Hydrogen (H₂), a kind of clean energy, is regarded as an ideal substitute for fossil fuels, which is nonrenewable and the combustion products are harmful for the environment. One of the most promising ways to produce H₂ energy is to split water into H₂ by photocatalytic process, because photocatalysis technology could realize the conversion and utilization of abundant solar energy [1]. Semiconductor-based photocatalyst materials with high activity and good stability are the core of photocatalytic process and great influence photocatalytic efficiency. To date, substantial research efforts have focused on developing a variety of visible-light responded photocatalysts, in order to extend the utilization of visible light in solar spectrum [2, 3].

Relative to inorganic semiconductors, organic polymers usually contain no or few metal elements, possess abundant synthetic materials, and easy to realize processing modification. Based on these advantages, polymers with comparative band gaps have been developed and applied in photocatalytic reactions for H₂ evolution, such as heptazine-based microporous polymer networks (HMPs) and covalent triazine-based frameworks (CTFs) [4, 5]. Graphitic carbon nitride (g-C₃N₄), an old but novel polymer semiconductor has been considered as a promising photocatalytic material. With appropriate band gap (~ 2.7 eV) and energy band positions (E_CB = –1.3 V and E_VB = 1.4 V, vs NHE at pH = 7), it can thermodynamically suitable for water reduction into H₂ [6]. However, the lower electron transport properties determines that a mass of works are still needed to modify and improve the photocatalytic performance of g-C₃N₄ semiconductors.

To address the inherent limitations, so many strategies have been applied to modify physicochemical structures of g-C₃N₄, such as semiconductor coupling, copolymerization, nanostructured design, and non-metal doping [7-19]. Doping is one of the most effective and convenient methods among these strategies. Non-metal doping is an effective way not only can improve charge separation due to electronegativity difference of atoms, but also keep non-metallic properties of semiconductor. For this, doping nonmetals by O, B, P and S could expand the visible-light response and adjust the energy band positions so as to improve the photoactivities [20-23]. In addition to the common single non-metal doping, codoping strategies with metal and nonmetal elements or multiple nonmetal elements have potential to develop high-performance g-C₃N₄-based photocatalysts [24, 25]. Choi et al. [26] prepared potassium (K) and phosphate (P) codoped carbon nitride (CN) framework from in situ incorporation method. The obtained catalyst exhibited a high apparent quantum yields (8.0 % at 420 nm) for H₂ production, and was about 25 times from those of bare CN. Liu et al. [27] prepared P-O codoped g-C₃N₄ from hydrothermal synthesis assisted by dissolution-precipitation process. Increases anoxic photocatalytic RhB degradation constants by approximately 27 times under visible light was achieved. In our previous work [28, 29], boron (B) and fluorine (F) codoped g-C₃N₄ was prepared from direct calcination, and the precursors were mixed by imidazole tetrafluoroborate ionic liquid. The separated doping state by B and F in skeleton and surface of catalyst significantly improved the electronic properties of pristine g-C₃N₄, hence got much enhanced photocatalytic H₂ production rate.

Carbon (C) and iodine (I) are well known species to improve electrical conductivity and light absorption properties of semiconductors [30]. Up to now, although C and I doped g-C₃N₄ photocatalysts have been explored, little work has been done on the C and I codoped g-C₃N₄, and the effects of the codoping on its electronic structure, band gap, and photocatalytic activity for H₂ evolution have not been investigated yet [31-33]. In this work, porous C-I codoped g-C₃N₄ photocatalysts were synthesized via two-step methods including thermal polymerization of a mixture of dicyandiamide (DCDA), urea and 1-ethyl-3-methy limidazolium iodide ([Emim]I), then post-thermal treated in air. The characteristics of the codoped materials were analyzed and the results show that the post-thermal activation and codoped C-I both play important role in the enhancement of photocatalytic H₂ evolution activity, the former of which contributes to the compact conjugate accumulation and the latter improves visible light response and charge transport. This work might provide feasible route for developing visible-light nanostructured g-C₃N₄ photocatalysts for efficient H₂ production from water reduction.

All of the materials were purchased from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China) and used as received without further purification.

C-I codoped g-C₃N₄ was synthesized as follows: 4.0 g of dicyandiamide (DCDA), 4.0 g of urea and measurable 1-ethyl-3-methy limidazolium iodide ([Emim]I) were dissolved in H₂O with vigorous stirring, and then the resulting solution was evaporated to dryness at 80 ℃. The solid mixtures were placed into covered porcelain crucibles and direct calcined at 550 ℃ in a muffle furnace for 2 h with a ramp rate of 5 ℃ /min. The process yield was about 31.7%. The obtained samples were denoted as CNI_x. Then, the CNI_x products were taken into open porcelain crucibles and further calcined at 550 ℃ for 2 h. The samples were denoted as CNIN_x and the process yield was about 59.6% to CNI_x. Here, x refers to the used mass of [Emim]I (g).

