Citation for published version:
Deng, Y, Eames, C, Chotard, JN, Laleìre, F, Seznec, V, Emge, S, Pecher, O, Grey, CP, Masquelier, C & Islam,
MS 2015, 'Structural and mechanistic insights into fast lithium-ion conduction in Li
4
SiO
4
-Li
3
PO
4
solid
electrolytes', Journal of the American Chemical Society, vol. 137, no. 28, pp. 9136-9145.
https://doi.org/10.1021/jacs.5b04444
DOI:
10.1021/jacs.5b04444
Publication date:
2015
Document Version
Peer reviewed version
Link to publication
This document is the Accepted Manuscript version of a Published Work that appeared in final form in Journal of
the American Chemical Society, copyright © American Chemical Society after peer review and technical editing
by the publisher. To access the final edited and published work see
http://pubs.acs.org/doi/abs/10.1021/jacs.5b04444
University of Bath
Alternative formats
If you require this document in an alternative format, please contact:
openaccess@bath.ac.uk
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners
and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
Take down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.
Download date: 10. Aug. 2022
1 / 12
Structural and mechanistic insights into fast lithium-ion conduc-
tion in Li
4
SiO
4
-Li
3
PO
4
solid electrolytes
Yue Deng
†‡
, Christopher Eames
‡
, Jean-Noël Chotard
†
, Fabien Lalère
†
, Vincent Seznec
†
, Steffen
Emge
§
, Oliver Pecher
§
, Clare P. Grey
§
, Christian Masquelier
†
, M. Saiful Islam
‡
*
†
Laboratoire de Réactivité et Chimie des Solides (UMR CNRS 7314), Université de Picardie Jules Verne, 33 rue Saint Leu, 80039 Amiens Cedex, France
‡
Department of Chemistry, University of Bath, Bath, BA2 7AY, UK
§
Department of Chemistry, University of Cambridge, Cambridge, CB2 1EW, UK
ABSTRACT: Solid electrolytes that are chemically stable and have a high ionic conductivity would dramatically enhance the
safety and operating lifespan of rechargeable lithium batteries. Here, we apply a multi-technique approach to the Li-ion con-
ducting system (1-z)Li
4
SiO
4
–(z)Li
3
PO
4
with the aim of developing a solid electrolyte with enhanced ionic conductivity. Previ-
ously unidentified superstructure and immiscibility features in high purity samples are characterized by X-ray and neutron
diffraction across a range of compositions (z = 0.0 to 1.0). Ionic conductivities from AC impedance measurements and large-
scale molecular dynamics (MD) simulations are in good agreement, showing very low values in the parent phases (Li
4
SiO
4
and Li
3
PO
4
), but orders of magnitude higher conductivities (10
-3
S/cm at 573 K) in the mixed compositions. The MD simula-
tions reveal new mechanistic insights in the mixed Si/P compositions in which Li
ion conduction occurs through 3D pathways
and a cooperative interstitial mechanism; such correlated motion is a key factor in promoting high ionic conductivity. Solid
state
6
Li,
7
Li and
31
P NMR experiments reveal enhanced local Li-ion dynamics and atomic disorder in the solid solutions, which
are correlated to the ionic diffusivity. These unique insights will be valuable in developing strategies to optimize the ionic
conductivity in this system and to identify next-generation solid electrolytes.
1. Introduction
The revolution in portable electronic devices has been
powered by rechargeable lithium-ion batteries. Such bat-
teries with liquid electrolytes, however, have cycle life and
safety issues, whereas all-solid-state batteries with inor-
ganic electrolytes may be regarded as a safer long-term so-
lution.
1–6
Many structural families
7–12
have been investi-
gated to identify potential ion-conducting solid electrolytes,
including framework-type materials based on NASICON,
LISICON, thio-LISICON and garnet structures. Recently, a
sulfur-based Li
2
S-P
2
S
5
glass-ceramic solid electrolyte has
been reported to show an ionic conductivity (1.7 x 10
-2
S/cm at room temperature) higher than that of many com-
monly-used liquid electrolytes.
