Overall water splitting by Pt/g-C3N4 photocatalysts without using sacrificial agents† †Electronic supplementary information (ESI) available: Characterization and experimental detail. See DOI: 10.1039/c5sc04572j

Direct splitting of pure water into H2 and O2 in a stoichiometric molar ratio of 2 : 1 by conjugated polymers via a 4-electron pathway was established for the first time, as demonstrated here using a g-C3N4 polymer and redox co-catalysts of Pt and Co species.

Using photocatalysts to produce hydrogen sustainably by water splitting is the "holy grail" in modern science. Over the past 40 years, inorganic semiconductors, such as metal oxides and metal (oxy)nitrides, have been utilized as photocatalysts for hydrogen production. [1][2][3][4][5][6][7][8] However, direct water splitting in a wireless powder photocatalytic system to produce gaseous hydrogen and oxygen has not yet been achieved using conjugated polymers (CPs). These materials have already shown great promise for use in organic electronics and photovoltaic devices, such as solar cells, light-emitting diodes, and eld-effect transistors, due to their good processability and tuneable electronic structures. [9][10][11][12][13] The key challenge to using pristine CPs for direct water splitting is the insufficient hopping charge transport of the chains (usually below 10 À4 cm 2 V À1 s À1 ) and a poor stability in water and under light irradiation. 12 Increasing the structural dimensions of the CPs (e.g., from 1D chains to 2D architectures) is desirable because the hole mobility is greatly increased (up to 0.1 cm 2 V À1 s À1 ) by the remarkably reduced binding energies of the Frenkel-type excitons and the robust stability of the 2D extended p-conjugated units. 14 However, further progress in direct water splitting by CPs will rely on breakthroughs in combining stable CP light transducers with suitable redox cocatalysts (usually noble metals) to promote charge separation and to reduce charge build-up on the polymer surface to prevent photocorrosion. Indeed, the promise of this type of system has been demonstrated by the successful development of 2D graphitic carbon nitride (g-C 3 N 4 ) polymer and metal-based redox cocatalyst systems for CO 2 reduction, organic synthesis and water half-splitting reactions using sacricial reagents. [15][16][17][18][19][20][21][22] In contrast, it is difficult to achieve overall water splitting without using sacricial reagents because it depends not only on a rational chemical synthesis to tune the textural properties of the polymer but also on a rational design of the composite to control the reaction kinetics on the polymer surface. [23][24][25][26][27] Photocatalytic water splitting by a prototypical g-C 3 N 4 polymer was shown to be thermodynamically possible because the C 2p and N 2p orbital bands straddle the water splitting redox potentials, 15-22,28-34 but pure g-C 3 N 4 is typically limited by sluggish kinetics in photocatalyzing overall water splitting due to a lack of surface redox active sites. By optimizing the g-C 3 N 4 bulk and morphological properties and employing suitable redox cocatalysts (e.g., Pt for H 2 evolution and Co(OH) 2 for oxygen evolution), activities for the water half-splitting reactions (water reduction and oxidation) can be dramatically increased. [28][29][30][31][32][33][34] Therefore, if the appropriate water redox cocatalysts are simultaneously deposited on g-C 3 N 4 , pure water splitting to produce gaseous hydrogen and oxygen could be achieved. However, the rough deposition of cocatalysts by traditional chemical reduction (e.g., H 2 and NaBH 4 ) cannot fully amplify the activity. Besides, the densely stacked graphitic layer also causes trouble for charge separation and migration due to a long bulk diffusion distance, resulting in a low photocatalytic quantum efficiency. 15 It is advisable to reduce the diffusion distance by rational synthesis of a g-C 3 N 4 nanosheet together with suitable cocatalyst modication to achieve water splitting. Up to now, direct water splitting photocatayzed by g-C 3 N 4 CPs in the absence of sacricial reagents has never been realized and still remains a signicant basic science challenge. Here, we demonstrate that light-excited g-C 3 N 4 CPs can induce a one-step water splitting reaction via a four-electron pathway to generate gaseous H 2 and O 2 in a stoichiometric molar ratio of 2 : 1 when their morphology is modied and the reaction kinetics are improved by modication with Pt, PtO x , and CoO x via photodeposition. The optimal g-C 3 N 4 -based nanocomposite had a turnover number of 3.1 moles of H 2 and O 2 per mole of g-C 3 N 4 photocatalyst for the overall water splitting reaction. The nanocomposite was stable in water and under light irradiation.
