Full paper

Electrocatalysts www.advenergymat.de

Unraveling Geometrical Site Confinement in Highly

Efficient Iron-Doped Electrocatalysts toward Oxygen
Evolution Reaction
Sung-Fu Hung, Ying-Ya Hsu, Chia-Jui Chang, Chia-Shuo Hsu, Nian-Tzu Suen,
Ting-Shan Chan, and Hao Ming Chen*

to develop OER electrocatalysts with a low

Introduction of iron in various catalytic systems has served a crucial function overpotential in order to scale down the
to significantly enhance the catalytic activity toward oxygen evolution reaction energy expenditure of water electrolysis.
To date, noble metal oxides, such as
(OER), but the relationship between material properties and catalysis is still
iridium oxide and ruthenium oxide, are
elusive. In this study, by regulating the distinctive geometric sites in spinel, regarded as the benchmarked OER elec-
Fe occupies the octahedral sites (Fe3+(Oh)) and confines Co to the tetrahedral trocatalysts,[10–13] but their scarcity and
site (Co2+(Td)), resulting in a strikingly high activity (ηj = 10 mA cm−2 = 229 mV high cost hinder them from the wide uti-
and ηj = 100 mA cm−2 = 281 mV). Further enrichment of Fe ions would occupy lization.[14] For these reasons, it is of the
the tetrahedral sites to decline the amount of Co2+(Td) and deteriorate the essence to explore earth-abundant sub-
stances showing comparable performance
OER activity. It is also found that similar tafel slope and peak frequency in
as costly noble-metal electrocatalysts.
Bode plot of electrochemical impedance spectroscopy indicate that Co2+(Td) Recently, earth-abundant transition-metal-
ions are primarily in charge of water oxidation catalytic center. By means of based materials exhibited high perfor-
electrochemical techniques and in situ X-ray absorption spectroscopy, it is mance and superb stability toward OER,
proposed that Fe3+(Oh) ions mainly confine cobalt ions to the tetrahedral site especially for the cobalt and nickel-based
oxides/nonoxides.[15–22] Furthermore, var-
to restrain the multipath transfer of cobalt ions during the dynamic structural
ious metal dopants would greatly affect
transformation between spinel and oxyhydroxide, continuously activating the the activities and intrinsic attributes.
catalytic behavior of Co2+(Td) ions. This material-related insight provides an For example, tungsten, chromium, iron,
indication for the design of highly efficient OER electrocatalysts. and zinc-doping have been discovered to
improve the activities in contrast to pris-
tine metal oxides owing to the optimi-
1. Introduction zation of adsorption energy for surface intermediates or the
increment of roughness factor.[23–29] Interestingly, among these
The rapid exhaustion of fossil fuels has raised considerable elements, iron can commonly perform significant enhance-
attention and triggered research interests in alternative energy, ment in catalytic activities toward oxygen evolution reaction as
in which water electrolysis with generating hydrogen and compared to other elements.[30,31] Boettcher’s group proposed
oxygen gases provides a potential way to produce clean and that Fe ions granted the catalytic activities while Co ions acted
renewable fuels.[1–6] Hydrogen and oxygen gases are generated as the conductive oxides to transport the charge carriers in an
through two half reactions, hydrogen evolution reaction (HER) Fe-Co metal oxide system.[32] Friebel et al. conducted a series of
and oxygen evolution reaction (OER), in water electrolysis. In operando experiments and calculations to demonstrate that Fe
comparison with HER reaction, OER reaction shows very slug- ions in Ni1-xFex OOH system would alter the adsorption ener-
gish kinetics and exhibits high overpotentials due to a com- gies of OER intermediates over the electrocatalytic surface and
plicated reaction mechanism, leading to the limited practical thus reduce the overpotential of OER, whereas the formation of
application for water electrolysis.[7–9] Therefore, it is important low-activity FeOOH declined the resulting activities in the cases
of higher Fe content.[33] Howbeit, the behavior of iron based on
the material insight was not elaborated. Toward this end, it is
S.-F. Hung, C.-J. Chang, C.-S. Hsu, Dr. N.-T. Suen, Prof. H. M. Chen
Department of Chemistry crucial to establish the direct relationship between the material
National Taiwan University characteristics and the catalytic activities.
Taipei 106, Taiwan Recently, we reported that the geometrical sites in spinel
E-mail: haomingchen@ntu.edu.tw cobalt oxide served distinct functions. In the case of Co3O4, the
Dr. Y.-Y. Hsu, Dr. T.-S. Chan cobalt ions in octahedral site (Co3+(Oh)) contributed to surface
National Synchrotron Radiation Research Center
Hsinchu 300, Taiwan
double layer capacitance while those in tetrahedral site (Co2+(Td))
were able to adsorb oxygen ions onto the surface for being
DOI: 10.1002/aenm.201701686 the active species.[34,35] It suggested that even for the identical

Adv. Energy Mater. 2017, 1701686 1701686  (1 of 11) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