The following specific analysis methods were used to test the performance of the sample. X-ray diffraction (XRD) patterns were collected on a Bruker D8 Advance diffractometer with Cu K_α radiation (λ = 1.5406 Å). Scanning electron microscope with energy-dispersive X-ray spectroscopy analysis (EDS) was performed using a Nova NanoSEM 230 instrument. Nitrogen adsorption-desorption isotherms were collected at –196 ℃ using a Micromeritics ASAP 2020 surface area and porosity analyzer. X-ray photoelectron spectroscopy was made using a Thermo ESCALAB250Xi spectrometer with Mg K_α radiation as the excitation source. Ultraviolet-visible (UV-Vis) diffuse reflectance spectroscopy (DRS) was performed on a Varian Cary 500 Scan UV-visible system. The photoluminescence (PL) emission spectra were recorded on an QuantaMaster^TM fluorescence spectrometer. The electrochemical experiments were taken on a CHI660D workstation. The working electrode was prepared on a fluorine-doped tin oxide (FTO) transparent conductive film glass. Electrochemical measurements were performed in a typical three electrode cell, using a Pt piece and calomel electrode as counter electrode and reference electrode, respectively. The electrolyte was 0.2 mol/L Na₂SO₄ aqueous solution.

Photocatalytic activity tests were evaluated by the photocatalytic hydrogen evolution from water under visible light irradiation (λ > 400 nm). Reactions were carried out in a Pyrex top-irradiation reaction vessel connected to a glass closed gas system. In each experiment, 50 mg of photocatalyst powder was dispersed in an aqueous solution (100 mL) containing triethanolamine (10 vol%) as a sacrificial electron donor. 3 wt% Pt was loaded on the surface of the catalyst by in situ photodeposition method using H₂PtCl₆. The reaction solution was evacuated several times to remove air completely prior to irradiation under a 300 W xenon-lamp (CEL-HXF300). The wavelength of the incident light was controlled by applying appropriate cut-off filters. The temperature of the reaction solution was maintained at 6 ℃ by the flow of cooling water during the reaction. The evolved gases were analyzed by gas chromatography equipped with a thermal conductive detector (TCD) and a 5 A molecular sieve column, using nitrogen as carrier gas.

The morphology of the CNI and CNIN samples was investigated by SEM observations and the results were shown in Fig. 1. Clearly, without participation of ILs, the obtained bulk sample CNI₀ features typical dense slate-like lamellar agglomerates. For CNI_0.2, the laminar particles became smashed to small pieces, suggesting that the incorporated ILs decomposed during the synthesis process and caused disorder degree of conjugated plane of CN materials due to particle fragmentation and size reduction. In the meantime, there are significant fractured slice can be seen in CNIN₀ after post-thermal treatment, derived from the further thermal decomposition of unpolymerized CN fragments. By conjunct effects of ILs incorporation and recalcination, the CNIN_0.2 possesses unconsolidated porous accumulation, which is important for increased surface contact in heterogeneous catalytic reaction.

EDS elements analysis for CNIN_0.2 is inserted in Fig. 1(d). In addition to C, N and O, which were regular necessary in g-C₃N₄, I was also observed. SEM elements mapping images for C, N. O and I also confirmed the existence of indine. The relative element contents (wt%) of prepared samples were analyzed and shown in Fig. 1. Although few iodine can be found uniformly distributed in CNI samples synthesized from ILs doping, and with the increase of ILs mass used, the iodine contents in corresponding products apparent increased. After post-thermal treatment, the volatilization of some I was inevitable, and the content reduced. Moreover, the content of C also increases with the increase amount of used ILs after secondary heat-treatment. The increased C content resulted from high-carbon ILs. These results could identify that C-I codoped g-C₃N₄ products were prepared from iodized ionic liquid as dopant source. From previous reports, single C doped and I doped could improve the optical absorption and electronic property, hence in favour of the improvement of photocatalytic performance of g-C₃N₄ [28, 29]. Therefore, the C-I co-doped samples could show predictable optimized structure and photocatalytic properties.

