Journal of The Electrochemical

Society

OPEN ACCESS

Oxygen Evolution on Iron Oxide Nanoparticles: The Impact of

Crystallinity and Size on the Overpotential
To cite this article: Zhongbin Zhuang et al 2021 J. Electrochem. Soc. 168 034518

View the article online for updates and enhancements.

This content was downloaded from IP address 41.250.243.157 on 03/02/2022 at 19:29

 Journal of The Electrochemical Society, 2021 168 034518

Oxygen Evolution on Iron Oxide Nanoparticles: The Impact of

Crystallinity and Size on the Overpotential
Zhongbin Zhuang,1,2 Stephen A. Giles,1,3,6 Glen R. Jenness,1,3,4 Reza Abbasi,1,z Xi Chen,5
Bo Wang,5 Dionisios G. Vlachos,1,3,z and Yushan Yan1,3,*
1
Department of Chemical and Biomolecular Engineering and Center for Catalytic Science and Technology, University of
Delaware, 150 Academy Street, Newark, Delaware 19716, United States of America
2
State Key Lab of Organic-Inorganic Composites and Beijing Advanced Innovation Center for Soft Matter Science and
Engineering, Beijing University of Chemical Technology, Beijing 100029, People’s Republic of China
3
Catalysis Center for Energy Innovation (CCEI), University of Delaware, Newark, Delaware 19716, United States of
America
4
Environmental Laboratory, US Army Engineer Research and Development Center, Vicksburg, Mississippi 39180, United
States of America
5
GEIRI North America, San Jose, California 95134, United States of America
6
CFD Research Corporation, Huntsville, Alabama 35806, United States of America

The oxygen evolution reaction (OER) is a critical reaction in electrolysis and photoelectrolysis of water to generate and store clean
energy. Therefore, the development of low-cost and efficient electrocatalysts for the OER is of great scientific and technological
importance. Although promising iron oxide-based electrocatalysts have been recently developed for the OER, an in-depth
experimental and theoretical analysis of the OER mechanism on iron oxide-based electrocatalysts is still needed to provide
guidelines to optimize the performance of iron oxide-based electrocatalysts further. To address this need, we synthesized a series of
monodisperse iron oxide nanoparticles to analyze their intrinsic OER activities. Using nanoparticles of the same size but different
crystallinity, we show that amorphous iron oxide nanoparticles have better OER activity than crystalline ones. The size effect
studies further revealed that the edge/defect sites are the active sites for the OER. Density functional theory calculations
demonstrated that the edge/defect sites provide bridge sites to adsorb OER intermediates, resulting in low OER overpotential.
These calculations confirm that the high OER activity of amorphous nanoparticles results from a high concentration of defect sites
on their surface. These results provide novel strategies to increase the performance of iron oxide-based and likely other oxide-based
OER electrocatalysts.
© 2021 The Author(s). Published on behalf of The Electrochemical Society by IOP Publishing Limited. This is an open access
article distributed under the terms of the Creative Commons Attribution 4.0 License (CC BY, http://creativecommons.org/licenses/
by/4.0/), which permits unrestricted reuse of the work in any medium, provided the original work is properly cited. [DOI: 10.1149/
1945-7111/abef47]

Manuscript submitted February 8, 2021; revised manuscript received March 7, 2021. Published March 26, 2021.
Supplementary material for this article is available online

The electrolysis or photoelectrolysis of water to generate metal (PGM)-free electrocatalysts and cheaper construction
hydrogen gas provides a promising route to store the intermittent materials.1,2,5,6,8,9 This interest motivates investigating and under-
renewable energy required to power a sustainable civilization.1–6 standing the PGM-free materials that catalyze the OER in an alkaline
The produced hydrogen gas could later be burnt directly as a fuel or environment.1,2,9
could be used to generate electricity in a fuel cell.1–5,7 Besides, Until now, ruthenium (Ru) and iridium (Ir) oxides are considered
hydrogen is already an important feedstock for chemical processes the benchmark electrocatalysts for the OER, but their high price and
such as oil refining, the Fischer–Tropsch synthesis of liquid low availability are limiting factors for their use in large-scale
hydrocarbons, the Haber–Bosch generation of ammonia, and metals production of hydrogen.1,2,4,6,8,10,11 Substitution of these PGM-
refining.4,6,8 Hydrogen is mostly produced by centralized steam- based electrocatalysts with cheap and readily available transition
reforming of fossil fuels, with the greenhouse gas carbon dioxide metal-based electrocatalysts with comparable activity and stability in
(CO2) formed as an undesirable byproduct.4,6,8 The hydrogen gas an alkaline environment would be ground-breaking progress for
produced by water electrolysis is CO2 emission-free provided that making solar hydrogen generators and electrolyzers commercially
the electricity was derived from renewable energy sources such as viable.1,2,4–6,8,9 Iron (Fe) is one of the most abundant elements in the
solar or wind.1,2,6 Notably, water electrolysis/photoelectrolysis must Earth’s crust, it is also cheaper and less toxic than other transition
have increased efficiency and decreased cost to compete with the metals such as cobalt (Co) and nickel (Ni).12 Furthermore, Fe-based
hydrogen generated from the centralized steam-reforming of fossil enzymes/complexes have extensive biological/biomimetic activity
fuels.1–4,6,8 for activation of O2.13,14 Based on these considerations, Fe-based
The process of water electrolysis/photoelectrolysis consists of compounds can potentially be very effective electrocatalysts for the
two half-reactions, namely, the hydrogen evolution reaction (HER) OER. Unfortunately, the early studies on the activity of Fe oxides
and the oxygen evolution reaction (OER). The HER is a relatively showed that they have low OER activity, which has been attributed
simple reaction and occurs at low overpotentials in the presence of to their poor electrical conductivity.10,11 The low activity of Fe
several electrocatalysts.1,2,6,9 The OER, on the other hand, involves oxides hampered their consideration as the OER electrocatalysts in
four successive electron transfers, and therefore, it has sluggish an alkaline environment.1,6,10,11
kinetics.1,2,6,9 The kinetics of the OER mostly limits the efficiency In sharp contrast, it is well-established that when Fe is incorpo-
of water electrolysis/photoelectrolysis at the anode in both acidic rated as a dopant in Ni-based OER electrocatalysts, it significantly
and alkaline environments.1,2,5,6,9 There is a significant interest improves the OER performance.15–20 Interestingly, recent studies
in performing water electrolysis/photoelectrolysis in an alkaline show that Fe is the active site for the OER in these electrocatalysts,
environment as it opens the possibility of using platinum group and Ni provides an electrically conductive and high surface area
matrix for the Fe active sites.15,20 Particularly, it has also been
shown that doping of Fe in Ni modifies the coordination environ-
*Electrochemical Society Fellow. ment of Fe and therefore it optimizes the energetics of interactions
z
E-mail: [email protected]; [email protected] between Fe and active intermediates formed during the OER which
 Journal of The Electrochemical Society, 2021 168 034518