The g-C 3 N 4 polymers used for photocatalytic water splitting were typically prepared by thermally polymerizing urea into heptazine units at 550 C which pack together like graphitic crystals. This structure was conrmed by X-ray diffraction (XRD), Fourier transform infrared (FT-IR) spectroscopy, and Raman spectroscopy (Fig. S1 †). 15,[35][36][37] The g-C 3 N 4 optical properties measured by UV-vis diffuse reection spectroscopy (DRS) were characteristic of a semiconductor; g-C 3 N 4 had an optical absorption edge at 442 nm due to the excitation of electrons from its valence band to its conduction band (Fig. 1a). The conduction band minimum (CBM) and valence band maximum (VBM) of the g-C 3 N 4 semiconductor were determined to be À1.31 V and 1.49 V (vs. NHE, pH ¼ 7), respectively, from electrochemical Mott-Schottky plots ( Fig. 1b and c), where an estimated at potential was directly used as the conduction band potential. Density functional theory (DFT) calculations revealed that the band gap was 2.56 eV with the CBM and VBM located at À1.0137 and 1.5505 V (vs. NHE, pH ¼ 7), respectively, which enables g-C 3 N 4 to act as a redox shuttle for the water splitting reaction (Fig. S2 †). This calculated band gap is consistent with the experimental data and further demonstrates that in theory, g-C 3 N 4 could be used to split water.
First, the effects of g-C 3 N 4 morphology on the photocatalytic activity were investigated. We prepared three types of g-C 3 N 4 using dicyandiamide (DCDA), ammonium thiocyanate (ATC) and urea as precursors. The results showed that aer in situ photo-deposition with Pt, the urea-derived g-C 3 N 4 exhibited signicant photocatalytic activity for the overall water splitting reaction, while the other samples were inactive for overall water splitting (Table S1 †). It should be noted that all pure g-C 3 N 4 polymers showed no activity for overall water splitting in the absence of cocatalysts, implying that surface kinetic control using Pt species was indispensable to achieve overall water splitting by g-C 3 N 4 based photocatalysts. N 2 sorption measurements revealed that the DCDA-and ATC-derived g-C 3 N 4 samples had smaller surface areas than the urea-derived samples (ca. 10 m 2 g À1 vs. 61 m 2 g À1 ). However, mpg-C 3 N 4 with a surface area of ca. 67 m 2 g À1 also exhibited no water splitting activity. This indicated that surface area was not the major factor in controlling the water splitting activity and the splitting of water on densely stacked g-C 3 N 4 polymers was indeed very difficult to achieve. To better understand the real mechanism of water splitting on the so surface of the CPs, we characterized the morphology of the above different polymers. TEM images of DCDA-and ATC-derived g-C 3 N 4 and mpg-C 3 N 4 samples revealed densely stacked polymer layers, which were very different from the silk-like thin nanosheets of the urea-derived one (Fig. S3 †). The fast evolution of O in the form of CO 2 or CO could accelerate the deamination rate. Thus, the texture, morphology and electronic properties of the CNU samples were optimized, and contributed to creating the active Pt/g-C 3 N 4 photocatalysts for overall water splitting. Evidently, accelerated charge separation and migration on the nanosheets can be obtained in comparison with densely stacked graphitic layers, which is elucidated well by the corresponding large decrease of PL emission intensity (Fig. S4 †). The nanosheet structure can also be certi-ed by an AFM experiment. As shown in Fig. 2a, the thickness of the nanosheet is determined as $2 nm. One can now easily conclude that the ultrathin 2D geometry of urea-derived g-C 3 N 4 is crucial for achieving overall water splitting as demonstrated by the fact that g-C 3 N 4 samples prepared from urea at different  temperatures all have remarkable water splitting activities (Fig. S5 †) due to their similar thin nanosheet structures (Fig. S6 †). The CNU samples prepared at 550 C showed optimum activities. This is because when the temperature is lower than 550 C, the heptazine cycle doesn't completely form, while partial decomposition occurs when the temperature is higher than 550 C. Both of these two aspects may generate inactive CNU samples. DCDA-and ATC-derived g-C 3 N 4 and mpg-C 3 N 4 samples revealed densely stacked polymer layers, and the deposition rate of Pt nanoparticles on the surface of the polymer was very slow, in the absence of organic sacricial agents to react with the holes. Optimization of the deposition technique of Pt is needed to enhance the overall water splitting activities of this bulky g-C 3 N 4 .
