Na3+xMxP1?xS4 (M = Ge4+, Ti4+, Sn4+) enables high rate all-solid-state Na-ion batteries Na2+2δFe2?δ(SO4)3|Na3+xMxP1?xS4|Na2Ti3O7

Article abstract of DOI:10.1039/c6ta09809f

Electrolytes in current Na-ion batteries are mostly based on the same fundamental chemistry as those in Li-ion batteries-a mixture of flammable liquid cyclic and linear organic carbonates leading to the same safety concerns especially during fast charging. All-solid-state Na-ion rechargeable batteries utilizing non-flammable ceramic Na superionic conductor electrolytes are a promising alternative. Among the known sodium conducting electrolytes the cubic Na₃PS₄ phase has relatively high sodium ion conductivity exceeding 10^?4 S cm^?1 at room temperature. Here we systematically study the doping of Na₃PS₄ with Ge⁴⁺, Ti⁴⁺, Sn⁴⁺ and optimise the processing of these phases. A maximum ionic conductivity of 2.5 × 10^?4 S cm^?1 is achieved for Na_3.1Sn_0.1P_0.9S₄. Utilising this fast Na⁺ ion conductor, a new class of all-solid-state Na_2+2δFe_2?δ(SO₄)₃|Na_3+xM_xP_1?xS₄ (M = Ge⁴⁺, Ti⁴⁺, Sn⁴⁺) (x = 0, 0.1)|Na₂Ti₃O₇ sodium-ion secondary batteries is demonstrated that is based on earth-abundant safe materials and features high rate capability even at room temperature. All-solid-state Na_2+2δFe_2?δ(SO₄)₃|Na_3+xM_xP_1?xS₄|Na₂Ti₃O₇ cells with the newly prepared electrolyte exhibited charge-discharge cycles at room temperature between 1.5 V and 4.0 V. At low rates the initial capacity matches the theoretical capacity of ca. 113 mA h g^?1. At 2C rate the first discharge capacity at room temperature is still 83 mA h per gram of Na_2+2δFe_2?δ(SO₄)₃ and at 80 °C it rises to 109 mA h per gram with 80% capacity retention over 100 cycles.

Full text of DOI:10.1039/c6ta09809f

View Article Online
View Journal
Journal of
Materials Chemistry A
Accepted Manuscript
This article can be cited before page numbers have been issued, to do this please use: R. Prasada Rao, H.
Chen, L. L. Wong and S. Adams, J. Mater. Chem. A, 2017, DOI: 10.1039/C6TA09809F.
This is an Accepted Manuscript, which has been through the
Volume
4 Number 1 7 January 2016 Pages 1–330
Royal Society of Chemistry peer review process and has been
accepted for publication.
Journal of
Materials Chemistry A
Accepted Manuscripts are published online shortly after
Materials for energy and sustainability
www.rsc.org/MaterialsA
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
author guidelines.
Please note that technical editing may introduce minor changes
to the text and/or graphics, which may alter content. The journal’s
standard Terms & Conditions and the ethical guidelines, outlined
in our author and reviewer resource centre, still apply. In no
event shall the Royal Society of Chemistry be held responsible
for any errors or omissions in this Accepted Manuscript or any
consequences arising from the use of any information it contains.
ISSN 2050-7488
PAPER
Kun Chang, Zhaorong Chang et al.
Bubble-template-assisted synthesis of hollow fullerene-like
MoS2 nanocages as a lithium ion battery anode material
rsc.li/materials-a

Page 1 of 12
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C6TA09809F
Na3+xMxP1ꢀxS4 (M = Ge4+, Ti4+, Sn4+) Enables High Rate AllꢀSolidꢀState
Naꢀion Batteries Na2+2δFe2ꢀδ(SO4)3|Na3+xMxP1ꢀxS4|Na2Ti3O7.
Rayavarapu Prasada Raoa, Haomin Chen, Lee Loong Wong, Stefan Adams,*,a
Received (in XXX, XXX) Xth XXXXXXXXX 20XX, Accepted Xth XXXXXXXXX 20XX
5
DOI: 10.1039/b000000x
Electrolytes in current Naꢀion batteries are mostly based on the same fundamental chemistry as those in
Liꢀion batteries – a mixture of flammable liquid cyclic and linear organic carbonates leading to the same
safety concerns especially during fast charging. Allꢀsolidꢀstate Naꢀion rechargeable batteries utilizing
nonꢀflammable ceramic Na superionic conductor electrolytes are a promising alternative. Among the
10 known sodium conducting electrolytes cubic Na3PS4 phase has relatively high sodium ion conductivity
exceeding 10−4 S cm−1 at room temperature. Here we systematically study the doping of Na3PS4 with
Ge4+, Ti4+, Sn4+ and optimise the processing of these phases. A maximum ionic conductivity of 2.5×10ꢀ4 S
cmꢀ1 is achieved for Na3.1Sn0.1P0.9S4. Utilising this fast Na+ ion conductor, a new class of allꢀsolidꢀstate
Na2+2δFe2ꢀδ(SO4)3 | Na3+xMxP1ꢀxS4 (M = Ge4+, Ti4+, Sn4+) (x = 0, 0.1) | Na2Ti3O7 sodiumꢀion secondary
15 batteries is demonstrated that is based on earthꢀabundant safe materials and features a high rate capability
even at room temperature. Allꢀsolidꢀstate Na2+2δFe2ꢀδ(SO4)3 | Na3+xMxP1ꢀxS4 | Na2Ti3O7 cells with the
newly prepared electrolyte exhibited charge–discharge cycles at room temperature between 1.5 V to 4.0
V. At low rates the initial capacity matches the theoretical capacity of ca. 113 mAh/g. At 2C rate the first
discharge capacity at room temperature is still 83 mAh per gram of Na2+2δFe2ꢀδ(SO4)3 and at 80 ⁰C it rises
20 to 109 mAh per gram with 80% capacity retention over 100 cycles.
temperatures have been published (For an overview see Table 1).
Allꢀsolidꢀstate Naꢀion rechargeable batteries utilizing nonꢀ
flammable ceramic Na superionic conductor electrolytes
55 constitute a promising alternative to conventional liquid Naꢀion
electrolytes that are mostly inspired by the fundamental chemistry
of seemingly analogous Liꢀion batteries, though the
predominating weakly coordinating alkali salts dissolved in
mixtures of cyclic and linear organic carbonates are typically
60 flammable and have limited electrochemical windows. The use of
a solid, rather than liquid electrolyte simplifies the fabrication and
integration of batteries into systems and is beneficial for the shelf
and cycle life of the batteries.10ꢀ12
The main function of the solid electrolyte is obviously to
65 transport ions, while the electronic conductivity should be low to
prevent a shortꢀcircuiting of the cell. Moreover, the electrolyte
must be electrochemically stable under operating conditions. In
most devices the electrolyte does not need to support a sizeable
external mechanical load, but it needs to withstand stresses that
70 arise due to thermal cycling or expansion/contraction of
electrodes during charge/discharge.
1. Introduction
Sodiumꢀion batteries are promising low cost electrochemical
energy storage systems to overcome voltage fluctuations and
thereby to enhance the commercial value of renewable energy
25 sources such as solar and wind energy, but appear also of interest
for
a wide variety of applications starting from portable
applications to (hybrid) electric vehicles. The regionally restricted
occurrence of economically recoverable lithium ore resources for
LIBs has increased the level of interest in alternatives with
30 comparable or higher energy densities such as sodiumꢀbased
batteries. These include high temperature molten salt designs
such as sodium–sulphur (or Zebraꢀtype NaꢀNiCl2) cells, wherein
liquid sodium is separated from liquid sulphur (or liquefied
NiCl2) by a ceramic sodium ionꢀconductor tube.1ꢀ6 A main
35 disadvantage of these molten salt designs is that they do not work
at room temperature and the cells may be irreversibly damaged
on quick temperature changes. For room temperature sodium ion
battery applications so far mainly polymer electrolytes, such as
poly(ethylene oxide)7 and poly(vinylidene difluoride),8, have
9
Oxideꢀbased Na solid electrolytes such as βꢀalumina and
NAtrium Superionic CONductors (NASICON) are wellꢀstudied,
but despite their high bulk Na+ conductivities the corresponding
75 polycrystalline oxide ceramics yield sufficiently high total ionic
conductivities and low charge transfer impedances only at
elevated temperatures. Here we report the synthesis,
characterization and application of a solid electrolyte that from
parallel computational studies appears to enable high rate
80 capability of allꢀsolidꢀstate rechargeable Naꢀion batteries both at
room temperature and elevated temperatures. Recently, the cubic
phase of Na3PS4 (cꢀNa3PS4, space group: I4ꢀ3m) was reported to
show a room temperature Na+ ion conductivity of 0.2 mS/cm,13, 14
which is considerably higher than previously found for the stable
40 been explored, but their limited temperature stability rules out the
design of batteries that can operate over wide temperature ranges.
Thus for applications under extreme temperature variations (e.g.
in marine drilling, for sterilisable medical devices, etc.) allꢀsolidꢀ
state batteries with inorganic solid sodium ion conductors will be
45 required that exhibit sufficient room temperature ionic
conductivity and are thermally stable over the entire operating
temperature range. Only a few such systems have been reported
in the literature so far and among these most reports are
concerning “half cells” using sodium or sodium alloy anodes,
50 while to the best of our knowledge for full cells with insertion
type anodes and cathodes only performance data at elevated