results in a marked increase in the activity of Fe to well beyond that For the synthesis of 6.0 nm C-Fe3O4 nanoparticles, another
of Ni.15 These studies imply that by changing the coordination procedure from the literature was followed.30 Iron(III) acetylaceto-
environment of Fe using strategies such as doping, altering the nate (0.35 g, 1.0 mmol), 1,2-hexadecanediol (1.3 g, 5.0 mmol), oleic
surface facets/crystalographic orientations, introducing surface de- acid (0.85 g, 3.0 mmol), and oleylamine (0.8 g, 3.0 mmol) were
fects/edges/vacancies, and evolving the crystallinity of Fe oxides, introduced in a three-neck flask with benzyl ether (10 ml) at room
much more efficient Fe oxide-based electrocatalysts can be manu- temperature, and the mixture was magnetically stirred. Next, the
factured in the near future.15,20 Inspired by the above results which mixture was flowed with N2 and heated to 200 °C. After two hours
highlight the enormous potential of Fe to act as a cheap and efficient of reaction at this temperature, the mixture was heated to 300 °C and
OER electrocatalyst, a notable increase in the number of the kept at this temperature for one hour before it was cooled down to
scientific papers that are devoted to Fe oxide-based electrocatalysts room temperature. Ethanol (50 ml) was added, and the precipitate
has been observed since 2015.21–27 Unfortunately, the mechanism of was collected by centrifugation at 8000 rpm for 10 min. The product
the OER on Fe-based electrocatalysts is still not clear and an in- was washed with ethanol and redispersed in hexane.
depth experimental and theoritical analysis of the OER mechanism
on Fe oxide-based electrocatalysts will be invaluable for the rational The procedure used to synthesize amorphous FeOx nanoparti-
design of more active Fe oxide-based and likely other oxide-based cles (A-FeOx).—The A-FeOx nanoparticles were obtained by mild
electrocatalysts in the future. oxidation of Fe nanoparticles in the air. The Fe nanoparticles were
In this work, we synthesized a series of size-controlled Fe oxide synthesized following the procedure in the literature.31 In a typical
nanoparticles. We compared the OER activities for the same size synthesis of Fe nanoparticles for the transformation to 10.4 nm
nanoparticles with narrow size distributions and varying crystallinity amorphous FeOx nanoparticles, a mixture of octadecene (20 ml) and
and showed that the amorphous nanoparticles have advantages in oleylamine (0.3 ml, 0.9 mmol) was degassed under N2 at 120 °C for
intrinsic OER electrocatalytic activity. We also performed a sys- 30 min. The temperature was raised to 180 °C, and Fe(CO)5 (0.7 ml,
tematic study on the nanoparticle size-dependent OER performance 5.2 mmol) was injected into the solution. The mixture was kept at
for both crystalline and amorphous nanoparticles. We found that the 180 °C for 10 min before it was cooled down to room temperature.
OER mass activity was linearly correlated to the volume-normalized Ethanol (50 ml) was added, and the precipitate was collected by
edge length of the crystalline nanoparticles and the volume-normal- centrifugation at 8000 rpm for 10 min. The product was washed with
ized surface area of the amorphous nanoparticles. This finding ethanol and redispersed in hexane. Then, the Fe nanoparticles were
allowed us to identify the edge/defect sites on the crystalline/ gradually transformed into amorphous FeOx nanoparticles by
amorphous Fe oxide nanoparticles as the OER active sites. We heating in the air.
also used density functional theory (DFT) calculations to investigate For the 5.2 nm, 6.4 nm, 8.2 nm, and 13.0 nm amorphous FeOx
the OER mechanism on Fe oxide nanoparticles. These calculations nanoparticles, a similar procedure was used, except with a different
elucidated the mechanism of the observed linear correlation between reaction time at 180 °C after Fe(CO)5 was injected. The mixture was
the electrocatalytic activity and the number of edge/defects sites on kept at 180 °C for 45 s, 2 min, 5 min, and 20 min for the synthesis of
the crystalline/amorphous Fe oxide nanoparticles. Our findings 5.2 nm, 6.4 nm, 8.2 nm, and 13.0 nm amorphous FeOx nanoparticles,
demonstrate that increasing the number of edge/defect sites by respectively.
decreasing the size/crystallinity of Fe oxide nanoparticles is a viable
strategy to increase the activity of Fe oxide-based OER electro- The procedure used to synthesize defective crystalline Fe3O4
catalysts. To our best knowledge, this is the first in-depth experi- nanoparticles (D-Fe3O4).—For the synthesis of 10.5 nm D-Fe3O4
mental and theoretical analysis of the OER mechanism on Fe oxide nanoparticles, Fe(CO)5 (0.2 ml, 1.5 mmol) was added to a three-neck
nanoparticles with controlled size and crystallinity, and it provides flask with dioctyl ether (10 ml) and oleic acid (0.85 g, 3.0 mmol) at
novel invaluable strategies to increase the performance of Fe oxide- room temperature, and the mixture was magnetically stirred. Then
based and likely other oxide-based OER electrocatalysts. the mixture was flowed with N2 and heated to 300 °C. After keeping
at this temperature for one hour, the mixture was cooled to room
Experimental temperature, and anhydrous trimethylamine N-oxide (0.17 g,
2.3 mmol) was added. Then the mixture was heated to 250 °C and
The procedure used to synthesize crystalline Fe3O4 nanoparti-
was kept at this temperature for one hour before it was cooled down
cles (C-Fe3O4).—The synthesis followed a slightly modified proce-
to room temperature. Ethanol (50 ml) was subsequently added, and
dure from the literature.28,29 For a typical synthesis of 10.7 nm
the precipitate was collected by centrifugation at 8000 rpm for
C-Fe3O4 nanoparticles, pentacarbonyl iron Fe(CO)5 (0.2 ml,
10 min. The product was washed with ethanol and redispersed in
1.5 mmol) was added to a three-neck flask with dioctyl ether
hexane.
(10 ml) and oleic acid (0.85 g, 3.0 mmol) at room temperature and
the mixture was magnetically stirred. The mixture was subsequently
Preparation of carbon-supported nanoparticle electrocata-
flowed with N2 and heated to 300 °C. After maintaining this
lysts.—High surface area carbon Vulcan XC-72 (VC-72) (25 mg)
temperature for one hour, the mixture was cooled to room
was dispersed in hexane (10 ml) by sonication for 30 min. The
temperature, and anhydrous trimethylamine N-oxide (0.34 g,
nanoparticle (~10 to 20 wt% to carbon support) dispersion in hexane
4.5 mmol) was added. The mixture was then heated to 130 °C under
was added, and then the mixture was sonicated for one hour. Then
N2 atmosphere. After two hours of reaction at this temperature, the
the dispersion was transferred to a ceramic boat, and hexane was
mixture was heated to 300 °C again with a heating rate of 2 °C
evaporated naturally in the air. The ceramic boat was then placed in
min−1. After one hour of reaction at 300 °C, the mixture was cooled
a tube furnace under dry airflow (c.a. 100 ml min−1), heated to 180 °
to room temperature. Ethanol (50 ml) was subsequently added, and
C, and maintained at this temperature for five hours before it was
the precipitate was collected by centrifugation at 8000 revolutions
cooled down to room temperature. The carbon-supported nanopar-
per minute (rpm) for 10 min. The product was washed with ethanol
ticle electrocatalysts were collected and stored.
and redispersed in hexane.
For the 7.9 nm, 8.8 nm, and 13.3 nm C-Fe3O4 nanoparticles, a
similar procedure was followed, except for different oleic acid Physical characterizations.—The sizes and morphologies of the
amounts. The amount of oleic acid introduced was 0.68 g nanoparticles were determined by transmission electron microscopy
(2.4 mmol), 0.77 g (2.7 mmol), and 1.28 g (4.5 mmol) for the (TEM) performed on a JEOL JEM-2010F TEM operating at 200 kV
synthesis of 7.9 nm, 8.8 nm, and 13.3 nm C-Fe3O4 nanoparticles, or a JEOL JEM-3010 TEM working at 300 kV. The size distribution
respectively. was measured based on at least 200 nanoparticles. The energy-
 Journal of The Electrochemical Society, 2021 168 034518

