Facet‐specific Active Surface Regulation of Bi x MOy (M=Mo, V, W) Nanosheets for Boosted Photocatalytic CO2 reduction

Abstract Photocatalytic performance can be optimized via introduction of reactive sites. However, it is practically difficult to engineer these on specific photocatalyst surfaces, because of limited understanding of atomic‐level structure‐activity. Here we report a facile sonication‐assisted chemical reduction for specific facets regulation via oxygen deprivation on Bi‐based photocatalysts. The modified Bi2MoO6 nanosheets exhibit 61.5 and 12.4 μmol g−1 for CO and CH4 production respectively, ≈3 times greater than for pristine catalyst, together with excellent stability/reproducibility of ≈20 h. By combining advanced characterizations and simulation, we confirm the reaction mechanism on surface‐regulated photocatalysts, namely, induced defects on highly‐active surface accelerate charge separation/transfer and lower the energy barrier for surface CO2 adsorption/activation/reduction. Promisingly, this method appears generalizable to a wider range of materials.

with water or ethanol. The products were collected and washed with water and ethanol three times.
Following drying in the freeze-dryer, the sample was cleaned with plasma cleaner to remove carbon impurities on the surface. Resulting samples were labelled as BWO.

Plasma cleaning
Following drying of the powders in the freeze-dryer, they were put in the plasma cleaner, vacuumed to 150 mTorr and purged with oxygen (BOC). This was repeated three times until the cleaner was stable at 450 mTorr. The plasma was conducted for 5 min. The powders were washed with water and dried in the freeze-dryer.

Sonication
Sonication was conducted under an ultra-sonic probe. Fifty (50) mg as-prepared samples were dispersed in 100 mL Na2SO3 solution (80 mM) for 2 h. The experiment is conducted at 0 ℃ in an ice-water mixture cooling system. The ultrasonic probe was operated for 2 s and stopped for 4 s. The samples were washed with water and dried in the freeze-dryer. Samples following sonication were labelled, respectively, BMO-R, BVO-R and BWO-R.

Material characterizations
X-ray diffraction, XRD data were determined on a powder X-ray diffractometer (Miniflex, Rigaku) using Cu Kα radiation. Raman spectroscopy data were determined using a confocal Raman microscope (Horiba LabRAM HR Evolution) with a 10X objective and a 532 nm laser. Morphology determination was conducted on a Tecnai G2 transmission electron microscopy (TEM). HRTEM, EDX mapping and EELS spectra were obtained on Transmission electron microscopy under STEM mode (FEI Titan Thermis, 200 kV). XPS measurement was on a VGESCALAB 210 XPS spectrometer with Mg K source. Binding energies were referenced to the C 1s peak at 284.6 eV. UV visible diffuse reflectance spectra were obtained on a UV-Vis spectrophotometer (UV2600, Shimadzu, Japan). An RF-5301PC spectrofluorophotometer (Shimadzu, Japan) was used to determine steady-state photoluminescence (PL) spectra at room temperature (25 ℃). Transient-state PL decay curves were determined on an FLS1000 fluorescence lifetime spectrophotometer (Edinburgh Instruments, UK). Continuous-wave (CW) X-band (ca. 9.385 GHz) electron paramagnetic resonance (EPR) spectra were recorded on a Bruker Elexsys E500 spectrometer equipped with an ElexSys Super High Sensitivity Probehead and He cooling using a cryogen-free cryostat (Bruker waveguide Cryogen-free system with recirculator, WVGD SYS 5K F70H wRCRC 2). The magnetic field was calibrated with a Gauss meter and measurements were carried out using a modulation amplitude of 0.8 mT, a modulation frequency of 100 kHz and a microwave power of SUPPORTING INFORMATION 4 0.5 mW (26 dB of 200 mW, non-saturating condition). Temperature was set to 20 K. Powder samples for EPR were weighted to 0.1 mg and packed in OD 4mm quartz EPR tubes. The height of these samples ranged from 5.9 to 13 mm because of available sample quantity. EPR signal intensity was calibrated by reference to frozen solutions of copper triflate in methanol and included data to cover the variable height of the samples. X-ray absorption was obtained from the XAS beamline of Australian Synchrotron (ANSTO, Melbourne). Data collection was conducted under the transmission mode. XAS raw data were background-substracted, normalized and Fourier-transformed with Athena.