Journal of Materials Chemistry A
Page 2 of 12
View Article Online
DOI: 10.1039/C6TA09809F
Table 1. Comparison of literature performance data on allꢀsolidꢀstate Naꢀion and selected Naꢀbased half cells combining Na metal or Na alloys with
insertionꢀtype cathodes.
Anode
Cathode
Solid electrolyte
Capacity /
C Rate
Reference
mAhgꢀ1 (Temp.)
Na or Na alloy-based all-solid-state half cells
15
16
17
18
19
Na
NaꢀSn
TiS2
TiS2
Na2.9375PS3.9375Cl0.0625
Na3PS4
Na3SbS4
Na3PS4
Na3PS4
80
90
108
60
(RT)
(RT)
(RT)
(RT)
(RT)
0.1 C
0.05 C
0.1 C
0.05 C
0.05 C
NaꢀSn
NaꢀSn
NaꢀSn
NaCrO2
NaCrO2
aꢀTiS3
300
All-solid-state Na-ion full cells
0
(RT)
ꢀꢀ
0.1 C
ꢀꢀ
0.05 C
2 C
20
21
Na3+xV2(PO4)3
Na3ꢀxV2(PO4)3
Na1+xZr2SixP3ꢀxO12
82 (200 °C)
(RT)
0
Na2Ti3O7 ꢀLa0.8Sr0.2MnO3
Na2/3[Fe1/2Mn1/2]O 2 Naꢀβ′′ꢀAlumina
152 (350 °C)
80 (350 °C)
114
83
(RT)
(RT)
0.1 C
2 C
Na2Ti3O7
Na2+2xFe2ꢀx(SO4)3
Na3.1Sn0.1P0.9S4
(This work)
2 C
109 (80 °C)
roomꢀtemperature tetragonal phase (space group Pꢀ421c).22ꢀ24 In 45 a minimal volume change that needs to be accommodated during
5
line with ab initio studies of Na3PS4,25 the closely related
cycling plays an important role. Alloy anodes thus have to be
excluded and graphite anodes only work with specific liquid
electrolytes that can be coꢀintercalated.41 Sodium titanate,
26
27
Na3PSe4 and recent findings for Na3SbS4 we observe that the
conductivity of this phase crucially depends on the availability of
defects in the Na pathways. In order to optimise the ionic
Na2+xTi3O7, can be reversibly intercalated by sodium with a
conductivity of the material we thus conducted further ab initio 50 plateau centred at 0.3 V vs. Na/Na+42,
and limited volume
change of 6% for when varying x from 0 to 244 and particularly
favourable rate performance (yet slightly higher average voltage)
when x is restricted to remain <1.45 In combination with suitable
cathodes such as alluaudite, the low voltage of Na2Ti3O7 yields a
43
10 and empirical computational studies and experimentally prepared
Na3+xMxP1ꢀxS4 (M = Ge4+, Ti4+, Sn4+; x = 0, 0.1) using mechanical
milling followed by annealing.
To realise a competitive allꢀsolidꢀstate sodiumꢀion battery
(SIB) a range of SIB cathode materials have been explored. P2ꢀ 55 high cell voltage and thus enhances energy density, compared to
15 and O3ꢀtype layered oxide cathode materials, which are mostly
based on costly transition metals such as Co and Ni, have been
studied extensively. Largeꢀscale Naꢀbatteries will, however, be
batteries relying on the previously studied anode materials with
more positive potentials (1.6 V for Na3V2(PO4)3, 1.75 V for
NaXVO2,46, 47 0.94 V for Ni3S2,48 or 2.1 V for NaTi2(PO4)349).
Based on the above considerations here we design and realise
commercially more viable with earthꢀabundant transition metals
28
such as Fe, but both O3ꢀtype NaFeO2
and P2ꢀtype 60 an allꢀsolidꢀstate full Na+ ion cell using low cost earthꢀabundant
29
20 Nax[Fe1/2Mn1/2]O2 suffer from limited reversible capacity and
low operating potential despite using Fe4+/Fe3+ redox couple.
Using the inductive effect in polyanion framework systems, the
potential of the Fe3+/Fe2+ redox couple can be pushed up with full
materials
Na2+2δFe2ꢀδ(SO4)3|Na3.1Sn0.1P0.9S4|Na2Ti3O7.
We
demonstrate that it can operate over a wide temperature range and
is the first allꢀsolidꢀstate SIB full cell that achieves a favourable
rate performance comparable to those reported for liquid
utilization of the oneꢀelectron reaction.30 In this pursuit, a wide 65 electrolyte cells even at room temperature.
3ꢀ
25 range of ~3V Feꢀbased phosphate PO4 insertion compounds
have been reported (e.g. Na2FePO4F 3.06 V,31 NaFePO4 2.7 V,32
Na2FeP2O7 3 V,33 and Na4Fe3(PO4)2P2O7 3.2 V34). Replacing
3ꢀ
2ꢀ
phosphate PO4 with sulphate SO4 units led to alluauditeꢀtype
cathodes, Na2+2δFe2ꢀδ(SO4)3, which combine the unusually high
30 Fe redox potential ~3.8 V versus Na with good economy.35 As
seen from Figure 1 that compares a wide range of recently
studied sodium insertion cathode materials with respect to the C
rate dependence of their capacity, many cathode materials lose
already a significant fraction of their lowꢀrate capacity when
35 cycled at 2 C, while for others including the alluaudite only a
minor decrease of the capacity is observed up to this C rate.
Detailed computational analyses of the structural origin of
different rate capabilities of sodiumꢀion cathode materials in
general and the role of nonꢀstoichiometry in alluaudite are
40 discussed elsewhere.36,37 Experimental studies on the interrelation
of nonꢀstoichiometry, transport processes and electrochemical
Figure 1. Rate capability of a wide range of sodium cathode materials
(from literature data based on tests in liquid electrolyte half cells)
displayed as capacity Q vs. C rate. Lines are fits to the displayed Q(C
rate) data as the sum of two stretched exponential functions. For
comparison, the rate capability of the selected anode material and the rate
capability found for the cathode material in this work in the room
temperature allꢀsolidꢀstate battery (ASSB) are also shown.35, 50ꢀ65
70
performance have been reported by the Yamada group38, 39
.
In the search for matching anode materials40 for bulk allꢀsolidꢀ
state batteries, besides the theoretical capacity and redox potential