dispersive X-ray spectroscopy (EDS) results were obtained on JEM- The unit cell for the bulk Fe3O4 is composed of 24 Fe atoms and
2010F TEM kV equipped with the EDS system. The crystallinity 32 O atoms. Within the unit cell, there are two distinct Fe species:
and phase of the samples were determined by X-ray diffraction tetrahedrally coordinated Fe (Fetet) composes 1/3 of the Fe atoms,
(XRD) collected on a Bruker D8 Discovery powder diffractometer and octahedrally coordinated Fe (Feoct) composes the remainder of
with Cu Kα radiation (λ = 0.15418 nm) operating at 40 kV and the Fe atoms. The Fetet atoms have a formal charge of +3, while the
40 mA. The thermogravimetric analysis (TGA) was done on a Feoct atoms have formal charges of +2 and +3 at a 1:1 ratio.
Mettler Toledo TGA/DSA 1 STARe System under airflow Additionally, the Fetet and Feoct atoms have an experimentally
(80 ml min−1) with a heating rate of 5 °C min−1. measured magnetic moment of −3.9 μB and +4.0 μB, respectively.
The experimental lattice constant of Fe3O4 is measured to be
Electrode preparation.—Thin-film electrodes of carbon-sup- 8.391 Å (JCPDS card No. 65–3107).
ported Fe oxide nanoparticles were prepared by casting electro- The (100), (111), and (211) surfaces were studied to simulate a
catalyst inks on glassy carbon electrodes (5 mm in diameter, Pine cuboctahedral nanoparticle. The former two correspond to the
Instruments, polished to a mirror-finishing with 0.05 μm alumina). terrace sites of the nanoparticle, while the (211) facet approximates
The electrocatalyst inks were prepared by dispersing 2 mg of carbon- the edges of a nanoparticle. For each surface, different surface
supported electrocatalyst in 1 ml of isopropanol with 0.05 wt % terminations of the bulk Fe3O4 crystal structure were tested, and the
Nafion. The dispersion was sonicated for at least 30 min to form a termination with the lowest surface energy was selected. The energy
homogeneous ink. Next, 4 μl of the electrocatalyst ink was deposited of surface i, γi, was referenced to the energy of the bulk and defined
on a glassy carbon electrode and dried in air at room temperature. as:
This procedure was repeated four more times so that a total amount
of 20 μl electrocatalyst ink was loaded on the electrode, with a total Ei - E Fe 3O4 NFe 3O4
gi = [1]
loading (Fe oxide and carbon support) of 0.2 mg(electrocatalyst + carbon) 2Si
cm−2. The metal contents in the electrocatalysts were measured by
TGA, which were in a range of ∼10 to 20 wt % for different where Ei is the total energy of surface i, E Fe3O4 is the energy of
samples. Thus, the metal loading of the electrode is in a range of the Fe3O4 unit cell in the bulk phase, NFe3O4 is the number of Fe3O4
∼0.02 to 0.04 mg metal cm−2. The OER activates in this paper are units in the supercell, and Si is the area of surface i. The zero-point
normalized to the mass of Fe. energy (ZPE) and entropic corrections for all species were taken
from previously published data at 298.15 K.42 Due to the difficulty
Electrochemical measurements.—Electrochemical studies were of simulating charged systems within the standard DFT framework,
carried out in a standard three-electrode system controlled by a the free energy of the proton and electron is represented by the free
multi-channel potentiostat (Princeton Applied Research). The as- energy of the hydrogen molecule at standard conditions (298 K,
prepared thin film electrode, which served as the working electrode, 1 bar, pH = 0).15,42 Furthermore, since DFT has well-known issues
was mounted onto a rotator and was immersed into a 0.1 M KOH with the treatment of the O2 molecule,15,42 O2 is represented using
solution. A silver-silver chloride (Ag/AgCl) electrode was used as the following free energy relation, also at standard conditions:
the reference electrode, and a platinum (Pt) wire was used as the
counter electrode. All potentials reported in this paper are referenced 2 H2 + O 2  2 H2 O (DG = -4.92 eV) [2]
to the reversible hydrogen electrode (RHE) potential. The zero volt
potential vs RHE was calibrated in the same electrolyte by G (O2) = 2G (H2 O) - 2G (H2) + 4.92 eV [ 3]
measuring HER/HOR currents on a Pt disk electrode, whereby the
potential at zero current corresponds to 0 V vs RHE.32 The The quantity of 4.92 eV results from the free energy of the
polarization curves were recorded in an O2-saturated electrolyte overall reaction (ΔG) used to approximate the O2 molecule.15,42
with a rotation rate of 2500 rpm. All the polarization curves were
corrected for solution resistance, which was measured using AC-
Results and Discussion
impedance spectroscopy from 200 kHz to 100 mHz and a voltage
perturbation of 10 mV. Synthesis and characterization of Fe oxide nanoparticles with
the same sizes but different crystallinities.—To carry out a
Computational methodology.—DFT+U calculations were per- monothetic analysis on the influence of the crystallinity of Fe oxide
formed within the Vienna ab initio simulation package (VASP, on the OER performance, we synthesized three types of Fe oxide
v.5.3.2).33 A 1 × 1 unit cell with two repeated cells in the z-direction nanoparticles with narrow size distributions (average size ∼10 nm,
was used for each surface, with 15 Å of vacuum included in the z- standard deviation less than 10%) and differing crystallinity: crystal-
direction to avoid interaction of the surface with its periodic image. line Fe3O4 (C-Fe3O4), defected crystalline Fe3O4 (D-Fe3O4), and
Valence electrons of Fe and O were represented with a plane-wave amorphous Fe oxide (A-FeOx). Figures 1a, 1f, and 1k show the
basis set and a 400 eV energy cutoff. Core electrons of Fe and O scheme of the three types of nanoparticles, respectively. The Fe
were represented with the projector augmented wave (PAW) oxide nanocrystals were oxidized from Fe nanoparticles. The degree
pseudopotential.34,35 To represent the electronic exchange and of crystallinity was controlled by the application of the oxidant
correlation, the revised Perdew–Burke–Ernzerhof (RPBE) base trimethylamine N-oxide (Me3NO) and the heating rate during
functional was used due to its improved prediction of adsorption synthesis.28,29,31 When Me3NO was added, crystalline Fe oxide
energetics.36 For the 3d electrons of Fe, the Hubbard-type U on-site was obtained, otherwise, amorphous Fe oxide (A-FeOx) was
term was implemented according to the rotationally invariant produced by gradual oxidizing in air. By controlling the heating
approach introduced by Dudarev et al.37 An on-site interaction of rate when the reaction was performed with Me3NO, the defects on
Ueffective = U − J = 3.3 eV was included following previously used the surface could be manipulated. Slow heating results in octahedron
values for Fe.15,38,39 Aspherical corrections to the PAW spheres shape nanoparticles (C-Fe3O4), and fast heating results in irregular
were also included since the use of DFT+U frequently results in nanoparticles with a defected surface (D-Fe3O4). Figures 1b, 1g, and
aspherical charge densities. The Brillouin zone of each surface was 1l show the transmission electron microscopy (TEM) images of the
sampled with a 3 × 3 × 1 Monkhorst-Pack k-point mesh.40 A three types of as-obtained Fe oxide nanoparticles, respectively. They
convergence threshold of 10–4 eV was used for electronic optimiza- are all uniform nanoparticles with narrow size distribution and a
tion. For geometry optimization, the conjugate gradient method was diameter of ca. 10 nm (see the inset size distribution histograms).
used, and all ionic forces were converged within 0.05 eV Å−1. The The selected area electron diffraction (SAED) patterns show that
Davidson-block iteration scheme41 was found to provide the best the C- and D-Fe3O4 are crystalline (Figs. 1c, 1h) in the magnetite
combination of speed and stability for the electronic optimizations. phase (JCPDS card No. 65–3107). In contrast, only diffuse rings
 Journal of The Electrochemical Society, 2021 168 034518

