Origin of the Overpotential for Oxygen Reduction at a Fuel-Cell Cathode
J. K. Nørskov, J. Rossmeisl, A. Logadottir, L. Lindqvist, J. R. Kitchin, T. Bligaard, and H. Jónsson, “Origin of the Overpotential for Oxygen Reduction at a Fuel-Cell Cathode”, The Journal of Physical Chemistry B 108(46), 17886-17892 (2004). https://doi.org/10.1021/jp047349j
This summary was written by a language model from the paper and has not been checked line by line. Rely on the published paper for details.
Question
In low-temperature fuel cells the cathode reaction, in which O2 is reduced by protons and electrons to water, is far slower than hydrogen oxidation at the anode, and even on Pt it carries a substantial overpotential. When the paper was written there was no consensus on why. The authors ask whether density functional theory (DFT) energetics of the surface intermediates can explain this overpotential on Pt(111), and whether the same picture accounts for the trends in oxygen reduction activity across transition and noble metals (Nørskov et al. 2004).
Approach
The central methodological contribution is a general recipe for putting electrochemical intermediates on a free energy scale from ordinary DFT adsorption energies. The proton-electron pair is referenced to half of gaseous H2 by taking the standard hydrogen electrode as the zero of potential, so at U = 0 and pH 0 the free energies of water dissociation into adsorbed OH or O equal those of the reverse reduction steps. States involving an electron are shifted by -eU, a pH correction of kT ln 10 per pH unit is applied to protons, and zero point energy and entropy corrections are added. A water layer is included to capture hydrogen bonding, which stabilizes OH but barely affects O. The overall reaction free energy is taken from experiment (2.46 eV per O atom, equivalent to 1.23 V) to avoid the poor DFT description of the O2 molecule (Nørskov et al. 2004).
The calculations use a plane wave code with ultrasoft pseudopotentials and the RPBE functional on close-packed slabs. The authors first analyze a dissociative mechanism (O2 splits, then O is hydrogenated to OH and to water) on Pt(111) at zero and one-half monolayer oxygen coverage, then compute O and OH binding energies on the close-packed surfaces of Ag, Au, Co, Cu, Fe, Ir, Mo, Ni, Pd, Pt, Rh, Ru and W. A Sabatier-type measure of maximal activity, the minimum over elementary steps of the log rate constant relative to a nonactivated proton transfer, is used to build volcano plots. Finally they consider an associative (peroxo) mechanism in which molecular O2 is hydrogenated to OOH before the O-O bond breaks (Nørskov et al. 2004).
Key findings
On Pt(111) at low coverage, O and OH are so stable at the equilibrium potential of 1.23 V that both proton-electron transfer steps are uphill by about 0.45 eV, which the authors identify with the observed overpotential. All steps become downhill only below about 0.78 V, which coincides with the measured onset of water dissociation, so the two phenomena are presented as one. Assuming the barrier equals the largest uphill step and taking a prefactor of about 200 per second per site from measured proton transfer rates, a model with no parameters fitted to oxygen reduction data gives a polarization curve similar to fuel cell measurements and a Tafel slope of 60 mV at 300 K (71 mV at 357 K). At half monolayer oxygen coverage the energies shift but the limiting barrier is nearly unchanged (Nørskov et al. 2004).
Across metals, activity plotted against oxygen binding energy forms a volcano with Pt and Pd at the top, consistent with experiment. Metals such as Ni bind O and OH too strongly, so proton transfer is strongly activated, while on Au oxygen is less stable than in the gas phase and O2 dissociation has a large barrier. The volcano implies that a surface binding oxygen somewhat more weakly than Pt should be more active. DFT values for Pt overlayers on Ni, Co and Fe, and the subsurface 3d metal calculations of (Kitchin et al. 2004), support this, rationalizing the higher activity of Pt-skin alloys, while steps such as Pt(110), which bind O about 0.5 eV more strongly, should not help. O and OH binding are roughly linearly correlated, and the authors note that breaking this correlation could yield new catalysts. The associative mechanism has lower barriers on Au(111) at low coverage and on Pt(111) at high coverage, dominates below roughly 0.8 V, and makes the most noble metals follow a peroxide route, yet the volcano maximum hardly changes (Nørskov et al. 2004).
Significance
The authors present the free energy method as completely general and applicable to any electrochemical reaction; it is the scheme now commonly referred to as the computational hydrogen electrode, and it underlies later work in this wiki on hydrogen and oxygen evolution. The paper gave a thermodynamic explanation for the Pt overpotential and for why moderately weakening oxygen binding on Pt, for example in Pt-skin alloys with 3d metals, improves activity.