You are on page 1of 8

Journal of Power Sources 412 (2019) 29–36

Contents lists available at ScienceDirect

Journal of Power Sources


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

A lithium argyrodite Li6PS5Cl0.5Br0.5 electrolyte with improved bulk and T


interfacial conductivity
Heng Wanga,b,∗,1, Chuang Yub,1, Swapna Ganapathyb, Ernst R.H. van Eckc, Lambert van Eijckb,
Marnix Wagemakerb
a
College of Science, Guilin University of Technology, Guilin, 541004, PR China
b
Department of Radiation Science and Technology, Delft University of Technology, Mekelweg 15, 2629JB, Delft, the Netherlands
c
Institute for Molecules and Materials, Radboud University Nijmegen, Heyendaalseweg 135, 6525 AJ, Nijmegen, the Netherlands

H I GH L IG H T S G R A P H I C A L A B S T R A C T

• Liductivity
6PS Cl Br
5 0.5 shows an ionic con-
0.5
of 3.6 mS cm−1
at room
temperature.
• The rate of the halogen occupancy on
4c site decreased with increased po-
larizability.
• All-solid-state cells exhibited a capa-
city of 820 mAh g−1 up to 30 cycles at
0.02 C.
• A higher interfacial conductivity is
achieved between the Li6PS5Cl0.5Br0.5-
Li2S.

A R T I C LE I N FO A B S T R A C T

Keywords: The increased safety associated with all-solid-state batteries using inorganic ceramic electrolytes make it a
Argyrodite promising technology, with potential to replace current commercial battery systems. The key challenges to
Superionic conductor realize this technology are the development of new solid electrolytes with high ionic conductivity and optimi-
Impedance zation of the ionic transport pathways across the multiple phases of the battery. In this study an optimal
All-solid-state
composition of the argyrodite i.e. Li6PS5Cl0.5Br0.5 is synthesized via the mechanical milling method. This ma-
NMR
terial possesses a higher bulk ionic conductivity and reduced activation energy than the single halogen doped
argyrodites i.e. Li6PS5X (X = Cl and Br), assessed by temperature-dependent impedance spectroscopy and
Nuclear Magnetic Resonance (NMR) relaxometry. A combined X-ray and neutron diffraction analysis reveals an
influence of the composition and distribution of halogen atoms on the Li-ion conductivity. All-solid-state bat-
teries fabricated using Li2S as cathode show a high reversible capacity of 820 mAh g−1 for up to 30 cycles. In
addition, the Li-ion diffusion across the interface between the Li2S cathode and Li6PS5Cl0.5Br0.5 electrolyte is
probed by exchange NMR spectroscopy. It reveals that Li-ion diffusion across this interface was the main factor
limiting the performance of Li6PS5Cl0.5Br0.5 in the battery, despite its high bulk ionic conductivity.

1. Introduction portable electronics and electric vehicles (EVs) [1–4] use organic liquid
electrolytes. As a consequence, they suffer from safety issues inherent to
Conventional commercial lithium ion batteries widely applied in the use of these liquid electrolytes which include fire hazards and gas


Corresponding author. College of Science, Guilin University of Technology, Guilin, PR China.
E-mail address: wangh@glut.edu.cn (H. Wang).
1
These authors contributed equally to this work.

https://doi.org/10.1016/j.jpowsour.2018.11.029
Received 21 August 2018; Received in revised form 16 October 2018; Accepted 9 November 2018
Available online 29 November 2018
0378-7753/ © 2018 Elsevier B.V. All rights reserved.
H. Wang et al. Journal of Power Sources 412 (2019) 29–36

