CHEMELECTROCHEM
ARTICLES www.chemelectrochem.org
ducting polymers used in PEM electrolysers could be replaced
by hydroxide-conducting membranes; indeed, several materials
have recently been discovered in alkaline electrolytes with
a catalytic activity at least as high as that of RuOx and IrOx.9–12
In Figure 1b, the best-performing OER catalysts in alkaline
media are shown in a unified Tafel plot (it should be noted
that in this plot the activity of some of the catalysts has been
reported as a current density normalised according to the microscopic
surface area, denoted by the symbol # in the plot,
whereas other catalysts have been normalised according to
the geometric surface area). It is clear from the plot that the
most active catalysts are primarily based on Ni or Co. Even so,
the use of alkaline polymeric membranes comes at the cost of
increased overpotential for hydrogen evolution, and lower
membrane conductivity and stability.13–15
Regardless of the pH of the electrolyte, robust OER catalyst
benchmarks are required that allow a straightforward comparison
of catalyst performance between different experimental
groups.4, 7 Critical parameters include: 1) the geometric activity,
that is, the current normalised according to the geometric
or projected area, 2) the mass activity, that is, the current per
unit mass precious metals, 3) the specific activity, the current
normalised according to the microscopic area, and 4) the stability
of the catalyst.
In a commercial device, it is essential that the geometric activity
is maximised, to minimise overpotentials. Moreover,
should precious metal oxides be employed, a high geometric
activity should not be reliant on significant loadings of scarce
elements, that is, the mass activity should also be maximised.
Both the geometric activity and mass activity can be maximised
by employing materials with a high specific activity and
a high surface area. However, to judge whether a material is intrinsically
active for a reaction, knowledge of the specific activity
becomes important. This metric, in turn, is dependent on an
accurate knowledge of the electrochemically active surface
area, which is challenging to measure on oxides. Alternatively,
the catalyst activity can be assessed using smooth thin films,
where the microscopic surface area is as close as possible to
the geometric surface area.20, 26, 27
The procedures for assessing the stability of OER catalysts
are not well established in the literature; this is in contrast to
the reverse of the OER, the oxygen reduction reaction (ORR),
where detailed tests for assessing the catalyst stability have
been developed to simulate the conditions required for automotive
applications28 . Thus far, most researchers have assessed
the stability under OER conditions for a limited number
of hours, by performing chronopotentiometry at a constant
current density or chronoamperometry at a constant potential.
However, it remains questionable whether such measurements
can provide the basis upon which one could judge the longterm
performance of a catalyst in a real device over the required
lifetime, that is, a number of years.
A number of methods exist to monitor catalyst corrosion.29
Microscopic techniques, such as scanning tunneling microscopy
30, 31 and transmission electron microscopy,32–35 can monitor
changes in the electrode morphology and structure. On the
other hand, macroscopic techniques can be applied to determine
the corrosion rates; these include the rotating ring disk
electrode (RRDE; for example, for monitoring the anodic dissolution
of RuO2
36, 37), the quartz crystal microbalance,38 and inductively
coupled plasma mass spectrometry (ICP–MS).39–41
Nonetheless, as of yet, no standardized protocols for assessing
the stability under OER conditions have emerged.
Herein, we present guidelines for establishing the stability of
OER electrocatalysts. By combining standard RDE tests with
electrochemical quartz crystal microbalance42 (EQCM) measurements
and inductively coupled plasma mass spectrometry
(ICP–MS), we provide a detailed description of corrosion processes
that take place in parallel to the OER. The catalysts investigated
are RuO2 and MnOx. RuO2 is an extensively studied
material with a high activity in acidic electrolyte.16, 43–45 However,
the stability of RuO2 is limited at high overpotentials. MnOx
has been proposed as a more abundant and inexpensive alternative
to RuO2 ;46–49 not only is it active for the OER, but also
for the ORR, opening up possibilities for its use in regenerative
fuel cells.50 Manganese can form numerous oxides and many
of these have been reported active for OER in alkaline and
neutral electrolytes.51 However, as for ruthenium dioxide, the
stability can be an issue at high overpotentials.
Experimental Section
Preparation of Thin Films
Thin MnOx and RuO2 films were prepared by reactive sputter deposition
on Au polycrystals and EQCM crystals. The deposition rates
were calibrated with an in-chamber QCM. Prior to deposition, the
samples were sonicated in acetone, isopropanol, and then Millipore
water (18.2 MW). RuO2 films were deposited at 3008C and
3 mTorr with a power of 50 W using an argon and oxygen flow at
a ratio of 5:2, with a metallic Ru target. MnOx films were deposited
at 2008C, 5 mTorr, and 140 W with an argon and oxygen flow at
a ratio of 5:1, and a metallic Mn target. The EQCM crystals were
purchased from Stanford Research Systems (QCM200) and consist
of a gold film deposited onto AT-cut quartz with a titanium layer in
between for improved adhesion. The top electrode, functioning as
working electrode for the electrochemical measurements, has
a geometrical surface area of 1.37 cm2. The bottom electrode is
smaller, 0.38 cm2, and the QCM is sensitive only in the overlapping
region of the top and bottom electrodes. This means that approximately
28% of the electrochemically active layer is sensitive to the
QCM measurement. The frequency change is converted to mass
change using the Sauerbrey equation, as explained below. In this
equation, a homogeneous mass change across the electrode is assumed.
Because of the semiconducting nature of manganese
oxides, we used a mask to confine the MnOx area to where the
QCM is sensitive (the central 0.38 cm2 of the top electrode). This
was done to diminish effects of local potential differences, for example,
caused by gradients in film thickness. The remaining part of
the gold film was covered with TiO2, which introduced negligible
currents and no frequency change during the stability tests. As
RuO2 is expected to be a metallic conductor, all of the gold film
was covered with RuO2. The Au polycrystalline electrodes
(0.196 cm2) used in RDE tests were polished prior to deposition
with 0.25 mm diamond paste, then plasma cleaned in argon and
annealed to 7008C in two consecutive cycles. The targets for sputtering
had a 99.95% purity and were purchased from AJA
International.
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemElectroChem 0000, 00, 1–8 &2&
These are not the final page numbers!