A robust Fe-based heterogeneous photocatalyst for the visible-light-mediated selective reduction of an impure CO2 stream

The transformation of CO2 into value-added products from an impure CO2 stream, such as flue gas or exhaust gas, directly contributes to the principle of carbon capture and utilization (CCU). Thus, we have developed a robust iron-based heterogeneous photocatalyst that can convert the exhaust gas from the car into CO with an exceptional production rate of 145 μmol g−1 h−1. We characterized this photocatalyst by PXRD, XPS, ssNMR, EXAFS, XANES, HR-TEM, and further provided mechanistic experiments, and multi-scale/level computational studies. We have reached a clear understanding of its properties and performance that indicates that this highly robust photocatalyst could be used to design an efficient visible-light-mediated reduction strategy for the transformation of impure CO2 streams into value-added products.


S1. Experimental procedures S1.1 Chemicals
Dry acetonitrile (99.9%), dicyandiamide (DCDA) and other reagent grade solvents were purchased from thermo-scientific.Iron (III) nitrate nonahydrate and triethanolamine (TEOA, ≥99%) and 1-benzyl 1,4-dihydronicotinamide (BNAH) were obtained from Sigma-Aldrich.Triethylamine (TEA, ≥99%) was obtained from Acros.All reagents were used without further purification except triethylamine.In catalytic reaction the triethylamine is used after purification.S1.2 Preparation of catalyst S1.2.1.Preparation of g-C 3 N 4 9 g of DCDA was calcined at 550 °C for 4h (temp.increasing rate = 2.2°C/min) in a tube furnace under aerobic conditions.The sample was removed when the furnace temperature was cooled down to room temperature (24 °C).After calcination, around 3g of g-C 3 N 4 catalyst was prepared and then was grinded to fine powder in an algae mortar.

S1.2.2 Preparation of f-gC 3 N 4
A 100 mL beaker containing 9 g of DCDA, and 150 mg of 2-amino-5-trifluoromethyl benzonitrile and deionized water (45 mL) was stirred at 95 °C until it's completely dried.Resulting mixture was then grinded in an algae mortar and was calcined at 550 °C for 4h (temp increasing rate = 2.2°C/min) under aerobic conditions.The sample was removed when the furnace temperature was cooled down to room temperature (24 °C).After the calcination, the around 3g prepared material was grinded in an algae mortar.S1.2.3 Preparation of Fe@f-gC 3 N 4 (0.2Fe@f-gC 3 N 4 , 0.5Fe@f-gC 3 N 4, 0.7Fe@f-gC 3 N 4 and 1Fe@f-gC 3 N 4 ) In a 20 mL glass tube, 18 mg of Fe(NO 3 ) 3 •9H 2 O and 500 mg of f-gC 3 N 4 were added in 10 mL deionized water and the reaction mixture was stirred at 100°C until it's completely dried.

Figure S 1.
Photographs of the different catalysts after synthesis.

S1.3 Purification of triethylamine:
100 mL of triethylamine was stirred and distilled in the presence of 2.6 gm of calcium hydride.
After the distillation, TEA was stored under nitrogen atmosphere in a dry schlenk flask over molecular sieves.

S1.4 Photocatalytic CO 2 reduction
Figure S 2. Scheme of the experimental photocatalytic system.Photochemical reactor: gasliquid system where the catalyst is heterogeneous (solid), the sacrificial reagent is homogeneous.
A 28 mL Schlenk tube containing 1 mg of the catalyst was capped with the rubber septum after the vacuum (3 minutes) and N 2 flow (30 seconds) to make anaerobic atmosphere inside the tube.This process was repeated for three times.After that, 0.2 mL of distilled TEA and 3.8 mL of dry ACN were added in the Schlenk tube.The solvent and TEA were added to the Schlenk tube in the presence of N 2 atmosphere.The reaction solution was then bubbled with CO 2 (99.9% pure) in the dark for 25 minutes.After CO 2 bubbling, a CO 2 balloon was kept on this Schlenk tube using a needle (needle above on the solution).The CO 2 saturated solution was then irradiated at λ = 427 nm kessil lamp (light intensity was 100 mW/cm 2 and distance between lamp to the Schlenk tube was ~1 cm, distance of fan from light set up was ~15 cm) with vigorous stirring at 30 • C temperature for 18 h.The light set up containing two lights where each light was face to face parallel to each other and one Schlenk tube was placed in front of another Schlenk tube.After the reaction, gaseous products were measured by headspace GC machine (GC, Isbuan 345, model no: 1300, Global Analyser solution).Reaction conditions: 1Fe@f-gC 3 N 4 (1 mg), TEA (0.8 mL), ACN (3.2 mL), Time = 18 h, Wavelength of light (390 -427 nm), Reaction temp = 30˚C.The components of exhaust gas vary depending on the type of engine (e.g., gasoline, diesel) and the fuel used, but typically include a mixture of gases and particulate matter.