evolution. These safety concerns could be mitigated by the use of solid- reported [38], which originated from the diffusion of non-lithium ions
state inorganic electrolytes. These materials also provide the possibility and/or the chemical reaction between the cathode and electrolyte [39].
of improving the energy density and cycling performance of the battery These scenarios make it imperative to investigate Li-ion diffusion
by the use of more efficient cell packing and their higher electro- through the interface for guiding the design of all-solid-state batteries.
chemical stability against high voltage cathodes [5–11]. Alternating current (AC) impedance spectroscopy is widely em-
General prerequisites for a good solid-electrolyte are high ionic ployed to investigate ionic transport through multiphasic systems. This
conductivity, low electronic conductivity, and electrochemical stability is due to the different frequency response obtained for different trans-
towards the electrode materials. Currently, several structural families of port processes. However in a complete battery system, it is still very
inorganic electrolytes are being investigated for use in solid-state bat- challenging to separate the contribution from the components, either
teries including the perovskite-type Li3xLa2/3-xTiO3 [12,13], NASICON- due to frequency limitations or due to the overlapping or indis-
type AM2(PO4)3(A = alkali ion, and M = Ge, Zr, or Ti) [14–16], garnet- tinguishable response observed in the investigated frequency region.
like Li7La3Zr2O12 [17–19], and sulfide-based solid electrolytes. Gen- Solid-state NMR offers the opportunity to determine the bulk mobility
erally, sulfide-based electrolytes show a higher ionic conductivity than of 6,7Li nuclei with high sensitivity, in a non-destructive manner
that of oxide-based electrolytes. For instance, the thiophosphates without the influence of grain boundaries [40]. Additionally, solid-state
Li10GeP2S12 and Li7P3S11 show conductivities exceeding 10−3 S cm−1 NMR can be used to measure the spontaneous Li-ion exchange between
[20,21]. The glass-ceramic P2S5eLi2S has been reported to have a different Li-ion containing materials, such as a mixture of electrode and
conductivity exceeding 10 mS cm−1 [22], comparable to that of liquid electrolyte phases, to obtain unique selectivity for charge transport over
electrolytes. Sulfur-containing lithium argyrodites Li6PS5X (X = Cl, Br, phase boundaries, as recently reported for the Li6PS5Cl/BreLi2S solid
I) are considered to be a promising candidate for all-solid-state battery electrolyte-electrode combination [41,42].
application because they possess a high ionic conductivity of In this study, a lithium argyrodite electrolyte doped with both
∼10−3 S cm−1 for X = Cl and Br at room temperature, and have the chlorine and bromine with a composition of Li6PS5Cl0.5Br0.5 was syn-
added advantage of utilizing cheap precursors for their synthesis thesized via mechanical milling followed by annealing. The bulk con-
[23,24]. The lithium argyrodites are isostructural to the fast-ion con- ductivity of this material was higher than either of the single halogen
ductor Ag-argyrodite Ag8GeS6 [25,26]. The structure of the lithium- Li6PS5Cl/Br analogues. A combined X-ray and neutron diffraction
argyrodite consists of the tetrahedral close packing of the sulfur and analysis was employed to reveal the role of the composition and dis-
halogen anions, where the phosphorus atoms occupy a part of the tet- tribution of sulfur and halogen atoms over the 4a and 4c sites on the
rahedral interstices (4b) forming the isolated PS4 tetrahedra. The li- conductivity. The bulk conductivity of Li6PS5Cl0.5Br0.5 was character-
thium atoms are randomly distributed over the remaining sites, which ized by temperature-dependent impedance spectroscopy and spin-lat-
is in part responsible for their high ionic conductivity. The conductivity tice relaxation (SLR) NMR. Interfacial charge transport between the
also depends on the halogen atoms present in the structure. It is com- Li6PS5Cl0.5Br0.5 solid electrolyte and the nano-Li2S cathode was char-
parable for structures containing either chlorine or bromine, and higher acterized by 1D-exchange 7Li NMR. All-solid-state cells, assembled with