atoms in the lattice, the bonding environment of metal ions

would considerably govern the catalytic behavior. On the other
hand, because oxygen evolution reaction commonly required
a high anodic potential, i.e., theoretical potential for splitting
water molecules (1.23 V vs standard hydrogen electrode) plus
the overpotential for kinetic reaction mechanism to drive the
reaction, the spinel phase would convert to form oxyhydroxide
phase according to the Pourbaix diagram and in situ observa-
tion.[36,37] The former phase possesses simultaneous octahedral
and tetrahedral sites while the latter one has purely octahedral
site in the lattice.[38] The phase transformation between two
phases could result in the geometrical-site alteration of Co ions,
followed by the lattice volume expansion/contraction and sub-
sequently the structural collapse.[36] From the viewpoint of lat-
tice geometrical site and the structural conversion, it is highly
desirable to realize how the material engineering and manipu-
lation dominate the catalytic activities and structural stabilities.
Herein, we exploited an occupancy preference of Fe ions
to manipulate the geometric site of active metal ions (i.e.,
Co2+(Td)), in which X-ray absorption spectroscopy (XAS) was uti-
lized to investigate the occupied geometric site of each metal
ion and the electron in the configuration of d-band character. A
strong correlation between the catalytic activities toward OER
and the proportional amount of Co2+(Td) ions was evidently elu-
cidated to ravel a geometrical-site confinement by iron cations,
achieving the lowest overpotential (ηj = 10 mA cm−2 = 229 mV and
ηj = 100 mA cm−2 = 281 mV) for iron-doped cobalt oxides. This con-
finement suppressed the multipath transition of Co2+(Td) during Figure 1.  SEM images of Fe-doped cobalt oxides: a) CoFe-0, b) CoFe-0.16,
the dynamic structurally transform, confirmed by in situ X-ray c) CoFe-0.28, d) CoFe-0.38, e) CoFe-0.44, and f) CoFe-0.49.
absorption spectroscopy, and thence the activating Co2+(Td)
ions significantly enhanced the catalytic activity and stability. formula of each catalyst is also listed in Table SI-1 (Supporting
The result of electrochemical impedance spectroscopy (EIS) Information). Nearly 3.5 mg cm−2 of catalyst on the substrate
bolstered the activation of Co2+(Td) and the fact that Co2+(Td) ions (as shown in Table SI-1, Supporting Information), which was
still governed the catalysis of water oxidation. Our finding could carefully weighed by microbalance, indicated that the overall
offer a guiding concept for introducing foreign metal ions to growth rate was not affected by the introduction of iron ions in
achieve efficient catalysis. the precursor solution.
Due to the large disparity in Fe content among these sam-
ples, it is essential to examine the conditions of Fe ions within
2. Results and Discussion the lattice since the presence of FeOOH might be a possible
phase to deteriorate the catalytic activity. Regarding attempts
To achieve an electrocatalyst for empirical utilization, not only to investigate the aforementioned lattice environment in these
low overpotential but also high current density in unit geo- samples, a high-resolution X-ray diffraction (XRD) with a syn-
metrical surface area should be concerned. According to this chrotron radiation light source was conducted to examine the
criterion, nanostructural electrocatalyst was directly deposited subtle difference. All peak positions, as illustrated in Figure 2
on 3D conductive framework by a chemical bath deposition to and Figure SI-1 of the Supporting Information, well matched
maximize the density of catalyst on the substrate. In the typical with reference of spinel phase referred to JCPDS No. 80-1545,
reaction, owing to a slow decomposition of urea, it facilitated a where all peaks gradually shifted toward lower angle region
heterogeneous deposition on the substrate instead of a homo- with increasing Fe content. Due to the fact that the radius
geneous agglomeration in the solution. Hence, needle-shaped of Fe3+ (78.5 pm) is larger than that of Co3+ (68.5 pm) for
nanowires were observed in pristine cobalt oxide whereas the oxides,[39] with increasing the Fe content, the lattice would
nanostructure converted into nanosheet once Fe ions were intro- expand as revealed by the shift of 2θ value toward lower angle.
duced into the precursor solution (Figure 1). This morphological High-resolution X-ray diffraction results clearly indicated a
transform was associated with the formation of nano­flake-like solid solution feature of Fe-doped cobalt oxide without sepa-
binary metal hydroxide in the stage of nucleation.[25] The ele- rated crystalline FeOOH. Notably, in Figure 3, X-ray absorption
mental analysis obtained by five trials as listed in Table SI-1 O K-edge spectrum showed that the oxygen bonding status of
(Supporting Information) coincided with the amount of CoFe-0.44 sample was evidently different from that of FeOOH,
Fe precursor. The samples were denoted as CoFe-0, CoFe-0.16, implying that the formation of amorphous FeOOH could be
CoFe-0.28, CoFe-0.38, CoFe-0.44, and CoFe-0.49 according ruled out. Moreover, the bond distance and orbital condition
to the average Fe content in each sample and the chemical of iron in CoFe-0.44 were also dissimilar with those in FeOOH

