The Design of Supramolecular Assemblies with Metal Salt as Precursors Enables The Growth of Stable Polymeric Carbon Nitride Photoanodes

Polymeric carbon nitrides (CN) have gained significant interest as photoanodes in photoelectrochemical cells (PEC). A widely researched approach for synthesizing CN films with controlled optical and photoelectrochemical properties relies on using supramolecular assemblies as the precursors for thermal polymerization over a transparent conductive substrate. However, the formation of supramolecular assemblies is highly dependent on the temperature and solubility in a given solvent, limiting the full potential of this method. Moreover, the intercalation of metal ions is challenging due to the use of polar solvents. Here, this study shows a new way of synthesizing supramolecular assemblies with metal ions using a solvothermal approach. The solvent, monomer composition, salt quantity, reaction temperature, and film thickness are varied in this study. As a result, well‐attached, uniform CN films with good optoelectronic properties are achieved. The synthesized photoactive CN films exhibit very low onset potentials and reach ≈0.13, ≈0.15, and 0.30 ± 0.01 mA cm−2 photocurrents in 0.1 m phosphate buffer (neutral) solution, 0.1 m KOH(aq) (basic) solution, and 0.1 m KOH solution containing 10 vol.% triethanolamine as the hole scavenger, respectively.


Introduction
Polymeric carbon nitrides (CNs) are a family of low-cost semiconductor materials, [1] posing great promise to constitute the photoabsorbing layer in photoelectrochemical (PEC) devices.[3][4][5] To construct an efficient photoanode based on CN, the first goal would be achieving a successful deposition of a CN layer on the surface of a transparent conductive electrode, such as fluorine-doped tin oxide coated glass (FTO), which is relatively stable both at the high temperatures used for CN synthesis and at the alkaline conditions commonly employed in water-splitting PEC devices.Recent years have seen tremendous progress in the development of deposition techniques of CN layers and films, [5] including but not limited to seeded crystallization of CN monomers, [6,7] thermal vapor condensation, [8] liquid-mediated growth, [9] and deposition of supramolecular assemblies via doctor-blading [10] followed by thermal polymerization.
Limiting the discussion to the photoabsorbing CN film, the main obstacles hindering these photoanodes from achieving their full potential are: [i] the contact (i.e., attaining robust physical attachment to the substrate); [ii] manipulation of the intrinsic properties of the CN layer while controlling the physical dimensions of the film (thickness and so forth); and [iii] integration with additives. [11]nspired by the vast knowledge accumulated for CN powder photocatalysts, a common way to control the CN's properties (obstacle [ii]) is the rational design of supramolecular aggregates as the precursors for CN materials with improved properties and tailored functionalities (e.g., C/N ratio, heteroatom incorporation, porosity and morphology, optoelectronic properties). [1,12]he synthesis process typically involves the self-assembly of organic building blocks through non-covalent interactions such as H-bonding or - stacking.These building blocks are carefully chosen to incorporate both carbon and nitrogen elements, ensuring the formation of a CN framework [13] and allowing CN formation via thermal polymerization. [14,15]Our group has demonstrated the dispersion of supramolecular precursors (and possible additives; obstacle [iii]) in H-bonding dispersant (ethylene glycol, EG) into a paste that can then be easily doctor-bladed on an FTO substrate and in situ polymerized into the final CN, [16][17][18][19] thus tackling the three described obstacles simultaneously.Scheme 1. Illustration of the synthetic procedure.First, cyanuric acid and melamine are ground in the presence of KCl; In an appropriate solvent (W, MeOH, MeCN, or NMP), they undergo a solvothermal reaction to form CMK-s-solvent supramolecular assembly.The powder is first dispersed into a paste using ethylene glycol, then doctor-bladed onto a substrate (FTO), and dried to form CN precursor films.After calcination in the presence of melamine vapor, the CN films are formed over the FTO; these electrodes are denoted as CN-CMK-s-solvent.