for structures doped with iodine [24,27]. The Li-ion transport network this solid electrolyte in combination with a nano-Li2S cathode and an
in the Li-argyrodite consists of interconnected low energy local cages Indium foil anode exhibited good electrochemical performance, with a
around the halogen and sulfur sites [27,28]. Kraft et al. [29] demon- high reversible capacity of around 820 mAh g−1 for up to 30 cycles.
strated that the softness of the lattice, a consequence of the increased
polarizability of the halogen ions, simultaneously leads to a lower ac- 2. Experimental section
tivation energy for diffusion, as earlier suggested [30], but also effec-
tively to a lower attempt frequency for diffusion. Furthermore, it was 2.1. Sample preparation
also shown that the halogen distribution over the sulfur sub-lattice has
a critical influence on the Li-ion conductivity [24]. Molecular dynamics Li6PS5Cl0.5Br0.5 was prepared using a ball-milling method followed
simulations suggested that the distribution of the halogen atoms over by an annealing process as described elsewhere [43]. In a typical pre-
the 4a and 4c sulfur sites determines the jump frequencies of all three paration process, appropriate amounts of reagent-grade Li2S (99.98%,
kinds of Li-ion jumps i.e. doublet, intra-cage, and inter-cage, max- Sigma-Aldrich), P2S5 (99%, Sigma-Aldrich), LiCl (99%, Sigma-Aldrich)
imizing the slowest jump frequency at 75% halogen occupancy of the 4c and LiBr (99%, Sigma-Aldrich) crystalline powder were weighed and
sites [27]. From this perspective, optimized compositions and occu- mixed in a WC coated (inner) stainless steel jar with 8 WC balls (8g/
pancies should be found in order to obtain the highest possible con- ball) in an argon filled glovebox (H2O, O2 < 0.3 ppm) to avoid reac-
ductivities for the Li-argyrodites. tion between the materials and air. The total weight of the mixture was
Despite considerable efforts devoted to exploring new materials about 2.0 g. The weight ratio of all the balls used to the total amount of
with high Li-ion conductivities, the overall performance of solid-state starting materials was set at 32 to 1. Ball milling was performed at
batteries made with these electrolytes in terms of cycling performance 550 rpm for 15 h with intermediate grindings by hand every hour to
and power density is still subpar, compared to Li-ion batteries made ensure the homogeneity of the mixture. After ball milling, the mixture
with conventional liquid electrolytes. The high interfacial resistance is was sealed in a quartz tube and annealed at 550 °C for 5 h to obtain the
recognized as the main factor limiting the performance of these bat- final Li6PS5Cl0.5Br0.5 solid electrolyte. For the 1D 7Li NMR exchange
teries. Hence the real issue impeding the performance of all-solid-state experiments, the mixture of the nano-Li2S electrode and the
lithium ion batteries is the interface, both in terms of the transport of Li- Li6PS5Cl0.5Br0.5electrolyte was prepared as follows: Ball milled Li2S
ions across the interface between the electrode and electrolyte as well (500 rpm/5 h) was mixed with an equal amount of milled
as the chemical and electrochemical stability of the interface itself Li6PS5Cl0.5Br0.5 (450 rpm/4 h) and the mixture was subsequently milled
[31–35]. at 400 rpm for 2 h to obtain the final mixture. To improve inter-particle
The mechanism leading to a high resistance interface involves sev- contact the obtained mixture was additionally cold pressed under a
eral aspects and will strongly depend on the electrolyte-electrode in- pressure of 6 tons cm−2 into a pellet and which was then crushed into
terface. Poor contact has been recognized as one source of high re- small pieces and filled in a custom-made Teflon tube in the glove box.
sistance, originating from the mismatched lattice of the two different
materials and/or the volume changes occurring during cycling [36]. A 2.2. Characterization
space charge region was also reported to form at the sulfide-based
electrolyte interface [37], which could deplete the Li-ions at the in- Crystal structure of the samples was determined by X-ray diffraction
terface resulting in a high resistance. An interphase layer was also on an X'Pert Pro X-ray diffractometer (PANalytical) using a copper