S2.3 The collection procedure and composition of exhaust gases
Others contains: ~12% water vapor.
Trace amounts of sulfur dioxide (SO₂) Some amounts of particulate Matter (PM).The data was processed using the Athena program. 3Scans were calibrated, aligned and normalized with background removal.

S3.3 Solid state NMR spectroscopy
Magic-angle-spinning (MAS) NMR experiments were performed at a magnetic field of 14.1 T (Larmor frequencies of 600.12, 150.92, 60.83, and 564.69 MHz for 1 H, 13 C, 15 N, and 19 F respectively) on a Bruker Avance-III spectrometer.The 1 H and 19 F MAS NMR spectra were acquired using a 1.3 mm probe head and a 60 kHz MAS rate.This acquisition involved a use of a rotor-synchronized, double-adiabatic spin-echo sequence with a 90° excitation pulse of 1.25 µs (2.30 µs for 19 F) followed by a pair of 50.0 µs tanh/tan short high-power adiabatic pulses (SHAPs) with 5 MHz frequency sweep. 4,5 ll pulses operated at the nutation frequency of 200 kHz (110 kHz for 19 F). 64 signal transients (10240 for 19 F) were acquired using a relaxation delay ranging from 0.1 to 5 s.Cross-polarization (CP) 1 H- 13 C and 1 H- 15 N CPMAS NMR spectra were recorded using a 7 mm probe head with a 7 kHz MAS rate, 65 kHz spinal and 64 proton decoupling.For 1 H- 13 C CPMAS acquisition, Hartmann-Hahn matched radiofrequency fields were applied for a contact interval of 1.5 ms and 2048 signal transients were collected using a relaxation delay of 5 s.The 1 H-15 N CPMAS acquisition involved contact interval of 5ms, and 16384 scans collected with relaxation time of 5 s.Chemical shifts are reported with respect to TMS ( 1 H, 13 C), nitromethane ( 15 N), and trichlorofluoromethane ( 19 F).

S3.4 XPS spectra of the fresh and reused samples
The surface composition was studied by X-ray photoelectron spectroscopy (XPS).The spectra were collected on a Prevac spectrometer equipped with a hemispherical VG SCIENTA R3000 analyzer (pass energy of 100 eV) and a monochromatized aluminum source AlKα (1486.6 eV).Binding energies were calibrated using the C 1s line at 284.8 eV.After a Shirley background subtraction, raw spectra were fitted using Gaussian-Lorentzian peak shapes in the Casa XPS software.S3.5 EPR spectroscopy EPR measurements were conducted at RT using a Bruker EMX CW micro-X-band spectrometer (microwave frequency ≈ 9.87 GHz) equipped with an ER 4119HS-WI highsensitivity optical resonator with a grid in the front side for irradiation the sample inside the EPR cavity.All the samples were irradiated by a Kessil lamp (λ = 427 nm) and measured by EPR before and after irradiation using a modulation frequency of 100 kHz.

Fluorescence spectroscopy
The emission spectra were recorded on an Edinburgh FLS980 under 365 nm excitation by a Xenon lamp.The TCSPC is equipped with a pulsed laser, Nd: YAG (Quanta-Ray INDI-40, Spectra-Physics), and an optical parametric oscillator.The pulse duration is fixed at around 2 ps, and the time resolution of each experiment was about 100 ns.A beam splitter is used in the pathway to split the beam towards a photodiode to generate a start signal and excite the sample.The photons are collected, filtered, and focused on the entrance slit of a 30 cm focal length spectrograph (SpectroPro-300i Acton) and detected through a photomultiplier tube (Hamamatsu, R928).The transient electrical signal is then displayed by an oscilloscope connected to the control computer.All the samples were excited at 360 nm with the excitation power intensity set at 12 μW cm -2 (0.2Fe@f-gC 3 N 4 and f-gC 3 N 4 were excited at the power of 4 μW cm -2 due to high photon counts).The decays were fitted using a tri-exponential fitting equation, and the average PL lifetimes ( av ) were calculated using an intensity-weighted  equation: Here, A i is the amplitude fraction and is the fluorescence lifetime. 

