.2 mm， 7 Ω·sq?1） coated glass waspurchased from Wuhan Lattice Solar Technology Co.， LTD.

2.2 Synthesis of KCN and PCN photoanode

To prepare high-crystalline PHI photoanodes， 5 g of melaminewere placed into a 50 mL crucible and subjected to thermaltreatment in an air muffle furnace. The temperature was rampedat a rate of 2 °C ?min?1 until reaching 400 °C， where it was heldfor 1 h.

The resulting powder was labelled as M400. For eachprecursor， 0.4 g of M400， 2.0 g of NH4SCN， and varyingamounts of K2CO3 （0.1 g， 0.2 g， 0.3 g， and 0.4 g） were blendedaccordingly. The washed FTO conductive glass was placed at thebottom of a corundum boat， and the precursor powder wasspread evenly over it. The corundum boat was annealed to500 °C at a rate of 2 °C?min?1 for 4 h under a nitrogenatmosphere.

The resulting photoanodes were named based on the amountof K2CO3 as precursor： 1KCN， 2KCN， 3KCN， and 4KCN，representing 0.1 g， 0.2 g， 0.3 g， and 0.4 g of K2CO3， respectively.

Similarly， the preparation process for the PCN carbon nitridephotoanode used 0.4 g of M400 and 2.0 g of NH4SCN asprecursors， following the same procedure as for the KCNsamples.

2.3 Characterizations

Scanning electron microscope （SEM， JSM-6700F， Japan） andtransmission electron microscopy （TEM， FEI Talos， USA） wereapplied to study the structure and morphology of photoanodes.X-ray diffraction （XRD， Miniflex 600 diffractometer， Japan）was applied to study the crystallinity. Fourier transform infrared（FTIR， Nicolet 670， USA）， Raman data （HORIBA LabRAM HREvolution system， Japan， laser source 325 nm）， and X-rayphotoelectron spectroscopy （XPS， Thermo Esca lab 250， USA）were applied to study the chemical properties of thephotoanodes. Ultraviolet-visible diffusion reflector spectroscopy（UV-Vis DRS， Varian Cary 500 Scan， USA） was applied to studythe bandgap of the photoanodes.

Transient room-temperature photoluminescence （PL，HORIBA Fluorolog-3） and time-resolved PL （TRPL） wereconducted by a laser source with the wavelength of 325 nm. Theaverage lifetime of excitons can be calculated from PL spectrumby Eq. （1）：

Photoelectrochemical （PEC， BioLogic VSP-300， France）system contains a photoanode as working photoanodes and a Ptfoil as the counter electrode. A Ag/AgCl electrode was appliedas reference electrode. The reaction was performed in theelectrolyte solution of NaOH （1.0 mol?L?1） with illuminatingAM 1.5G （solar simulator， Newport， USA）. Nyquist plots andMott-Schottky test were carried in electrochemical impedancespectroscopy （EIS） system in the electrolyte solution of Na2SO4（0.2 mol?L?1）. The related flat band potential was calculated bythe Mott-Schottky equation （Eq. （2）） as following 32：

where Csc stands as the space charge region capacitance， e standsas the elementary charge （1.602 × 10?19 coulombs）， ε0 is thepermittivity number of vacuum （ε0 = 8.854 × 10?14 F?cm?1）， εstands as the dielectric constant of PCN （ε ≈ 8）， A stands as theelectrochemically active surface area. V is the applied voltage incircuit and Vfb is the flat-band potential which close toconduction band minimum. Nd is the donor density in thephotoanode.

A Xenon （300 W） with a monochromator （Newport， USA）was used to value the incident light-to-electron conversionefficiency （IPCE）， and Eq. （3） was used for the calculation asfollows：

where the jp（λ） is the photocurrent density under the appliedvoltage in circuit is 1.23 VRHE in a certain wavelength（mA?cm?2）， and the Eλ（λ） is the strength of every wavelengthfrom the incident light source （mW?cm?2）.

Electrochemically active surface areas （ECSA） were appliedto show the nonlinear changes in PEC properties 33. Thephotoanodes electrochemical double-layer capacitance （Cdl） waslinearly dependent on the ECSA. The difference of currentdensity （ΔJ） was conducted by using the equation of ΔJ = Ja ?Jc， where Ja is the anodic current density and Jc is the cathodiccurrent density. The linear slope is linearly dependent on theECSA. In our work， ECSA was conducted by a three-electrodesystem， and he applied voltage ranging from 0.95–1.0 VRHE within the electrolyte solution of NaOH （1 mol?L?1）.

3 Results and discussion

The synthesis of KCN from melamine is illustrated in Fig. 1a.Thermal treatment of melamine yields Melem， and KCN issynthesized ionothermally from Melem using NH4SCN andK2CO3 as mediators 34. The digital photographs of thephotoanodes were shown in Fig. S1. 2.0 g of NH4SCN wasapplied as standard， and K2CO3 was regulated for optimization.SEM images in Fig. 1b，c present the top-view and crosssectionalview of KCN films， respectively， revealing ananosheet-like morphology uniformly anchored on the FTOlayer 35. Noted that blending an increased amount of salt into theprecursor causes the resulting KCN films to become rigid andresemble nanosheets （Fig. S2）. The thickness of the 3KCN filmswas determined to be ca. 1.5 μm， as shown in Fig. 1c. In contrast，the bulky structure of amorphous PCN films can be observed（Fig. S3a）.