30
H. Wang et al. Journal of Power Sources 412 (2019) 29–36

Fig. 1. The (a) X-ray diffraction and (b) neutron


diffraction data and fits resulting from the si-
multaneous Rietveld refinement of
Li6PS5Cl0.5Br0.5 annealed at 550 °C for 5 h.
Symbols indicate the experimental data, red the
fit, green the background fit, and blue the dif-
ference between the data and the fit. (For in-
terpretation of the references to colour in this
figure legend, the reader is referred to the Web
version of this article.)

target (1.5406 Å at 45 kV and 40 mA) in the 2θ range of 10 to 140° in an Li6PS5Cl0.5Br0.5 (450 rpm/4 h) for 2 h with the weight ratio of 1:1. The
airtight sample holder. Neutron diffraction data at ambient temperature obtained mixture was then mixed with super P (TIMCAL) and carbon
was collected on the new neutron powder diffractometer PEARL of the nano-fiber (CNF) with a weight ratio of 8:1:1 at 400 rpm for 1 h to
TU Delft using the (533) reflection of the germanium monochromator obtain the final cathode mixture. The particle sizes of Li2S and
(λ = 1.665 Å) in the 2θ range of 10 to 160° using a 6 mm airtight Li6PS5Cl0.5Br0.5 in the cathode mixture were estimated to be 32 and
sample can. The data treatment consisted of a relative correction for 20 nm, respectively, via the refinement of the XRD patterns given in the
detection efficiency of the 1408 detector pixels and a subtraction of the supporting information in detail. For the assembly of the all-solid-state
background caused by the instrument and the sample can. batteries, 120 mg electrolyte powder was first pressed into a pellet with
The ionic conductivities of the samples were measured using AC a diameter of 10 mm under pressure of 4 tons cm−2. Next 5–8 mg of the
impedance spectroscopy. Samples were pressed uni-axially into pellets cathode mixture was evenly spread and pressed on the obtained elec-
with a diameter of 10 mm and about 1 mm in thickness. Stainless steel trolyte pellet under 6 tons cm−2. Finally, a piece of Indium foil was
disks were attached on both sides of the pellets as electrodes. AC im- attached on the other side of the bilayer pellet under 4 tons cm−2 to
pedance measurements were performed using an Autolab obtain the all-solid-state battery. Subsequently, the assembled battery
(PGSTAT302N) in the frequency range of 0.1 Hz–1 MHz, the data points was (dis)charged with the current density of 0.064 and 0.64 mA cm−2
being collected on a logarithmic scale with an oscillation voltage of between 0 and 3.5 V vs. In (0.62 and 4.12 V vs. Li/Li+) over multiple
500 mV in the temperature range from 30 to 150 °C. The impedance cycles to evaluate its electrochemical performance. The capacity was
data was analyzed with an equivalent circuit model using the program normalized to the weight of Li2S in the cathode mixture. Cyclic vol-
of Zsimpwin (PerkinElmer Instruments). tammetry (CV) measurements of the battery were performed within
Solid-state NMR measurements were performed on a Chemagnetics different voltage windows, applying a sweep rate of 0.5 mV s−1.
400 Infinity spectrometer (B0 = 9.4 T), operating at a 7Li resonance
frequency of 155.506 MHz. The π/2 pulse length was determined to be
3.2 μs with an RF field strength of 89 kHz. Chemical shifts were refer- 3. Results and discussion
enced with respect to a 0.1 M LiCl solution. Variable temperature
measurements were performed using a 5 mm static goniometer probe. 3.1. Phase and structural analysis
For the Li6PS5Cl0.5Br0.5 sample, the T1 relaxation times were de-
termined by using a saturation recovery experiment from −100 to Fig. S1 given in the supporting information shows the XRD patterns
180 °C. The temperature calibration has been done using solid lead of the ball-milled and annealed Li6PS5Cl0.5Br0.5 at given temperatures.
nitrate, which is commonly used as a ‘thermometer’ for variable-tem- After 550 rpm for 15 h, all reflections of the milled sample can be at-
perature NMR due to the 207Pb chemical shift being incredibly sensitive tributed to the argyrodite Li6PS5Cl0.5Br0.5 structure, except for a weak
to temperature. The temperature variations are controlled manually peak at 26.9° most likely originating from the Li2S precursor material.
and each target temperature point was kept for 5 minutes before test. Previous research has shown that a higher degree of crystallinity results
The corresponding T1ρ measurement was performed using the spin-lock in a higher ionic conductivity of the argyrodite electrolyte [43]. As
method at three different lock frequencies, ω1/(2π) ≈ 9, 19, and shown in Fig. S1, the Li2S reflection decreases gradually with an in-
39 kHz, respectively. Additional single pulse experiments were per- creasing annealing temperature and the pure phase of Li6PS5Cl0.5Br0.5 is
formed at different temperatures to determine the evolution of line obtained after annealing at 550 °C for 5 h, which is the electrolyte
width with temperature. In each case, a recycle delay of 3T1 was used. material described henceforth in this study.
For the mixture of Li2S and Li6PS5Cl0.5Br0.5, 1D exchange measure- Fig. 1 shows the result of the simultaneous Rietveld refinement of X-
ments were performed from −55 to 180 °C. The pulse sequence used ray and neutron powder diffraction data of Li6PS5Cl0.5Br0.5. The GSAS
has been described in detail elsewhere [44]. An echo time τ ranging [45] program was employed for the refinement. The parameters refined
from 200 to 400 μs was utilized so as to preserve the intensity of the within the cubic F-43m space group are summarized in Table 1. For the
narrow Li6PS5Cl0.5Br0.5 resonance and filter out the broad Li2S re- refinement, the sum of the occupancies on the 4a site (S1, Cl1, and Br1)
sonance, effectively functioning as a T2 filter. These 1D exchange ex- and 4c site (S2, Cl2, and Br2) were constrained to 1 and the isotropic
periments were performed for a range of mixing times, tmix, to follow temperature factors (Uiso) of the three atoms on the same site were
the spontaneous equilibrium exchange of the Li between the coupled. Note that the refinement of the occupancy of the three atomic
Li6PS5Cl0.5Br0.5 and Li2S phase. species demands two different contrasts, here provided by simultaneous
refinement of both neutron and X-ray diffraction data. The occupancies
2.3. Electrochemical performance evaluation at the 4b and 16e sites were not refined and restricted to their stoi-
chiometric values, assuming that the PS4 tetrahedron is not affected by
Laboratory-scale all-solid-state Li2S/Li6PS5Cl0.5Br0.5/In batteries the halogen substitution. The Li-ion occupancy at 48h site and the Uiso
were fabricated in the following manner: First, commercial Li2S was were refined without constraints. The results of the refinement given in
milled at 500 rpm for 5 h and then mixed with the milled Table 1 indicate a stoichiometry of Li6.4PS5Cl0.65Br0.35. For comparison,

