CHEMCATCHEM
FULL PAPERS www.chemcatchem.org
In the case of Mn2O3 the binding to all the intermediates are
stronger, resulting in a close to ideal binding of the *O and
*OH intermediates. Upon assuming hydrogen transfer to an
adjacent Au=O site at a Au nanoparticle, the overpotential is
lowered to only 0.2 V. Here the recovery of the hydrogen acceptor
unit and the OO bond formation step require approximately
the same potentials. Further improvements are only
possible with a more ideal hydrogen acceptor. This analysis is
summarized in Figure 4, which includes the volcano plot showing
the activity dependence based on a single descriptor,
DG*=ODG*OH. This volcano arises from limitations of the scaling
relations that result in an overpotential of 0.3–0.4 V at the
peak, as described in the Introduction. Pure Mn2O3 is predicted
to be less active than pure MnO2, but with Au as hydrogen acceptor
the order shifts as more ideal binding energies are
available for reactions (1) and (2) on Mn2O3.
Thermodynamically it is expected that MnO2 is the most
stable phase at OER-relevant potentials, at which a Mn2O3 surface
would be oxidized.27, 42 For Mn2O3, reaction (3) is potential
determining and essentially corresponds to a reduction of
the active site. However, as the binding energy to *OOH is so
weak, the lowest potential path for OER on Mn2O3 is through
an oxidation to MnO2. In the presence of Au (hydrogen acceptors),
the lowest potential would instead occur in the OER on
the Mn2O3 site itself. This suggests that during OER the Mn2O3
sites near Au could exist simply because they can perform the
reduction of the catalytic site. This reducing effect agrees with
the indication from ex situ X-ray absorption spectroscopy that
a lower oxidation state of Mn forms in the vicinity of the Au.38
It also indicates that a very small subset of improved sites are
responsible for the overall increase in current, meaning that
these special sites must be very active. Unfortunately, it is difficult
to assess the quantity of sites with improved catalytic activity
due to the presence of Au, which complicates estimations
of the real decrease in overpotential.
Extending the concept to CoOx, Co3
O4 binds the intermediates
similarly to MnO2, which results in a very similar reaction
profile. Due to some scatter in the binding energies, the overpotential
for the reaction proceeding via CoOOH becomes
only 0.3 eV.10 This is lowered to 0.2 V on assuming a hydrogen
transfer to Au=O. Under OER conditions the most stable phase
for CoOx is b-CoOOH and the most active of the facets is
(0 114).43 Despite significant structural differences between
these cobalt oxides and hydroxides, the redox potentials for
the different oxidation steps are very similar, that is, an overpotential
of 0.40 V is found on assuming a CoOOH intermediate.
This is lowered to 0.3 V when considering the possibility of hydrogen
transfer to Au=O. Similar results are also found for the
(0 00 1) facet. In case of (0 11 2) the oxidation of water to *O is
potential-determining. Correspondingly, no improvements can
be achieved by stabilizing the *OOH intermediate.
Similar to CoOx and MnOx, improvements from using a Au
support have also been reported for nickel oxides.44 For NiO
and NiO2, which lie on the weak binding side of the volcano
plot, reaction (2) is potential-determining. Thus, stabilization of
the NiOOH intermediate through a hydrogen acceptor no
longer results in a lower overpotential. Instead, an improvement
could originate from the same property of Au, that is,
the oxidation potential at which a Au site forms Au=O. Alternatively,
Au can act as an electron sea so that reaction (2) can
proceed at a lower potential. This would be similar to the
effect of doping in, for example, TiO2.45
Conclusions
We propose that hydrogen transfer to an adjacent site significantly
improves catalytic activity in the oxygen evolution reaction
(OER) on Mn and Co oxides. Such an effect can explain
the beneficial interactions between Au and the oxides reported
experimentally. The absolute values of potentials described
here may not be directly transferable to the experimental conditions,
however, the trends indicate enhancements in overpotential
in the order of 100 mV for MnO2 and 300 mV for Mn2O3.
For both Co3O4 and CoOOH the enhancement is approximately
100 mV. These trends are qualitatively consistent with the experimental
results. As an unknown fraction of the total amount
of sites is affected by the addition of Au it is complicated to
compare these results directly to experimental work. It is likely
that, since a small subset of sites is improved, the experimental
enhancement is dampened in comparison to what the theoretical
calculations suggest. Potentially, the OER sites on Mn and
Co oxides close to Au approach the thermodynamic limit for
OER just like the special sites that have an increased OER activity
due to the Ni and Co incorporation on Ni and Co-modified
RuO2.23 Therefore, a huge challenge remains in increasing the
density of these special catalytic sites and stabilizing the
surface.
Figure 4. The theoretical volcano plot obtained for OER proceeding via
*OH and *OOH (c) by using the difference in binding free energies between
the *=O and the *OH, established in Ref. 10 as a descriptor for the
theoretical overpotential in V.a: Potential of a Au=O hydrogen acceptor
that is also the lower limit for overpotentials obtained from interaction with
such a site. *: Theoretical overpotential without a hydrogen acceptor, *: overpotential
including the hydrogen acceptor. Mn2O3 (&/&), MnO2 (*/*), and
Co3O4 (^/^) are placed on the strong binding branch of the volcano. For b-
CoOOH only the (0 11 4) facet (!/!) is on the strong binding, for which an
effect of Au interaction can be expected.
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemCatChem 0000, 00, 1–7 &4&
These are not the final page numbers!