Adv. Energy Mater. 2017, 1701686 1701686  (2 of 11) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
Figure 2.  PXRD patterns of Fe-doped cobalt oxides with synchrotron light source: a) (111),
b) (220), c) (311), and d) (400) peaks.
(Figure SI-2, Supporting Information), suggested that neither
crystalline FeOOH nor amorphous form dominated the cata-
lytic nature. It was worth noting that the catalytic activities
arisen from individual components (i.e., separated phases) can
be ruled out in the present study, since both XRD and XAS
results concluded that all resulting samples showed character-
istic of solid solution rather than individual phases and further
indicated that no undesired component was present except for
spinel phase.
Additionally, high-resolution transmission electron micro­
scopy (TEM) photographs were taken to identify the doping
behavior by examining the interfacet distance (Figure 4). The
Figure 3.  X-ray absorption near edge spectroscopy for O K-edge.
Adv. Energy Mater. 2017, 1701686 1701686  (3 of 11)
www.advenergymat.de
d-spacing of (220) facet in CoFe-0.44 sample
(2.87 Å) was larger than that in pristine
cobalt oxide (2.76 Å) and no obvious sepa-
rated phase was generated, which was in
accord with the result of XRD. The selected
area electron diffraction patterns verified
the polycrystalline nature of these oxides,
while the d-spacing of each facet in Fe-doped
cobalt oxide was larger than that in the pris-
tine cobalt oxide (the detailed comparison of
d-spacing is listed in Table SI-2, Supporting
Information). Above observations, including
XRD patterns, high-resolution TEM, and
selected area electron diffraction results, are
in line with our expectations for the forma-
tion of solid solution in Fe-doped cobalt
oxide, even though the higher Fe content was
employed. The formation of solid solution
even in high Fe content could result from
the slow release of hydroxide ions from the
thermal decomposition of urea during the
synthesis, followed by the sufficient reaction
of binary metallic ions with the hydroxide
ions, leading to the well-mixing state in the
prepared film.
XAS was a potent technique to investigate
the electronic[12,40–42] and structural nature
of materials.[38,43,44] First, as illustrated by
Fourier transformed extended X-ray absorption fine structure
(EXAFS) spectra, there were two distinctive geometrical sites
(i.e., tetrahedral site and octahedral site) in spinel structure
(Figure 5). In Figure 5a, the intensity of Co-Co(Oh) peak, pro-
portional to the coordination number, gradually descended
with increasing Fe content, indicating that most Fe ions (Fe3+)
Figure 4.  High-resolution TEM images and electron diffraction patterns
of a,c) pristine cobalt oxide and b,d) Fe-doped cobalt oxide (CoFe-0.44).
© 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
Figure 5.  Fourier transformed extended X-ray absorption fine structure of a) Co K-edge, assigning Co-O (I), Co-Co(Oh) (II), and Co-Co(Td)/Fe-Co(Td)/
Co-O(2nd shell) (III), and b) Fe K-edge, assigning Fe-O (I), Fe-Fe(Oh) (II), and Fe-Fe(Td)/Fe-Co(Td) (III). The interatomic distance is shown without phase
correction.
substituted for the host cobalt in octahedral site (Co3+(Oh)). Note
that this behavior was able to confine Co ions in the lattice to
tetrahedral site and cause a decrement in relative oxidation
number of Co (evidenced in Figure SI-3a, Supporting Informa-
tion, of the X-ray absorption near edge spectra (XANES) spectra
of Co from K-edge). A similar observation from O K-edge
spectra could be found as well, in which the decreased oxida-
tion state of Co significantly reduced the electron density of
oxygen in energy of 531 eV (Figure SI-4, Supporting Informa-
tion). On the other hand, the EXAFS spectra of Fe K-edge of
Figure 5b elucidated that in the situation of heavy doping con-
dition, Fe ions tended to occupy tetrahedral sites owing to the
present inverse spinel preference (Figure 5b), while most Fe
occupied octahedral sites in light doping condition. This obser-
vation also accounted for the unchangeableness in XANES of
Fe (Figure SI-3b, Supporting Information). The preference sub-
stitution of Co3+(Oh) by Fe3+ ions might be attributed that non-
charged defect FeCo0 was more stable than the charged defect
FeCo+1 (appeared on the replacement of Co2+(Td) by Fe3+ ions)
in the lattice during the substitution. Yet, in the case of heavy
doping condition, the serious lattice distortion due to sub-
stantial defect FeCo0 in the lattice (the crystal radius of Fe3+ is
78.5 pm and that of Co3+(Oh) is 68.5 pm) will destabilize the lat-
tice, leading to the initiation of the substitution of Co2+(Td) by
Fe3+ to ease the lattice distortion due to the larger crystal radius
of Co2+(Td) (72 pm). In the meanwhile, the induced electron
was transferred to Co3+(Oh) to form the inverse spinel structure.
Second, in Co L3-edge of Figure 6a, the signal around energy of
778 eV represented the situation of Co2+(Td) while that of 780 eV
expressed the condition of Co3+(Oh).[45,46] By normalizing the
peak intensity in the energy of 780 eV, it was clear that the elec-
tronic structure changed by increasing occupation of Co2+(Td),
consistent with the EXAFS result. In the case of CoFe-0.49, the
Co2+(Td) occupancy decreased because Fe ions tended to occupy
tetrahedral sites. In Figure 6b, the electronic structure of Fe
was surprisingly unvaried despite the difference of occupied
sites of Fe. This could be ascribed to the crystal field diagrams
of Co and Fe at the different sites (Figure 6c). The crystal field
diagrams of Fe3+(Oh), Fe3+(Td), and Co2+(Td) were in a high-spin
Adv. Energy Mater. 2017, 1701686 1701686  (4 of 11)
www.advenergymat.de
arrangement,[47] except for that of Co3+(Oh) with a low-spin
arrangement.[48] The half-full configuration in crystal field dia-
grams of Fe3+(Oh)/Fe3+(Td) and the energetical similarity in Δoct
and Δtet resulted in an identical electronic structure, and a theo-
retical study of absorption spectra in varying the corresponding
crystal field stabilization energies predicted this argument as
well.[49] Thus, the electronic structure as well as oxidation state
of Fe3+ ions remained while those of Co2+ ions were affected
by the introduction of Fe3+ ions. In the case of cobalt ions, for
the reason that t2g orbitals were related to the π-bonding with
oxygen-related intermediate,[50,51] a full occupation of t2g orbitals
in Co3+(Oh) suppressed the adsorption of intermediates and the
reaction of active species for OER. Thus, the catalytic cycle for
OER hardly proceeded on the surface of Co3+(Oh), resulting in
the fact that massive hydroxides ions accumulated on the sur-
face by coulomb attraction and provided great surface double
layer capacitance. On the other hand, Co2+(Td) possessed a maxi­
­mum single-spin configuration. This analysis consisted in a
recent work, in which μ-OO peroxide intermediate was directly
detected in a Co2+(Td)-only spinel but no species was found in
a Co3+(Oh)-only spinel.[35] Therefore, we believed that Co2+(Td)
should be paramount toward OER while Fe3+ dopant served the
function of geometrical confinement.
In order to determine the relationship between the elec-
tronic characteristics of different geometrical sites and the
catalytic activities, a series of electrochemical experiment were
performed to examine the catalytic performance toward OER.
As illustrated in Figure 7a, Fe-doped cobalt oxides (CoFe-0.44
sample) exhibited the smallest overpotential for OER (281 mV
of overpotential at 100 mA cm−2) than that of pristine cobalt
oxide (the overpotential at the current density of 100 mA cm−2
was marked by the dashed line and listed in Table 1). It was
worth saying that after further introduction of Fe ions (CoFe-
0.49), its performance toward OER evidently decayed due to
decreasing Co2+(Td) ions in the lattice. Because separated phases
were not observed in this composition, the reason of deteriora-
tion by formation of inactive phase (i.e., iron oxides) could be
ruled out. It unraveled the strong connection between Co2+(Td)
ions and the catalytic activities rather than Fe3+ ions. Next,
© 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
Figure 6.  X-ray absorption near edge spectroscopy for a) Co L3-edge and b) Fe L3-edge, normalized by peak intensity. c) The electronic structures of
the Co and Fe ions in different geometrical sites and the corresponding crystal field diagrams for pristine and Fe-doped cobalt oxides. HS and LS are
referred to high spin and low spin, respectively.
overpotential at 10 mA cm−2 was commonly reported in the
literatures due to the average current density in solar-to-fuel
conversion.[7] Owing to the plentiful amount of catalyst grown
on the substrate, not only the current density for OER but also
that of oxidation peak augmented, leading to the unattainable-
ness of overpotential at 10 mA cm−2 in Figure 7a. Hence, the
chronopotentiometry was conducted to evaluate the overpoten-
tial at 10 mA cm−2 during the long-term measurement. At the
initial stage, an abrupt jump in each curve was observed since
the capacitive current was eliminated. Following this jump,
there were two distinguishable behaviors. The applied voltage
of pristine cobalt oxide gradually raised while that of Fe-doped
cases decreased. We believed that this phenomenon was highly
related to the activation process of Co2+(Td) and would be dis-
cussed later. Eventually, the detected voltage attained a steady
state in each sample. A value of 229 mV for the overpotential
was achieved in the CoFe-0.44 sample, which is the smallest
one among the cobalt-iron oxides-based electrocatalysts as
compared with the previous literatures.[52–55] The overpotential
raised in the case of further increase in iron dopant (CoFe-0.49
sample). A similar phenomenon was also exhibited in chrono-
potentiometry, in which the activation process was noted in the
initial measurement, and the samples after activation exhibited
decent catalytic stabilities (Figure SI-5, Supporting Informa-
tion). The activation process also depended on the amount of
Adv. Energy Mater. 2017, 1701686 1701686  (5 of 11)
www.advenergymat.de
Co2+(Td); this suggested that the catalytic activities were associ-
ated with the amount of Co2+(Td) ions in the catalyst rather than
that of Fe ions.
In order to consider an auxiliary role of Fe ions in OER,
the influences of Fe ions on the reaction mechanism of OER
and the interaction between Co and Fe ions should be fur-
ther analyzed. The value of Tafel slope was directly related
to the rate-determining step in the reaction mechanism of
OER.[56–59] Interestingly, a similar value of ≈60 mV dec−1 was
obtained from all samples in Figure 7c, explicating an iden-
tical rate-determining step for all samples (i.e., the step of
another hydroxide ion adsorbed by metal-oxygen (MO) inter-
mediate),[56,60] in agreement with the in situ identification of
OO bond on the surface by means of Raman analysis in our
previous report.[35] Thus, the introduction of Fe ions into the
lattice did not alter the reaction path, supporting the argument
that the active sites remained the original element rather than
the foreign one, but merely reduced the conductivity of elec-
trocatalysts to cause the slight increment of Tafel slope. On the
other hand, the interaction between Co and Fe could be verified
by cyclic voltammogram (Figure 7d). Two redox couples were
observed in the cyclic voltammogram of pristine cobalt oxide:
one was Co3+/Co4+ redox pair at 0.42 V (anodic) and 0.36 V
(cathodic) versus a saturated calomel electrode (SCE) and the
other was Co2+/Co3+ redox pair at 0.32 V (anodic) and 0.23 V
© 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

