Laras Fadillah, Damian Kowalski, Yoshitaka Aoki, Hiroki Habazaki: Compositional variations in anodic

nanotubes/nanopores formed on Fe 100, 110 and 111 single crystals., Electrochimica Acta 364
(2020) 137316. https://doi.org/10.1016/j.electacta.2020.137316. IF: 6.2, Q1

Laras Fadillah, Kentaro Takase, Hikaru Kobayashi, Sylwia Turczyniak-Surdacka, Marcin Strawski,
Damian Kowalski, Chunyu Zhu, Yoshitaka Aoki, Hiroki Habazaki: The Role Of Tungsten Species In
The Transition Of Anodic Nanopores To Nanotubes Formed On Iron Alloyed With Tungsten.,
Electrochimica Acta 309 (2019) 274-282, https://doi.org/10.1016/j.electacta.2019.03.206. IF:
6.2, Q1

Chunmei Tang, Katsuya Akimoto, Ning Wang, Laras Fadillah, Sho Kitano, Hiroki Habazaki, and
Yoshitaka Aoki.; Effect of Anode Functional Layer on Steam Electrolysis Performances of
Protonic Solid Oxide Cells., Journal of Materials Chemistry A (2021).
https://doi.org/10.1039/D1TA02848K. IF: 11.3, Q1

Ruijie Zhu, Huijun Yang, Laras Fadillah, Zetao Xiong, Damian Kowalski, Chunyu Zhu, Sho Kitano,
Yoshitaka Aoki, Hiroki Habazaki.; Lithiophilic carbon scroll as Li metal host with low tortuosity
design and “Dead Li” self-cleaning capability., Journal of Materials Chemistry A (2021).
https://doi.org/10.1039/D1TA02491D. IF: 11.3, Q1

Nan Sheng, Jiahui Lu, Jingdong Hu, Ruijie Zhu, Laras Fadillah, Cheng Wang, Chunyu Zhu, Zhonghao
Rao, Hiroki Habazaki; Synthesis of Sn@SnO2 core-shell microcapsules by a self-oxidation
strategy for medium temperature thermal storage., Chemical Engineering Journal 420 (2021)
129906. https://doi.org/10.1016/j.cej.2021.129906. IF: 10.7, Q1
 Electrochimica Acta 309 (2019) 274e282

Contents lists available at ScienceDirect

Electrochimica Acta
journal homepage: www.elsevier.com/locate/electacta

The role of tungsten species in the transition of anodic nanopores to

nanotubes formed on iron alloyed with tungsten
Laras Fadillah a, Kentaro Takase b, Hikaru Kobayashi b, Sylwia Turczyniak-Surdacka c,
Marcin Strawski c, Damian Kowalski b, c, *, Chunyu Zhu b, Yoshitaka Aoki b,
Hiroki Habazaki b, **
a
Graduate School of Chemical Sciences and Engineering, Hokkaido University, Kita-Ku Kita 13, Jo Nishi 8, Sapporo, Hokkaido 060-8628, Japan
b
Division of Applied Chemistry, Faculty of Engineering, Hokkaido University, Kita-Ku Kita 13, Jo Nishi 8, Sapporo, Hokkaido 060-8628, Japan
c
Faculty of Chemistry and Biological and Chemical Research Centre, University of Warsaw, Zwirki i Wigury 101, 02-089, Warsaw, Poland

a r t i c l e i n f o a b s t r a c t

Article history: The effect of alloying of sputter-deposited Fe with 9 at.% tungsten on the growth of nanoporous anodic
Received 5 January 2019 oxide was studied in ethylene glycol electrolyte containing 0.1 mol dm
3 ammonium fluoride and
Received in revised form 1.5 mol dm
3 water. The classic nanoporous anodic film (Al2O3-like) was developed on pure Fe while the
29 March 2019
transition of nanopores to nanotubes (TiO2-like) was observed for anodizing of Fe-W alloy. The pores/
Accepted 30 March 2019
Available online 2 April 2019
nanotubes having average diameter 50e110 nm and 30e60 nm on pure Fe and Fe-W alloy anodized at
voltage 40e60 V, respectively. Both nanoporous/nanotubular anodic films grow in line with the field
assisted flow model with a few fundamental details: i) transition of nanopores to nanotubes is observed
Keywords:
Anodizing
upon anodizing of Fe-W alloy, ii) significant reduction of the cell size (nanotube diameter) is obtained on
Iron oxide Fe-W alloy, iii) relatively thick layer is produced at Fe-W alloy/oxide interface. The primary reason of this
Nanotubes transition to nanotubes as well as chemical changes is discussed in view of effective modification of the
Anodic oxide cell boundary region with tungsten species, probably WF6 compound, upon growth of anodic film under
influence of high electric field strength. The possible reason of developing the space in between nano-
tubes is faster kinetics of WF6 reaction with water over the presence of low solubility FeFx species.
Alloying of iron is one of the effective ways to modify the nanostructure of the anodic film on iron.
© 2019 Elsevier Ltd. All rights reserved.

1. Introduction
Anodizing is a high voltage electrochemical conversion process
that forms barrier-type oxide layers or self-organized nanoporous/
nanotubular structures. So far, the Al2O3-like nanopores and TiO2-
like nanotubes could be successfully synthesized on valve metals
and alloys such as tantalum, niobium, zirconium, hafnium, etc [1].
Very recently, the formation of self-organized nanoporous and
nanotubular anodic films has also been reported on non-valve
* Corresponding author. Division of Applied Chemistry, Faculty of Engineering,
Hokkaido University, Kita-Ku Kita 13, Jo Nishi 8, Sapporo, Hokkaido 060-8628,
Japan.
** Corresponding author. Division of Applied Chemistry, Faculty of Engineering,
Hokkaido University, Kita-Ku Kita 13, Jo Nishi 8, Sapporo, Hokkaido 060-8628,
Japan.
E-mail addresses: damian.kowalski@chem.uw.edu.pl (D. Kowalski), habazaki@
eng.hokudai.ac.jp (H. Habazaki).
metals, including iron and stainless steel [2e9]. The surface
modification of iron and its alloys is highly demanded in view of its
essential role in the field of corrosion and large steel construction
market [10]. In addition to corrosion resistance of iron alloys, the
surface modifications are necessary to provide additional func-
tionalities such as superhydrophobic and superoleophobic prop-
erties for anti-ice and anti-snow sticking, self-cleaning, anti-
fouling, etc [11].
Anodizing of iron is practically more complex in processing in
relation to other valve metals such as Ti, Ta, Nb, etc., due to diffi-
culty of keeping the passive sate of iron. Growth of porous anodic
film is therefore not favored in aqueous electrolytes since at high
potential, formation of soluble ferrate ions is promoted [12,13] as
well as anodic gas generation. The key for growth of thick porous
anodic layer on iron is therefore good control over the passivity/
corrosion which can be effectively provided in organic electrolytes
containing small amounts of water and fluorides. The morphology
of the anodic layer is mainly controlled by the voltage/current in

https://doi.org/10.1016/j.electacta.2019.03.206
0013-4686/© 2019 Elsevier Ltd. All rights reserved.
 L. Fadillah et al. / Electrochimica Acta 309 (2019) 274e282
the anodizing process. In ethylene glycol electrolyte containing
0.1 mol dm
3 ammonium fluoride and 1.5 mol dm
3 water, the
nanopores (Al2O3-like) and nanotubes (TiO2-like) can be formed on
iron. The nanoporous-type anodic film is preferentially formed on
iron, possibly due to relatively slow kinetics of dissolution of
fluoride-enriched cell boundary regions [8]. The partial transition
of nanopores to nanotubes is typically observed at enhanced volt-
ages/currents. The primary reason of this transition is formation of
fluoride rich layer at metal/anodic-film interface [14] and its
transfer to the cell boundary region, once the porous film grows by
field assisted plastic flow [8]. Although the formation of nanopores
on aluminum is quite well understood, some features of the
nanotubular film formation are still the subject of debate.
The insight into the growth mechanism may be achieved by
tracer studies as demonstrated for the alumina case by using 18O
isotope tracer technique [15]. The key data obtained by means of
the nuclear microanalysis is the information on the contribution of
oxygen transport in the growing of the film and therefore direct
indication on relative ratio of film formed at metal/oxide interface
to that formed at oxide electrolyte interface. It was demonstrated
that both Al3þ and O2
ions migrate in anodic alumina, with 40% of
the ionic current carried by the aluminum ions while about 60%
transported by oxygen anions. The pore formation resulted from
the loss of aluminum cations to the electrolyte with reduced
growth efficiency close to 60% for porous film [15]. A number of
information related to growth mechanism were obtained by
introducing W tracer into the structure of anodic film by Garcia-
Vergara et al. [16e22]. The mechanical stress essentially triggered
by electrostriction forces and change of the oxide volume ratio
described by PillingeBedworth ratio is presumed to be responsible
for the flow of film material from pore base to pore wall and hence
the formation of pores [23]. Based on the tungsten tracer studies,
the mechanism of pore generation and development in porous
anodic alumina films was explained by flow mechanism based on
direct experimental evidences. Plastic flow of alumina beneath the
pore in the presence of ion transport and comprehensive stresses
displaces the material from the barrier layer towards the cell wall
region, and is predicted to contribute to an increased thickness of
porous film relative to that of metal consumed. The STEM obser-
vation showed that tungsten band is distorted, flowing upward
toward the cell wall regions from the pore base as pore develops.
These results indicate the flow of film material owing to the plas-
ticity and high compressive stresses in barrier layer of porous
anodic films.
In the present study, the anodizing of Fe-W alloy was studied for
the first time. The results were compared with that of anodizing of
pure Fe for a better understanding the growth mechanism of the
anodic films. Moreover, the influence of operating conditions,
especially anodizing voltage on the geometry of nanopores
including pore diameter, was analyzed. The primary interest in this
study was to examine the influence of tungsten on the growth
process as well as chemical changes of formed oxide.
2. Experimental
Magnetron-sputtered Fe and Fe-W (9 at.%) films having the
thickness of approximately 450 nm were used for anodizing. The
target consisted of 99.99% pure iron disk of 100 mm in diameter
and one 99.9% tungsten disk of 20 mm in diameter placed on the
sputter erosion region of the iron disk. Sputter deposition was
carried out in 99.999% pure argon atmosphere at 2.5  10
3 Torr
with the current 0.5 A and voltage 350 V for 15 min. The compo-
sition of the alloy was evaluated by electron probe micro analyzer
(EPMA) and equaled to 9 at.% tungsten.
The Fe and Fe-W alloys were anodized at 40e60 V for 300 s in
275
ethylene glycol electrolyte containing 0.1 mol dm
3 ammonium
fluoride and 1.5 mol dm
3 deionized water at room temperature in
o-ring type cell with a two-electrode system using platinum
counter electrode without stirring of electrolyte. After anodizing,
the specimens were rinsed in ethylene glycol, ethanol, and then
dried in nitrogen stream.
The surface morphology and cross sections of the specimens
were examined by a Zeiss Sigma 500 field emission scanning
electron microscope (SEM) instrument. Elemental depth profiles of
the specimens were obtained by a Horiba Jobin Yvon 5000 RF glow
discharge optical emission spectrometer (GDOES) in neon atmo-
sphere at 800 Pa by applying 35 W RF power under a pulse mode.
The data were recorded at a sampling interval 0.02 s. The diameter
of sputtered area was 4 mm. The cross-sections of anodic films were
examined by a JEOL, JEM-ARM200F scanning transmission electron
microscope (STEM) with energy dispersive X-ray spectrometry
(EDS) facilities. Electron-transparent sections were prepared by a
Hitachi FB-2100 focused ion beam (FIB) system employing a Gaþ
ion beam. X-ray photoelectron spectroscopy (XPS) spectra were
acquired using a Kratos Axis Supra spectrometer, equipped with a
monochromatic Al Ka radiation (1486.7 eV) source. The instrument
work function was calibrated to give a BE of 84.0 eV ± 0.1 eV for the
4f7/2 line of metallic gold and the spectrometer dispersion was
adjusted to give a BE of 932.62 eV for the Cu 2p3/2 line of metallic
copper. Energy resolution was examined on silver sample. Survey
(wide) spectra were collected with a quality corresponding to the
full width at half maximum (FWHM) parameter for Ag 3d line equal
to 0.71 eV at energy step size equal to 0.5 eV. For high resolution
spectra the FWHM parameter for Ag 3d line was equal to 0.58 eV at
step size was: 0.1 eV. Kratos charge neutralizer system was used to
reduce charge compensation. The peaks fitting was conducted us-
ing CasaXPS software version 2.3.18 on a Shirley background. All
spectra were calibrated using the adventitious C 1s peak with a
fixed value of 285.0 eV. The in-depth composition of the films was
probed by using secondary-ion mass spectroscopy (SIMS). As-
received samples were transferred without special pre-treatment
to the analytical chamber where the pressure was 7  10
10 Torr.
Distribution of elements was obtained with a time-of-flight SIMS
apparatus (TOF SIMS 5, ION-TOF GmbH) operating in dual beam
mode. Samples were sputtered by Csþ ions (2 keV, 135 nA), rastered
over 320 mm by 320 mm area. Exposed this way, internal layers of
films were analyzed with use of Biþ ions (at 64 mm by 64 mm central
region only). The primary ion (target) current was typically 1.2 pA.
Low energy electron flood gun was used to reduce sample charging
during analysis. The internal mass calibration was performed using
several ions from the mass range starting from C
up to W2O
6.
3. Results and discussion
3.1. j-t curves and morphology of anodic films
Anodizing of sputter-deposited Fe and Fe-W (9 at.%) alloy was
performed at a constant voltage between 40 V and 60 V in ethylene
glycol electrolyte containing 1.5 mol dm
3 water and

3
0.1 mol dm NH4F at room temperature for 300 s. Fig. 1 shows the
current-time (j-t) transients obtained during anodizing of both
specimens. Two distinct stages are clearly visible on j-t curves; the
first stage of anodizing, observed within the first tens of seconds is
characterized by rapid current drop and is typically associated with
initial formation of barrier-type oxide at the alloy electrolyte
interface [1] and possible pore initiation. The second stage of
anodizing is a quasi-steady state where the current density is
nearly constant. The time scale for the stage I is independent of the
applied voltage, but the magnitude of current drop slightly in-
creases with anodizing voltage. The current drop in stage I region is
276 L. Fadillah et al. / Electrochimica Acta 309 (2019) 274e282
Fig. 1. Current-time (j-t) responses for anodizing of sputter deposited (a) Fe and (b) Fe-
W (9 at.%) alloy at 40 V, 50 V and 60 V in an ethylene glycol electrolyte containing
1.5 mol dm
3 of water and 0.1 mol dm
3 of NH4F at room temperature.
seven times faster for Fe-W alloy than that for pure Fe so that a
quasi-steady state anodizing condition is more rapidly established
on the Fe-W alloy. An initial anodizing stage is more distinct by
plotting j-t curves in the logarithmic scale [24,25]. This is apparent
when one looks at the j-logt graph (Fig. 1) and compares charge
passed during the first stage of anodizing of 2.2 and 0.9 C cm
2 for
Fe and Fe-W, respectively, estimated at 40 V where the contribution
of oxygen evolution reaction, involved during anodizing of Fe and
Fe-W alloy in this electrolyte, is significantly reduced [8]. During
anodizing of Fe and Fe-W alloy, oxygen evolution was observed by
naked eye for anodizing at 60 V and became less significant at lower
voltages. The shape of j-t curve for Fe in Fig. 1a strongly depends on
the applied voltage and is characterized by a steady state current
density at 40 V while the current density gradually increases at 50
and 60 V probably due to Joule heat effect [8]. The current density
for anodized Fe-W alloy in Fig. 1b is slightly higher than the current
density for pure Fe in Fig. 1a due to the higher electric field applied
over a thinner barrier layer formed on Fe-W alloy as discussed later.
The total charge passed during 300 s anodizing of Fe and Fe-W at
40 V equals to 10 and 12 C cm
2, respectively. The current drop
observed during last 20 s of anodizing of Fe-W alloy at 60 V results
from consumption of the alloy above glass substrate and subse-
quent narrowing of the surface area available for oxidation of Fe-W,
having negligible effect on the final structure.
The cross-section SEM images for Fe-W alloys anodized at
40e60 V for 300 s are demonstrated in Fig. 2 with an anodic layer
formed on pure Fe at 60 V shown as a reference. The thickness of
anodic layer varies with applied voltage and equals 0.98, 1.28 and
1.76 mm for alloy anodized at 40, 50 and 60 V, respectively, aligned
with higher electric field at higher anodizing voltage. The thickness
variations correlate very well with the applied total electric charge
(Qt) of 12, 15 and 20 C cm
2 upon anodizing at 40, 50 and 60 V,
respectively. This suggests that the enhanced oxygen gas genera-
tion observed at higher anodizing voltage is due to the increased
current density, not the increased current efficiency for gas gen-
eration. Anodizing of Fe-W alloy at 60 V results in formation of
1.76 mm thick anodic layer which is approximately the same as
anodic film formed on pure Fe shown as a reference in Fig. 2. The
total electric charge of 17 C cm
2 is required for anodic film
development on pure Fe which is 15% less than that for Fe-W case at
the same anodizing conditions. The slight difference in Qt value
may be related to the thickness/morphology changes and/or oxy-
gen evolution reaction aspects.
The classic Al2O3-like nanopores are developed on pure Fe as
expected for short anodizing time and relatively low voltage im-
pacts according to the previous study [8]. The detailed analysis of
the anodic layer formed on Fe-W alloy uncovers the shape of TiO2-
like nanotubes (Fig. 2). The inset of SEM image of Fe-W alloy at 60 V
in Fig. 2 shows that the bottom part of the anodic film is composed
of the array of nanotubes.
The surface morphology of nanoporous/nanotubular anodic film
formed by anodizing of pure Fe and Fe-W alloy at 40e60 V is shown
in Fig. 3. The microstructure of the as-deposited substrate surface
(Fig. 3 a,e) has an impact on the surface morphology and self-
organization of pores/tubes particularly observed at lower anod-
izing voltages for Fe-W case where the formed nanopores/nano-
tubes are organized along the longitudinal grains (Fig. 3 f-h). Such
arrangement can be clearly understood in terms of preferential
nucleation of the pores in the dimples [26] of the textured alloy
which slightly differs from that of pure Fe. The pore diameter lin-
early increases with anodizing voltage 40e60 V from 50 to 110 nm
and from 30 to 60 nm for anodic film formed on pure Fe and Fe-W
alloy, respectively (Fig. 3). The effective cell size reduction is
therefore observed with alloying of the iron with tungsten.
3.2. Composition of anodic films
Fig. 4a and b shows GDOES elemental depth profiles, displaying
the change in the optical emission intensities from oxygen, iron,
fluorine and tungsten with the time of sputtering for Fe and Fe-W
anodized at 60 V. The depth distribution of Fe, W and O clearly
identifies the interface between alloy and formed anodic film. The
total time required to sputter through the thickness of the porous
anodic film is approximately 22 and 54 s for pure Fe and Fe-W alloy,
respectively, resulting mainly from different sputtering rates of Fe
and W. The GDOES depth profile revealed the incorporation of
fluorine species from electrolyte and their distribution in anodic
layer strongly depending on alloying element. The enrichment of
fluoride in porous anodic film immediately above the metal/film
interface formed on pure Fe should be associated with faster
migration of F
ions relative to that of O2
ions under the high
electric field during film growth [14]. The accumulation of fluorides
at the metal/oxide interface for pure Fe case is evident from high
resolution STEM/EDS mapping studies shown in a previous study
[8]. The essential difference in the element distribution for porous
films developed on Fe and Fe-W alloy is the unusual distribution of
fluorine, i.e. the absence of F peak at alloy/oxide interface for Fe-W
system. Another interesting feature for anodic film developed on
Fe-W alloy is the presence of oxygen peak at the first 10 s of sput-
tering time suggesting modification of surface region, both dis-
cussed in the next section of this paper.
The qualitative analysis of the surfaces for Fe and Fe-W alloy
anodized at 60 V was performed based on the XPS wide scan
spectra. Results allowed to confirm that the nanoporous/nano-
tubular anodic films contained Fe, W, F, C and O elements (ESI). The
 L. Fadillah et al. / Electrochimica Acta 309 (2019) 274e282 277

Fig. 2. Cross-section SEM images of nanopores/nanotubes formed by anodizing of Fe-W alloy at 40, 50 and 60 V. The anodic layer formed on pure Fe at 60 V is shown as a reference.
The metal/alloy layers were anodized in ethylene glycol electrolyte containing 1.5 mol dm
3 of water and 0.1 mol dm
3 of NH4F at room temperature.

Fig. 3. Surface morphology images for (a,e) as deposited Fe and Fe-W alloys, (bed) Fe and (feh) Fe-W (9 at.%) alloy anodized at 40e60 V in ethylene glycol electrolyte containing
1.5 mol dm
3 of water and 0.1 mol dm
3 of NH4F at room temperature.

Fig. 4. Glow discharge optical emission spectroscopy (GDOES) depth profiles for anodic layers formed at 60 V on a) pure Fe and b) Fe-W alloy.

high-resolution Fe 2p region spectra of anodic films formed on Fe
and Fe-W, shown in Fig. 5 a,b (ESI), consist of two peaks assigned to
the Fe 2p3/2 and Fe 2p1/2 spin-orbits components (D ¼ 13.2 eV). Iron
oxidation state was examined by curve-fitting the Fe 2p3/2 region
with two well resolved components located at 711.3 and 714.2 eV.
The resulting fit suggests coexistence of Fe(II) and Fe(III) oxidation
states. The relative concentration of these two components has not
been determined since Fe 2p region overlaps with F 1s plasmons,
however, the shoulder at 714.2 eV is more pronounced for anodic
porous film formed on pure Fe. According to literature data, analysis
of the lower binding energy peak is not unequivocal. This peak can
be related to the presence of Fe(II) in FeF2 [27e29], however, Fe(III)
in Fe2O3 and FeOOH were also reported close to this position
[29,30]. So far, published results revealed that signal around 710 eV
suggests the existence of FeO, therefore the presence of FeO cannot
be excluded. The peak at energy even higher than 714.2 eV, which
278 L. Fadillah et al. / Electrochimica Acta 309 (2019) 274e282

Fig. 5. Fe2p, F1s, O1s, W4f and C1s core level spectra for a) pure Fe anodized at 60 V, b) Fe-W alloy anodized at 60 V. Negative ion ToF-SIMS mass spectra (c,d) and depth profile (e)
for Fe-W alloy anodized at 60 V for 300s in ethylene glycol electrolyte containing 1.5 mol dm
3 of water and 0.1 mol dm
3 of NH4F at room temperature.

was found in this work, is assigned to the FeF3 [27,28]. In addition,
binding energy of F 1s (648.8 ± 0.2 eV) and its symmetrical shape is
in accordance with the literature given for metal fluoride [27,28,31].
The presence of FeFx is consistent with identification of fluoride-
rich-layer by Shahzad et al. [8] by means of STEM/EDS elemental
mapping in the transition of anodic nanopores to nanotubes
formed on bulk iron. The primary location of FeFx is the cell
boundary and oxide/metal interface [8]. In the XPS spectrum a very
intense signal of O 1s region was recorded. The spectrum can be
resolved into two peaks for anodic Fe or three peaks for anodic Fe-
W. The high binding energy component O 1s (532.0 eV) recorded
for anodic Fe is related to the presence of organic impurities, which
 L. Fadillah et al. / Electrochimica Acta 309 (2019) 274e282 279

is confirmed in C 1s region (signal from C-O, C-O-C, C-C]O, car-

bonates). The other peak, at 530.3 eV comes from lattice oxygen
which confirm the presence of FexOy. The position of this peak
slightly shifts into higher binding energy (530.5 eV) after addition
of tungsten, which is similar to the O 1s values for WO3 [32].
Additionally, 531.7 and 533.5 eV appear. High binding energy peaks
can be assigned, like previously, to the presence of organic and
adsorbed water, respectively. Based on the W 4f region it can be
excluded that the peak around 531.7 eV is related to the presence of
WO3-x phase, caused by the existence of hydroxyl groups on the
surface. The W 4f has well separated spin-orbit components
(D ¼ 2.15 eV). The position of W 4f 7/2 (35.8 eV) and its symmetrical
shape suggest presence of only WO3 phase [33e35].
ToF-SIMS mass spectra, cumulated during the depth profiling
measurement, were acquired in negative mode to avoid the influ-
ence by an external oxygen source (Fig. 5 c,d). The spectra show
many components which can be assigned to the iron oxides, iron
fluorides, tungsten oxides and tungsten oxide fluorides ion frag-
ments. For detailed analysis the 225e325 m/z region was chosen as
the most characteristic ions produced at the sample surface are
WOx clusters [36]. Anodizing of tungsten or its alloys typically re-
sults in formation of WO3 domains in anodic film structure [37,38].
The complexity of the spectrum is affected by the presence of
oxide-fluoride forms. Tungsten has five isotopes of which four are
present in the predominant quantity and relatively similar amount.
This isotopic fingerprint allowed us to distinguish the contribution
of individual fragments from groups of peaks that occurs at 230,
246, 271 and 293 m/z values. Fig. 5c shows that WO
3 and various
WOxF



y , such as WOF4 and WO2F2 , could be assigned. Since WF6
form was our main point of interest a zoomed data at
292.5e300.5 m/z range is given in Fig. 5d proving that WF6 mole-
cule is present in the film structure. The depth profile experiment
allowed to monitor the content of above fragments in the nano-
tubular film structure. Fig. 5e shows that various fluorides forms are
stable among the film depth. A typical behavior for ToF-SIMS
measurements in the depth profiling mode is the stabilization of
the etching process, therefore, the content of individual fragments
in the first stage undergo non-specific changes during the first 60 s.
It is obvious that the only component that intensity change differs
from the other fragments is WF
6 . A clear disappearance of this Fig. 6. High angle annular dark field (HAADF) images of FIB lamella for Fe-W alloy
component in the near-surface region is visible. We assume that anodized at 60 V in ethylene glycol electrolyte containing 1.5 mol dm
3 of water and
0.1 mol dm
3 of NH4F at room temperature.
this part of the film has been exposed to air/water and this reactive
compound has decomposed as discussed in the next section of the
paper. 3.4. Growth of nanoporous/nanotubular anodic films

The above findings reveal that alloying of iron with 9 at.% of

3.3. Analysis of film by STEM
The scanning transmission electron micrograph (STEM) of a FIB
lamella for Fe-W alloy anodized at 60 V is shown in Fig. 6. The shape
of nanotubes is evident from high angle annular dark field (HAADF)
images. The thickness of the anodic film is approximately 1.7 mm
with the inner nanotube diameter of 60 nm, consistent with pre-
vious SEM observations. The regions with lighter appearance on
STEM micrograph (Fig. 6b) are corresponding to nanotube walls
composed of mixed tungsten/iron oxide containing fluorides.
Relatively thick layer at the alloy/oxide interface was detected
(Fig. 6c). Its thickness varies across the FIB lamella from 40 to
130 nm as demonstrated in low resolution STEM image (Fig. 6d).
Such large thickness variations cannot be caused by electro-
chemical reaction as the high electric field is primarily localized in
the barrier-layer forming uniform anodic layer depending on the
applied voltage. Those thickness variations supported by unusual
GDOES elemental depth distribution suggest chemical nature of its
formation.
tungsten results in structural changes upon anodizing in ethylene
glycol electrolyte containing 1.5 mol dm
3 water and 0.1 mol dm
3
ammonium fluoride. Essentially, two types of anodic structures are
observed: Al2O3-like nanopores developed on pure Fe, and TiO2-
like nanotubes developed on Fe-W alloy. Schematic illustration of
formed structures is demonstrated in Fig. 7.
Two fundamental mechanisms have been proposed for forma-
tion of nanoporous/nanotubular anodic films: i) field assisted
dissolution model [39], and ii) field assisted flow model [20,22,40].
Very recently Shahzad et al. showed that porous anodic film on iron
may grow according to both models critically depending on the
water concentration in electrolyte [8]. In the classic field assisted
dissolution model the thickness of barrier-layer between metal and
porous layer is kept constant under constant voltage. The formed
structure is typically characterized by i) presence of less organized
nanopores, ii) thick barrier-layer, and iii) flat metal/oxide interface.
In the field assisted flow model, which was proposed using tung-
sten tracer studies [21,22,40,41], the resulting structure is typically
280 L. Fadillah et al. / Electrochimica Acta 309 (2019) 274e282

Fig. 7. Schematic illustrations for proposed growth of nanopores/nanotubes on a) iron, and b) iron tungsten alloy, c) illustrates post reaction with WFx compounds leading to
formation of thick layer at metal/oxide interface.

characterized by i) presence of nanopores or nanotubes with high

degree of self-organization, ii) thin barrier layer, iii) scalloped
metal/oxide interface.
The key mechanism for the growth of anodic film on Fe-W alloy
is the field assisted flow model, as only the plastic flow may rear-
range the initial horizontal position of anodic layers into vertical
position and therefore develop nanotube with the composition of
the -outer and -inner shell corresponding to -bottom and -top of
the barrier layer, respectively [1]. The general growth is analogous
to that formed on pure Fe in electrolyte containing 1.5 mol dm
3
water [8] except a few fundamental aspects: i) the transition of
nanopores to nanotubes is observed by alloying with tungsten, ii)
significant reduction of the cell size (nanotube diameter) is ob-
tained on Fe-W alloy, iii) relatively thick layer is produced at alloy/
oxide interface.
The nanotubes instead of nanopores are considered to be
formed due to faster kinetics of WF6 (or similar compound)
dissolution. WF6 domains and their decomposed forms have been
detected on ToF-SIMS mass spectra in the nanotubular film (Fig. 5).
Taking into account fast inward-migration of F
[14] and very slow
outward-migration of W6þ species [37] in anodic film, the primary
location of WF6 compound should be the bottom of barrier-layer Fig. 8. Pores size as a function of anodizing voltage for anodic films formed on Fe and
which is switched to the cell boundary position due to stress Fe-W alloy anodized in ethylene glycol electrolyte containing 1.5 mol dm
3 of water
generated at the barrier-layer and material flow according to the and 0.1 mol dm
3 of NH4F at room temperature.