Furthermore， TEM and EDS analyses were conducted toinvestigate the structure of 3KCN. The EDS images in Fig. 1e?ireveal the presence of elements C， N， S， and K， respectively.Their homogeneous distributions indicate successful synthesisof highly crystallized 3KCN 36. The TEM image in Fig. 1g showsdistinct lattice fringes with 1.08 nm and 0.32 nm， which is owingto crystal facets of PHI， specifically the [100] and [002]planes 37， respectively. Comparatively， in Fig. S4， it is evidentthat PCN exhibits lower crystallinity.

XRD was carried out to investigate the crystallinity of PCNand KCN photoanodes （Fig. 2a，b）. Most of the XRD peaks areattributed to FTO films （JCPDS No. 46-1088）. A typical peak at27.5° can be ascribed to the interlayer stacking of graphene-likecarbon nitride layers 38. By increasing the quantity of K2CO3 inthe molten salt， the diffraction angle shifts to 28.1°， indicating asharper peak. This result suggests that by increasing theconcentration of K ion， the enhanced π–π stacking andintermolecular forces were achieved within these photoanodes，resulting in shorter interlayer distances and a more orderlystructure between the layers 39. Meanwhile， it also affects thecondensation process in the in-plane direction. As shown in Fig.2b， the increase in K2CO3 also shifts the peak at 13° to 8°，indicating further crystallization of Melem units into heptazinestructure 40，41. Generally， charge transfer through the interlayeris crucial for determining the performance of photocatalyticreactions 42. Therefore， enhancing intermolecular forces canfacilitate photoexcited charge transfer.

FTIR was applied to study the photoanodes， and the resultswere presented in Fig. 2c. The bending mode of heptazine unitin PCN and KCN， were assigned to the ranging of thewavenumbers from 1200 to 1700 cm?1 43. Furthermore， a distinctpeak at ca. 2177 cm?1 was presented in the KCN spectrum，representing the stretching vibrations of terminal cyanogroups 44，45. Raman spectroscopy was also utilized to analyzethe photoanode （Fig. 2d）. The peaks at 707 and 765 cm?1 wereattributed to the in-plane C―N＝C bending vibrations. Thebroad peaks spanning 1100–1800 cm?1 were owing to thestretching vibrations of C―N rings 46. Additionally， thepresence of the cyano group was confirmed by the peak observedat 2180 cm?1. These findings elucidate the structure of KCN andwere consistent with the XRD patterns.

XPS measurement was carried out to study the photoanodes.Fig. 3a displays the high-resolution C1s spectra， which was fittedwith three peaks at 284.8， 286.3， and 288.1 eV. The peaks areowing the C＝C， C―S， and C―N＝C groups 47， respectively.Additionally， KCN exhibited two distinct peaks at bindingenergies of 292.9 and 295.6 eV in the high-resolution K 2pspectra， which were attributed to K 2p3/2 and K 2p1/2，respectively.

In the N1s spectra （Fig. 3b）， four peaks were assigned at398.6， 400.2， 401.1， and 404.0 eV. Among them， the peaks at398.6 and 400.2 eV were owing to the C―N＝C and N―（C）3groups， respectively 48. These results highlight the molecularstructures of PCN and KCN， which were consistent withfindings from FTIR and Raman spectra. In the S 2p XPS spectra（Fig. 3c）， the peak at 163.7 and 162.5 eV corresponds to theS―C group 49. While peaks at 162.4 and 161.2 eV werecorrespond to S―Sn interactions， the peaks at 169.7， and 168.5eV were due to sulfur oxides. In the Sn 3d XPS spectra asexhibited in Fig. 3d， two peaks at 494.9 and 486.6 eV in KCNwere characteristic of S―Sn interactions. Notably， an increasein salt quantity caused a shift in the Sn 3d peak towards higherenergies， indicating enhanced chemical interaction between Snand the PHI films. However， excessive reaction of Sn withcarbon nitride films can increase resistance.

Subsequently， PCN and KSCN photoanodes wereinvestigated by UV-DRS （Fig. S5a） and the correspondingKubelka-Munk plots were conducted （Fig. S5b?f）. Both PCNand KCN photoanodes exhibited ability for harvesting visiblelight. The conduction band level has been calculated by flat bandpotential in the Mott-Schottky test （Fig. S6） and Eq. （2）.Accompanied by the band gap results， the band structure of eachphotoanode has been illustrated in Table S1 50. Theoretically， themigration of photoexcited charges on the photoanode can thusbe illustrated in Fig. S7.

