Perovskite Photocatalytic CO2 Reduction or Photoredox Organic Transformation?

Abstract Metal‐halide perovskites have been explored as photocatalysts for CO2 reduction. We report that perovskite photocatalytic CO2 reduction in organic solvents is likely problematic. Instead, the detected products (i.e., CO) likely result from a photoredox organic transformation involving the solvent. Our observations have been validated using isotopic labeling experiments, band energy analysis, and new control experiments. We designed a typical perovskite photocatalytic setup in organic solvents that led to CO production of up to ≈1000 μmol g−1 h−1. CO2 reduction in organic solvents must be studied with extra care because photoredox organic transformations can produce orders of magnitude higher rate of CO or CH4 than is typical for CO2 reduction routes. Though CO2 reduction is not likely to occur, in situ CO generation is extremely fast. Hence a suitable system can be established for challenging organic reactions that use CO as a feedstock but exploit the solvent as a CO surrogate.


General materials and instruments
All commercially available reagents and solvents used in this study were purchased from TCI, Fisher or Sigma Aldrich and used without further purification. The labeled solvent was purchased from Cambridge Isotope Laboratories (CIL). Labeled CO2 gas was from Sigma-Aldrich. Respective gas tanks, including ultra-pure CO2, were from Air Gas. Gas Flowmeters, 0.1 To 1 Lpm, Viton Seal were purchased from Grainger.
GC measurements were conducted on Shimadzu GC-2010 Plus Tracera with a barrier discharge ionization detector using He (99.9999%) as a carrier gas. PL lifetime measurements were conducted with a DeltaPro TCSPC Lifetime Fluorometer with a 371 nm excitation wavelength. PL spectra were recorded with a Horiba Jobin Yvon Model FluoroMax-4.

Perovskite CsPbBr3 NC synthesis
The NC synthesis procedure is adopted from our previous publications. [1] CsPbBr3 nanocrystals (NCs) were synthesized by modifying the hot injection method previously reported. First, a Csoleate solution was prepared by charging a 100 ml 3-neck flask with Cs2CO3 (0.16g) along with octadecene (6 mL, ODE) and oleic acid (2.5 mL, OA) and dried for one hour under vacuum at 120 o C. The Cs-oleate solution was then followed by N2 sparging at 150 o C until all of the Cs2CO3 dissolved in ODE. In a separate 100 ml 3-neck flask, 10 mL ODE and PbBr2 (0.178 g, 0.486 mmol) are dried for one hour under vacuum at 120 o C and subsequently purged with N2 followed by an injection of both oleylamine (1 mL) and OA (1 mL) after the temperature was raised to 150 o C. Once the Pb salts dissolved, the temperature was raised to 180 o C and the prepared Cs-oleate solution (1 mL) was swiftly injected into the reactor. After five seconds, the yellow-green reaction mixture was cooled by an ice bath and subsequently washed with methyl acetate (10 mL). After centrifuging at 9000 rpm for five minutes, a yellow-green precipitate was obtained and was further washed with hexanes (10 mL) prior to centrifugation again. The NCs were then vacuum oven dried overnight. The yield was less than 10%.

Grinding bulk perovskite CsPbBr3 synthesis
Ground CsPbBr3 was produced by grinding 1mmol of CsBr with 1 mmol PbBr2 using a mortar pestle. After 30 minutes of firm grinding, ~578 mg of yellow microcrystals were produced with moderate green PL under UV-light. Such bulk perovskite power was annealed in the oven at 130°C to remove any possible organics before applying to the photocatalytic reaction.

Characterization
Perovskite CsPbBr3 NCs and grinded bulk CsPbBr3 were characterized according to our previous method. [1] The respective characteristic XRD, absorption, photoluminescence, and TEM images are illustrated below. Hitachi H-7500 transmission electron microscope was utilized to measure the TEM images. A Philips Empyrean X-Ray Diffractometer was employed to measure the powder XRD samples.   After light irradiation for a desired period of time, a Hamilton gas-tight 50 µL syringe was used to inject 30 uL of the reaction headspace into the GC. The inlet, 0.53mm ID column, and detector were maintained at 200°C, 40°C, and 235°C, respectively. The respective CO or CH4 amount was determined by the detected respective gas area using the calibration curve above.

Calculation of gas rate: umol/g/h
Gas production rate, umol/g/h was determined using the following equation:

Perovskite photocatalytic experimental setup
A typical photocatalytic experiment has been set up as follows: 1 mg of catalyst is added to a 4 ml vial as well as 2 ml of solvent (i.e., ethyl acetate) containing a stir bar. (Note: 6 uL of water was used if the experiment needed water). Gentle sonication was then used to suspend the catalyst in the solution. The desired gas (i.e., CO2 or O2) was then bubbled into the reaction at a flow rate of ~0.1 L/min for 10 minutes to saturate the system. The reaction was then sealed with septum and placed on a stir plate while irradiated by 456 nm Kessil LEDs, intensity (40W).

Band tuning of perovskite for photocatalytic CO generation.
Perovskite CsPbBr3 has been tuned with TMSI using our previous method [1] , leading to the formation of CsPbBrxI3-x. Highly exchanged CsPbBrxI3-x NCs with x approaching 3, has also been achieved using largely excess amount of iodide source enabling a VB band close to +1.1 vs. RHE. CsPbI3 NCs were also synthesized to in a similar manner except PbI2 was used instead of PbBr2. [2] Such perovskite has been employed in photocatalytic reaction set up as follows: 1 mg of catalyst is added to a 4 ml as well as 2 ml of solvent (i.e., ethyl acetate) containing a stir bar. (Note: 6 uL of water was used if the experiment needed water). Gentle sonication was then used to suspend the catalyst in the solution. The desired gas (i.e., CO2) was then bubbled into the reaction at a flow rate of ~0.1 L/min for 10 minutes to saturate the system. The reaction was then sealed with septum and placed on a stir plate while irradiated by 456 nm Kessil LEDs. After LED illumination, a significant amount of CO has also been observed using this band-tuned perovskite materials with rates reaching up to 2.5 umol/g/h respectively.

Additional Control experiments to explore O2's impact on the photocatalytic outcome.
Air or pure oxygen environment has been employed for control studies with comparison to CO2, and N2 atmosphere under the same photocatalytic setup.
In addition, O2's specific impact has been explored. Before LED illumination, the photocatalytic reaction vial was purged with two needles with two controlled gas flowmeter rates for over 20 minutes. And then the reaction vial was then carefully sealed for photocatalytic illumination. And the headspace gas was detected and measured in the same way as a typical photocatalytic experiment. Note that the CO production rate in this comparison was measured on a 6-hour time scale.

Label experiments.
Conditions: 1mL ethyl acetate solvent with 1mg perovskite nanocrystals (~10nm), saturated using various gas for 5 min and sealed with septum, illuminated under 456 nm Kessil LED. Aliquots of the gas from headspace were applied using a gas-tight syringe and detected by GCMS. 13 C label using 99 atom% ethyl acetate (From CIL, Cambridge Isotope Laboratory) or 99%atom 13 CO2 (from Sigma). The details for GCMS data were shown below.

Photoluminescence quenching experiments
The samples were prepared by suspending 1 mg of CsPbBr3 NCs in 3mL of hexane. Then, they were saturated using various gases and sealed with septum for further PL measurement.