High-temperature liquid-phase reactions for different CN polymorphs are common in sub-critical organic solvents [20,21] and in ionothermal reactions [22,23] ; however, the synthesis of supramolecular assemblies of CN monomers is habitually limited to room temperature (rt).Inspired by our recent work where EG was used as part of a hydrothermal treatment to supramolecular assemblies, [19] in the current work, we explore the solvothermal synthesis of supramolecular aggregates from precursor molecules [24] in various solvents for the synthesis of CN.It enables simple and precise control over structure or morphology by manipulating multiple reaction parameters such as temperature, time, surfactants, and solvents. [25]Additionally, the solvothermal approach facilitates utilizing a wide range of solvents and monomers since it overcomes miscibility limits, and different non-covalent interactions may lead to supramolecular aggregation of hitherto unknown complexes at higher yields.
Furthermore, we integrate a solvothermal process incorporating KCl as an additive [15] to serve as the first step toward fabricating uniform CN layers with excellent contact to conductive substrates.This integration modifies the synthesized supramolecular CN precursors and allows the formation of photoanode films with merits in both adhesion and overall PEC performance.

Results and Discussion
CN photoelectrodes were synthesized by thermal polymerization of doctor-bladed precursor pastes on FTO substrates in the presence of melamine vapor. [5,26]The supramolecular precursor powders were prepared by a solvothermal treatment of a mixture of cyanuric acid, melamine, and KCl.These powders, after drying, are dispersed in ethylene glycol to form the viscous precursor paste, [10] used for doctor-blading, as shown in Scheme 1.
[31][32] Here, we note that without the addition of KCl during the synthesis of the solvothermal supramolecular precursors, the ensuing CN layers (after calcination) have a poor contact with the FTO substrate (see Figure S1, Supporting Information).Therefore, one of the roles KCl plays as an additive is to improve the contact between the CN layers and FTO, a prerequisite for a photoelectrode.
We hypothesized that harnessing the solvothermal methodology would allow exploring a variety of organic solvents of different polarities and potential interactions with the dissolved monomers, benefiting from the high temperature to overcome solubility limitations of both monomers and additives.Thus, new forms of supramolecular aggregates could be used for CN fabrication with tunable properties and possibly with improved contact and deposition consistency.
We started with the well-established cyanuric acid-melamine combination, which is formed via up to three hydrogen bonds between each pair in the complex. [13,33]Since the aggregates' morphology depends on the solvent, we have chosen testing water [34] (W) and three additional polar solvents to allow the dissolution of the monomers and polar additives.The selected solvents-methanol (MeOH), acetonitrile (MeCN), and Nmethyl-2-pyrrolidone (NMP)-represent both protic and aprotic families of polar molecules covering a range of solvation and hydrogen bonding abilities (as measured by the Hansen polar and hydrogen bonding parameters, respectively) as well as different boiling points (see Table S1, Supporting Information for the relevant physical parameters).

Solvothermal Supramolecular Precursors Characterization
As expected, a supramolecular structure (after a 4 h solvothermal reaction at 180 °C) forms in all solvents as indicated by Fourier transform infrared (FTIR) spectroscopy (Figure 1a).At room temperature, CM-rt-MeCN fails to form a supramolecular structure (Figure S2, Supporting Information).However, when subjected to solvothermal conditions, it acquires the supramolecular structure, confirming that the solvothermal approach allows the formation of structures not attainable at rt.