13
However, sulfide-based
electrolytes are very hydroscopic
14
and must be prepared
in a water-free environment. They also operate in limited
voltage windows. Although oxides do not currently have as
high ionic conductivity as sulfides, they exhibit higher sta-
bility and are easier to synthesize and handle.
The Li
4
SiO
4
-Li
3
PO
4
solid solution system and the end-
member parent phases have been identified as potential
solid electrolytes,
15–27
but not all compositions have been
fully characterized. It has been reported that the ionic con-
ductivity can be increased by three orders of magnitude for
Li
4-x
Si
1-x
P
x
O
4
or -Li
3+y
Si
y
P
1-y
O
4
compositions, compared
with the two end members, Li
4
SiO
4
and –Li
3
PO
4
.
18
Figure 1. Schematic representations of the crystal structures of the
end member phases Li
4
SiO
4
(subcell) and -Li
3
PO
4
. Key: SiO
4
and PO
4
-
blue and red tetrahedra; lithium - green. Partially occupied Li sites are
represented by partial shading. Unit cells are shown as solid lines.
This drastic enhancement of conductivity is believed to be
caused by the increased concentration of Li-ion vacancies
γ-Li
3
PO
4
Li
4
SiO
4
2 / 12
in Li
4-x
Si
1-x
P
x
O
4
and Li-ion interstitials in Li
3+y
Si
y
P
1-y
O
4
, with
the ionic defects acting as charge carriers. However, an
atomic-scale understanding of the conduction mechanisms
and local structures in this system is still lacking; such de-
tail is important for developing strategies for optimizing
the conductivity, as well as identifying next-generation ma-
terials.
Both Li
4
SiO
4
and –Li
3
PO
4
are related to the LISICON-
type structure with XO
4
-based (X=Si or P) tetrahedral units
(Figure 1), and Li-O polyhedra. The crystal structure of
Li
4
SiO
4
was initially reported
28
to be monoclinic (space
group P2
1
/m, with β very close to 90°) and to contain six
partially occupied crystallographically independent lithium
sites. All lithium sites are fully occupied, giving rise to a
complex 7-fold superstructure.
29
–Li
3
PO
4
crystallizes in a
similar structure to Li
4
SiO
4
, but with orthorhombic sym-
metry (space group Pnma), as illustrated in Figure 1. The
positions of the Si and P atoms are quite similar, but the re-
spective orientations of the SiO
4
and PO
4
tetrahedra are dif-
ferent: up-down-up-down for Li
4
SiO
4
, and down-down-up-
up for -Li
3
PO
4
(along
).
However, there have been limited structural studies of
the complete solid solution. Early work of Hu
15
and West
30
determined that the Li
4
SiO
4
-Li
3
PO
4
system did not yield one
single continuous solid solution, as both end members do
not possess the same crystal structure, despite being
closely related to each other. More recent work of Arachi et
al.
18
proposed x
max
=0.40 for Li
4-x
Si
1-x
P
x
O
4
and y
max
=0.40 for
Li
3+y
Si
y
P
1-y
O
4
, while the existence of single-phase
Li
3.5
Si
0.5
P
0.5
O
4
remains uncertain.
Here, we investigate the crystal chemistry and ion
transport properties of solid solution compositions within
the Li
4
SiO
4
—Li
3
PO
4
system, using a powerful combination
of experimental and computational techniques. Very high
purity Li
4
SiO
4
, Li
3.75
Si
0.75
P
0.25
O
4
, Li
3.5
Si
0.5
P
0.5
O
4
,
Li
3.25
Si
0.25
P
0.75
O
4
and -Li
3
PO
4
samples were obtained and
further investigated by various diffraction (single crystal,
powder X-ray, powder neutron) techniques. Solid-state
6
Li,
7
Li and
31
P NMR spectroscopy is used to provide insights
into local structural ordering and Li-ion dynamics. The
crystal structures and ion transport are then investigated
by atomistic molecular dynamics techniques, in which the
relative energies of possible defect arrangements are con-
sidered, as well as large-scale simulations of ion transport
mechanisms.
