CO Poisoning Effects on FeNC and CN_x ORR Catalysts: A Combined Experimental–Computational Study

Abstract

CO poisoning as a probe for oxygen reduction reaction (ORR) active sites was examined on both iron-nitrogen coordinated catalysts supported on carbon (FeNC) and nitrogen-doped carbon nanostructures (CN x ). Rotating disk electrode (RDE) experiments show a partial decrease in the ORR activity of FeNC catalysts after CO exposure and no decrease for CN x catalysts. Diffuse Reflectance Infrared Fourier transform (DRIFT) Spectroscopy experiments after exposure to CO show a peak at 2033 cm -1 (associated with linearly bound CO) on the FeNC catalysts and no peaks associated with adsorbed CO on the CN x catalyst surface. Density functional theory (DFT) calculations incorporating dispersion interactions of the adsorption energy of O 2 and CO were performed on a total of sixteen proposed active sites for the FeNC and CN x catalysts. On CN x , all the sites show weak CO adsorption and only O 2 molecules are expected to adsorb, which matches the experimental observation of no poisoning. Several FeNC sites show CO binding energies similar in strength to that seen for Pt(111), but only two sites, namely FeN 4 /C pyridinic and FeN 4 /C pyrrolic, are found to bind CO stronger than O 2 . DFT results suggest that the partial poisoning (instead of complete poisoning as found on Pt catalysts) observed in experiments is due to only some fraction of the active sites being blocked. DFT-derived CO stretch frequencies on FeNC show a similar redshift as observed in the DRIFTS experiments, which further confirms that the FeNC catalysts adsorb CO while the CN x catalysts only show weak physisorption.

Improving the Activity of M−N4 Catalysts for the Oxygen Reduction Reaction by Electrolyte Adsorption

“…A possible explanation for this could be that lower concentrations of the halide were employed in these studies, which, according to our calculations for F − and Cl − , could result in only one halide atom binding. The experimental results for CO are likewise inconclusive, with some studies showing no effect, whereas others report a partial decrease in performance . Our calculations suggest that CO would adsorb on one or both sides of the Fe−N 4 motif, depending on the partial pressure, and this would lead to a reduced limiting potential or poisoning, respectively.…”

“…It is of interest to compare our results with experiments reported in the literature, keeping in mind the variation in catalyst preparation and poisoning procedures across these studies. An experiment comparing the activities of M−N/C catalysts with M=Mn, Fe, and Co found that they followed the order Fe>Co>Mn; the half‐wave potentials of the three metals differed by 0.08 V. To match this observation with theoretical calculations, Zelenay et al.…”

“…We furthermore note that our results might not be representative of all types of active sites found in M−N/C catalysts. Previous studies have shown that adsorption energies for different models of the M−N 4 site (e.g., M−N 4 C 10 , M−N 4 C 12 ) can differ by more than 0.50 eV, but that trends are often similar …”

“…The potentially beneficial presence of an adsorbate on one side of the M−N 4 motif is closely related to the detrimental poisoning of the catalyst, in which adsorbates block both sides of the metal atom, leading to a loss of catalytic activity. Poisoning of M−N/C catalysts has been studied experimentally for a number of potential poisons, including halide ions, F 2 , NCS − , H 2 S, CN − , CO, NO x , and H 3 PO 4 . As an example, Wang et al.…”

“…Such variation in the obtained results probably arises as a result of differences in the poisoning procedure, as well as differences in the catalyst preparation schemes, and thereby in the detailed structure of the active site. Indeed, DFT calculations have shown that different types of Fe−N x C sites behave differently towards the adsorption of CO …”