The reaction between melamine and cyanuric acid leads to Hbonding and results in distinct changes in the stretching and bending vibrations observed in the FTIR spectra of the two compounds. [13,35]The FTIR analysis reveals notable differences in the stretching vibrations of the amino groups of melamine, which were observed initially at 3469 and 3414 cm −1 .Additionally, the C=O stretching modes of cyanuric acid exhibit a slight shift to higher wavenumbers, ranging from 1685 to 1730 cm −1 , in all four solvothermally prepared supramolecular assemblies. [13]urthermore, the characteristic bending modes of the triazine ring in melamine display a shift to lower wavenumbers, specifically, from 812 to 764 cm −1 upon forming the supramolecular assembly with cyanuric acid. [14]The FTIR spectra display two distinct sharp peaks at ≈3370 and 3228 cm −1 , which correspond to the characteristic stretching vibrations of H-bonded amines.The intensities of these two peaks vary significantly between the four samples, as expected from the differences in H-bonding in the final structure, possibly with residual solvent molecules. [19]he X-ray diffraction (XRD) of the solvothermal precursor powders (Figure 1b) reflects: (i) the varying supramolecular structure upon changing the solvent from water to MeOH, MeCN, and NMP; and (ii) the presence of KCl(s) within the powders when solvent ≠ water (see also XRD patterns over extended range normalized to KCl's (200) in Figure S3, Supporting Information).Unlike those observed in melamine and cyanuric acid alone, the presence of diffraction signals centered ≈10.9°, 11.8°, and 27.9°i n the CMK-s-W pattern, provides compelling evidence of the formed aggregate. [13,36,37]However, the intensity of these signals in the other samples indicates that the supramolecular arrangement is altered as a consequence of the chosen solvent (see also the XRD patterns of CM-s-solvent in Figure S4, Supporting Information).We ascribe these variations to the difference between the hydrogen bonding ability (at the elevated temperature) and deprotonation degree in each solvent forming the rossete CM structure between cyanuric acid and melamine units, the - stacking between the rosette layers, and even the dissolved C-to-M ratio (which has been shown via calculations to affect the order crystallization in water). [38]Additionally, the strongest diffraction intensity of melamine (2 ≈26.15°) is absent in CMK-s-W, indicating that melamine completely reacted at the hydrothermal conditions while in organic solvents unreacted melamine remains in the powders.
The yield (mass basis) of the CMK-s-solvent precursor powders after the solvothermal reactions are: 71.29% (W), 91.20% (MeOH), 97.11% (MeCN), and 81.89% (NMP).The lower yield in water further supports the fact that the higher solubility of KCl (33% of the total reactants mass) and monomers in water limits their conversion into a supramolecular aggregate with no embedded KCl(s).
Based on FTIR and XRD results, we suggest that the formed supramolecular aggregates contain K + (see Figure S5, Supporting Information for the proposed supramolecular structure for CMK-s-solvent), possibly residual Cl − as ionic pairs or adsorbates, and KCl crystals (except in the hydrothermal case, CMK-s-W, where no KCl(s) is present and probably only K + ions interact strongly with the organic backbone, see discussion in Section 3).
As the next synthetic step toward photoanode preparation, an H-bond forming dispersant, ethylene glycol (EG; 300 μL in the case of powders prepared in MeOH and MeCN; 350 μL in the case of water and NMP), is used to form a viscous paste. [10]A doctorblade application technique is used to coat the FTO substrates uniformly. [4]canning electron microscopy (SEM) images (top view) of all four ensuing precursor films in Figure 2 show a consistent coating.The main difference between samples is the morphology of the aggregates formed in water (i.e., the hydrothermal synthesis of CMK-s-W).][41][42][43][44] The relative abundance of rods diminishes according to the Hansen H-bonding parameter ( H , Table S1, [45,46] since the hexagonal structure forms as a result of hydrogen-bond based reorganization of the melamine and cyanuric acid building blocks).Furthermore, the chosen solvothermal solvents allow the formation of smaller aggregates, as the formation of an extended network of H-bonds is less favorable.Thus, the non-aqueous environment allows forming a more uniform deposited film of precursors.We were unable to distinguish the unreacted KCl(s) crystals in the supramolecular aggregates films, and they look similar to films without the addition of KCl (Figure S6, Supporting Information).

CN Photoanodes Synthesis and Characterization
Calcination of the coated substrates in a glass tube with melamine at 520 °C for 2 h (inert N 2 atmosphere, see Scheme 1) allows the formation of a thick porous CN layer with intimate contact to the substrate.
FTIR and XRD measurements confirm the successful fabrication of a CN layer.Notably, distinct ethylene glycol peaks are absent in the FTIR spectra of CN photoanodes.This absence can be attributed to the elevated calcination temperature of 520 °C, which leads to the complete evaporation of ethylene glycol during the synthesis of CN photoanodes. [16,19,47]The characteristic breathing mode of the s-triazine units found in CN results in a pronounced absorption peak at 804 cm −1 (Figure 3a). [13]urthermore, several intense bands within the 1200-1600 cm −1 range confirm the presence of the typical triazine ring present in CN heterocycles. [17,48]The electrodes prepared from CMK-s-MeOH, CMK-s-MeCN, and CMK-s-NMP precursors exhibit an additional stretching vibration at 2165 cm −1 , attributed to cyano groups. [49,50]This vibration is significantly weaker in CN-CMK-s-W, where fewer ─CN groups remain after polymerization.Additionally, a broad absorption band ca.3100 cm −1 range is observed, indicating N─H stretching vibrations originating from uncon-densed or partially polymerized terminal amino groups, as well as O─H stretching vibrations from adsorbed water molecules.