2. Methods
Synthesis The raw materials used in this work were all
purchased from Sigma-Aldrich, including LiOH·H
2
O (98%),
SiO
2
(fumed, dried at 500°C for 3 hours) and β-Li
3
PO
4
pow-
ders. Stoichiometric mixtures of raw materials were dis-
persed in distilled water, the quantity of which was ad-
justed to give an overall concentration of 0.6 mol of Li per
H
2
O liter. The solution was then slowly heated up to 80°C to
evaporate water. The resulting solid was pulverized and
cold pressed into pellets under 40 MPa. The pellets were
then heated in alumina combustion crucibles at 900°C for
10 hours under Ar flow, followed by slow cooling to room
temperature, and then pulverized. Energy Dispersive X-ray
analysis (EDX) indicates no contamination of Al from cruci-
ble. SEM shows the synthesized samples have aggregates of
about 100 µm in size and smaller individual particles of
about 10 to 50 µm. Since these samples are quite reactive
with CO
2
,
31
they were stored in a glove box. Single crystals
of Li
4
SiO
4
were prepared by a similar method, except that
the final heating temperature was set to 1200°C and the
cooling rate was set to 10°C/hour. To produce pure -Li
3
PO
4
,
commercially available β-Li
3
PO
4
was heated at 850
o
C for 10
hours, cooled down to room temperature and pulverized.
Diffraction X-ray powder diffraction (XRPD) patterns
were collected from a Bruker D8 diffractometer (Cu-Kα ra-
diation, θ-θ configuration). High quality diffraction patterns
were recorded overnight between 2θ ranges of 10 to 100°,
with a step size of 0.009° and a scan rate of 3.8 seconds per
step. For Li
3.75
Si
0.75
P
0.25
O
4
a neutron powder diffraction ex-
periment was carried out at the SINQ spallation source
32
of
the Paul Scherrer Institute (Switzerland) using the high-
resolution diffractometer for neutrons HRPT
33
(λ = 1.494Å).
Neutron powder diffraction patterns were collected for
Li
3.5
Si
0.5
P
0.5
O
4
and Li
3.25
Si
0.25
P
0.75
O
4
using the high-resolution
D2B diffractometer at Institute Laue-Langevin (Grenoble,
France). High quality diffraction patterns were recorded
between 2θ ranges of 10 to 160°, with a step size of 0.05°,
accumulated over 6 hours. X-ray single crystal diffraction
measurements were carried out at 293K using a Bruker D8
Venture diffractometer, with Mo Kα radiation (multilayer
optics monochromator). Data collecting conditions, crystal
data and refinement parameters are listed in Supplemen-
tary Information (SI) Table S1.
AC Impedance Spectroscopy For ionic conductivity
measurements, powder of Li
4
SiO
4
, Li
3
PO
4
and their solid so-
lutions were cold pressed into disk-shaped pellets. In each
case, about 150 mg of powder was placed in a graphite ma-
trix (10 mm in diameter) and cold pressed at 40 MPa. The
pellets were then sintered in a FCT Spark Plasma Sintering
apparatus at 70K min
-1
up to 700°C for 3 minutes under an
applied force of 8 kN. The resulting dense pellets were pol-
ished and metalized on both sides by gold sputtering using
a Bal-Tec SCD 050. The sintering process and metallization
step were carried out in Ar atmosphere. Pellets were then
dried under primary vacuum at 100°C overnight before
measurements, and immediately transferred into a glove
box. The sample was then introduced into the impedance
measurement cell directly in the glove box to avoid any air
contamination. Impedance measurements were performed
over a frequency range of 0.1 Hz to 200 kHz, between 25
and 300°C, both upon heating and cooling, under static Ar.
34
Solid-State NMR Spectroscopy Powder samples of
Li
4
SiO
4
, Li
3.75
Si
0.75
P
0.25
O
4
, Li
3.5
Si
0.5
P
0.5
O
4
and Li
3.25
Si
0.25
P
0.75
O
4
,
and Li
3
PO
4
were packed in 1.3, 4.0, or 7.0 mm ZrO
2
rotors
(Bruker) and closed with Kel-F or BN caps depending on the
temperature of the experiment. All sample handling was
done under argon atmosphere in a glove box with p(H
2
O, O
2
)
< 0.1 ppm. Ambient and high temperature (320 - 875 K)
7
Li
magic angle spinning (MAS) NMR experiments were per-
formed at 9.4 T (Avance I console) using a Bruker double
resonance 7.0 mm MAS probe with laser heating of the sam-
ple. Temperature calibration using KBr was done before
3 / 12
measuring the samples.