Figure 7.  a) Linear sweep voltammograms, b) chronopotentiometry at current density of 10 mA cm−2, c) corresponding Tafel plots, and d) cycling
voltammetry curve at scan rate of 10 mV s−1 of various samples. e) Illustrative representation of crystal field diagrams for the interaction between Co
and Fe via the bridged oxygen in the lattice.

(cathodic) versus SCE (the detailed information is shown in
Figure SI-6, Supporting Information). It revealed that the Co2+/
Co3+ oxidation peak[61] shifted anodically with increasing Fe
content and an obvious increase in integrated charge of this
peak. The first meant that a strong interaction between Co and
Fe ions gave rise to an anodic shift of Co2+ oxidation peak in
Fe-doped cobalt oxides, similar to the findings by Boettcher
and co-workers.[32] This could be interpreted by an interaction
between Co and Fe cations via a bridged oxygen in the lattice
(as illustrated in Figure 7e). In the case of pristine cobalt oxide,
there was an intensive electron–electron repulsion between the
full occupation of t2g in Co3+(Oh) and O 2p πu orbitals due to
the excessive electron density in these orbitals, leading to an
increasing electron donation of O 2p πu orbital toward t2g in
Co2+(Td) and thereby intensifying the repulsion between O and
Co2+(Td) orbitals and elevating the energy level. In the Fe-doped

Adv. Energy Mater. 2017, 1701686 1701686  (6 of 11) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

Table 1. Overpotential at current density of 10 and 100 mA cm−2,

obtained from chronopotentiometry and linear sweep voltammograms,
respectively, Tafel slope, charge transfer resistance, and roughness factor
for Fe-doped cobalt oxides.

Sample η10 η100 Tafel slope RCT Roughness factor

[mV] [mV] [mV dec−1] [Ω]
CoFe-0 291 341 63.1 3.60 4439
CoFe-0.16 249 314 58.8 0.66 1937
CoFe-0.28 233 296 60.7 0.61 2782
CoFe-0.38 231 288 62.5 0.59 2886
CoFe-0.44 229 281 64.2 0.44 2604
CoFe-0.49 237 300 67.6 0.52 2469

case, the replacement by Fe3+(Oh) with a lower electron den-

sity in t2g weakened the electron–electron repulsion between
Fe3+(Oh) and O 2p πu orbitals, thereby reducing electron dona-
tion of O 2p πu orbitals toward t2g of Co2+(Td) and descending
the energy level. As a result, due to the decrease in energy level,
the electrons occupied at the t2g in Co2+(Td) were hardly removed
(i.e., oxidation). That is, the oxidation peak of Co2+/Co3+ gradu-
ally shifted toward higher voltage with increasing Fe content
was verified and a similar behavior could be observed in the
case of nickel-iron oxides.[62] The second showed that Co2+ oxi-
dation boosted with increasing Fe content. This phenomenon
was attributed to two possibilities: one was the expansion of
surface area to provide more Co2+ ions and the other was the
activation of Co2+ ions to improve the ability of oxidation.[63]
Thus, the electrochemical active surface area (ECSA) or rough-
ness factor was obtained by scan-rate-dependent cyclic voltam-
mogram at non-Faradaic region (see Figures SI-7 and SI-8,
Supporting Information) and is listed in Table 1. It was found
that ECSA of Fe-doped cobalt oxides was much lower than that
of pristine one, directly related to the morphologies shown
in Figure 1; i.e., the surface area of 2D nanosheet should be
smaller than that of 1D nanowire. Therefore, the surface area
did not dominate the Co2+ oxidation. It should be the improved
properties of Co2+ ions by the presence of Fe ions to promote
the Co2+ oxidation, which was strongly associated with the cata-
lytic activity toward water oxidation.
Charge transfer between the catalyst and electrolyte was an
important indicator to evaluate the catalytic properties. EIS was
utilized to analyze the behavior of charge transfer at the inter-
face or the surface. A qualitative Bode plot (phase angle vs log
frequency; Figure 8 and Figure SI-9, Supporting Information)
showed that the phase peaks appeared at region below 101 Hz,
referring to the Faradaic oxygen reaction upon the outer sur-
face of intermediates in Co2+, and the phase angle decreased
with increasing applied bias.[34,64,65] Furthermore, Lyons and
Brandon pointed out that the frequency of maximum phase
angle for cobalt oxides was different from that for iron oxides
at low frequency region.[66] In our results, the peak frequency
slightly shifted toward higher frequency region (≈100 Hz)
and the phase angle greatly reduced with the presence of Fe
dopant in the lattice. This suggested that Co2+(Td) was mainly
in charge of the water oxidation catalytic centers while Fe3+
ions improved the nature of Co2+(Td) ions in the lattice through Figure 8. 2D contour of Bode plots of various samples with varying
the electronic interaction and geometrical confinement. On applied voltage referred to SCE.