[53][54] The (100) diffraction ca.2 = 13.1°frominterplanar structural packing of tri-s-triazine units is weak yet more significant in CN-CMK-s-W (hydrothermal preparation of the precursors).This might suggest more abundant induced inplane porosity in the CN layers [43,44,55,56] or other defects in the intrinsic crystal structure of CN [56,57] when the supramolecular organization occurs in organic solvents.Moreover, the crystalline XRD peaks of KCl [58,59] are prominent only in CN electrodes that were prepared using non-aqueous solvothermally-prepared precursors (Figure S7, Supporting Information shows patterns normalized to KCl's (200) signal).This reflects the situation in the supramolecular precursors, i.e., KCl crystals remain within the CN films after dispersion in EG and calcination.The combined XRD and FTIR analysis indicates that incorporating KCl in the synthetic procedure does not prevent the formation of the CN framework, yet induces defects and porosity.In summary, the CN films from solvothermal preparation in non-aqueous solvents with KCl exhibit more cyano groups and lower packing of the CN sheets.
Top view SEM images of CN-CMK-s-solvent layers (Figure 3c,d; Figures S8 and S9, Supporting Information) indicate the coverage of the FTO with a CN layer containing some broken elongated structures on the surface, originating mainly from melamine vapor condensation, [19,26] and a glomerular lump network with variable porosity and compactness.Since CN-CMK-s-MeOH results in the highest photoelectrochemical activity, we focus on its morphological characterization, as presented in Figure 3c-f.The thickness and connection of the synthesized CN layers with the FTO substrate were studied by cross-sectional SEM imaging (Figure 3e,f; Figure S10, Supporting Information).For all electrodes, a similar thickness of ca.70-80 μm is measured and a good continuous contact of CN-CMK-s-MeOH layers is seen (Figure 3f), explaining the good physical adhesion and structural stability (Section 3.1); it should allow fast charge transfer kinetics across this interface under operational conditions (Section 3.2).
To examine the photophysical properties of the CN films, absorption spectra were obtained using UV-vis diffuse reflectance spectroscopy (DRS, Figure 4a), and the optical bandgaps were estimated using a Tauc plot analysis [60] and a direct bandgap assumption (inset of Figure 4a).The estimated optical bandgaps of CN-CMK-s-MeOH and CN-CMK-s-NMP exhibit somewhat lower energies (also manifesting in a slight red shift in the absorption onset) in comparison to CN-CMK-s-W and CN-CMK-s-MeCN photoelectrodes.Of note is that the narrower bandgap when MeOH is the solvent during supramolecular precursor prepa-ration occurs only under solvothermal conditions in the presence of KCl (see Figure S11, Supporting Information for UVvis DRS of CN-CM-rt-solvent).Thus, in accordance with previous reports, partial alkali metal incorporation may improve optical absorbance. [31,61]hotoluminescence (PL) spectroscopy (Figure S12, Supporting Information) shows an emission peak ca.64] A Mott-Schottky analysis of CN-CMK-s-solvent electrodes confirms that the CN layers exhibit the expected n-type semiconductor behavior (Figure S13, Supporting Information).The flat-band potentials were determined by extrapolating the linear part of the measurement to intersect the x-axis.Since the flat-band potential typically lies 0.1 V below the conduction band (CB) minima for n-type SCs, [65,66] the CB edges were estimated accordingly.Combining this analysis with the optical bandgap estimation allows proposing an approximate band diagram for the CN electrodes (Figure 4b), showing that all are suitable for PEC water-splitting.