7
Li NMR signal line shapes were de-
termined by one-pulse experiments with high power pulses
of 2.1 µs (7.0 mm MAS) and 0.9 µs (4.0 mm MAS) and a rep-
etition time of 5.0 s. A saturation recovery pulse sequence
was applied to determine
7
Li spin-lattice relaxation time
constants (T
1
) at variable temperatures. Ambient tempera-
ture
6
Li MAS NMR experiments were performed at 11.7 T
(Avance III HD console) using a Bruker triple resonance 4.0
mm MAS probe including a temperature calibration on the
207
Pb shift in lead nitrate before the actual experiment.
35
6
Li
pulse optimization was done on
6
Li
2
CO
3
and followed by the
NMR signal line shape measurements on Li
3.75
Si
0.75
P
0.25
O
4
,
Li
3.5
Si
0.5
P
0.5
O
4
and Li
3.25
Si
0.25
P
0.75
O
4
using a one-pulse se-
quence with high power pulses of 4.45 µs and a repetition
time of 25 s. The
6;7
Li NMR shifts were referenced to a 1 M
LiCl solution in D
2
O.
36
Activation energies were derived by
both BPP fits
37,38
and fitting of the linear regimes in Arrhe-
nius plots (see SI Table S2). Due to the low natural abun-
dance of
6
Li (7.59 % vs.
7
Li 92.41 %)
36
and the significantly
longer relaxation times, we focused on
7
Li NMR experi-
ments in this study for the sake of higher sensitivity and
shorter measurement times.
31
P MAS NMR experiments
were performed at ambient temperature in a magnetic field
of 16.4 T using a Bruker 1.3 mm triple resonance MAS
probe on an Avance III console. Pulse optimization was
done using ammonium dihydrogen phosphate (ADP).
39
A
one-pulse sequence with high power 1.8 µs pulses and a re-
cycle delay of 3.0 s was applied to acquire the spectra for
the line-shape measurements. The
31
P NMR signals were
referenced to 85% H
3
PO
4
.
36
Atomistic modeling Interatomic potentials-based
methods, which are well established and detailed else-
where, were employed.
40–43
The effective potentials de-
scribing the interatomic forces include a long-range Cou-
lomb term, short-range Morse function and repulsive con-
tribution. The parameters were taken from the extensive li-
brary of potentials developed by Pedone et al,
44
which have
been shown to perform well in molecular dynamics (MD)
simulations of silicates and polyanion-type materials. Fur-
ther details can be found in SI Table S3. Modeling of crystal
structures and different defect ordering schemes was car-
ried out using energy minimization methods (GULP
code
41,45
). For ion diffusion modeling we have used MD
methods (DL_POLY 4 code
42
). A time step of 1 fs for MD runs
of up to 5 ns with supercells (60 x 60 x 60 Å
3
in three di-
mensions) containing 20,000 to 30,000 ions were em-
ployed. Simulations were carried out at several tempera-
tures (300 - 673 K). Each set of calculations was repeated
three times to confirm good statistics. Such computational
methods have been applied successfully to other Li-ion bat-
tery materials.
43,46–53
To facilitate comparison with experi-
mental data the calculated diffusion coefficients (D) were
used to derive the ionic conductivity σ using the Nernst-
Einstein relationship:
where n is the number of particles per unit volume, q is the
charge of an electron, k is the Boltzmann constant and T is
the temperature; H
R
is the correlation factor (or Haven ra-
tio), defined as the ratio of the tracer diffusion coefficient to
a diffusion coefficient dependent upon the ionic conductiv-
ity. In this work we have used the methods of Morgan and
Madden
54
to determine H
R
. A Haven ratio of 1.0 suggests un-
correlated ion hopping, whereas high values (> 2) are ob-
served in fast-ion conductors with highly correlated ionic
motion.