Adv. Energy Mater. 2017, 1701686 1701686  (7 of 11) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

Figure 9.  The ratio between coordination number of octahedral site and tetrahedral site, extracted from the in situ X-ray absorption spectroscopy for
a) CoFe-0 and b) CoFe-0.44. The catalyst underwent oxidation process and then reduction process during the in situ measurement. O. C. meant open-
circuit state. c) Schematic representation of multipath during phase conversion in cobalt oxides.

the other hand, the Nyquist plot in Figures SI-10 and SI-11
(Supporting Information) showed that RCT gradually decreased
with increasing Fe content but increased in the case of CoFe-
0.49, matching well with the relative magnitude of Co2+(Td). As
a consequence, we could conclude that the Co2+(Td) ions still
acted as the major active species even if half Co ions were
replaced with Fe ions.
Based on above observations, we proposed a model for
this geometrical site confinement in iron-doped electrocata-
lysts. In the onset region of OER, a reversibly structural con-
version between spinel and oxyhydroxide phase occurred at
high applied voltage as revealed in Pourbaix diagram.[67,68]
Dynamic structural conversion is clearly evidenced by in
situ X-ray absorption spectroscopy in Figure 9a,b (the values
are derived from Figure SI-12, Supporting Information). The
ratio between the coordination number of octahedral sites
and tetrahedral sites gradually increased in the region of oxi-
dation and subsequently decreased during the reduction.
It indicated that the structural conversion was a dynamic
equilibrium rather than an irreversible process. That is to
say, Co3+(Oh) in oxyhydroxide phase could transform back to
either Co2+(Td) (active site) or Co3+(Oh) (inactive site) in pris-
tine cobalt oxides. In this case, a multipath conversion has to
be occurred (illustrated in Figure 9c) and the overall activa-
tion should be disturbed because a random transforming of
cobalt ions migration possibly caused the original active site
to be inactive. This phenomenon, called a deactivation pro-
cess, was predicted to decline the resulting catalytic activity,
thereby elevating the applied voltage to maintain the current
density of 10 mA cm−2 (Figure 7b) and reducing the initial cur-
rent density in chronoamperometry (Figure SI-5, Supporting
Information). In the Fe-doped case, with the presence of geo-
metrical-site confinement by iron ions, supported by the fact
of unvaried bonding situation of Fe in the in situ X-ray absorp-
tion spectroscopy (Figure SI-12, Supporting Information), the
transformation back into octahedral site could be consider-
ably suppressed to recover compulsorily the Co2+(Td), and thus
these sites were continuously activated. This situation, called
an activation process, led to the decrease of applied voltage in
chronopotentiometry and enhancement of current density in
chronoamperometry. The results of in situ measurement and
electrochemical analysis suggested the auxiliary function of Fe
as the geometrical-site confinement and the catalytic center of
cobalt ions. Recently, Li et al. proposed a role of Fe as a Lewis
acid in catalyst to further oxidize nickel ions to enhance the
catalytic activity.[69] Chen et al. also proved that the presence of
Fe ions would stabilize NiOOH lattice, which was responsible
for the catalytic activity.[62] Therefore, we believed that the acti-
vation of cobalt oxides by the preferred occupancy of Fe ions
in the lattice would offer an implication for the role of Fe in
OER electrocatalysts and a guide for designing a highly effi-
cient electrocatalyst for oxygen evolution reaction.