To gain further insight into the fate of the potassium and chloride from the supramolecular assemblies in the ensuing CN materials, energy-dispersive X-ray spectroscopy (EDS) analysis was conducted on the CN electrodes (Figure S14, Supporting Information).In the case of the CN-CMK-s-MeOH electrode, the EDS elemental mapping (Figure S15, Supporting Information) demonstrates a uniform spatial distribution of carbon and nitrogen, as expected.More interestingly, K and Cl are non-evenly distributed-potassium is always present where chlorine is detected, but the reverse does not occur.It means that some areas are rich with KCl, but potassium is also distributed throughout the film.The EDS quantification (Table S2, Supporting Information) and XRD analysis support this claim: KCl(s) is present in all samples except CN-CMK-s-W, but free potassium is also there (specifically, in CN-CMK-s-MeOH, ≈1.5 at % of K and 1.0 at % Cl, indicating K + bound to the CN backbone in addition to KCl(s)).Of note is that in all samples the amount of free potassium is similar, with no Cl detected in CN-CMK-s-W.
X-ray photoelectron spectroscopy (XPS) was employed to investigate the chemical environment of the CN photoanodes' constituents.The survey spectra show the presence of C, N, O, K, and Cl (Figure S16, Supporting Information).The O 1s signal could be attributed to oxygen originating from unreacted cyanuric acid and surface-adsorbed moisture or CO 2 , indicating minimal Ocontaining impurities.
High-resolution C 1s XPS spectra of CN-CMK-s-solvent electrodes were analyzed (Figure S17a, Supporting Information) and deconvoluted into five distinct peaks.In the CN-CMK-s-MeOH electrode, the peak at 284.8 eV corresponds to graphitic carbons (C─C─H). [19][69] The 287.9 eV peak represents sp 2 -bonded carbons within heptazine or triazine rings (N═C─N), [70] while the peak at 288.5 eV can be attributed to sp 2 carbons in the aromatic ring (C─(N) 3 ). [19,71]Lastly, the peak at 289.2 eV corresponds to C─O bonds (surface-adsorbed atmospheric oxygen) or C─N bonds. [17]The high-resolution N 1s spectra (Figure S17b, Supporting Information) reveal five distinct peaks at 398.8 eV, 399.2 eV, 399.7 eV, 400.9 eV, and 401.9 eV.][74] A visual depiction of the chemical motifs and structural arrangement of CN is illustrated in Figure S18 (Supporting Information) based on this analysis in conjunction with the FTIR analysis discussed before.

Adhesion Test of CN-Layer
To evaluate the mechanical stability and adhesion properties of the CN film over FTO in a CN-CMK-s-MeOH electrode, we employed adhesive scotch tape and subsequently exerted external force. [75,76]This process was carried out periodically, with the scotch tape being removed at 5 and 85 min intervals, respectively (see scheme in Figure S19, Supporting Information).
Removing the scotch tape after 5 min revealed only an impression on the tape, primarily attributed to the CN powder on the surface.The subsequent application of a fresh tape for an additional 85 min and its removal shows minimal evidence of impression, exemplifying the film's sustained integrity through intimate contact with the FTO substrate and uniformity across its entire surface.

PEC Performance Metrics of CN-Photoanodes
We investigated the PEC properties of CN photoanodes, examining the influence of the solvothermal preparation of the supramolecular CN precursors (solvent, KCl amounts, and reaction temperature) on the photocurrent density.CN electrodes prepared from assemblies from only melamine and cyanuric acid at room temperature (no KCl, shaken for 24 h) were labeled as CN-CM-rt-solvent and showed relatively low photocurrent densities, j ph , with CN-CM-rt-MeOH slightly exceeding 100 μA cm −2 (Figure S20, Supporting Information) under 1 sun illumination at 1.23 V versus reversible hydrogen electrode (RHE) in 0.1 M KOH (pH ≈13).The addition of KCl (CN-CMK-rt-MeOH) improved the value by ≈25% (j ph reaching almost 130 μA cm −2 ).As was mentioned before, CN films do not form a good adhesion with the FTO substrate in the absence of KCl in the solvothermal preparation of supramolecular precursors; thus, KCl was used in all solvothermal reactions for CN photoanode preparation and electrochemical measurements to test whether it improves performance as hypothesized.