3. Results and discussion
3.1 Structures of (1-z)Li
4
SiO
4
- (z)Li
3
PO
4
solid solutions
Our study confirms that the compositions Li
4
SiO
4
and
Li
3.75
Si
0.75
P
0.25
O
4
can be indexed in space group P2
1
/m while
Li
3.5
Si
0.5
P
0.5
O
4
and Li
3.25
Si
0.25
P
0.75
O
4
adopt the
-Li
3
PO
4
structure (Figure 2a). Lattice parameters refined
from powder X-ray diffraction data are collected in Table 1.
Table 1. Cell parameters determined from X-ray powder diffraction
in the (1-z)Li
4
SiO
4
-(z)Li
3
PO
4
system. (*Subcell parameters of Li
4
SiO
4
are used for ease of comparison.)
Li
4
SiO
4
(z=0)*
z=0.25
z=0.50
z=0.75
Li
3
PO
4
(z=1)
S.G.
P2
1
/m
P2
1
/m
Pnma
Pnma
Pnma
a (Å)
5.1504(3)
5.1094(2)
10.5990(7)
10.5356(3)
10.4763(3)
b (Å)
6.1012(4)
6.1135(4)
6.1155(4)
6.1169(2)
6.1193(2)
c (Å)
5.2998(3)
5.3002(4)
5.0114(3)
4.9697(2)
4.9245(2)
β (
o
)
90.321(5)
90.378(4)
90
90
90
V/Z(Å
3
)
83.268(3)
82.777(2)
81.207(2)
80.067(2)
78.924(2)
To obtain deeper insights into the phase stabilities
within the Li
4
SiO
4
-Li
3
PO
4
system, we prepared several (1-
z)Li
4
SiO
4
—(z)Li
3
PO
4
compositions in steps of z=0.1. We
found that the immiscibility zone is around 0.35 < z < 0.45.
Between the two end members, the a, b and c parameters
vary smoothly except for an abrupt discontinuity between
the a parameter of the P-substituted Li
4
SiO
4
structure type
and the c-parameter of the Si-substituted Li
3
PO
4
phase at z
= 0.4 where two phases coexist (Figure 2b; note that
).
The XRD powder pattern of Li
4
SiO
4
was indexed and the
lattice parameters refined using the unit-cell proposed by
Völlenkle
28
in the monoclinic space group P2
1
/m. Most of
the diffraction peaks of our sample could be indexed but, as
shown in Figure 3, many small intensity contributions re-
mained unidentified. These are signatures of lithium order-
ing within Li
4
SiO
4
producing the supercell previously re-
ported from single crystal diffraction work,
29
and seen here
for the first time in powder XRD data.
Tranqui et al.
29
described the structure of Li
4
SiO
4
using
the P2
1
/m space group (a = 11.546 Å, b = 6.090 Å, c = 16.645
Å, β = 99.5° and Z = 14). The structure contains SiO
4
tetra-
hedra and LiO
n
(n = 4, 5, 6) polyhedra. The 19 fully occupied
crystallographic sites for Li are distributed over 9 4f
Wyckoff positions while the other 10 are at 2e positions:
this results in 56 lithium atoms per unit cell. A similar struc-
ture was proposed by de Jong
55
, using results from single
crystal X-ray diffraction and XPS measurements. Their
4 / 12
structure was basically the same as that in previous work,
with the main difference being the splitting of the Li(51)
and Li(65) sites (using the notation of Tranqui) over two
additional sites. It is noted that these sites in Tranqui’s
work
29
had thermal displacement parameters of 4.79 and
4.04 Å
2
, respectively, approximately twice as large as those
for the other Li atoms. All four positions then have an occu-
pancy of 0.5, which introduces a small degree of positional
disorder over the Li sublattice.
Figure 2. a) XRD patterns of powders obtained within the Li
4
SiO
4
-
Li
3
PO
4
system. Li
4
SiO
4
-based patterns are in blue, Li
3
PO
4
-based
patterns are in red. Stars (*) indicate superstructure peaks. b) Lattice
parameters variation as a function of z in (1-z) Li
4
SiO
4
– (z)Li
3
PO
4
.