31
H. Wang et al. Journal of Power Sources 412 (2019) 29–36

Table 1 most likely originates from a sum of the grain and grain boundary be-
Fractional atomic coordinates, site occupancies and isotropic atomic displace- cause the contribution of the grain boundary could not be deconvoluted
ment parameters (Uiso) derived from the simultaneous refinement of XRD and from the present spectra.
ND patterns of Li6PS5Cl0.5Br0.5. The lattice parameters a = b = c = 9.9063 Å Temperature dependence of the conductivity of Li6PS5Cl0.5Br0.5 is
and unit cell volume V = 972.15 Å3 were obtained from the refinement.
shown in Fig. 3b in terms of the modified Arrhenius law σT = σ0exp
Atom x y z Wyckoff Occupancy Uiso (-Ea/kBT), where for comparison the conductivities of the single halogen
doped Li6PS5Cl and Li6PS5Br are also shown. The Li6PS5Cl0.5Br0.5
Li 0.1885 0.1885 −0.0037 48h 0.5307 0.2189
composition shows the highest conductivity from among the three
P 0.5000 0.5000 0.5000 4b 1 0.0332
S3 0.6195 0.6195 0.6195 16e 1 0.0457 compositions, reaching 3.63 mS cm−1 at 32 °C while the values for
S1 0.0000 0.0000 0.0000 4a 0.4288 0.0346 Li6PS5Cl and Li6PS5Br are 0.79 and 0.36 mS cm−1 respectively. The Li-
Cl1 0.0000 0.0000 0.0000 4a 0.3233 0.0346 ion conductivity of the Li6PS5Cl0.5Br0.5 composition is also higher than
Br1 0.0000 0.0000 0.0000 4a 0.2478 0.0346
conductivities reported elsewhere [24,49] for Li6PS5Cl and Li6PS5Br
S2 0.2500 0.2500 0.2500 4c 0.5603 0.0313
Cl2 0.2500 0.2500 0.2500 4c 0.3329 0.0313
obtained using the same synthesis route, and comparable to the con-
Br2 0.2500 0.2500 0.2500 4c 0.1068 0.0313 ductivity reported by Kraft et al. for Li6PS5Cl0.5Br0.5 synthesized by
direct annealing [29]. The conductivity of the mixed halogen doped
argyrodite is however lower than that of the superionic conductors
the data of single halogen doped samples Li6PS5Cl and Li6PS5Br have Li10GeP2S12 [21] and Li9.54Si1.74P1.44S11.7Cl0.3 [50], which have con-
been included in the supporting information Tables ST1 and ST2. The ductivities of 12 and 25 mS cm−1 respectively at room temperature.
present material showed excess lithium content compared to the Fitting the temperature dependence of the conductivity by the modified
Li6PS5X stoichiometry, similar to previous findings [26]. Besides, the Arrhenius law σT = σ0exp(-Ea/kBT), where σ0 is the pre-factor, Ea is the
less content of Br than that of Cl could be due to the loss of LiBr during activation energy for ion conduction, and kB is Boltzmann constant,
the annealing process for its relative poor chemical stability at high results in Ea = 0.31 eV for Li6PS5Cl0.5Br0.5, 0.44 eV for Li6PS5Cl, and
temperature compared to that of LiCl. Fig. 2a shows the unit cell vo- 0.39 eV for Li6PS5Br (Fig. 3b).
lume and the occupancy of the halogen ions at the 4c site of Li6PS5X for The lithium diffusion in the argyrodite Li6PS5X (X = Cl, Br, and I)
the different halogen compositions. The nearly linear increase in unit occurs via three types of jumps. The first kind comprises of jumps be-
cell volume from Li6PS5Cl to Li6PS5Br is due to the larger ionic radius of tween the paired 48h sites which are separated by about 1.9 Å. Three
Br− compared to Cl−. However, an obvious deviation from the Ve- adjacent 48h pairs are interconnected via three interstitial sites to form
gards's law [46] could be due to the excess Li content in the present a pathway hexagon. Four hexagons are interconnected via a second
materials obtained by the Rietveld refinement. In addition, the ratio of interstitial site to form a cage-like structure surrounding the 4c site. The
halogen to sulfur occupancy of the 4c site decreases from Cl to Br. second type is the jump which occurs between the different 48h site
pairs over a distance of 2.25 Å, which in combination with the paired
48h sites results in intra-cage diffusion. The final kind of jump occurs
3.2. Conductivity and diffusion properties
between these cages (inter-cage jumps). All of the three jumps con-
tribute towards the macroscopic ionic conductivity of the crystal and
The ionic conductivity was determined using AC impedance spec-
the one with the smallest jump rate will limit macroscopic diffusion
troscopy. Typical complex impedance plots for Li6PS5Cl0.5Br0.5 at given
[27]. Two factors appear to play a role in explaining the impact of the
temperatures are shown in Fig. 3a. The impedance data were almost
different halogens on the conductivity in argyrodites, the polarizability
composed of a spike in the investigated temperature range. An in-
of the halogen ions [29] and the distribution of the halogen over the 4c
complete semi-circular arc can only be observed at 32 °C as shown in
sulfur sub lattice [27]. In general, the increased polarizability of the
the inset of Fig. 3a, which is due to the frequency limitation of the
lattice would facilitate ionic conductivity from the weaker attractive
instrument. The impedance data were fit with an equivalent circuit
force between the ligand and the mobile ions and/or the increased
model that contains a resistance R and a parallel combination of a re-
volume of the diffusion pathway [30]. For instance, the conductivity of
sistance R and a constant phase element Q connected in series with
LiX (X = F, Cl, Br, and I) increases while the activation energy Ea de-
impedance ZQ = [A(jω)n]−1, where A and n are constants for a given set
creases from F to I [51,52]. For the superionic conductor Li10GeP2L12
of experimental data, and ω is the angular frequency. The exponent, n,
(L = O, S, and Se), the conductivity increases from O to Se, with a
ranges from 0 to 1, which characterizes the constant phase elements of
concurrent decrease in Ea [53,54]. However, for the argyrodite Li6PS5X
the (RQ) sub circuit [47]. The Rg means the bulk resistance while the Ri
it has been shown that polarizability has a strong influence on both the
and Qi represent the resistance and capacitor of the interface between
activation energy Ea and the prefactor σ0. On increasing the polariz-
the sample and the contact electrode, respectively. The value of the Qi is
ability both the activation energy and prefactor σ0 decreases from Cl to
of 2.8 × 10−6 F, typical value for the electrode contact [48]. It should
Br [29], which has been attributed to the softness of the lattice with the
be noted that the resistance Rg obtained in the high-frequency region

Fig. 2. (a) Unit cell volume and the occupancy of halogen ions at the sulfur (4c) site, and (b) activation energy and prefactor σ0 for the Li-ion conductivity in Li6PS5X
(X = Cl, Cl0.5Br0.5, and Br).

32
H. Wang et al. Journal of Power Sources 412 (2019) 29–36

Fig. 3. (a) Complex impedance plots for


Li6PS5Cl0.5Br0.5 measured at given tempera-
tures. Solid lines represent the fit using an
equivalent circuit model illustrated in the inset.
(b) Temperature dependence of conductivities
for Li6PS5Cl0.5Br0.5 shown in the modified
Arrhenius law, σT = σ0exp(-Ea/kBT). The solid
lines were calculated using the Arrhenius law. Ea
denotes the activation energies determined for
Li-ion conduction. The data of Li6PS5Cl and
Li6PS5Br have also been included for comparison
(unpublished data).