Chronoamperometric measurements were used to determine the solvothermal reaction temperature (160-200 °C) that would result in the highest and most stable j ph (Figure S21, Supporting Information).The optimal thickness of CN-CMK-s-MeOH films was also optimized via modulation of the height of the barrier during doctor-blading of the precursor pastes (denoted as "single layer", "double layer", and "triple layer" according to the thick-ness of the used scotch tape).The optimal photocurrent density ca.150 μA cm −2 at 1.23 V versus RHE in 0.1 M KOH (pH ≈13) was observed for the double layer (see Figure S22, Supporting Information).Therefore, this thickness was adopted as the constant parameter during CN photoanode preparation.Figure 5a shows chronoamperograms of CN photoanodes prepared from supramolecular assemblies synthesized at the optimized conditions of 180 °C for 4 h and a "double layer" thickness.
The effect of KCl content in CN-CMK-s-MeOH electrodes was also examined (Figure S23b, Supporting Information).A small amount (10 mmol, see experimental section) was enough to achieve adhesion of the CN film and improve performance relative to CN-CM-rt-MeOH.Further increase results in the optimized performance (33 mmol KCl), which is used for all the characterizations hereinafter, while even higher amounts result in diminishing photocurrent density.
Linear sweep voltammetry (LSV) curves of CN-CMK-s-solvent electrodes (Figure S24, Supporting Information) reflect the discussed trends among the solvothermal media.Remarkably, all electrodes exhibit very low onset potentials, ≈0 V versus RHE for CN-CMK-s-MeOH.We ascribe it to the intimate contact formed between the CN and the FTO, aided by the incorporated K + within the layer.LSV curves in the presence of a hole scavenger (Figure S24, Supporting Information) further confirm the chronoamperometric measurements and exhibit doubling of the photocurrent density in the case of CN-CMK-s-MeOH; besides, three separate electrodes (with and without TEOA) were measured in LSV and chronoamperometry to confirm the reproducibility of the PEC results (Figure S25, Supporting information); the average photocurrent of CN-CMK-s-MeOH at 1.23 V versus RHE in the presence of a hole scavenger is 0.30 ± 0.01 mA cm −2 .These metrics are of the magnitude to the best pristine CN photoanodes and are on par with the best supramolecular-based CN films (Table S3, Supporting Information).
The incident photon-to-current conversion efficiency (IPCE) of the CN-CMK-s-MeOH electrode under alkaline operational conditions [77] (in the presence of a hole scavenger and without it) was evaluated between 310 and 500 nm (Figure 5d).Non-zero photoresponse is measured at long wavelengths > 460 nm attributed to sub-bandgap defect states.In accordance with the determined optical bandgap, significant photoresponse begins at  <450 nm with a measured increase in the presence of TEOA due to the improved charge transfer kinetics when TEOA is the oxidized species.The maximum IPCE values are 4.4% at 374 nm in 0.1 m KOH and almost double in the presence of TEOA (7.9%), in accordance with the photocurrent measurement under 1 sun illumination.The decline in IPCE values at short wavelengths stems from lower photoabsorption of CN and the absorbance of the FTO substrate (glass and tin oxide).
Figure 5e shows the stability of CN-CMK-s-MeOH during continuous operation for 5 h under 1 sun illumination in alkaline environment in the presence of TEOA hole scavenger.Within 1 h, a significant decrease in current density was observed, decreasing by nearly 40% to ≈185 μA cm −2 .Subsequently, a modest decline ca.15% is recorded.This abrupt decrease in photocurrent density during the 1 h of testing may be attributed to the partial degradation of heptazine units, dissolution of embedded KCl(s), and leaching out of intercalated potassium during the PEC operation.XPS analysis of the CN electrode after 1 h of operation (see Figure S26, Supporting Information) shows a decrease in the relative abundance of (N─(C) 3 ) in the N 1s spectrum, which can be attributed to partial degradation of heptazine units in the CN structure.
Further characterization using SEM, FTIR, XRD, and UVvis DRS (Figures S27-S29, Supporting Information), shows no significant changes in the morphology and film thickness, and the intimate contact between the CN layer and FTO substrate remains intact even after the PEC stability measurement.The significant difference is the absence of a KCl XRD signal after the stability measurement, confirming the dissolution of KCl(s).The FTIR spectrum shows a broad peak in the 2300-3700 cm −1 region, which can be related to strongly-adsorbed TEOA or hydroxyls as a result of interaction with the electrolytes during operation, as well as an indication of oxidation of terminal amino groups.In addition, the slight blue-shift in the absorbance after operation may arise from a combination of partial degradation of CN surface, photo-corrosion, surface contamination, and intercalated K + leaching.
