Speaker
Description
Transition metal-nitrogen-doped carbons (TM-N-C) are promising catalysts for several important electrochemical processes, including CO$_{2}$ electrocatalytic reduction (CO$_{2}$RR) [1]. In these catalysts, nitrogen is incorporated into a carbon matrix, creating binding sites for metal species. The latter are believed to be the active sites for CO$_{2}$RR. Among different TM-N-Cs, Ni-N-C is most studied for the CO$_{2}$RR reduction to CO, exhibiting high selectivity (Faradaic efficiency FE$_{CO}$ >90%) and significant current density. Several studies suggest that Co-N-C catalysts can have comparable performance for CO$_{2}$RR. However, while there is a consensus regarding the Ni-N-C's performance, the selectivity of Co-N-C catalysts varies widely in literature, with FE$_{CO}$ values ranging from 20% to 100% [2]. This discrepancy suggests that dynamic processes in Co-N-C catalysts is more complex than in Ni-N-C, with more than one Co species coexisting during CO$_{2}$RR conditions.
In this study, operando time-resolved XANES data were used to unveil the local structure around Co sites in their as prepared states, but also under realistic working conditions during CO$_{2}$RR. A multi-step approach has been used for the interpretation of the collected XANES data. First, we identified the number of different coexisting Co species, their corresponding kinetic profiles and XANES spectra using unsupervised machine learning methodologies, such as the principal component analysis combined with a multivariate curve resolution technique [3]. In the second step, we deduced the atomistic structures for each of the identified species through a XANES fitting procedure facilitated by a supervised machine learning approach [4]. Afterwards, we validated the predicted structures by comparing the simulated pre-edge regions with experimental results, gaining additional information on the types of ligands appearing and evolving during reaction conditions.
Our results confirm that, similarly to the Ni-N-C system [5], the single Co sites are the active species for the CO$_{2}$RR, but also reveal the dynamic, heterogeneous nature and adaptation of the system to the reaction conditions. In particular, our data suggest that the local environment around Co is affected by the interactions between the Co site and CO adsorbates but also highlight the formation of of partially reduced Co clusters [6]. The presence of these species, which we believe are not active for CO$_{2}$RR but facilitate parasitic HER, could explain the large discrepancies between the CO$_{2}$RR activities for Co-N-C materials reported in the literature.
Overall, these results demonstrate the potential of XAS spectroscopy in combination with advanced data modelling to access an unprecedented level of understanding of a complex multi-component catalytic system, yielding novel insights into the formation of Co-active sites for the CO$_{2}$RR reaction.
References
[1] J. Timoshenko et al. Chem. Review. 2021, 121 (2)
[2] Q. Fan et al. Adv. En. Mat. 2020, 10 (5)
[3] A. Martini et al. Crystals 2020, 10 (8)
[4] A. Martini et al. Comput. Phys. Commun. 2020, 250.
[5] A. Martini et al J. Am. Chem. Soc. 2023, 145 (31)
[6] A. Martini et al. J. Synchr. Rad. 2024, 31(4).