increased polarizability. On the other hand, it has also been shown that
the halogen distribution over the sulfur sub-lattice has a critical influ-
ence on the Li-ion conduction. Independent of the halogen dopant, it
has been predicted that the inter-cage jump frequency decreases and
the activation energy increases with the decreasing halogen occupancy
on the 4c site [27]. The origin of this may be the stronger interaction
between the S2− and mobile Li-ion species compared to that with the
halogen ions [24]. The present results depicted in Fig. 2b show that
both the activation energy Ea and prefactor σ0 decrease from Cl to
Cl0.5Br0.5 and increase for Br, unlike the continuous decrease shown by
Kraft et al. [29], which may be due to the lower 4c occupancy of the
dopant prepared by us (Fig. 2a). In addition, the increased activation
energy and prefactor observed in Li6PS5I could be explained by the no
occupancy of halogen at 4c [24,27]. Nevertheless, a maximum in the
conductivity is found at the composition of Li6PS5Cl0.5Br0.5, consistent
with what was reported [29].
7
Li solid-state NMR experiments were applied to further study the
intrinsic Li-ion mobility and diffusion in the bulk of the argyrodite
Li6PS5Cl0.5Br0.5. Fig. S2 shows evolution of the static 7Li NMR spectrum
of the Li6PS5Cl0.5Br0.5 electrolyte as a function of temperature. At low
Fig. 4. Temperature dependence of the 7Li SLR rate of Li6PS5Cl0.5Br0.5 mea-
temperatures, a broadened Gaussian line was observed due to the low
sured in the laboratory frame of reference. The Larmor frequency (ω0/2π) is
mobility of the 7Li spins in the lattice and the consequent strong dipolar
155.506 MHz. The data of the Li6PS5Cl and Li6PS5Br were from elsewhere.
interaction between these spins. With increasing temperature, the mo-
[41,42].
bility of lithium increases and consequently the dipolar interactions are
gradually averaged out leading to narrower lines, i.e. motional nar-
rowing is observed [40,55]. This is shown in Fig. S3, where the width of processes as predicted by the MD simulations [27]. The low-T flank of
the 7Li NMR line has been plotted as a function of temperature. 1/T1 in Fig. 4 most likely reflects short-range diffusion, i.e. jumps be-
In order to further determine the jump rate and activation energy of tween paired 48h sites and intra-cage jumps. The high-T flank may in
Li-ion diffusion, SLR rates both in the laboratory (R1) and rotating (R1ρ) addition to longer-range diffusion of Li-ions also include inter-cage
frames of reference were determined for Li6PS5Cl0.5Br0.5 at various jumps in the present system [42,55]. Assuming Arrhenius behaviour of
temperatures as shown in Fig. 4 (R1) and in Fig. S4 (R1ρ) in the sup- the SLR rate versus temperature, R1 = R0exp(-Ea/kBT), the activation
porting information. It is well documented that the Li-ion jump rate 1/τ energy Ea for short-range jumps in Li6PS5Cl0.5Br0.5 was found to be
is directly related to the spectral density function [56]. When the SLR 0.06 eV, while for single chlorine and bromine doped samples the va-
rate R1(ρ) reaches the maximum versus the temperature, the mean jump lues of 0.09 and 0.1 eV, respectively were found. For the high-T flank
rate of the Li-ions is of the order of the Larmor frequency (ω0/2π) or the the activation energies were 0.14, 0.29, and 0.14 eV for those three
spin-lock frequency (ω1/2π) [57,58]. As shown in Fig. 4, the SLR rate of samples. It should be noted that the activation energies determined
Li6PS5Cl0.5Br0.5 reaches the maximum at 323 K, at which temperature from SLR NMR are significantly smaller than those obtained from im-
the condition of ω0τ ≈ 1 yields a corresponding jump frequency of 1/ pedance spectroscopy. This difference could in part be due to the
τ ≈ 9.8 × 108 s−1 based on the Larmor frequency of ω0/ aforementioned difficulties in decoupling the bulk and grain boundary
2π = 155.506 MHz utilized in this work. The corresponding maxima for contributions from the impedance data whereas SLR NMR is bulk se-
Li6PS5Cl and Li6PS5Br occur at 348 and 363 K, respectively. This im- lective.
plies that the same Li-ion jump rate in Li6PS5Cl0.5Br0.5 is achieved at a
significantly lower temperature than that for the Li6PS5Cl and Li6PS5Br 3.3. Electrochemical performance
materials, suggesting a higher Li-ion diffusivity, which is in good
agreement with the AC impedance results. In addition, the SLR rate The electrochemical performance of the electrolyte was evaluated in
peak is asymmetric, which is most likely due to the disorder of the a solid-state battery using Indium metal as anode material, and Li2S as
present system and/or the structural complexity and corresponding the cathode material. Indium was used as the anode due to the well
complexity of the diffusion processes [55]. As discussed, Li-ion diffu- documented electrochemical instability of the argyrodite versus Li-
sion in the argyrodite Li6PS5X involves several kinds of motional metal [43,59]. Cyclic voltammetry (CV) was employed to optimize the
potential window for the charge-discharge process. The CV curves of

33
H. Wang et al. Journal of Power Sources 412 (2019) 29–36

Fig. 5. (a) Charge and discharge voltage profiles for selected cycles under different (dis)charge rates.