Figure 5f illustrates hydrogen production during a PEC measurement of CN-CMK-s-MeOH electrode in the presence of TEOA.The decline in the rate over time reflects the decrease in the current density over time.Therefore, we have calculated the Faradaic efficiency (FE) of the hydrogen evolution reaction (HER) only during the first half hour.H 2 (g) amounts were detected at t = 0 and at t = 30 min (2.27 μmol cm −2 ; normalized to the geometric area of the CN layer).The estimated average HER FE during the first 30 min in the presence of a hole scavenger exceeds 94%; this high value corroborates the fact that the main kinetic barrier for PEC activity is the sluggish OER, and future efforts should be concentrated on improving the stability of the photoanode on the one hand, and improving the OER kinetics, on the other hand.

Conclusion
We have demonstrated a straightforward method for synthesizing supramolecular aggregates in the presence of salts at solvothermal conditions.This approach allows overcoming miscibility limits by choice of the solvent.It can be tailored to include monomers for synthesizing polymeric carbon nitrides-in this case, cyanuric acid and melamine supramolecular complex in the presence of KCl.The resulting powders are easily doctorbladed and allow the formation of a CN layer after calcination with good attachment to an FTO substrate, which is impossible under conventional conditions when a salt is introduced.The thick, uniform, and porous CN films exhibit significant watersplitting photoelectrochemical activity in both alkaline and neutral environments.The improved contact improves charge transfer through the FTO to the counter-electrode and is one of the main contributors to the low onset potential under illumination.The best photoanode, CN-CMK-s-MeOH, results in an impressive photocurrent density of ≈0.15 mA cm −2 under 1 sun illumination at 1.23 V versus RHE in alkaline electrolyte, which doubles with the addition of a hole scavenger and allows relatively long stability and high HER Faradaic efficiency of 94.4%.Overall, these findings highlight the merits of solvothermal supramolecular precursor preparation and the benefits of simple and cheap salt addition, paving the way further to improve CN-based PEC devices toward water-splitting or other reactions.

Experimental Section
Further details regarding chemicals and all characterization equipment and procedures (e.g., analytical, electrochemical, and photoelectrochemical) are given in the Supporting Information.
Solvothermal Synthesis of Supramolecular Assemblies as Precursors: Cyanuric acid (20 mmol) (C), melamine (M) (20 mmol), and potassium chloride (33 mmol) (denoted henceforth as K in all abbreviations) were ground using an agate mortar-and-pestle, yielding a fine powder mixture (CMK).Subsequently, this mixture was carefully transferred into a polytetrafluoroethylene (PTFE)-lined stainless-steel autoclave (100 mL nominal internal volume of the liner) and filled with 40 mL of the appropriate solvent, namely, DI water (W), methanol (MeOH), acetonitrile (MeCN), and Nmethyl-2-pyrrolidone (NMP).The autoclave was placed into a thermallycontrolled oven for 4 h for the solvothermal reaction to occur at the designated temperature (160-200 °C range; unless indicated otherwise, T solvothermal = 180 °C).The autoclave was allowed to naturally cool down to room temperature for ≈4-5 h, and the white slurry settled at the bottom was collected by decanting the supernatant and vacuum drying the slurry overnight at ≈60 °C.This material is denoted as CMK-s-solvent, where CMK stands for the cyanuric acid-melamine supramolecular assembly (CM) prepared in the presence of potassium chloride (K); s stands for a solvothermal process; solvent stands for the reaction medium (solvent = W, MeOH, MeCN, or NMP).Control experiments were conducted (i) at the same experimental conditions without the addition of potassium chloride (and denoted as CM-s-solvent), and (ii) at rt with shaking (denoted as CMK-rt-solvent or CM-rt-solvent, when KCl is added or absent, respectively).

