CHEMELECTROCHEM
ARTICLES www.chemelectrochem.org
both tests. The losses evaluated by ICP–MS are higher than
what is seen from the EQCM method. This could be due to the
fact that EQCM is only sensitive to about 28% of the catalytically
active area, as descibed in the experimental section. We
assume a homogeneous mass loss across the electrode but an
inhomogenous current distribution could result in a wrong estimation.
On the other hand, ICP–MS is sensitive to losses from
the entire electrode area, which is likely to give a more accurate
evaluation.
2.2. Stability of MnOx
Chronopotentiometry and chronoamperometry
tests were also
carried out for MnOx in alkaline
solution, 1m KOH, as shown on
Figure 5 a,b. Compared to the results
for RuO2, the chronopotentiometry
test results in a larger
increase of potential during the
two hours. The graph in Figure
5 a also shows the importance
of choosing an axis length
spanning only the relevant data
range so the change is easily
spotted. However, on the basis of the chronoamperometry
measurement in Figure 5b, the deactivation is even clearer because
of the exponential behaviour of the current density as
a function of the overpotential. However, in both measurements,
a constant mass loss takes place in parallel with the
oxygen evolution current. At 1.9 VRHE, the loss of 1128 ngcm2
is equivalent to about 3.9 ML per hour, assuming the density
and lattice parameters of rutile (110) MnO2 layers,59 while
a constant current at 20 mAcm2 leads to a loss of about
2.6 ML per hour.1 Assuming that the losses are due to anodic
dissolution, that is, MnO2+2H2O!MnO4
(aq)+4H+ +3e,60
this rate would be equivalent to a current density of
0.35 mAcm2, more than four orders of magnitude lower than
the total current.2 The mass losses were also evaluated by ICP–
MS after each measurement, as shown in Figure 5c. We note
that the error bars are rather large. Nonetheless, the two methods
show an overall agreement. Extrapolating the data here,
the time required to completely corrode a 40 nm-thick MnOx
film would be approximately 36 h at a constant potential of
1.9 VRHE. From another perspective, a confirmation of stability
for a given catalyst would require that a specific lifetime can
be ensured. As an example, a lifetime of five years for a 40 nm
film corresponds to a maximum of 0.02 dissolved material in
a two-hour test. This rate for a RuO2 catalyst on a 1 cm2 electrode
in 100 mL of electrolyte results in a less than 2 ppt concentration
in the ICP–MS analysis. Such concentrations approach
the limit of detection, that is, 0.4 ppt for Ru,62 which
complicates meaningful extrapolation. Therefore, a comprehensive
lifetime evaluation should be accompanied by a long-term
test. In Table 1, the relevant stability metrics and standard deviations
are listed together with the OER activity of the thin
films.
From these measurements, it is clear that solely examining
current or potential changes for a small number of hours is insufficient
to establish the long-term performance of an OER
catalyst in an electrolyser. On the contrary, the anodic dissolution
of a catalyst may actually manifest itself over a short-term
measurement as an improvement in current density or decreased
overpotential, due to an increased surface area.34 OER
conditions may lead to an increase in the microscopic surface
area, a decrease in the catalyst surface area and a structural
change to a more stable phase; without prior knowledge, it is
not possible to determine which of these processes would predominate.
Therefore, we emphasise that explicit analyses of
mass changes are needed to quantify the stability of these
catalysts.
3. Conclusions
In conclusion, we have shown that the stability of catalysts for
the oxygen evolution reaction can be assessed by means of
short-term tests based on a combination of EQCM and ICP–
MS. It is clear that it is not possible to even roughly estimate
the long-term performance of a catalyst on the basis of shortterm
chronopotentiometry or chronoamperometry measurements
alone. Benchmarking and standardising research efforts
are still at an early stage for this reaction. Nonetheless, when
a new catalyst is discovered, rigorous and transparent criteria
should be applied to establish whether or not the material is
stable. While the end goal should be to test catalysts over the
long term in actual devices, the quantification of mass losses
using well-defined electrodes combined with EQCM and ICP–
MS provides a less-time-consuming, albeit meaningful, alternative.
Finally, although we have focused on the oxygen evolu-
1 We assumed a MnO2 composition on the basis of our XPS analysis. Although
the XRD experiments suggested that the films are amorphous, we take the
view that the rutile (110) plane provides a reasonable approximation of the
surface termination. Should we have chosen a different structure, the interplanar
distance would always be between 2–4 , varying the loss in monolayers
by less than a factor of two. The exact surface termination will not
change our overall conclusions.
2 It is conceivable that MnO2 dissolves via a two-electron process to MnO4
2,60
which would lead to a corrosion current density of 0.26 mAcm2, rather than
0.35 mAcm2. Regardless, this will not change the picture presented herein, as
the anodic current would still be negligible in comparison to the overall dissolution
current.
Table 1. Stability metrics from EQCM and ICP–MS.
Samplea h10 mAcm2 mV
from RDE
DM1.8V(RHE) 2 h
ngoxidecm2
DM1.9V(RHE) 2 h
ngoxidecm2
DM30 mAcm2 2 h
ngoxidecm2
DM20 mAcm2 2 h
ngoxidecm2
RuO2 3608 EQCM 146413/ – 1566110/ –
ICP-MS 191569 2624346
MnOx 4906 EQCM 462131/ 1128229/ – 73593/
ICP-MS 332108 1570447 793194
a For both oxides the measured mass losses are shown with corresponding standard deviation from four independent
measurements. The overpotentials listed here are from RDE tests based on two independent measurements.
Mass losses from EQCM are calculated from the frequency change using the Sauerbrey equation.
Values for ICP–MS are corrected to the corresponding dioxide masses.
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemElectroChem 0000, 00, 1–8 &6&
These are not the final page numbers!