Figure 1. Electron microscopy of the C-Fe3O4 (a)–(e), D-Fe3O4 (f)–(j), and A-FeOx (k)–(o). (a)–(e) Scheme (a), TEM image (b), SAED pattern (c), HRTEM
image (d), and TEM image of the nanoparticle after being supported on carbon (e). (f)–(j) Scheme (f), TEM image (g), SAED pattern (h), HRTEM image (i), and
TEM image of the nanoparticle after being supported on carbon (j). (k)–(o) Scheme (k), TEM image (l), SAED pattern (m), HRTEM image (n), and TEM image
of the nanoparticle after being supported on carbon (o). The histograms inset in (b), (g), and (l) are the size distributions of the nanoparticles.

were observed in the pattern for A-FeOx (Fig. 1m). The crystal- To examine the electrochemical activity of the Fe oxide
linities of the nanoparticles were further confirmed by the X-ray nanoparticles, they were loaded onto a conductive and high surface
diffraction (XRD) patterns shown in Fig. 2a. To isolate the crystal- area carbon support (VC-72). The supported nanoparticles were
linity of Fe oxide electrocatalyst as the key variable in the current further calcined in air to remove the surfactant on the surface of the
study, the electronic structure of A-FeOx must be similar to C-Fe3O4 nanoparticles. Figures 1e, 1j, and 1o show the TEM images of the
and D-Fe3O4. In this regard, Wang and co-authors have reported that calcined supported nanoparticles of C-Fe3O4, D-Fe3O4, and A-FeOx,
amorphous FeOx formed by air oxidation of Fe has similar electron respectively. They demonstrate that the nanoparticles are well
energy loss spectra (EELS) fine structure features on the Oxygen (O) dispersed on the carbon support, and almost no aggregation was
K-edge compared to C-Fe3O4.43 This observation confirms that observed after calcination. The XRD patterns (Fig. 2b) illustrate that
C-Fe3O4 and D-Fe3O4 have a similar electronic structure to that of the crystallinity of the nanoparticles was maintained after the
A-FeOx. calcination. The A-FeOx did not become crystalline after this
The high-resolution TEM (HRTEM) images were further em- thermal treatment. The HRTEM images and SAED patterns (Fig.
ployed to investigate the detailed structure of the nanoparticles. S1 (available online at stacks.iop.org/JES/168/034518/mmedia))
Fig. 1d shows C-Fe3O4 has a cuboctahedral morphology with flat also confirmed that the structure of the nanoparticles did not change.
facets exposing (100) and (111) surfaces. It also indicates a well- The Fe oxide nanoparticle loading on carbon support was tested by
defined lattice fringe with interplanar spacing consistent with that of TGA under an air atmosphere. Figure S2a shows a typical TGA
magnetite. D-Fe3O4 also shows evidence of a lattice fringe; curve. Carbon supports were burned off, and the Fe oxide was
however, numerous defects are present on the surface (indicated further oxidized to α-Fe2O3, which was confirmed by the XRD
by arrows in Fig. 1i). A-FeOx did not show any evidence of a lattice pattern (Fig. S2b). Thus, the Fe content can be obtained from the
fringe due to the lack of crystallinity (Fig. 1n). Hence, the C-Fe3O4 TGA curves, which was in a range of ∼10 to 20 wt.% for different
exposes mainly terrace sites and some edge sites. In addition to those samples.
sites, the D-Fe3O4 has some defect sites. In contrast, the A-FeOx
does not have terrace sites on the surface, and it can be considered as The effect of crystallinity on the OER activity of the same size
a surface full of defects. Fe oxide nanoparticles.—The OER activity was measured by using
 Journal of The Electrochemical Society, 2021 168 034518

Figure 3. Mass normalized polarization curves tested in O2-saturated 0.1 M

KOH at a scan rate of 10 mV s−1. For C-Fe3O4/C, D-Fe3O4/C, and
A-FeOx/C, the curves were normalized to the mass of Fe. For Vulcan XC-
72 (VC-72), the curve was normalized to the weight of 10% of the carbon.

of the VC-72 support is shown to be negligible when testing the

OER performances of the Fe oxide nanoparticles.
The above results demonstrate that crystallinity is an important
parameter in determining the OER activity of Fe oxide nanoparti-
cles. Also, due to the similar particle sizes and narrow particle size
distributions, we conclude that the differences in the intrinsic OER
activities of Fe oxide nanoparticles come from the inherent proper-
ties of the nanoparticles, most probably the distribution of the
different active catalytic sites. More specifically, we hypothesize that
the observed trend in the electrochemical activity of Fe oxide
nanoparticles results from the presence of surface edge/defects sites.
The validity of this hypothesis can be checked by investigating
whether the OER activity of Fe oxide nanoparticles with different
sizes is correlated with the number of edge/defect sites on their
surface.