The PEC WOR was conducted using a three-electrode system.Linear sweep voltammetry （LSV） was performed across avoltage bias range of 0.5 to 1.5 VRHE under illumination of AM1.5G （Fig. S8）. The photocurrent density is a critical indicator ofthe efficiency of the WOR， and 3KCN demonstrated superiorperformance compared to other photoanodes. Fig. 4a shows thechronoamperometry （CA） curve collected at applied potential of1.23 VRHE. Consistent with the LSV findings， the 3KCNphotoanode exhibited the highest photocurrent density with thevalue of ca. 365 μA?cm–2 at 1.23 VRHE. This represents anapproximately 18 folds increase compared to the pristine PCNphotoanode， and this performance is comparable to the recentreported work （Table S2）， highlighting the significant impact ofimproved crystallinity on carbon nitride materials. In addition，the hydrogen evolution ability was tested （Fig. S9）， with about50 μA?cm?2 photocurrent density in 3KCN sample at ?0.7 VRHE.

The IPCE test was measured under the same conditions as theCA experiment but with different incident wavelengths. Theresults are shown in Fig. 4b， which closely correlate with theoptical properties （Fig. S5a）， indicating absorption edges around550 nm 51. Moreover， the IPCE results for 3KCN generallyexceed those of the other samples， with the highest IPCE valuereaching approximately 18% when illuminated at a wavelengthof 360 nm.

The enhanced performance was further investigated throughPL analysis of charge transfer in PCN and KCN films （Fig. S10）.When excited by 360 nm light， KCN films predominantly emitlight around 460 nm， closely matching UV-DRS results （Fig.S5a）. TRPL spectra of the photoanodes illuminated at 360 nmwere also recorded， as shown in Fig. 5a. The reduced lifetimesindicate faster transfer rates for photoexcited charges.Specifically， lifetimes were 1.69， 1.15， 0.85， 0.55 and 0.52 ns forPCN， 1KCN， 2KCN， 3KCN and 4KCN， respectively. Generally，enhancing the crystallinity of the photoanode led to shorterlifetimes， demonstrating significant improvements in chargetransfer efficiency compared to PCN 52–54.

The ECSA was performed by a double layer capacitancemethod as shown in Fig. 5b and Fig. S11 55，56. The ECSA valueis associated with the charge transfer. The improved active siteon KCN surface is mainly due to the improved crystal structure.In addition， the contact angle of water of each photoanode hasbeen achieved in Fig. S12， the low angle of KCN photoanodesexhibit better absorption of water， also indicating the improvedphysicochemical interaction between the surface of PHI andwater molecules.

The charge carrier density was further investigated using EISand analyzed through Mott-Schottky plots （Fig. S13a）.According to the Eq. （2）， charge carrier density of 3KCNphotoanode was estimated to be about 1.5 × 1015 cm?3， higherthan that of the general amorphous carbon nitride films basedphotoanodes. The circuit photovoltage decay test was conductedto estimate the surface charge concentration （Fig. S13b） 57. The3KCN sample shows a noticeably longer relaxation time，indicating longer lifetimes of photoexcited charges 58. Nyquistplots of EIS have been shown in Fig. 5c under both theconditions of with and without AM 1.5G illumination （Fig. S14）.The fitted data were presented in Table S3. The charge transferresistances （Rct） of PCN and KCN films in Fig. 5d decreasedunder AM 1.5G illumination， revealing the photo-response fromthe PCN photoanode. The lower Rct of KCNs （Fig. 5d） wereattributed to high conductivity of crystal polymer with a moreordered in-plane structure for charge transfer compared to PCNfilms 56，59，60. However， an increase in the precursor salt led to ahigher Rct， which can be attributed to the formation of SnS? onthe surface， causing erosion of the FTO. Therefore， 4KCNexhibited the largest Rct among KCN samples， and reducedphotocurrent density.

4 Conclusions

In summary， an ionothermal approach using binary salts，specifically K2CO3 and NH4SCN， was employed to synthesizePHI films on FTO glass as photoanodes. The salt quantities wereadjusted to optimize photocurrent density in WOR， with thehighest photocurrent density achieved using 0.3 g of K2CO3 and2.0 g of NH4SCN. This optimal value of 365 μA?cm?2 at 1.23VRHE under illumination of AM 1.5G， which is ca. 18 folds to thepristine PCN photoanode. This improvement was mainlyattributed to the promoted transfer of photoexcited charges. Thisstudy paves the way for understanding the role of salt mediatorsin synthesizing the PHI photoanode， demonstrating the potentialfor designing semiconducting devices with crystallized carbonnitrides.

Author Contribution： Conceptualization， Yuanxing Fang;Methodology， Jiaxin Su; Software， Jiaxin Su; Validation， JiaqiZhang， Shuming Chai and Yankun Wang; Formal Analysis，Jiaxin Su and Yuanxing Fang; Investigation， Jiaxin Su andYuanxing Fang; Resources， Yuanxing Fang; Data Curation，Jiaxin Su， Shuming Chai and Jiaqi Zhang; Writing-OriginalDraft Preparation， Jiaxin Su; Writing-Review amp; Editing， JiaxinSu， Yuanxing Fang and Sibo Wang; Visualization， Jiaxin Su andJiaqi Zhang; Supervision， Yuanxing Fang; ProjectAdministration， Yuanxing Fang and Sibo Wang; FundingAcquisition， Yuanxing Fang and Sibo Wang.

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