field assisted flow model as discussed above. The difficulty in

detection of WF6 is its decomposition due to hydrolysis reaction
The origin of a thick layer formed at metal/nanotubular-film
[42] in contrast to FeF2 which has low solubility constant of
interface observed on STEM image (Fig. 6a) may be misidentified
8.3  10
3 mol dm
3 [43]. WF6 is generally considered to be a gas at
and at first glance could suggests that alloying of iron with tungsten
standard conditions [42,44], however, may be stabilized by the
leads to change of growth mechanism from flow model into field
oxide structure as the WF6 domains and their decomposed forms,
assisted dissolution model as similar barrier layers are formed. The
have been identified by ToF-SIMS analysis. Its distribution in the
clear shape of developed nanotubes is however against that
depth (Fig. 5e) suggest faster decomposition in near surface region
concept and suggests another origin of its formation. Variations in
of the nanotubular layer. Whereas WF6 has been identified in the
thickness of this layer 40e130 nm (Fig. 6d) suggests chemical na-
film structure, it is not completely understood if the compound is
ture of formation since electric filed located in barrier layer pro-
entirely transformed to oxide structure or partially evolved as a gas
duces homogeneous layers of regular structure and thickness. The
during anodizing process. In any case, the primary consequence of
possible reason of the layer formation would be post reaction of
alloying of iron with tungsten is modification of the cell boundary
WF6 or similar compound remaining at the bottom of nano-
region and full transition to nanotubes as observed on SEM and
structure, with moisture after completed anodizing process. WF6 is
STEM images.
typically unstable and easily hydrolyze with water though three
The cell size, which is corresponding to the developed nanotube
elementary steps (1e3):
diameter, is effectively modified by alloying element as demon-
strated in Fig. 8. The reduction of cell size may be understood as an
WF6 þ H2O / WOF4 þ 2HF (1)
effect of thinner barrier layer formed on Fe-W alloy during the
initial growth as observed in stage I of the i-t curves in Fig. 1 and
WOF4 þ H2O / WO2F2 þ 2HF (2)
therefore development of smaller pore/tube cells by means of
horizontal to vertical oxide reorganization due to material flow
WO2F2 þ H2O / WO3 þ 2HF (3)
effect. Since the same voltage is applied, over a thinner barrier layer
on Fe-W, to some extent a stronger electric field may be applied
The overall reaction is WF6 þ 3H2O ¼ WO3 þ 6HF where
leading to morphological alternations.
 L. Fadillah et al. / Electrochimica Acta 309 (2019) 274e282
1 mol dm
3 WF6 produces 6 mol dm
3 hydrofluoric acid. Such
harsh acidic conditions may drive chemical reaction with alloy at
anodic film base, and therefore thickening the layer at the alloy/
nanotubular-film interface as demonstrated in Fig. 6c. The
absence of characteristic fluorine peak in the GDOES depth profile,
in contrast to that for pure Fe case (Fig. 4), at alloy/nanotubular-film
interface and consequently uniform distribution of fluorides sug-
gest that originally formed fluoride rich layer is consumed in post
reaction process. The fluoride rich layer, FeFx formed on pure iron is
practically free of oxygen species as demonstrated elsewhere [8].
The composition of the layer obtained by STEM/EDS analysis on Fe-
W alloy shows 17.2, 3.4, 35.4 and 44.0 at.% for Fe, W, F and O,
respectively, indicating rather complex structure of the layer con-
taining compounds such as metal oxides and metal oxyfluorides as
a result of post reaction.
4. Conclusions
The present study shows that alloying of iron with 9 at.% tung-
sten has meaningful implications on the growth and resulting
structure of anodic film. Based on the results obtained the following
conclusions can be drawn:
(i) Alloying of iron with 9 at.% of tungsten results in transition of
nanoporous oxide to nanotubular nanostructure. The
possible reason of this transition is faster kinetics of WF6
reaction with water over the presence of low solubility FeFx
species, easily developing the space in between nanotubes.
(ii) The pore size of the anodic films on the Fe-W alloy is smaller
than that on pure Fe and the quasi-steady state anodizing
condition is more rapidly established on the Fe-W alloy. Such
influences of alloying with tungsten are most likely due to
formation of thinner barrier layer formed on the Fe-W alloy.
(iii) The anodic nanopores developed on pure Fe and anodic
nanotubes developed on Fe-W alloy composed of FexOy and
WO3 domains grow in line with the field assisted flow model.
The thick layer formed at the alloy/oxide interface is possibly
a result of chemical reaction due to exposition of structure to
moisture.
Acknowledgements
The present study was supported in part by JFE 21 stCentury
Foundation. A part of this study was conducted at Hokkaido Uni-
versity Microstructure Characterization Platform supported by
Ministry of Education, Culture, Sports, Science and Technology,
Japan. D.K. would like to acknowledge for financial support from
National Science Centre grant number 2013/11/B/ST4/02151. ToF-
SIMS and XPS studies were carried out with the equipment pur-
chased by Biological and Chemical Research Centre, University of
Warsaw, established within the project co-financed by European
Union from the European Regional Development Fund under the
Operational Programme Innovative Economy, 2007e2013.
Appendix A. Supplementary data
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.electacta.2019.03.206.
References
[1] D. Kowalski, D. Kim, P. Schmuki, TiO2 nanotubes, nanochannels and meso-
sponge: self-organized formation and applications, Nano Today 8 (2013)
235e264.
[2] H.E. Prakasam, O.K. Varghese, M. Paulose, G.K. Mor, C.A. Grimes, Synthesis and
photoelectrochemical properties of nanoporous iron (III) oxide by
281
potentiostatic anodization, Nanotechnology 17 (2006) 4285e4291.
[3] S.K. Mohapatra, S.E. John, S. Banerjee, M. Misra, Water photooxidation by
smooth and ultrathin alpha-Fe2O3 nanotube Arrays, Chem. Mater. 21 (2009)
3048e3055.
[4] H. Habazaki, Y. Konno, Y. Aoki, P. Skeldon, G.E. Thompson, Galvanostatic
growth of nanoporous anodic films on iron in ammonium fluoride-ethylene
glycol electrolytes with different water contents, J. Phys. Chem. C 114
(2010) 18853e18859.
[5] Y. Konno, E. Tsuji, P. Skeldon, G.E. Thompson, H. Habazaki, Factors influencing
the growth behaviour of nanoporous anodic films on iron under galvanostatic
anodizing, J. Solid State Electrochem. 16 (2012) 3887e3896.
[6] K. Kure, Y. Konno, E. Tsuji, P. Skeldon, G.E. Thompson, H. Habazaki, Formation
of self-organized nanoporous anodic films on Type 304 stainless steel, Elec-
trochem. Commun. 21 (2012) 1e4.
[7] H. Tsuchiya, T. Suzumura, Y. Terada, S. Fujimoto, Formation of self-organized
pores on type 316 stainless steel in organic solvents, Electrochim. Acta 82
(2012) 333e338.
[8] K. Shahzad, D. Kowalski, C.Y. Zhu, Y. Aoki, H. Habazaki, Ex situ evidence for the
role of a fluoride-rich layer switching the growth of nanopores to nanotubes:
a missing piece of the anodizing puzzle, Chemelectrochem 5 (2018) 610e618.
[9] S.P. Albu, A. Ghicov, P. Schmuki, High aspect ratio, self-ordered iron oxide
nanopores formed by anoclization of Fe in ethylene glycol/NH4F electrolytes,
Phys. Status Solidi RRL 3 (2009) 64e66.
[10] N. Sato, Toward a more fundamental understanding of corrosion processes,
Corrosion 45 (1989) 354e368.
[11] A. Kasuga, A. Koyama, K. Nakayama, D. Kowalski, C. Zhu, Y. Aoki, H. Habazaki,
Fabrication of superoleophobic surface on stainless steel by hierarchical sur-
face roughening and organic coating, ISIJ Int. (2018) advpub.
[12] F. Beck, R. Kaus, M. Oberst, Transpassive dissolution of iron to ferrate (Vi) in
concentrated alkali hydroxide solutions, Electrochim. Acta 30 (1985)
173e183.
[13] M. Pourbaix, Atlas of Electrochemical Equilibria in Aqueous Solutions, NACE,
Houston, 1974.
[14] H. Habazaki, K. Fushimi, K. Shimizu, P. Skeldon, G.E. Thompson, Fast migration
of fluoride ions in growing anodic titanium oxide, Electrochem. Commun. 9
(2007) 1222e1227.
[15] C. Cherki, J. Siejka, Study by nuclear microanalysis and O-18 tracer techniques
of oxygen-transport processes and growth laws for porous anodic oxide
layers on aluminum, J. Electrochem. Soc. 120 (1973) 784e791.
[16] S.J. Garcia-Vergara, H. Habazaki, P. Skeldon, G.E. Thompson, Tracer studies
relating to alloying element behaviour in porous anodic alumina formed in
phosphoric acid, Electrochim. Acta 55 (2010) 3175e3184.
[17] S.J. Garcia-Vergara, D. Le Clere, T. Hashimoto, H. Habazaki, R. Skeldon,
G.E. Thompson, Optimized observation of tungsten tracers for investigation of
formation of porous anodic alumina, Electrochim. Acta 54 (2009) 6403e6411.
[18] S.J. Garcia-Vergara, P. Skeldon, G.E. Thompson, H. Habazaki, Behaviour of a fast
migrating cation species in porous anodic alumina, Corros. Sci. 50 (2008)
3179e3184.
[19] S.J. Garcia-Vergara, P. Skeldon, G.E. Thompson, H. Habazaki, A tracer investi-
gation of chromic acid anodizing of aluminium, Surf. Interface Anal. 39 (2007)
860e864.
[20] S.J. Garcia-Vergara, P. Skeldon, G.E. Thompson, T. Hashimoto, H. Habazaki,
Compositional evidence for flow in anodic films on aluminum under high
electric fields, J. Electrochem. Soc. 154 (2007) C540eC545.
[21] S.J. Garcia-Vergara, P. Skeldon, G.E. Thompson, H. Habazaki, Tracer studies of
anodic films formed on aluminium in malonic and oxalic acids, Appl. Surf. Sci.
254 (2007) 1534e1542.
[22] S.J. Garcia-Vergara, P. Skeldon, G.E. Thompson, H. Habazaki, A flow model of
porous anodic film growth on aluminium, Electrochim. Acta 52 (2006)
681e687.
[23] N. Sato, A theory for breakdown of anodic oxide films on metals, Electrochim.
Acta 16 (1971) 1683e1692.
[24] D. Kowalski, J. Mallet, S. Thomas, J. Rysz, B. Bercu, J. Michel, M. Molinari, Self-
organization of TiO2 nanotubes in mono-, di- and tri-ethylene glycol elec-
trolytes, Electrochim. Acta 204 (2016) 287e293.
[25] D. Kowalski, J. Mallet, J. Michel, M. Molinari, Low electric field strength self-
organization of anodic TiO2nanotubes in diethylene glycol electrolyte,
J. Mater. Chem. 3 (2015) 6655e6661.
[26] H. Masuda, M. Yotsuya, M. Asano, K. Nishio, M. Nakao, A. Yokoo, T. Tamamura,
Self-repair of ordered pattern of nanometer dimensions based on self-
compensation properties of anodic porous alumina, Appl. Phys. Lett. 78
(2001) 826e828.
[27] Y. Bai, L.W. Yang, F. Wu, C. Wu, S. Chen, L.Y. Bao, W.L. Hu, High performance
FeFx/C composites as cathode materials for lithium-ion batteries, J. Renew.
Sustain. Energy 5 (2013).
[28] L.P. Li, J.H. Zhu, M.W. Xu, J. Jiang, C.M. Li, In situ engineering toward core
regions: a smart way to make applicable FeF3@Carbon nanoreactor cathodes
for Li-ion batteries, ACS Appl. Mater. Interfaces 9 (2017) 17992e18000.
[29] A.P. Grosvenor, B.A. Kobe, M.C. Biesinger, N.S. McIntyre, Investigation of
multiplet splitting of Fe 2p XPS spectra and bonding in iron compounds, Surf.
Interface Anal. 36 (2004) 1564e1574.
[30] M.C. Biesinger, B.P. Payne, A.P. Grosvenor, L.W.M. Lau, A.R. Gerson, R.S. Smart,
Resolving surface chemical states in XPS analysis of first row transition
metals, oxides and hydroxides: Cr, Mn, Fe, Co and Ni, Appl. Surf. Sci. 257
(2011) 2717e2730.
282 L. Fadillah et al. / Electrochimica Acta 309 (2019) 274e282
[31] R. Thorpe, S. Rangan, R. Whitcomb, A.C. Basaran, T. Saerbeck, I.K. Schuller,
R.A. Bartynski, The solid state conversion reaction of epitaxial FeF2(110) thin
films with lithium studied by angle-resolved X-ray photoelectron spectros-
copy, Phys. Chem. Chem. Phys. 17 (2015) 15218e15225.
[32] H.S. Shim, J.W. Kim, Y.E. Sung, W.B. Kim, Electrochromic properties of tung-
sten oxide nanowires fabricated by electrospinning method, Sol. Energy
Mater. Sol. Cells 93 (2009) 2062e2068.
[33] S.V. Green, A. Kuzmin, J. Purans, C.G. Granqvist, G.A. Niklasson, Structure and
composition of sputter-deposited nickel-tungsten oxide films, Thin Solid
Films 519 (2011) 2062e2066.
[34] G.J. He, J.M. Li, W.Y. Li, B. Li, N. Noor, K.B. Xu, J.Q. Hu, I.P. Parkin, One pot
synthesis of nickel foam supported self-assembly of NiWO4 and CoWO4
nanostructures that act as high performance electrochemical capacitor elec-
trodes, J. Mater. Chem. 3 (2015) 14272e14278.
[35] N.V. Alov, XPS study of MoO3 and WO3 oxide surface modification by low-
energy Arþ ion bombardment, Phys. Status Solidi C 12 (2015) 263e266.
[36] C. Bittencourt, M.P. Felicissimo, A. Felten, L.A.O. Nunes, P. Ivanov, E. Llobet,
J.J. Pireaux, L. Houssiau, Characterization of WO3 : Ag films: ToF-SIMS studies
of ammonia adsorption, Appl. Surf. Sci. 250 (2005) 21e28.
[37] S.J. Garcia-Vergar, I.S. Molchan, F. Zhou, H. Habazaki, D. Kowalski, P. Skeldon,
A.E. Thompson, Incorporation and migration of phosphorus species within
anodic films on an Al-W alloy, Surf. Interface Anal. 43 (2011) 893e902.
[38] D. Kowalski, Y. Aoki, H. Habazaki, High proton conductivity in anodic ZrO2/
WO3 nanofilms, Angew. Chem. Int. Ed. 48 (2009) 7582e7585.
[39] T.P. Hoar, N.F. Mott, A mechanism for the formation of porous anodic oxide
films on aluminium, J. Phys. Chem. Solids 9 (1959) 97e99.
[40] J.E. Houser, K.R. Hebert, The role of viscous flow of oxide in the growth of self-
ordered porous anodic alumina films, Nat. Mater. 8 (2009) 415e420.
[41] P. Skeldon, G.E. Thompson, S.J. Garcia-Vergara, L. Iglesias-Rubianes,
C.E. Blanco-Pinzon, A tracer study of porous anodic alumina, Electrochem.
Solid State Lett. 9 (2006) B47eB51.
[42] H. Jung, J. Hwang, H. Chun, B. Han, Elucidation of hydrolysis reaction mech-
anism of tungsten hexafluoride (WF6) using first-principles calculations,
J. Ind. Eng. Chem. (2018).
[43] D.R. Lide, CRC Handbook of Chemistry and Physics, 84th Edition, CRC Press,
2003.
[44] Y. Kajikawa, T. Tsumura, S. Noda, H. Komiyama, Y. Shimogaki, Nucleation of W
during chemical vapor deposition from WF6 and SiH4, Jpn. J. Appl. Phys. Part
1-Regular Papers Brief Commun. Rev. Papers 43 (2004) 3945e3950.
 Electrochimica Acta 364 (2020) 137316

Contents lists available at ScienceDirect

Electrochimica Acta
journal homepage: www.elsevier.com/locate/electacta

Compositional variations in anodic nanotubes/nanopores formed on Fe

100, 110 and 111 single crystals
Laras Fadillah a, Damian Kowalski b,c,∗, Yoshitaka Aoki b, Hiroki Habazaki b
a
Graduate School of Chemical Sciences and Engineering, Hokkaido University, Kita-Ku Kita 13, Jo Nishi 8, Sapporo, Hokkaido 060-8628, Japan
b
Division of Applied Chemistry, Faculty of Engineering, Hokkaido University, Kita-Ku Kita 13, Jo Nishi 8, Sapporo, Hokkaido 060-8628, Japan
c
Faculty of Chemistry and Biological and Chemical Research Centre, University of Warsaw, Zwirki i Wigury 101, 02-089 Warsaw, Poland

a r t i c l e i n f o a b s t r a c t

Article history: Anodizing of iron (100), (110) and (111) single crystals in mono-ethylene-glycol electrolyte contain-
Received 14 May 2020 ing 1.5 mol dm−3 water and 0.1 mol dm−3 ammonium fluoride leads to formation of anodic iron
Revised 14 October 2020
nanopores/nanotubes where the nanopores are essentially composed of oxide nanotubes separated by
Accepted 15 October 2020
iron fluoride matrix. It was found that electrochemical thermodynamics for nanoporous/nanotubular film
Available online 18 October 2020
formation apparently depends on the index number of facet on which the anodic film is formed. The film
Keywords: formation on (100) facet is associated with extended gas evolution upon anodizing and consequently cor-
Anodizing responds to a shift of current-time curve towards higher current density values comparing with those
Iron oxide formed on higher index number facets. The nanotubes formed on (100) facet have a general chemical
Nanopores formula of Fe2 O3 .FeF2 , whereas those formed on higher index number are composed of Fe3 O4 .FeF2 . The
Single crystal anodic films formed on Fe (110) and Fe (111) are essentially amorphous whereas the one formed on (100)
facet shows high degree of crystallinity. The results are discussed in view of anisotropic properties of iron.
© 2020 Elsevier Ltd. All rights reserved.

Introduction ping [10]. The full transition of nanopores to nanotubes may be

Anodizing is a high voltage electrochemical process commonly
used to form protective oxide films on metals and alloys. The
barrier-type, nanoporous and nanotubular anodic films can be
formed depending on anodizing conditions, such as composition of
the electrolyte, temperature of the electrolyte, current/voltage pro-
tocol and presence of alloying elements [1-3]. Anodizing of iron
and its alloys have been performed in mono-ethylene-glycol elec-
trolytes containing desired amount of fluorides and water [4-10].
Shahzad et al. [10] established that at low water concentrations,
anodic film grows through classical field-assisted dissolution model
whereas at high water contents the growth mechanism is switched
to field-assisted flow model. At high electric field strength con-
ditions typically shape of nanopores can be obtained being simi-
lar to that observed for anodizing of aluminum (Al2 O3 -like). Those
nanopores are essentially composed of iron oxide nanotubes (TiO2 -
like) separated by iron fluoride located at the cell boundary re-
gion [10]. Extended anodizing process typically leads to transition
of nanopores to nanotubes due to dissolution of metal fluoride as
evidenced ex-situ by high resolution TEM/EDS compositional map-
∗
Corresponding author at: Faculty of Chemistry and Biological and Chemical Re-
search Centre, University of Warsaw, Zwirki i Wigury 101, 02-089 Warsaw, Poland.
E-mail address: damian.kowalski@chem.uw.edu.pl (D. Kowalski).
facilitated by addition of alloying elements such as tungsten which
promotes formation of low stability form of WFx over FeF2 at the
cell boundary region [11].
Whereas the effects of applied voltage, current, temperature,
and electrolyte composition are well known to influence the
growth of anodic film, the effects of crystallographic structure of
iron on anodizing process has not been studied so far. The crys-
tallographic structure of oxides grown on iron has been a subject
of immense debate for the last few decades by corrosion com-
munity with the aim to understand the passivity of iron. For the
passive film the presence of Fe2 O3 or Fe3 O4 has been proposed
[12-14]. Nagayama et al. found that the passive oxide can be com-
posed of Fe3 O4 inner-layer and Fe2 O3 outer-layer [15-17]. The pas-
sive film may be amorphous having FeOOH-like structure [18-24].
In contrast to the above findings, the other studies showed that
passive oxide is composed of Fe2 O3 or Fe3 O4 [25,26]. Toney et al.
[27] and Davenport et al. [28] evaluated the structure of passive
oxide formed on iron single crystals of (001) and (110) by means of
in-situ surface X-ray diffraction technique, and found that the pas-
sive oxide has the LAMM phase, analogous to Fe2 O3 or Fe3 O4 . The
oxide formed on (001) plane was less defective, has larger crys-
tallites and grains than that formed on (110) single crystal surface
[27,28]. Takabatake et al. found the passive film on iron has bi-
layered structure, with Fe3 O4 as an inner-layer and Fe2 O3 as an

https://doi.org/10.1016/j.electacta.2020.137316
0013-4686/© 2020 Elsevier Ltd. All rights reserved.
L. Fadillah, D. Kowalski, Y. Aoki et al.
outer-layer, varying in the proportion of -inner to -outer layers de-
pending on the surface energy of the substrate related to its crys-
tallographic orientation [29]. The mechanical properties of passive
oxides formed on iron (100) and (110) single crystals were con-
ducted by Seo and Chiba, the hardness of the passive iron (100)
orientation was lower by 10% than the passive iron of (110) orien-
tation [30]. Taking into account the nature of oxide formed on iron,
immensely investigated by corrosion community, one may expect
that the structure of oxide formed by anodizing may also depend
on the crystallographic structure of metal on which it is formed.
There are only few works referring to the nature of anodic ox-
ide formed on different grains of polycrystalline metal substrate
such as titanium [31-37] typically studied by means of electron
backscatter diffraction (EBSD) technique and few works reporting
the structure of anodic films on single crystals for the aluminum
case [38-40]. The effect of crystallographic structure of iron on an-
odizing has not been studied so far, either on grains of polycrys-
talline iron or iron single crystals. Whereas the EBSD technique is
useful regarding the nature of individual grains on polycrystalline
specimen, the miss-indexing of the plains may be an obstacle lead-
ing to inaccurate conclusions, therefore we choose to evaluate on
how the crystallographic structure of iron affects the growth of an-
odic film on individual single crystals for iron body centered cubic
(BCC) structure.
Experimental
Iron single crystal (100), (110) and (111) disk electrodes of
10 mm in diameter, grown by Czochralski process, were purchased
from Goodfellows (purity 99.98%). Polycrystalline iron specimen
purchased from Nilaco, Japan (purity 99.99%) was used as a refer-
ence. Before anodizing the specimens were mechanically polished
with SiC grinding paper up to 1500 grade, then polished with set
of alumina suspensions having particle size of 3 μm, 1 μm, and
0.05 μm. Then, the specimens were electropolished in a mixture
of perchloric acid and ethanol (40:60 v/v) followed by cleaning in
acetone and ethanol in an ultrasonic bath for 15 min. Prior to an-
odizing the iron specimens were observed by electron backscatter
diffraction (EBSD) mapping. Iron single crystals were anodized at
60–100 V for 900 s in mono-ethylene-glycol electrolyte contain-
ing 0.1 mol dm−3 ammonium fluoride and 1.5 mol dm−3 deion-
ized water in o-ring type electrochemical cell (V= 4 cm3 ) with a
two-electrode system using platinum counter electrode. The dis-
tance between working and counter electrodes was kept constant
at 20 mm. The anodizing was carried out at 20 ± 0.1 °C with
the temperature of substrate controlled by Peltier element. After
anodizing, the specimens were rinsed in ethylene glycol, ethanol
(99,8%), and then dried in nitrogen stream. Anodizing at selected
voltage was repeated ten times on each single crystal specimen to
ensure the effect of crystallographic plane orientation.
The surfaces and cross-sections of specimens were observed us-
ing a JEOL JSM-6500F field emission scanning electron microscope
equipped with a TSL-EBSD system. EBSD maps were analyzed with
respect to grain size distribution and crystallographic texture us-
ing the OIM5.22 software. Electron-transparent cross-sections were
prepared by a Hitachi FB-2100 focused ion beam (FIB) system
employing a Ga+ ion beam. Cross-sections of the anodized spec-
imens were observed by a JEOL, JEM-ARM200F scanning trans-
mission electron microscope (STEM) with energy dispersive X-ray
spectrometry (EDS) facilities. The Raman spectra were collected at
10 0–120 0 cm−1 with Horiba XploRA Raman spectrometer (Kyoto,
Japan) equipped with an Olympus microscope (Tokyo, Japan) us-
ing 100x magnification. Calibration of the spectra were obtained
using a silicon wafer with a main peak at 520.7 cm−1 as a refer-
ence. The spectra were collected with a 532 nm line, solid-state
lasers (12 mW at the sample); 2400 lines/mm grating in back-
2
Electrochimica Acta 364 (2020) 137316
scattering, with exposure time 10 s with 3 integration time. The
crystallographic structure of the specimens was investigated by X-
ray diffractometer (using Rigaku, RINT-20 0 0) using Cu-Kα radia-
tion (λ = 0.15418 nm) with range of 10–80° 2θ at a scan speed
of 0.5°/min and collected by a tube voltage of 40 kV and 20 mA
tube current. For the analysis of thin film, grazing incidence XRD
mode was used to minimize the contribution of the substrate ma-
terial. The high-resolution transmission electron microscopy (HR-
TEM) was performed on Jeol JEM-2010 electron microscope oper-
ated at an accelerating voltages 200 kV. Auger spectroscopy analy-
sis was performed on iron (100) and (110) anodized samples. The
Auger spectroscopy point analysis used 10 kV primary electron en-
ergy, 10.2 nA beam current, with a cylindrical mirror analyzer of
incidence with respect to the surface normal of 30° slope and per-
formed at a vacuum pressure of ~10−6 Pa.
Results and discussion
The effect of crystallographic orientation of iron on anodic ox-
ide growth was investigated by means of anodizing of single crys-
tals grown by the Czochralski method. The atomic structure of each
single crystal having plane orientation of (100), (110) and (111)
is schematically shown in Fig. 1 with the emphasis on the first
three atomic layers. The surface energy of each crystal significantly
differs as a result of variations in atomic density of 1.2 × 1019 ,
1.75 × 1019 , and 0.7 × 1019 atom.m−2 for (100), (110) and (111)
surface, respectively. The surface state of iron specimens after pol-
ishing with 0.05 μm alumina suspension and subsequent by elec-
tropolishing was investigated by electron backscatter diffraction
(EBSD) analysis. The inverse pole figures (IPF) and corresponding
Kikuchi patterns obtained after background subtraction form each
analysis of iron single crystals (100), (110) and (111) are shown
in Fig. 2. The IPF image for polycrystalline iron (Fig. 2a) depict-
ing grain size of 2–10 μm is shown as a reference. The IPF im-
ages for single crystals after surface pretreatment are character-
ized with one mono-crystalline domain indicating proper crystallo-
graphic plane exposed at solid electrolyte interface in the anodiz-
ing process.
Anodizing j-t curves
Anodizing of iron single crystals and polycrystalline reference
specimen was performed at a constant voltage of 60 V in mono-
ethylene-glycol electrolyte containing 1.5 mol dm−3 water and
0.1 mol dm−3 NH4 F at 20 ± 0.1 °C for 900 s. Fig. 3a shows the
current-time (j-t) transients which are characterized by current
spike observed in the first tens of seconds followed by quasi-steady
state where the current density is nearly constant. The first stage
of anodizing typically corresponds to initial barrier-type oxide for-
mation at the metal-electrolyte interface and possible pore initi-
ation, while the second stage is related to a steady state growth
of nanoporous structure [11]. Significant differences in the cur-
rent density values were found on j-t curves depending on index-
ing number of single crystal specimen. The current density in the
steady state growth on iron (100) is larger than that on (110) and
(111) of by a factor of 1.4. This is more apparent when one com-
pares the total charge (Q) passed in the anodizing process at 60 V
for QFe(100) = 26 C cm−2 with QFe(111) = QFe(110) =19 C cm−2 . The dif-
ference in anodizing on low index facet is much more pronounced
at higher voltages. Fig. 3b,c shows the current density as a function
of voltage applied in anodizing of single crystal specimens. The
slope of j=f(U) for Fe (100) is 4.3 A.V−1 .m−2 significantly higher
than that of 0.8 A.V−1 .m−2 for Fe (110) and Fe (111) giving the cur-
rent density for low index facet larger by a factor of 2.5 at 90 V
(Fig. 3c). By looking at the j-t transients in Fig. 2a one may as-
sume that the ionic current and electronic current may be involved
L. Fadillah, D. Kowalski, Y. Aoki et al. Electrochimica Acta 364 (2020) 137316

Fig. 1. Schematic illustration of atomic structure for a) Fe (100), b) Fe (110) and c) Fe (111) orientation, grey color represent first atomic layer, purple color represent second
atomic layer and green color represent third atomic layer.(For interpretation of the references to color in this figure legend, the reader is referred to the web version of this
article.)

Fig. 2. Electron backscatter diffraction (EBSD) maps for (a) polycrystalline iron and iron single crystals having (b) (100), (c) (110) and (d) (111) orientation.

in the anodizing process. The ionic current leads to atomic migra- Fe → Fex+ + xe− (4)
tion of ions during the oxidation process leading to film develop-
ment in the form of Fe3 O4 , Fe2 O3 , FeFx on both metal/film and film and consequently, at film/electrolyte interface:
electrolyte interfaces according to reactions (1-6) and/or injection
of ionic species to the electrolyte at film/electrolyte interface. The 3Fex+ + 4H2 O → Fe3 O4 + 8H+ + (8-3x)e− (5)
possible reactions at metal/film interface may be expressed as fol-
lows: 2Fex+ + 3H2 O → Fe2 O3 + 6H+ + (6-2x)e− (6)

3Fe + 4O2− → Fe3 O4 + 8e− (1) 2H2 O → O2 (↑) + 4H+ + 4e− (7)

2Fe + 3O2− → Fe2 O3 + 6e− (2) The presence of electronic current strongly depends on elec-
tronic structure of the developed oxide of which the conduc-
Fe + xF− → FeFx + xe− (3) tive/resistive nature controls injection of electrons from electrolyte

3
L. Fadillah, D. Kowalski, Y. Aoki et al.
Fig. 3. a) Current-time (j-t) curves for anodizing of iron (100), (110) and (111) single crystals at 60 V for 900 s in mono-ethylene-glycol electrolyte containing 1.5 mol dm−3
of water and 0.1 mol dm−3 of ammonium fluoride at 20 ± 0.1 °C; the anodizing curve for polycrystalline iron is shown as a reference. The effect of anodizing voltage on
current density for anodizing time of b) 450 s and c) 900 s.
across the oxide to the metal back contact. If true electronic cur-
rent is present, the decomposition of water takes place resulting in
oxygen generation according to reaction (7). The kinetics of oxygen
evolution strongly depends on electronic nature of the anodic ox-
ide; for instance for anodizing of aluminum no oxygen evolution
is typically observed due to resistive nature of Al2 O3 ; for titanium
case, oxide formation is typically accompanied with O2 evolution
as a result of semi-conductive nature of developed TiO2 . The vi-
sual observation of the iron anodizing process indicated massive
gas evolution during oxide formation on (100) facet, whereas very
little gas evolution was observed on high index facets in Fig. 4.
Fast kinetics of gas evolution suggest dominant role of reaction (7)
in the shift of j-t curve towards higher current density values on
iron (100).
Morphology of anodic films
The surface morphology and corresponding cross-section im-
ages for anodic oxides formed on iron (100), (110) and (111) at
60 V depicting the presence of classic nanoporous structure anal-
ogous to that formed on polycrystalline iron surface [10,11] are
demonstrated in Fig. 5. The average inner-diameter of nanopore
was found to be 99 ± 1 nm, 102 ± 1 nm, and 103 ± 1 nm, for
anodic oxide formed on iron (100), (110) and (111), respectively.
The thickness of anodic film formed on Fe (100) is 2.3 ± 0.1 μm,
followed by (111) 2.2 ± 0.1 μm and (110) orientation around 2.2 ±
0.1 μm. Increase in the anodizing voltage results in more apparent
4
Electrochimica Acta 364 (2020) 137316
changes in surface morphology as well as the presence of defects
on the surface depending on the indexing number of the facet (ESI,
Fig. S1 and S2). For further more detailed studies on the impact of
exposed facet we choose the conditions at which oxygen evolution
is considerably low and do not cause damages on the anodic film
surface. The scanning transmission electron micrograph (STEM) of
a focused ion beam (FIB) lamellas for Fe (100) and Fe (110) an-
odized at 60 V are shown in Fig. 6 a,b. The results of high angle
annular dark field (HAADF) have shown that the thickness of the
barrier layer is 97 ± 1 nm and 104 nm ± 1 nm formed on (100)
and (110) facets, respectively. The thickness of pore wall for anodic
film formed on (110) is approximately ten nanometers thicker than
that formed on (100) facet. A critical point of observation between
anodic film formed on high and low index facets is the presence of
roughness in the barrier-layer with dark spots on formed on (100)
surface possibly associated with the higher degree of crystallinity.
Analysis of composition
Fig. 7 displays the HAADF images and corresponding high reso-
lution energy dispersive X-ray spectrometry (EDS) elemental maps
for anodic films formed at 60 V on (100) and (110) facets. The EDS
maps reveal the fluorine, oxygen, and iron at the bottom regions of
anodic oxides. A more detailed analysis of EDS compositional maps
for oxide formed on polycrystalline iron is demonstrated elsewhere
[10]. The structure of anodic film formed on iron, observed by
SEM and TEM, is typically considered to be nanoporous (Al2 O3 -
L. Fadillah, D. Kowalski, Y. Aoki et al. Electrochimica Acta 364 (2020) 137316

Fig. 4. Gas evolution during anodizing of (a) Fe (100), (b) Fe (110), and (c) Fe (111) single crystals at 60 V in mono-ethylene-glycol electrolyte containing 1.5 mol dm−3 water
and 0.1 mol dm−3 of ammonium fluoride at 20 ± 0.1 °C. The dimension of inner diameter of the electrochemical cell corresponds to 20 mm.

Fig. 5. Surface morphology and cross-section images for porous anodic films formed on (a,d) Fe (100), (b,e) Fe (110) and (c,f) Fe (111) single crystals at 60 V in mono-
ethylene-glycol electrolyte containing 1.5 mol dm−3 of water and 0.1 mol dm−3 of ammonium fluoride at 20 ± 0.1 °C.

like nanopore). If one compares, however, TEM image with oxy-
gen elemental map in Fig. 7 it is easy to understand that the true
structure of oxide is nanotubular (TiO2 -like nanotube). The fluorine
enrichment at the cell boundaries and at the metal/anodic-film is
apparent from fluorine elemental map. Both regions are practically
free from oxygen species implying formation of iron fluoride. The
detailed compositional EDS analysis of the inner-part of barrier
layer, denoted “2” in Fig. 7 c,d, reveal composition of 32 at.% of Fe
and 65 at.% of F for film formed on (100) facet, 34 at.% of Fe and 63
at.% of F for film formed on (110) facet. The composition of the flu-
oride rich layer located at metal/anodic-film interface is in agree-
ment with atomic ratio of 1:2 for FeF2 compound. According to the
field assisted flow model, [10] FeF2 layer is switched from horizon-
tal position to vertical position located at the cell boundary, ap-
parently well observed on superimposed iron-fluorine-oxygen EDS
images in Fig. 7. By choosing appropriate anodizing conditions,
[10] the FeF2 may be dissolved from cell boundaries and therefore
the nanopores can be further transformed into nanotubes follow-
ing the shape of nanotube visible on oxygen elemental map. The
outer part of barrier-layer (denoted “1” in Fig. 7c,d) refers to the
inner-shell of oxide nanotube, is composed of Fe, O, F with atomic
ratio of 38, 38, 24 at.%, and 40, 43, 17 at.% for anodic film formed
on (100) and (110) surface, respectively. If we assume that: i) all
fluorine present in the outer part of barrier-layer is bounded to
iron and forms iron-fluoride compound, ii) iron bounded to fluo-
ride has an oxidation number 2+, the calculated anodic structure

5
L. Fadillah, D. Kowalski, Y. Aoki et al. Electrochimica Acta 364 (2020) 137316

Fig. 6. TEM images for focused ion beam (FIB) lamellas performed for iron (100) and (110) single crystals anodized at 60 V in mono-ethylene-glycol electrolyte containing
1.5 mol dm−3 of water and 0.1 mol dm−3 of ammonium fluoride at 20 ± 0.1 °C.

Table 1
Calculation of chemical composition for the outer part of barrier layer of porous/tubular anodic film indicated as
region 1 in Fig. 7.

EDS results for anodic nanopore/nanotube Fe O - F

formed on Fe (100): 38 38 - 24

Calculated atomic ratio: Fe(oxide) O Fe(FeF2 ) F

26 38 12a 24
2 3 1 2
Chemical formula: Fe2 O3 .FeF2
EDS results for anodic nanopore/nanotube Fe O - F
formed on Fe (110): 40 43 - 17
Calculated atomic ratio: Fe(oxide) O Fe(FeF2 ) F
31 43 9∗ 17
3 4 1 2
Chemical formula: Fe3 O4 .FeF2
a
Calculated for FeF2 formation.

formed on iron (100) facet can be expressed by the following for-
mula: Fe2 O3 .FeF2 . Consequently, the calculated structure of the an-
odic outer part of barrier-layer formed on (110) facet would be
Fe3 O4 .FeF2 (Table 1). The initial formation of FeF3 has been con-
sidered as a result of ions migration under high electric field con-
ditions, however, its presence in the anodic film structure is ne-
glected due to its faster dissolution in electrolyte over FeF2 due to
difference in solubility constants of kFeF =5.2 × 10−1 mol.dm−3 and
3
kFeF =8.3 × 10−3 mol.dm−3 [10]. The x-ray diffraction (XRD) spectra
2
for anodic oxides formed on (100), (110) and (111) single crystals
are shown in Fig. 8. The most intense 2θ diffraction peak for iron is
that at 44.70° corresponding to 110 reflection (Fig. 8a), and there-
fore (110) single crystal gives most intense XRD pattern. The other
single crystal specimens show a small 2θ peak at 65.16° which is
assigned to 200 reflection, whereas the (111) single crystal essen-
tially gives no diffraction line. The anodic films formed on all three
single crystal surfaces show 2θ peaks at 26.82° corresponding to
110 reflection of FeF2 (ICSD no. 81-2271) (Fig. 8b). The 2θ peaks
at 35.40° and 56.94° for anodic films formed on (110) and (111)
single crystals may be indexed to 113 and 115 reflections of Fe3 O4
(ICSD no. 75-1372). The presence of FeOOH due to hydrolysis re-
action with humid air for anodic films formed on (110) and (111)
iron (ICSD no. 75-1594) with reference 2θ peak at 11.90° (110),
and other peaks that overlapping with those observed for Fe3 O4
35.30° corresponding to 211 reflection and 56.61° corresponding
to 521 reflection cannot be excluded. Surprisingly, additional 2θ
peaks at 24.10°, 33.10° and 35.60° appear on the diffraction pattern
obtained for anodic film formed on (100) single crystal, and could
be assigned to 012, 104 and 110 reflections of Fe2 O3 (ICSD no. 80-
2377). The Fe2 O3 diffraction peaks are essentially absent for anodic
films formed on high index number single crystals which can be
more clearly seen on magnified patterns in Fig. 8c. The formation
of Fe2 O3 on (100) surface and Fe3 O4 on (110) and (111) surfaces,
detected by XRD, is in agreement with the stoichiometry of anodic
films calculated from high resolution compositional EDS elemental
maps in Fig. 7.

6
L. Fadillah, D. Kowalski, Y. Aoki et al. Electrochimica Acta 364 (2020) 137316

Fig. 7. High resolution energy dispersive X-ray spectrometry (EDS) maps obtained for bottom regions of focused ion beam (FIB) cross-sections of anodic porous film formed
on (a, c) (100) and (b, d) (110) single crystals at 60 V in mono-ethylene-glycol electrolyte containing 1.5 mol dm−3 of water and 0.1 mol dm−3 of ammonium fluoride for
900s at 20 ± 0.1 °C. The star symbol “∗ ” indicates the calculated chemical composition at selected regions of nanopore/nanotube.

Fig. 8. X-ray diffraction spectra for anodic porous films formed on (100), (110) and (111) single crystals at 60 V for 900 s in mono-ethylene-glycol electrolyte containing 1.5
mol dm−3 of water and 0.1 mol dm−3 of ammonium fluoride at 20 ± 0.1 °C.

Fig. 9 shows high resolution TEM images obtained for FIB cross-
sections of anodic films formed on (100) and (110) iron single crys-
tals. Comparing both TEM micrographs one may notice that, funda-
mentally, higher degree of crystallinity was observed for the film
formed on (100) single crystal. The composition of the inner part
of the barrier-layer, i.e. metal/anodic-film interface was found to
be essentially the same for both cross-sections. The lattice fringes
having 3.3 Å and 2.3 Å corresponding to 110 and 111 planes of FeF2
were found in the inner part of barrier layer for both anodic films
(ICSD no. 81-2271). The outer part of barrier-layer (Fig. 9a) for film
formed on iron (100) single crystal is characterized by presence of
large crystalline domains with the lattice fringes of 3.7 Å, 2.5 Å and
2.7 Å which may be indexed to 012, 110 and 104 planes of Fe2 O3 ,
respectively (ICSD no. 80-2377). The anodic film formed on (110)
single crystal is essentially of amorphous nature having some little
nanocrystals with lattice fringes at 2.5 Å, 1.6 Å and 4.8 Å which
may be assigned to 113, 115 and 111 planes of Fe3 O4 (ICSD no. 75-
1372).
Fig. 10 shows Raman spectra recorded for each anodized sin-
gle crystal in order to get more insight into the structural varia-
tions in anodic porous films. The spectra for anodic films formed
on (110) and (111) facets show typical magnetite (Fe3 O4 ) features
which are attributed to the A1g mode at around 700 cm−1 in
Fig. 10 [41-43]. The weak spectral features at 300 and 550 cm−1
are assigned to the T1g vibrational mode [41,44]. The Raman spec-
trum for anodized (100) single crystal exhibit the spectral signa-

7
L. Fadillah, D. Kowalski, Y. Aoki et al.
Fig. 9. High resolution TEM images of focused ion beam (FIB) lamellas performed for iron (100) and (110) single crystals anodized at 60 V for 900 s in mono-ethylene-glycol
electrolyte containing 1.5 mol dm−3 of water and 0.1 mol dm−3 of ammonium fluoride at 20 ± 0.1 °C.
Fig. 10. Raman spectra of anodic porous films formed on iron (100), (110) and (111)
single crystals anodized at 60 V for 900 s in mono-ethylene-glycol electrolyte con-
taining 1.5 mol dm−3 of water and 0.1 mol dm−3 of ammonium fluoride at 20 ±
0.1 °C.
tures characteristic for hematite (Fe2 O3 ) in addition to the peaks
associated with the presence of magnetite (Fe3 O4 ). More specifi-
cally, the peaks at 229 cm−1 and 500 cm−1 are assigned to the A1g
modes of hematite (Fe2 O3 ) [41,42] with the remaining five peaks
at 249, 295, 302, 414, and 615 cm−1 assigned to the Eg modes
of hematite (Fe2 O3 ) [41]. The strong peak seen at approximately
660 cm−1 has been widely observed in Raman spectra for mag-
netite (Fe3 O4 ) or maghemite (γ -Fe2 O3 ) and also attributed with
the presence of hematite (Fe2 O3 ) [41-44]. Except obvious differ-
ences in the chemical structure for anodic porous films formed
on high and low index facets, the spectrum obtained for anodized
(100) single crystal is characterized by the presence of more sharp
and more intense Raman peaks which is directly related to high
degree of crystallinity for anodic film formed on low index facet.
Auger spectroscopy has been applied to identify the compo-
sition changes in anodic porous films formed on iron (100) and
(110) single crystals (Fig. 11). The Fe LMM-peaks in the Auger spec-
tra for metallic Fe, FeO, Fe2 O3 and Fe3 O4 typically show similar
8
Electrochimica Acta 364 (2020) 137316
Fig. 11. Auger spectra for anodic porous film formed on iron (100) and (110) single
crystals anodized at 60 V for 900 s in mono-ethylene-glycol electrolyte containing
1.5 mol dm−3 of water and 0.1 mol dm−3 of ammonium fluoride at 20 ± 0.1 °C.
in shape spectra having the variations in the intensity of individ-
ual peaks [45-47]. The Fe LMM peaks at 595 and 703 eV can be
used to distinguish between iron oxides having oxidation state of
Fe2+ and Fe3+ [48]. Except typical features characteristic for Fex Oy
compound, the Auger spectrum in Fig. 11 shows additional peak
at ~650 eV corresponding to FeF2 . The basic curve fitting of the
spectrum using standard shift peak database of Fe2 O3 , FeO, and
combination of both corresponding to standard line of Fe3 O4 could
not be performed with satisfactory fitting due to complexity of the
spectrum because presence of intense peak at ~650 eV. However,
close inspection of the spectrum obtained for anodic films formed
on Fe (100) and that formed on Fe (110) reveal higher intensity
peaks at 595 and 703 eV for films formed on low index number
facet whereas the peak at ~650 eV essentially remains unchanged.
This suggest the contribution of Fe2 O3 for film formed on low in-
dex number facet whereas the shape of the spectrum obtained for
L. Fadillah, D. Kowalski, Y. Aoki et al.
higher index number specimen is analogous to Fe3 O4 . Several stud-
ies have reported analogous changes in the shape of spectra for
passive oxide films formed on Fe (100) and (110) [49-51].
Anisotropy of iron
Anodizing of iron single crystals, namely (100), (110) and (111),
indicated that the electrochemical thermodynamics of the anodic
porous film growth depends on the index number of facet on
which the film is formed. The variations in the growth may be
possible if one considers following physicochemical properties of
iron: i) planar atomic density and corresponding ii) surface en-
ergy, iii) magnetic anisotropy, and/or iv) electronic properties. The
j-t transients obtained during anodizing of single crystals indicated
that there is a significant difference in the formation of anodic
film on (100) facet and the other single crystals. The porous an-
odic film formed on (100) facet is composed of Fe2 O3 .FeF2 whereas
that formed on higher index facet consists of Fe3 O4 .FeF2 . The con-
version of iron into iron oxide essentially results in volume ex-
pansion which is described by Pilling-Bedworth ratio [52]. Un-
der influence of high electric field strength, the volume change
typically generates mechanical stress at the metal/oxide interface
which, together with electrostriction forces, is responsible for the
change of the horizontal oxide growth into vertical one develop-
ing nanoporous/nanotubular shape [53]. The very first compari-
son of composition for both oxides reveals that essentially higher
difference in the volume for Fe2 O3 than that for Fe3 O4 is ex-
pected, which simply means that higher mechanical stress at the
metal/oxide interface could be present for (100) single crystal case.
The presence of high internal stress may favor crystals growth
in anodic film [54-57]. For the barrier-type anodic films inter-
nal stress has been demonstrated to induce formation of defects
within the oxide, localized increase of current and heat genera-
tion within the defect area leading to oxide crystallization [54-56].
For the nanotubular/nanoporous structures certain degree of crys-
tallization is occasionally observed [53] although the mechanism of
crystals nucleation and growth has not been studied in detail. High
degree of crystallinity for the oxide formed on low index num-
ber iron facet, as evidenced by XRD and Raman spectroscopy re-
sults in modification of electronic properties of the oxide, i.e. in-
crease of conductance which leads to electronic current increase
and consequently evolution of oxygen as observed in Fig. 4, ac-
cording to reaction (7). The gas evolution is typically observed for
electrically conductive anodic oxides such that formed on titanium
[58] in contrast to non-conductive oxides such as those formed on
aluminum [53].
The anisotropy of crystal in the sense of variations in physical
properties along different molecular axes such as difference in pla-
nar density shown in Fig. 1, should be discussed in view of its ef-
fect on the anodizing process. The surface energy of a single crys-
tal is directly related to the planar density of the substrate be-
cause of the number of unsatisfied bonds [38,40]. In the case of
iron single crystals, the calculated atomic density is 1.75 × 1019 ,
1.20 × 1019 , and 0.70 × 1019 atom.m−2 for (110), (100) and (111)
surface, respectively, and therefore the surface energy would be in
the following sequence (110) < (100) < (111). Our experimental re-
sults, however, show that anodizing conditions differ only on (100)
surface, having the middle surface energy estimate, therefore we
rule out the effect of surface density in this particular case. The
other physical properties such as elastic modulus, work function
and magnetic anisotropy show dependence on the index number
of the crystal [59-66]. For instance, an elastic modulus of iron (100)
has been reported to be half lower than the higher index num-
ber facets, according to the following sequence: 129, 221 and 276
GN.m−2 for (100), (110) and (111) facet, respectively [59,67]. In the
anodizing process the presence of compressive-stress and tensile-
9
Electrochimica Acta 364 (2020) 137316
stress is critical on the effects on convex geometry and therefore
formation of nanoporous/nanotubular shape of anodic film [68].
Twice lower elastic modulus value means that iron single crys-
tal is more prone for deformation in (100) direction. The motion
at the metal/film interface would therefore be dependent on the
number of index facet in line with the tendency of the material
towards elastic deformation. Consequently, the stress distribution
would be different on (100) facet than that on the higher index
number facets as a result of variations in the material motion. The
magnetic anisotropy of iron is another aspect which may have an
influence on the presence of magnetic field in the electrochem-
ical cell and therefore affect electrochemical process [60,62-66].
The easiest direction of magnetization for iron is (100) orientation,
followed by (110) and (111) orientation [60]. In our experimental
work, presented here, there is no indicator linking the magnetic
anisotropy with anodizing of iron, however, the effect of magnetic
field cannot be ruled out and should be explored more in depth in
further work.
Conclusions
The anodic iron nanopores/nanotubes were formed on iron
(100), (110) and (111) single crystals in mono-ethylene-glycol elec-
trolyte containing 1.5 mol dm−3 of water and 0.1 mol dm−3 ammo-
nium fluoride. It was found that electrochemical thermodynamics
of anodic film growth depends on the index number of facet on
which the film is formed. Based on the results obtained the fol-
lowing conclusions can be drawn:
Nanoporous/nanotubular anodic film formed on (100) iron facet
was formed with electric charge ~30% larger than that on (110) and
(111) single crystals. The electric charge increase was accompanied
with extensive gas evolution on (100) facet.
Chemical composition of anodic film formed on (100) facet was
Fe2 O3 .FeF2 whereas film formed on higher index number was com-
posed of Fe3 O4 .FeF2 , evaluated by high resolution TEM/EDS ele-
mental mapping analysis of FIB lamella.
The anodic films formed on (111) and (110) single crystals were
of amorphous nature whereas those formed on (100) facet were
found to be crystalline. The crystallization upon anodic film growth
is a possible reason for modification of electronic properties fol-
lowed by extensive oxygen evolution on Fe2 O3 nanocrystals ob-
served on (100) facet.
CRediT author statement
Laras Fadillah: Investigation, Formal analysis, Data curation,
Writing- Original draft preparation.: Damian Kowalski: Conceptu-
alization, Supervision, Writing - Review & Editing, Funding acquisi-
tion.: Yoshitaka Aoki: Visualization.: Hiroki Habazaki: Supervision,
Funding acquisition.
Declaration of Competing Interest
None.
Acknowledgments
Financial support from National Science Centre grant number
2016/21/B/ST5/03387 is highly acknowledged. The study was sup-
ported in part by JSPS KAKENHI Grant Number 19H02469 and JFE
21st Century Foundation. A part of this study was conducted at
Hokkaido University Microstructure Characterization Platform sup-
ported by Ministry of Education, Culture, Sports, Science and Tech-
nology (MEXT), Japan.
L. Fadillah, D. Kowalski, Y. Aoki et al.
Supplementary materials
Supplementary material associated with this article can be
found, in the online version, at doi:10.1016/j.electacta.2020.137316.
References
[1] S. Ono, K. Kuramochi, H. Asoh, Effects of electrolyte pH and temperature on
dielectric properties of anodic oxide films formed on niobium, Corros. Sci. 51
(2009) 1513–1518.
[2] P. Skeldon, K. Shimizu, G.E. Thompson, G.C. Wood, Fundamental studies elu-
cidating anodic barrier-type film growth on aluminium, Thin Solid Films 123
(1985) 127–133.
[3] G.E. Thompson, G.C. Wood, 5 - anodic films on aluminium, in: J.C. Scully (Ed.),
Treatise on Materials Science and Technology, Elsevier, 1983, pp. 205–329.
[4] H. Habazaki, Y. Konno, Y. Aoki, P. Skeldon, G.E. Thompson, Galvanostatic
growth of nanoporous anodic films on iron in ammonium fluoride-ethylene
glycol electrolytes with different water contents, J. Phys. Chem. C 114 (2010)
18853–18859.
[5] A. Kasuga, A. Koyama, K. Nakayama, D. Kowalski, C. Zhu, Y. Aoki, H. Habazaki,
Fabrication of Superoleophobic Surface on Stainless Steel by Hierarchi-
cal Surface Roughening and Organic Coating, ISIJ International, 2018 ISI-
JINT-2018-2569.
[6] Y. Konno, E. Tsuji, P. Skeldon, G.E. Thompson, H. Habazaki, Factors influencing
the growth behaviour of nanoporous anodic films on iron under galvanostatic
anodizing, J. Solid State Electrochem. 16 (2012) 3887–3896.
[7] K. Kure, Y. Konno, E. Tsuji, P. Skeldon, G.E. Thompson, H. Habazaki, Formation
of self-organized nanoporous anodic films on Type 304 stainless steel, Elec-
trochem. Commun. 21 (2012) 1–4.
[8] S.K. Mohapatra, S.E. John, S. Banerjee, M. Misra, Water photooxidation by
smooth and ultrathin α -Fe2 O3 nanotube arrays, Chem. Mater. 21 (2009)
3048–3055.
[9] H.E. Prakasam, O.K. Varghese, M. Paulose, G.K. Mor, C.A. Grimes, Synthesis and
photoelectrochemical properties of nanoporous iron (III) oxide by potentio-
static anodization, Nanotechnology 17 (2006) 4285–4291.
[10] K. Shahzad, D. Kowalski, C. Zhu, Y. Aoki, H. Habazaki, Ex situ evidence for the
role of a fluoride-rich layer switching the growth of nanopores to nanotubes:
a missing piece of the anodizing puzzle, ChemElectroChem 5 (2018) 610–618.
[11] L. Fadillah, K. Takase, H. Kobayashi, S. Turczyniak-Surdacka, M. Strawski,
D. Kowalski, C. Zhu, Y. Aoki, H. Habazaki, The role of tungsten species in the
transition of anodic nanopores to nanotubes formed on iron alloyed with tung-
sten, Electrochim. Acta 309 (2019) 274–282.
[12] I. Iitaka, S. Miyake, T. Iimori, Examination of passive iron by electron diffrac-
tion, Nature 139 (1937) 156 -156.
[13] J.E.O. Mayne, M.J. Pryor, The mechanism of inhibition of corrosion of iron by
chromic acid and potassium chromate, J. Chem. Soc. (Resumed) 392 (1949)
1831–1835.
[14] M. Cohen, An electron diffraction study of films formed by sodium nitrite so-
lution on iron, J. Phys. Chem. 56 (1952) 451–453.
[15] M.i. Nagayama, M. Cohen, The anodic oxidation of iron in a neutral solution: I.
The nature and composition of the passive film, J. Electrochem. Soc. 109 (1962)
781–790.
[16] C.L. Foley, J. Kruger, C.J. Bechtoldt, Electron diffraction studies of active, pas-
sive, and transpassive oxide films formed on iron, J. Electrochem. Soc. 114
(1967) 994–1001.
[17] K. Kuroda, B.D. Cahan, G. Nazri, E. Yeager, T.E. Mitchell, Electron diffraction
study of the passive film on iron, J. Electrochem. Soc. 129 (1982) 2163–2169.
[18] W.E. O’Grady, Mössbauer study of the passive oxide film on iron, J. Elec-
trochem. Soc. 127 (1980) 555–563.
[19] J. Eldridge, R.W. Hoffman, A Mössbauer spectroscopy study of the potential
dependence of passivated iron films, J. Electrochem. Soc. 136 (1989) 955–961.
[20] M. Kerkar, J. Robinson, A.J. Forty, In situ structural studies of the passive film
on iron and iron/chromium alloys using X-ray absorption spectroscopy, Faraday
Discuss. Chem. Soc. 89 (1990) 31–40.
[21] R.W. Hoffman, Study of passivity of iron by in-situ methods: Mossbauer and
EXAFS, in: M. Froment (Ed.), Passivity of Metals and Semiconductors, Elsevier,
1983, pp. 147–162.
[22] G.G. Long, J. Kruger, D.R. Black, M. Kuriyama, Structure of passive films on
iron using a new surface-EXAFS technique, J. Electroanal. Chem. Interfac. Elec-
trochem. 150 (1983) 603–610.
[23] J. Gui, T.M. Devine, In situ vibrational spectra of the passive film on iron in
buffered borate solution, Corros. Sci. 32 (1991) 1105–1124.
[24] J.C. Rubim, J. Dünnwald, Enhanced Raman scattering from passive films on
silver-coated iron electrodes, J. Electroanal. Chem. Interfac. Electrochem. 258
(1989) 327–344.
[25] A.J. Davenport, M. Sansone, High resolution in situ XANES investigation of the
nature of the passive film on iron in a pH 8.4 borate buffer, J. Electrochem.
Soc. 142 (1995) 725–730.
[26] M.P. Ryan, R.C. Newman, G.E. Thompson, An STM study of the passive film
formed on iron in borate buffer solution, J. Electrochem. Soc. 142 (1995)
L177–L179.
[27] M.F. Toney, A.J. Davenport, L.J. Oblonsky, M.P. Ryan, C.M. Vitus, Atomic struc-
ture of the passive oxide film formed on iron, Phys. Rev. Lett. 79 (1997)
4282–4285.
10
Electrochimica Acta 364 (2020) 137316
[28] A.J. Davenport, L.J. Oblonsky, M.P. Ryan, M.F. Toney, The structure of the pas-
sive film that forms on iron in aqueous environments, J. Electrochem. Soc. 147
(20 0 0) 2162–2173.
[29] Y. Takabatake, K. Fushimi, T. Nakanishi, Y. Hasegawa, Grain-dependent pas-
sivation of iron in sulfuric acid solution, J. Electrochem. Soc. 161 (2014)
C594–C600.
[30] M. Seo, M. Chiba, Nano-mechano-electrochemistry of passive metal surfaces,
Electrochim. Acta 47 (2001) 319–325.
[31] H. Sopha, K. Tesar, P. Knotek, A. Jäger, L. Hromadko, J.M. Macak, TiO2 nan-
otubes grown on Ti substrates with different microstructure, Mater. Res. Bull.
103 (2018) 197–204.
[32] S. Leonardi, A. Li Bassi, V. Russo, F. Di Fonzo, O. Paschos, T.M. Murray, H. Efs-
tathiadis, J. Kunze, TiO nanotubes: interdependence of substrate grain orienta-
tion and growth characteristics, J. Phys. Chem. C 116 (2012) 384–392.
[33] M.L. Vera, M.C. Avalos, M.R. Rosenberger, R.E. Bolmaro, C.E. Schvezov, A.E. Ares,
Evaluation of the influence of texture and microstructure of titanium sub-
strates on TiO2 anodic coatings at 60V, Mater. Character. 131 (2017) 348–358.
[34] S. Kudelka, A. Michaelis, J.W. Schultze, Effect of texture and formation rate
on ionic and electronic properties of passive layers on Ti single crystals, Elec-
trochim. Acta 41 (1996) 863–870.
[35] M.V. Diamanti, M.P. Pedeferri, C.A. Schuh, Thickness of anodic titanium oxides
as a function of crystallographic orientation of the substrate, Metall. Mater.
Trans. A 39 (2008) 2143–2147.
[36] E. Matykina, R. Arrabal, P. Skeldon, G.E. Thompson, H. Habazaki, Influence of
grain orientation on oxygen generation in anodic titania, Thin Solid Films 516
(2008) 2296–2305.
[37] A. Michaelis, J.W. Schultze, Anisotropy micro-ellipsometry for in-situ determi-
nation of optical and crystallographic properties of anisotropic solids and lay-
ers with TiO2 as an example, Thin Solid Films 274 (1996) 82–94.
[38] G. Beck, R. Bretzler, Regularity of nanopores in anodic alumina formed on ori-
entated aluminium single-crystals, Mater. Chem. Phys. 128 (2011) 383–387.
[39] G. Beck, K. Petrikowski, Influence of the microstructure of the aluminum sub-
strate on the regularity of the nanopore arrangement in an alumina layer
formed by anodic oxidation, Surf. Coat. Technol. 202 (2008) 5084–5091.
[40] L. Sacco, I. Florea, M. Châtelet, C.-S. Cojocaru, Investigation of porous anodic
alumina templates formed by anodization of single-crystal aluminum sub-
strates, Thin Solid Films 660 (2018) 213–220.
[41] A.M. Jubb, H.C. Allen, Vibrational spectroscopic characterization of hematite,
maghemite, and magnetite thin films produced by vapor deposition, ACS Appl.
Mater. Interfaces 2 (2010) 2804–2812.
[42] I. Chamritski, G. Burns, Infrared- and Raman-active phonons of magnetite,
maghemite, and hematite: a computer simulation and spectroscopic study, J.
Phys. Chem. B 109 (2005) 4965–4968.
[43] D. Bersani, P.P. Lottici, A. Montenero, Micro-Raman investigation of iron ox-
ide films and powders produced by sol–gel syntheses, J. Raman Spectrosc. 30
(1999) 355–360.
[44] L.V. Gasparov, D.B. Tanner, D.B. Romero, H. Berger, G. Margaritondo, L. Forró,
Infrared and Raman studies of the Verwey transition in magnetite, Phys. Rev.
B 62 (20 0 0) 7939–7944.
[45] A.G. Sault, Quantitative analysis of Auger lineshapes of oxidized iron, Appl.
Surf. Sci. 74 (1994) 249–262.
[46] J. Kovač, M. Bizjak, B. Praček, A. Zalar, Auger electron spectroscopy depth pro-
filing of Fe-oxide layers on electromagnetic sheets prepared by low tempera-
ture oxidation, Appl. Surf. Sci. 253 (2007) 4132–4136.
[47] C. Ruby, J. Fusy, Oxidation of ultrathin iron layers grown on Cu(111), Appl. Surf.
Sci. 99 (1996) 393–400.
[48] F. Qin, N.P. Magtoto, M. Garza, J.A. Kelber, Oxide film growth on Fe(111)
and scanning tunneling microscopy induced high electric field stress in
Fe2 O3 /Fe(111), Thin Solid Films 444 (2003) 179–188.
[49] V.S. Smentkowski, J.T. Yates, The adsorption of oxygen on Fe(110) in the tem-
perature range of 90–920 K, Surf. Sci. 232 (1990) 113–128.
[50] M. Langell, G.A. Somorjai, The composition and structure of oxide films grown
on the (110) crystal face of iron, J. Vac. Sci. Technol. 21 (1982) 858–866.
[51] S.J. Roosendaal, A.M. Vredenberg, F.H.P.M. Habraken, Oxidation of iron: the re-
lation between oxidation kinetics and oxide electronic structure, Phys. Rev.
Lett. 84 (20 0 0) 3366–3369.
[52] N. Pilling, The oxidation of metals at high temperature, J. Inst. Met. 29 (1923)
529–582.
[53] D. Kowalski, D. Kim, P. Schmuki, TiO2 nanotubes, nanochannels and
mesosponge: self-organized formation and applications, Nano Today 8 (2013)
235–264.
[54] J.S.L. Leach, B.R. Pearson, Crystallization in anodic oxide films, Corros. Sci. 28
(1988) 43–56.
[55] D.A. Vermilyea, Nucleation of crystalline Ta2 O5 during field crystallization, J.
Electrochem. Soc. 104 (1957) 542.
[56] K. Nagahara, M. Sakairi, H. Takahashi, K. Matsumoto, K. Takayama, Y. Oda,
Change in the structure and dielectric properties of niobium anodic oxide films
during potentiostatic anodizing, Electrochemistry 72 (2004) 624–632.
[57] Z. Su, L. Zhang, F. Jiang, M. Hong, Formation of crystalline TiO2 by anodic oxi-
dation of titanium, Progr. Nat. Sci.: Mater. Int. 23 (2013) 294–301.
[58] H. Habazaki, M. Uozumi, H. Konno, K. Shimizu, P. Skeldon, G.E. Thompson,
Crystallization of anodic titania on titanium and its alloys, Corros. Sci. 45
(2003) 2063–2073.
[59] T.H. Courtney, Mechanical Behavior of Materials, Second Edition, Waveland
Press, 2005.
L. Fadillah, D. Kowalski, Y. Aoki et al.
[60] H. Kronmüller, S.S.P. Parkin, Handbook of Magnetism and Advanced Magnetic
Materials, John Wiley & Sons, Hoboken, NJ, 2007.
[61] G.N. Derry, M.E. Kern, E.H. Worth, Recommended values of clean metal surface
work functions, J. Vac. Sci. Technol. A 33 (2015) 060801.
[62] R.C. Hall, Single-crystal magnetic anisotropy and magnetostriction studies in
iron-base alloys, J. Appl. Phys. 31 (1960) 1037–1038.
[63] B. Hillebrands, P. Baumgart, G. Güntherodt, In situ Brillouin scattering from
surface-anisotropy-dominated Damon-Eshbach modes in ultrathin epitaxial
Fe(110) layers, Phys. Rev. B 36 (1987) 2450–2453.
[64] G. Rupp, W. Wettling, R.S. Smith, W. Jantz, Surface magnons in anisotropic fer-
romagnetic films, J. Magn. Magn. Mater. 45 (1984) 404–408.
11
Electrochimica Acta 364 (2020) 137316
[65] Y. Ueda, M. Takahashi, Structure and magnetic properties in single-crystal iron
film electrodeposited on a (110) copper crystal, J. Magn. Magn. Mater. 71
(1988) 212–218.
[66] X. Liu, M.M. Steiner, R. Sooryakumar, G.A. Prinz, R.F.C. Farrow, G. Harp, Ex-
change stiffness, magnetization, and spin waves in cubic and hexagonal phases
of cobalt, Phys. Rev. B 53 (1996) 12166–12172.
[67] M. Yamamoto, On the elastic constants of iron single crystals, J. Jpn. Inst. Met.
7 (1943) 346–348.
[68] J.E. Houser, K.R. Hebert, The role of viscous flow of oxide in the growth of
self-ordered porous anodic alumina films, Nat. Mater. 8 (2009) 415–420.
 View Article Online
View Journal

Journal of
Materials Chemistry A
Materials for energy and sustainability

Accepted Manuscript

This article can be cited before page numbers have been issued, to do this please use: C. Tang, K.
Akimoto, N. Wang, L. Fadillah, S. kitano, H. Habazaki and Y. Aoki, J. Mater. Chem. A, 2021, DOI:
10.1039/D1TA02848K.
Volume 6
Number 12
28 March 2018
This is an Accepted Manuscript, which has been through the
Pages 4883-5230
Royal Society of Chemistry peer review process and has been
Journal of accepted for publication.
Materials Chemistry A
Materials for energy and sustainability
rsc.li/materials-a
Accepted Manuscripts are published online shortly after acceptance,
before technical editing, formatting and proof reading. Using this free
service, authors can make their results available to the community, in
citable form, before we publish the edited article. We will replace this
Accepted Manuscript with the edited and formatted Advance Article as
soon as it is available.

You can find more information about Accepted Manuscripts in the

Information for Authors.

Please note that technical editing may introduce minor changes to the
text and/or graphics, which may alter content. The journal’s standard
ISSN 2050-7488 Terms & Conditions and the Ethical guidelines still apply. In no event
COMMUNICATION
Zhenhai Wen et al.
An electrochemically neutralized energy-assisted low-cost
shall the Royal Society of Chemistry be held responsible for any errors
acid-alkaline electrolyzer for energy-saving electrolysis
hydrogen generation
or omissions in this Accepted Manuscript or any consequences arising
from the use of any information it contains.

rsc.li/materials-a
 Page 1 of 11 Journal
Please of
doMaterials Chemistry
not adjust marginsA

View Article Online

DOI: 10.1039/D1TA02848K

PAPER

Journal of Materials Chemistry A Accepted Manuscript

Received 00th January 20xx,
Accepted 00th January 20xx
DOI: 10.1039/x0xx00000x
Published on 26 May 2021. Downloaded by Hokkaido Daigaku on 5/26/2021 1:59:31 PM.
Effect of Anode Functional Layer on Steam Electrolysis
Performances of Protonic Solid Oxide Cells
Chunmei Tang,a Katsuya Akimoto,a Ning Wang,a Laras Fadillah,a Sho Kitano,b Hiroki Habazaki,b and
Yoshitaka Aoki*b

Protonic solid oxide steam electrolysis cells (P-SOECs) based on BaZrxCe0.8-xYb0.1Y0.1O3−δ proton conductors are promising to
produce “green” hydrogen from renewable energy at intermediate temperatures. Herein, we demonstrate that the
electrolysis performances of a cell with a high-Zr-content electrolyte, BaZr0.6Ce0.2Y0.1Yb0.1O3−δ (BZCYYb6211), can be
significantly improved by using a La0.5Sr0.5CoO3−δ (LSC) thin film (~90 nm) as an anode functional layer (AFL). Electrochemical
measurements indicated that LSC-AFL significantly reduced the barrier height for the electrochemical proton incorporation
reaction at the gas-electrolyte-electrode triple-phase boundary. Hence, both ohmic and polarization resistances of the
BZCYYb6211 cell decreased from 0.52 and 0.98 Ω cm2 to 0.26 and 0.57 Ω cm2, respectively, with the LSC-AFL at 600 C. In
addition, the BZCYYb6211 cell achieved a high electrolysis current of 1.22 A cm−2 at 1.3 V with Faraday efficiency of
approximately 80%, which was equivalent to that (1.13 A cm−2) of the cell with a state-of-the-art electrolyte
BaZr0.1Ce0.7Y0.1Yb0.1O3−δ (BZCYYb1711). BZCYYb6211 with LSC-AFL showed good durability at 500 °C under high steam
conditions with an applied current of 1 A cm−2 for 100 h. These results revealed that the introduction of the AFL is an effective
method to obtain P-SOECs with excellent performances and durability.

aspects of the water splitting reaction. First, the reaction enthalpy

Introduction (rH) of steam electrolysis is smaller than that of water
electrolysis because of the vaporization heat. Second, the Gibbs
Decarbonizing the planet is one of the goals set by countries
energy (rG) decreases with temperature; therefore, the standard
around the world for 2050. To achieve this, one of the key
potential (E0) decreases owing to the increase in heat term, TrS,
measure is decarbonizing the production of hydrogen, resulting
where rS and T are the reaction entropy and temperature,
in “green” hydrogen, as this process is currently responsible for
respectively. Thus, solid oxide electrolysis cells (SOECs) enable
more than 2% of the total global CO2 emissions. To date,
a highly efficient process for converting renewable energy to
hydrogen required for industrial applications is usually produced
green hydrogen. 3–5
from hydrocarbons by steam reforming, which is costly and leads
Traditional SOECs use an oxide ion conductor, such as
to CO2 emission.1, 2 Instead of such “gray” hydrogen process, the
yttrium-stabilized zirconia (YSZ), as a solid electrolyte (O-
renewable energy grid integrated water electrolysis system
SOECs), and thus require operation in the high-temperature
makes carbon-free, green hydrogen production feasible. Hence,
region (700−1000 °C) to achieve significantly high ion
renewable energy storage in the form of hydrogen via water
conductivity. However, the severe operating conditions involve
electrolysis is a key technology for next-generation energy
various problems: thermal cell degradation, material corrosion,
carriers and conversion. There are three main types of water
difficulty in heat management and so on.3, 6–8 Protonic solid
electrolysis technologies: alkaline water electrolysis, polymer
oxide electrolysis cells (P-SOECs) based on proton-conducting
electrolyte membrane water electrolysis, and solid oxide
perovskites, that is, Ba(Ce, Zr, M)O3− (M = Y, Yb, etc.) are
electrolysis. The first two processes are operated near room
attractive alternatives to O-SOECs because the relatively low
temperature, and the third is normally conducted at a high
activation energy of proton conduction (~0.5 eV) enables
temperature with steam. Alkaline water electrolysis is the most
operation in the intermediate temperature region (400−600 °C). 9–
mature technology and has been commonly utilized at a 12
Ba(Ce, Zr, M)O3– tends to be more thermodynamically stable
commercial level. Steam electrolysis has some advantages in
under high H2O partial pressure (pH2 O) with increasing Zr
comparison to water electrolysis based on the thermodynamic
content,13–15 and thus, Zr-rich side materials are desirable as a
practical electrolyte for P-SOECs. Nevertheless, the electrolysis
performances of P-SOECs with Zr-rich side electrolyte is
a. Graduate School of Chemical Sciences and Engineering, Hokkaido University,
N13W8, Kita-ku, Sapporo, 060-8628, Japan.
b. Faculty of Engineering, Hokkaido University, N13W8, Kita-ku, Sapporo, 060-8628,

Japan. E-mail: y-aoki@eng.hokudai.ac.jp

†Electronic Supplementary Information (ESI) available. See
DOI: 10.1039/x0xx00000x

Please do not adjust margins

 Journal
Please of
doMaterials Chemistry
not adjust marginsA
Paper
Published on 26 May 2021. Downloaded by Hokkaido Daigaku on 5/26/2021 1:59:31 PM.
Fig. 1 Schematic illustration of the fabrication of thin-film electrolysis cell.
considerably inferior compared to that of P-SOECs with Ce-rich
side. For instance, Duan et al. conducted a study on
BaZr0.1Ce0.7Y0.1 Yb0.1O3−δ (BZCYYb1711) can yield an electric
current density of ~1.00 A cm−2 at a thermal neutral voltage of
1.30 V at 600 °C, 12 which is considerably higher than the
performances of O-SOECs with YSZ electrolyte. 16, 17 However,
the cell with Ba(Zr0.5Ce0.4)8/9Y0.2O2.9 can produce only 0.50 cm−2
at 1.45 V at 600 C.18 One major reason for the deteriorated
performances in Zr-rich side electrolytes is the relatively large
ohmic resistance of the electrolyte. This is because Ba(Ce, Zr,
M)O3− tends to show larger grain boundary resistances with
increasing Zr content due to the highly refractory nature of
BaZrO3 moieties.
Recently, Choi et al. reported on the significant improvement
of the ohmic loss and thus electrolysis performances of P-SOECs
with relatively high Zr content BaZr 0.4Ce0.4Y0.1Yb0.1O3−
(BZCYYb4411) by implementing an interfacial layer of
H+/O2−/e− triple-conducting PrBa0.5Sr0.5Co1.5 Fe0.5O5+δ (PBSCF)
between the electrolyte and the anode.19 Thus, the ohmic loss of
the 15 m-thick electrolyte film decreased from 0.18 to 0.10 Ω
cm2 at 600 C under SOEC conditions by depositing a dense thin
film (~100 nm) of PBSCF on the electrolyte surface at the anode
side with pulsed laser deposition (PLD) technique. Therefore, the
P-SOECs reached ~1.80 A cm−2 at 1.3 V at 600 C, which is
higher than the performances of the aforementioned
BZCYYb1711 cells.12 This result implies the importance of the
anode functional layer (AFL) developed between porous anode
and electrolyte in improving the performances of P-SOECs.
Herein, we demonstrate that the well-known O2−/e− double
conducting oxide La0.5 Sr0.5CoO3−δ (LSC) is a promising AFL for
P-SOECs with the state-of-the-art BZCYYb1711 electrolyte and
the Zr-rich side BaZr0.6Ce0.1Y0.1Yb0.1 O3−δ (BZCYYb6211), as
well as triple-conducting PBSCF. The implementation of LSC-
AFL decreased the ohmic resistances due to proton bulk
conduction even though LSC was not highly proton conductive.
It also decreased the resistances related to electrochemical proton
incorporation at anode-electrolyte-gas triple phase boundaries by
a factor of 50% at approximately 600 C. BZCYYb6211 base P-
SOECs yielded 1.22 A cm−2 at 600 C at 1.3 V with the aid of
LSC-AFL, which is close to the electrolysis current of a similar
type BZCYYb1711 cell.
2 | J. Name., 2012, 00, 1-3
Please do not adjust margins
Page 2 of 11
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/D1TA02848K
Journal of Materials Chemistry A Accepted Manuscript
Experimental
Synthesis of La0.8Sr0.2Co0.7NiO3−δ anode powders
We chose La0.8Sr0.2Co0.7 Ni0.3O3−δ (LSCN8273) as the anode of
P-SOECs because it has been demonstrated to be efficient with
excellent activity and durability.20 The powders were
synthesized via a citrate precursor route, in which the required
amounts of nitrate precursors of La(NO3)3·6H2O, Sr(NO3)2,
Co(NO3)2·6H2O, and Ni(NO3)2·6H2O were dissolved in Milli-Q
H2O. The chelating agent citric acid (CA; C6H8O7·H2O) was
added at a molar ratio of CA:LSCN = 2:1. Gelatinous products
were obtained by heating the citrate solution at 60 °C for 3 h with
vigorous stirring to evaporate H2O and promote polymerization.
The gel was calcined at 500 °C for 1 h in air and then annealed
at 800 °C for 15 h in O2 to obtain single-phase LSCN8273
powders. In a separate experiment La0.5Sr0.5CoO3−δ (LSC)
powders were also applied to the anode, which was prepared by
the similar method as LSCN8273.
Synthesis of BaZrxCe0.8−xY0.1Yb0.1O3−δ
BaZrxCe0.8−xY0.1Yb0.1O3−δ (x = 0.1, 0.6; BZCYYb1711,
BZCYYb6211, respectively) electrolyte powders were prepared
with stoichiometric quantities of BaCO3 (High Purity Chemicals,
99.95%), ZrO2 (High Purity Chemicals, 98%), CeO2 (High
Purity Chemicals, 99.99%), Y2O3 (High Purity Chemicals,
99.99%), and Yb2O3 (High Purity Chemicals, 99.9%). The
mixture was first ball-milled for 10 h using ethanol as the milling
medium, and then the ethanol was removed by heating the
mixture at 80 °C. The powders were calcined at 1300 °C for 10
h after light grounding by hand. To ensure the formation of
phases, milling and calcination were repeated.
Fabrication of P-SOECs
Cathode-supported P-SOECs were fabricated as shown in Fig. 1.
The porous cathode was prepared by ball-milling NiO (High
Purity Chemicals, 99.97%), electrolyte powders (either
BZCYYb1711 or BZCYYb6211), and starch with a weight ratio
of 60:40:7, respectively, for 10 h in ethanol. After drying, the
mixture powders were uniaxially pressed into pellets (~12 mm
This journal is © The Royal Society of Chemistry 20xx
 Page 3 of 11 Journal
Please of
doMaterials Chemistry
not adjust marginsA
Journal of Materials Chemistry A
diameter, ~1.8 mm thickness) under 20 MPa and subsequently
isostatically pressed under a hydrostatic pressure of 100 MPa.
The electrolyte layer was spin-coated on both surfaces of the
porous cathode pellets with a slurry, which was prepared by
dispersing electrolyte powders with 1 wt.% NiO into a solution
containing a dispersant (20 wt.% polyethyleneimine (Mw 28 000)
dissolved in α-terpineol) and a binder (5 wt.% surfactant
dissolved in α-terpineol). Subsequently, the pellet was first
exposed to 1450 °C for 10 min and then at 1400 °C for 8 h in an
air atmosphere to form a half-cell. The back side of the sintered
pellet was polished with SiC paper. A LSC thin film was
deposited on the electrolyte layer by radio frequency (RF)
sputtering at a base pressure of less than 1 × 10 −4 Pa, performed
in an ultrahigh-vacuum chamber system (ULVAC ACS-3000).
Published on 26 May 2021. Downloaded by Hokkaido Daigaku on 5/26/2021 1:59:31 PM.
RF sputtering was performed at a sputtering power of 50 W
under a flow of 4%−O2/Ar gas at 50 sccm, and the substrate
temperature was maintained at 500 °C. Finally, a LSCN8273
slurry prepared by dispersing the powders into a mixture of
dispersant and binder was screen-printed on the electrolyte or
LSC-AFL as a porous anode.
Characterization
The phase was identified by X-ray powder diffraction (XRD,
Rigaku, Ultima IV) with Cu Kα radiation at a scan rate of 5°
min−1 in the 2θ range from 20° to 80°. For the XRD
measurements, the sintered body specimens were pulverized in a
mortar. The structure of the LSC thin film deposited on a silicon
plate was characterized using a grazing incidence X-ray
diffractometer (GIXRD, Rigaku, RINT-2000) at a scan speed of
0.5° min−1 and a grazing incidence angle of 2°. Field emission
scanning electron microscopy (FESEM, SIGMA500, ZEISS)
was employed to examine the microstructures of the cells. High-
resolution transmission electron microscopy (HR-TEM) was
performed using a field-emission TEM (Jeol JEM-2010)
equipped with an energy dispersive X-ray spectrometer (EDX).
For steam electrolysis, P-SOECs were mounted in a laboratory-
constructed test station with the seal of a molten glass ring
gasket. The cathode was reduced at 700 °C for 3 h by supplying
60%−H2/Ar gas to the cathode side and 3%−H2O/Ar to the anode
side to form a porous Ni-BZCYYb cermet cathode. For steam
electrolysis, steamed air was fed to the anode at a total flow of
62 sccm, and humidified hydrogen gas with a ratio of H2O/H2/Ar
= 2/10/90 was fed to the cathode at a total flow rate of 20.4 sccm.
Steamed air was prepared using a temperature-controlled water
bubbler with an inlet of 20%−O2/Ar mixed gas. The water partial
pressure (pH2 O) of steamed air was set to 0.3 atm. Humidified
hydrogen gas was prepared by bubbling 10%−H2/Ar = 10/90 in
pure water maintained at 25 C. The impedance spectra of P-
SOECs were measured using a Solartron 1260A frequency
response analyzer implemented with a Solartron 1287
potentiostat in the frequency range of 106−10−1 Hz with an
alternating current (AC) amplitude of 30 mV under open circuit
voltage (OCV) conditions and different direct current (DC)
potentials. Current-voltage (I-V) curves were recorded using the
same apparatus.
The amount of hydrogen evolution through electrolysis was
determined by analyzing the exhaust gas from the cathode side
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins
Paper
View Article Online
DOI: 10.1039/D1TA02848K
Journal of Materials Chemistry A Accepted Manuscript
Fig. 2 (a) Powder XRD patterns of a sintering cake comprising a dense BZCYYb film and
BZCYYb-NiO composite bulk. (b) and (c) show surface SEM images of BZCYYb1711 and
BZCYYb6211 electrolyte films, respectively.
using a gas chromatograph (490 Micro GC, Agilent
Technologies). The Faraday efficiency, η, was calculated
according to:
nH2 , measured nH2 , measured
η= = × 100%
nH2 , theoretical I × (n × F)−1
Here, nH2 , measured is the measured hydrogen evolution rate, I is the
applied current, n (2) is the electron transport number of steam
electrolysis, and F is Faraday’s constant (96485 C mol −1).
Results and discussion
Material characterization
In general, water tolerance of proton conducting perovskite is
increased with Zr contents of B site cations.13–15, 21, 22 In this
regard, BaZr0.6Ce0.1Y0.1Yb0.1O3−δ (BZCYYb6211) is attractive
electrolyte for steam electrolysis conducted under high water
partial pressure (pH2 O) condition. BaZr0.1Ce0.7Y0.1Yb0.1O3−δ
(BZCYYb1711) has been recognized as a state-of-the-art
electrolyte of proton conducting perovskites with satisfactory
high proton conductivity, as mentioned before.[12, 23–25] Here the
effects of AFL was examined on both BZCYYb1711 and
BZCYYb6211 electrolytes. Both electrolyte powders were
successfully prepared as single-phase monoclinic and cubic
perovskites, respectively (BaCe0.8Y0.2 O3−δ PDF-01-070-6753;
BaZrO3 PDF-06-0399; Fig. S1). Fig. 2a shows the powder XRD
patterns of a sintering cake for dense BZCYYb films on
BZCYYb-NiO composite supports. All the diffraction peaks
were consistent with the BZCYYb perovskites and NiO. The
peaks of BZCYYb6211 emerged at higher angles than the
corresponding peaks of BZCYYb1711, indicating that the lattice
constants of the former were smaller than those of the latter due
to the substitution of Ce4+ (0.87 Å) by Zr4+ (0.72 Å). The surface
SEM images of BZCYYb1711 and BZCYYb6211 electrolyte
films are shown in Figs. 2b and c, respectively. Due to the easy
sinterability of the Ce-rich phase compared to that of the Zr-rich
phase,26, 27 the average grain size of BZCYYb1711 (ca. 20−25
μm) was considerably larger than that of BZCYYb6211 (ca. 3−4
μm).
J. Name ., 2013, 00, 1 -3 | 3
 Journal
Please of
doMaterials Chemistry
not adjust marginsA
Paper
Published on 26 May 2021. Downloaded by Hokkaido Daigaku on 5/26/2021 1:59:31 PM.
Fig. 3 (a) Cross-section SEM image of cathode-supported P-SOECs of BZCYYb6211
electrolyte. (b) TEM of interface between LSC-AFL and BZCYYb6211 electrolyte. (c) STEM-
EDX mapping and (d) HR-TEM image of the LSC thin film deposited on BZCYYb6211
electrolyte. The inset of (d) shows the fast Fourier transform patterns of the lattice fringe
image.
Fig. 3a shows the cross-sectional SEM image of a
BZCYYb6211 P-SOEC comprising a porous cathode support,
electrolyte film, AFL, and La0.8Sr0.2Co0.7Ni0.3O3−δ (LSCN8273)
anode. Ni-BZCYYb6211 cathode supports retain
interpenetrating networks of macro-and micropores, which must
be formed via the combustion of starch and the reduction of NiO,
thus providing a sufficient gas diffusion pathway. A highly dense
electrolyte film of BZCYYb6211 was uniformly formed over a
wide area of porous Ni-BZCYYb cathode with a thickness of
~14 μm. The TEM image of the interface between LSC-AFL and
Fig. 4 Electrochemical performances of (a) and (b) BZCYYb1711 base cells and (c) and (d) BZCYYb6211 base cells. (a) and (c) are the cells without LSC-AFL, and (c) and (d) are cells
with LSC-AFL. Electrolysis current density at 1.3 V for (e) BZCYYb1711 and (f) BZCYYb6211 base cells.
4 | J. Name., 2012, 00, 1-3
Please do not adjust margins
Page 4 of 11
Journal of Materials Chemistry A
BZCYYb6211 electrolyte revealed that the LSC-AFL with
View Article a
Online
thickness of ~90 nm was uniformly formed DOI:over
10.1039/D1TA02848K
the surface of
the electrolyte (Fig. 3b). The GIXRD patterns of the LSC thin
film deposited on a silicon wafer by RF sputtering were identical
to those of rhombohedral La0.5 Sr0.5CoO2.91 (R-3c, PDF-48-0122;
Journal of Materials Chemistry A Accepted Manuscript
Fig. S2). Further characterization was conducted for the LSC-
AFL of the cell before printing the porous anode. Fig. 3c shows
the scanning TEM (STEM)-EDX mapping analysis, confirming
that La, Sr, and Co atoms were uniformly distributed throughout
the film while preserving the same molar ratio of La:Sr:Co =
1:1:2 as the parent phase. HR-TEM results exhibited a clear
lattice fringe with an interplanar lattice distance of 0.271 nm (Fig.
3d), which corresponds to the (104) crystal plane, indicating that
a well-defined LSC thin film was uniformly formed over the
electrolyte surface. The LSCN8273 anode layer was a porous
agglomerate of oxide particles with a diameter of several tens of
nanometers (Figs. S3 and 4b).
Electrochemical performances of P-SOECs without and with LSC-
AFL
The electrochemical performances of P-SOECs based on
BZCYYb1711 and BZCYYb6211 electrolytes were evaluated in
the temperature range of 700−500 °C by supplying
30%−H2O/Air and humidified 10%−H2/Ar gases to the anode
and cathode, respectively. Fig. 4 shows the current-voltage
relationships of the P-SOECs with and without LSC-AFL (~90
nm). For the cell without LSC-AFL layer, the OCVs of
BZCYYb1711 and BZCYYb6211 cells were 0.93 and 0.92 V,
respectively, at 600 ℃ (Figs. 4a and c, respectively), which were
slightly lower than the theoretical value (ENernst) of 0.96 V as
calculated by conventional Nernst equation. The steam
electrolysis currents of the BZCYYb1711 cell were equal to 2.10,
1.11, 0.57, 0.30, and 0.15 A cm−2 at 700, 650, 600, 550, and
500 ℃, respectively, at the thermal neutral point (approximately
This journal is © The Royal Society of Chemistry 20xx
 Page 5 of 11 Journal
Please of
doMaterials Chemistry
not adjust marginsA

Journal of Materials Chemistry A Paper

1.3 V in the temperature range; Fig. 4a). BZCYYb6211 exhibited higher than those of cells without LSC-AFL, respectively
View Article(Figs.
Online
inferior performances to BZCYYb1711 at relatively high 4e and f). DOI: 10.1039/D1TA02848K
temperatures, with electrolysis currents of 1.50 and 0.96 A cm −2, The electrolysis currents of the cells with LSC-AFL were
at 700 and 650 °C, respectively, at 1.3 V. The electrolysis considerably higher than those of most previously reported P-
currents of the former were similar to those of the latter at SOECs, as summarized in Table 1. Fig. S5 shows I-V curves for

Journal of Materials Chemistry A Accepted Manuscript

temperatures below 600 °C, yielding 0.55, 0.30, and 0.14 A cm−2
at 600, 550, and 500 °C, respectively (Fig. 4c). The electrolysis
currents of P-SOECs significantly increased with the
introduction of the LSC-AFL between the electrolyte film and
anode at all temperatures analyzed (Figs. 4b and d). With the
LSC-AFL, the currents of BZCYYb1711 and BZCYYb6211
base cells provided OCVs similar to those of cells without LSC-
AFL are 3.07 and 3.07 A cm−2, respectively, at 700 ℃ by
applying 1.3 V cell voltage. These values were 46.19% and
Published on 26 May 2021. Downloaded by Hokkaido Daigaku on 5/26/2021 1:59:31 PM.
several samples of BZCYYb6211 cells with LSC-AFL at 600 °C,
indicating that the curves are very similar to each other, which
confirms the excellent performances of the current cells.
Fig. S6 shows the current-voltage-power (I-V-P) curves of
BZCYYb1711 and BZCYYb6211 base cells with and without
LSC-AFL in the fuel cell mode. The peak power densities (PPDs)
of BZCYYb1711 and BZCYYb6211 cells with LSC-AFL were
considerably higher than those of cells without LSC-AFL. For
instance, the PPDs of the BZCYYb6211 cell with LSC-AFL

104.46% higher than the corresponding values of the cells
without LSC-AFL, respectively (Figs. 4e and f, respectively).
Similarly, the currents at 1.3 V were 1.13 and 1.22 A cm−2 for
BZCYYb1711 and BZCYYb6211 base cells, at 600 ℃, which
were 97.55% and 120.61% higher than the values of the
corresponding cells without LSC-AFL, respectively. Even at the
relatively low temperature of 500 ℃, the BZCYYb1711 and
BZCYYb6211 base cells with LSC-AFL exhibited 0.31 A cm−2
and 0.30 A cm−2 at 1.3 V, and 0.65 and 0.76 A cm -2 at 1.5 V,
respectively. The values at 1.3 V were 110.81% and 117.73%
were 316, 297, 233, 171, and 110 mW cm −2 at 700, 650, 600, 550,
and 500 °C, respectively, which were considerably higher than
the values of the cell without LSC-AFL at each temperature.
These results indicate that AFL promotes both oxygen evolution
and reduction reactions in protonic ceramic electrolytes.
Fig. 5 shows the voltage responses and H2 evolution rates of
BZCYYb1711 and BZCYYb6211 cells without and with LSC-
AFL during galvanostatic electrolysis at 500 C. Here, the
constant current was set to 0.15 A cm−2 for the cells without AFL
(Fig. 5a) and 0.24 A cm−2 for the cells with AFL (Figs. 5b and c)

Table 1 Performances of steam electrolysis based on P-SOECs at 600 ℃ under 1.3 V in recent years.

Configuration of electrolysis cell: Inlet gas composition OCV/E Nernst I

Ref.
anode/electrolyte (thickness)/cathode Anode Cathode (V) (A cm −2)
LSCN8273/BZCYYb1711 (14 μm)/Ni-BZCYYb1711 Air (30% H 2 O) 10% H 2 /Ar (3% H 2 O) 0.93/0.96 0.57 This
LSCN8273/BZCYYb6211 (14 μm)/Ni-BZCYYb6211 Air (30% H 2 O) 10% H 2 /Ar (3% H 2 O) 0.92/0.96 0.55 This
LSCN8273/LSC/BZCYYb1711 (14 μm)/Ni-BZCYYb1711 Air (30% H 2 O) 10% H 2 /Ar (3% H 2 O) 0.89/0.96 1.13 This
LSCN8273/LSC/BZCYYb6211 (14 μm)/Ni-BZCYYb6211 Air (30% H 2 O) 10% H 2 /Ar (3% H 2 O) 0.91/0.96 1.22 This
SFM/BZY82 (16 μm)/Ni-BZY82 Air (3% H 2 O) 10% H 2 /N 2 0.86/1.05 0.21 28
SEFC/BZCY172 (15 μm)/Ni-BZCY172 Air (10% H 2 O) 97% H 2 (3% H 2 O) 0.99/1.05 ~0.40 29
SLF/BZCY352 (20 μm)/Ni-BZCY352 Air (20% H 2 O) 97% H 2 (3% H 2 O) 0.96/1.04 0.46 30
NBSCF-BZCYYb1711/BZCYYb1711 (20 μm)/Ni-BZCYYb1711 Air (10% H 2 O) 97% H 2 (3% H 2 O) 1.03/1.05 0.75 24
PNO-BZCY262/ BZCY262 (20 μm)/ Ni-BZCY262 Air (40% H 2 O) 100% H 2 0.98/1.02 0.60 31
LSN/BZCY172 (15 μm)/Ni-BZCY172 Air (20% H 2 O) 97% H 2 (3% H 2 O) 1.02/1.04 0.42 32
PBSCF(3D)/BZCYYb (20 μm)/Ni-BZCYYb O 2 (12% H 2 O) 5% H 2/Ar 0.99/1.00 0.85 25
PBCC95/BZCYYb4411 (20 μm)/Ni-BZCYYb4411 O 2 (20% H 2 O 100% H 2 1.06/1.16 0.72 33
PNC/BZCYYb4411 (15 μm)/Ni-BZCYYb4411 Air (10% H 2 O) 10% H 2 /Ar 1.01/1.00 1.18 34
PNO/LCO|BZCYYb1711 (bilayer 20 μm)/Ni-BZCYYb1711 Air (60% H 2 O) 100% H 2 0.95/1.00 0.33 35
LSN/BZCYYbCu (13 μm)/Ni-BZCYYbCu Air (20% H 2 O) 97% H 2 (3% H 2 O) 0.99/1.06 0.59 36
PBSCF(PLD-modified)/BZCYYb4411 (15 μm)/Ni-BZCYYb4411 Air (3% H 2 O) 97% H 2 (3% H 2 O) 1.03/1.08 1.80 19
BCFZY/BZCYYb1711 (12 μm)/Ni-BZCYYb1711 Air (10% H 2 O) 100%Ar − 1.00 12
PNC/BZCYYbGd (25 μm)/Ni-BZCYYbGd O 2 (20% H 2 O) 100% H 2 1.02/1.09 0.56 37
BZCY36-BCFZY/BZCYSm (25 μm)/Ni-BZCYSm Air (12% H 2 O) 97% H 2 (3% H 2 O) 1.05/1.08 0.37 38
LSN- BZCYYbC2/BZCYYbCu (13 μm)/Ni-BZCYYbCu Air (20% H 2 O) 97% H 2 (3% H 2 O) 0.98/1.08 1.03 39
SFM-BZY82/BZY82 (18 μm)/Ni-BZCY172 Air (3% H 2 O) 20% H 2 /N 2 0.92/1.08 0.38 40
PBSCF/BHCYYb3511 (10 μm)/Ni-BHCYYb3511 Air (3% H 2 O) 97% H 2 (3% H 2 O) 1.04/1.13 1.45 41

Abbreviations: Sr2Fe1.5Mo0.5O6−δ (SFM); BaZr0.8Y0.2O3−δ (BZY82); SrEu2Fe1.8Co0.2O7−δ (SEFC); BaZr0.1Ce0.7Y0.2O3−δ (BZCY172); Sr2.8La0.2Fe2O7−δ (SLF); BaZr0.3Ce0.5Y0.2O3−δ (BZCY352);
NdBa0.5Sr0.5Co1.5Fe0.5O5+δ (NBSCF); Pr2NiO4+δ (PNO); BaZr0.2Ce0.6Y0.2O3−δ (BZCY262); La1.2Sr0.8NiO4 (LSN); PrBa0.5Sr0.5Co2−xFexO5+δ (PBSCF), BaZr0.1Ce0.7Y0.2−xYbxO3−δ (BZCYYb);
(PrBa0.8Ca0.2)0.95Co2O6−δ (PBCC95); BaZr0.4Ce0.4Y0.1Yb0.1O3−δ (BZCYYb4411); PrNi0.5Co0.5O3−δ (PNC); BaZr0.1Ce0.68Y0.1Yb0.1Cu0.02O3−δ (BZCYYbCu); BaCo0.4Fe0.4Zr0.1Y0.1O3−δ (BCFZY);
BaZr0.2Ce0.5Y0.1Yb0.1Gd0.1O3−δ (BZCYYbGd); BaZr0.1Ce0.7Y0.1Sm0.1O3−δ (BZCYSm).

This journal is © The Royal Society of Chemistry 20xx J. Name ., 2013, 00, 1 -3 | 5

Please do not adjust margins

 Journal
Please of
doMaterials Chemistry
not adjust marginsA
Paper
Published on 26 May 2021. Downloaded by Hokkaido Daigaku on 5/26/2021 1:59:31 PM.
Fig. 5 Responses of cell voltages and transients of H 2 gas evolution rates during
galvanostatic electrolysis at 500 ℃. (a) BZCYYb1711 cell without LSC-AFL, (b)
BZCYYb1711 cell with LSC-AFL, and (c) BZCYYb6211 cell with LSC-AFL. The constant
current was set to 0.15 A cm−2 for (a) and 0.24 A cm−2 for (b) and (c). Blue lines show the
cell voltage and red symbols show the H2 evolution rate determined by gas
chromatography. Black dashed lines indicate the rate calculated with 100% Faradaic
efficiency.
to investigate the behavior near the thermal neutral point around
1.3 V. The hydrogen concentrations in the cathode exhaust gases
were monitored by gas chromatography, which allowed the
determination of the H2 evolution rate. The concentrations were
in equilibrium for approximately 1 h after the beginning of
electrolysis in every cell, and thus the average rates after 1 h were
used to calculate the Faraday efficiency (see Experimental
section). The efficiency of BZCYYb1711 without and with LSC
was approximately 76% and 71%, respectively. The efficiency
of less than 100% is due to the hole conductivity of
Ba(Zr,Ce,Y)O3−δ electrolytes, 42–44 which is attributed to the
relatively low steam pressure (pH2 O) at the anode and hydrogen
pressure (pH2) at the cathode because the transport number of
holes in the BaZr0.7Ce0.2YO2.95 electrolyte is close to 0 with
increasing pressure of pH2 O and pH2 to 25 and 50 atm,
respectively.42 Both BZCYYb6211 and BZCYYb1711 cells with
LSC achieved an efficiency of 70%. These values of Faraday
efficiency were consistent with the corresponding values ranging
from 40% to 86% for the P-SOECs at 1.3 V under low pH2 O and
pH2 conditions. 19, 34 The slight degradation in Faraday efficiency
indicates the LSC-AFL promotes the conduction of both proton
6 | J. Name., 2012, 00, 1-3
Please do not adjust margins
Page 6 of 11
Journal of Materials Chemistry A
View Article Online
DOI:to
and hole in electrolyte, which is probably due 10.1039/D1TA02848K
the modification
of hydrogen and/or oxygen potential at the interface.
Nevertheless, the hydrogen production rate clearly increases
from ~40 to ~60 μmol min−1 cm−2 at 1.3 V by LSC-AFL (Fig. 5).
Journal of Materials Chemistry A Accepted Manuscript
Impacts on ohmic and polarization resistances by LSC-AFL
Electrochemical impedance spectra (EIS) were measured
under various DC conditions and temperatures, as shown in Fig.
6 and Figs. S7 and S8, to deconvolute the ohmic and polarization
resistances of the P-SOECs. Usually, the impedance spectra of
solid electrochemical cells include the high-frequency x-
intercept region, corresponding to ohmic loss, that is, the bulk
resistances (Ro) and the following arcs are associated with the
interfacial polarization resistances (Rp) on the anode side. 16, 29, 44
The P-SOECs yield two distinct arcs in the EIS spectra: an SHF
arc (Fig. 6) in the high-frequency region of 105−103 Hz, and an
SLF (Fig. 6) arc in the low-frequency region of 103−10-1 Hz.
Hence, the EIS spectra of P-SOECs at 600 and 500 °C were fitted
with the equivalent circuit Ls-Ro–(RHF-CPEHF)–(RLF-CPE LF)
depicted in the inset of Fig. 6c, where L, R, and CPE are the
inductance, resistance, and constant phase element, that is,
pseudo-capacitance, respectively. Ls comes from the electrical
metal lead, and Ro is mainly attributed to proton conduction in
the electrolyte. Parallel components of (RHF-CPEHF) and (RLF-
CPELF) were used to represent SHF and SLF, respectively.
Therefore, RHF and RLF provide the polarization resistance
related to SHF and SLF, respectively. In all cells, the observed EIS
spectra were well fitted with the equivalent circuit model, as
shown in Figs. 6a-d, and the results are summarized in Table S1.
In case of the cells without LSC-AFL, Ro of BZCYYb1711
was lower than that of BZCYYb6211 by 20−50% in the entire
measured temperature range (Fig. S7). The values of
BZCYYb1711 were 0.30 and 0.61 Ω cm2 at 600 and 500 ℃,
respectively, whereas those of BZCY6211 cells were 0.52 and
0.83 Ω cm2 at 600 and 500 ℃, respectively. The relatively large
Ro of BZCY6211 could be attributed to the grain boundary
resistances because the BZCYYb6211 film had a smaller grain
size than the BZCYYb1711 film (Figs. 2b and c). Ro was
considerably reduced with increasing cell voltage in both cells
(Fig. S8), which is probably due to the increment of both proton
and hole currents as mentioned before.46, 47 Ro of BZCYYb1711
decreased from 0.30 to 0.27 Ω cm2 with switching from OCV to
1.2 V at 600 C (Fig. S8a), and similarly, Ro of BZCYYb6211
decreased from 0.52 to 0.45 Ω cm2 (Fig. S8c). Figs. 7a and b
show the Arrhenius plots of Ro−1, revealing that the activation
energies of bulk proton conduction were 0.37 and 0.28 eV for
BZCYYb1711 and BZCYYb6211, respectively, which were
similar to the early corresponding reports of BZCYYb1711
(~0.40 eV) and BZCY622 (~0.33 eV).41, 48
The RHF and RLF significantly decreased with increasing cell
voltage (Fig. S8), which confirms that the concentration
overpotential owing to the slow gas diffusion was relatively
small; thus, both resistances were correlated with the anode
reaction kinetics. Without LSC-AFL, BZCYYb1711 cells
exhibited larger RHF than BZCYYb6211 cells, although the RLF
This journal is © The Royal Society of Chemistry 20xx
 Page 7 of 11 Journal
Please of
doMaterials Chemistry
not adjust marginsA

Journal of Materials Chemistry A Paper

View Article Online

DOI: 10.1039/D1TA02848K

Journal of Materials Chemistry A Accepted Manuscript

Published on 26 May 2021. Downloaded by Hokkaido Daigaku on 5/26/2021 1:59:31 PM.

Fig. 6 EIS of BZCYYb1711 and BZCYYb6211 cells without (black) and with (red) LSC-AFL under OCV condition. (a) and (b) are the spectra at 600 ℃, and (c)−(d) are the spectra at 500
℃. Circles are the observed data and solid lines are the fitting results with the equivalent circuit depicted in the inset of Fig. 6(c).

of both cells were similar (Figs. 6a and b). At 600 C, RHF of
BZCYYb1711 and BZCYYb6211 were 0.52 and 0.21 Ω cm2,
respectively, whereas RLF were 0.56 and 0.77 Ω cm2 at OCV,
respectively. Based on the reverse mode of the cathode reactions
on protonic solid oxide fuel cells, 49, 50 the anode reactions in P-
SOECs were roughly represented as shown in Fig. 9a. Step-1 is
the dissociative adsorption of water on the gas-electrode-
electrolyte triple phase boundary (TPB), Step-2 is the
electrochemical proton incorporation into BZCYYb electrolytes,
Step-3 is the electrochemical diffusion of oxygenic species in
LSCN8273, and Step-4 is the associative desorption of oxygen.51
Each elementary step can be given as follows:
Step-1: H2 O(g)⟶ O2− (TPB) + 2H+ (TPB)
Step-2: O2− (TPB)+2H+ (TPB)+h+ ⟶O− (TPB)+2H+ (ele) (2)
Step-3: O− (TPB)+h+ ⟶O(an)
Step-4: O(an)⟶1/2 O2 (g)
Here, “an” indicates that a species is on the anode surface, “TPB”
indicates species adsorbed at the electrode-gas-electrolyte triple
phase boundary, and “ele” indicates the species in the electrolyte.
Several authors have reported that P-SOECs based on Ba(Zr,
Ce, Y)O3− electrolytes exhibit two distinct semiarcs at
approximately 105–102 Hz and 102−10−1 Hz,30, 31, 45, 51 mainly due
to the polarization of the electrochemical reactions at the
anode/electrolyte interface. In general, a low-frequency semiarc
could be primarily related to the mass transfer on the anode, that
is, the surface diffusion or associative desorption of oxygenic
species on the anode (Step-3 and 4, respectively). The high-
frequency semiarc is probably associated with the charge transfer
at the TPB, that is, electrochemical proton incorporation (Step-
2).45, 49, 52 In fact, these general descriptions fit the features of RHF
and RLF in our cells. The RLF of BZCYYb1711 cells was
equivalent to that of BZCYYb6211 cells, indicating that the
same LSCN8273 anode purely contributed to RLF (Figs. 6a and
b). In contrast, the charge transfer kinetics at TPB must vary with
the electrolyte materials, such that the different RHF values
among both cells confirm the correlation of RHF with coupled
hole/proton transfer at TPB (Step-2).
Arrhenius plots of RHF−1 and RLF −1 under OCV indicated
activation energies of 0.70 and 1.04 eV, respectively, for
BZCYYb1711 (Figs. 7c and e), and 1.15 and 0.95 eV,
respectively, for BZCYYb6211 (Figs. 7d and f). The activation
energy of RLF was very close to the related energy (0.77−1.21 eV)
for oxide ion diffusion on cobaltite perovskite, 53 which proves
(1) the assignment of RLF to the oxygen diffusion on LSCN8273.
The activation energy of RHF was similar to those (0.51−1.00 eV)
of the corresponding resistance components for P-SOECs with
Ba(Zr, Ce, Y)O3−δ electrolytes. 31, 32
(3) EIS revealed that Ro, RHF, and RLF significantly decreased due
to the use of LSC-AFL in both BZCYYb1711 and
(4)
BZCYYb6211 cells. The impact of LSC-AFL on Ro was clearly
demonstrated by the Arrhenius plots as shown in Figs. 7a and b.
Although the activation energies of proton conduction remained
unchanged (0.37 and 0.28 eV for BZCYYb1711 and
BZCYYb6211, respectively), the values of Ro almost decreased
to half for BZCYYb1711 and BZCYYb6211, with the use of
LSC-AFL. For instance, Ro of BZCYYb1711 decreased from
0.30 to 0.12 Ω cm2 and that of BZCYYb6211 from 0.52 to 0.26
Ω cm2 at 600 C under OCV (Figs. 6a–b and Table S1). These
results imply that Ro, that is, ohmic loss, includes large
contributions of proton conduction near the anode/electrolyte
interface, and thus, LSC-AFL sufficiently increases the number
of mobile protons or conduction paths near the interface (Fig. 9b).
In contrast, the activation energy for RHF drastically decreased
with LSC-AFL (Figs. 7c and d). RHF of BZCYYb1711 decreased
with LSC from 0.52 to 0.40 Ω cm 2, and from 1.29 to 0.58 Ω cm2

This journal is © The Royal Society of Chemistry 20xx J. Name ., 2013, 00, 1 -3 | 7

Please do not adjust margins

 Journal
Please of
doMaterials Chemistry
not adjust marginsA Page 8 of 11

Paper Journal of Materials Chemistry A

View Article Online

DOI: 10.1039/D1TA02848K

Journal of Materials Chemistry A Accepted Manuscript

Published on 26 May 2021. Downloaded by Hokkaido Daigaku on 5/26/2021 1:59:31 PM.

Fig. 7 Arrhenius plots of ohmic resistances (Ro) and polarization resistances (RHF and RLF) of (a), (c), (e) BZCYYb1711 and (b), (d), (f) BZCYYb6211 cells at OCV condition, as determined
by equivalent circuit analysis. Black and red symbols show the cells without and with LSC-AFL, respectively.

at 600 and 500 C, respectively (Table S1), and the related
activation energy decreased from 0.70 to 0.44 eV with LSC (Fig.
7c). The reduction in RHF was more evident in BZCYYb6211
than BZCYYb1711 cells (Fig. 7d and Table S1). RHF decreased
from 0.21 to 0.06 and from 0.72 to 0.07 Ω cm2 at 600 and 500
C, respectively, by using LSC-AFL. Moreover, RHF of the
BZCYYb6211 cell with LSC-AFL exhibited a less-pronounced
temperature dependence, and thus, the related activation energy
was equal to 0.09 eV, which is one order of magnitude smaller
than the values of the cell without LSC-AFL. These results
indicate that LSC-AFL can significantly promote hole/proton
transfer to water adsorption at the TPB (Fig. 9b). It is warrant
noticing that the activation energy of RHF of BZCYYB6211 is
much smaller than that of BZCYYb1711 with LSC-AFL. This
must give insights to the functionality of AFL, and thus would
be addressed in next reports.
The changes in RLF with LSC were small in comparison to RHF
for both BZCYYb1711 and BZCYYb6211 cells (Figs. 7e and f).
From the Arrhenius plots of RLF−1 under OCV, the activation
energies slightly decreased from 1.04 and 0.95 eV to 0.88 and
0.79 eV for BZCYYb1711 and BZCYYb6211, respectively.
This indicates that the RLF is purely related to the surface kinetics
of the LSCN8273 anode.
Because the BZCYYb electrolyte is covered by a dense layer
of LSC (Fig. 3b), the anode reactions occur mainly near the gas-
AFL-electrode triple phase boundary (hereafter denoted as a-
TPB). This strongly suggests that LSC-AFL exhibits partial
proton conductivity under the anode conditions of steam
electrolysis, which is consistent with the recent report that a
La1−xSrxCoO3−δ series show minor proton conduction via
hydration under a relatively high pH2 atmosphere. 54 Based on
this, we ascribe LSC-AFL as a proton-electron-oxide ion triple
conducting phase.
The electrochemical proton incorporation (reaction (2)) can be
rewritten for the cell with AFL as follows:
O2− (a-TPB)+2H+ (a-TPB)+h+ ⟶ O− (a-TPB)+2H+ (AFL) (5)
EIS revealed that LSC significantly decreased the activation
energy of RHF (Figs. 7c and d), which indicates that LSC-AFL
involves a decrease in the energy barrier height of
electrochemical proton incorporation, rather than an increase in
the effective reaction area. Although the rate-determining step of
reaction (5) is still unclear, the activation energy of reaction (5)
must be smaller than that of reaction (2). EIS also confirmed that
Ro was sufficiently decreased by the aid of AFL. When
electrochemical proton incorporation is encouraged, the proton
concentration near the underlayer of the anode can be increased,
which may lead to a reduction in the proton-conducting
resistance.
To evaluate the activity of LSC as an anode, the BZCYYb6211
cell applied by LSC porous anode without AFL was also
constructed. The LSC anode cell exhibited a current of 0.65 and
0.14 A cm−2 at 600 and 500 ℃ under 1.3 V, which are similar to
the values of the corresponding cell using LSCN8273 anode
without AFL (Fig. S9). The impedance features of the former are
identical to the latter’s ones (Fig. S7c), confirming that the LSC
anode has similar activity as LSCN8273 anode. These results
proved that the effect of LSC-AFL does not rely on the activity
of LSC and thus AFL has inherent role to promote the
electrochemical proton incorporation.
Durability test
Finally, the long-term durability of the BZCYYb6211
electrolysis cell with LSC-AFL was examined by galvanostatic
electrolysis at 1 A cm−2 at 500 °C for 100 h (Fig. 8a). It is known
that Ba(Zr, Ce)O3 solid solutions can show a higher tolerance to
steam with increasing Zr content. As mentioned earlier,
BZCYYb6211 cells exhibited performances similar to those of
BZCYYb1711 cells with LSC-AFL despite the relatively high Zr

8 | J. Name., 2012, 00, 1-3 This journal is © The Royal Society of Chemistry 20xx

Please do not adjust margins

 Page 9 of 11 Journal
Please of
doMaterials Chemistry
not adjust marginsA
Journal of Materials Chemistry A
Published on 26 May 2021. Downloaded by Hokkaido Daigaku on 5/26/2021 1:59:31 PM.
Fig. 8 (a) Long-term durability tests of cell BZCYYb6211 with LSC-AFL in 20%−H2O/Air at
500 ℃. (b) Comparison of ohmic resistances (Ro) and polarization resistances (RHF and
RLF) before and after ~100 h operation, which was determined from the impedance
spectra measured under OCV condition (c).
content. Hence, BZCYYb6211 is very attractive as an electrolyte
for steam electrolysis cells operating under high H2O conditions.
The cell showed excellent durability with only a 1% increase in
cell voltage for 100 h. The EIS of the cells before and after the
long-term durability tests (Fig. 8c) confirmed that the changes in
the ohmic resistance and polarization resistance at high-
frequency were relatively small (Fig. 8b). Bulk resistance (Ro)
increased from 0.31 to 0.33 Ω cm 2, whereas RHF in high-
frequency region of 105−103 Hz slightly decreased from 0.09 to
0.08 Ω cm2 after 100 h of operation. RLF value in low-frequency
region of 103−10−1 Hz became 2.5 times higher than the value
before the durability test, indicating that the deterioration of cell
performances was mainly because of the degradation of the
anode material, rather than the electrolyte and AFL. Anyway,
these results indicate that LSC-AFL can persist for a long time
under the anode conditions of P-SOECs. There is a simultaneous
improvement in the ohmic resistance and interfacial charge
transfer resistances of P-SOECs.
The results clearly demonstrate that the LSC-AFL developed
in this study is significantly advantageous for use in P-SOECs to
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins
Paper
View Article Online
DOI: 10.1039/D1TA02848K
Journal of Materials Chemistry A Accepted Manuscript
Fig. 9 Schemes for the anode reactions of P-SOECs (a) without LSC-AFL and (b) with LSC-
AFL at the triple-phase boundary.
accelerate charge transfer to water adsorption at the TPB and
increase the number of mobile protons or conduction paths near
the anode/electrolyte interface, as shown in Fig. 9b. In
conclusion, an anode functional thin layer is a promising
technology for P-SOECs and offers an opportunity to explore
other active materials to improve the steam electrolysis
performances at intermediate temperatures.
Conclusions
P-SOECs with or without LSC-AFL were successfully fabricated
using BZCYYb1711 or BZCYYb6211 as the electrolyte and
LSCN8273 as the anode. Here, the novel LSC-AFL is
demonstrated to be a promising technology for P-SOECs owing
to the established connection between the anode/electrolyte.
BZCYYb6211 with LSC-AFL exhibited an excellent current of
1.22 A cm−2 at 600 °C at 1.3 V, despite the larger grain boundary
surfaces compared with BZCYYb1711. This result was
attributed to the decreased ohmic loss caused by the increased
number of mobile protons or conduction paths near the
anode/electrolyte interface and the polarization resistance at high
frequencies by promoting hole or proton transfer to water
adsorption at the anode-electrolyte-gas triple phase boundary.
The BZCYYb6211 cell showed a Faraday efficiency of ~70%
for H2 evolution and excellent stability with only ~1% increment
in cell voltage for 100 h. The current results offer an opportunity
J. Name ., 2013, 00, 1 -3 | 9
 Journal
Please of
doMaterials Chemistry
not adjust marginsA
Paper
to explore more active anode functional materials to improve the
electrochemical performances of steam electrolysis at
intermediate temperatures.
Conflicts of interest
There are no conflicts to declare.
Acknowledgements
This work was supported by the JST MIRAI “Realization of a
low carbon society, global issue” No. JPMJM17E8; JSPS
KAKENHI: Grant-in-Aid for Scientific Research (B), No.
Published on 26 May 2021. Downloaded by Hokkaido Daigaku on 5/26/2021 1:59:31 PM.
18H02066; and the “Nanotechnology Platform” program of
MEXT Japan.
Notes and references
1 R. Ma, B. Xu and X. Zhang, Catal. Today, 2019, 338, 18–30.
2 S. Chen, C. Pei and J. Gong, Energy Environ. Sci., 2019, 12,
3473–3495.
3 M. Ni, M. Leung and D. Leung, Int. J. Hydrogen Energy, 2008,
33, 2337–2354.
4 Y. Zheng, J. Wang, B. Yu, W. Zhang, J. Chen, J. Qiao and J.
Zhang, Chem. Soc. Rev., 2017, 46, 1427–1463.
5 A. Pandiyan, A. Uthayakumar, R. Subrayan, S. W. Cha and S.
B. Krishna Moorthy, Nanomater. Energy, 2019, 8, 2–22.
6 A. Hauch, S. D. Ebbesen, S. H. Jensen and M. Mogensen, J.
Mater. Chem., 2008, 18. 2331–2340.
7 X. Yang and J. T. S. Irvine, J. Mater. Chem., 2008, 18. 2349–
2354.
8 T. Matsui, R. Kishida, J. Y. Kim, H. Muroyama and K. Eguchi,
J. Electrochem. Soc., 2010, 157, B776–B781.
9 L. Lei, J. Zhang, Z. Yuan, J. Liu, M. Ni and F. Chen, Adv. Funct.
Mater., 2019, 29, 1903805.
10 N. Wang, S. Hinokuma, T. Ina, C. Zhu, H. Habazaki and Y.
Aoki, J. Mater. Chem. A, 2020, 8, 11043–11055.
11 C. Duan, J. Huang, N. Sullivan and R. O'Hayre, Appl. Phys.
Rev., 2020, 7, 011314.
12 C. Duan, R. Kee, H. Zhu, N. Sullivan, L. Zhu, L. Bian, D.
Jennings and R. O’Hayre, Nat. Energy, 2019, 4, 230–240.
13 J. LÜ, L. Wang, L. Fan, Y. Li, L. Dai and H. Guo, J. Rare Earths,
2008, 26, 505–510.
14 Z. Zhong, Solid State Ionics, 2007, 178, 213–220.
15 J. Lagaeva, D. Medvedev, A. Demin and P. Tsiakaras, J. Power
Sources, 2015, 278, 436–444.
16 J. Lin, L. Chen, T. Liu, C. Xia, C. Chen and Z. Zhan, J. Power
Sources, 2018, 374, 175–180.
17 K. Joong Yoon, M. Biswas, H. J. Kim, M. Park, J. Hong, H. Kim,
J. W. Son, J. H. Lee, B. K. Kim and H. W. Lee, Nano Energy,
2017, 36, 9–20.
18 K. Leonard, Y. Okuyama, Y. Takamura, Y. S. Lee, K. Miyazaki,
M. E. Ivanova, W. A. Meulenberg and H. Matsumoto, J.
Mater. Chem. A, 2018, 6, 19113–19124.
19 S. Choi, T. C. Davenport and S. M. Haile, Energy Environ. Sci.,
2019, 12, 206–215.
20 N. Wang, H. Toriumi, Y. Sato, C. Tang, T. Nakamura, K.
Amezawa, S. Kitano, H. Habazaki and Y. Aoki, ACS Appl.
Energy Mater., 2021, 4, 554–563.
21 P. Sawant, S. Varma, B. N. Wani and S. R. Bharadwaj, Int. J.
Hydrogen Energy, 2012, 37, 3848-3856.
22 E. Fabbri,A. Depifanio,E. Dibartolomeo,S. Licoccia and E.
Traversa, Solid State Ionics, 2008, 179, 558-564.
10 | J. Name., 2012, 00, 1-3
Please do not adjust margins
Page 10 of 11
Journal of Materials Chemistry A
23 L. Yang, S. Wang, K. Blinn, M. Liu, Z. Liu, Z. Cheng and M. Liu,
View Article Online
Science, 2009, 326, 126–129. DOI: 10.1039/D1TA02848K
24 J. Kim, A. Jun, O. Gwon, S. Yoo, M. Liu, J. Shin, T. H. Lim and G.
Kim, Nano Energy, 2018, 44, 121–126.
25 W. Wu, H. Ding, Y. Zhang, Y. Ding, P. Katiyar, P. K. Majumdar,
T. He and D. Ding, Adv. Sci., 2018, 5, 1800360.
Journal of Materials Chemistry A Accepted Manuscript
26 K. Katahira, Y. Kohchi, T. Shimura and H. Iwahara, Solid State
Ionics, 2000, 138, 91–98.
27 Y. Guo, Y. Lin, R. Ran and Z. Shao, J. Power Sources, 2009,
193, 400–407.
28 L. Lei, Z. Tao, X. Wang, J. P. Lemmon and F. Chen, J. Mater.
Chem. A, 2017, 5, 22945–22951.
29 D. Huan, N. Shi, L. Zhang, W. Tan, Y. Xie, W. Wang, C. Xia, R.
Peng and Y. Lu, ACS Appl. Mater. Interfaces, 2018, 10, 1761–
1770.
30 D. Huan, W. Wang, Y. Xie, N. Shi, Y. Wan, C. Xia, R. Peng and
Y. Lu, J. Mater. Chem. A, 2018, 6, 18508–18517.
31 W. Li, B. Guan, L. Ma, S. Hu, N. Zhang and X. Liu, J. Mater.
Chem. A, 2018, 6, 18057–18066.
32 S. Yang, Y. Wen, J. Zhang, Y. Lu, X. Ye and Z. Wen,
Electrochim. Acta, 2018, 267, 269–277.
33 W. Tang, H. Ding, W. Bian, W. Wu, W. Li, X. Liu, J. Y. Gomez,
C. Y. Regalado Vera, M. Zhou and D. Ding, J. Mater. Chem. A,
2020, 8, 14600–14608.
34 H. Ding, W. Wu, C. Jiang, Y. Ding, W. Bian, B. Hu, P. Singh, C.
J. Orme, L. Wang, Y. Zhang and D. Ding, Nat. Commun., 2020,
11, 1907.
35 W. Li, B. Guan, L. Ma, H. Tian and X. Liu, ACS Appl. Mater.
Interfaces, 2019, 11, 18323–18330.
36 S. Yang, S. Zhang, C. Sun, X. Ye and Z. Wen, ACS Appl. Mater.
Interfaces, 2018, 10, 42387–42396.
37 S. Rajendran, N. K. Thangavel, H. Ding, Y. Ding, D. Ding and
L. M. Reddy Arava, ACS Appl. Mater. Interfaces, 2020, 12,
38275–38284.
38 Y. Meng, J. Gao, H. Huang, M. Zou, J. Duffy, J. Tong and K. S.
Brinkman, J. Power Sources, 2019, 439, 227093.
39 C. Sun, S. Yang, Y. Lu, J. Wen, X. Ye and Z. Wen, J. Power
Sources, 2020, 449, 227498.
40 L. Lei, J. Zhang, R. Guan, J. Liu, F. Chen and Z. Tao, Energy
Conver. Manag., 2020, 218, 113044.
41 R. Murphy, Y. Zhou, L. Zhang, L. Soule, W. Zhang, Y. Chen and
M. Liu, Adv. Funct. Mater., 2020, 30, 2002265.
42 E. Vøllestad, R. Strandbakke, M. Tarach, D. Catalan Martinez,
M. L. Fontaine, D. Beeaff, D. R. Clark, J. M. Serra and T.
Norby, Nat. Mater., 2019, 18, 752–759.
43 S. M. Babiniec, S. Ricote and N. P. Sullivan, Int. J. Hydrogen
Energy, 2015, 40, 9278–9286.
44 M. Dippon, S. M. Babiniec, H. Ding, S. Ricote and N. P.
Sullivan, Solid State Ionics, 2016, 286, 117–121.
45 H. Toriumi, T. Kobayashi, S. Hinokuma, T. Ina, T. Nakamura,
K. Amezawa, C. Zhu, H. Habazaki and Y. Aoki, Inorg. Chem.
Front., 2019, 6, 1587–1597.
46 S. Jeong, T. Yamaguchi, M. Okamoto, C. Zhu, H. Habazaki, M.
Nagayama and Y. Aoki, ACS Appl. Energy Mater., 2020, 3,
1222–1234.
47 R. J. Kee, H. Zhu, B. W. Hildenbrand, E. Vøllestad, M. D.
Sanders, R. P. O’Hayre, J. Electrochem. Soc., 2013, 160,
F290−F300.
48 S. Jeong, T. Yamaguchi, M. Okamoto, C. Zhu, H. Habazaki, M.
Nagayama and Y. Aoki, ACS Appl. Energy Mater., 2020, 3,
1222–1234.
49 F. He, T. Wu, R. Peng and C. Xia, J. Power Sources, 2009, 194,
263-268.
50 E. Fabbri, L. Bi, D. Pergolesi and E. Traversa, Adv. Mater.,
2012, 24, 195-208.
51 T. Kobayashi, K. Kuroda, S. Jeong, H. Kwon, C. Zhu, H.
Habazaki and Y. Aoki, J. Electrochem. Soc., 2018, 165, F342–
F349.
This journal is © The Royal Society of Chemistry 20xx
 Page 11 of 11 Journal
Please of
doMaterials Chemistry
not adjust marginsA

Journal of Materials Chemistry A Paper

52 Y. Zhang, D. Xie, B. Chi, J. Pu, J. Li and D. Yan, Asia-Pac. J. View Article Online
Chem. Eng., 2019, 14, E2322. DOI: 10.1039/D1TA02848K
53 M. V. Ananyev, N. M. Porotnikova and E. K. Kurumchin, Solid
State Ionics, 2019, 341, 115052.
54 D. Han, Y. Okumura, Y. Nose and T. Uda, Solid State Ionics,
2010, 181, 1601-1606.

Journal of Materials Chemistry A Accepted Manuscript

Published on 26 May 2021. Downloaded by Hokkaido Daigaku on 5/26/2021 1:59:31 PM.

This journal is © The Royal Society of Chemistry 20xx J. Name ., 2013, 00, 1 -3 | 11

Please do not adjust margins

 View Article Online
View Journal

Journal of
Materials Chemistry A
Materials for energy and sustainability

Accepted Manuscript

This article can be cited before page numbers have been issued, to do this please use: R. Zhu, H. Yang, L.
Fadillah, Z. Xiong, D. Kowalski, C. Zhu, S. kitano, Y. Aoki and H. Habazaki, J. Mater. Chem. A, 2021, DOI:
10.1039/D1TA02491D.
Volume 6
Number 12
28 March 2018
This is an Accepted Manuscript, which has been through the
Pages 4883-5230
Royal Society of Chemistry peer review process and has been
Journal of accepted for publication.
Materials Chemistry A
Materials for energy and sustainability
rsc.li/materials-a
Accepted Manuscripts are published online shortly after acceptance,
before technical editing, formatting and proof reading. Using this free
service, authors can make their results available to the community, in
citable form, before we publish the edited article. We will replace this
Accepted Manuscript with the edited and formatted Advance Article as
soon as it is available.

You can find more information about Accepted Manuscripts in the

Information for Authors.

Please note that technical editing may introduce minor changes to the
text and/or graphics, which may alter content. The journal’s standard
ISSN 2050-7488 Terms & Conditions and the Ethical guidelines still apply. In no event
COMMUNICATION
Zhenhai Wen et al.
An electrochemically neutralized energy-assisted low-cost
shall the Royal Society of Chemistry be held responsible for any errors
acid-alkaline electrolyzer for energy-saving electrolysis
hydrogen generation
or omissions in this Accepted Manuscript or any consequences arising
from the use of any information it contains.

rsc.li/materials-a
 Page 1 of 11 Journal of Materials Chemistry A

View Article Online

Received 00th January 20xx,
Accepted 00th January 20xx
DOI: 10.1039/x0xx00000x
DOI: 10.1039/D1TA02491D
Lithiophilic carbon scroll as Li metal host with low tortuosity
design and “Dead Li” self-cleaning capability

Journal of Materials Chemistry A Accepted Manuscript

Ruijie Zhu,a Huijun Yang,b Laras Fadillah,a Zetao Xiong,a Damian Kowalski, d,e Chunyu Zhu,c, d * Sho
Kitano,a,d Yoshitaka Aoki,a,d Hiroki Habazaki. a,d

ABSTRACT On the way to achieve practical lithium (Li) metal anode for next-generation batteries, the formation and
accumulation of inactive “Dead Li” is an unavoidable issue. The accumulation of “Dead Li” leads to increased internal mass-
transfer resistance which seriously deteriorates the performance of Li metal batteries during long-term cycling. In this study,
by accommodating Li metal into a copper oxide coated carbon scroll host with a vertically aligned framework which
Published on 14 May 2021. Downloaded by Hokkaido Daigaku on 5/18/2021 11:22:28 AM.

possesses a unique low-tortuosity structure, the cyclic stability of Li anode can be significantly improved. It is demonstrated
that the mass-transfer resistance and the concentration polarization near the Li metal surface can be greatly alleviated by
using this low-tortuosity anode structure design. “Dead Li” that is formed on the electrode’s surface can automatically fall
into the inner tunnel of the carbon host, endowing the anode with the capability of “Dead Li” self-cleaning. As a result, our
new Li electrode can remain electrochemically active even after 1000 h in a symmetric cell measurement at 1 mA cm-2 to 1
mA h for 500 cycles. The as-reported structure design of Li anode in this work is compatible with most of the modification
technologies that have been applied to conventional Li foil electrodes, making this new Li anode with a great potential to be
applied in subsequent Li anode researches.

1. Introduction
Lithium (Li) metal is widely considered to be the most promising
anode material for developing high energy density batteries. In
comparison with other anode candidates, Li metal has the lowest
negative potential (E° (Li+/Li) =-3.04 V vs standard hydrogen
electrode (SHE)),1 by which the nominal voltage of batteries can be
greatly increased. Moreover, as benefiting from its ultra-high
capacity (3860 mA h g-1), the use of Li metal anode will significantly
reduce the overall weight of batteries, so as to meet the requirement
for high energy density.2
To achieve viable Li metal anodes, many roadblocks are still lying
in front of us. Firstly, the large change in volume could be virtually
infinite, because Li metal itself is a “no-host” electrode,
electrodeposition of Li metal will cause a huge geometric
deformation of the electrode.3 Secondly, on the surface of Li metal,
heterogeneous unstable solid-electrolyte-interphase (SEI) that
generates from the decomposition of electrolyte often fails to
protect Li metal from unceasing side reactions.4,5 The SEI layer
cannot withstand the deformation of the electrode, thereby the
continuous consumption of electrolyte and Li metal occur along with
the Li electrodeposition-dissolution process. Thirdly, the growth of
dendritic Li crystal increases the risk of internal short circuit, and the
thermal runaway may cause severe accidents.6 Finally, being broken
from the large Li crystals and subsequently wrapped by SEI, some
unreacted metallic Li will be separated from electronically
conductive substrate, resulting in the formation of inactive Li metal
a. Graduate School of Chemical Sciences and Engineering, Hokkaido University,
Sapporo, Hokkaido 060-8628, Japan
b. National Institute of Advanced Industrial Science and Technology (AIST), 1-1-1,
Umezono, Tsukuba 305-8568 Japan.
c. School of Electrical and Power Engineering, China University of Mining and
Technology, Xuzhou, 221116, China
d. Division of Applied Chemistry, Faculty of Engineering, Hokkaido University,
Sapporo 060-8628, Japan.
e. Faculty of Chemistry and Biological and Chemical Research Centre, University of
Warsaw, Zwirki i Wigury 101, 02-089 Warsaw, Poland
† * E-mail: zcyls@cumt.edu.cn,
Electronic Supplementary Information (ESI) available: [details of any supplementary
information available should be included here]. See DOI: 10.1039/x0xx00000x
lumps which are also known as “Dead Li”.7, 8 The formation of “Dead
Li” reduces the reversibility of batteries, so that excessive Li metal
and electrolyte must be added to batteries to achieve expected
service life.
Current researches for optimizing Li metal anode are focusing on
the following strategies: 1) accommodating Li metal in conductive 3D
hosts, in order to adapt the volume change of electrode9-14; 2)
designing lithiophilic substrates for seeding uniform nucleation of Li
metal, in order to eliminate the dendritic crystal growth;15-18 3)
adjusting the concentration or composition of electrolytes to form
“dense Li deposition” or stable SEI, in order to improve the
Coulombic Efficiency (CE) during electrochemical cycling;19-21 4)
constructing artificial SEI with high ionic conductivity and high
modulus on the surface of Li metal, in order to improve the interface
stability at the metal-electrolyte interface.22-26 Thanks to these
efforts made by many researchers, the CE of current Li metal
batteries can readily achieve a value higher than 99%.27 But what is
interesting is that only very few studies about “Dead Li” are involved,
and most studies have evaded this issue. It is confirmed that the
accumulation of “Dead Li” will severely increase the concentration
polarization near the Li metal surface and block the Li+ flux, which
negatively influence the already adopted Li metal protection
strategies and inevitably damage the performance of the cell.28,29
Fang et al. has proved that by depositing Li crystals with granular size
and less tortuosity the trend of forming “Dead Li” can be reduced,
but inactive Li will generate in nearly all cases of Li metal batteries.30
Valid strategies for alleviating the influence from “Dead Li”
accumulation still worth studying.
In this study, we innovate a lithiophilic carbon scroll (LCS), which
consists of vertically aligned carbon fibers that are decorated with
lithiophilic copper oxide (CuOx) nanoparticles (NPs). The LCS is used
as host for Li metal to develop Li@LCS composite electrode. The
carbon scroll host with vertically aligned carbon fibers provides a
channel structure which can accelerate the diffusion of Li+.
Considering that the transporting behaviour of ion in the porous
electrode network can be described by electrode tortuosity, which is
a parameter for describing that how the geometric structure of the
electrode hinders ion diffusion. Hence, we consider the LCS skeleton
 Journal of Materials Chemistry A
ARTICLE
with channel structure which has no obstruction on the way of ion-
transport can be called a low-tortuosity electrode. On one hand, with
the synergetic effect between lithiophilic CuOx and low-tortuosity
structure, Li metal can be deposited and accommodated in the
carbon matrix with low over-potential and suppressed dendrite
growth. On the other hand, the inactive Li lumps that are
Published on 14 May 2021. Downloaded by Hokkaido Daigaku on 5/18/2021 11:22:28 AM.
Fig. 1 Schematic illustration for the design concept of vertically aligned
lithiophilic carbon scroll host.
formed on the surface of the carbon host will fall into the inner
tunnel structure, which plays an important role as a Li reservoir. In
this way, the active sites for Li deposition can be always exposed
during long-term cycling, as depicted in Fig. 1. As a comparison,
conventional Li metal foil with planar structure is easily covered by
“Dead Li”, which obstructs the ion-transport and severely damages
the performance of battery. As a result, the novel low-tortuosity
structured Li@LCS electrode can eliminate most of the negative
influences from the accumulation of “Dead Li”, which is evidenced by
the reduced concentration polarization and over-potential. Even
after 500 cycles in a symmetric cell measurement at 1 mA cm-2 to 1
mA h cm-2 for 1000 h, the Li@LCS electrode can still maintain a very
low hysteresis voltage of 40 mV vs Li+/Li. In a Li metal|LiCoO2 full cell,
Li@LCS anode ensures the cell capacity hardly be affected by mass
transfer resistance as-caused by “Dead Li” accumulation. In contrast,
planar Li foil is quickly deactivated due to the influence of “Dead Li”
accumulation, by which the cell shows a low-capacity retention.
2. Results and discussion
Fig. 2 Schematic diagram for the preparation of low-tortuosity designed vertically carbon matrix and the low-tortuosity designed Li metal anode (Li@LCS).
Please do not adjust margins
Page 2 of 11
Journal Name
2.1 Preparation and material properties of the LCS framework and Li
View Article Online
anodes DOI: 10.1039/D1TA02491D
The LCS was prepared through a cotton self-template method. As
introduced in Fig. 2, a cotton sheet with one-dimensionally aligned
fibers was rolled to a scroll, which had a naturally anisotropic
Journal of Materials Chemistry A Accepted Manuscript
structure with vertically aligned fibers. Subsequently, the cotton
scroll was absorbed with the reported “variant of Schweizer’s
reagent”.31 The above cotton scroll soaked with [Cu(NH3)4]2+ was
pyrolyzed under Ar flow, and the carbon scroll decorated with Cu NPs
(CS@Cu) was obtained. The sample was further calcined under air
atmosphere at 250 ℃ to make the oxidation of Cu NPs. With the
presence of CuOX species, which were known as the lithiophilic
species,9,33 Li metal could be easily infused into the carbon scroll,
forming the Li@LCS composite electrode. In comparison to the
conventional isotropic carbon host for Li metal, the anisotropic
carbon scroll, which was consisted of vertically aligned carbon fibers,
endowed the Li@LCS electrode with a low-tortuosity characteristic,
by which the Li+ transport would be unimpeded.
The structural details of the vertically aligned carbon scroll could
be seen in Fig. 3. Fig. 3a-3c present the SEM images of CS@Cu, which
has a unique anisotropic structure with vertically arranged carbon
fibers, resulting in vertical tunnels of carbon scroll. Due to the
decoration of Cu NPs, the colour of CS@Cu is black charcoal mixed
with red copper. Thanks to the strong coordination between
cellulose and [Cu(NH3)4]2+, by which the Cu source could be firmly
fixed on the fibers surface during heating, the Cu NPs generated from
the decomposition of “variant of Schweizer’s reagent” are uniform
and fine as shown in Fig. 3b. After the oxidation treatment for
preparing LCS, the anisotropic structure of carbon scroll was
maintained (Fig. 3d), while the colour of LCS was changed to black.
EDS spectra confirmed changes in oxygen (O) content of the NPs. A
fine Cu NPs decorated carbon substate is shown in Fig. 3e. Mole ratio
of O in this region was calculated to be 3.2%, indicating that there is
only a small amount of natural oxides layer on the Cu NPs. The Cu
NPs were oxidized to larger CuOx particles after treatment, resulting
in the formation of a lithiophilic layer. XRD analysis confirms the
composition changes during the preparation, which clearly shows
the peaks of Cu (Fig. 3g) that were converted to the peaks of Cu2O
and CuO (Fig. 3h). It is worth pointing out that both CuO and Cu2O
have been confirmed to have lithiophilicity.9,34 The simultaneous
presence of peaks that are corresponded to Cu2O and CuO in XPS
spectra corroborated composition of the LCS (Fig. 3i, 3j). N2
adsorption behaviour of the samples is shown in Fig. 3k and Fig. S2
 Page 3 of 11 Journal of Materials Chemistry A

Journal Name ARTICLE

View Article Online

DOI: 10.1039/D1TA02491D

Journal of Materials Chemistry A Accepted Manuscript

Published on 14 May 2021. Downloaded by Hokkaido Daigaku on 5/18/2021 11:22:28 AM.

Fig. 3 Material properties of CS@Cu and LCS. SEM images of (a) Top view and (b) a single fiber of the CS@Cu (The inset figure is an optical photo of the CS@Cu
disk). Cross sectional SEM observation of (c) CS@Cu. (d) Top view of the LCS (The inset figure is an optical photo of the LCS disk). High magnification SEM and
EDS images of (e) a single fiber decorated by fine Cu nanoparticles of CS@Cu and (f) a single fiber decorated by CuOx nanoparticles of LCS disk. XRD patterns
of (g) CS@Cu and (h) LCS. (i) Cu 2p and (j) O 1s XPS spectra for LCS. (k) N2 adsorption isotherms of CS@Cu and LCS. (l) Pore size distribution of CS@Cu (red) and
LCS (black). The material properties for pristine CS are shown in Fig S2 and S3.

Both the CS@Cu and LCS samples exhibit large specific surface area
(SSA) of 142.9 m2 g-1 and 185.1 m2 g-1, respectively. In contrast, the
pristine carbon scroll without coating presents a very low SSA of 1.6
m2 g-1. The increase in SSA for the Cu decorated carbon scrolls could
be attributed to the vigorous pyrolysis of [Cu(NH3)4]2+ absorbed
cotton and the reduction formation of Cu NPs, resulting in the
formation of many nanopores. As shown in Fig. S4, Cu NPs were
embedded in the nanopores, however, after oxidization treatment,
the NPs were squeezed out of the pores due to the volume expansion
from Cu to CuOx and the oxidizing consumption of carbon, resulting
in a further SSA increase. The pore-size distribution was calculated
by the BJH method as shown in Fig. 3l, indicating the presence of
numerous mesopores in the range of around 2 nm to 10 nm. At the
same time, in order to understand the low-tortuosity characteristics
of LCS framework, we compared the difference in tortuosity between
LCS and other 3D hosts (Fig. S5-Fig. S8). The relevant results are
summarized in Table S1 and Table S2. The tortuosity of conventional
carbon cloth is about 6-fold that of LCS, while the tortuosity of
horizontally aligned carbon fiber is around 8-fold that of LCS. This
shows that LCS can greatly alleviate the hindrance of Li-ion
transporting, which makes it a low-tortuosity host.
The lithiophilicity of LCS as-endowed by CuOx ensures that the
carbon scroll host could be directly used to host Li metal. By
employing the commonly used melting-infusion method,35 molten Li
could gradually flow into the carbon skeleton and occupy the empty
spaces between the fibers (Fig. 4a), thus forming the Li@LCS
composite electrode. The surface of the composite electrode

Please do not adjust margins

 Journal of Materials Chemistry A
ARTICLE
exhibits exposed fibers, as shown in Fig. 4b. This is due to the
influence of gravity during the cooling process, in which the liquid Li
sank to the bottom of the carbon skeleton. Fig. 4c is a 3D
reconstructed image obtained by using a laser scanning confocal
microscope (LSCM). It can be seen that the upright carbon fibers
maintains a large number of channels on the surface of the electrode,
which allows Li-ion can transport smoothly at the electrode surface.
A larger magnification SEM image is shown in Fig. S9 to observe the
structural details of the electrode surface. The thickness of the
Li@LCS composite electrode is around 1300 μm (Fig. 4d). A careful
observation of the cross-section of Li@LCS in Fig. 4e indicates that
the vertically aligned fibers are wrapped inside Li metal.
Published on 14 May 2021. Downloaded by Hokkaido Daigaku on 5/18/2021 11:22:28 AM.
Fig. 4 Morphology and microstructure of the Li@LCS composite electrode.
Optical photos (a) for illustrating the melting-infusion process for preparing
Li@LCS composite electrode. (b) SEM image and (c) LSCM image of Li@LCS
composite electrode from top view. (d, e) SEM images of Li@LCS composite
electrode obtained from cross sectional observation.
2.2 Electrochemical properties as measured in symmetric cells and
GITT technique
To grasp the uniqueness of the Li@LCS composite electrode in
improving the performance of Li anode, the electrochemical cycling
performance of symmetric cells was firstly investigated. The
electrolyte used in symmetric cell test was 1M LiTFSI in DOL+DME co-
solvent, which is a commonly employed electrolyte for Li-S battery.
As can be seen in Fig. 5a, the Li@LCS | Li@LCS symmetric cell could
maintain a stable cycling performance of at least 500 cycles at a
current density of 1 mA cm-2 with an areal capacity of 1 mA h cm-2.
Here, we stopped the test at the 500th cycle. As shown in Fig. 5b, the
Li@LCS symmetric cell could maintain almost the same hysteresis
voltage during the 100-400 cycles, with an approximate value of 40
mV. As a comparison, the hysteresis voltage of the Li foil symmetric
cell at the 100th cycle was as large as 61 mV. It is also seen that the
overpotential of Li foil cell was significantly higher than the Li@LCS
cell during the whole cycles. Theoretically, the overpotential of a cell
is mainly caused by three parts, namely the ohmic polarization, the
activation polarization, and the concentration polarization.
Please do not adjust margins
Page 4 of 11
Journal Name
Considering that for the symmetric Li metal cells, View
the Article
activation
Online
polarization mainly comes from the followingDOI:
reactions:
10.1039/D1TA02491D
  
𝐋𝐢 + + 𝐞 ― 𝐋𝐢𝟎 (Reduction) (1)
𝐋𝐢𝟎
𝐋𝐢 + + 𝐞 ― (Oxidation) (2)
Journal of Materials Chemistry A Accepted Manuscript
Therefore, when the concentration polarization is eliminated, the
overpotential difference of the symmetric cells should be only
controlled by the ohmic resistance. It is known that the
concentration polarization on the surface of Li metal is mainly from
the slowing down of Li+ diffusion caused by the “Dead Li”
accumulation.29 Based on this point of view, in order to reveal the
influence that is from the concentration gradient of Li+ as-caused by
“Dead Li”, a galvanostatic intermittent titration technique (GITT) was
employed. In this way, the difference in ohmic polarization between
different symmetric cells can be revealed. At the same time, the
influence from “Dead Li” can also be evaluated. As shown in Fig. 5c,
the GITT cycle was obtained after the 500th cycle of Li@LCS
symmetric cell corresponding to Fig. 5a. The GITT measurement was
conducted by charging (or discharging) at 1 mA cm-2 for a short time,
which subsequently had a rest for a certain time that was enough for
the completion of Li+ diffusion on the electrode surface. In this
manner, the mass-transfer resistance could be eliminated, and the
Li+ concentration near the surface could return to the initial state at
each intermittent.36 This capacity reached a value of 1 mA h cm-2. In
this way, we were able to compare the Li-ion concentration
polarization of different Li metal electrodes, and to have a deep
insight into the role of the LCS skeleton.
Fig. 5 Electrochemical performance of Li@LCS in ethe r-based LiTFSI
electrolyte. (a) Plating/stripping curves in symmetric cell test at a current
density of 1 mA cm-2 with a capacity of 1 mA h cm-2. (b) Voltage profile of
Li@LCS symmetric cell and Li foil symmetric cell at the 100th, 200th, 300th,
400th cycle. GITT measurement of (c) Li@LCS symmetric cell and (d) Li foil
symmetric cell (corresponding to Fig. 4a). Here, the GITT measurement was
performed after the first 500 cycles (100 cycles for Li foil).
In view of that the Li+ concentration near the electrode surface is
strongly affected by ion mobility and diffusion coefficient. High mass
transfer resistance can influence the diffusion coefficient near the
electrode surface, leading to a significant Li+ concentration gradient.
In order to understand the changes in the value and trend of the
time-dependent overpotential after which the mass transfer
resistance was eliminated, a simulation model was employed to
describe the relationship between the overpotential and Li-ion
concentration distribution on the electrode surface. Hence, we used
the Butler-Volmer equation which as modified by Chen et al. as the
 Page 5 of 11 Journal of Materials Chemistry A

Journal Name ARTICLE

electro-deposition model for Li metal.29 It is reported that when the cLi（0 surf) (1- )F c  F
applied current density is not very high (such as 1 mA cm-2), the j =j0 [ exp(  ) - Li（surf） exp(  )] View Article Online
(Oxidation) (4)
cLi0 (bulk ) RT cLi (bulk ) RT DOI: 10.1039/D1TA02491D
modified Butler-Volmer model is more suitable than the
conventional Marcus-Hush model.37 The modified equations are Here, j refers to current density and j0 refers to exchange current
shown as below: density, and β is symmetry factor. cLi+(bulk) and cLi0(bulk) are the bulk
concentration of Li+ and metallic Li, while cLi+(surf) and cLi0(surf) are the

Journal of Materials Chemistry A Accepted Manuscript

cLi（ surf） F c 0 (1- )F
j =j0 [ exp(  )- Li（surf） exp(  )] (Reduction) (3) surface concentration of Li + and metallic Li, respectively. η is
cLi (bulk ) RT cLi0 (bulk ) RT overpotential and R, T, F are the ideal gas constant, temperature,
Faradays’ constant, respectively. Relationship between the applied
Published on 14 May 2021. Downloaded by Hokkaido Daigaku on 5/18/2021 11:22:28 AM.

Fig. 6 (a) Schematic illustration for explaining the potential and Li+ concentration profiles (upper plots) near the electrode surface at initial stage (left), later
cycles (mid) and GITT cycle (right), and the corresponding Li-ion concentration distribution near the electrode surface (bottom illustration) in a Li foil symmetric
cell. (b) Schematic illustration of the electrode morphologies at initial stage (left) and later cycles (mid), in a Li foil symmetric cell, corresponding to scheme
(a). (c) Schematic illustration for explaining the voltage and Li-ion concentration profiles (upper plots) near the electrode surface at initial stage (left), later
cycles (mid) and GITT cycle (right), and the corresponding Li-ion concentration distribution near the electrode surface (bottom illustration), in a Li@LCS
symmetric cell. (d) Schematic illustration of the electrode morphologies at initial stage (left) and later cycles (mid), in a Li@LCS symmetric cell, corresponding
to scheme (c). Potential response of the cell and Li-ion concentration at the reducing side (Li+ → Li0) were plotted vs. time in a half cycle. Deep-blue, cyan-blue
and red zones shown in the bottom of schematic (a) and (c) referred to low Li+ concentration gradient, moderate Li+ concentration gradient and high Li+
concentration gradient, respectively.

current, overpotential and Li-species concentration distribution at
non-equilibrium conditions is shown intuitively through those
equations. Moreover, j0 can be expressed by the following equation:
j0 = γFk 0eff [cLi (bulk ) ](1  ) [cLi0 (bulk ) ]
In which, γ is a parameter representing the roughness of electrode
surface, and keff0 is an effective heterogeneous rate constant. Both γ
and keff0 are depending upon the electrode morphology, namely, the
morphology of deposited Li metal.
The time-dependent overpotential in a Li metal symmetric cell is
influenced mainly by two aspects, i.e., the Li+ concentration gradient
on the surface of electrode and the morphology of the electrode
surface. Hence, we summarized the possible situations of Li metal
symmetric cells that were tested by us, based on the model given by
Chen et al,29 As shown in Fig. 6(a), when the cell is cycled at an early
stage, the electrode surface is not covered by “Dead Li”. Once a
current is applied to the cell, it will cause a significant concentration
(5) polarization on the electrode surface at the initial time, and a high-
value initial overpotential will be observed. Then, the Li+
concentration gradient will gradually recover and become stable,
resulting in a relatively low overpotential. Therefore, the time-
dependent overpotential of a Li foil | Li foil symmetric cell always
shows a “peak” shaped profile at the early stage (Fig. 6(a) left). After
being cycled for a long-time, the electrode surface will be covered by
“Dead Li”, which will cause a significant Li+ concentration gradient
throughout the cycles, because the diffusion of Li+ on the electrode
surface is hindered. Hence, an arc-shaped potential curve is observed

Please do not adjust margins

 Journal of Materials Chemistry A
ARTICLE
(Fig. 6(a) middle). In the GITT cycle, the influence from “Dead Li”
during mass-transferring is eliminated, and the overpotential profile
will return to the early-stage status. Here, the overpotential
difference is mainly caused by “Dead Li”, which is represented as η
(Dead Li caused) in the figure (Fig. 6(a) right). The influence from the
morphology of electrode surface is shown in Fig. 6(b). When the cell
is cycled at the early stage, the growth of Li dendrites will cause a
gradually changed γkeff0, hence the potential profile in this process
presents a change of shape from “peak” to “concave arc”. But when
the cell is cycled at the later stage, the value of γkeff0 will become
relatively stable due to the formation of “Dead Li” layer, thus the
shape of the overpotential curve is mainly governed by Li+
concentration variation.
In the case of Li@LCS electrode, the low-tortuosity structure
allows Li+ concentration gradient near the electrode surface to
Published on 14 May 2021. Downloaded by Hokkaido Daigaku on 5/18/2021 11:22:28 AM.
quickly reach equilibrium after applying current, as depicted in Fig.
6(c). The influence from the morphology of Li electroplating/
stripping is depicted in Fig. 6(d). The dendrites-suppressed Li
deposition on Li@LCS (shown in Fig. S10, S11 and S13) will result in a
stable γkeff0, therefore the shape of potential curve of Li@LCS
symmetric cell should be similar to the “arc-shaped curve” that is
shown in Fig. 6(a). However, as benefiting from the low-tortuosity
structure, the potential of Li@LCS cell will be lower than Li foil cell.
Finally, if the influence of “Dead Li” accumulation is negligible in the
case of Li@LCS symmetric cell, the potential curve after long-term
cycling and the potential curve of GITT cycle should be consistent
with the potential curve of their early stage (Fig. 6(c) right).
The experimental results confirmed our description above. After
eliminating the influence from concentration polarization near the
electrode surface, the Li@LCS symmetric cell still shows an “arc-
plateau” shaped voltage profile, which is similar with the normal
voltage profile as shown in Fig. 5b. The overpotential in every
“intermittent” is around 16.5 mV, which is close to the over-potential
in the initial cycles (~ 20 mV). Although the influence from Li+ mass-
transfer resistance is eliminated in the Li@LCS symmetric cell (during
GITT cycle), its overpotential does not show large difference in
comparison with the overpotential in normal cycle. This result
indicates that the Li@LCS electrode is only affected slightly by the
“Dead Li" accumulation during cycling. Furthermore, it is reasonable
to consider that the diffusion coefficient of Li+ near the surface of
Li@LCS electrode and the diffusion coefficient in separator
(containing electrolyte) are almost the same. At the same time,
although overpotential of the Li@LCS cell is not influenced by Li+
mass-transfer resistance, its “arc-plateau” voltage profile does not
change in the GITT cycle, demonstrating that Li deposition on Li@LCS
is consistently uniform,38 because the value of γkeff0 does not change
so much. As a comparison, the GITT cycle for the Li foil symmetric cell
was tested after the 100th cycle as shown in Fig. 5d, which indicates
an overpotential of 16.0 mV with a “peak” shaped voltage curve. The
Li foil symmetric cell has a “Peak” in its voltage profile, and the
overpotential drops after the initial concentration gradient achieving
an equilibrium, indicating that the time-dependent overpotential in
this case is greatly governed by γkeff0.29,39 In other words, Li metal
crystals that are electrodeposited on Li foil will grow in the form of
dendrites. One thing has to be aware of is that the Li+ diffusion
coefficient of ether-based LiTFSI electrolyte is much higher than the
diffusion coefficient of carbonate-based LiPF6 electrolyte,40,41 hence
the “Peak” in our work is not as pronounced as those reported in the
mentioned literatures.
Electrochemical impedance spectroscopy (EIS) analysis further
confirms the abovementioned conclusions. We selected the
Please do not adjust margins
Page 6 of 11
Journal Name
commonly accepted equivalent circuit to describe the information
View Article Online
as-included in EIS. Here, Rsol is the electrolyte resistance,
DOI: RSEI is the
10.1039/D1TA02491D
resistance of SEI and the RCT is the charge transfer resistance. In Fig.
7a, the Li@LCS symmetric cell shows smaller values of RSEI and Rct
than the Li foil symmetric cell before cycling. This is consistent with
the conclusion that the Li foil symmetric cell has high overpotential
Journal of Materials Chemistry A Accepted Manuscript
in the initial several cycles. After cycling, RSEI and RCT of both Li@LCS
cell and Li foil cell are reduced (Fig. 7b), due to the reason that the
SEI generated in the ether-based LiTFSI electrolyte has a relatively
good stability and high Li+ conductivity.42,43 Moreover, both cells
show an overall AC impedance of about tens of ohms, which is
consistent with the overpotential value in the symmetric cells test.
Detailed observations were provided by directly analyse the
electrodes after cycling. As
Fig. 7 EIS of Li@LCS symmetric cell and Li foil symmetric cell (a) before and (b)
after GITT measurement corresponding to Fig. 5. SEM images of (c, d) Li@LCS
composite electrode and (e) Li foil after cycling, corresponding to Fig. 5. (f)
Schematic illustration for understanding the accumulation behavior of “Dead
Li”on different Li metal electrodes.
shown in Fig. 7c, after 1000 hours cycling, porous Li metal clusters on
the surface of the Li@LCS electrode were observed, which were
generally considered to be the omen of inactive Li metal
generation.28 This kind of morphology indicates that during long-
term Li metal plating/stripping cycling, break of the Li crystal results
in incomplete stripping of Li and the generation of “Dead Li”. But
surprisingly, those porous Li clusters and Li fragments (“Dead Li”) did
not cover the entire electrode surface. It is observed that most of the
vertically aligned fibers are still exposed on the surface of the
electrode, and importantly, the tunnels between the fibers are not
blocked, indicating that this electrode still has enough active sites for
Li deposition and the Li+ transport path is unhindered. Additionally,
the cross-sectional morphology of the cycled Li@LCS electrode is
shown in Fig. 7d. A compact and dense Li metal layer that is formed
by accumulation of the fallen Li metal blocks can be observed under
the yellow line. This observation result is similar with many reported
Li metal non-planar 3D hosts,44,45 indicating that LCS does have the
capability to accommodate cycled Li metal into its 3D space, by which
the losing of electrochemical activity of Li clusters is avoided.46 In
contrast, the planar Li foil is covered by highly porous “Dead Li” on
the surface after cycling, as shown in Fig. 7e. Fig. 7f illustrates the
schemes for the formation and accumulation patterns of “Dead Li”
during long-term cycling. “Dead Li” is inevitably formed on both
Li@LCS and Li foil. In the case of Li@LCS electrode, the deposited Li
can be accommodated inside the spaces of the carbon host, at the
same time, “Dead Li” will sink into the tunnels under the effect of
gravity. Therefore, the LCS host has a unique “Dead Li” self-cleaning
capability, and the electrode surface is always active for Li deposition.
In addition, the unique structure of LCS ensures unobstructed Li+
transport during long-term cycle, and the low-tortuosity structure
design of LCS can also eliminate the Li+ concentration gradient. In
contrast, the planar Li foil has to face the hindrance of Li+ transport
 Page 7 of 11 Journal of Materials Chemistry A
Journal Name
as-caused by “Dead Li” accumulation on the electrode’s surface. At
the same time, the large SSA of the LCS skeleton also dispersed the
ion flux and reduced the local current density, which can also
efficiently alleviate the dendritic growth of Li metal. The eliminated
Li+ concentration gradient and the lithiophilicity further endow LCS
with the capability in suppressing Li dendrite growth.47 The dendrite-
suppressed Li deposition is evidenced by the morphology
observation as shown in Fig. S13. Both the dendrite-suppressed
morphology and the merging of Li clusters are consistent with
previous report given by L. A. Archer’s group.46
In more symmetric cell tests at harsh conditions, Li@LCS also
shows good cycling stability. As shown in Fig. 8a, 8b, when cycled at
a current density of 1 mA cm-2 with an areal capacity of 2 mA h cm-2,
Li@LCS could maintain a superior cyclic stability of more than 400
cycles (> 1600 hours). The overpotential of the Li@LCS symmetric cell
Published on 14 May 2021. Downloaded by Hokkaido Daigaku on 5/18/2021 11:22:28 AM.
is always kept at a low value of around 20 mV from the 100th to 400th
cycle. Even when cycled at a higher current density or higher areal
capacity as shown in Fig. S14, Li@LCS electrode still presents far
better cyclic performance than Li foil. More comparison of the SEM
images of the Li anodes after cycling is shown in Fig. S15. In addition,
when the orientation of the carbon fiber changed from vertical to
horizontal, “Dead Li” accumulation cannot be avoided due to the lack
of the space for the “Dead Li” to fall down, as we depicted in Fig. S16.
Fig. 8 (a) Symmetric cells tested at a current density of 1 mA cm-2 with a
capacity of 2 mA h cm-2. (b) Voltage profiles of Li@LCS symmetric cell at 100th,
200th, 300th, 400th cycle.
2.3 Electrochemical performance in full cell measurement
In order to further assess the performance of Li@LCS anode in a
full cell, we assembled Li metal|LiCoO2 (LCO) cells by using two kinds
of Li anodes, i.e., Li foil and Li@LCS. The electrolyte was LiPF6 in
EC/DMC. Compared with electrolytes that are based on ether
solvents, electrolytes based on carbonate solvents are less effective
in protecting Li metal, especially in the absence of fluorine-
containing species (e.g. fluoroethylene carbonate, FEC) as
additives .48 This is to say that the self-protection capability for the Li
metal anode will be particularly important to achieve a good cyclic
performance in the above full cell measurement. As shown in Fig. 9a,
the Li@LCS | LCO cell and the Li foil | LCO cell present close initial
capacities of around 148 mA h g-1 as-calculated based on the cathode
active materials. Here, the corresponding amount of Li metal was
excessive both cells. After 300 cycles at a charge/discharge rate of 1
C (150 mA g-1), Li@LCS | LCO cell can maintain a capacity of 102 mA
h g-1, while the Li foil | LCO cell shows a low value of only 69 mA h g-
1. It is reported that the accumulation of “Dead Li” and the thickening
of SEI layer will increase the internal resistance of Li metal | LCO
cells.29,49 The discharging/charging performance of the cathode will
be damaged because the discharging/charging voltage platform
gradually approach the cut-off voltage during the cycling. This means
that once the cell voltage achieves the cut-off voltage, the LCO
Please do not adjust margins
ARTICLE
cathode will not be lithiated/delithiated anymore, even when the
View Article Online
LCO is still at an incomplete state of charge. TheDOI:EIS results as-shown
10.1039/D1TA02491D
in Fig. 9b confirm that both cells had high internal resistance after
300 cycles. The influence of internal resistance on the cell voltage can
be seen intuitively in Fig. 9c, 9d. Since LixCoO2 is considered to be
fully lithiated at 3.88 V (vs. Li/Li+, x=1),50,51 the difference between
Journal of Materials Chemistry A Accepted Manuscript
discharging platform voltage (Vdp) and 3.88 V can be regarded as the
overall overpotential (ηo) in the cell. The ηo is consisted of two parts
as shown below:
ηo = 3.88- Vdp = η + IR sol (6)
Here, η refers to the electrode overpotential, IRsol is the
overpotential from electrolyte. Since the IRsol is very small in
comparison with η, therefore the IRsol is considered negligible, and
the change of ηo approximately reflects the change of electrode
overpotential (η). As shown in Fig. 9c, at the 10th cycle, the η10th
shows a low value of 0.06 V, demonstrating that the Li@LCS | LCO
cell is in good health at the initial several cycles. At the 300th cycle,
the η300th is 0.17 V, indicating that the internal resistance of the
Li@LCS | LCO cell is increased. In view of the unique "Dead Li" self-
cleaning capability possessed by Li@LCS electrode and its low-
tortuosity nature, we also adopted the GITT method to eliminate the
influence from mass-transfer resistance near the electrode surface,
in order to observe the influence from mass-transfer resistance on
the overpotential in the Li@LCS | LCO cell. Thus, the ηGITT represents
the overpotential caused by internal resistance excluding the mass-
transfer resistance. The GITT cycle is conducted after the 300th cycle,
and the ηGITT here shows a value of 0.10 V which is close to the η300th.
The difference between 0.17 V and 0.10 V indicates that the mass-
transfer resistance caused by “Dead Li” accumulation in the Li@LCS
| LCO cell should be around 0.07 V. It is also worth pointing out that
the Li | LCO cells are two-electrodes cell rather than three-electrodes
cell. Hence, during the GITT cycle, the GITT method also inevitably
eliminates the same influence from the cathode side. But according
to previous reports,29 the LCO cathode has a stable performance in
LiPF6-carbonate electrolyte. The used testing condition also
guaranteed that mainly the influences from the Li metal anode was
eliminated during the GITT cycle. Hence, it is reasonable to believe
that the result is reliable. As for the voltage profile of the Li foil| LCO
cell that is shown in Fig. 9d, the η10th is 0.05 V, indicating that the Li
foil| LCO cell also has a good cyclic stability during the first several
cycles. Moreover, the η300th is 0.58 V, implying that the cyclic
performance of the Li foil| LCO cell gradually deteriorates in
subsequent cycles. With the help of GITT, the ηGITT of Li foil| LCO cell
turns back to 0.15 V, demonstrating that the mass-transfer
resistance plays an important role in internal resistance of the Li foil|
LCO cell. The obtained values of η are in good agreement with the
results form EIS, which strengthens the reliability of our inferences
(Fig. 9l). It is also worth pointing out that capacity of the tested cells
increases significantly in the GITT cycle. This phenomenon confirms
that the internal mass-transfer resistance does cause great damage
to the battery capacity, which is also supported by previous study.29
Voltage profiles of the GITT cycle are shown in Fig. 9e,9f for the
Li@LCS | LCO cell and Li foil| LCO cell, respectively. It is clearly to see
that the underlying kinetic behaviour revealed by GITT makes the
voltage curve turn back to similar shape as the 10th cycle. However,
due to the severe ohmic polarization inside the Li foil| LCO cell, which
can be caused by the thickening of SEI and cathode electrolyte
interface (CEI), the charging platform voltage of the Li foil cell can
hardly achieve 4.2 V. The Li foil cell also exhibits an impaired
Coulombic Efficiency owing to the poor Li utilization of Li foil. Cross-
sectional SEM observation of the Li metal anodes after cycling further
 Journal of Materials Chemistry A Page 8 of 11

ARTICLE Journal Name

confirmed our speculation. As shown in Fig. 9g, 9h for the Li@LCS
electrode, the lithiated carbon fibers remain exposed, and “Dead Li”
is not found. As a comparison, the surface of Li foil is covered by
“Dead Li” layer with a thickness of around 50 μm (Fig. 9j). The
morphology observation results also explain why the ηGITT of Li@LCS
is smaller than that of Li foil after 300 cycles. As benefiting from the
is within 0.1 M. Moreover, the low-tortuosity design similar with the
View Article Online
tunnel structure makes the Li+ concentration gradient at the
DOI: 10.1039/D1TA02491D
electrode surface be greatly alleviated.47,52,53 This result is also
confirmed by the corresponding voltage profile shown in Fig. 9c. The
simulated discharge voltage profile shows a similar Vdp as the actual
measured one in 300th cycle, demonstrating the superiority of

Journal of Materials Chemistry A Accepted Manuscript

3D internal space and the dendrite suppressed characteristics, the
thickening of SEI is alleviated in the Li@LCS | LCO cell, making the cell
exhibits a lower ηGITT.
Finally, through the multi-physical simulation of the Li metal |
LCO cells, the superiority of Li@LCS composite electrode in
alleviating the Li+ concentration gradient at the electrode surface is
intuitively demonstrated. The electrode parameters for simulation
were obtained from the 300th cycle corresponding to Fig. 9a. The
electrode structure was simplified as presented in Fig. S1. As shown
Published on 14 May 2021. Downloaded by Hokkaido Daigaku on 5/18/2021 11:22:28 AM.
Li@LCS in alleviating concentration polarization. In comparison, as
for the Li foil with a thickness of 50 μm, the “Dead Li” layer shows an
extremely strong concentration polarization (Fig. 9k). The simulated
voltage curve also indicates a bad performance that is similar with
the actual measured curve. This result also illustrates that the
blocking of Li+ transport at the anode side will seriously damage the
battery capacity, while the unique low-tortuosity structure design is
effective on the cycling stability.
Another full cell measurement in Li-S battery was also conducted

in Fig. 9i, for the Li@LCS | LCO cell, during discharging the Li+ with the results shown in Fig. S18, which presents the obvious
concentration difference from the electrode surface to the inner part superiority of the Li@LCS anode.

Fig 9. Electrochemical performance of Li metal | LCO cells in carbonate-based LiPF6 electrolyte. (a) Cycling capacities of Li metal | LCO full cells tested at a
discharge rate of 1 C (150 mA g-1). (b) Nyquist plot of different Li metal | LiCoO2 full cells after cycling. (c) Voltage-capacity curves of Li@LCS | LiCoO2 full cell
at 10th cycle (black), 150th cycle (green), 300th cycle (blue), the GITT cycle (red) and the simulated 300th cycle (cyan-blue). (d) Voltage-capacity curves of Li
foil | LiCoO2 full cell at 10th cycle (black), 150th cycle (green), 300th cycle (blue), the GITT cycle (red) and the simulated 300th cycle (cyan-blue). Voltage profile
of GITT tests for (e) Li@LCS | LiCoO2 full cell and (f) Li foil| LiCoO2 full cell. (g, h) Cross-sectional SEM images of Li@LCS anode after cycling. (i) Simulation results
of Li+ concentration distribution near anode surface in the Li@LCS| LiCoO2 full cells at 300th cycle (discharge). (j) Cross-sectional SEM images of Li foil anode

Please do not adjust margins

 Page 9 of 11 Journal of Materials Chemistry A
Journal Name
after cycling. (k) Simulation results of Li+ concentration distribution near anode surface in the Li foil | LiCoO2 full cells at 300th cycle (discharge). (l) Internal
resistance values and over-potential (η) values of different cells.
3. Experimental section
3.1 Preparation of “variant of Schweizer’s reagent” coated cotton
scroll precursor
The “variant of Schweizer’s reagent” was prepared by dissolving
copper acetate hydrate (Cu(CH3COO)2·H2O, Labchem-Wako) into
ammonia solution as reported in our previous studies.31,32 The
concentration of Cu ion was 0.6 M in this report, and the optimization
of the concentration is shown in the Supplementary Information.
Cotton sheet with aligned fibers was purchased from Suzuran
company (Japan, Nagoya). In a typical preparation, a cotton sheet
Published on 14 May 2021. Downloaded by Hokkaido Daigaku on 5/18/2021 11:22:28 AM.
with a length 36 cm was rolled up to a scroll with a diameter of 1.5
cm. The cotton scroll was cut into disks with a thickness of 3 mm.
Subsequently, the cotton scroll disks were soaked into the “variant
of Schweizer’s reagent” for 12 h. After being taken out from the
solution, the disks were washed by tetrahydrofuran (THF) for several
times in order to remove the excess solution at the surface and the
gap. The difference brought by the material synthesis conditions is
shown in Fig. S19 and S20.
3.2 Preparation of LCS and Li@LCS composite electrode
The above “variant of Schweizer’s reagent” coated cotton scrolls
were annealed in Ar flow at 850 ℃ for 1.5 h to obtain the CS@Cu
disks, which were polished to a thickness of around 1.3 mm.
Afterwards, the CS@Cu disks were calcined in a muffle furnace at 250
℃ for 10 h, in this way, the LCS disks were obtained. The comparable
pristine CS was also prepared by pyrolyzing and annealing the same
cotton scroll at 850 ℃ in a tube furnace for 1.5 h. The LCS disks were
transferred into a glove-box filled with Ar gas, in which the oxygen
level was kept below 0.1 ppm. Pre-polished Li metal was melted on
a hot plate at 350 ℃. The LCS disk was put in contact with the molten
Li through the bottom in order to make molten Li infiltrate into the
skeleton slowly. After carefully taking out the composite, the Li@LCS
electrode was obtained.
3.3 Electrochemical test
Metallic Li foil and Li@LCS electrodes were cut into disks with a
diameter of 10 mm. Swagelok cells with a 10 mm inner diameter
were employed. Symmetric cells were assembled with two identical
electrodes. 1 M lithium bis(trifluoromethanesulfonyl)imide (LiTFSI,
Kishida Chemical, 99.90%) in 1,3-dioxolane/1,2-dimethoxyethane
(DOL/DME, volume ratio 1:1, purchased from Sigma-Aldrich with
purity of 99.8% and 99.5% respectively) was used as the electrolyte,
and 1 wt% lithium nitrate (LiNO3, Kanto Chemical) was used as an
additive. A glass-fiber separator was used, while the added
electrolyte amount was 125 μL per cell. Symmetric cells were tested
through a galvanostatic Li plating/stripping method, during which
the time-dependent overpotential was recorded.
The galvanostatic intermittent titration technique (GITT) test for
symmetric cells was conducted by dividing a whole cycle into 480 s.
Here, the symmetric cell was charged for 15 s at a current density of
1 mA cm-2, then rested for 3 min for the completion of Li+ diffusion
on the electrode surface. When the overall areal capacity reached 1
mA h cm-2, i.e., after 240 times of charging for a total time of 3600 s,
the cell was switched to the discharge mode, and the
abovementioned steps were repeated.
Please do not adjust margins
ARTICLE
View Article Online
DOI: 10.1039/D1TA02491D
Li metal | LCO full cells were assembled by using LiCoO2 cathode
Journal of Materials Chemistry A Accepted Manuscript
and different Li metal anodes. LiCoO2 slurry was prepared by firstly
mixing LiCoO2 powders, poly-vinylidene difluoride (PVDF) and
acetylene black (AB) with a weight ratio of 8:1:1 in N-methyl-2-
pyrrolidone (NMP). The slurry was pasted to an aluminium (Al) foil.
After drying, the active material loaded on Al foil was around 2.5 mg
cm-2. A disk of LiCoO2 loaded Al foil with a diameter of 10 mm was
used as cathode, Li foil or Li@LCS electrode with a diameter of 10
mm was used as Li metal anode. 1 M lithium hexafluorophosphate
(LiPF6) in ethylene carbonate/ dimethyl carbonate (EC/DMC 1:1 in
volume ratio, Chameleon Reagent) was used as the electrolyte. A
glass-fiber separator was used, while the added electrolyte was 125
μL per cell. Li metal | LCO full cells were discharged from 4.2 V to 3.0
V, and charged from 3.0 V to 4.2 V at a constant current. The current
density was dependent upon the cathode capacity.
The GITT test for full cells was conducted by intermittently charging
(or discharging) the cell until the cell voltage reached the cut-off
voltage. Here, the cells were firstly discharged at a current density of
150 mA g-1 (calculated by cathode active material) for 15 s which was
defined as one section, then the cells rested for 3 min for the
completion of Li+ diffusion on the Li-metal surface. When the cell
voltages achieved the cut-off voltage of 3.0 V, the cells were switched
to the charge mode, and the abovementioned steps were repeated.
For the charge mode, the cells were stopped until that the cell
voltages achieved the cut-off voltage of 4.2 V. Otherwise the cells
were stopped until the charge section lasted for 480 times, in order
to avoid the case that the cells with upper internal resistance cannot
achieve the cut-off voltage. The capacity of GITT cycle was the sum
of the capacities of all charge or discharge sections. To further verify
the Li+ concentration distribution near anode surface, a multi-
physical simulation of the Li metal | LCO cells was conducted using
COMSOL Multiphysics (see details in the Supplementary
Information).
The electrochemical impedance spectra (EIS) measurement was
carried out on a Princeton apparatus with a frequency range from
100 kHz to 0.01 Hz. A multi-channel battery tester (Hokuto Denko,
8CH) was used to test the electrochemical cyclic stability for different
Li metal cells. In addition, Li-S full cells, i.e., Li metal | S, were also
assembled to further verify the advantages of Li@LCS electrode, and
the detailed experiment and results are shown in the Supplementary
Information.
3.4 Materials Characterization
Scanning electron microscope (SEM, ZEISS Sigma 500) was used to
investigate the morphological information for different samples. For
the ex-situ observation of Li@LCS composite electrodes both before
and after electrochemical test, the cells were firstly disassembled in
Ar filled glove-box, then the Li@LCS electrodes were separated from
the sandwiches and washed by DOL/DME co-solvent for several
times. For laser scanning confocal microscope (LSCM) observation,
the sample is first placed into a chamber with a quartz window to
avoid contact between the sample and the air. The sample was then
observed through the optical window. X-ray diffraction (XRD)
patterns of the samples were recorded by a Rigaku MiniFlex600 X-
ray diffractometer. N2 absorption/desorption measurement was
used to analyse the porous characteristics of the carbon-based
materials by using a Microtrac-BEL (BELSORP-mini) surface area
 Journal of Materials Chemistry A
ARTICLE
analyser. The Brunauer-Emmett-Teller (BET) method was used to
calculate the specific surface areas (SSA) of the samples, while the
Barrett-Joyner-Halenda (BJH) method was employed to estimate the
pore-size distribution.
4. Conclusions
In summary, we prepared a carbon scroll as a novel Li metal host by
the pyrolysis of cotton template. The carbon scroll was consisted of
vertically aligned carbon fibers, presenting low-tortuosity structure
characteristic. Lithiophilic CuOX NPs were coated on the carbon fibers,
making Li metal easily infiltrate into the carbon host to form Li@LCS
composite electrode. Li@LCS electrode showed superior
electrochemically cyclic stability to the conventional Li foil electrode,
especially in alleviating the accumulation of “Dead Li” and reducing
Published on 14 May 2021. Downloaded by Hokkaido Daigaku on 5/18/2021 11:22:28 AM.
the mass-transfer resistance. It was investigated that the inactive Li
blocks could naturally sink into the framework of LCS, resulting in an
unimpeded Li+ transport path even in long-term cycling. Moreover,
thanks to the low-tortuosity nature and the “Dead Li” self-cleaning
capability of the Li@LCS electrode, the decrease in battery capacity
which was caused by the upper mass-transfer resistance in the Li |
LCO full cell could be significantly reduced. As a result, the Li@LCS
electrode delivered a stable cyclic performance of > 1000 hrs in the
symmetric cell test at a current density of 1 mA cm-2, and ensured
that the Li metal | LCO full cell could maintain a low overpotential
during 300 cycles. This work provided a new solution to the problem
of “Dead Li” accumulation during the use of Li metal anodes through
a simple structural design strategy. In particular, the structure design
of Li@LCS electrode is compatible with most of the optimization
strategies that have been applied to conventional Li foil, making this
research provide efficient inspiration for developing high-
performance Li metal anode.
Conflicts of interest
There are no conflicts to declare.
Acknowledgements
A part of the characterization work was conducted at Hokkaido
University, supported by the “Nanotechnology Platform”
Program of the Ministry of Education, Culture, Sports, Science
and Technology (MEXT), Japan. R. Zhu would like to thank the
support by Program for Leading Graduate Schools (Hokkaido
University "Ambitious Leader’s Program"). C. Zhu also thanks
the supports by National Natural Science Foundation of China
(No. 52006238), Natural Science Foundation of Jiangsu Province
(No. BK20200635), China Postdoctoral Science Foundation
(2020M681764), and the Fundamental Research Funds for the
Central University (Grant no. 2020ZDPYMS24).
Notes and references
1 B. Dunn, H. Kamath and J. M. Tarascon, Science 2011,334,
928-935.
2 P. G. Bruce, S. A. Freunberger, L. J. Hardwick and J. M.
Tarascon, Nat. Mater. 2011, 11, 19-29.
Please do not adjust margins
Page 10 of 11
Journal Name
3 G. Zheng, S. W. Lee, Z. Liang, H. W. Lee, K. Yan, H. Yao, H.
View Article Online
Wang, W. Li, S. Chu and Y. Cui, Nat. Nanotechnol. 2014, 9, 618-
DOI: 10.1039/D1TA02491D
623.
4 G. Bieker, M. Winter and P. Bieker, Phys. Chem. Chem. Phys.
2015, 17, 8670-8679.
5 K. Nishikawa, Y. Fukunaka, T. Sakka, Y. H. Ogata and J. R.
Journal of Materials Chemistry A Accepted Manuscript
Selman, Journal of The Electrochemical Society 2007, 154, 6.
6 Ulrich von Sacken, Eric Nodwell, Avtar sundher and J. R. Dahn,
Journal of Power Sources 1995, 54, 240-245.
7 C. Fang, X. Wang and Y. S. Meng, Trends in Chemistry 2019, 1,
152-158.
8 A. Kushima, K. P. So, C. Su, P. Bai, N. Kuriyama, T. Maebashi,
Y. Fujiwara, M. Z. Bazant and J. Li, Nano Energy 2017, 32, 271-
279.
9 S. Wu, Z. Zhang, M. Lan, S. Yang, J. Cheng, J. Cai, J. Shen, Y.
Zhu, K. Zhang and W. Zhang, Adv. Mater. 2018, 30, 1705830.
10 S. Li, Q. Liu, J. Zhou, T. Pan, L. Gao, W. Zhang, L. Fan and Y. Lu,
Advanced Functional Materials 2019, 29, 1808847.
11 Y. Ma, B. Yao, M. Zhang, H. Bai and G. Shi, Journal of Materials
Chemistry A 2018, 6, 15603-15609.
12 G. Yang, J. Chen, P. Xiao, P. O. Agboola, I. Shakir and Y. Xu,
Journal of Materials Chemistry A 2018, 6, 9899-9905.
13 C. P. Yang, Y. X. Yin, S. F. Zhang, N. W. Li and Y. G. Guo, Nat.
Commun. 2015, 6, 8058.
14 J. Gao, X. Cai, J. Wang, M. Hou, L. Lai and L. Zhang, Chemical
Engineering Journal 2018, 352, 972-995.
15 H. Duan, J. Zhang, X. Chen, X. D. Zhang, J. Y. Li, L. B. Huang, X.
Zhang, J. L. Shi, Y. X. Yin, Q. Zhang, Y. G. Guo, L. Jiang and L. J.
Wan, J. Am. Chem. Soc. 2018, 140, 18051-18057.
16 B. Yu, T. Tao, S. Mateti, S. Lu and Y. Chen, Adv. Funct. Mater.
2018, 28, 1803023.
17 R. Zhang, X. R. Chen, X. Chen, X. B. Cheng, X. Q. Zhang, C. Yan
and Q. Zhang, Angew. Chem. Int. Ed. 2017, 56, 7764-7768.
18 X. Chen, X. R. Chen, T. Z. Hou, B. Q. Li, X. B. Cheng, R. Zhang
and Q. Zhang, Sci. Adv. 2019, 5, 7728.
19 X. Fan, L. Chen, X. Ji, T. Deng, S. Hou, J. Chen, J. Zheng, F. Wang,
J. Jiang, K. Xu and C. Wang, Chem 2018, 4, 174-185.
20 Y. Yan, Y. M. Zhang, Y. T. Wu, Z. Z. Wang, A. Mathur, H. C. Yang,
P. Chen, S. Nair and N. Liu, ACS Applied Energy Materials 2018,
1, 6345-6351.
21 X. Ren, S. Chen, H. Lee, D. Mei, M. H. Engelhard, S. D. Burton,
W. Zhao, J. Zheng, Q. Li, M. S. Ding, M. Schroeder, J. Alvarado,
K. Xu, Y. S. Meng, J. Liu, J.-G. Zhang and W. Xu, Chem 2018, 4,
1877-1892.
22 N. W. Li, Y. X. Yin, C. P. Yang and Y. G. Guo, Adv. Mater. 2016,
28, 1853-1858.
23 Y. Liu, D. Lin, P. Y. Yuen, K. Liu, J. Xie, R. H. Dauskardt and Y.
Cui, Adv. Mater. 2017, 29, 1605531.
24 H. Duan, Y. X. Yin, Y. Shi, P. F. Wang, X. D. Zhang, C. P. Yang, J.
L. Shi, R. Wen, Y. G. Guo and L. J. Wan, J. Am. Chem. Soc. 2018,
140, 82-85.
25 M. S. Kim, J.-H. Ryu, Deepika, Y. R. Lim, I. W. Nah, K.-R. Lee, L.
A. Archer and W. Il Cho, Nat. Energy 2018, 3, 889-898.
26 T. Zhang, Y. Wang, T. Song, H. Miyaoka, K. Shinzato, H.
Miyaoka, T. Ichikawa, S. Shi, X. Zhang, S. Isobe, N. Hashimoto
and Y. Kojima, Joule 2018, 2, 1522-1533.
27 J. Wang, Y. Yamada, K. Sodeyama, E. Watanabe, K. Takada, Y.
Tateyama and A. Yamada, Nat. Energy 2017, 3, 22-29.
28 C. Jin, T. Liu, O. Sheng, M. Li, T. Liu, Y. Yuan, 
J. Nai, Z. Ju, W. Zhang, Y. Liu, Y. Wang, Z. Lin, J. Lu and
X. Tao, Nature Energy, 2021, 6, 378-387.
29 K.-H. Chen, K. N. Wood, E. Kazyak, W. S. LePage, A. L. Davis, A.
J. Sanchez and N. P. Dasgupta, Journal of Materials Chemistry
A 2017, 5, 11671-11681.
30 C. Fang, J. Li, M. Zhang, Y. Zhang, F. Yang, J. Z. Lee, M. H. Lee,
J. Alvarado, M. A. Schroeder, Y. Yang, B. Lu, N. Williams, M.
Ceja, L. Yang, M. Cai, J. Gu, K. Xu, X. Wang and Y. S. Meng,
Nature 2019, 572, 511-515.
 Page 11 of 11 Journal of Materials Chemistry A
Journal Name
31 R. Zhu, N. Sheng, Z. Rao, C. Zhu, Y. Aoki and H. Habazaki,
Journal of Materials Chemistry A 2019, 7, 27066-27073.
32 R. Zhu, C. Zhu, N. Sheng, Z. Rao, Y. Aoki and H. Habazaki,
Chemical Engineering Journal 2020, 388, 124256.
33 Q. Zhang, J. Luan, Y. Tang, X. Ji, S. Wang and H. Wang, Journal
of Materials Chemistry A 2018, 6, 18444-18448.
34 Y. Liu, S. Zhang, X. Qin, F. Kang, G. Chen and B. Li, Nano Lett.
2019, 19, 4601-4607.
35 Z. Liang, D. Lin, J. Zhao, Z. Lu, Y. Liu, C. Liu, Y. Lu, H. Wang, K.
Yan, X. Tao and Y. Cui, Proc. Natl. Acad. Sci. 2016, 113, 2862-
2867.
36 Y. Zhu and C. Wang, Journal of Physical chemistry C 2010, 6,
2830-2841.
37 D. T. Boyle, X. Kong, A. Pei, P. E. Rudnicki, F. Shi, W. Huang, Z.
Bao, J. Qin and Y. Cui, ACS Energy Letters 2020, 5, 701-709.
38 C. Niu, H. Pan, W. Xu, J. Xiao, J. G. Zhang, L. Luo, C. Wang, D.
Mei, J. Meng, X. Wang, Z. Liu, L. Mai and J. Liu, Nat.
Published on 14 May 2021. Downloaded by Hokkaido Daigaku on 5/18/2021 11:22:28 AM.
Nanotechnol. 2019, 14, 594-601.
39 K. N. Wood, E. Kazyak, A. F. Chadwick, K. H. Chen, J. G. Zhang,
K. Thornton and N. P. Dasgupta, ACS Cent. Sci. 2016, 2, 790-
801.
40 A. Nyman, M. Behm and G. Lindbergh, Electrochimica Acta
2008, 53, 6356-6365.
41 R. Cao, J. Chen, K. S. Han, W. Xu, D. Mei, P. Bhattacharya, M.
H. Engelhard, K. T. Mueller, J. Liu and J.-G. Zhang, Adv. Funct.
Mater. 2016, 26, 3059-3066.
42 B. Wu, J. Lochala, T. Taverne and J. Xiao, Nano Energy 2019,
40, 34-41.
Please do not adjust margins
ARTICLE
43 M. Ue and K. Uosaki, Current Opinion in Electrochemistry
View Article Online
2019, 17, 106-113. DOI: 10.1039/D1TA02491D
44 Y. Wang, Z. Wang, D. Lei, W. Lv, Q. Zhao, B. Ni, Y. Liu, B. Li, F.
Kang and Y. B. He, ACS Appl. Mater. Interfaces 2018, 10,
20244-20249.
45 Q. Yun, Y. B. He, W. Lv, Y. Zhao, B. Li, F. Kang and Q. H. Yang,
Journal of Materials Chemistry A Accepted Manuscript
Adv. Mater. 2016, 28, 6932-6939.
46 J. Zheng, T. Tang, Q. Zhao, X. Liu, Y. Deng and L. A. Archer, ACS
Energy Letters 2019, 4, 1349-1355.
47 H. Chen, A. Pei, J. Wan, D. Lin, R. Vilá, H. Wang, D. Mackanic,
H.-G. Steinrück, W. Huang, Y. Li, A. Yang, J. Xie, Y. Wu, H. Wang
and Y. Cui, Joule 2020, 4, 938-952.
48 H. Zhang, G. G. Eshetu, X. Judez, C. Li, L. M. Rodriguez-
Martinez and M. Armand, Angew. Chem. Int. Ed. 2018, 57,
15002-15027.
49 T. T. Beyene, H. K. Bezabh, M. A. Weret, T. M. Hagos, C.-J.
Huang, C.-H. Wang, W.-N. Su, H. Dai and B.-J. Hwang, Journal
of The Electrochemical Society 2019, 166, A1501-A1509.
50 G. G. Amatucci, J. M. Tarascon and L. C. Klein, 1996, 143, 1114-
1123.
51 K. Mizushima, P. C. Jones, P. J. Wiseman and J. B.
Goodenough, Solid State Ionics 1981, 3/4, 171-174.
52 C. Chen, Y. Zhang, Y. Li, Y. Kuang, J. Song, W. Luo, Y. Wang, Y.
Yao, G. Pastel, J. Xie and L. Hu, Adv. Energy Mater. 2017, 7,
1700595.
53 Y. Zhang, W. Luo, C. Wang, Y. Li, C. Chen, J. Song, J. Dai, E. M.
Hitz, S. Xu, C. Yang, Y. Wang and L. Hu, Proc. Natl. Acad. Sci.
2017, 114, 3584-3589.
 Chemical Engineering Journal 420 (2021) 129906

Contents lists available at ScienceDirect

Chemical Engineering Journal

journal homepage: www.elsevier.com/locate/cej

Synthesis of Sn@SnO2 core-shell microcapsules by a self-oxidation strategy

for medium temperature thermal storage
Nan Sheng a, Jiahui Lu a, Jingdong Hu a, Ruijie Zhu b, Laras Fadillah b, Cheng Wang c,
Chunyu Zhu a, *, Zhonghao Rao a, Hiroki Habazaki b
a
School of Electrical and Power Engineering, China University of Mining and Technology, Xuzhou 221116, China
b
Division of Applied Chemistry, Faculty of Engineering, Hokkaido University, Kita-Ku Kita 13, Jo Nishi 8, Sapporo, Hokkaido 060-8628, Japan
c
School of Petroleum Engineering, Changzhou University, Chang Zhou 213164, China

A R T I C L E I N F O A B S T R A C T

Keywords: Latent heat storage using metals as solid-liquid phase change materials (PCMs) have been concerned for medium-
Phase change material temperature thermal energy storage. However, due to the leakage and corrosion problems during phase trans­
Latent heat storage formation, the use of metallic PCMs are greatly limited. Encapsulation of PCM microparticles to form micro­
Thermal energy storage
encapsulated PCMs (MEPCMs) is an effective strategy, which is unfortunately technologically difficult for
Thermal management
metallic PCMs. In this study, we develop metallic Sn MEPCMs coated by a stable SnO2 shell through a facile self-
Phase change capsule
oxidation method. Sn@SnO2 core-shell MEPCMs are fabricated by two steps: firstly, Sn microspheres are pre-
treated by vapor or water to form an oxide precursor shell on Sn microspheres; secondly, heat oxidation treat­
ment under O2 atmosphere is conducted to form a stable SnO2 shell. The Sn@SnO2 microcapsules exhibit a
melting point of ~232 ◦ C and latent heat of ~53 J/g. Importantly, the capsules present an excellent thermal
cycling stability, in which after 100 cycles of melting-freezing the phase change properties and core-shell
structure could be well retained. These results reinforce the promising application of microencapsulated Sn in
medium-temperature thermal storage.

1. Introduction heat storage [2,3].

In order to alleviate the energy crisis and climate warming due to the
excessive exploitation and utilization of fossil fuels, clean energies such
as solar energy and wind energy have recently received great attentions.
However, the wide application of those renewable energies is limited by
their intermittent and unstable natures, which induce the mismatch
between energy demand and supply in both time and location [1]. The
development of reliable thermal energy storage (TES) technologies is
very important for the efficient utilization of renewable energies, which
can fill the gaps between energy demand and supply. Latent heat storage
(LHS) using phase change material (PCM) is a promising TES technol­
ogy, which is based on the storage or release of latent heat when the
PCM undergoes the phase transition between the solid phase and liquid
phase. The obvious advantages of LHS, including high thermal storage
density, constant phase transition temperature and reversible/repeat­
able phase change behavior, have reinforced LHS-based TES systems
much more reliable and efficient as compared to conventional sensible
Currently, the technologies for low-temperature PCMs below 100 ◦ C
have been well established in preparation and application, while the
medium and high temperature PCMs over 100 ◦ C have been greatly
concerned which require the devotion of more efforts. The medium and
high temperature PCMs include mainly molten salts such as carbonates
[4,5], chlorides [6,7], nitrates [8] et al., and metal/alloys such as Cu [9],
Al-Si [10,11], Zn [12,13], Sn [14,15] et al. The medium and high tem­
perature PCMs, especially the metallic PCMs, have high heat storage
densities and high thermal conductivity, which could be used for me­
dium and high temperature TES systems in waste heat recovery,
concentrated solar power plants and so on. However, the use of metallic
PCMs as thermal storage media is significantly limited, because the
liquified metals could leak out and are highly corrosive and chemically
active [16,17]. Shape-stabilization strategies, including compositing
with porous supporting additives and encapsulation to make core-shell
capsules, have been widely investigated for the liquid-solid PCMs
[18–20]. The use of encapsulation technology to make core-shell nano/

* Corresponding author.
E-mail address: zcyls@cumt.edu.cn (C. Zhu).

https://doi.org/10.1016/j.cej.2021.129906
Received 31 January 2021; Received in revised form 22 March 2021; Accepted 14 April 2021
Available online 21 April 2021
1385-8947/© 2021 Elsevier B.V. All rights reserved.
N. Sheng et al. Chemical Engineering Journal 420 (2021) 129906

Fig. 1. Experiment procedures for synthesizing Sn@SnO2 core-shell MEPCMs. (a) Vapor pretreatment and (b) water pretreatment.

micro/macro-capsules, which are consisted of PCM cores and stable
shells, has particularly got a lot of attentions. The core PCM is well
encapsulated by solid and stable shells, as a result the capsules can be
considered as solids even when the core PCM is in the liquid state. So far,
the preparation and application of capsules for low temperature PCMs
such as paraffin and salt hydrates have achieved many successes
[21,22]. However, the encapsulation of metallic PCMs is very difficult
because of their high corrosivity to the shell materials and the large
volume change during phase transition at high utilization temperature.
Recently, Zhang et al. [23] prepared Fe-core/SiC/C-shells capsules of
around 500 μm by a pouted bed CVD method, and the capsules can be
used at a high temperature more than 1100 ◦ C. Wei et al. [11] suc­
cessfully prepared the core(=Al–Si/Bi)/void/shell(=Al2O3) composite
microcapsules (with melting temperature of ~571.9 ◦ C to 631.9 ◦ C) by
using Al/Bi immiscible alloy powders as starting material and tetrae­
thoxysilane as SiO2 source, which smartly utilized the in-situ displace­
ment reaction between SiO2 and molten Al. Metallic Sn has a melting
point of 231.9 ◦ C, which is proper for medium temperature heat storage
with potential applications in medium temperature solar energy utili­
zation and waste heat recovery. Lai et al. [24] successfully coated silica
shells on particles of Sn and Sn-Zn alloys by a sol-gel method, and added
them to HITEC molten salt to form latent heat fluid, which was applied
to Stirling engine. The results shown that the heat transfer and heat
Table 1
Naming of the samples prepared in this study.
Pretreating Time After Pretreatment After Heat Oxidation Treatment
0 min pure-Sn pure-Sn-0
30 min vapor-30-pre vapor-30
90 min vapor-90-pre vapor-90
20 min water-20-pre water-20
storage properties of the microcapsules added molten salt fluid were
improved. These attempts on metallic phase change capsules give us
many encouragements for continuing to explore a facile and consoli­
dated method to prepare medium temperature Sn microcapsules.
In this study, we are focusing on the preparation of Sn@SnO2 core-
shell micro-encapsulated PCM (MEPCM) for medium temperature
thermal storage. The microcapsules are produced by two steps,
including the first vapor or water treatment and the subsequent heat
oxidation treatment. The formation mechanism, morphology/structure,
phase change properties and cycling stability are examined in details.
2. Experimental section
2.1. Preparation of MEPCMs
The raw material of Sn microspheres used in this study was pur­
chased from Shanghai Xiangtian Nano Materials Co., Ltd. The as-
obtained Sn microspheres were filtered to 45–50 μm before use.
Deionized water used in the experiments was purified by using a water
purification system (EPED, EQ201).
The whole synthesis procedure is illustrated in Fig. 1. As shown in
Fig. 1, the Sn@SnO2 core-shell MEPCMs were prepared by two steps. The
first step was the pretreatment of Sn microspheres, which involved two
different methods: (a) vapor pretreatment and (b) water pretreatment. For
vapor pretreatment (Fig. 1a), Sn particles were homogeneously dispersed
over a nylon mesh with hole size of 0.5 μm, which was fixed on a beaker
filled with deionized water. The beaker was heated on a magnetic stirring
hot plate and the deionized water was kept boiling for 30 min or 90 min,
respectively, to make steam go through the particles. For the water pre­
treatment (Fig. 1b), Sn particles were dispersed into deionized water and
heated to 80 ◦ C, keeping for 20 min. Subsequently, the samples were

(a) (b) (c)

vapor-90-pre vapor-90

water-20
Intensity(a.u.)

Intensity(a.u.)

Intensity(a.u.)

vapor-30-pre vapor-30

water-20-pre

pure-Sn-0
JCPDS No. 41-1445 SnO2
pure-Sn
JCPDS No. 41-1445 SnO2
JCPDS No. 04-0673 Sn JCPDS No. 04-0673 Sn
JCPDS No. 04-0673 Sn

25 30 35 40 45 50 55 60 65 70 75 25 30 35 40 45 50 55 60 65 70 75 25 30 35 40 45 50 55 60 65 70 75
2θ(degree) 2θ(degree) 2θ(degree)

Fig. 2. XRD patterns of the microparticles (a) pure-Sn, vapor-30-pre and vapor-90-pre, (b) pure-Sn-0, vapor-30 and vapor-90 and (c) water-20-pre and water-20.

2
N. Sheng et al. Chemical Engineering Journal 420 (2021) 129906

Pretreating Time Raw material of pure Sn After Heat Oxidation Treatment

a-1 a-2 b-1 b-2 b-3

0 min

After Vapor Pretreatment After Heat Oxidation Treatment

c-1 c-2 d-1 d-2 d-3

30 min

e-1 e-2 f-1 f-2 f-3

90 min

After Water Pretrentment After Heat Oxidation Treatment

g-1 g-2 h-1 h-2 h-3

20 min

Fig. 3. SEM images of the microparticles after vapor or water pretreatment and after heat oxidation treatment. (a) Raw material of pure-Sn, (b) pure-Sn-0, (c) vapor-
30-pre, (d) vapor-30, (e) vapor-90-pre, (f) vapor-90, (g) water-20-pre and (h) water-20.

washed, filtrated and dried at 60 ◦ C over 12 h for further use. The second
step was heat oxidation treatment, during which the pretreated particles
were placed in an Al2O3 crucible and heated in a tube furnace from room
temperature to 500 ◦ C at a rate of 10 ◦ C min− 1 and kept at 500 ◦ C for 30
min under an oxygen gas flow of 50 mL min− 1. After the completion of heat
oxidation treatment, the samples were naturally cooled to room tempera­
ture. Besides, pure Sn particles were directly undergone the heat oxidation
treatment without any pretreatment as a reference sample. Table 1 listed
the names of all samples.
Fig 4. Cross sectional SEM image of the typical sample obtained after heat
oxidation treatment.
2.2. Characterization
The powder X-ray diffraction patterns of the samples were measured
by an X-ray diffractometer (XRD, BRUKER, D8 ADVANCE) using Cu Kα
radiation (λ = 0.15405 nm). The X-ray tube was operated at 40 kV and
30 mA. The X-ray diffractogram was recorded at 0.01◦ interval with a
scanning rate of 6◦ /min for the scanning angle (2θ) from 5◦ to 75◦ . The
morphology of the samples was observed by high resolution field-
emission scanning electron microscope (SEM, TESCAN, MAIA3 LMH).
X-ray energy dispersive spectrometry (EDS) analysis combined with
SEM observation was conducted to observe the elemental mapping of
the microcapsules. The cross section elemental distribution of Sn@SnO2
core-shell MEPCM was also analyzed by electron prob micro-analyzer
(EPMA, SHIMADZU, 8050G). As for the cross section observation, the
MEPCMs were mounted in resin and were carefully polished to get a
smooth cross-section surface by using abrasive paper, polishing fluid
and cloth in sequence.
To carefully identify the shell properties of the sample after water
treatment, the following confirmatory experiment was conducted by X-
ray photoelectron spectroscopy (XPS, JEOL JPS-9200). A piece of pure
Sn plate was polished carefully by abrasive paper, which was subse­
quently treated by hot water at 80 ◦ C. Both the polished Sn plate and the
hot water treated Sn plate were parallelly observed by XPS. By con­
trolling the Ar ion etching time, a series of spectra at different depths (0,
400 and 800 nm) were obtained for both samples.
To observe the heat oxidation behavior of the formation of final
microcapsules, the samples obtained after vapor and water treatment
were measured by a simultaneous differential scanning calorimetry
analyzer combined with a thermal gravimetric analyzer (DSC-TGA,
NETZSCH, STA449F5). The samples were placed in an Al2O3 crucible
and heated in O2 or N2 flow of 50 mL min− 1 respectively, which were
conducted at a heating rate of 10 ◦ C min− 1 to 500 ◦ C and kept at 500 ℃
for 30 min.
The thermal change behavior and phase change enthalpies of the
samples were measured by using a differential scanning calorimetry

3
N. Sheng et al. Chemical Engineering Journal 420 (2021) 129906

(a)

(d)

10μm
Scan
line

(b) Sn
1 μm

Oxide layer
400
(e) Sn
15000
10μm
O
300
Intensity (counts)

Intensity (counts)
10000

(c) O 200

5000
100

0 0
0 1 2 3 4 5 6
Distance ( m)
10μm

Fig. 5. Cross sectional morphology and elemental analysis of the pre-treated sample water-20-pre. (a, d) cross-sectional image of water-20-pre, (b) EPMA mapping of
Sn, (c) EPMA mapping of O, (f) elemental line scan of Sn and O corresponding to image (d).

analyzer (DSC, TA900). In this measurement, 20 mg samples were
placed in an Al crucible and analyzed in a N2 flow rate of 250 mL min− 1
and a heating/cooling rate of 2 ◦ C min− 1. The cyclic stability of the
MEPCMs was evaluated by 100 times cycling test using DSC. For cycle
test, the MEPCMs were heated from 90 ◦ C to 280 ◦ C at a heating/cooling
rate of 20 ◦ C min− 1 in the N2 atmosphere. The thermal conductivity of
the samples was analyzed by a thermal conductivity analyzer (TC 3000,
Xiatech, Xian), during which the pure Sn powders and the Sn@SnO2
microcapsules were pressed at 10 MPa in a module to form pellets for
measurements. Each sample was measured more than five times to
achieve an average value.
3. Results and discussion
3.1. Material properties of the Sn@SnO2 MEPCMs
XRD analysis was carried out to identify the crystal phase composi­
tion of the samples obtained at different stages, as shown in Fig. 2. The
XRD patterns for pure-Sn, vapor-30-pre and vapor-90-pre are presented
in Fig. 2a, in which only a single phase of Sn (JCPDS No. 04-0673) was
detected for all samples. Even after the vapor treatment of Sn micro­
spheres, the samples do not indicate a second crystal phase. This might
be due to the amorphous nature and low content of the shell precursor
through the reaction between Sn and water. The careful observation of
the shell precursor will be discussed in following XPS and EPMA anal­
ysis. After heat oxidation treatment to 500 ◦ C under O2 flow, the samples
are consisted of SnO2 (JCPDS No. 41-1445) and Sn, as confirmed by XRD
patterns in Fig. 2b. As a comparison, the Sn microspheres were also
treated by another process in hot water at 80 ◦ C. Similar to the case of
vapor treatment, the XRD pattern (Fig. 2c) after water treatment in­
dicates a single crystal phase of metallic Sn, and after the subsequent
heat oxidation treatment, the XRD pattern indicates two crystal phases
of SnO2 and Sn.
The micromorphology and structure of the samples were observed by
SEM, with the results shown in Fig. 3. As shown in Fig. 3a, the raw
material of Sn particles are microspheres of around 50 μm with smooth

4
N. Sheng et al. Chemical Engineering Journal 420 (2021) 129906

Fig. 6. XPS spectra of (a) O 1s and (b) Sn 3d for the pure Sn plate and the water-pretreated Sn plate.

Fig. 7. TG and DSC curves of the samples measured in O2 and N2 atmosphere.

surface, which also exhibits good dispersion. After the direct heat
oxidation treatment at 500 ◦ C, the pure-Sn-0 particles become rough,
wrinkled and agglomerated together as shown in Fig. 3b. The melting
point of Sn is around 232 ◦ C. During the heat treatment to 500 ◦ C under
O2 flow, the particles were oxidized on the surface to form a thin oxide
layer, and at above the melting point of Sn, the liquid Sn leaked from the
thin oxide layer, making the particles conglutinate and agglomerate.
After the heat oxidation treatment for pure-Sn, the particles cannot be
dispersed. Fig. 3c–f shows the SEM images of the samples formed by
using vapor pretreatment and after heat oxidation treatment under
different vapor treating time, respectively. The surfaces of vapor-30-pre
and vapor-90-pre are still smooth, in addition, the particles also show

5
N. Sheng et al. Chemical Engineering Journal 420 (2021) 129906

Fig. 8. DSC results for pure-Sn. (a) is the melting-freezing curves for 5 cycles, (b, c) are the enlarged freezing and melting peak curves.

Table 2
The summary of phase change temperatures and enthalpy values of pure-Sn.
Sample Cycle Melting Freezing

Ts/◦ C TP/◦ C TE/◦ C Enthalpy/(J/g) TS/◦ C TP/◦ C TE/◦ C Enthalpy/(J/g)

pure-Sn 1st 227.5 233.2 237.2 55.2 173.1 146.3 135.4 55.2
2nd 228.7 233.2 237.0 55.0 171.0 146.4 133.7 55.0
3rd 228.9 233.2 237.0 54.9 169.0 146.7 132.9 54.9
4th 229.1 233.2 236.7 54.8 168.3 146.9 132.4 54.8
5th 229.0 233.2 236.7 54.8 167.7 147.1 132.1 54.8

TS is the starting temperature, TP is peak temperature and TE is ending temperature for melting or freezing.

Fig. 9. DSC results for the Sn@SnO2 capsules. DSC curves at the 5th, 50th and 100th cycles for (a) vapor-30, (b) vapor-90 and (c) water-20. (d) Is the 100th cycle DSC
curves of vapor-30, vapor-90 and water-20.

6
N. Sheng et al. Chemical Engineering Journal 420 (2021) 129906

Table 3
The summary of phase change temperatures and enthalpy values of the microcapsules.
Sample Cycle Melting Freezing

T S/ C
◦
T P/ C
◦
TE/ C
◦
Enthalpy/(J/g) TS/◦ C TP/◦ C TE/◦ C Enthalpy/(J/g)

vapor-30 5th 226.6 232.9 237.4 53.9 166.4 143.9 134.7 53.8
50th 227.5 233.0 236.5 53.8 164.6 143.9 134.7 53.5
100th 227.7 233.0 236.4 53.0 163.0 143.9 134.7 53.0
vapor-90 5th 228.6 233.0 236.8 52.5 164.6 143.8 133.0 52.5
50th 228.7 233.0 236.3 52.5 162.6 143.9 130.9 52.5
100th 228.7 233.0 236.0 52.4 160.9 143.8 130.3 52.4
water-20 5th 226.8 233.2 237.3 53.1 166.9 143.6 137.6 53.1
50th 226.9 233.2 236.4 52.9 164.6 143.4 136.8 52.9
100th 228.1 233.2 236.4 52.8 164.7 142.8 136.4 52.8

(a )
(a) (b)
(b) (cc))

Fig. 10. A bar chart of the enthalpy values at different cycles for (a) vapor-30, (b) vapor-90 and (c) water-20.

good dispersion state, as presented in Fig. 3c, e. After heat oxidation
treatment at 500 ◦ C which was higher than the phase change tempera­
ture of Sn, the particles could still remain the original particle size and
shape, as shown in Fig. 3d, f, although the surfaces of the particles
exhibited rougher morphology than those of raw Sn particles (Fig. 3a, b).
From these images, no leakage can be found from the as-prepared
products, and the particles also exhibit good dispersion state.
Comparing vapor-30 and vapor-90, a similar surface morphology could
be observed for both samples, as shown in Fig. 3d, f. According to XRD
(Fig. 2a, b) and SEM (Fig. 3c–f) analysis, it is confirmed that the SnO2
shell was successfully formed on the Sn microspheres, forming the
Sn@SnO2 MEPCM. As for another treating method using hot water
pretreatment, very similar results to that by vapor treatment were ob­
tained. The SEM images for the samples obtained after water treatment
and subsequent heat oxidation are shown in Fig. 3g, h. After 20 min
water pretreatment, the surface morphology of the particles was also
still smooth. After heat oxidation treatment, the particle surface of
water-20 become rougher than before. The surface morphologies of the
MEPCMs are similar no matter the different treating process. The typical
cross section SEM image of the final product after heat oxidation
treatment is shown in Fig. 4, which indicates the obvious core-shell
structure with shell thickness of around 1–2 μm. The careful observa­
tion of the core-shell structure for the water treated sample is also shown
in Fig. 5. It is also observed that the shell become rough for the capsules
after heat oxidation treatment than that after water treatment, which is
consistent with the surface morphology observation in Fig. 3. These
results suggest that the Sn@SnO2 core-shell MEPCMs were successfully
prepared by the two-step methods, including vapor or water pretreat­
ment and the subsequent heat oxidation treatment.
3.2. Formation mechanism of the oxide precursor by the pretreatment
with water
Fig. 5 presents the cross sectional images and elemental analysis of
the pre-treated sample water-20-pre. The oxide shell of the water treated
sample is very thin as compared to the final heat oxidized sample. The
cross sectional SEM images (Fig. 5a, d) indicate that shell is smooth
compared to that in Fig. 4. Fig. 5b, c shows the EMPA mapping of Sn and
O, indicating that the surface of the particle is accumulated with O,
while the concentration of Sn is lower on the surface than in the core
part. The elemental line analysis from the core to surface is shown in
Fig. 5e, which further confirms the presence of a thin oxide layer on the
surface of a particle. From Fig. 5d, e it can be confirmed that the
thickness of oxide layer is around several hundreds of nanometers.
In order to further demonstrate the characteristics of the oxide pre­
cursor, we did the same hot water treatment of a carefully polished Sn
plate, which was further analyzed by XPS. It is noted that the charac­
teristics of the oxide precursor layer by vapor treatment are similar to
the hot water treated one. Fig. 6 exhibits the XPS depth profiles (0, ~400
and ~800 nm) of O1s and Sn3d for the samples of water treated Sn plate
and polished pure Sn plate. At the depth of 0 nm, the water treated Sn
plate shows stronger peaks of O1s and Sn3d, compared to those of pure
Sn plate. The O1s curves cover the bonding positions of O-Sn (530.9 eV)
and –O–H (532.8 eV) [25,26]. The O–Sn band refers to the lattice oxygen
in tin oxide, while the –O–H band corresponds to hydroxyl groups which
are attributed to the adsorbed water and hydrate water. The Sn3d curves
cover the bonding positions of Sn0 (e.g. ~484.9 eV) and Sn–O (e.g.
~486.8 eV), which refer to the bands of metallic Sn and tin oxide
respectively.
As for the polished pure Sn plate at the depth of 0 nm, the XPS curves
indicate obvious O1s peaks which is mainly located at the –O–H position
(the –O–H band is stronger than the O–Sn band), whereas the XPS curves
for Sn3d are very weak both for the Sn0 and Sn–O bands. At the depth of
~400 and ~800 nm, the peaks for O1s disappear, while the Sn3d peaks
are very strong which are located at Sn0 band. Those observation in­
dicates that the surface of the polished pure Sn plate was only weakly
oxidized with a very thin layer, and at the depth of ~400 nm the oxide
layer was not detectable. As for the water treated Sn plate at the depth of
0 nm, the XPS curves indicate very strong O1s peaks, in which the O-Sn
band is stronger than the –O–H band. In addition, the XPS curves for
Sn3d are also obviously observed, in which the Sn–O band is stronger
than the Sn0 band. At the depth of ~400 and ~800 nm, the peaks for O1s

7
N. Sheng et al.
vapor-30
(a-1) (a-2)
vapor-90
(b-1) (b-2)
water-20
(c-1) (c-2)
Fig. 11. SEM images of the capsules after 100 melting-freezing cycles. (a) vapor-30, (b) vapor-90 and (c) water-20.
become weaker both for the O–Sn and –O–H bands, while the Sn3d
peaks become stronger at the Sn0 band and the Sn–O band peak become
weaker. Even at ~800 nm, the O1s peak is still detectable. Therefore, the
oxide layer is around several hundreds of nanometers, which is similar
to the EPMA analysis in Fig. 5. Considering that the absorbed water is
only observed on the surface for pure Sn plate, however, the peaks for
the –O–H bands can be still observed for the water treated Sn plate at the
depth of 800 nm, therefore the –O–H bands must be mainly attributed
hydrate water. Therefore, through the water/vapor treatment, the oxide
precursor shell should be in the form of SnO2⋅xH2O. This is also indi­
cated by the weight decrease in the TG analysis at temperature below the
melting point of Sn. The formation of the thin oxide precursor layer is a
weak reaction with the help of hot water or vapor, which can be written
in the following formula:
Sn + O2 + xH2O → SnO2⋅xH2O
3.3. Formation mechanism of the Sn@SnO2 MEPCMs by heat oxidation
treatment
To investigate the formation of SnO2 shell, the heat oxidation process
8
Chemical Engineering Journal 420 (2021) 129906
(a-3)
(b-3)
(c-3)
to form the final capsules was also monitored by DSC-TGA analysis
under O2 flow, as shown in Fig. 7. As a comparison, the pre-treated
samples by water or vapor were also undergone the DSC-TGA analysis
under N2 flow. The samples were heated in a DSC-TGA apparatus at
10 ◦ C min− 1 to 500 ◦ C and kept at 500 ◦ C for 30 min. In the DSC curves,
all samples presented the typical endothermic peaks at around 232 ◦ C
during the temperature rise period, corresponding to the solid-liquid
phase change of Sn. In N2 atmosphere, TG curves for the water/vapor
pretreated samples could be divided into two stages. The first stage was
that the weight slightly decreased from room temperature to around
232 ◦ C, which might due to loss of absorbed/hydrated water on the pre-
formed shell. The second stage was that the weight increased linearly
and gradually above ~232 ◦ C, which might due to the slight oxidation of
Sn (weight increase of around 0.5%) because there was a small amount
of O2 in N2 flow (purity: 99.999%). In O2 atmosphere, TG curves for the
water/vapor pretreated samples could be divided into three stages. The
first stage was the slight weight decrease from room temperature to
around 232 ◦ C due to the loss of absorbed/hydrated water on the pre-
formed shell. The second stage was that the weight increased (~0.2%)
slightly and gradually until near 500 ◦ C due to the gradual oxidation and
thickening of the shells. The third stage was that the weight increased
dramatically (~2%) during the holding period at 500 ◦ C, indicating the
N. Sheng et al. Chemical Engineering Journal 420 (2021) 129906

SEM image EDS-mapping of Sn EDS-mapping of O

(a-1) (a-2) (a-3)

vapor-30

(b-1) (b-2) (b-3)

water-20

Fig. 12. Cross sectional SEM images and EDS elemental mapping of (a) vapor-30 and (b) water-20 after 100 cycles.

SEM image EPMA-mapping of Sn EPMA-mapping of O

(a) (b) (c)

10μm 10μm 10μm

counts counts
347 1547 2747 23 117 211

Fig. 13. Cross-sectional SEM image and EPMA Sn/O element distribution images for water-20 after 100 cycles.

quick oxidation and thickening of the SnO2 shell. Pure-Sn was also un­
dergone the same heat oxidation process as a comparison. Pure-Sn
shows a gradual weight increase during the temperature rise period
until 500 ◦ C, and a dramatic weight increase during the holding period
at 500 ◦ C. The oxidation of pure-Sn during the temperature rise period
was severer than those of the pre-treated samples. The pre-treated
samples with pre-formed thin shells could induce a gradual thickening
and stabilization of SnO2 shells, avoiding the breakage or crack of shells,
thereby the Sn@SnO2 MEPCMs were successfully prepared by the two-
step method, as confirmed by SEM images in Fig. 3d, f, h. However,
without the protection of pre-formed shells, the direct oxidation of pure-
Sn was sudden and uneven, and above the melting point of Sn, liquid Sn
would leak, making conglutinated and agglomerated particles, as shown
in Fig. 3b.
3.4. Heat storage and thermal cycling stability of the MEPCMs
DSC was used to measure the phase change temperature and latent
heat capacity of pure-Sn and MEPCMs as-obtained after heat oxidation
treatment. The DSC results of pure-Sn were shown in Fig. 8 and
summarized in Table 2. The latent heat capacity of pure-Sn was 55.2 J/g
and the peak temperature of melting was 233.2 ◦ C, as shown in Table 2.
The melting curves during the 5 cycles were almost overlapped. The
melting heat capacity and the melting phase change temperatures,
including starting temperature, peak temperature and ending tempera­
ture, showed no significant changes during the 5 cycles. However, for
the freezing process, with the increase of cycling number, the single
freezing curve gradually split into two peaks, presenting gradually
decreased freezing temperature. This indicates the instability of pure Sn
powders during cycling, presenting increased supercooling degree. It is
commonly recognized the encapsulated particles have higher super­
cooling degree due to the confinement effect at small size [27]. The pure
Sn powders of ~45–50 μm present a supercooling degree of around
87 ◦ C based on the melting/freezing peak temperatures. In comparison,
the supercooling degree for the Sn@SnO2 MEPCMs are around 89 ◦ C as
shown in the following measurement. The pure Sn powders were not
covered by a good protective oxide layer, thereby during cycling mea­
surement the powders could be partially oxidized due to the oxygen
residue in the DSC chamber, especially for the smaller particles. When
the smaller particles were oxidized and encapsulated, the freezing

9
N. Sheng et al.
temperature of those smaller oxidized particles would shift to the lower
temperature side, thereby with increasing cycling number the freezing
curve was split into two peaks in which a new small peak appeared at the
lower temperature side.
The melting-solidification cycling stability of the capsule samples
was conducted using DSC measurement. Fig. 9a–c show the DSC curves
at the 5th, 50th and 100th cycles for vapor-30, vapor-90 and water-20,
respectively. For each sample, the DSC curves were very similar during
the 100 cycles, indicating the good cycling stability for the capsules.
Fig. 9d presents the comparison of the DSC curves of the three samples at
the 100th cycle, in which the three curves were almost overlapped,
indicating the similarity of phase change properties for the three capsule
samples. The phase change parameters including phase change tem­
peratures and enthalpies were summarized in Table 3. These samples
exhibited melting peak temperature of around 233 ± 1 ◦ C, freezing peak
temperature of around 143 ± 1 ◦ C, and melting/freezing enthalpies
around 53 ± 1 J/g. For each sample, with the increase of cycling
number, the melting/freezing enthalpies decreased slightly, for
example, melting enthalpy was 53.9 J/g at the 5th cycle while the value
is 53.0 J/g at the 100th cycle for sample vapor-30. This decreasing ratio
of around 1.7% is in an acceptable range. These results indicate the good
thermal cycling stability of the as-prepared Sn@SnO2 core-shell
MEPCM. The comparison of enthalpy values in term of bar chart is
shown in Fig. 10.
To further investigate the thermal cycling stability of the capsules,
the morphology and structure observation were conducted. Fig. 11
shows the SEM images of those sample after 100 cycles. From these
images, it is seen that no breakage or crack of a capsule could be
observed. Those capsules could still present good dispersion state, and
no conglutination or agglomeration of particles were observed. The
surface morphologies were also similar to those of the as-prepared
samples as-shown in Fig. 3.
Fig. 12 shows the cross-sectional SEM images and the Sn/O
elemental distribution for the typical samples vapor-30 and water-20
obtained after 100 cycles of melting-freezing in DSC measurement.
Clear core-shell structure could be observed for both samples in the
cross-sectional SEM images (a-1, b-1). Elemental distribution further
confirmed the accumulation of Sn (a-2, b-2) in the cores and O (a-3, b-3)
in the shells, indicating the characteristic of metallic core and oxide shell
that is the Sn@SnO2 core-shell capsule. The core-shell structure was
further clearly observed by EPMA technique by selecting sample water-
20 after 100 cycles as an example, as shown in Fig. 11. As for a typical
cross section of a MEPCM as shown in Fig. 11a, the EMPA mapping in
term of Sn and O are presented in Fig. 11b, c. The color from black to
orange indicated the increasement of element content, as shown by the
count bars. From Fig. 11b, it was observed that the Sn element could be
detected in all areas of the microsphere. The core part with dark orange
color of high counts represented high content of Sn, while at the surface
the color was lighter with low counts which represented a relatively
ratio of Sn. Fig. 11c shows the EPMA mapping of O, in which O element
was accumulated on the edge of the microsphere, forming an oxide shell
coated on the Sn core. The thickness of the oxide shell was around 1–2
μm. All of these characterizations further confirmed the good cycling
stability of Sn@SnO2 MEPCMs, which is promising for long term
repeated use (Fig. 13.).
Finally, the thermal conductivity of the samples was also measured
by pressing the capsule powders to disks. The Sn@SnO2 MEPCMs have a
thermal conductivity of around 4.2 W/m⋅K, which is lower than the
counterpart of pure Sn powders of 5.6 W/m⋅K. It is normal that the
thermal conductivity of the compacted powders is lower than that of
pure bulk Sn (67 W/m⋅K).
4. Conclusion
In summary, the Sn@SnO2 core-shell microcapsules were success­
fully prepared using Sn microspheres of 45–50 μm by a facile two-step
10
Chemical Engineering Journal 420 (2021) 129906
self-oxidation strategy including vapor or water pretreatment and sub­
sequent heat oxidation treatment. The crystal phase, composition and
microstructure observation by XRD, SEM-EDS and EPMA analysis sug­
gested that the metallic Sn core was well coated with SnO2 shell of 1–2
μm. DSC thermal analysis indicated that the Sn@SnO2 capsules exhibi­
ted a melting point of ~232 ◦ C and latent heat of ~53 J/g. Thermal
cycling measurement illustrated that after 100 cycles of melting-
freezing, the phase change properties and core-shell structure of the
capsules could be well remained, particularly, no crack or leakage was
observed. Therefore, by solving the critical problems of leakage and
corrosion of metallic PCM, the newly developed Sn@SnO2 core-shell
MEPCMs are promising for application in middle-temperature thermal
energy storage.
Declaration of Competing Interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
the work reported in this paper.
Acknowledgments
This work was supported partially by National Natural Science
Foundation of China (No. 52006236), Natural Science Foundation of
Jiangsu Province (No. BK20200656), the Fundamental Research Funds
for the Central University (Grant no. 2020QN08) and China Postdoctoral
Science Foundation (Grant no. 2020M681764).
References
[1] A. Dahash, F. Ochs, M.B. Janetti, W. Streicher, Advances in seasonal thermal
energy storage for solar district heating applications: a critical review on large-
scale hot-water tank and pit thermal energy storage systems, Appl. Energy 239
(2019) 296–315.
[2] C. Suresh, R.P. Saini, Thermal performance of sensible and latent heat thermal
energy storage systems, Int. J. Energy Res. 44 (6) (2020) 4743–4758, https://doi.
org/10.1002/er.5255.
[3] H. Nazir, M. Batool, F.J. Bolivar Osorio, M. Isaza-Ruiz, X. Xu, K. Vignarooban,
P. Phelan, Inamuddin, A.M. Kannan, Recent developments in phase change
materials for energy storage applications: a review, Int. J. Heat Mass Transf. 129
(2019) 491–523, https://doi.org/10.1016/j.ijheatmasstransfer.2018.09.126.
[4] C. Li, Q.i. Li, Y. Ding, Investigation on the effective thermal conductivity of
carbonate salt based composite phase change materials for medium and high
temperature thermal energy storage, Energy 176 (2019) 728–741, https://doi.org/
10.1016/j.energy.2019.04.029.
[5] H. Zhang, D. Shin, S. Santhanagopalan, Microencapsulated binary carbonate salt
mixture in silica shell with enhanced effective heat capacity for high temperature
latent heat storage, Renew. Energy 134 (2019) 1156–1162, https://doi.org/
10.1016/j.renene.2018.09.011.
[6] C. Wickramaratne, J.S. Dhau, R. Kamal, P. Myers, D.Y. Goswami, E. Stefanakos,
Macro-encapsulation and characterization of chloride based inorganic Phase
change materials for high temperature thermal energy storage systems, Appl.
Energy 221 (2018) 587–596, https://doi.org/10.1016/j.apenergy.2018.03.146.
[7] D. Singh, W. Yu, W. Zhao, T. Kim, D.M. France, R.K. Smith, Development and
prototype testing of MgCl2/graphite foam latent heat thermal energy storage
system, Sol. Energy 159 (2018) 270–282, https://doi.org/10.1016/j.
solener.2017.10.084.
[8] Y. Ren, P. Li, M. Yuan, F. Ye, C. Xu, Z. Liu, Effect of the fabrication process on the
thermophysical properties of Ca(NO3)2–NaNO3/expanded graphite phase change
material composites, Sol. Energy Mater. Sol. Cells 200 (2019), 110005.
[9] G. Zhang, J. Li, Y. Chen, H. Xiang, B. Ma, Z. Xu, X. Ma, Encapsulation of copper-
based phase change materials for high temperature thermal energy storage, Sol.
Energy Mater. Sol. Cells 128 (2014) 131–137, https://doi.org/10.1016/j.
solmat.2014.05.012.
[10] Z. Wang, H. Wang, M. Yang, W. Sun, G. Yin, Q. Zhang, Z. Ren, Thermal reliability
of Al-Si eutectic alloy for thermal energy storage, Mater. Res. Bull. 95 (2017)
300–306, https://doi.org/10.1016/j.materresbull.2017.07.040.
[11] H. Wei, C. Wang, S. Yang, J. Han, M. Yang, J. Zhang, Y. Lu, X. Liu, Novel core/
void/shell composite phase change materials for high temperature thermal energy
storage, Chem. Eng. J. 391 (2020), 123539.
[12] E. Risueño, A. Faik, A. Gil, J. Rodríguez-Aseguinolaza, M. Tello, B. D’Aguanno,
Zinc-rich eutectic alloys for high energy density latent heat storage applications,
J. Alloys Compd. 705 (2017) 714–721, https://doi.org/10.1016/j.
jallcom.2017.02.173.
[13] S. Cingarapu, D. Singh, E.V. Timofeeva, M.R. Moravek, Use of encapsulated zinc
particles in a eutectic chloride salt to enhance thermal energy storage capacity for
N. Sheng et al.
concentrated solar power, Renew. Energy 80 (2015) 508–516, https://doi.org/
10.1016/j.renene.2015.02.026.
[14] S. Zhu, M.T. Nguyen, T. Tokunaga, T. Yonezawa, Size-tunable alumina-
encapsulated Sn-based phase change materials for thermal energy storage, ACS
Appl. Nano Mater. 2 (6) (2019) 3752–3760, https://doi.org/10.1021/
acsanm.9b00649.s001.
[15] C.C. Lai, W.C. Chang, W.L. Hu, Z.M. Wang, M.C. Lu, Y.L. Chueh, A solar-thermal
energy harvesting scheme: enhanced heat capacity of molten HITEC salt mixed
with Sn/SiO(x) core-shell nanoparticles, Nanoscale 6 (2014) 4555–4559.
[16] R. Fukahori, T. Nomura, C. Zhu, N. Sheng, N. Okinaka, T. Akiyama, Macro-
encapsulation of metallic phase change material using cylindrical-type ceramic
containers for high-temperature thermal energy storage, Appl. Energy 170 (2016)
324–328, https://doi.org/10.1016/j.apenergy.2016.02.106.
[17] C. Zhou, S. Wu, Medium- and high-temperature latent heat thermal energy storage:
material database, system review, and corrosivity assessment, Int. J. Energy Res.
(2018).
[18] F. Xue, Y.u. Lu, X.-D. Qi, J.-H. Yang, Y. Wang, Melamine foam-templated graphene
nanoplatelet framework toward phase change materials with multiple energy
conversion abilities, Chem. Eng. J. 365 (2019) 20–29, https://doi.org/10.1016/j.
cej.2019.02.023.
[19] K. Yuan, J. Liu, X. Fang, Z. Zhang, Novel facile self-assembly approach to construct
graphene oxide-decorated phase-change microcapsules with enhanced photo-to-
thermal conversion performance, J. Mater. Chem. A 6 (10) (2018) 4535–4543,
https://doi.org/10.1039/C8TA00215K.
[20] C. Liu, J. Zong, J. Zhang, D. He, C. Guo, B. Xu, Z. Rao, Knitting aryl network
polymers (KAPs)-embedded copper foam enables highly efficient thermal energy
11
Chemical Engineering Journal 420 (2021) 129906
storage, J. Mater. Chem. A 8 (30) (2020) 15177–15186, https://doi.org/10.1039/
D0TA04626D.
[21] S.B. Ye, Q.L. Zhang, D.D. Hu, J.C. Feng, Core-shell-like structured graphene aerogel
encapsulating paraffin: shape-stable phase change material for thermal energy
storage, J. Mater. Chem. A 3 (2015) 4018–4025.
[22] S.L. Bhanawase, G.D. Yadav, Green synthesis of vanillyl mandelic acid (sodium
salt) from guaiacol and sodium glyoxylate over novel silica encapsulated
magnesium hydroxide, ACS Sustain. Chem. Eng. 4 (4) (2016) 1974–1984, https://
doi.org/10.1021/acssuschemeng.5b01219.s001.
[23] F. Zhang, Y. Zhong, X.u. Yang, J. Lin, Z. Zhu, Encapsulation of metal-based phase
change materials using ceramic shells prepared by spouted bed CVD method, Sol.
Energy Mater. Sol. Cells 170 (2017) 137–142, https://doi.org/10.1016/j.
solmat.2017.05.061.
[24] C.-C. Lai, S.-M. Lin, Y.-D. Chu, C.-C. Chang, Y.-L. Chueh, M.-C. Lu, Tunable
endothermic plateau for enhancing thermal energy storage obtained using binary
metal alloy particles, Nano Energy 25 (2016) 218–224, https://doi.org/10.1016/j.
nanoen.2016.02.058.
[25] V. Bonu, A. Das, M. Sardar, S. Dhara, A.K. Tyagi, Surface functionalization-
enhanced magnetism in SnO2 nanoparticles and its correlation to
photoluminescence properties, J. Mater. Chem. C 3 (2015) 1261–1267.
[26] M. Kwoka, B. Lyson-Sypien, E. Comini, M. Krzywiecki, K. Waczynski, J. Szuber,
Surface properties of SnO2 nanolayers prepared by spin-coating and thermal
oxidation, Nanotechnology 31 (2020), 315714.
[27] X. Tong, N. Li, M. Zeng, Q. Wang, Organic phase change materials confined in
carbon-based materials for thermal properties enhancement: Recent advancement
and challenges, Renew. Sustain. Energy Rev. 108 (2019) 398–422, https://doi.org/
10.1016/j.rser.2019.03.031.