Photocatalytic CO2 reduction
Photocatalytic performance tests were conducted in a 287 mL reactor sealed with silicone-rubber septa at ambient conditions. A 300 W Xenon lamp was used as the light-source. No light filter was used for BMO, BMO-R, BWO and BWO-R whilst 420 nm filter was used for BVO and BVO-R. In a typical test, 20 mg of photocatalyst was dispersed in 5 mL water and sonicated for 5 min then loaded on a glass-fibre filter to reduced the stack effect and dried under infrared light. Prior to illumination, the reactor was purged with laser-grade CO2 for 1 h. Product was collected from the reactor h -1 via syringe and analysed via gas chromatograph (GC, 7890B, Agilent). GC was equipped with plot-Q and a 5 Å sieve column (Agilent) in series, TCD and methanizer/FID detectors and UHP Ar (BOC) as carrier-gas. Each test was conducted in triplicate. Blank experiments were conducted under conditions, namely, photocatalysts in ultra-high purity Ar under illumination, photocatalysts in laser-grade CO2 without illumination, and laser-grade CO2 under illumination without photocatalysts. Trace/no products were apparent. Importantly, these blank experiments confirmed there were no carbon impurities in the experimental system.

Electrochemical and photoelectrochemical testing
Testing was conducted in a standard, three-electrode system with the as-prepared samples as the working electrode, a Pt wire as the counter electrode, and Ag/AgCl (saturated KCl) as reference electrode. In the identical three-electrode system, EIS measurement was carried out in the range 1 to 2  10 5 Hz with an AC amplitude of 20 mV. 0.5 M Na2SO4 was used as electrolyte. Polarization curves were obtained in the three-electrode system. The bias sweep range was -1.5 to -0.8 V vs. Ag/AgCl with a step size of 5 mV. 0.5 M Na2SO4 was the electrolyte. In the same three-electrode system the TPC response measurement was carried out. A 300 W Xenon light was used as the light source. 0.5 M Na2SO4 aqueous solution was the electrolyte. Working electrodes were prepared as follows: 5 mg sample, 960 μL of mixed solvent (Visoproponal: Vwater = 1: 2 ) with the addition of 40 μL of 5% Nafion. The dispersion was vigorously sonicated for 6 h to form a homogenous ink. A doctor-blade method was used to coat the slurry onto a 2  1.5 cm FTO glass electrode.

In situ diffuse reflectance infrared spectroscopy (DRIFTS)
All IR spectra were determined using a Nicolet iS20 spectrometer equipped with an HgCdTe (MCT) detector cooled with liquid nitrogen and a VeeMax III (PIKE technologies) accessory. The in situ DRIFTS were determined using a Praying Mantis DRIFTS accessory and a reactor (Harrick Scientific, HVC-DRP).
A 300 W Xe lamp (ZhongJiaoJinYuan) is connected to a liquid light guide for irradiation.
Samples were purged with wet CO2 for 40 min until the sample spectrum was stable. CO2 adsorption in dark was recorded until it was stable. Spectra under irradiation were recorded as a function of time to determine the dynamics of surface carbon contamination.

Theoretical computation
DFT computations were performed with the Vienna Ab Initio Simulation Package (VASP) code. [1] The Perdew-Burke-Ernzerhof (PBE) functional was employed for electron exchange-correlation within the generalized gradient approximation. [2] The projector-augmented wave (PAW) method was used to describe the ionic cores. [3] Geometry optimizations were performed with a 400 eV cut-off energy for plane wave expansion. Ionic relaxations were conducted until all forces were < 0.01 eV·Å -1 . A Gaussian smearing was used with 0.2 eV width and a (2 × 2 × 1) Monkhorst-Pack k-point grid was applied. The Tkatchenko-Scheffler method was employed to describe long-range van der Waals interactions. [4] Kinetic barriers were computed via climbing-image nudged elastic band (CI-NEB). [5] Four images were interpolated between initial (IS) and the final state (FS) to determine minimum energy path, and geometry of the transition state (TS). TS was confirmed through frequency analysis to ensure only one imaginary frequency existed, assigned to the unstable mode of minimum energy path.
BMO and BMO-R (010) surfaces were optimized in a 1 × 2 × 1 supercell. A vacuum space of 20 Å was applied to separate the interactions between neighbouring slabs. The computational hydrogen electrode (CHE) model was employed for free energy computations. [6] Free energies for intermediates                     Binding energy shifts remain stable evidencing that the modified surface remains stable during photocatalytic CO2 reduction testing.         [21] Cu/Bi/BiVO4 H2O vapor Xe lamp CO 11.15 [22] Supporting References