Fig. 2 shows XRD patterns of pristine CNI₀ and post-thermal treated C-I Co-doped (g-C₃N₄) graphitic carbon nitride samples. Two characteristic peaks of graphite-like carbon nitride were observed in all samples. The small peaks at about 13.1^o corresponded to (100) in-planar repeat tri-s-triazine units and the strong peaks around 27.4^o corresponded to (002) interlayer reflection of g-C₃N₄ [34, 35]. Compared with pure g-C₃N₄, there was no obvious change in the (001) diffraction peak of samples doped with iodide-containing liquids, indicating that there was no change in the interplanar spacing. As shown in Fig. 2(b), the (002) peak shifts from 27.4^o of CNI₀ to 27.5^o of CNIN₀, reflecting the XRD d-spacing changes from 0.325 nm for CNI₀ to 0.324 nm for CNIN₀, also the peak intensity was enhanced obviously. It indicates an optimized compactness and higher polymerization degree due to improved post-polycondensation, which could strongly avails the charge transport in organic semiconductors [36]. With the increasing content of inserted ILs, the (002) peaks for CNIN slightly lower but no shift. That is to say, more defects occurred after co-doped by C and I, but no significant change for the lamellar stacking characteristic, which is consistent with the morphology observation.

Increased porosity is of great concern for heterogeneous photocatalysis. The textural properties of CNI and CNIN samples were analyzed using a nitrogen adsorption-desorption spectrometer. The isotherms and the Barrett-Joyner-Halenda (BJH) pore size distributions were shown in Fig. 3. The isotherm of the samples shows type IV with H3 hysteresis loops, and the adsorption branch of nitrogen isotherms shows a sharp increase at high relative pressures, proving the existence of mesopores. The surface areas measured by the BET technique for CNIN₀ is 37.3 m²/g, about 2 times higher than that of CNI₀, because of the post-thermal decomposition effect. However, the surface area of CNI_0.2 (13.8 m²/g) slight decrease, which may resulted from the heteroatomic doping. It is expected that the surface area for CNIN_0.2 increased to 54.4 m²/g, nearly 4 times to that of CNI_0.2. The increased surface area of CNIN_0.2 may resulted from intensive decomposition of CNI_0.2, which possessed lower degree of polymerization than CNI₀ due to heteroatomic doping. Additionally, the pore-size distribution curves of all samples are wide around pore diameters of 10–50 nm, indicating the co-existence of meso- and macropores. Obviously, the pore volumes of samples treated by secondary calcination and doping increased. The results show that the increase of surface area and porosity for C-I co-doped samples could increase the surface active sites, hence improve photocatalytic performance. From the comparison results, post-thermal treatment is more crucial to enlarge porosity. Heteroatomic co-doping realized by ILs inserting take no obvious affect for the texture of CN catalysts. Although CNIN_0.2 has no significant increase in the absorption of visible light compared with CNI₀, the increased specific surface area adds more active sites for surface catalytic reaction. So the increase of porous properties plays important role in optimizing the photocatalytic performance. Nevertheless, microstructure photoelectric property changes could not be ignored, which are equal or more important for photocatalysis.

To further reveal the microstructure and the chemical environment of surface elements, X-ray photoelectron spectra (XPS) were characterized and shown. Fig. 4(a) shows XPS scan spectra of CNIN_0.2 and signals for C, N, O and I can be detected. The high resolution spectra for I 3d are shown in Fig. 4(b). I 3d region can be fitted into four peaks in the spectrum around 617.5, 618.7, 629.1 and 630.8 eV. The two peaks about 629.1 eV (I 3d_3/2) and 617.5 eV (I 3d_5/2) show the existence of I^–, while the other two weaker peaks at 621.6 (I 3d_5/2) and 633.1 eV (I 3d_3/2) indicate the oxidation state of doped iodine I⁵⁺ during the reaction process [33, 37, 38]. The high-resolution C 1s and N 1s XPS spectra were also analyzed. The mainly carbon species centered at 284.6, 286.3 and 288.1 eV are assigned to C–C, C–O and N-containing aromatic ring (N–C=N), respectively [39]. The N1s spectrum of CNIN_0.2 is deconstructed into three peaks located at 398.6, 400.0 and 401.0 eV, which are assigned to sp²-hybridized nitrogen (N1) in triazine rings (C–N=C), tertiary nitrogen N₂-(C)₃ groups (N2) and amino functions carrying hydrogen (C-N-H, N3) [40].

