Designing covalent organic frameworks with Co-O4 atomic sites for efficient CO2 photoreduction

Cobalt coordinated covalent organic frameworks have attracted increasing interest in the field of CO2 photoreduction to CO, owing to their high electron affinity and predesigned structures. However, achieving high conversion efficiency is challenging since most Co related coordination environments facilitate fast recombination of photogenerated electron-hole pairs. Here, we design two kinds of Co-COF catalysts with oxygen coordinated Co atoms and find that after tuning of coordination environment, the reported Co framework catalyst with Co-O4 sites exhibits a high CO production rate of 18000 µmol g−1 h−1 with selectivity as high as 95.7% under visible light irradiation. From in/ex-situ spectral characterizations and theoretical calculations, it is revealed that the predesigned Co-O4 sites significantly facilitate the carrier migration in framework matrixes and inhibit the recombination of photogenerated electron-hole pairs in the photocatalytic process. This work opens a way for the design of high-performance catalysts for CO2 photoreduction.


Materials characterization
Fourier transform infrared (FTIR) spectra were recorded on a Perking Elmer Spectrum were pretreated under He atmosphere at 150 o C for 1 h (temperature increasing rate for 10 o C min -1 ), and then cooled to 50 o C. After this procedure, CO 2 was absorbed by COF for 30 min at 50 o C, and then CO 2 desorption was analyzed at different temperature (from 50 to 350 o C). The CO 2 -TPD curve was obtained by a TCD detector. X-ray photoelectron spectroscopy (XPS) measurements were performed on the Thermo Scientific K-Alpha electron energy spectrometer with Al Kα (1486.6 eV) radiation as the X-ray excitation source. All binding energies were referenced to the C 1s peak (284.6 eV) based on adventitious carbon.
The inductively coupled plasma-mass spectrum (ICP-MS) of metal elements was recorded on the Agilent 7700 spectroscopy. Electrochemical tests were performed by the CHI 760E electrochemical working station (Shanghai). The steady-state photoluminescence (PL) spectra and PL decay spectra were measured by FLS980 Fluorescence Spectrometer (UK). Solid-state UV-vis diffuse reflectance spectra of the samples were collected on a Perkin Elmer Lambd 950 spectroscopy (USA) using BaSO 4 as the reference standard. Thermogravimetric analysis (TGA) was conducted from 25 o C to 800 o C under N 2 protection with a heating rate of 10 o C min -1 using a NETZSCH STA449C thermal analyzer. The ultra-fast femtosecond time-resolved transient absorption (fs-TA) spectra were measured by an Helios pump-probe system (Ultrafast Systems LLC) coupled with an amplified femtosecond laser system (Coherent, 35 fs, 1kHz, 800 nm).
The probe pulses (from 400 to 700 nm) were produced by focusing a small portion (around 10 μJ) of the fundamental 800 nm laser pulses into 1 mm thickness of rotated S4 CaF 2 . The 365 nm pump pulses were generated from an optical parametric amplifier (TOPAS-800-fs).

The effect of different dosages of Co-2,3-DHTA-COF on the CO production
The photocatalytic CO 2 RR experiments with different dosages of Co-2,3-DHTA-COF (0.50, 1.00, 1.50, 2.00, 5.00 and 10.00 mg) were completed to systematically examine the effect of catalyst dosage on the production of CO. It was found that the CO production in 4 hours was 0.078, 0.072 and 0.072 mmol when the amount of Co-2,3-DHTA-COF used in this photocatalytic system was 0.50, 1.00 and 1.50 mg, respectively, indicating that the increase in catalyst dosage did not significantly affect CO production. However, the CO production decreased to 0.060, 0.054 and 0.051 mmol when the dosage of Co-2,3-DHTA-CO was increased to 2.00, 5.00 and 10.00 mg. This result suggests that when catalyst dosage was >1.50 mg, the catalytic performance of the COF powder was reduced, which means that the catalyst did not fully take effect in the catalysis.
In fact, the CO 2 photoreduction by COF is a heterogeneous reaction and was conducted at the gas-solid-liquid interfaces. Importantly, the COF material is very light and fluffy, which usually floats on the top of reaction mixture in the liquid-solid mode reaction. Even under stirring, the contact of COF catalyst with components in the catalytic system is very limited. These results in two main problems: one is the low light utilization efficiency due to the strong light-shading effect; and another problem is the limited contact of the COF catalyst with the reactant and intermediate. Thus, the observed decrease in catalytic efficiency is not only related to the light-shading effect due to the increased dosage of the COF powder, but also related to the limited contact area of the catalyst with the reactant and intermediate 1 . In other words, optimal dosage of COF catalyst is necessary for the photoreduction reaction, however, excess catalyst would have a negative effect for the reaction owing to the factors mentioned above. This may be the possible reason why low dosage of catalysts was usually used in the previous studies and in this work

Error analysis of CO production rate
In order to obtain reliable CO production rate data, accurate mass determination of photocatalyst Co-2,3-DHTA-COF is very important. Therefore, a high precision S5 electronic balance with resolution of ±0.00001g (±0.01mg) was used in this work. We estimated the related experimental errors in weighing and production rate of CO by performing five independent reduction experiments catalyzed by Co-2,3-DHTA-COF, and the result was listed in Supplementary Table 8. The calculated standard deviation was ± 0.02 mg for the mass of Co-2,3-DHTA-COF and ± 0.210 mmol·g -1 ·h -1 for the CO production rate.

Turnover number and CO selectivity
The turnover number (TON) was calculated by supplementary Eq. S1 2-3 : where n co and n active site are molar number of the CO product and active sites of the photocatalyst, respectively, and t is the reaction time (hour). The selectivity of CO was calculated by Eq. S2: 4 where n co and n H 2 refer to the molar number of CO and H 2 products under visible light irradiation (λ > 420 nm).

Apparent quantum efficiency
The apparent quantum efficiency (AQE) of the catalysts was measured by different bandpass filters (including 420 nm, 450 nm, 500 nm, 520 nm, 550 nm, 600 nm and 630 nm) under the same photocatalytic reaction conditions, and the light intensity was detected by a Newport 91150-2000 optical power meter (USA Newport corporation). The AQE values were calculated by Eq. S3 [5][6] : where n co is the molar number of the CO, N A is Avogadro's constant (6.022×10 23 mol -1 ), h is the Planck's constant (6.63×10 -34 m 2 kg s -1 ), c is the speed of light (3×10 8 m· s -1 ), S is the irradiation area (cm 2 ), P is irradiation intensity (W cm -2 ), t is irradiation time (s), and λ is the wavelength of the light source, respectively.

DFT calculations
All the calculations were performed at the DFT level by using the B3LYP-D3 hybrid S6 functional 7,8 . In B3LYP-D3, a semiempirical dispersion potential to the traditional Kohn pseudo-DFT energy was included. For metal Co, the LANL2DZ basis set was used 9 , while the basis set 6-31G(d, p) was used for nonmetals H, C, N and O elements 10 .
Geometry optimizations were conducted with the CPCM solvation model (CH 3 OH, ɛ 32.70) 11 . All the DFT calculations were carried out by the Gaussian 09 program 12 . The redox potential E 1/2 corresponding to the half-reaction (O + ne → R) was given by the Nernst equation 13 : where ∆G m θ is the standard Gibbs free energy change of the half-reaction (O + ne → R), F is the Faraday constant, and E SHE   Tables 1-8 Supplementary