Synthesis of Polymeric Carbon Nitride (CN) Layers Over FTO:
To form CN films on FTO, 0.5 g of the supramolecular precursor (CMK-s-solvent) was ground and mixed with EG (300 μL in the case of MeOH and MeCN; 350 μL in the case of W and NMP), using an agate mortar-and-pestle, resulting in the formation of a cohesive and uniform supramolecular paste.Subsequently, this paste was applied (doctor-blade method) [5,10,16] onto a clean FTO substrate using a glass rod with an elevation of 1-3 layers of scotch tape (3M company, cat.810, 19 mm width; the standard procedure is using an elevation of two layers).After drying on a hot plate (surface T ≈85 °C) for 20 min, the precursor-coated FTO substrates were transferred into 16 mm diameter glass test tubes (three substrates per tube).Additionally, 1 g of melamine powder was introduced at the bottom of each tube to facilitate the formation of a CN layer. [5,26]Before calcination, all the glass test tubes were purged with Ar for 2 min, tightly closed with an Al foil, and placed in a tube furnace under constant N 2 flow.The samples were heated from rt at a constant rate of 5 °C min -1 to the reaction setpoint of T thermal polymerization = 520 °C and maintained at this temperature for 2 h.The CN-coated FTO substrates are used for characterization and utilization as photoanodes in a photoelectrochemical cell are denoted with a "CN-" prefix to the abbreviation of the supramolecular assembly, which was used for its synthesis, e.g., CN-CMK-s-MeOH is a film originating from a solvothermal synthesis of cyanuric acid, melamine, and KCl in methanol after thermal polymerization in the presence of melamine vapor (see also Scheme 1).

Figure 1 .
Figure 1.Characterization of the obtained supramolecular assemblies (powders) acting as the precursors for CN synthesis.a) FTIR spectra of CMK-ssolvent powders and the pristine monomer powders-melamine and cyanuric acid, and b) XRD patterns of CMK-s-solvent (larger 2 range and normalized intensity appear in Figure S3, Supporting Information).The FTIR spectra and XRD patterns are offset for clarity.

Figure 3 .
Figure 3. Structural and morphological characterization of CN electrodes.a) FTIR spectra, b) XRD patterns normalized to CN's (002) intensity ca.2 = 27.6°; the FTO substrate pattern is normalized so the strongest SnO 2 (200) diffraction intensity at 2 = 37.7°coincides with CN-CMK-s-W signal for ease of comparison.SEM images of CN-CMK-s-MeOH: c) top view, and d) higher magnification; e) cross-section, and f) higher magnification image focusing on the interface with the FTO-coated glass substrate.See Figure S7 (Supporting Information) for complete diffraction signals from KCl.The FTIR spectra and XRD patterns are vertically offset for clarity.

Figure 4 .
Figure 4. a) UV-vis diffuse reflectance spectra of CN electrodes (inset: the corresponding Tauc plot analysis assuming a semiconductor with a direct bandgap), and b) Proposed energy band positions of CN electrodes.

Figure 5 .
Figure 5. PEC performance metrics of CN-CMK-s-solvent electrodes.Chronoamperometry (current densities versus time) in a) 0.1 m KOH at 1.23 V versus RHE under 1 sun illumination and b) 0.1 m KOH aqueous solution containing 10 vol.% triethanolamine (TEOA) as a hole scavenger.c) Chronoamperometric current density comparison of CN-CMK-s-MeOH electrode with (dashed line) and without (complete line) TEOA as a hole scavenger.d) Incident photon-to-current conversion efficiency (IPCE) of the CN-CMK-s-MeOH electrode in 0.1 m KOH (complete line), and 0.1 m KOH containing 10 vol.%TEOA (dashed line) at 1.23 V versus RHE.e) Stability measurement of CN-CMK-s-MeOH electrode at 1.23 V versus RHE in 0.1 m KOH containing 10 vol.%TEOA for 5 h (current density recorded under continuous 1 sun illumination).f) Hydrogen evolution measurements of the headspace above CN-CMK-s-MeOH electrode under operational conditions-1.23V versus RHE in 0.1 m KOH containing 10 vol.%TEOA in 0.1 m KOH; the quantified amount of H 2 (g) is normalized to the surface area of the geometrically-defined photoelectrode; the lines linking the experimental data points serve only as a visual reference.