Adv. Energy Mater. 2017, 1701686 1701686  (8 of 11) © 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
3. Conclusion
In conclusion, iron-doped cobalt oxide with a solid solution
form was directly synthesized on 3D conductive framework as a
highly efficient electrocatalyst toward OER, fulfilling the criteria
of high current density (over 400 mA cm−2) and low overpo-
tential (ηj = 10 mA cm−2 = 229 mV and ηj = 100 mA cm−2 = 281 mV)
for practical utilization. Similar behavior in both Tafel slope and
Bode plot of electrochemical impedance spectroscopy revealed
that Co2+(Td) was primarily in charge of the water oxidation cata-
lytic centers, while the Fe3+ ions altered the nature of Co2+ in
the lattice through the electronic interaction and geometrical
confinement. This confinement of Fe3+(Oh) was able to restrain
the multipath conversion during the reversible structural
transformation in OER, evidenced by the in situ X-ray absorp-
tion spectroscopy, to perform a distinguished electrochemical
nature. This revelation provides a connection between the
intrinsic material properties and the electrocatalysis for OER.
4. Experimental Section
Chemicals: Cobalt(II) nitrate hexahydrate (99%) and urea (99.5%)
were purchased from ACROS. Iron(II) chloride (98%) was purchased
from Alfa Aesar. Potassium hydroxide was purchased from Fisher
Scientific.
Iron-Doped Cobalt Oxides Synthesis: Cobalt oxides were synthesized
by means of chemical bath deposition and postcalcination, described
in detail in our previous report.[25] In a typical synthesis, preclean nickel
foam was put perpendicularly into a sample vial with 1 cm of depth of
immersion in precursor solution (10 mL). The solution containing urea
(0.625 g), cobalt nitrate (1.5 mmol), and iron chloride was heated to 90 °C
and reacted for 3 h. In order to tune the amount of iron dopant, the various
amounts of iron salts (0, 0.3, 0.6, 0.9, 1.2, and 1.5 mmol) were introduced
in each solution. Subsequently, the substrate was rinsed with deionized
water and further calcinated at 300 °C for 2 h with the heating rate of 2 °C
min−1 in order to insert iron ions into the lattice of cobalt oxides. Loading
amount of catalysts on the substrate was weighed by microbalance
(Mettler Toledo XS 105, d = 0.01 mg) and obtained by subtracting the
weight of catalyst on the blank substrate from catalyst-coated substrate.
Structural Characterization: The morphologies and elemental
analysis were investigated with field-emission scanning electron
microscopy (SEM, JEOL JSM-6700F) equipped energy dispersive X-ray
spectroscopy (Oxford Instrument XMax 150 mm2). The lattice fringes
and selected-area electron diffraction were collected with field-emission
TEM (JEOL-2100F). The crystal structures were determined by powder
X-ray diffraction (PXRD) analysis (the incident X-ray wavelength of
0.7749 Å) using a large Debye–Scherrer camera in BL-01C2 at National
Synchrotron Radiation Research Center (NSRRC, Hsinchu, Taiwan) in
which the electron storage ring was operating at 1.5 GeV with a beam
current of 360 mA. The XRD patterns were calibrated by CeO2 standard
and altered to that with wavelength of 1.5413 Å using the software called
“Winplotr.” The samples were detached from the substrate and loaded
into the holder of capillary to perform the analysis for the reason of
avoiding the interference of intense background signal by substrate.
Electrochemical Characterization: Electrochemical properties were
investigated on Bio-logic VSP potentiostat in a standard three-electrode
setup using electrocatalyst as the working electrode, a platinum plate as
the counter electrode, and a saturated calomel electrode (SCE) as the
reference electrode. Aqueous solution of KOH (1 m, pH = 13.6) was used as
the electrolyte. The potentials were converted to values relative to reversible
hydrogen electrode (RHE) based on the following equation: ERHE = ESCE +
0.0591 × pH + E0SCE, where E0SCE is the standard potential of SCE relative
to standard hydrogen electrode (SHE) at 25 °C (0.240 V). The working
area of electrodes was fixed at 0.5 cm2 for each experiment. Linear sweep
Adv. Energy Mater. 2017, 1701686 1701686  (9 of 11)
www.advenergymat.de
voltammetry for oxygen evolution was collected at the scan rate of 5 mV s−1
with iR-correction. Chronopotentiometry was recorded at the current density
of 10 mA cm−2 while chronoamperometry was also performed at the initial
current density of 10 mA cm−2 to monitor the activation process and to
evaluate the stability. The electrochemical active surface area (ECSA)
was measured from double-layer capacitance using cyclic voltammetry
in a small potential range (0.15–0.20 V vs SCE), from which the double-
layer capacitance was determined from the slope of capacitive current
at the midpoint of the scan range (0.175 V vs SCE) versus the scan rate
according to Cdl = Icap(dEdt−1)−1, where Cdl is the double-layer capacitance
and dEdt−1 is the scan rate. ECSA was calculated through dividing the
average specific capacitance of 40 µF cm−2 for transition metal oxide in 1 m
alkaline solution.[70] EIS measurements were performed in a frequency
