Graphitic Carbon Nitride with Dopant Induced Charge Localization for Enhanced Photoreduction of CO2 to CH4

Abstract The photoreduction of CO2 to hydrocarbon products has attracted much attention because it provides an avenue to directly synthesize value‐added carbon‐based fuels and feedstocks using solar energy. Among various photocatalysts, graphitic carbon nitride (g‐C3N4) has emerged as an attractive metal‐free visible‐light photocatalyst due to its advantages of earth‐abundance, nontoxicity, and stability. Unfortunately, its photocatalytic efficiency is seriously limited by charge carriers′ ready recombination and their low reaction dynamics. Modifying the local electronic structure of g‐C3N4 is predicted to be an efficient way to improve the charge transfer and reaction efficiency. Here, boron (B) is doped into the large cavity between adjacent tri‐s‐triazine units via coordination with two‐coordinated N atoms. Theoretical calculations prove that the new electron excitation from N (2px, 2py) to B (2px, 2py) with the same orbital direction in B‐doped g‐C3N4 is much easier than N (2px, 2py) to C 2pz in pure g‐C3N4, and improves the charge transfer and localization, and thus the reaction dynamics. Moreover, B atoms doping changes the adsorption of CO (intermediate), and can act as active sites for CH4 production. As a result, the optimal sample of 1%B/g‐C3N4 exhibits better selectivity for CH4 with ≈32 times higher yield than that of pure g‐C3N4.

Electron spin resonance (ESR) signals were recorded at room temperature on a Bruker A300 spectrometer. X-ray photoelectron spectroscopy (XPS) were performed on Thermo Fisher Scientific-Escalab 250Xi. All the binding energies were calibrated by the C 1s peak at 284.8 eV. Solid state C 13 nuclear magnetic resonance (NMR) was measured on an Agilent 600M spectrometer. UV-visible diffuse reflection spectra were obtained on a UV-visible spectrophotometer (UV-2600, Shimadzu, Japan) with BaSO 4 as the reflectance sample. Nitrogen adsorption desorption isotherms were obtained on Micromeritics ASAP 2020 nitrogen adsorption apparatus.
Photoluminescence (PL) measurement was performed on Horiba LabRAM HREVO with excitation of 365 nm.
Time resolved fluorescence spectra were obtained on single photon counting (TCSPC) system (Picoquant "Timeharp 300") at wavelength of 450 nm. Output characteristic curves were measured on Keithley 4200 Semiconductor Characterization System. CO 2 adsorption isotherms were recorded on Micromeritics ASAP 3020 adsorption apparatus. CO temperature program desorption (TPD) curves were measured on Micromeritics AutoChem 2920.

Computational details
All DFT calculations were carried out using the Vienna ab initio simulation package (VASP). [1] The Perdew−Burke−Ernzerh of exchange-correlation functional was used for the generalized gradient approximation. [2,3] All geometry optimizations were carried out with a cutoff energy of 450 eV and a Monkhorst-Pack k-point mesh of 5 × 5 × 1. The iterative process considered was convergences, when the force on the atom was less than 0.01 eV Å -1 and the energy change was less than 10 -5 eV per atom. The Heyd-Scuseria-Ernzerhof (HSE06) method was employed to investigate the electronic properties, and obtain an accurate description of density of states (DOS). [4][5][6] The PBE+D2 method with the Grimme van der Waals correction was considered in calculation. [7] The free energy was defined as Where E is the total energy from DFT calulation, ZPE E is the zero-point energy, T is the temperature (298.15 K) and S is the enropy. Zero-point energies and entropies of the intermediates obtained from the vibrational frequencies. [8] The adsorption energies (E ads ) of CO on the g-C 3 N 4 and B-doped g-C 3 N 4 modes are respectively obtained with the following equations.
Where E(CO/Sub) and E(Sub) are the total energies of the different systems with and without CO molecule, and E(CO) is the energy of an isolated CO molecule. From this definition, a more negative value of E ads indicates that the adsorption is thermodynamically more stable and favorable.

Photocatalytic reduction of CO 2
Photocatalytic reduction of CO 2 was carried out by gas-solid surface reactions. CO 2 and H 2 O vapor were come from the chemical reaction of H 2 SO 4 solution and NaHCO 3 powder. Specific operation process can refer to the previous works. [9 -11] Gas products were tested by gas chromatography (GC-2014c, Shimadzu, Japan) with a thermal conductivity detector (TCD) and a flame ion detector (FID). The gas composition and concentration were calibrated by the mixture standard gas. A 300 W Xe arc lamp (Microsolar 300, Perfectlight Science Co. LtD, China) was used as UV/visible light source. We don't know how to calculate the exact value of the matrix element ( ). Based on the above formula, we focus on the density of states (DOS) of the final state, which can approximately indicate the transition probability ( ). We calculated the electronic structure of initial state and final state. For the DOS of g-C 3 N 4 ( Figure S1a and S1b), we further confirm the excitation of electrons from the twocoordinated N 2p orbitals (initial state) to C 2p orbitals (final state), for that the excited state (final state) is mainly contributed by the C 2p orbitals. For the DOS of B-doped g-C 3 N 4 ( Figure S1d and S1e), we

Results and
can observe the electrons of highest occupied states come from the N 2p orbitals (initial state) and the excited states (final state) is composed of B 2p, N 2p and C 2p orbitals. We found B 2p orbitals provided the main contribution on the excited state. Therefore, we propose that electrons are easily excited from N 2p to B 2p.
Moreover, we also calculated the partial charge density distribution of the initial and final state ( Figure S2). The electrons of initial state are mainly localized on two-coordinated N atoms and distributed in the x, y plane in both g-C 3 N 4 and B-doped g-C 3 N 4 . The final state is mainly distributed in the z direction of C and N atoms for pure g-C 3 N 4 . Thus, the electrons excite from N (2p x , 2p y ) to C 2p z orbitals in pure g-C 3 N 4 . Differently, we found the final state in B-doped g-C 3 N 4 are almost localized on B atom along the x, y plane. These results show the electrons excite from N (2p x , 2p y ) to B (2p x , 2p y ) orbitals is superior to original excite pathway (from N (2p x , 2p y ) to C 2p z orbitals) in B-doped g-C 3 N 4 .
Therefore, the new electron pathway from N (2p x , 2p y ) to B (2p x , 2p y ) in the same plane is much easier than N (2p x , 2p y ) to C 2p z .           Figure S14. Photocatalytic CO yield of CO 2 reduction with the as-prepared samples (x% represents x%B/g-C 3 N 4 , 0% represents pure g-C 3 N 4 ).     Table S3. The calculated adsorption energy of CO on pure g-C 3 N 4 and B-doped g-C 3 N 4 .