Synthesis of A-FeOx and C-Fe3O4 nanoparticle series with

different sizes.—Next, we synthesized two series (crystalline series
and amorphous series) of monodisperse Fe oxide nanoparticles with
different sizes to study the effect of size on the OER activity of the
nanoparticles. The nanoparticles with different sizes were synthe-
sized by carefully controlling the synthesis parameters described in
detail in the experimental section. In brief, the sizes of the C-Fe3O4
nanoparticles were mainly controlled by the amount of oleic acid
Figure 2. XRD pattern of the as-obtained Fe oxide nanoparticles (a) and
used in the reaction, and Fig. 4a to e show the TEM images of the as-
these nanoparticles supported on carbon after calcination in the air (b).
obtained nanoparticles. They were all uniform nanoparticles with
narrow size distributions. The sizes of the C-Fe3O4 series were
a standard three-electrode system in an O2-saturated 0.1 M KOH measured as 6.0 ± 1.0, 7.9 ± 0.5, 8.8 ± 0.6, 10.7 ± 0.8, and 13.3 ±
electrolyte. The OER activities were normalized by the metal mass 1.1 nm, respectively. The SAED patterns (Figs. 4a to 4e) confirm
of the electrocatalysts. Here, nanoparticles with the same sizes are that the nanoparticles had good crystallinity. After loading on the
utilized to exclude size effects and mass/charge transport limitations. VC-72 support, the nanoparticles were well dispersed on the support.
Due to the same size of the nanoparticles, they have similar specific The XRD patterns (Fig. S3a) show that the nanoparticles were all in
surface areas; thus, it is reasonable to compare the OER activities the magnetite phase.
directly using mass activity. This method bypasses the difficulty in For the A-FeOx nanoparticles, the sizes were controlled by the
measuring the electrochemical surface area of the supported Fe reaction time, and Figs. 4f to 4j show the TEM images of the as-
oxides nanoparticles. As shown in Fig. 3, the A-FeOx nanoparticles obtained nanoparticles. They were all uniform spherical nanoparti-
demonstrated the best performance, followed by D-Fe3O4 and cles with size of 5.2 ± 0.6, 6.4 ± 0.7, 8.2 ± 0.6, 10.4 ± 0.8, and 13.0 ±
C-Fe3O4. For example, at the current density of 100 mA mg−1, 0.6 nm, respectively. The SAED patterns (Figs. 4f to 4j) show that
A-FeOx shows an overpotential of ∼427 mV compared to ∼495 mV they were amorphous, and the amorphous structure was maintained
for D-Fe3O4 and ∼534 mV for C-Fe3O4. Furthermore, the oxidation after they were loaded on VC-72 support and were subsequently
 Journal of The Electrochemical Society, 2021 168 034518

Figure 4. TEM image, size distribution, SEAD pattern of the Fe oxide nanoparticles and TEM image of the carbon supported Fe oxide nanoparticles. (a)–(e)
C-Fe3O4 series: (a) 6.0 nm. (b) 7.9 nm. (c) 8.8 nm. (d) 10.7 nm. (e) 13.3 nm. (f)–(j) A-FeOx series: (f) 5.2 nm. (g) 6.4 nm. (h) 8.2 nm. (i) 10.4 nm. (j) 13.0 nm.

calcined in the air (Fig. S3b). The TEM images also confirm that the overpotential of 0.55 V, and those of A-FeOx nanoparticles were
A-FeOx nanoparticles were well dispersed on the VC-72 support. calculated at a lower overpotential (0.45 V) because A-FeOx nanopar-
ticles have higher OER activities. For the C-Fe3O4 (Fig. 5a), smaller
The effect of nanoparticle size on the OER activity of C-Fe3O4 nanoparticles showed a higher surface area-specific activity. This
and A-FeOx nanoparticles.—The OER activities of the C-Fe3O4 observation demonstrates that different sites exist on the surface of
and A-FeOx nanoparticle series were studied to provide insight into C-Fe3O4 nanoparticles and that more active sites per unit surface area
the distribution of active catalytic sites on the surface of the are located on the surface of the smaller C-Fe3O4 nanoparticles. By
nanoparticles. Specific activity, which is the activity normalized to contrast, the specific activity of A-FeOx nanoparticles does not show a
the surface area, is used to investigate the size effect. Due to the size effect (Fig. 5b), which indicates that the areal density of active
narrow size distributions of our obtained nanoparticles, the volume catalytic sites is constant for the A-FeOx nanoparticle series.
normalized surface areas can be calculated from the particle sizes To further understand the origin of the size effect for C-Fe3O4
(detailed calculations are shown in Tables SI–SIII). And the surface nanoparticles, we plotted the activities vs different types of sites. There
area of the electrocatalyst can be calculated from the volume are two different sites on the cuboctahedral C-Fe3O4 nanoparticles, the
normalized surface area and the mass of the electrocatalyst. terrace sites (111 and 110 facets) and the edge sites. The portions of
The calculated specific activities at a certain overpotential were these two types of sites are different when nanoparticle sizes are
plotted vs the nanoparticle sizes (Figs. 5a and 5b). We point out that the different (detailed calculations are shown in Table SI). Because our
OER activities of the C-Fe3O4 nanoparticles were calculated at an synthesized C-Fe3O4 nanoparticles have narrow size distributions, we
 Journal of The Electrochemical Society, 2021 168 034518

Figure 5. (a), (b) Surface area-specific activity of the nanoparticle series vs particle sizes. (a) C-Fe3O4 nanoparticles. (b) A-FeOx nanoparticles. (c) the plot of
mass activity at 0.55 V overpotential vs volume-normalized edge length for C-Fe3O4 nanoparticles. (d) the plot of mass activity at 0.45 V overpotential vs
volume-normalized surface area for A-FeOx nanoparticles. The overpotential is marked at the lower right corner. The actual current density normalized to the
geometric area of the rotating disk electrode (mA cm−2 Geometric ) for crystalline nanoparticles (vs normalized edge length) and amorphous nanoparticles (vs
normalized surface area) is shown in Fig. S9. Note, the current densities for A-FeOx nanoparticles are obtained at a relatively lower overpotential (0.45 V) than
the C-Fe3O4 nanoparticles (0.55 V) since the A-FeOx nanoparticles have higher OER activities. The particle size, the volume normalized surface area, and the
volume normalized edge length for crystalline and amorphous nanoparticles are provided in Tables SII and SIII.

can calculate their facet surface areas, the edge lengths between alkaline environments.42 The thermodynamics of the OER proceeds
different facets, and the number of different sites on their surface based via four sequential oxidation steps42:
on the statistical results of the nanoparticle sizes (Tables SI-SIII).
Interestingly, we found that a linear correlation exists between the OER H2 O + *HO* +H+ + e- [4]
mass activity of C-Fe3O4 nanoparticles with the volume-normalized
edge length edge (Fig. 5c), which indicates that the active sites are HO*O*+H+ + e- [5]
located on the edge of C-Fe3O4 nanoparticles. By contrast, the mass
activity of A-FeOx nanoparticles shows a linear dependence on the
O*+H2 O  HOO*+H+ + e- [6]
volume-normalized surface area of the nanoparticles, which is fully
covered by defect sites (Fig. 5d). From these results, we conclude that
only the low coordination sites (i.e., edge/defect sites) are the OER HOO**+O 2 + H+ + e- [7]
active sites on the surface of Fe oxide nanoparticles.
The reaction free energies can be calculated as follows42:
The OER mechanism on metal oxide surfaces.—To understand 1
the origin of the high activity of the edge/defect sites, we studied the DG 1 = G (HO*) + G (H2) - G (H2 O) - G (*) [8]
adsorption properties of different sites for the intermediates pro- 2
duced in the OER processes. Recent theoretical work on the OER
1
demonstrated that a single reaction pathway could universally DG 2 = G (O*) + G (H2) - G (HO*) [9]
describe the OER activity of an electrocatalyst in either acidic or 2
 Journal of The Electrochemical Society, 2021 168 034518