Page 3 of 12
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C6TA09809F
2. Experimental and Computational Methods
Preparation of electrolytes
Samples of Na3+xMxP1ꢀxS4 (M = Ge4+, Ti4+, Sn4+) (x = 0, 0.1)
were prepared from preheated Na2S⋅9H2O (see supplement for
details of Na2S preparation and supplement figure S1 for XRD of
Na2S), GeS2, SnS2, TiS2, P2S5. Stoichiometric amounts of the
respective required chemicals were mixed by planetary high
energy ballꢀmilling using zirconia pot and balls (Fritsch P7, 45 ml
bowl and 15 number of balls ø 10 mm) for 5 to 10 h. The samples
5
10 were annealed at 250 °C for 6 h under argon atmosphere. The ball
milled Na3PS4 was subsequently heated at 350 °C under vacuum
for 2h under Ar atmosphere and at 500 °C for different samples.
Preparation of electrodes
15 Na2+2δFe2ꢀδ(SO4)3 was synthesized by mechanical milling. 1.54 g
of Na2SO4 and 2.73 g of FeSO4 were mixed in a planetary high
energy ballꢀmill using zirconium pot and balls (Fritsch P7, 45 ml
bowl and 15 number of balls ø 10 mm) for 5 h followed by
annealing at 400 ºC under argon for 18 h. Pure Na2Ti3O7 was
20 prepared from anatase TiO2 (>99.8%, Aldrich) and anhydrous
Na2CO3 (>99.995%, Aldrich) mixtures with 10% excess of the
latter with respect to stoichiometric amounts. These mixtures
were milled and treated at 800 °C for 20 h with intermediate
regrinding.
60 Figure 2. Cross section of Scanning electron microscopy image for allꢀ
solidꢀstate Na2+2δFe2ꢀδ(SO4)3 | Na3.1Sn0.1P0.9S4 | Na2Ti3O7 battery.
analysis we employed a software package developed in our
group,36 while for the molecular dynamics simulations of 4×4×4
65 supercells of the cubic phase, Na392P120Si8S512 (i.e. Na3+xP1ꢀxSixS3
with x = 0.0625 to approximate the composition reported by
Hayashi et al.69) and the tetragonal phase Na384P128S512 the
software package GULP was employed as implemented in
Materials Studio 6.0. For the MD simulations short NPT
70 simulations to equilibrate the unit cell volume at the respective
temperature were followed by NVT simulation production runs
over 1 – 24 ns using a 1 fs time step. The observed Na diffusivity
is converted to ionic conductivity assuming validity of the Nernst
Einstein relationship.
25
The annealed samples were characterized by in situ Xꢀray
powder diffractometry using Cu Kα radiation (PANalytical X’Pert
PRO) in the 2θ° range 10ꢀ100° with a nominal scan rate of 120s
stepꢀ1 and a step size of 0.016° from 30°C to 350 °C. Rietveld
refinements of XRD powder patterns were performed with the
30 Generalized Structure Analysis System (GSAS), along with the
graphical user interface EXPGUI. Ionic conductivities of samples
were determined by impedance spectroscopy (Schlumberger
Solartron SI1260, frequency range 1 Hz ꢀ 1 MHz) using pressed
pellets in between stainless steel sheets in the temperature range
35 30°C ꢀ 150°C for all the samples. The samples were kept at each
temperature for 20 min for thermal equilibration before recording
impedance spectra. The bulk resistance Rb was then determined
from fitting impedance spectra. Nyquist plots (Z′ vs.−Z″) were
derived and analyzed using Z plot and Z view software (Version
40 2.2, Scribner Associates Inc., USA). Scanning electron
microscopy (SEM) of the all samples and battery were collected
using ZEISS NEON High Resolution Characteristics by applying
10 kV potential.
Electrodes for the allꢀsolidꢀstate batteries were prepared by
45 ball milling the composite consisting of electroactive material :
solid electrolyte : activated carbon in the ratio 60 : 30 : 10.
Cathode and anode composite powders were pressed on either
side of the electrolyte. The thickness of the electrolyte in a
representative test cell was about 540 ꢁm as shown in figure 2.
50 The cells are assembled into a Swagelok cell and cyclic
75
The formation energy of various Na3+xMxP1ꢀxS4 phases from
their constituting elements is determined by DFT simulation
methods as implemented in the Vienna ab initio simulation
package (VASP) software70 using a planewave basis set with the
Perdew–Burke–Ernzerhof
(PBE)
exchangeꢀcorrelation
80 parametrized general gradient approximation (GGA)71 and
projector augmentedꢀwave pseudopotentials (PAW)72, 73 The cutꢀ
off energy for planewave basis set is set to 400 eV and
convergence is by force < 0.02 eV/Å. Each compound has a kꢀ
mesh with a density of at least 30 k points per Å3.
85
To study the effect of dopant M (M = Ge, Si, Sn, Ti) and the
doping concentration on the stability of the compound, a 2×2×2
supercell is constructed for cubic Na3PS4 and both atomic
coordinates as well as lattice parameters are allowed to relax
during geometry optimization. One, two or three P atoms are
90 substituted by M (x =0.0625, 0.125, 0.1875), maximizing the
distance of dopant atoms. Correspondingly, one, two or three Na
atoms are added near the dopant site in order to maintain charge
neutrality. The stability is measured by dissociation energies,
with reference structures retrieved from or derived from the ICSD
95 database,74 using the method outlined in our recent paper on
argyroditeꢀtype thiophosphates.75
performance of the batteries was investigated using
a
potentiostat/galvanostat (Arbin BT2000) at room temperature in
the voltage range 1.5V and 4.0V vs. Na/Na+.
3. Results and discussion
55 Computational methods
Computational studies of ion transport pathways and molecular
dynamics simulations were performed using our bond valenceꢀ
based softBV forcefield parameters.36, 66ꢀ68 For the static pathway
Preparation of solid electrolyte
The variation of formation energies of dopedꢀNa3PS4 are plotted
100 in Figure 3, and the dissociation products are listed in Table 2