We then investigated the effect of cocatalyst loading techniques on the photocatalytic water splitting activity. Three different cocatalyst loading techniques, in situ photodeposition, and H 2 and NaBH 4 reduction, were developed to decorate the g-C 3 N 4 nanosheets. As shown in Fig. S7, † evident water splitting activity was generated for photodeposition of Pt on the surface of the g-C 3 N 4 nanosheets, while only very slow H 2 and no O 2 evolution were found for both H 2 and NaBH 4 reduction modi-ed ones. In the rst case, when g-C 3 N 4 was irradiated with light, photoexcited charge carriers were generated and then immediately migrated to the surface of the g-C 3 N 4 nanosheets without recombination. The surface adsorbed Pt 4+ was then reduced in situ by the excited electrons and deposited on the active sites, which can efficiently promote the water splitting. For the other investigated techniques, Pt 4+ was reduced by H 2 or NaBH 4 and then randomly deposited on the surface, resulting in poor activities. The selective photodeposition of Pt on thin g-C 3 N 4 nanosheets resulted in a uniform dispersion of ultrane Pt nanoparticles ($1-2 nm) with a (111) crystal lattice spacing of $0.23 nm (Fig. 2b and c). The homogeneous deposition of Pt can be further proved by STEM imaging (Fig. 2d). However, serious particle accumulation occurred when the Pt cocatalysts were deposited by H 2 and NaBH 4 reduction (Fig. S8 †), which was the major hindrance which led to decreased water splitting activity.
We also investigated the chemical composition and valence state of the Pt species. As shown in Fig. 3a and b, electron energy loss spectroscopy (EELS) and XRD analysis conrmed the existence of a Pt (111) plane. 38 Besides, no evident structure variation occurred aer modication with the Pt cocatalysts, implying a robust stability of the g-C 3 N 4 CPs. [39][40][41] Three pairs of XPS peaks corresponding to Pt 0 , Pt 2+ , and Pt 4+ with binding energy at 72.13, 74.26 and 78.17 eV, respectively, were measured (Fig. 3c). Pt 0 was effective for H 2 evolution while PtO x were able to promote O 2 evolution. 42 However, two pairs of XPS peaks were deconvoluted for a NaBH 4 reduction modied one (Fig. S9 †), indicating the complete reduction of Pt 4+ into Pt 2+ and Pt 0 . To conrm that PtO x were active for the promotion of a water oxidation reaction, we evaluated the photocatalytic water oxidation activities of the as-prepared PtO x /g-C 3 N 4 . As shown in Fig. S10, † this material showed enhanced activity for water oxidation in comparison with the pure one, emphasizing the positive role of PtO x in improving the water oxidation rate. In addition, the water splitting rates and evolved H 2 /O 2 gas ratio (Fig. S11 †) could be nely tuned by simply adjusting the total loading from 0.2 to 5 wt% due to the change of the ratio of Pt and PtO x intensities ( Fig. S12 and Table S2 †) and the alteration of particle size (Fig. S13 †). The creation of metal/polymer surface junctions promotes the interfacial redox reaction which can be conrmed by a rapidly decreased PL intensity (Fig. 3d). The optimum activity was achieved at a Pt loading content of 3 wt%. When Pt or PtO x were singly deposited on the g-C 3 N 4 nanosheets, the sample exhibited very poor activity in both cases, which once again highlighted that the simultaneous creation of both H 2 and O 2 evolution cocatalysts on the active sites was indeed essential for triggering the overall splitting of water.