Fe2O3 occurs at an oxygen chemical potential of -3.92 eV (i.e.,

0.25 V vs RHE).48 Therefore, the bulk oxidation process from Fe3O4
1
DG 3 = G (HOO*) + G (H2) - G (O*) - G (H2 O) [10] to Fe2O3 is much more energetically favorable than the surface
2 transition to a Feoct termination, and consequently, observation of
the Feoct-terminated surface is not expected. As mentioned above,
3 the as-synthesized Fe3O4 nanoparticles can be oxidized to Fe2O3 at
DG4 = G (*) + 2G (H2 O) - G (H2) - G (HOO*) + 4.92 eV the pH and electrode potential of relevance to the OER.48 We note,
2
[11] however, that the lowest energy termination of the Fe2O3 (100)
surface, across a broad range of chemical potentials, has a similar
In the equations above, * represents a vacant surface site. The Fetet termination to the Fe3O4 (100) surface considered in the present
Gibbs free energy G for species i is defined as Gi = EDFT + ZPE-TS, work.49 As a result, we believe the qualitative trends predicted from
where EDFT is the total energy of species i, ZPE is the zero-point our Fetet-terminated Fe3O4 (100) surface are still applicable even in
energy of species i, T is the temperature, and S is the entropy of the presence of bulk oxidation from Fe3O4 to Fe2O3. A possible
species i. ΔGn is the free energy change corresponding to the OER reaction pathway proceeding through a Mars-van Krevelen
reactions shown in Eqs. 4–7. mechanism,50 wherein a defect site is consumed and regenerated,
In the above mechanism, the step with the highest ΔG (ΔGmax) has also been calculated for the (211) surface (Fig. S8 and Table
is the rate-determining step for the overall OER, and the sites with SV). Overall, we find the HOO* decomposition step (which
the lowest ΔGmax are the most active sites.42 Next, we will discuss regenerates the vacancy) is highly endothermic; thus, we rule out
the calculated ΔGmax for different sites. the occurrence of a Mars-van Krevelen mechanism at the step sites.
We note that the step edge sites are analogous to the low-
Calculated OER activities for different sites.—The OER process coordination sites in crystalline nanoparticles. The amorphous
is examined at the DFT level to gain an atomistic understanding of nanoparticles, however, do not have long-range ordering and are
the origin of the enhanced OER activity of edge sites. Consistent full of surface defects. Thus, amorphous nanoparticles have high
with the TEM images, we considered a cuboctahedral Fe3O4 concentrations of low-coordination sites at their surface, which
nanoparticle, which exposes six (100) and eight (111) facets, appear analogous to the edge sites of crystalline nanoparticles.
resulting in 24 edges located at the boundary of the (100) and Prior work has demonstrated a significant decrease in the metal
(111) facets. Here, the stepped (211) facet shown in Fig. S4 was used coordination number, a frequently used descriptor for surface
to model the edge, a standard approximation made in the reactivity,46 when transitioning to the amorphous phase.51,52 Due
literature.44,45 The lowest energy terminations of the (100), (111), to the high concentration of these sites on the surface of the
and (211) facets are shown in Fig. S5. amorphous nanoparticles, they have higher OER activity than the
The calculated ΔG for the OER process on the different active corresponding crystalline nanoparticles. Furthermore, since the
sites are shown in Table I. Based on the calculation results, the amorphous nanoparticles are isotropic, no size effect is exhibited (
intermediates are strongly bound to the undercoordinated Fetet i.e., they have the same surface area-specific activity regardless of
surface atoms on the (100) and (111) surfaces. Consequently, the their size).
desorption of the HOO* intermediate is always highly endergonic
for these surfaces. The (100) and (111) surfaces require an over- Future directions/strategies for synthesizing more active Fe
potential of 1.89 V and 0.96 V, respectively. The strong adsorption oxide-based OER electrocatalysts.—We note that there are very
of intermediates observed at the undercoordinated sites is in intriguing and inspiring similarities between our observations and
agreement with recent work by Sautet and co-authors.46 For the those made with molybdenum disulfide (MoS2) as an electrocatalyst
(211) surface, the intermediates adopt a bridge configuration, as for the HER.53–59 Briefly, it was already known for quite some time
shown in Fig. S6. Adsorption is less favorable and is found to that bulk MoS2 was a poor electrocatalyst for the HER.53–59
decrease along the step edge resulting in an overpotential of 0.66 V. However, based on DFT calculations and experimental studies, it
In agreement with the experimental results, the DFT calculations was shown that the edge sites of MoS2 have a good HER
indicate that the step edges are the active site, with reaction activity.53–58 These ground-breaking studies stimulated significant
intermediates adsorbed at the Fetet-Feoct bridge location. Moreover, interest in MoS2 as a HER electrocatalyst. Since then, various
the free energies for the formation of HO* and HOO* are linearly material and electrode preparation methods have been used success-
related with a slope of unity, in agreement with the universal scaling fully to boost the HER activity of MoS2 electrocatalysts.53–59 From
relation proposed by Nørskov and co-authors.42 the viewpoint of materials design, two main promising strategies
It is important to note that the Feoct termination of the (100) have been used successfully to increase the activity of MoS2
surface (Fig. S7) can become more favorable at high oxygen electrocatalysts significantly. (i) Increasing the number of edge/
chemical potential (i.e., oxidative electrode potentials).47 The defect sites per unit surface area of MoS2 electrocatalysts can
Feoct-terminated (100) surface yields a similar overpotential (within produce more active MoS2 electrocatalysts.53,54,57,58 Unfortunately,
∼0.01 eV) to the (211) edge (Table SIV), indicating an enhanced this was very challenging as thermodynamics generally favor the
activity for the OER as the potential increases. As described by presence of facets, limiting the number of edge/defect sites at the
Łodziana, the Feoct-terminated (100) surface of Fe3O4 only becomes surface of a nanoparticle.53,54,58 There are two prominent examples
more energetically favorable than the Fetet surface above an oxygen in the literature that provided solutions to tackle this challenge. First,
chemical potential of −2.2 eV (i.e., 0.68 V vs RHE).47 On the other using nanostructured templates allowed the synthesis of mesoporous
hand, as shown by Morita and Sano, a transition from Fe3O4 to MoS2 electrocatalysts with high surface areas, preferential exposure

Table I. Calculated reaction free energy change for each OER elementary step at 0 V overpotential. The rate-determining step is shown in bold
font.