the Li2S/Li6PS5Cl0.5Br0.5/In solid-state battery with different cut-off the given current density. The capacity of the both batteries increases
voltage ranges at a scanning rate of 0.5 mV s−1 are shown in Fig. S5 in within the first a few cycles, which could be due to the activation
the supporting information. Two redox peaks are observed at 2.0 and process of the cathode Li2S [66]. This cycling performance is better
0.75 V, corresponding to the oxidation of Li2S and the reduction of S than that of the solid-state battery using Li2S as cathode material and
[60,61]. The capacity obtained in the voltage window of 0–3.5 V is the single halogen doped electrolyte Li6PS5Cl and Li6PS5Br as solid
larger than that in the 0–3 V window, however on further increasing the electrolyte [41,43,66] and also better than that using Li2SeP2S5 glass
upper cut-off voltage to 4.0 V, the differences in intensity (area) be- ceramics as electrolyte [67,68], however, better cycle performance has
tween charge and discharge become greater suggesting that electrolyte also been reported [69].
decomposition reactions are occurring [62]. Fig. 5 shows the charge
and discharge curves when the constant current density of
0.064 mA cm−2 (0.02 C) and 0.64 mA cm−2 (0.2 C) were applied in the 3.4. Lithium diffusion between Li6PS5Cl0.5Br0.5-Li2S
voltage window of 0–3.5 V vs In (0.62–4.12 V vs. Li/Li+). The capacity
was normalized by the weight of Li2S. The capacity of the first charge Charge transport over the solid-solid interface is anticipated to be
was 790 mAh g−1 while the initial discharge capacity was 422 mAh one of the major challenges for solid-state batteries [41]. Recently solid
g−1, yielding an initial coulombic efficiency of 53.5% at 0.02 C. This state NMR exchange experiments between argyrodite Li6PS5Cl/Br and
low coulombic efficiency may be due to the large decrease in volume of Li2S [41] were shown to be able to quantify the equilibrium Li exchange
Li2S during cycling, which is likely to result in the loss of contact be- over the electrode-electrolyte interface. It was shown that interfacial
tween the active material and the electrolyte and/or conductive carbon conductivity was orders of magnitude smaller than the bulk con-
black. The corresponding initial capacities decreased to 261 and ductivity, presumably because of poor wetting. However, for pristine
244 mAh g−1 when the C-rate increased to 0.2 C (Fig. 5b). Interesting (uncycled) mixtures the activation barrier for diffusion was found to be
the initial charge profile shows one charge plateau while the sub- comparable to that obtained for bulk diffusion, indicating that for the
sequent cycles show more, which could be due to the oxidation of the limited contact points that are present between Li6PS5Cl/Br and Li2S
solid electrolyte, potentially leading to the decomposition of facile Li-ion diffusion is possible. However, upon cycling these barriers
Li6PS5Cl0.5Br0.5 into S, Li2Sn, P2Sx (x ≥ 5), and LiX (X = Cl, Br) at the are raised dramatically, most likely due to the electrolyte decomposi-
interface between the cathode and electrolyte [62–65]. In addition, the tion and the loss of contact because of the large volumetric changes
products of the decomposition could also show electrochemical activity occurring in Li2S upon cycling. In this work, 7Li solid state NMR ex-
resulting the additional charge plateau at subsequent cycles. Similar change experiments are applied to quantify the equilibrium Li-transport
behaviors were also observed in other argyrodite-based solid state over the interface between the Li6PS5Cl0.5Br0.5 electrolyte and Li2S
batteries using the Li2S as active materials [42,66]. After a few cycles, active material in the cathodic mixture. The static 1D 7Li NMR spec-
the coulombic efficiency increases to nearly 100% and the capacity trum of the mixture is shown in Fig. 7a. The narrower resonance cor-
stabilizes at 819 mAh g−1 for up to 30 cycles and then start to decrease responds to the more mobile Li-ions in the Li6PS5Cl0.5Br0.5 electrolyte
(0.02 C). Fig. 6 shows the cycling performance of the batteries under and the broader resonance is from Li in the Li2S, consistent with pre-
vious work [42]. The difference in line-width allows us to initially filter
out the NMR signal of Li2S, while preserving the resonance intensity
representing Li-ions in the electrolyte. The normalized intensity of the
resonance for the 7Li in Li2S versus the mixing time tmix at selected
temperatures is shown in Fig. 7b. With increasing mixing times, the
intensity increases due to the exchange of Li-ions from Li6PS5Cl0.5Br0.5
to Li2S. The evolution of the magnetization versus mixing time of the
Li2S resonance quantifies the equilibrium exchange between the solid
electrolyte and the active material in cathode. The analysis of the ex-
change between Li2S and Li6PS5Cl0.5Br0.5 was performed by fitting the
growing Li2S signal to a diffusion model, derived from Fick's law for
diffusion as explained in detail in the supporting information.
For simplicity, we assume that Li-ion diffusion occurs from the
center of one particle to the other, which have the particle sizes of 20
and 32 nm respectively. This leads to an average diffusion distance of
26 nm between Li2S and Li6PS5Cl0.5Br0.5. The fit result is shown in
Fig. 7b, from which the diffusion coefficient D for diffusion across the
Fig. 6. The cycling performance of the all-solid-state battery Li2S/ interface between the electrode and electrolyte was qualitatively de-
Li6PS5Cl0.5Br0.5/In at different current densities. termined. The value obtained was 1.5 × 10−10 cm2 s−1 at 303 K, which

34
H. Wang et al. Journal of Power Sources 412 (2019) 29–36

Fig. 7. (a) The static 1D 7Li NMR spectrum of


the Li6PS5Cl0.5Br0.5 electrolyte and Li2S active
material mixture. (b) Normalized intensity of
the static 7Li NMR spectrum of Li2S, T1 cor-
rected, as a function of mixing time at different
temperatures. Solid lines were fitting curves
using the diffusion model described in detail
elsewhere [44,71] and in the supporting in-
formation. Inset shows the evolution of the dif-
fusion coefficient D versus kT. The indicated
activation energy is derived from the Arrhenius
law and represents the energy barrier for Li-ion
transport between the electrolyte and cathode.