Journal of Materials Chemistry A
Page 4 of 12
View Article Online
DOI: 10.1039/C6TA09809F
showing for comparison also data for homovalently (M = As) and 45 Na3PS4 (space group Iꢀ43m) with lattice parameter a = 6.9781(1)
trivalently (M = Al) doped samples. The stability of tetravalently
doped Na3PS4 with varying dopant ion M decreases in the
sequence Ge > Si > Ti ~ Sn. The formation energy, and
particularly the influence of different dopants on it, harmonises
with literature DFT results,25 where such values are available.
The slight difference in absolute values may be due to different
computational settings and different sets of reference compounds.
It may be noted that the stable dissociation products Na4MS4 for
Å at room temperature. Figure 6 compares the temperature
dependence of the cubic lattice parameter a obtained from
Rietveld refinements of in situ xꢀray diffraction data for the
undoped, Geꢀ and Snꢀdoped samples during stepwise heating
5
50
10 the different dopants M do not refer to the same structure type:
Na4GeS4 and Na4SnS4 refer to the tetragonal Na4SnS4 structure
type (space group Pꢀ421c)76, while Na4SiS4 and Na4TiS4 refer to
the orthorhombic Na4GeSe4 structure type (space group Pnma)77.
Given our experimental results of successfully doping 10% Ge,
15 Sn and Ti, it may be concluded that the minor difference in
dissociation energies (at 0 K) due to the choice of M is
insignificant at the synthesis temperature. The dissociation energy
of the 6.25% Al doped sample is comparable to the dissociation
energy of the 12.5% tetravalently doped samples.
Figure 3. Dissociation energies of Na3+xMxP1ꢀxS4 compounds versus
dopant concentration x in percent. (Less negative values indicate higher
stability of the doped phases).
20
Table 2 also lists results for analogous doping with a halide X
(X = F, Cl, Br, I) Na2.9375PS3.9375X0.0625, which according to
suggestions from de Klerk et al.78 should promote the
conductivity by generating Na vacancies. The dissociation
energies corresponding to a doping concentration of 1.56% X are
25 comparable to those of doping 6.25% M. It may be noted that the
dissociation products include elemental sulphur, which may
undesirably react with the electrode materials.
As seen in Fig. 4 any tetravalent doping increases the volume
because additional Na is inserted into the structure. For the same
30 dopant concentration, the volume increases in the sequence
Si < Ge < Ti < Sn in line with the atomic radii of the dopants. The
extrapolation of linear trend lines of the volumeꢀvs.ꢀdopant
concentration correlation to the undoped case leads to volumes
close to those of the cubic phase although the energyꢀminimised
35 supercells of the doped phases are inevitably slightly distorted
variants of the cubic phase.
55
Figure 4. Volume of Na3+xMxP1ꢀxS4 phases versus dopant concentration x
in percent based on DFT geometry optimisations.
Characterisation of solid electrolyte
Figure 5 shows the Rietveld refinement of Na3PS4 prepared by
40 ball milling and subsequent annealing under Argon atmosphere at
250 °C. The pattern indicates the formation of the cubic phase of
Table 2. Comparison for DFT simulated dissociation energies of various
compounds related to Na3PS4 in this work and from the literature.25
Figure 5. Rietveld refinement of Na3PS4 indicating the formation of the
cubic phase with space group Iꢀ43m with wRp = 6.2%.
Compound Dissociation product
ꢁE
Lit. ꢁE
60
(eV/atom) (eV/atom)
Na49P15GeS64 15 tꢀNa3PS4 + Na4GeS4
Na49P15SiS64 15 tꢀNa3PS4 + Na4SiS4
Na49P15SnS64 15 tꢀNa3PS4 + Na4SnS4
Na49P15TiS64 15 tꢀNa3PS4 + Na4TiS4
ꢀ0.010
ꢀ0.010
ꢀ0.012
ꢀ0.011
ꢀ0.005
ꢀ0.023
ꢀ0.006
ꢀ0.009
ꢀ0.009
ꢀ0.008
0.045
ꢀ0.013
ꢀ0.013
ꢀ0.015
Na48P15AsS64 15 tꢀNa3PS4 + tꢀNa3AsS4
Na50P15AlS64 15 tꢀNa3PS4 + Na3AlS3 + Na2S
Na47P16S63F 14 tꢀNa3PS4 + S + NaF + 2 Na2PS3
Na47P16S63Cl 14 tꢀNa3PS4 + S + NaCl + 2 Na2PS3
Na47P16S63Br 14 tꢀNa3PS4 + S + NaBr + 2 Na2PS3
Na47P16S63I 14 tꢀNa3PS4 + S + NaI + 2 Na2PS3
ꢀ0.006
ꢀ0.008
ꢀ0.009
ꢀ0.008
tꢀNa3PS4
cꢀNa3PS4
tꢀNa3AsS4
½ S + ½ Na2S + Na2PS3
tꢀNa3PS4
S + Na3AsS3
ꢀ0.006
0.010
ꢀ0.005
cꢀNa3AsS4 tꢀNa3AsS4
ꢀ0.004
0.013
ꢀ0.001
tꢀNa3PSe4
cꢀNa3PSe4
NaSe + Na2PSe3
tꢀNa3PSe4
Figure 6. Variation of cubic or pseudoꢀcubic lattice constant a with
temperature for Na3+xMxP1ꢀxS4 (M= Ge4+, Sn4+) (x=0, 0.1) samples.

Page 5 of 12
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C6TA09809F
under vacuum with a heating rate of 3 K/min. The average
3.498 Å to 3.527 Å by the coordinated displacement of 4 of the 6
Na per unit cell from the fully occupied octahedral site towards
Our bond valence site energy models of Na+ ion migration
paths in the average structure of the disordered cubic Na3PS4 (see
coefficient of thermal expansion of cꢀNa3PS4 over the range from
room temperature to 250 °C is found to be 4.5×10ꢀ5 Kꢀ1. Both
samples annealed at 350 ˚C and 500 ˚C under argon atmosphere
5
lead to the tetragonal phase. For the case of the sample annealed 50 Fig. 8a,b) phase show that “octahedral” Na(1) site is the only
at 500 °C we find lattice constants a=6.9676(1) Å, c= 7.0832(2)
Å according to the Rietveld refinement shown in figure S2.
When partially replacing phosphorus with Ge in
Na3.1Ge0.1P0.9S4, just ball milling will not lead to a cubic phase
equilibrium site and the reported “Na(2)” site is just the
saddlepoint along a Na(1)ꢀNa(2)ꢀNa(1) path. Thus the description
of Na distribution in the cubic phase by Na(1) and Na(2) should
be rather seen as an approximation to a large energy minimum
10 but to the tetragonal phase. The Rietveld refinement of a powder
pattern for a Na3.1Ge0.1P0.9S4 sample annealed at 250 °C is shown
in figure 7 (data collected at 200 °C), which demonstrates that the
sample is stable at least up to 200 °C. The lattice parameters at
room temperature for this compound are a=6.9950(5) Å,
15 c=7.031(1) Å. The ratio between lattice parameters a and c is
0.99 indicating the compound is pseudoꢀcubic. The compounds
Na3.1Sn0.1P0.9S4, Na3.1Ti0.1P0.9S4 with Sn and Ti dopants formed
cubic phases with room temperature lattice parameters of
7.0088(3) Å and 6.9894(3) Å, respectively. Rietveld refinements
20 of Na3.1M0.1P0.9S4 for M= Ge, Si, Sn shown in figures S3, S4 and
S5. From Rietveld refinements over 6 temperatures in the range
between room temperature and 250°C the coefficients of thermal
expansion were determined as 4.08×10ꢀ5 Kꢀ1 for Na3.1Ge0.1P0.9S4
and 4.11×10ꢀ5 Kꢀ1 for Na3.1Sn0.1P0.9S4. The variation of lattice
25 constants and thermal expansion coefficients appears plausible
based on the different size of the dopant ions.
55
(a)
(b)
3ꢀ
The bodyꢀcentred cubic arrangement of PS4 groups in the
cubic phase of Na3PS4 (space group Iꢀ43m) creates two distinct
Na site types: the 6b site Na(1) with the partial occupancy of 0.8
3ꢀ
30 that has a distorted octahedral coordination by 6 PS4 groups (or
4+4 coordination if the individual S2ꢀ are considered as the
counterions), and the 12d site Na(2) with partial occupancy of 0.1
3ꢀ
and a tetrahedral coordination by 4 PS4 groups (or 4 S2ꢀ at a
(c)
(d)
distance of only 2.07 Å). For the average structure the Na(2) site
35 is somewhat overbonded with a bond valence sum of 1.27
explaining why the “octahedral” Na(1) site is preferred. The
tetragonal room temperature phase is a just slightly distorted
variant of this cubic structure (not a heavily distorted fcc structure
as was suggested in the original structure determination26), where
40 the distortion allows to increase the average NaꢀNa distance from
the tetrahedral site.
60
(e)
(e)
Figure 8. Crystal structures of (a,b) the cubic phase and (c,d) the
tetragonal phase of Na3PS4 projected on the aꢀc plane. (green spheres: P,
yellow spheres: sulphur, red (magenta) spheres: Na(2) (Na(1)) in the
65 tetragonal phase; Na(1) in the cubic phase, blue crosses: Na(2) “sites” in
the cubic phase that turn out to be saddlepoints). The superimposed
isosurfaces in graphs (b, d) indicate bondꢀvalence models of Na+ ion
migration pathways of lowest activation energy. Red isosurfaces mark the
plateꢀshaped space that is explored by Na while vibrating on the
70 equilibrium site, the orange isosurface indicates the percolating pathways
of lowest energy, which is 2D in the aꢀb plane for the tetragonal phase,
3D for the cubic phase. For the tetragonal phase the yellow isosurface in
graph (d) shows the additional links perpendicular to the cꢀdirection that
allow for a 3D migration for Na+ with a migration barrier of ca. 0.45 eV.
75
(e) Migration barriers in tetragonal (black) and cubic (red) Na3PS4 as
determined from the bond valence site energy model.
Figure 7. Rietveld refinement of Na3.1Ge0.1P0.9S4 annealed at 250°C,
45
data collected at 200°C under vacuum, wRp=5%.