OER elementary step Reaction free energy change (100) (111) (211)
+ −
H2O+*→HO* + H + e ΔG1 (eV) −2.370 −1.121 −0.133
HO*→ O* + H+ + e− ΔG2 (eV) −0.161 0.960 −0.087
O*+ H2O → HOO* + H+ + e− ΔG3 (eV) 0.646 −0.501 0.660
HOO* → O2 + H+ + e− ΔG4 (eV) 1.885 0.662 −0.440
 Journal of The Electrochemical Society, 2021 168 034518

of edge sites, and excellent transport properties.53,54,58 s, controlling on crystalline/amorphous Fe oxide nanoparticles. DFT calculations
the reaction parameters (e.g., crystallization temperature) during the demonstrated that the edge/defect sites provide bridge sites to adsorb
synthesis introduced abundant defects into the facets of MoS2 the OER intermediates, resulting in high OER activity. The in-depth
nanosheets, which resulted in a marked increase in the activity of experimental and theoretical analysis of the OER mechanism
nanosheets.53,54,59 (ii) The electrocatalytic activity of edge/facet sites performed on Fe oxide-based nanoparticles in this paper provides
on a nanoparticle surface can be enhanced significantly by appro- invaluable insights for the rational design of more active Fe oxide-
priately tuning the electronic structure of the edge/facet sites to based electrocatalysts in the future.
optimize the binding energies of reaction intermediates.53–55,57 This
strategy has been successfully used to enhance the electrocatalytic Acknowledgments
activity of MoS2 nanoparticles by using a promoter (Co) that created
This experimental work was partly supported by GEIRI North
new active sites with comparable activity to active edge sites of
America under State Grid fund SGRI-DL-71-16-015. The computa-
MoS2 nanoparticles.53–55 As the behavior of MoS2 nanoparticles is
tional work was supported by the Catalysis Center for Energy
quite similar to the behavior of Fe oxide nanoparticles, one can
Innovation, an Energy Frontier Research Center funded by the US
simply envision that research efforts along the directions that were
Department of Energy, Office of Science, Office of Basic Energy
previously perused to increase the activity of MoS2 electrocatalysts
Sciences under Award No. DE-SC0001004. S.A.G. acknowledges a
for the HER can be used to enhance the OER activity of Fe oxide
UDEL fellowship. This research used resources of the National
electrocatalysts significantly and to develop new classes of low-cost
Energy Research Scientific Computing Center, a DOE Office of
and efficient OER electrocatalysts.
Science User Facility supported by the Office of Science of the US
Finally, we note that introducing edge/defect sites into transition
Department of Energy under Contract No. DE-AC02-05CH11231.
metal-based materials has resulted in remarkable improvements in
their performance for the OER, the HER, the oxygen reduction
Supporting information
reaction (ORR), the nitrogen reduction reaction (NRR), and the
lithium-ion battery (LIB) storage capacity in recent years.60–63 The Supporting Information is available online.
introduction of edge/defect sites is typically done through novel
synthetic methods, thermal annealing, NaBH4 and hydrogen reduc- ORCID
tion, plasma irradiation, and chemical/electrochemical etching.60–63 Reza Abbasi https://orcid.org/0000-0002-4938-4151
Herein, we mention two recent impactful and engaging papers Yushan Yan https://orcid.org/0000-0001-6616-4575
describing the OER electrocatalysts whose activities were enhanced
significantly by incorporating edge/defect sites. Zhao and coworkers References
recently reported the first successful synthesis of sub-3 nm ultrafine
1. M. Bodner, A. Hofer, and V. Hacker, Wiley Interdiscip. Rev.: Energy Environ., 4,
monolayer NiFe-layered double hydroxide (LDH) nanosheets.60 365 (2015).
They obtained the nanosheets in high yields with narrow size 2. M. Carmo, D. L. Fritz, J. Merge, and D. Stolten, Int. J. Hydrogen Energy, 38, 4901
distributions through an innovative process involving pulsed ultra- (2013).
sonication of monolayer LDH nanosheet precursors in formamide. 3. N. S. Lewis and D. G. Nocera, Proc. Natl Acad. Sci., 103, 15729 (2006).
4. B. Pivovar, N. Rustagi, and S. Satyapal, The Electrochemi. Soc. Interface, 27, 47
These LDH nanosheets contained an abundance of edge/defect sites, (2018).
which resulted in excellent OER performance with an overpotential 5. M. G. Walter, E. L. Warren, J. R. McKone, S. W. Boettcher, Q. Mi, E. A. Santori,
of 254 mV at a current density of 10 mA cm−2. In another similarly and N. S. Lewis, Chem. Rev., 110, 6446 (2010).
impressive work, Zhang and coworkers recently reported the 6. R. Abbasi, B. P. Setzler, S. Lin, J. Wang, Y. Zhao, H. Xu, B. Pivovar, B. Tian,
X. Chen, and G. Wu, Adv. Mater., 31, 1805876 (2019).
synthesis of defect-rich porous monolayer NiFe-LDH nanosheets 7. J. R. McKone, N. S. Lewis, and H. B. Gray, Chem. Mater., 26, 407 (2013).
by an innovative and straightforward one-step process.61 This 8. B. A. Pinaud, J. D. Benck, L. C. Seitz, A. J. Forman, Z. Chen, T. G. Deutsch, B.
synthesis process involved adding an aqueous solution containing D. James, K. N. Baum, G. N. Baum, and S. Ardo, Energy Environ. Sci., 6, 1983
Ni2+ and Fe3+ dropwise to an aqueous formamide solution at 80 °C (2013).
9. K. E. Ayers, E. B. Anderson, C. B. Capuano, M. Niedzwiecki, M. A. Hickner, C.-
and pH of 10, which resulted in the formation of defect-rich porous Y. Wang, Y. Leng, and W. Zhao, ECS Trans., 45, 121 (2013).
monolayer NiFe-LDH nanosheets within only 10 min. The porous 10. S. Trasatti, J. Electroanal. Chem. Interfacial Electrochem., 111, 125 (1980).
NiFe-LDH nanosheets displayed outstanding OER performance (an 11. S. Trasatti, Electrochim. Acta, 29, 1503 (1984).
overpotential of only 230 mV at a current density of 10 mA cm−2). 12. M. Chen, Y. Wu, Y. Han, X. Lin, J. Sun, W. Zhang, and R. Cao, ACS Appl. Mater.
Interfaces, 7, 21852 (2015).
Overall, the above studies, when combined with our observations, 13. I. G. Denisov, T. M. Makris, S. G. Sligar, and I. Schlichting, Chem. Rev., 105, 2253
indicate that introducing defect/edge sites into transition metal (2005).
oxides is a promising approach to develop more active OER 14. E. Y. Tshuva and S. J. Lippard, Chem. Rev., 104, 987 (2004).
electrocatalysts in the future. 15. D. Friebel, M. W. Louie, M. Bajdich, K. E. Sanwald, Y. Cai, A. M. Wise, M.-
J. Cheng, D. Sokaras, T.-C. Weng, and R. Alonso-Mori, J. Am. Chem. Soc., 137,
1305 (2015).
Conclusions 16. G. Młynarek, M. Paszkiewicz, and A. Radniecka, J. Appl. Electrochem., 14, 145
(1984).
Synthetic methods targeting nanometer-sized materials have 17. D. A. Corrigan, J. Electrochem. Soc., 134, 377 (1987).
significantly advanced during the past few years to an astonishing 18. E. L. Miller and R. E. Rocheleau, J. Electrochem. Soc., 144, 3072 (1997).
point where the material with desired size, shape, or crystallinity can 19. X. Li, F. C. Walsh, and D. Pletcher, Phys. Chem. Chem. Phys., 13, 1162 (2011).
be produced easily. These nanomaterials provide an intriguing 20. S. Klaus, Y. Cai, M. W. Louie, L. Trotochaud, and A. T. Bell, J. Phys. Chem. C,
119, 7243 (2015).
opportunity to investigate structure-activity and size-activity rela- 21. N. I. Andersen, A. Serov, and P. Atanassov, Appl. Catal., B, 163, 623 (2015).
tionships for a variety of electrocatalysts. Using this opportunity, in 22. H. Bandal, A. Jadhav, A. Chaugule, W. Chung, and H. Kim, Electrochim. Acta,
this paper, we prepared a series of high-quality monodisperse 222, 1316 (2016).
crystalline/amorphous Fe oxide nanoparticles with different nano- 23. S. Haschke, D. Pankin, Y. Petrov, S. Bochmann, A. Manshina, and J. Bachmann,
ChemSusChem, 10, 3644 (2017).
particle sizes, and we analyzed their OER activities. Using the 24. S. Haschke, Y. Wu, M. Bashouti, S. Christiansen, and J. Bachmann,
nanoparticles of the same size but different crystallinity, we showed ChemCatChem, 7, 2455 (2015).
that amorphous Fe oxide nanoparticles have higher OER activity 25. T. Sharifi, W. L. Kwong, H.-M. Berends, C. Larsen, J. Messinger, and T. Wågberg,
than the crystalline ones. Because electrocatalytic activity measure- Int. J. Hydrogen Energy, 41, 69 (2016).
26. M. Tavakkoli, T. Kallio, O. Reynaud, A. G. Nasibulin, J. Sainio, H. Jiang, E.
ments for the OER correlate linearly with the number of edge/defects I. Kauppinen, and K. Laasonen, J. Mater. Chem. A, 4, 5216 (2016).
sites on the crystalline/amorphous Fe oxide nanoparticles, we 27. R. Sinha, I. Tanyeli, R. Lavrijsen, M. van de Sanden, and A. Bieberle-Hütter,
conclude that the edge/defect sites are the active sites for the OER Electrochim. Acta, 258, 709 (2017).
 Journal of The Electrochemical Society, 2021 168 034518