range from 0.1 to 104 Hz with AC amplitude of 10 mV under applied bias
range from 0.4 to 0.6 V versus SCE with a scan step of 20 mV.
X-Ray Absorption Spectroscopy Analysis: X-ray absorption spectroscopy
including X-ray absorption near edge structure (XANES) and extended
X-ray absorption fine structure (EXAFS) at Co and Fe K-edge were collected
in total-fluorescence-yield mode at ambient air in BL-01C1 at NSRRC. The
scan range was kept in an energy range of 7600–8300 eV for Co K-edge
and 7000–7700 eV for Fe K-edge. Subtracting the baseline of pre-edge and
normalizing that of postedge obtained the spectra. EXAFS analysis was
conducted using Fourier transform on k3-weighted EXAFS oscillations to
evaluate the contribution of each bond pair to Fourier transform peak. On
the other hand, the XANES of Co L-edge, Fe L-edge, and O K-edge were
measured in total X-ray electron yield mode at room temperature using
BL-20A at NSRRC. The samples were subjected to an ultrahigh vacuum
chamber (1 × 10−9 torr). The area under spectra was fixed over an energy
range of 770–810 eV for Co L-edge and 700–730 eV for Fe L-edge and the
spectra were normalized by the maximum intensity of L3 peak.
In Situ X-Ray Absorption Spectroscopy: The measurement in a
typical three-electrode setup as the same condition in electrochemical
characterization case was performed in a specially designed Teflon
container with a window sealed by Kapton tape. X-ray was allowed to
transmit through the tape and electrolyte, so that the signal of X-ray
absorption spectroscopy could be collected in total-fluorescence-yield
mode in BL-01C1 in NSRRC.
Supporting Information
Supporting Information is available from the Wiley Online Library or
from the author.
Acknowledgements
The authors acknowledge support from the Ministry of Science and
Technology, Taiwan (Contract Nos. MOST 105-2923-M-002-004-MY3
and MOST 106-2119-M-002-031-, NTU-CESRP-106R7619, NTUERP-
106R880206, and NTUERP-106R880218). Technical support from
NanoCore, the Core Facilities for Nanoscience and Nanotechnology at
Academia Sinica in Taiwan, is acknowledged.
Conflict of Interest
The authors declare no conflict of interest.
Keywords
electrocatalysts, highly efficient, iron, oxygen evolution reaction, water
splitting
Received: June 20, 2017
Revised: August 10, 2017
Published online:
© 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
[1] S. Chu, A. Majumdar, Nature 2012, 488, 294.
[2] C. G. Morales-Guio, L.-A. Stern, X. Hu, Chem. Soc. Rev. 2014, 43,
6555.
[3] M. G. Walter, E. L. Warren, J. R. McKone, S. W. Boettcher, Q. Mi,
E. A. Santori, N. S. Lewis, Chem. Rev. 2010, 110, 6446.
[4] J. Duan, S. Chen, M. Jaroniec, S. Z. Qiao, ACS Catal. 2015, 5, 5207.
[5] C. F. Tan, S. A. B. Azmansah, H. Zhu, Q.-H. Xu, G. W. Ho,
Adv. Mater. 2017, 29, 1604417.
[6] M.-Q. Yang, Y.-J. Xu, W. Lu, K. Zeng, H. Zhu, Q.-H. Xu, G. W. Ho,
Nat. Commun. 2017, 8, 14224.
[7] C. C. L. McCrory, S. Jung, I. M. Ferrer, S. M. Chatman, J. C. Peters,
T. F. Jaramillo, J. Am. Chem. Soc. 2015, 137, 4347.
[8] G. Yilmaz, K. M. Yam, C. Zhang, H. J. Fan, G. W. Ho, Adv. Mater.
2017, 29, 1606814.
[9] M.-Q. Yang, J. Dan, S. J. Pennycook, X. Lu, H. Zhu, Q.-H. Xu,
H. J. Fan, G. W. Ho, Mater. Horiz. 2017, 4, 885.
[10] L. C. Seitz, C. F. Dickens, K. Nishio, Y. Hikita, J. Montoya, A. Doyle,
C. Kirk, A. Vojvodic, H. Y. Hwang, J. K. Nørskov, T. F. Jaramillo,
Science 2016, 353, 1011.
[11] H.-S. Oh, H. N. Nong, T. Reier, A. Bergmann, M. Gliech,
J. Ferreira de Araújo, E. Willinger, R. Schlögl, D. Teschner,
P. Strasser, J. Am. Chem. Soc. 2016, 138, 12552.
[12] S.-F. Hung, F.-X. Xiao, Y.-Y. Hsu, N.-T. Suen, H. B. Yang,
H. M. Chen, B. Liu, Adv. Energy Mater. 2016, 6, 1501339.
[13] T. Audichon, T. W. Napporn, C. Canaff, C. Morais, C. Comminges,
K. B. Kokoh, J. Phys. Chem. C 2016, 120, 2562.
[14] P. C. K. Vesborg, T. F. Jaramillo, RSC Adv. 2012, 2, 7933.
[15] F. Song, X. Hu, J. Am. Chem. Soc. 2014, 136, 16481.
[16] Y.-P. Zhu, T. Y. Ma, M. Jaroniec, S. Z. Qiao, Angew. Chem., Int. Ed.
Engl. 2016, 129, 1344.
[17] S. Dou, X. Li, L. Tao, J. Huo, S. Wang, Chem. Commun. 2016, 52,
9727.
[18] C. Taviot-Guého, P. Vialat, F. Leroux, F. Razzaghi, H. Perrot, O. Sel,
N. D. Jensen, U. G. Nielsen, S. Peulon, E. Elkaim, C. Mousty, Chem.
Mater. 2016, 28, 7793.
[19] P. Chen, K. Xu, Z. Fang, Y. Tong, J. Wu, X. Lu, X. Peng, H. Ding,
C. Wu, Y. Xie, Angew. Chem., Int. Ed. Engl. 2015, 54, 14710.
[20] V. Fidelsky, M. C. Toroker, J. Phys. Chem. C 2016, 120, 25405.
[21] X. Xu, F. Song, X. Hu, Nat. Commun. 2016, 7, 12324.
[22] T. Y. Ma, J. L. Cao, M. Jaroniec, S. Z. Qiao, Angew. Chem., Int. Ed.
Engl. 2015, 128, 1150.
[23] B. Zhang, X. Zheng, O. Voznyy, R. Comin, M. Bajdich,
M. García-Melchor, L. Han, J. Xu, M. Liu, L. Zheng,
F. P. García de Arquer, C. T. Dinh, F. Fan, M. Yuan, E. Yassitepe,
N. Chen, T. Regier, P. Liu, Y. Li, P. De Luna, A. Janmohamed,
H. L. Xin, H. Yang, A. Vojvodic, E. H. Sargent, Science 2016, 352,
333.
[24] C.-C. Lin, C. C. L. McCrory, ACS Catal. 2017, 7, 443.
[25] S.-F. Hung, C.-W. Tung, T.-S. Chan, H. M. Chen, CrystEngComm
2016, 18, 6008.
[26] C. Xiao, X. Lu, C. Zhao, Chem. Commun. 2014, 50, 10122.
