Synthesis and characterization of Bi-doped g-C3N4 for photoelectrochemical water oxidation

Abstract Photoelectrochemical (PEC) water splitting has emerged as a promising technology for the storage of renewable energy sources, via the production of hydrogen, a clean and multi-purpose chemical energy vector. The key component in a PEC cell is the photoanode where light energy is absorbed and transformed into electron-hole pairs of appropriate energy for water photo-oxidation. We report on the synthesis of g-C3N4 materials, with an elongated nano-structure, fabricated by the direct pyrolysis of supramolecular melamine used as a chemical precursor. The as-prepared material was used to host specific amounts of bismuth, a doping element used to adjust the band gap of the hosting matrix. The presence of Bi in the photoanodes was confirmed by energy dispersive x-ray analysis (EDX) analysis. Powder X-ray (p-XRD) and Fourier transform infrared (FT-IR) measurements performed on the photoanodes confirmed the absence of Bi-based oxides, and showed that bismuth may bonded to nitrogen atoms inside the voids of the g-C3N4 skeleton. Differential reflective spectroscopy (DRS) measurements revealed that the band gap energy was reduced upon introduction of Bi into g-C3N4. From photoluminescence (PL) plots, it was observed that the 2.5% Bi doping induced a 6-fold electron-hole separation, compared to the pristine g-C3N4. PEC water splitting measurements showed that 2.5% Bi doping approximately doubled the activity of g-C3N4 towards water oxidation. Electrochemical impedance spectroscopy (EIS) measurements showed that Bi doping was an effective method for decreasing the charge transfer across the electrode/electrolyte interface; 2.5% Bi-g-C3N4 was reduced by around 2.4 times compared to that of pristine g-C3N4. Bode-phase plots accompanied EIS spectra revealed that the lifetime of the photo-generated electrons in neat g-C3N4 was improved as a result of Bi doping. The band gaps and the positions of the valence and conduction bands were determined from Mott–Schottky plots.