To further investigate the influence of post-thermal treatment and C-I codoping modification, the surface elements chemical states of CNI₀, CININ₀ and CNIN_0.2 from XPS were parsed and summarized in Table 1. Clearly, the molar ratio of C/N increase. The raised C content on one hand signifies N deficiency, which is inevitable, on the other hand, states C doping dating from decomposition of carbon-rich ILs, proved by remarkable increased molar ratio of C1/C3 [41, 42]. More importantly, the increased N1/N2 after post-thermal treatment and C-I co-doping indicates much optimized in plane heterocyclic sp² conjugate structure. Meanwhile, the decreased N3/N1 values confirms the reduction of surface amino group. These results coincident confirms that the C-I co-doping modification did not change the basic skeleton, but optimize sp² aromatic ring polymerization structure. It is a significant factor and favorable for photocatalytic hydrogen production [43].

The photocatalytic activity of prepared samples were evaluated by H₂ evolution in a conventional closed-circulation system under visible light irradiation (λ > 400 nm). Fig. 5(a) shows the photocatalytic H₂ evolution of samples using triethanolamine and 3wt% of chloroplatinic acid as electron donors and co-catalysts, respectively. It can be observed that compared to CNI₀ (17.0 μmol/h), the H₂ evolution rate for CNI samples modified by C-I codoping first increase, and then decrease which is restrained by excess C content, and CNI_0.2 shows the best performance for 30.1 μmol/h. More significantly, the H₂ production over CNIN samples after post-thermal treatment was sharply enhanced, following the same trend that increase first and then decrease. The H₂ evolution rate achieved of CNIN_0.2 is as high as 168.2 μmol/h, which is more than 5.5 times higher than CNI_0.2 and 2.7 times higher than CNIN₀.

Fig. 5(b) shows the H₂ evolution performance of CNI₀, CNIN₀ and CNIN_0.2 under different wavelength of light illumination. The order of H₂ evolution rate is CNIN_0.2 > CNIN₀ > CNI₀ regardless of the wavelength of illumination, which further confirm that C-I co-doping and post-thermal treatment jointly promote the photocatalytic performance of CN catalysts. When the incident wavelength extends from visible to ultraviolet region, the H₂ production rate of all samples was enhanced. It is according to photoresponse of CN samples and confirms the photocatalytic reaction progress. Interestingly, the advantages for H₂ generation over CNIN_0.2 is more significant in visible light, especially λ > 420 nm. Therefore, optimized visible-light absorption can be speculated for CNIN samples. In addition, the energy band structure, electronic properties and photoelectric response were investigated to illustrate the enhanced photocatalytic H₂ evolution activity.

In order to evaluate the stability and reusability of the CNIN photocatalyst, H₂ production catalyzed by CNIN_0.2 was carried out cycled for four times under visible light (λ > 400 nm). As shown in Fig. 6, under continuous illumination, the amount of H₂ release shows an approximate linear. In addition, after four consecutive cycle's reactions, almost negligible deactivation could be observed, indicating that CNIN_0.2 has good H₂ generation stability under visible light irradiation.

The optical absorption of the samples was elucidated by UV-Vis diffuse reflection spectra (DRS), as shown in Fig. 7. Post-thermal treatment caused a slight blue shift in the spectrum for CNIN₀ due to quantum size effect. For CNIN samples, band edges redshift can be found, and the corresponding band gaps (E_g) from Tauc plot for CNIN_0.2 and CNIN₀ are 2.81 and 2.82 eV, respectively. Compared to CNI₀, although the band gap enlarges, the effect of C-I codoping remedied this shortage. On the other hand, the absorption sideband of CNIN samples is larger than CNI₀ (shown in Fig. 7(a)), which are derived from C-I codoping expounded in the aforementioned contents. Beyond this, enhanced light absorption, especially in visible light region (450–700 nm) was observed and gradually improve with the increased C-I doping content. The enhanced visible light absorption is beneficial for the photocatalytic reaction proved by H₂ evolution test. Compared with CNI₀ and CNI_0.2, the absorption capacity and response range of visible light increased, indicating that C-I doping could in favor of the photocatalytic performance of the catalyst. From previous reports, C and I doping are all conducive to enhance the light absorption of CN materials, but from above-mentioned results, the doping amount of I is much lower than C, so the improved light absorption is mainly due to C doping.