S4. Theoretical calculation Computational Details
To identify possible configurations of the Fe@f-gC 3 N 4 -CO 2 complex and reaction mechanisms leading to O 2 in acetonitrile (ACN), we used classical reactive molecular dynamics simulations (RMD) 17 based on a force field already tested 18 combined with DFT calculations.We used a polymeric carbon nitrides model consisting of fourteen melems with one f-substituent and three Fe ions (Figure S19).We surrounded the model with 886 ACN and 16 CO 2 molecules and placed some CO 2 close to the metal centers to favor adsorption.The simulation box (33x35x12 Å 3 ) was replicated in all directions (Figure S20).RMD equilibration simulations were carried out in the NVT/NPT ensembles at ambient temperature and pressure (298 K and 1 atm).Production trajectories were performed in the NVT ensemble, and the system configurations were sampled every 0.01 ps.The ReaxFF version, implemented in the Amsterdam Density Functional (ADF)/ReaxFF 2023.1 package, was used for all the simulations.The time step was 0.2 fs, and the temperature was controlled through the Hoover-Nosé thermostat with a relaxation constant of 0.1 ps.No constraints were imposed on the system.

Scheme S 1 .
Scheme S 1.The schematic of preparation procedure of f-gC 3 N 4 and the functionalization introduced in f-gC 3 N 4 samples.

Figure S 5 .
Figure S 5. Band position of different catalysts.

Figure S 6 .
Figure S 6.The collection procedure of exhaust gasses.

Figure S 7 .
Figure S 7. The exhaust gas components.

Figure S 14 .
Figure S 14. EPR spectra of a) f-gC 3 N 4 ; b) 1Fe@f-gC 3 N 4 in the dark and during irradiation; c) the double integration of the EPR-active photoexcited electrons signal after subtraction of the corresponding spectrum in dark.

Figure S 15 .
Figure S 15.EPR spectra of f-gC 3 N 4 (15 mg) suspension in CH 3 CN (0.5 ml) in the presence and absence of TEA (CH 3 CN: TEA = 19:1) measured in the dark and during irradiation.

Figure S 17 .
Figure S 17. Photoluminescence studies of the photocatalysts.(a) Emission spectra, and (b,c) lifetime profiles of gC 3 N 4 , Fe-gC 3 N 4 , and f-gC 3 N 4 (b), Fe-loaded f-gC 3 N 4 photocatalysts (c) with their corresponding fit shown with a solid yellow line.The powdered samples for lifetime studies were excited at 360 nm with an excitation power intensity of 12 μW cm -2 .

Figure S 19 .
Figure S 19.Initial catalyst structure used for the ReaxFF MD simulations.Ball and stick model Color code: C gray, N blue, F green, Fe orange, H white.The Fe ions maintained their chelated arrangements between the nitrogens of adjacent triazine rings.

Figure S 20 .
Figure S 20.Top.Perturbed RMD: catalI structure surrounded by ACN (grey lines) and CO 2 (ball and stick model) molecules.Color code: C gray, N blue, F green, Fe orange, O red, H

Figure S 21 .
Figure S 21.Top.Distributions of the atomic partial charges of all the Fe atoms during the perturbed RMD simulations.The black plot corresponds to the Fe atom with a CO 2 molecule coordinated through its carbon S20).The evolution of the charge is shown in the bottom plot.Bottom.Evolution of the partial charges of the carbon atom of the adsorbed CO 2 (green line) and the connected Fe (black line).

Figure S 22 .
Figure S 22. Representative Snapshots of the CO 2 reduction mechanism disclosed by the perturbed RMD simulations.ACN (grey lines) and melem/CO 2 molecules (ball and stick model).Color code: C gray, N blue, F green, Fe orange, O red, H white.

Figure S 23 .
Figure S 23.NBO charge analysis of states 1-5 is shown in Fig. 5 according to a partition in single atoms (a) or aggregated form (b).

Figure S 24
Figure S 24 Total and Projected Density of States (DOS) corresponding to Spin Up (left column) and Spin Down (right column) of states 1-5 of the high-coordination site shown in Fig. 5 of the main article (structures are also reported as in-set pictures).

Figure S 25 .
Figure S 25.Simulated UV-vis spectra of the complex minimum energy structures.Color code: C gray, N blue, F green, Fe orange, O red, H white.

Table S1 .
Photocatalytic reduction of CO 2 by different catalysts.

Table S 2
Photocatalytic CO 2 reduction at different wavelengths.

Table S 3
Fitting parameters for lifetime profiles of photocatalysts.

Table S 4
. Representative examples of reaction reported for photocatalytic CO 2 reduction.