Journal of Materials Chemistry A
Page 6 of 12
View Article Online
DOI: 10.1039/C6TA09809F
that extends from the central Na(1) towards four surrounding
Na(2) and is large enough to temporarily host more than one Na+
ion. If there is only one Na, then it will be predominantly located
at the Na(1) site, while the presence of a second Na requires both
of them to be displaced towards two of the Na(2) sites at opposite
ends of the minimum, where they will have a higher site energy
and therefore be considerably more mobile. This implies that for
this structure subvalent doping on the P site (and consequentially
an uptake of additional Na+) will be a more promising doping
5
10 strategy than the creation of vacancies by supervalent doping, as
it leads to lower effective migration barriers between the
overcrowded extended Na sites.
In the tetragonal phase (Fig. 8c,d) that differs from the cubic
phase mostly by the Na+ ordering into wellꢀdefined sites that can
15 host only one ion, Na+ transport occurs via direct hops among the
two nominally fully occupied equilibrium sites (an octahedral
Na(2) site and a Na(1) site that is shifted from the octahedral site
by ca. 0.6 Å towards one of the tetrahedral sites). Both sites
correspond to the octahedral Na(1) site of the cubic phase. Thus,
20 the achievable ionic conductivity will again crucially depend on
the presence of vacancies or interstitials. Moreover, the relaxation
of the immobile substructure in response to the Na+ ordering
restricts the low energy paths for Na+ motion to the aꢀb plane
with a substantially higher energy barrier for the motion along the
25 c direction.
To gain deeper insight into the ion transport in doped and
undoped Na3PS4, we conducted molecular dynamics simulations
of the cubic phase Na392P120Si8S512 (with the same 6% Si doping
proposed by Hayashi et al.) and the tetragonal phase Na384P128S512
30 over the temperature range 300 – 450 K. Fig. 9 compares the
results of these simulations to experimental literature conducꢀ
tivity data of the same phases.22, 50 The simulated conductivity of
the tetragonal phase agrees both in absolute value and activation
energy 0.42 eV with the available experimental data. In the fully
35 ordered tetragonal phase the observed activation energy not only
contains the migration energy (cf. Fig. 8e) but also a defect
formation contribution. In contrast the activation energy observed
for the Siꢀdoped cubic phase (0.27 eV) closely matches both the
experimental value and the one predicted in Fig. 7e from the
40 static bond valence models. It should also be noted that the
Arrhenius plot for the cubic phase is slightly curved so that the
60
Figure 9. Comparison of molecular dynamics simulation results (full
symbols) for Na+ ionic conductivity in cubic Na3+xP1ꢀxSixS4 (x=0.0625)
and tetragonal Na3PS4 in this work with literature experimental
conductivity data (open symbols)22, 26, 50
.
65 whether in a study reporting an enhanced conductivity of Na3PS4
by optimized heat treatment of glass ceramics (as well as in the
nominally stoichiometric samples produced in this work)
unintentional doping may have been overlooked.45
As the volume change on doping shown in Figure 10 should
70 lead to a corresponding increase in free volume for the Na+ ion
mobility, bond valence models yield a slight decrease of the
migration barriers in the order undoped phase > Siꢀ > Geꢀ > Tiꢀ >
Snꢀdoped phase. These bond valence migration barriers do not
yet factor in the effect of Na concentration on the effective
75 barriers of undoped and tetravalently doped phases. To achieve a
deeper insight into the influence of the Na site overcrowding on
the ion transport we calculated the Na site energies in pure and
doped Na3PS4 applying our softBV forcefield to the DFTꢀ
optimized structures. Here, the site energy is defined as the
80 interaction energy of an individual Na with the remaining
structure including the other Na. We first verified that site
energies in the tꢀNa3PS4 are lower than those of the majority site
in cꢀNa3PS4, consistent with the fact that tꢀNa3PS4 is more stable
and has a fully ordered Na arrangement. As seen in Figure 8 the
85 average site energies of Na in all dopedꢀNa3PS4 phases are higher
than in both undoped cases, because more Na ions cause larger
repulsion that is not fully compensated by the increase in unit cell
volume. Thus the empirical simulations show that the average site
energies of the Na ions decrease in the order Sn > Ti > Ge > Si,
90 which matches the trend of decreasing dopant ion size. In other
words, the larger the dopant ion is, the more the overcrowding of
Na sites raises the Na site energy, which has the beneficial side
effect of lowering the effective migration barrier for interstitialcyꢀ
like motion, while (as seen above) it will also tend to slightly
95 destabilize the doped phases. Hence, from our calculations
Snꢀdoping appeared particularly promising.
exact activation energy will exhibit
a minor temperature
dependence. The absolute value of the simulated conductivity
exceeds the experimentally observed value by a factor of two,
45 which may either be traced back to an inaccuracy of our softBV
forcefield parameters or suggest that the experimental sample by
Hayashi et al.22, 50 does not have the exactly same structure: The
latter case might e.g. involve incomplete crystallinity of the glass
ceramic, as the more recent set of experimental data (for a rather
50 limited temperature range) by Tanibata et al.23 from the same
group perfectly matches both in terms of conductivity and
activation energy.
When the tetragonal distortion is suppressed by enforcing the
lattice parameters of the cubic phase while maintaining the
55 stoichiometric composition, the simulated ionic conductivity
remains practically unchanged at the low values of the tetragonal
phase. This clarifies that the excess Na+ in the doped phase are
crucial to enhance the conductivity and casts some doubts on
A DFT calculation of the band structures of Na3PS4 as well as
of the isostructural Na3PSe4 (cf. Supplementary material Figs. S6
and S7) suggests that the band gap of Na3PS4 (>2eV in the
100 standard DFT approximation, which due to the systematic
underestimation of bandgaps by DFT should translate to roughly
4 eV) is large enough to allow for its use as an electrolyte, while
26, 79
for Na3PSe4
the band gap (ca. 1 eV in the DFT
approximation) may be too low and hence electronic conductivity
105 too high for practical applications as a solid electrolyte.

Page 7 of 12
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C6TA09809F
Figure 10. Site energies of Na ions in undoped Na3PS4 and doped
Na3+xMxP1ꢀxS4 (M=Si, Ge, Ti, Sn; x=0.0625), as calculated by the bond
valence site energy approach for DFT optimised structure models. The
35
5
site energies are listed relative to the most stable state (Na(1) in tꢀNa3PS4)
as the reference point.
Table 3. Refined lattice parameters and impedance parameters of the
studied solid electrolyte phases.
10 Compound lattice
Coefficient of
Ionic
Activation
constant thermal expansion conductivity energy
/ Å
×10ꢀ5 Kꢀ1
×10ꢀ4 S cmꢀ1
/ eV
Na3PS4 cubcic
Na3.1Ge0.1P0.9S4 6.9950(5) 3.99
15 Na3.1Ti0.1P0.9S4 6.989(3) 3.76
6.9781(1) 4.24
1.05
2.12
2.3
0.22
0.21
0.20
0.18
Na3.1Sn0.1P0.9S4 7.0088(3) 4.09
2.5
Doping
Predictions about the influence of doping on ionic conductivity
20 are verified experimentally by impedance spectroscopy. A series
of impedance plots of Na3.1Ge0.1P0.9S4 at different temperatures is
shown in figure 11 as an example. As seen from the summary of
the observed room temperature sodium ionic conductivities in
table 3, all three experimentally investigated dopants enhance the
25 ionic conductivity and the highest conductivity is found for Sn
doping. While this is qualitatively consistent with the theoretical
prediction in this work as well as by Zhu et al.,25 the extent of the
differences between the different dopants is only minor. SEM
images of all the studied solid electrolytes in figure 12 show that
30 all samples consisted of micronꢀsized agglomerated particles.
Figure 11. Nyquist plot of Na3.1Ge0.1P0.9S4 annealed at 250°C under argon
atmosphere.
Figure 12. SEM images of a) Na3PS4, b) Na3.1Ge0.1P0.9S4,
40
c) Na3.1Ti0.1P0.9S4, d) Na3.1Sn0.1P0.9S4

