Thanks to its wide availability, CO₂-neutrality and intrinsically renewable nature, biomass is emerging as a promising substitute to fossil fuels, and biomass electro-oxidation as an encouraging production route for hydrogen and value-added chemicals. To catalyse these electrochemical oxidation reactions, platinum group metals have emerged as promising candidates, showing good activity at low overpotentials. However, noble metals suffer from rapid and drastic activity drops, caused by the poisoning nature of some adsorptive species. The first step, to tackle this issue and to allow high reaction rates to be sustained for longer times, is the identification of these poisonous intermediates. This insight could guide either the development of mildly binding, poison-resistant catalyst, or the identification of potential pulsing protocols to recover the active sites.

For the specific case of platinum, while it is generally accepted that the formation of platinum oxides is responsible for the rapid Pt deactivation above 0.8V vs RHE, a similar activity decline is observed at lower potentials, the cause of which is less clear. In the case of glycerol oxidation, while the activity at 0.8V can be recovered by pulsing to mildly reductive potential (≈0.4V), this kind of pulsing has no effect on the catalytic activity at 0.6V, which can however be recovered by oxidative pulsing (1.2V). In a recent attempt to identify the poisonous intermediates, Chen et al. have proposed that glyceric acid could be blocking the surface at 0.6V, and reported increased glyceric acid selectivity by reductive pulsing.¹  Expanding on their work, here we combined surface enhanced infrared spectroscopy and potential pulsing to identify poisoning intermediates and improve the catalyst stability.

Our results show that not only carboxylate species, but also linear (1930-1970 cm^-1) and bridge-bonded (1700-1790 cm^-1) CO accumulate during both glycerol and formate oxidation. These CO intermediates show a blue shift compared to pure CO, of around 100 cm^-1, and twice as high Stark tuning slopes (of 120-50 cm^-1). This suggests the presence of partially hydrogenated CO adsorbates, previously observed in methanol oxidation.² What’s more, this hydrogenated form of CO appears to poison the active sites up to potentials as high as 0.9V vs RHE, largely above the oxidation potential of pure CO.

Our study shows how hydrogenated CO is an overlooked cause of Pt deactivation, and common to several oxidation reactions. The identification of the poisoning intermediates also provides insights into alleviating the blockage of sites, as for the case of CO, potential pulsing above 1.2V vs RHE was found to be an effective approach to reactive the catalyst.