Careful inspection of our Li
4
SiO
4
powder revealed that
it contained single crystals of sufficient size (~20 m) for
precise structural determination. Our refined structure
contains 19 independent lithium sites. The corresponding
Li(51) and Li(65) sites display isotropic thermal displace-
ment parameters as high as 6.48 and 3.21 Å
2
, indicating that
these two lithium ions are loosely bound to their ideal po-
sitions. We carried out a second set of refinements by split-
ting the two sites. This resulted in occupancy factors for the
sites split from Li(51) of 0.528(13) and 0.472(13) (and 0.73
Å from each other). The Li(65) was split into two sites, sep-
arated by 0.40 Å, with occupancies of 0.520(12) and
0.480(12).
Figure 3. Full pattern profile matching of X-ray powder diffraction
pattern of Li
4
SiO
4
. Red dots: experimental data; black line: profile
matching; blue bar: Bragg positions of the small cell; green bar: Bragg
positions of the 7-fold super cell.
Within the Li
4
SiO
4
-Li
3
PO
4
system, only the
Li
3.75
Si
0.75
P
0.25
O
4
composition has been previously investi-
gated
56
by single crystal X-ray diffraction. The refinement
converged to an overall composition of Li
3.43
Si
0.75
P
0.25
O
4
af-
ter summation of individual Li site occupancy factors. We
determined the crystal structures of the three solid solution
compositions Li
3.75
Si
0.75
P
0.25
O
4
, Li
3.5
Si
0.5
P
0.5
O
4
and
Li
3.25
Si
0.25
P
0.75
O
4
through Rietveld refinements of powder
neutron diffraction data. Unlike Li
4
SiO
4
, these three compo-
sitions do not exhibit long-range ordering of lithium ions at
room temperature and therefore can be described using the
original unit cells in Table 1.
The published atomic coordinates of each structure
type (Li
4
SiO
4
or -Li
3
PO
4
(ICSD-77095)
57
) were used as
starting models for the refinements of Li
3.75
Si
0.75
P
0.25
O
4
,
Li
3.5
Si
0.5
P
0.5
O
4
and
Li
3.25
Si
0.25
P
0.75
O
4
compositions. The main
challenge here was to localize the lithium atomic positions
and occupancy factors. The refinement strategy can be il-
lustrated here using Li
3.5
Si
0.5
P
0.5
O
4
as an example. -Li
3
PO
4
contains four formula units per unit cell, with one type of
PO
4
tetrahedron and two crystallographically independent
Li sites: Li1(8d) and Li2(4c), giving 12 lithium atoms per
unit cell. We first verified that the Li1 and Li2 sites of Li
3
PO
4
were appropriate for describing the structure of
Li
3.5
Si
0.5
P
0.5
O
4
by removing one or the other from the list of
atomic coordinates and calculating Fourier transformed
scattering density difference maps (Figure 4). These maps
clearly reveal sharp difference density peaks at the atomic
positions of Li1 and Li2, which were then subsequently
used to refine the overall structure, as well as other posi-
tions in the unit cell.
2θ(°), λ
Cu
γ-Li
3
PO
4
(Pnma)
Li
4
SiO
4
(P2
1
/m)
z = 0.25
z = 0.50
z = 0.75
Intensity (
a.u.)
a)
a_Li
4
SiO
4
c_Li
4
SiO
4
b_Li
4
SiO
4
c_Li
3
PO
4
(a_Li
3
PO
4
)/2
b_Li
3
PO
4
z in (1-z) Li
4
SiO
4
– z Li
3
PO
4
Cell parameters (Å)
b)
*
*
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
5.20
5.30
5.40
6.09
6.10
6.11
6.12
4.90
5.00
5.10
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
10 20 30 40 50
15 20 25 30 35 40 45 50
Li
4
SiO
4
Monoclinic
P2
1
/m
a = 11.608(1) Å
b = 6.099(6) Å
c = 16.681(2) Å
β = 99.49(1)°
V/z = 83.1 Å
3
2θ(°), λ
Cu
Intensity (
a.u.)
30 35 40 45