Journal of Materials Chemistry A
Page 8 of 12
View Article Online
DOI: 10.1039/C6TA09809F
Preparation and characterization of electrodes
The cathode material Na2+2δFe2ꢀδ(SO4)3 forms an alluauditeꢀtype
Thus all components of the envisaged allꢀsolidꢀstate battery
remain (kinetically) stable between room temperature and at least
structure in space group C2/c with lattice constants a = 12.654(1) 35 250 °C and show only moderate volume changes of 1 ꢀ 3 % over
Å, b = 12.761(9) Å, c = 6.503(5)Å, β=115.52(1)º close those
previously reported values for δ = 0.12 (see figure 13), the
secondary phase observed is about 4% βꢀFeSO4 and possibly
further amorphous impurities. The refined site occupancy of the
Fe site accords to δ ≈ 0.1 though with a large error margin.
The temperature dependence of the lattice permeates as
this temperature range (comparable to the volume changes of
5
10 determined from Rietveld refinements of in situ XRD powder
patterns is shown in figure 14. The coefficient of thermal
expansions of lattice parameters are αa = 2.8×10ꢀ5 Kꢀ1, αb = 2.0×
10ꢀ5 Kꢀ1, αc = 1.0 ×10ꢀ5 Kꢀ1, respectively, in accordance to a
previous report.37 Moreover the change in thermal expansion
15 above 350 °C may indicate structural changes that lead to the
thermal decomposition above 450 °C.
Figure 15. Rietveld refinement of room temperature XRD pattern for
Na2Ti3O7 (wRp=3%).
Figure 13. Rietveld refinement of room temperature XRD pattern for the
Na2+2δFe2ꢀδ(SO4)3 with wRp= 3.4%. In this sample ca 4% βꢀFeSO4 is
20
present as an impurity phase.
40
Figure 16 Variation of lattice parameters with temperature for
Na2Ti3O7.
insertion type active materials during cycling), so that it appears
feasible to operate the envisaged allꢀsolidꢀstate battery over this
45 entire temperature range.
Full cell performance test
Figure 17 shows selected charge discharge cycles of allꢀsolidꢀ
state Na2+2δFe2ꢀδ(SO4)3 | Na3.1Sn0.1P0.9S4 | Na2Ti3O7 battery cycled
50 between 1.5 and 4.0V at room temperate or 80 °C with currents
corresponding to rates of 0.1 or 2C. Figure 16 then displays the
corresponding variation of the specific capacity with the number
of cycles. When charged and discharged at 0.1 C rate the first
discharge capacity at room temperature reaches 113.5 mAh per g
55 of Na2+2δFe2ꢀδ(SO4)3 and the first charge capacity 107.5 mAh/g
(see Fig. 16(a)), both close to the theoretical capacity of 113
mAh/g and a discharge capacity of 94 mAh/g is retained after the
10th cycle. When charged and discharged at a rate of 2C (Fig. 16
(b)) still an initial discharge capacity of 84 mAh/g is achieved.
60 Over the first 30 room temperature cycles with a rate of 2C the
Figure 14. Temperature dependence of lattice parameters of
Na2+2δFe2ꢀδ(SO4)3 between room temperature and 400°C.
The Rietveld refinement of the anode material in figure 15
25 demonstrates the formation of phase pure Na2Ti3O7 with space
group P21/m, and the refined room temperature lattice parameters
a = 8.5651(2) Å, b = 3.8029(1) Å, c = 9.1279(1) Å, β =
43
101.60(2)° are similar to previously reported values.42,
The
temperature dependence of the lattice constants (see figure 16) up
30 to 350 °C translates into coefficients of thermal expansion of
1.9×10ꢀ5 Kꢀ1, 0.88×10ꢀ5 Kꢀ1 and 1.4×10ꢀ5 Kꢀ1 along the a, b and c
axis, respectively.

Page 9 of 12
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C6TA09809F
specific discharge capacity drops to 47 mAh/g and thereafter 20 onwards. The additional peaks in the first discharge cycles may
continues to fade to 16 mAh/g at the 100th cycle (see Fig. 18).
It should be mentioned that the voltage profile of the initial
cycles differ slightly different from those of the later cycles as
recently also found by Oyama et al.39 The differential
indicate some irreversible structural transformations, but might
also represent the different potentials for the removal of Na from
the three distinct Na sites in Na2+2δFe2ꢀδ(SO4)3. In contrast, the
charging process consistently shows a single peak around 3.1 V
25 only, which might suggest that the kinetics of the charging
process (as well as of the later cycles of the discharge) in our allꢀ
solid state cells is controlled by the charge transfer at the
interfaces of the alluaudites.
5
galvanostatic profiles (dQ/dV) of Na2+2δFe2ꢀδ(SO4)3
|
Na3.1Sn0.1P0.9S4 | Na2Ti3O7 battery (Fig. 19) show besides the
main peak at 2.7V vs. Na/Na+ two additional minor peaks (1.7,
2.0 V vs. Na/Na+) during the first discharge, one additional peak
10 for the second discharge (1.7 V) and only the main peak at 2.7 V
from the third discharge cycle
In summary, all components of the envisaged allꢀsolidꢀstate
30 battery remain (kinetically) stable between room temperature and
at least 250 °C and show only moderate volume changes of 1 ꢀ 3
% over this temperature range (comparable to the volume
changes of insertion type active materials during cycling), so that
it appears feasible to operate the envisaged allꢀsolidꢀstate battery
35 over this entire temperature range.
Figure 19. Differential capacity plots of Na2+2δFe2ꢀδ(SO4)3 |
Na3.1Sn0.1P0.9S4 | Na2Ti3O7 batteryfor selected cycles at 2C rate.
Figure 16c shows that the first discharge capacity at 80 ˚C and
40 a rate of 2C is with 109 mAh/g of Na2+2δFe2ꢀδ(SO4)3 close to the
theoretical capacity (ca. 113 mAh/g). Fig. 18 then demonstrates
that 91% of the initial capacity is retained after 50 cycles, and
82% over 100 cycles.
Figure 20 compares the current that can be drawn from the allꢀ
45 solidꢀstate sodiumꢀion battery demonstrated in this work with the
Figure 17. Selected chargeꢀdischarge profiles of
Na2+2δFe2ꢀδ(SO4)3 | Na3.1Sn0.1P0.9S4 | Na2Ti3O7 batteries at room
15
temperature (a, b) and 80°C (c) for 0.1C (a) and 2C (b,c).
Figure 20. Comparison of the current that can be drawn from allꢀsolidꢀ
state full Naꢀion cells in this work with data from the recent literature on
allꢀsolidꢀstate full Naꢀion cells and allꢀsolidꢀstate Naꢀion half cells. For
50
references see Table 1. Literature data for cathode materials tested in
liquid electrolytes are shown as crosses. As for most allꢀsolidꢀstate cells
only low rate capacity data were available, characteristic current data are
consistently determined as experimental cathode capacity times C rate.
For full cell data the respective operation temperatures are indicated;
Figure 18. Variation of specific capacity for allꢀsolidꢀstate
Na2+2δFe2ꢀδ(SO4)3 | Na3.1Sn0.1P0.9S4 | Na2Ti3O7 batterywith 2C rate at 80 ⁰C
(diamonds) and at room temperature (triangles).
55 other data points refer to room temperature values at a rate of 2C (or the
nearest C rate published).