[27] Z. Ren, V. Botu, S. Wang, Y. Meng, W. Song, Y. Guo, R. Ramprasad,
S. L. Suib, P.-X. Gao, Angew. Chem., Int. Ed. Engl. 2014, 53, 7223.
[28] X. Liu, Z. Chang, L. Luo, T. Xu, X. Lei, J. Liu, X. Sun, Chem. Mater.
2014, 26, 1889.
[29] J. Wang, C. F. Tan, T. Zhu, G. W. Ho, Angew. Chem., Int. Ed. Engl.
2016, 128, 10482.
[30] J. Duan, S. Chen, A. Vasileff, S. Z. Qiao, ACS Nano 2016, 10, 8738.
[31] Y. Li, C. Zhao, ACS Catal. 2017, 7, 2535.
[32] M. S. Burke, M. G. Kast, L. Trotochaud, A. M. Smith, S. W. Boettcher,
J. Am. Chem. Soc. 2015, 137, 3638.
[33] D. Friebel, M. W. Louie, M. Bajdich, K. E. Sanwald, Y. Cai,
A. M. Wise, M.-J. Cheng, D. Sokaras, T.-C. Weng, R. Alonso-Mori,
R. C. Davis, J. R. Bargar, J. K. Nørskov, A. Nilsson, A. T. Bell, J. Am.
Chem. Soc. 2015, 137, 1305.
Adv. Energy Mater. 2017, 1701686 1701686  (10 of 11)
www.advenergymat.de
[34] H.-Y. Wang, S.-F. Hung, H.-Y. Chen, T.-S. Chan, H. M. Chen, B. Liu,
J. Am. Chem. Soc. 2016, 138, 36.
[35] H.-Y. Wang, S.-F. Hung, Y.-Y. Hsu, L. Zhang, J. Miao, T.-S. Chan,
Q. Xiong, B. Liu, J. Phys. Chem. Lett. 2016, 7, 4847.
[36] C.-W. Tung, Y.-Y. Hsu, Y.-P. Shen, Y. Zheng, T.-S. Chan,
H.-S. Sheu, Y.-C. Cheng, H. M. Chen, Nat. Commun. 2015, 6,
8106.
[37] M. Bajdich, M. García-Mota, A. Vojvodic, J. K. Nørskov, A. T. Bell,
J. Am. Chem. Soc. 2013, 135, 13521.
[38] M. W. Kanan, J. Yano, Y. Surendranath, M. Dincă, V. K. Yachandra,
D. G. Nocera, J. Am. Chem. Soc. 2010, 132, 13692.
[39] R. D. Shannon, Acta Cryst. A 1976, 32, 751.
[40] S.-C. Lin, C.-S. Hsu, S.-Y. Chiu, T.-Y. Liao, H. M. Chen, J. Am. Chem.
Soc. 2017, 139, 2224.
[41] H. M. Chen, C. K. Chen, C.-J. Chen, L.-C. Cheng, P. C. Wu,
B. H. Cheng, Y. Z. Ho, M. L. Tseng, Y.-Y. Hsu, T.-S. Chan, J.-F. Lee,
R.-S. Liu, D. P. Tsai, ACS Nano 2012, 6, 7362.
[42] T. Ling, D.-Y. Yan, Y. Jiao, H. Wang, Y. Zheng, X. Zheng, J. Mao,
X.-W. Du, Z. Hu, M. Jaroniec, S. Z. Qiao, Nat. Commun. 2016, 7,
12876.
[43] C. H. M. van Oversteeg, H. Q. Doan, F. M. F. de Groot, T. Cuk,
Chem. Soc. Rev. 2016, 46, 102.
[44] G. S. Hutchings, Y. Zhang, J. Li, B. T. Yonemoto, X. Zhou, K. Zhu,
F. Jiao, J. Am. Chem. Soc. 2015, 137, 4223.
[45] A. M. Hibberd, H. Q. Doan, E. N. Glass, F. M. F. de Groot,
C. L. Hill, T. Cuk, J. Phys. Chem. C 2015, 119, 4173.
[46] J. Wang, J. Zhou, Y. Hu, T. Regier, Energy Environ. Sci. 2013, 6,
926.
[47] C. L. Muhich, V. J. Aston, R. M. Trottier, A. W. Weimer,
C. B. Musgrave, Chem. Mater. 2016, 28, 214.
[48] M. Al Samarai, M. U. Delgado-Jaime, H. Ishii, N. Hiraoka,
K.-D. Tsuei, J. P. Rueff, B. Lassale-Kaiser, B. M. Weckhuysen,
F. M. F. de Groot, J. Phys. Chem. C 2016, 120, 24063.
[49] G. Van der Laan, I. W. Kirkman, J. Phys.: Condens. Matter 1992, 4,
4189.
[50] J. Huang, J. Chen, T. Yao, J. He, S. Jiang, Z. Sun, Q. Liu, W. Cheng,
F. Hu, Y. Jiang, Z. Pan, S. Wei, Angew. Chem., Int. Ed. Engl. 2015,
127, 8846.
[51] J. O. Bockris, T. Otagawa, J. Electrochem. Soc. 1984, 131, 290.
[52] X.-F. Lu, L.-F. Gu, J.-W. Wang, J.-X. Wu, P.-Q. Liao, G.-R. Li,
Adv. Mater. 2016, 29, 1604437.
[53] Y. Liu, J. Li, F. Li, W. Li, H. Yang, X. Zhang, Y. Liu, J. Ma, J. Mater.
Chem. A 2016, 4, 4472.
[54] X. Lin, X. Li, F. Li, Y. Fang, M. Tian, X. An, Y. Fu, J. Jin, J. Ma,
J. Mater. Chem. A 2016, 4, 6505.
[55] C. G. Morales-Guio, L. Liardet, X. Hu, J. Am. Chem. Soc. 2016, 138,
8946.
[56] T. Shinagawa, A. T. Garcia-Esparza, K. Takanabe, Sci. Rep. 2015, 5,
13801.
[57] R. Guidelli, R. G. Compton, J. M. Feliu, E. Gileadi, J. Lipkowski,
W. Schmickler, S. Trasatti, Pure Appl. Chem. 2014, 86,
245.
[58] A. Damjanovic, A. Dey, J. O. Bockris, Electrochim. Acta 1966, 11,
791.
[59] J. O. Bockris, J. Chem. Phys. 1956, 24, 817.
[60] Y. Zheng, Y. Jiao, Y. Zhu, Q. Cai, A. Vasileff, L. H. Li, Y. Han, Y. Chen,
S. Z. Qiao, J. Am. Chem. Soc. 2017, 139, 3336.
[61] S. Palmas, F. Ferrara, A. Vacca, M. Mascia, A. M. Polcaro,
Electrochim. Acta 2007, 53, 400.
[62] J. Y. C. Chen, L. Dang, H. Liang, W. Bi, J. B. Gerken, S. Jin, E. E. Alp,
S. S. Stahl, J. Am. Chem. Soc. 2015, 137, 15090.
[63] N.-T. Suen, S.-F. Hung, Q. Quan, N. Zhang, Y.-J. Xu, H. M. Chen,
Chem. Soc. Rev. 2017, 46, 337.
[64] R. N. Singh, N. K. Singh, J. P. Singh, Electrochim. Acta 2002, 47,
3873.
© 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com
[65] C.-S. Hsu, N.-T. Suen, Y.-Y. Hsu, H.-Y. Lin, C.-W. Tung, Y.-F. Liao,
T.-S. Chan, H.-S. Sheu, S.-Y. Chen, H. M. Chen, Phys. Chem. Chem.
Phys. 2017, 19, 8681.
[66] M. E. G. Lyons, M. P. Brandon, J. Electroanal. Chem. 2009, 631,
62.
[67] N. Spataru, C. Terashima, K. Tokuhiro, I. Sutanto, D. A. Tryk,
S.-M. Park, A. Fujishima, J. Electrochem. Soc. 2003, 150,
E337.
Adv. Energy Mater. 2017, 1701686 1701686  (11 of 11)
www.advenergymat.de
[68] A. Bergmann, E. Martinez-Moreno, D. Teschner, P. Chernev,
M. Gliech, J. F. de Araújo, T. Reier, H. Dau, P. Strasser,
Nat. Commun. 2015, 6, 8625.
[69] N. Li, D. K. Bediako, R. G. Hadt, D. Hayes, T. J. Kempa, F. von Cube,
D. C. Bell, L. X. Chen, D. G. Nocera, Proc. Natl. Acad. Sci. USA 2017,
114, 1486.
[70] C. C. L. McCrory, S. Jung, J. C. Peters, T. F. Jaramillo, J. Am. Chem.
Soc. 2013, 135, 16977.
© 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim