Electrochimica Acta 182 (2015) 871–877

Contents lists available at ScienceDirect

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

Influence of the preparation methods on the electrochemical

properties and structural changes of alpha-sodium iron oxide as a
positive electrode material for rechargeable sodium batteries
R. Kataoka, K. Kuratani ∗ , M. Kitta, N. Takeichi, T. Kiyobayashi, M. Tabuchi
Research Institute for Electrochemical Energy, Department of Energy and Environment, National Institute of Advanced Industrial Science and Technology
(AIST), 1-8-31 Midorigaoka, Ikeda, Osaka 563-8577, Japan

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

Article history: A potential positive electrode material, ˛-NaFeO2 , can be viewed one of the base material for rechargeable
Received 7 August 2015 Na batteries. In this study, we found that its structural evolution during charge process differentiates
Received in revised form 8 September 2015 depending on its synthesis route. The ˛-NaFeO2 is prepared by three methods: one is by the hydrothermal
Accepted 17 September 2015
technique (HT), and the other two by solid-state reactions, SSn and SS , in which nano-sized Fe3 O4 and
Available online 26 September 2015
micro-sized Fe3 O4 are used as the Fe source, respectively. The synchrotron radiation X-ray diffraction
profiles of the initial state suggest that all three samples are apparently identified as the target product,
Keywords:
˛-NaFeO2 , with rhombohedral structures. Though all three samples remain stable for more than 50
Sodium ion battery
Iron-based material
cycles under the cycling conditions of a constant amount of charging (Na extraction) to 70 mA h g−1 , their
Layered structure structural evolution differs during the sodium extraction. The SS transitions from the rhombohedral
Cycle life structure to a spinel-like cubic structure, while the HT transforms into a monoclinic structure. On the
Phase transition other hand, both the monoclinic and cubic phases co-exist in SSn . What produces these differences among
the seemingly identical materials is discussed.
© 2015 Elsevier Ltd. All rights reserved.

1. Introduction ˛-NaFeO2 transitions to the electrochemically inert ˇ-NaFeO2 is

relatively low (∼1050 K), thus the material cannot be annealed
Rechargeable sodium batteries are drawing increasing attention at high temperatures to obtain a homogeneous phase, (ii) Unlike
[1–3] as an alternative to the rechargeable lithium batteries in the the Li counterparts, ˛-NaFeO2 is prone to deteriorate by reac-
hope of circumventing the following material resource problems ting with water, thus more rigorous control of the preparation
at the expense of the unfavorable 0.3 V higher redox potential of atmosphere is required. The difficulties in the preparation are pre-
Na than Li: (i) Lithium is unevenly distributed around the world, sumably reflected in the inconsistent observations concerning the
thus its supply can be precarious depending on the global geopo- phase transitions as reported in the literature [18,19], which are
litical volatility, (ii) Fe-based substances, e.g., NaFeO2 , can be used addressed in Section 3.5.
as a less resource-intensive material for the positive electrode of The objective of the present study is to describe our observations
the rechargeable Na batteries unlike the Li counterparts in which that, even if the X-ray diffraction profiles suggest the successful
LiFeO2 is not electrochemically active while costly LiCoO2 is. A synthesis of ˛-NaFeO2 , and their electrochemical cycle properties
layered metal oxide, ˛-NaFeO2 , with sheets of edge-sharing FeO6 are similar, the associated phase transitions of ˛-NaFeO2 are signif-
octahedron, can be thus viewed as one of the base materials to icantly influenced by the small amount of impurities in the samples
which a series of layered oxides is compared as a positive elec- or their crystallinities. Three samples were compared; one sample
trode material for the rechargeable Na batteries [4–17]. However, was prepared by a hydrothermal method, and two by the solid-
obtaining a qualified ˛-NaFeO2 is more laborious than LiCoO2 state method in which nano-sized Fe3 O4 or micro-sized Fe3 O4 was
because of the following reasons: (i) The temperature above which used as the iron source. The ˛-NaFeO2 samples were subjected to
synchrotron X-ray diffraction (SR-XRD), Rietveld analysis, scanning
electron microscopy (SEM), particle size distribution measurement,
etc. Aside from the conventional charge and discharge test, the
∗ Corresponding author. Tel.: +81 72 751 9489; fax: +81 72 751 9629. galvanostatic intermittent titration technique (GITT) was used to
E-mail address: [email protected] (K. Kuratani). evaluate the chemical diffusion coefficient.

http://dx.doi.org/10.1016/j.electacta.2015.09.092
0013-4686/© 2015 Elsevier Ltd. All rights reserved.
872 R. Kataoka et al. / Electrochimica Acta 182 (2015) 871–877

2. Experimental 2.4. Electrochemical properties

2.1. Sample Preparation The electrochemical properties of the samples were evaluated as
a positive electrode material by forming the two-electrode setup in
Hydrothermal method. The ˛-NaFeO2 sample was prepared an IEC R2032 coin-type cell. The working electrode was prepared by
by modifying our previously reported hydrothermal method mixing the ˛-NaFeO2 sample, Ketjenblack as the conductive addi-
[21]. A 134.67 g (0.33 mol) sample of iron(III) nitrate nonahy- tive and polytetrafluoroethylene (PTFE) as the binder in the weight
drate (Fe(NO3 )3 · 9H2 O, Wako Pure Chemical Industries) and 60 g ratio of 84: 8: 8 along with adding a few drops of hexane to facil-
(1.5 mol) of sodium hydroxide (NaOH, Wako) were dissolved into itate the mixing. The mixture was pressed on an Al mesh current
500 and 1000 mL of distilled water, respectively. A precipitate collector (16 mm) and kept at 493 K for 10 h under a vacuum to
was formed by dropwise adding Fe(NO3 )3 (aq) into NaOH(aq) thoroughly remove the hexane and adsorbed water. The amount
with bubbling air at room temperature. The precipitate was of ˛-NaFeO2 in each electrode was ca. 10 mg. A lump of metallic
then transferred to 100 mL of highly concentrated NaOH(aq) in sodium (Aldrich) was pressed into a foil and punched out to form
which 270 g of NaOH was dissolved, then hydrothermally treated the counter electrode (12 mm, ca. 40 mg) which provides more
at 493 K for 20 h in an autoclave. The product was repeat- than a 20 times excess amount of Na to serve as a reservoir for the
edly washed with ethanol to eliminate any residual NaOH, then working electrode. The electrolyte was 1.0 mol L−1 NaPF6 in ethyl-
dried overnight in air at 323 K. The obtained sample was labeled ene carbonate/diethyl carbonate (EC/DEC, 1:1 by volume, Kishida),
“HT”. and the separator was a glass filter (Advantec, GA-100). The coin cell
Solid-state method. We used homemade nano-sized Fe3 O4 and was assembled in a dry-box of which the dew point was maintained
purchased micro-sized triiron tetraoxide (Fe3 O4 , Kishida Chemi- below 223 K.
cal) as the iron source, and sodium peroxide (Na2 O2 , Wako) as the Cyclic voltammetry (CV) was applied to three electrodes using
sodium source. The nano-sized Fe3 O4 was prepared by dropwise an electrochemical analyzer (Solartron 1280, SOlartron Co. Ltd.) The
adding 1.0 mol L−1 of FeSO4 (aq, 100 mL) to 1.0 mol L−1 of NaOH(aq, voltammograms was recorded at the scan speed of 1 mV s−1 in the
200 mL) at room temperature. A small amount of citric acid(∼0.3 g) potential range between 1.5 and 3.8 V(vs. Na+ /Na).
was added into the FeSO4 solution in order to prevent Fe2+ from The cell was charged and discharged using a BLS5500 (Keisokuki
oxidizing to Fe3+ . The precipitate was washed with distilled water Center) at the current density of 10 mA (g-NaFeO2 )−1 in the elec-
until the filtrate was neutral, then subsequently dried overnight at trode. The other conditions of the electrochemical measurement,
353 K in air to yield nano-sized Fe3 O4 . A 0.77 g sample of Fe3 O4 such as the cut-off potentials, etc., are specified in the results and
was hand-milled with 0.39 g of Na2 O2 in an Ar-filled glove box to discussion section.
avoid its reaction with water and CO2 . The mixture placed in a gold The chemical diffusion coefficient of the samples during the ini-
crucible was calcined in air for 10 h at 903 K for the nano-sized tial charge process was evaluated by GITT [23]. The titrations were
Fe3 O4 and at 923 K for micro-sized Fe3 O4 . The calcined product was carried out by applying square current pulses of 3 mA g−1 for 2 hs,
pulverized and used without further treatment. The two samples followed by potential relaxation step of 1 h at the open circuit volt-
prepared with nano-sized Fe3 O4 and with micro-sized Fe3 O4 were age. The procedure was continued until 0.5 Na ion was extracted
labeled “SSn ” and “SS ”, respectively. from ˛-NaFeO2 .

2.2. Synchrotron radiation X-ray diffraction 3. Results and discussion

The crystallographic structures of the samples were charac- 3.1. The Fe3 O4 starting ingredient
terized by high-resolution synchrotron radiation at the X-ray
diffraction (SR-XRD) facility (BL19B2 beam line, SPring-8, Japan). Fig. 1 compares the SEM images of the homemade and the com-
The samples were placed in a glass capillary with the outer diameter mercially available Fe3 O4 . Both samples are confirmed to be Fe3 O4
of 0.5 mm. The SR-XRD data were collected using a Debye-Scherrer with Fd3m from their XRD patterns (see Fig. S1 in supplementary
camera with an imaging plate. The diffraction angle, 2, ranged information). The primary particle size of the homemade Fe3 O4 is
from 5.00 to 78.00◦ requiring 300 s. The X-ray wavelength was nano-sized (less than 60 nm). On the other hand, that of the pur-
calibrated to /pm = 69.95 using CeO2 as the standard material. chased one is micro-sized (about 1 m). The primary particles of
The XRD profiles of the as-prepared samples were subjected to the nano-sized Fe3 O4 aggregate to form the secondary particles of
a Rietveld analysis, of which the details are described in Section more than 3 m, while the purchased Fe3 O4 particles do not form
3.2. For the XRD measurement after the electrochemical tests, the any secondary particles.
coin cell (see Section 2.4) was disassembled in a dry-box to remove
the electrode, which was then washed with dimethyl carbonate to
3.2. Crystallographic structure of the as-prepared NaFeO2
remove the residual electrolyte and peeled off the Al mesh current
collector.
Fig. 2 shows the SR-XRD profiles of the as-prepared samples pre-
pared by three methods; i.e., (a) the hydrothermal method (HT),
2.3. Surface area, particle morphology and size distribution (b) the solid-state method using nano-sized Fe3 O4 (SSn ) and (c)
the solid-state method using micron-sized Fe3 O4 (SS ) as the iron
The particle morphology of the as-prepared samples was source. The XRD profiles indicate that the three samples contain
observed by a scanning electron microscope (SEM, JEOL-6301F). the rhombohedral ˛-NaFeO2 (JCPDS #20-1115, space group R3m,
The specific surface areas of the samples were determined by or the “O3” structure in the classification of layered oxides pro-
the Brunauer-Emmett-Teller (BET) method using a BELSORP-mini posed by Delmas et al.[5]) as the main phase. While no extra peaks
(BEL Japan, Inc.). The particle size distribution was measured ascribed to impurity phases are detected in the HT, some tiny extra
by the dry method using an SPR-7340 (NIKKISO Co., Ltd.). peaks attributable to ˇ-NaFeO2 and Fe3 O4 are detected in SSn and
For the surface area and particle size distribution, we con- SS (see below for more information about the impurities). Calcu-
firmed the reproducibility of the result based on several sample lated from the peak widths of the (003) plane based on the Scherrer
batches. equation, one finds that the crystallite sizes of the HT and SSn are
 R. Kataoka et al. / Electrochimica Acta 182 (2015) 871–877 873

Fig. 1. Scanning electron microscope images of (a) homemade nano-sized Fe3 O4 and (b) commercially available micro-sized Fe3 O4 .

Table 2
Crystallographic structure model for the Rietveld refinement of ˛-NaFeO2 belonging
to the R3m space group.

Atom Site Occupancy Coordinates

X Y Z

Na 3a g(Na) 0 0 0
Fe 3b 1.0 0 0 1/2
O 6c 1.0 0 0 z(O)

The detailed crystallographic structures of the as-prepared sam-

ples were analyzed by the Rietveld modeling based on the XRD
profiles in Fig. 2. In the structure model of ˛-NaFeO2 , given in
Table 2, the occupancy of Na, g(Na), was treated as a variable to
see whether or not Na is partially lost during the preparation.
We also evaluated the mass fraction of Fe3 O4 and ˇ-NaFeO2 con-
tained as impurities in the samples. The structural parameters of
the impurity phases, such as the atomic coordinates, etc., were fixed
Fig. 2. Synchrotron radiation X-ray diffraction profiles of NaFeO2 prepared by according to the literature [24,25] during the analysis.
(a) hydrothermal method (HT), (b) solid-state method with nano-sized Fe3 O4 (SSn )
and (c) with micro-sized Fe3 O4 (SS ). /pm = 69.95. The inset figure is the magnified
Table 3 summarizes the results of the Rietveld analysis. The
profiles in the 2 range between 14.0◦ and 17.5◦ . The peaks indicated by the solid reliability factors of weighted pattern (Rwp ), of the Bragg(RB ) and
and dashed arrows are of the Fe3 O4 and ˇ-NaFeO2 impurities, respectively. the S-value, that reflect the degree of convergence of the refine-
ment, are around 6.7-8.5%, 1.8-3.0% and 2-2.7, respectively, which
we consider satisfactory. The mass fractions of the Fe3 O4 impurity
comparable, while they are significantly greater than that of SS
in SSn and SS are 0.3 and 5.5 wt.%, respectively. In addition, SS
(see Table 1). The higher degree of crystallite growth of SSn than of
contains 1.6 wt.% ˇ-NaFeO2 , which is an electrochemically inactive
SS , despite the similar calcination temperature, implies the higher
phase. The lattice parameters of ˛-NaFeO2 for the three samples
reactivity of the nano-sized Fe3 O4 than the micro-sized Fe3 O4 in the
well coincide with those previously reported [18,19], We consider
solid-state reaction.
x ∼ 1 in Nax FeO2 for all the samples, since the Na occupancy, g(Na),
Table 1 in the ˛-NaFeO2 phase is calculated to be ≥96%.
Particle characteristics of the as-prepared ˛-NaFeO2 powders.

HT SSn SS 3.3. Particle morphology of the as-prepared ˛-NaFeO2 powders

FWHMa , W 003/deg 0.042 0.050 0.84
Crystallite sizeb , Lcr /nm 90 75 45 The SEM images, as shown in Fig. 3, reveal that the particle sizes
Surface areac , ABET /m2 g−1 0.83 2.98 2.14 of HT and SS are relatively uniform while that of SSn is dispersed.
a
Full width at half maximum of the diffraction peak from the (003) plane in Fig. 2. Comparing the SEM images of SSn and SS with those in Fig. 1 in
b
Calculated from W003 based on the Scherrer relation. Section 3.1, one finds that the particle morphology of each sample
c
Determined by the N2 adsorption at 77 K. prepared by the solid-state reaction resembles that of the starting
874 R. Kataoka et al. / Electrochimica Acta 182 (2015) 871–877

Table 3
Result of the Rietveld refinement of the as-prepared ˛-NaFeO2 .

HT SSn SS

Lattice parameters in nm
a 0.30255(1) 0.30254(2) 0.30235(1)
c 1.60985(1) 1.60954(7) 1.60836(6)
Site occupancies
g(Na) 1.00(3) 0.983(2) 0.959(5)
Atomic coordinates
z(O) 0.2338(1) 0.2376(2) 0.2369(2)
Impurities in wt%*
Fe3 O4 — 0.3 5.5
ˇ-NaFeO2 — 1.6 —
Quality of the refinement
Rwp (%) 6.78 7.553 8.48
RB (%) 1.81 1.85 3.01
S ≡ Rwp /Re 2.00 2.39 2.69

Structure model and optimized variables are given in Table 2. *Entries without value
(—) signify that the inclusion of each impurity into the Rietveld analysis resulted in a
negative scale factor. Rwp : R-factor of the weighted pattern. RB : R-factor of the Bragg
pattern. Re : Theoretically expected lower limit of the R-factor.

material, Fe3 O4 ; that is, SSn consists of nano-sized primary par-

ticles which aggregate to form the secondary particles of m-size
whereas SS consists of 1 m-sized primary particles which barely
aggregate.
Regarding the particle size distribution (f(D)), Fig. 4 suggests that
the HT and SS powders are mono-dispersed and well follow the
logarithmic normal distribution reflecting the size of the primary
particles. On the other hand, f(D) of SSn is irregularly and widely dis-
persed in a micrometers-range reflecting the size of the secondary
particles. The particle size distributions of SSn (b) and SS (c) exactly
replicate that of the nano-sized Fe3 O4 (d) and micro-sized Fe3 O4 (e),
respectively.
The specific surface areas of samples were evaluated based on
BET method and the results are summarized in Table 1 (The N2 -
adsorption isotherms are shown in Fig. S2.)
The trend in the specific surface area, is SSn > SS > HT, appar-
ently reflecting the inverse order of the primary particle size.
The fact that SSn and SS resemble their precursor Fe3 O4
in particle size (either primary or secondary) corroborates that
previously suggested by Takeda et al.; namely, the solid-state
reaction topotactically proceeds on the basis of Fe3 O4 [20]. The
hydrothermal method, on the other hand, passes through a
dissolution-precipitate process, so that the product particle size is
considered independent of the starting ingredients.

3.4. Electrochemical properties

Fig. 5 shows the cyclic voltammograms of the cells with the posi-
tive electrodes made of the three ˛-NaFeO2 samples. During anodic Fig. 3. Scanning electron microscope images of as-prepared ˛-NaFeO2 powders: (a)
hydrothermal method (HT), (b) solid-state method with nano-sized Fe3 O4 (SSn ) and
sweep, oxidization current of three samples starts to increase at
(c) solid-state method with micro-sized Fe3 O4 (SS ). Scale bar indicates 3 m.
around 3.3 ∼ 3.4 V (vs. Na+ /Na), indicating extraction of Na ion from
˛-NaFeO2 . The corresponding reduction current were also detected
during cathodic sweep in the potential range between 2.3 and reported [19,22], the potential plateau reflects the oxidation of
3.4 V(vs. Na+ /Na), but the profiles are different from each other. Fe3+ in ˛-NaFeO2 to Fe4+ and the concurrent Na+ extraction during
The two peaks are detected in the profiles of SSn and SS while a charging, and vice versa during discharging, i.e., Fe4+ → Fe3+ and Na+
single peak is observed from that of HT. The difference in the CV insertion. The above mentioned conditions of 70 mA h g−1 charg-
profiles is discussed later together with the change in the charge ing corresponds to 0.29 mole equivalent of Na+ extraction from the
and discharge curves. NaFeO2 formula. The changes in the charge and discharge curves
Fig. 6 shows the charge and discharge curves of the cells with through cycling shown in Fig. 6 are discussed in the next paragraph.
positive electrodes made of the three ˛-NaFeO2 samples. The cell The result of the charge and discharge cycling test is shown
was charged with a fixed amount of charging to 70 mA h g−1 and in Fig. 7. The cell was cycled with a constant amount of charg-
discharged to a fixed cut-off potential of 1.5 V(vs. Na+ /Na). The test ing to 70 or 100 mA h g−1 (i.e., 0.29 or 0.42 mole Na extraction
conditions were switched to the constant cut-off potential mode from NaFeO2 , respectively) unless the charging potential reached
ranging from 1.5 to 4.0 V(vs. Na+ /Na) when the electrode failed 4.0 V(vs. Na+ /Na). The discharge process was fixed to a cut-off
to deliver 70 mA h g−1 during the charge. For all three samples, potential of 1.5 V (vs. Na+ /Na) throughout the cycling. The three
a potential plateau is observed at 3.3 V(vs. Na+ /Na). As already electrodes remain stable for more than 50 cycles under the
 R. Kataoka et al. / Electrochimica Acta 182 (2015) 871–877 875

Fig. 4. Diameter distribution of the secondary particle of the as-prepared NaFeO2

powders and the starting materials, Fe3 O4 , for the solid-state method for compari- Fig. 7. Cycle trends of the positive electrodes with ˛-NaFeO2 prepared by three
son. methods. Cycle conditions: charging 70 or 100 mA h g−1 , discharging ≥1.5 V(vs.
Na+ /Na).

conditions of 70 mA h g−1 charging, but their discharge curves were

different after the cycling. As shown in Fig. 6, only the increase in
the electrochemical polarization is observed in the charge and dis-
charge curve of HT after 50 cycles. For SSn and SS , on the other
hand, the width of the potential plateau at 3.2 V(vs. Na+ /Na) sig-
nificantly decreases and an inflection, which is also detected from
CV profile, appears at 2.5 V(vs. Na+ /Na) in the discharge curves after
50 cycles, suggesting a side reaction within the active material other
than the Na insertion into ˛-NaFeO2 . In contrast, the electrodes
tested under the conditions of 100 mA h g−1 charging significantly
deteriorate and fail to be charged to 100 mA h g−1 within a few
cycles (Fig. 7).

3.5. Chemical diffusion of Na+ ion in ˛-NaFeO2

Fig. 5. Cyclic voltammograms of ˛-NaFeO2 prepared by three method in the poten-
tial range between 1.5∼3.8 V(vs. Na+ /Na) at the scan speed of 1 mV s−1 .
The kinetics of the Na+ at each state of charge and its influence
on the electrochemical performance were investigated by the GITT.
Based on the GITT measurement, the diffusion coefficient of the
sodium ion (DNa+ ) is calculated as follows [23]:

4
 m V 2  E 2
B M s
DNa + = (1)
 MB S Et
where mB , MB and VM are the mass, formula weight and molar vol-
ume of the host material through which Na diffuses, respectively,
and S is the area of the sample-electrolyte interface. Et and Es
are the transient potential changes responding to the current pulse
of  as illustrated in Fig. 8 (a). Fig. 8 (b) shows experimental GITT
curves of HT, SSn and SS and (c) shows the calculated D Na+ of
the samples at various Na concentrations, in which we approxi-
mately considered the host material to be NaFeO2 and S/mB = ABET .
We can find, by and large, that DNa+ of each sample falls in the
range of 10−14 -10−13 cm2 s−1 when x is in the range of 0 and
0.3 in Nax FeO2 , which correspond to the capacity range of 0 and
70 mA h g−1 , respectively. During a further Na extraction, DNa+ of
all three electrodes starts to decrease when x exceeds 0.3 (SSn and
SS ) and 0.4 (HT), above which the decay in the cycle stability is
pronounced as shown in Fig. 7.

3.6. Crystallographic evolution of ˛-NaFeO2 through the Na

extraction and insertion
Fig. 6. Charge and discharge profiles of the positive electrodes with ˛-NaFeO2 pre-
pared by three methods: (a) hydrothermal (HT), (b) solid-state with nano-sized
Fe3 O4 (SSn ) and (c) solid-state with micro-sized Fe3 O4 (SS ). Profiles of the 1st, 2nd, The SR-XRD profiles in Fig. 9 were taken after 0.50 mole equiva-
10th and 50th-cycle are presented. Cycle conditions: charging of 70 mA h g−1 and lent of Na was extracted from (a) HT, (b) SSn and (c) SS by charging
discharging ≥1.5 V(vs. Na+ /Na).
the cells to 120 mA h g−1 . The structures of the three samples
876 R. Kataoka et al. / Electrochimica Acta 182 (2015) 871–877

Table 4
Lattice parameters of present phases for the samples after 0.5 Na extraction.

present phase Space Group a/nm b/nm c/nm ˇ/◦

(a) HT
˛-NaFeO2 R3m 0.295(2) 0.295(2) 1.617(2) 120*
monoclinic phase C2/m 0.5268(10) 0.28887(11) 0.592(4) 108.5(3)
(b) SS
˛-NaFeO2 R3m 0.2974(4) 0.2974(4) 1.544(5) 120*
Monoclinic phase C2/m 0.528(4) 0.2902(4) 0.5883(12) 108.9(3)
Spinel-like Fd3m 0.848(3) 0.848(3) 0.848(3) 90*
(c) SSn
˛-NaFeO2 R3m 0.292(6) 0.292(6) 1.62(6) 120*
Monoclinic C2/m 0.5258(2) 0.29028(6) 0.5908(2) 109.7(4)
*
Values are uniquely determined depending on their space group.

Fig. 9. Synchrotron radiation X-ray diffraction profiles of the three positive elec-
trodes after charging to 120 mA h g−1 , corresponding to 0.5 Na extraction from
NaFeO2 : (a) HT, (b) SSn and (c) SS. /pm = 69.95. The peaks marked with open cir-
cle, closed triangle and closed square in figures are rhombohedral phase, monoclinic
phase and spinel-like cubic phase, respectively.

The lattice constants of the phases observed from three diffrac-

tion patterns were summarized in Table 4
Tables 1 and 3 imply a tendency that the low crystallinity
of ˛-NaFeO2 and the Fe3 O4 residue in it seem to induce the
phase transition from R3m to Fd3m. Notice that, different though
their lattice parameters are, Fe3 O4 also belongs to Fd3m. While
Fig. 8. (a) Schematic illustration of a single step of the galvanostatic intermittent the crystallographic structure of HT starts to change when about
technique (GITT). (b) Chemical diffusion coefficient of Na+ in ˛-NaFeO2 prepared by 0.4 mole Na ion is extracted, the phase transition of SSn is observed
the three methods. (circle: HT, triangle: SSn and square: SS ).
after the 0.3 mole Na extraction (see Supplementary informa-
tion). When compared with the result of GITT, one finds the
evolve differently. HT transforms from the rhombohedral R3m to a state of charge at which the crystal structure of the samples
monoclinic structure belonging to the C2/m space group with some starts to change roughly coincides with that at which the DNa+
amount of the R3m phase still remaining. The phase transition of starts to decline. The significantly lower diffusivity of Na+ in
NaFeO2 (R3m) to Na0.5 FeO2 (C2/m) was reported by Takeda et al. either C2/m or Fd3m than in R3m seems to eventually result in
[19] who prepared the initial ˛-NaFeO2 by the solid-state method the capacity fading when more than 0.3-0.4 Na is extracted from
using the same materials as SSn and SS in the present study; i.e., NaFeO2 .
Na2 O2 and Fe3 O4 . On the other hand, SS transforms from the R3m
phase to a spinel-like structure belonging to the Fd3m after the 4. Conclusion
0.5 Na extraction. SSn behaves intermediately between HT and SSn ,
i.e., both the C2/m and the Fd3m phases are detected. Yabuuchi As shown in the XRD profiles (Fig. 2), by either the hydro-
et al. [18] also prepared ˛-NaFeO2 by the solid-state method from thermal method (HT) or the solid-state method (SSn and SS ),
the same materials as Takeda et al.[19] and as our sample pre- we succeeded in preparing ˛-NaFeO2 that can serve as a posi-
pared by solid state method. They did not observe, however, the tive electrode active material for rechargeable Na batteries (Fig. 6).
phase transition from the R3m to the C2/m through charging. They However, their crystallographic structures evolve differently dur-
instead suspect that an Fe ion migration during the Na-extraction ing the charge (Na-extraction) process. When 0.5 mole equivalent
gives rise to the irreversible structural change in ˛-NaFeO2 . If we of Na is extracted from NaFeO2 by charging to 120 mA h g−1 , HT
interpret the Fe-migration as the formation of a spinel-like phase, transitions from R3m to C2/m, while SS changes to Fd3m, and
Yabuuchi’s sample resembles our SS . SSn results in the mixture of C2/m and Fd3m (Fig. 9). A small
 R. Kataoka et al. / Electrochimica Acta 182 (2015) 871–877 877

amount of Fe3 O4 as a residual impurity in the sample and/or the [4] C. Fouassier, G. Matejka, J.-M. Reau, P. Hagenmuller, J. Solid State Chem. 6 (1973)
crystallinity of the sample seems to affect their structural changes. 532–537.
[5] C. Delmas, C. Fouassier, P. Hagenmuller, Physica B+C 99 (1980) 81–85.
Namely, HT has a high crystallinity and contains few impurities, [6] C. Delmas, J.-J. Braconnier, C. Fouassier, P. Hagenmuller, Solid State Ionics 3/4
while SS has a low crystallinity and contains a few percentage (1981) 165–169.
of the Fe3 O4 residue (Tables 1 and 3). In addition, the HT sam- [7] Z. Lua, J.R. Dahn, J. Electrochem. Soc. 148 (2001) A1225–A1229.
[8] S. Komaba, C. Takei, T. Nakayama, A. Ogata, N. Yabuuchi, Electrochem. Comm.
ple is able to undergo more Na extraction, while maintaining its 12 (2010) 355–358.
R3m structure(Supplementary Information). The phase transition [9] N. Yabuuchi, M. Kajiyama, J. Iwatate, H. Nishikawa, S. Hitomi, R. Okuyama, R.
from R3m to either C2/m or Fd3m causes a decrease in DNa+ of Usui, Y. Yamada, S. Komaba, Nature Mater. 11 (2012) 512–517.
[10] S. Komaba, N. Yabuuchi, T. Nakayama, A. Ogata, T. Ishikawa, I. Nakai, Inorg.
the ˛-NaFeO2 samples, resulting in the significant degradation of Chem. 51 (2012) 6211–6220.
the ˛-NaFeO2 electrodes. Therefore, we assume that it is impor- [11] M. Sathiya, K. Hemalatha, K. Ramesha, J.-M. Tarascon, A.S. Prakash, Chem. Mater.
tant to prepare the sample without impurities and with a high 24 (2012) 1846–1853.
[12] B. Mortemard de Boisse, D. Carlier, M. Guignard, C. Delmas, J. Electrochem. Soc.
crystallinity in order to improve the electrode performance of ˛-
160 (2013) A569–A574.
NaFeO2 . Investigations using the transmission electron microscope, [13] J.S. Throne, R.A. Dunlap, M.N. Obrovac, J. Electrochem. Soc. 160 (2013)
57 Fe Mösbbauer spectroscopy, etc., are under way to further clarify A361–A367.
the mechanism of the structural evolution and its relation to the [14] D. Buchholz, A. Moretti, R. Kloepsch, S. Nowak, V. Siozios, M. Winter, S. Passerini,
Chem. Mater. 25 (2013) 142–148.
electrochemical properties. [15] X. Wu, J. Guo, D. Wang, G. Zhong, M.J. Mcdonald, Y. Yang, J. Power Sources. 281
(2015) 18.
Appendix A. Supplementary Data [16] G. Singh, F. Aguesse, L. Otaegui, E. Goikolea, E. Gonzalo, J. Segalini, T. Rojo, J.
Power Sources 273 (2015) 333–339.
[17] H. Guo, Y. Wang, W. Han, Z. Yu, X. Qi, K. Sun, Y.-S. Hu, Y. Liu, D. Chen, L. Chen,
Supplementary data associated with this article can be found, Electrochim. Acta 158 (2015) 258–263.
in the online version, at http://dx.doi.org/10.1016/j.electacta.2015. [18] N. Yabuuchi, H. Yoshida, S. Komaba, Electrochem. 80 (2012) 716–719.
09.092. [19] Y. Takeda, K. Nakahara, M. Nishijima, N. Imanishi, O. Yamamoto, M. Takano, R.
Kanno, Mater. Res. Bull. 29 (1994) 659–666.
[20] Y. Takeda, J. Akagi, A. Edagawa, M. Inagaki, S. Naka, Mat. Res. Bull. 15 (1980)
References 1167–1172.
[21] M. Tabuchi, K. Ado, H. Sakaebe, C. Masquelier, H. Kageyama, O. Nakamura, Solid
[1] S.-P. Ong, V.L. Chevrier, G. Hautier, A. Jain, C. Moore, S. Kim, X. Ma, G. Ceder, State Ionics 79 (1995) 220–226.
Energy Environ. Sci. 4 (2011) 3680–3688. [22] J. Zhao, L.-W. Zhao, N. Dimov, S. Okada, T. Nishide, J. Electrochem. Soc. 160
[2] V. Palomares, P. Serras, I. Villaluenga, K.B. Hueso, J. Carretero-González, T. Rojo, (2013) A3077–A3081.
Energy Environ. Sci. 5 (2012) 5884–5901. [23] W. Weppner, R.A. Huggins, J. Electrochem. Soc. 124 (1977) 1569–1578.
[3] A. Abouimrane, D. Dambournet, K.W. Chapman, P.J. Chupas, W. Weng, K. Amine, [24] I.E. Grey, R.J. Hill, A.W. Hewat, Z. Kristallographie 193 (1990) 51–69.
J. Am. Chem. Soc. 134 (2012) 4505–4508. [25] M.E. Fleet, Acta Crystallographica B 37 (1981) 917–920.