Journal of Materials Chemistry A
Page 10 of 12
View Article Online
DOI: 10.1039/C6TA09809F
11.
12.
13.
M. Chen, R. P. Rao and S. Adams, Solid State Ionics, 2014,
268, 300ꢀ304.
M. Chen, R. P. Rao and S. Adams, Solid State Ionics, 2014,
262, 183ꢀ187.
A. Hayashi, K. Noi, A. Sakuda and M. Tatsumisago, Nature
Communications, 2012, 3, 856.
N. Tanibata, K. Noi, A. Hayashi and M. Tatsumisago, RSC
Advances, 2014, 4, 17120ꢀ17123.
I.ꢀH. Chu, C. S. Kompella, H. Nguyen, Z. Zhu, S. Hy, Z.
Deng, Y. S. Meng and S. P. Ong, Scientific Reports, 2016, 6,
33733.
M. Tatsumisago and A. Hayashi, International Journal of
Applied Glass Science, 2014, 5, 226ꢀ235.
A. Banerjee, K. H. Park, J. W. Heo, Y. J. Nam, C. K. Moon, S.
M. Oh, S. T. Hong and Y. S. Jung, Angewandte Chemie, 2016,
128, 9786ꢀ9790.
A. Hayashi, K. Noi, N. Tanibata, M. Nagao and M.
Tatsumisago, Journal of Power Sources, 2014, 258, 420ꢀ423.
N. Tanibata, T. Matsuyama, A. Hayashi and M. Tatsumisago,
Journal of Power Sources, 2015, 275, 284ꢀ287.
F. Lalère, J. B. Leriche, M. Courty, S. Boulineau, V. Viallet,
C. Masquelier and V. Seznec, Journal of Power Sources,
2014, 247, 975ꢀ980.
T. Wei, Y. Gong, X. Zhao and K. Huang, Advanced
Functional Materials, 2014, 24, 5380ꢀ5384.
M. Ribes, B. Barrau and J. Souquet, Journal of Non-
Crystalline Solids, 1980, 38, 271ꢀ276.
S. Susman, L. Boehm, K. Volin and C. Delbecq, Solid State
Ionics, 1981, 5, 667ꢀ669.
M. Jansen and U. Henseler, Journal of Solid State Chemistry,
1992, 99, 110ꢀ119.
Z. Zhu, I.ꢀH. Chu, Z. Deng and S. P. Ong, Chem. Mat., 2015,
27, 8318ꢀ8325.
S.ꢀH. Bo, Y. Wang, J. C. Kim, W. D. Richards and G. Ceder,
Chem. Mat., 2015, 28, 252ꢀ258.
A. Banerjee, K. H. Park, J. W. Heo, Y. J. Nam, C. K. Moon, S.
M. Oh, S.ꢀT. Hong and Y. S. Jung, Angewandte Chemie
International Edition, 2016, 55, 9634ꢀ9638.
available literature allꢀsolidꢀstate sodiumꢀion full and half cell
data (listed in Table 1). Therefrom it can be seen that this is the
first reported full cell allꢀsolidꢀstate sodiumꢀion battery that
achieves a current capability comparable to those in liquid
electrolyte cells at room temperature.
60
5
65 14.
15.
Conclusions
Na3+xMxP1ꢀxS4 phases with various tetravalent dopants M= Ge4+,
Ti4+, Sn4+ (x = 0, 0.1) have been prepared using ball milling
followed by annealing at 250°C. In line with our parallel
70 16.
17.
10 computational studies, we find the highest sodium ionic
conductivity when using Sn as the dopant Na3.1Sn0.1P0.9S4. The
computational studies also demonstrate that the overcrowding of
Na sites is more important than the transition from tetragonal to
cubic symmetry in enhancing the ionic conductivity and that the
15 effective migration barrier will in this compound slightly
decrease with increasing dopant size. The high potential of Snꢀ
doped cubic Na3PS4 as a solid electrolyte is demonstrated by
realising for the first time a full cell allꢀsolidꢀstate sodium ion
battery with a first discharge capacity near theoretical cathode
20 capacity of 113 mAh/g both at room temperature (when cycled
with 0.1C) and at 80 ˚C (when cycled with 2C rate). Over 100
cycles with 2C rate 80% of the capacity is retained at 80 ˚C;
while the two previous reports in literature on full allꢀsolid state
cells could only demonstrate up to 10 or 26 cycles, respectively
25 and worked only at high temperatures. All components of our cell
are stable to beyond 250 ˚C with moderate thermal expansion
<3%, so that the cell should be able to operate over this much
wider temperature range.
75 18.
19.
20.
80
21.
22.
85
23.
24.
90 25.
26.
27.
95
28.
29.
S. Okada, Y. Takahashi, T. Kiyabu, T. Doi, J.ꢀI. Yamaki and
T. Nishida, 2006.
Acknowledgements
K. Trad, D. Carlier, L. Croguennec, A. Wattiaux, M. Ben
Amara and C. Delmas, Chem. Mat., 2010, 22, 5554ꢀ5562.
A. Padhi, K. Nanjundaswamy, C. Masquelier and J.
Goodenough, Journal of the Electrochemical Society, 1997,
144, 2581ꢀ2586.
B. Ellis, W. Makahnouk, Y. Makimura, K. Toghill and L.
Nazar, Nat. Mater., 2007, 6, 749ꢀ753.
P. Moreau, D. Guyomard, J. Gaubicher and F. Boucher, Chem.
Mat., 2010, 22, 4126ꢀ4128.
P. Barpanda, T. Ye, S.ꢀi. Nishimura, S.ꢀC. Chung, Y. Yamada,
M. Okubo, H. Zhou and A. Yamada, Electrochemistry
Communications, 2012, 24, 116ꢀ119.
H. Kim, I. Park, D.ꢀH. Seo, S. Lee, S.ꢀW. Kim, W. J. Kwon,
Y.ꢀU. Park, C. S. Kim, S. Jeon and K. Kang, Journal of the
American Chemical Society, 2012, 134, 10369ꢀ10372.
P. Barpanda, G. Oyama, S.ꢀi. Nishimura, S.ꢀC. Chung and A.
Yamada, Nature communications, 2014, 5.
L. L. Wong, H. M. Chen and S. Adams, submitted, 2016.
L. Wong, H. Chen and S. Adams, Physical Chemistry
Chemical Physics, 2015, 17, 9186ꢀ9193.
G. Oyama, S.ꢀi. Nishimura, Y. Suzuki, M. Okubo and A.
Yamada, ChemElectroChem, 2015, 2, 1019ꢀ1023.
G. Oyama, O. Pecher, K. J. Griffith, S.ꢀi. Nishimura, R.
Pigliapochi, C. P. Grey and A. Yamada, Chem. Mat., 2016, 28,
5321ꢀ5328.
30 This research was supported by the National Research
Foundation, Prime Minister’s Office, Singapore under its
Competitive Research Programme (CRP Award NRFꢀCRP 10ꢀ
2012ꢀ6) and the NUS “Centre for Energy Research” seed grant.
100
30.
31.
32.
33.
105
110
115
120
125
130
Notes and references
35 a Department of Materials Science and Engineering, National University
of Singapore, Singapore-117575, Singapore. Fax:+65- 6776 3604;
Tel: +65-6516 6869; E-mail: mseasn@nus.edu.sg.
34.
35.
1.
2.
3.
J. Sudworth and A. Tiley, Sodium Sulphur Battery, Springer
Science & Business Media, 1985.
T. Oshima, M. Kajita and A. Okuno, International Journal of
Applied Ceramic Technology, 2004, 1, 269ꢀ276.
B. L. Ellis and L. F. Nazar, Current Opinion in Solid State and
Materials Science, 2012, 16, 168ꢀ177.
40
36.
37.
45 4.
5.
J. Sudworth, Journal of power sources, 1994, 51, 105ꢀ114.
M. Hosseinifar and A. Petric, Journal of Power Sources, 2012,
206, 402ꢀ408.
38.
39.
6.
50 7.
8.
T. Girija and A. V. Virkar, Journal of Power Sources, 2008,
180, 653ꢀ656.
C.ꢀW. Park, H.ꢀS. Ryu, K.ꢀW. Kim, J.ꢀH. Ahn, J.ꢀY. Lee and
H.ꢀJ. Ahn, Journal of power sources, 2007, 165, 450ꢀ454.
J.ꢀS. Kim, H.ꢀJ. Ahn, I.ꢀP. Kim, K.ꢀW. Kim, J.ꢀH. Ahn, C.ꢀW.
Park and H.ꢀS. Ryu, J Solid State Electrochem, 2008, 12, 861ꢀ
865.
40.
41.
W. Luo, F. Shen, C. Bommier, H. Zhu, X. Ji and L. Hu,
Accounts Chem. Res., 2016, 49, 231ꢀ240.
H. Kim, J. Hong, G. Yoon, H. Kim, K.ꢀY. Park, M.ꢀS. Park,
W.ꢀS. Yoon and K. Kang, Energy & Environmental Science,
2015, 8, 2963ꢀ2969.
55 9.
C.ꢀW. Park, J.ꢀH. Ahn, H.ꢀS. Ryu, K.ꢀW. Kim and H.ꢀJ. Ahn,
Electrochemical and solid-state letters, 2006, 9, A123ꢀA125.
R. P. Rao, N. Sharma, V. K. Peterson and S. Adams, Solid
State Ionics, 2013, 230, 72ꢀ76.
42.
P. Senguttuvan, G. Rousse, V. Seznec, J.ꢀM. Tarascon and M.
R. Palacin, Chem. Mat., 2011, 23, 4109ꢀ4111.
10.

Page 11 of 12
Journal of Materials Chemistry A
View Article Online
DOI: 10.1039/C6TA09809F
43.
44.
5
A. Rudola, K. Saravanan, C. W. Mason and P. Balaya, Journal
of Materials Chemistry A, 2013, 1, 2653ꢀ2662.
J. NavaꢀAvendaño, A. MoralesꢀGarcía, A. Ponrouch, G.
Rousse, C. Frontera, P. Senguttuvan, J.ꢀM. Tarascon, M.
Arroyoꢀde Dompablo and M. Palacín, Journal of Materials
Chemistry A, 2015, 3, 22280ꢀ22286.
A. Rudola, N. Sharma and P. Balaya, Electrochemistry
Communications, 2015, 61, 10ꢀ13.
C. Didier, M. Guignard, C. Denage, O. Szajwaj, S. Ito, I.
Saadoune, J. Darriet and C. Delmas, Electrochemical and
Solid-State Letters, 2011, 14, A75ꢀA78.
D. Hamani, M. Ati, J.ꢀM. Tarascon and P. Rozier,
Electrochemistry communications, 2011, 13, 938ꢀ941.
J.ꢀS. Kim, H.ꢀJ. Ahn, H.ꢀS. Ryu, D.ꢀJ. Kim, G.ꢀB. Cho, K.ꢀW.
Kim, T.ꢀH. Nam and J. H. Ahn, Journal of Power Sources,
2008, 178, 852ꢀ856.
72.
75 73.
P. E. Blöchl, Physical Review B, 1994, 50, 17953ꢀ17979.
G. Kresse and D. Joubert, Physical Review B, 1999, 59, 1758ꢀ
1775.
M. Hellenbrandt, Crystallography Reviews, 2004, 10, 17ꢀ22.
H. M. Chen, C. Maohua and S. Adams, Physical Chemistry
Chemical Physics, 2015, 17, 16494ꢀ16506.
J.ꢀC. Jumas, E. Philippot, F. VermotꢀGaudꢀDaniel, M. Ribes
and M. Maurin, Journal of Solid State Chemistry, 1975, 14,
319ꢀ327.
K. O. Klepp, Zeitschrift für Naturforschung B, 1985, 40, 878ꢀ
882.
N. J. J. de Klerk and M. Wagemaker, Chem. Mat., 2016, 28,
3122ꢀ3130.
74.
75.
45.
80 76.
46.
10
77.
85 78.
79.
47.
48.
L. Zhang, K. Yang, J. Mi, L. Lu, L. Zhao, L. Wang, Y. Li and
H. Zeng, Energy Mater, 2015, 5, 1501294.
15
49.
S. I. Park, I. Gocheva, S. Okada and J.ꢀi. Yamaki, Journal of
The Electrochemical Society, 2011, 158, A1067ꢀA1070.
K. Saravanan, C. W. Mason, A. Rudola, K. H. Wong and P.
Balaya, Advanced Energy Materials, 2013, 3, 444ꢀ450.
Y.ꢀU. Park, D.ꢀH. Seo, H.ꢀS. Kwon, B. Kim, J. Kim, H. Kim,
I. Kim, H.ꢀI. Yoo and K. Kang, Journal of the American
Chemical Society, 2013, 135, 13870ꢀ13878.
50.
20
51.
52.
H. Kim, G. Yoon, I. Park, K.ꢀY. Park, B. Lee, J. Kim, Y.ꢀU.
Park, S.ꢀK. Jung, H.ꢀD. Lim, D. Ahn, S. Lee and K. Kang,
Energy & Environmental Science, 2015, 8, 3325ꢀ3335.
S. Y. Lim, H. Kim, J. Chung, J. H. Lee, B. G. Kim, J.ꢀJ. Choi,
K. Y. Chung, W. Cho, S.ꢀJ. Kim, W. A. Goddard, Y. Jung and
J. W. Choi, Proceedings of the National Academy of Sciences,
2014, 111, 599ꢀ604.
25
53.
30
54.
C. S. Park, H. Kim, R. A. Shakoor, E. Yang, S. Y. Lim, R.
Kahraman, Y. Jung and J. W. Choi, Journal of the American
Chemical Society, 2013, 135, 2787ꢀ2792.
55.
35
P. Barpanda, G. Liu, M. Avdeev and A. Yamada,
ChemElectroChem, 2014, 1, 1488ꢀ1491.
56.
Y. Kawabe, N. Yabuuchi, M. Kajiyama, N. Fukuhara, T.
Inamasu, R. Okuyama, I. Nakai and S. Komaba,
Electrochemistry Communications, 2011, 13, 1225ꢀ1228.
X. Lin, X. Hou, X. Wu, S. Wang, M. Gao and Y. Yang, RSC
Advances, 2014, 4, 40985ꢀ40993.
57.
40
58.
Y. Zhu, Y. Xu, Y. Liu, C. Luo and C. Wang, Nanoscale, 2013,
5, 780ꢀ787.
59.
P. P. Prosini, C. Cento, A. Masci and M. Carewska, Solid State
Ionics, 2014, 263, 1ꢀ8.
45 60.
J. Song, L. Wang, Y. Lu, J. Liu, B. Guo, P. Xiao, J.ꢀJ. Lee, X.ꢀ
Q. Yang, G. Henkelman and J. B. Goodenough, Journal of the
American Chemical Society, 2015, 137, 2658ꢀ2664.
C.ꢀY. Yu, J.ꢀS. Park, H.ꢀG. Jung, K.ꢀY. Chung, D. Aurbach,
61.
Y.ꢀK. Sun and S.ꢀT. Myung, Energy
Science, 2015, 8, 2019ꢀ2026.
& Environmental
50
62.
Y.ꢀL. Ruan, K. Wang, S.ꢀD. Song, X. Han and B.ꢀW. Cheng,
Electrochimica Acta, 2015, 160, 330ꢀ336.
63.
M. Jayakumar, K. Hemalatha, K. Ramesha and A. S. Prakash,
Physical Chemistry Chemical Physics, 2015, 17, 20733ꢀ
20740.
55
64.
H. Pan, X. Lu, X. Yu, Y.ꢀS. Hu, H. Li, X.ꢀQ. Yang and L.
Chen, Advanced Energy Materials, 2013, 3, 1186ꢀ1194.
H. Chen, Q. Hao, O. Zivkovic, G. Hautier, L.ꢀS. Du, Y. Tang,
Y.ꢀY. Hu, X. Ma, C. P. Grey and G. Ceder, Chem. Mat., 2013,
25, 2777ꢀ2786.
65.
60
66.
S. Adams and R. P. Rao, in Bond Valences, eds. I. D. Brown
and K. R. Poeppelmeier, Springer Berlin Heidelberg, Berlin,
Heidelberg, 2014, DOI: 10.1007/430_2013_137, pp. 129ꢀ159.
S. Adams and R. P. Rao, physica status solidi (a), 2011, 208,
1746ꢀ1753.
67.
65
68.
S. Adams and R. Prasada Rao, Solid State Ionics, 2011, 184,
57ꢀ61.
69.
70 70.
71.
A. Hayashi, K. Noi, N. Tanibata, M. Nagao and M.
Tatsumisago, Journal of Power Sources, 2014, 258, 420ꢀ423.
G. Kresse and J. Furthmüller, Physical Review B, 1996, 54,
11169ꢀ11186.
J. P. Perdew, K. Burke and M. Ernzerhof, Physical Review
Letters, 1996, 77, 3865ꢀ3868.

Journal of Materials Chemistry A
Page 12 of 12
View Article Online
TOC entry
DOI: 10.1039/C6TA09809F
Fast ion conducting Na3.1P0.9Sn0.1S4 is combined with alluaudite and Na2Ti3O7 to achieve a low-cost
high power all-solid state Na-ion battery