Mott-Schottky analysis has been employed to determine the semiconductor type and flat potential of prepared samples. Fig. 7(c) gives Mott-Schottky curves of CNI₀, CNIN₀ and CNIN_0.2 samples measured at 800 Hz. It can be seen that the plot slopes are positive for all samples, indicating n-type behavior for all samples. The plot of CNIN₀ and CNIN_0.2 show bigger slop than that of CNI₀, indicating higher electron donor density for the former catalysts. It is helpful for promoting photocatalytic performance because of increased electrical conductivity and mobility of charge carriers [44]. The flat potentials for CNI₀, CNIN₀ and CNIN_0.2 samples are –1.35, –1.37 and –1.40 eV vs SCE, respectively. Therefore, the conduction band (CB) of samples are deduced since the CB potentials of n-type semiconductors are close to the flat band potentials [45]. Combined with the Eg valus gained from UV-Vis plot, the band positions of these samples are exhibit in Fig. 7(d). According to the thermal dynamics principle of photocatalytic water for H₂ evolution, C-I codoped CNIN_0.2 sample with negative shifted CB has higher H₂ evolution power.

In addition to bang gap structure, the effects of C-I co-doping and post-thermal treatment on the charge transfer property change of the catalysts were also investigated. Fig. 8(a) gives the room temperature steady-state PL emission spectra of samples excited at 400 nm. All samples exhibit broad emission peaks in the range of about 470 nm, which can be attributed to the band-band PL phenomenon with the light energy approximately equal to the band gap energy of g-C₃N₄. Compared with CNI₀, the PL intensity of CNIN₀ raise. It may be resulted from the increased defects which could became recombination centers. After modification by C-I co-doping, the PL intensity of both CNI and CNIN samples significantly reduced, and gradually decreases with increasing doping amount, indicating that C-I co-doping inhibit the recombination of photo-generated charges, lower the disadvantages of increased defects.

It is well known that a longer charge carrier lifetime usually implies a better photocatalytic activity because of higher participation possibility of carriers in photocatalytic reactions. Calculated results of time-resolved transient PL spectroscopy (Fig. 8(b)) shows that the average lifetime of the excited charge carriers in the case of CNIN_0.2 is 5.203 ns, whereas CNI₀ and CNIN₀ exhibit average lifetimes of 0.671 and 4.762 ns, respectively. These fitting data solidly corroborate that the lifetimes of charge carriers in CNIN_0.2 is significantly lengthened by the combinative post-thermal treatment and C-I codoping.

Fig. 9(a) displays the transient photocurrent responses of CNI₀, CNIN₀ and CNIN_0.2 with several on-off cycles. CNIN_0.2 produces significant higher photocurrent than that of CNI₀ and CNIN₀, verifies the results of PL. It is a good indication of the improved mobility of the photo-excited charge carriers. First, it is attributed to the close-grained layer stacking and optmizied sp² conjugate structure, which can shorten the diffusion length of charge migration and promote electron relocalization on the surface. Secondly, heteroatomic doping by C and I both beneficial to the improvement of carrier transmission in semiconductors. The photocurrent response has a slight attenuation might due to the less stability of iodine in the electrochemical process.

The above PL and photocurrent techniques highlight the high separation and transfer efficiency of photo-generated electron-hole paris in CNIN_0.2 upon light irradiation. In order to inspect the capacity to shuttle and convey charge carriers to the targeted reactive sites, EIS measurement were carried out under dark. Fig. 9(b) gives the EIS Nyquist plots. Among these samples, CNIN_0.2 shows the smallest diameter, suggesting its lowest resistance for interfacial charge transfer.

The interrelation of several structure parameters of CNI and CNIN samples with H₂ evolution rate (HER) was summarized and listed in Table 2. It can be found that enlarged surface area, negative shifted conduction band position and optimized separation/transfer efficiency of photo-induced carriers all give push to the enhanced photocatalytic H₂ evoulution activity over C-I codoped porous carbon nitride.

Porous C-I co-doped g-C₃N₄ materials (CNIN) were successfully prepared through post-thermal treat and doped by iodized ionic liquid co-polymerization. The secondary calcination makes for higher degree of polymerization, more porous and larger surface area. Synergistic with C-I co-doping modification, light absorption response of the prepared catalysts were much enhanced for prepared catalysts, especially in visible-light region. Compared to pristine g-C₃N₄, although the bang gaps of C-I codoped porous CNIN materials enlarged, the improved reduction potential due to negative shifted conduction band position provides powerful impetus for the H₂ production from water photolysis. In addition, the faster transfer and lower recombination rate of photo-induced carriers for CNIN samples renders much enhanced photocatalytic H₂ evolution rate under visible light irradiation. The CNIN_0.2 with moderate C-I doing content shows best photocatalytic activity (168.2 μmol/h), which is nearly 5.5 times higher than merely C-I doped CNI_0.2 and 2.7 times higher than post-thermal treated porous CNIN₀. This is an easily-operated route for preparation non-metallic doped porous CN-based polymer semiconductors using ILs as dopant source, and the photocatalytic properties of samples could be convenient modified and enhanced.