The g-C 3 N 4 nanosheets modied by other noble metals (e.g., Rh, Ru, or Au) via in situ photodeposition all just showed trace H 2 and no O 2 evolution (Fig. S14 †), implying the importance of Pt for water splitting. The pH value and amount of polymer powders used for water splitting were also optimized ( Fig. S15 and S16 †). The optimum water splitting rate was obtained for samples prepared by photodepositing 3 wt% Pt on 0.2 g of g-C 3 N 4 nanosheets under neutral conditions. We then evaluated their stability for long term reaction.
As shown in Fig. S17, † the optimized Pt/g-C 3 N 4 showed good water splitting stabilities under both UV and visible light irradiation for 580 hours of continuous reaction. It should be noted that N 2 gas was evolved along with H 2 and O 2 at the initial stage of the reaction. This arises from the self-oxidation of the surface un-condensed amino groups (-NH) by excited holes. [43][44][45] As the reaction proceeded, aer 80 hours almost no N 2 evolution was observed, suggesting a complete consumption of the -NH groups. When the Xe lamp was turned off, the amounts of the evolved gases quickly diminished in just four hours (Fig. S18 †), indicating a fast occurrence of the backward reaction of water splitting on the Pt species (H 2 and O 2 recombination for water formation). Thus, to further enhance the overall water splitting activity of the system, an efficient restraint of the backward reaction via rational structural design of the cocatalysts (e.g., core/shell nanostructure) should be considered.
The addition of cobalt species for in situ formation of cobaltbased cocatalysts can also sufficiently promote the water oxidation selectivity and efficiency of metal-free semiconductors such as g-C 3 N 4 and h-BCN. [43][44][45][46][47] As expected, the simultaneous evolution of H 2 and O 2 gases in a stoichiometric ratio of 2 : 1 by Pt-Co/g-C 3 N 4 under UV (l > 300 nm) (12.2 and 6.3 mmol h À1 ) (Fig. 4a) and visible light irradiation (l > 420 nm) (1.2 and 0.6 mmol h À1 ) (Fig. 4b) was signicantly enhanced aer 1 wt% CoO x were further modied for use as O 2 evolution cocatalysts, which can be determined by XPS analysis (Fig. S19 †). The slightly decreased activity in each run of reaction may be attributed to the stacked samples on the inner side of the reactor (Fig. S20 †). Furthermore, no obvious deactivation was observed aer 510 hours of reaction ( Fig. S21 †), demonstrating the robust resistance of the composites to water and light corrosion at the so interface. The total amount of gaseous H 2 and O 2 collected reached $6.2 mmol, which corresponded to turnover numbers (TON) of 3.1 and 111.3 based on g-C 3 N 4 and Pt, respectively. The apparent quantum yield (AQY) for the overall water splitting reaction was calculated to be 0.3% at 405 nm (Fig. S22 †) and was monitored by an on-line gas chromatograph (Fig. S23 †). This is lower than the value of 2.5% of (Ga 1Àx Zn x ) (N 1Àx O x ) inorganic photocatalysts. However, it is a remarkable rst observation that photocatalytic overall water splitting can occur on the surface of an organic/polymer semiconductor via a 4-electron pathway. Optimization of the system to further improve the efficiency is ongoing in our lab.

Conclusions
The discovery of Pt/g-C 3 N 4 CPs that can split pure water without the use of sacricial reagents establishes a new chemical paradigm for exploiting clean, renewable solar energy using organic semiconductor light-energy transducers. Ongoing efforts are focused on modifying the electronic and textural structures of g-C 3 N 4 CPs and coupling them to low-cost kinetic promoters to facilitate photocatalytic cascade processes for water splitting and CO 2 xation that are relevant to sustainable energy production via articial photosynthesis. [48][49][50]