28. J. Park, E. Lee, N.-M. Hwang, M. Kang, S. C. Kim, Y. Hwang, J.-G. Park, H.- 46. F. Calle-Vallejo, J. Tymoczko, V. Colic, Q. H. Vu, M. D. Pohl, K. Morgenstern,
J. Noh, J.-Y. Kim, and J.-H. Park, Angew. Chem., 117, 2932 (2005). D. Loffreda, P. Sautet, W. Schuhmann, and A. S. Bandarenka, Science, 350, 185 (2015).
29. T. Hyeon, S. S. Lee, J. Park, Y. Chung, and H. B. Na, J. Am. Chem. Soc., 123, 47. Z. Łodziana, Phys. Rev. Lett., 99, 206402 (2007).
12798 (2001). 48. K. Morita and N. Sano, Fundamentals of Metallurgy (Woodhead Publishing,
30. S. Sun, H. Zeng, D. B. Robinson, S. Raoux, P. M. Rice, S. X. Wang, and G. Li, Cambridge, United Kingdom) p. 82 (2005).
J. Am. Chem. Soc., 126, 273 (2004). 49. X. Huang, S. K. Ramadugu, and S. E. Mason, J. Phys. Chem. C, 120, 4919 (2016).
31. S. Peng and S. Sun, Angew. Chem. Int. Ed., 46, 4155 (2007). 50. P. Mars and D. W. V. Krevelen, Chem. Eng. Sci., 3, 41 (1954).
32. R. Abbasi, H. Wang, J. R. Lattimer, H. Xu, G. Wu, and Y. Yan, J. Electrochem. 51. D. B. Buchholz, Q. Ma, D. Alducin, A. Ponce, M. Jose-Yacaman, R. Khanal, J.
Soc., 167, 164510 (2020). E. Medvedeva, and R. P. Chang, Chem. Mater., 26, 5401 (2014).
33. G. Kresse and J. Furthmüller, Comput. Mater. Sci., 6, 15 (1996). 52. T. Yoshida, T. Tanabe, and H. Yoshida, Phys. Scr., 2005, 435 (2005).
34. P. E. Blöchl, Phys. Rev. B, 50, 17953 (1994). 53. M. S. Faber and S. Jin, Energy Environ. Sci., 7, 3519 (2014).
35. G. Kresse and D. Joubert, Phys. Rev. B, 59, 1758 (1999). 54. C. G. Morales-Guio, L.-A. Stern, and X. Hu, Chem. Soc. Rev., 43, 6555 (2014).
36. B. Hammer, L. B. Hansen, and J. K. Nørskov, Phys. Rev. B, 59, 7413 (1999). 55. J. Bonde, P. G. Moses, T. F. Jaramillo, J. K. Nørskov, and I. Chorkendorff, Faraday
37. S. Dudarev, G. Botton, S. Savrasov, C. Humphreys, and A. Sutton, Phys. Rev. B, 57, Discuss., 140, 219 (2009).
1505 (1998). 56. B. Hinnemann, P. G. Moses, J. Bonde, K. P. Jørgensen, J. H. Nielsen, S. Horch,
38. Y.-F. Li and A. Selloni, ACS Catal., 4, 1148 (2014). I. Chorkendorff, and J. K. Nørskov, J. Am. Chem. Soc., 127, 5308 (2005).
39. X. Yu, Y. Li, Y.-W. Li, J. Wang, and H. Jiao, J. Phys. Chem. C, 117, 7648 (2013). 57. T. F. Jaramillo, K. P. Jørgensen, J. Bonde, J. H. Nielsen, S. Horch, and
40. H. J. Monkhorst and J. D. Pack, Phys. Rev. B, 13, 5188 (1976). I. Chorkendorff, Science, 317, 100 (2007).
41. E. Davidson, NATO Advanced Study Institute, Series C (Plenum, New York) 113, p. 58. J. Kibsgaard, Z. Chen, B. N. Reinecke, and T. F. Jaramillo, Nat. Mater., 11, 963 (2012).
95 (1983). 59. J. Xie, J. Zhang, S. Li, F. Grote, X. Zhang, H. Zhang, R. Wang, Y. Lei, B. Pan, and
42. I. C. Man, H. Y. Su, F. Calle-Vallejo, H. A. Hansen, J. I. Martínez, N. G. Inoglu, Y. Xie, J. Am. Chem. Soc., 135, 17881 (2013).
J. Kitchin, T. F. Jaramillo, J. K. Nørskov, and J. Rossmeisl, Chem. Cat. Chem, 3, 60. Y. Zhao, X. Zhang, X. Jia, G. I. Waterhouse, R. Shi, X. Zhang, F. Zhan, Y. Tao, L.
1159 (2011). Z. Wu, and C. H. Tung, Adv. Energy Mater., 8, 1703585 (2018).
43. C. Wang, D. R. Baer, J. E. Amonette, M. H. Engelhard, J. Antony, and Y. Qiang, 61. X. Zhang, Y. Zhao, Y. Zhao, R. Shi, G. I. Waterhouse, and T. Zhang, Adv. Energy
J. Am. Chem. Soc., 131, 8824 (2009). Mater., 9, 1900881 (2019).
44. I. A. Filot, S. G. Shetty, E. J. Hensen, and R. A. van Santen, J. Phys. Chem. C, 115, 62. L. Xiaoqing, C. Shoufu, W. Xiaofei, L. Shaoren, and W. Shuxian, Acta Chim. Sin.,
14204 (2011). 78, 1001 (2020).
45. G. R. Jenness and J. Schmidt, ACS Catal., 3, 2881 (2013). 63. X. Yan, Y. Jia, and X. Yao, Small Structures, 2, 2000067 (2021).