is several orders of magnitude lower than that for diffusion in the include diffusion on longer length scales such as diffusion over grain
electrolyte. This indicates that the wetting of the solid electrolyte is boundaries as discussed.
relatively poor, resulting in limited contact area that allows Li-ion In addition, the conductivity across the Li6PS5Cl0.5Br0.5-Li2S inter-
diffusion. Note that the diffusion coefficient D is directly correlated to face deduced from the 7Li 1D exchange NMR is also shown in Fig. 8.
the Li-ion diffusion distance d, and the assumption that the diffusion The conductivity was two orders of magnitude lower than that obtained
occurs between the centres of the two particles will lead an over- from impedance spectroscopy, suggesting that the interface is much
estimation of d due to the very low conductivity of Li2S [70], then re- more resistive for the diffusion of Li-ions than the bulk solid-electrolyte
sulting in an overestimation of the diffusion coefficient. The tempera- and also the main factor restricting the electrochemical performance of
ture dependence of the diffusivity has been presented in the inset of the all-solid-state battery. As seen from the electrochemical perfor-
Fig. 7b, from which the activation energy for Li-ion diffusion across the mance of the all-solid-state battery, the use of the Li6PS5Cl0.5Br0.5
interface was estimated. The activation energy for interfacial diffusion electrolyte resulted in a better electrochemical performance than bat-
was found to be about 0.19 eV, slightly larger than that for bulk dif- teries of the same configuration where the single halogen doped
fusion in the solid electrolyte obtained from SLR NMR (0.14 eV) and Li6PS5X (X = Cl or Br) electrolytes were used. In order to understand
much smaller than that determined by impedance spectroscopy. The this, we also compared the interfacial conductivities across all three
small activation energy indicates that Li-exchange over the electrolyte interfaces as shown in the Fig. S6 in the supporting in-
Li6PS5Cl0.5Br0.5-Li2S interface is facile, and that only Li-ions near these formation. The Li-ion conductivity across the Li6PS5Cl0.5Br0.5-Li2S in-
interfaces are involved in the exchange process. However, the small terface is much higher than that for the Li6PS5CleLi2S and
diffusivity implies that the current density through the interface was Li6PS5BreLi2S interfaces, which could be responsible for the better
low, so a larger effective contact area was essential for improving the battery performance of the former.
interfacial conductivity of the cathode mixture.
The conductivities derived from impedance, SLR NMR, and ex-
4. Conclusion
change NMR spectroscopy are compared in Fig. 8. The conductivity
obtained from the SLR measurements (8 × 10−2 S cm−1 at 323 K) is
Argyrodite Li6PS5Cl0.5Br0.5 was synthesized via a mechanical mil-
one order of magnitude higher than that obtained from the impedance
ling method followed by annealing. The lithium diffusion was studied
spectra (6.7 × 10−3 S cm−1). This is because NMR is sensitive to local
by impedance and NMR spectroscopy. A conductivity of 3.63 mS cm−1
ionic diffusion in the bulk lattice while impedance spectra may also
at 32 °C from impedance spectroscopy and a conductivity of
∼10−2 S cm−1 from NMR T1 relaxation measurements was obtained,
which are higher than values obtained for the single halogen doped
Li6PS5Cl and Li6PS5Br argyrodites. The higher conductivity and the
lower activation energy can be attributed to both the optimal polariz-
ability introduced by the halogens as well as the distribution of the
halogen dopants over the 4c sulfur sub lattice in Li6PS5Cl0.5Br0.5. The
all-solid-state battery performance with Li2S cathode showed a re-
versible discharge capacity of 820 mAh g−1 for up to 30 cycles. The 7Li
1D exchange NMR measurements demonstrated that the rate of Li-ion
transport across the Li6PS5Cl0.5Br0.5-Li2S interface is higher than that
across either of the Li6PS5X (X = Cl, Br)eLi2S interfaces, which in
combination with the higher bulk conductivity rationalizes the better
solid-state battery performance. Nevertheless, the interface remains the
stumbling block, where solid electrolyte decomposition and contact loss
due to volumetric changes in the cathode are expected to lead to an
increasing resistance and limited cycle life.

Acknowledgement

Fig. 8. The comparison of conductivities derived from impedance spectra, SLR This research was financially supported by the National Natural
NMR, and 1D exchange NMR. The symbols represent experimental data and the Science Foundation of China (No. 11404072), Guangxi Natural Science
lines represent fits using the Arrhenius law. The activation energy of the bulk Foundation (No. 2014GXNSFBA118014), the young teacher ability
conduction was obtained through fitting the high-T flank of temperature-de- enhancement project of Guangxi (2018KY0263), and Chinese
pendent SLR rate shown in Fig. 4. Scholarship Council (No. 201508455030). Support from the Dutch

35
H. Wang et al. Journal of Power Sources 412 (2019) 29–36

organization of scientific research (NWO) for the solid-state NMR fa- 10909.
cility for advanced materials science in Nijmegen is gratefully ac- [30] J.C. Bachman, S. Muy, A. Grimaud, H.H. Chang, N. Pour, S.F. Lux, O. Paschos,
F. Maglia, S. Lupart, P. Lamp, L. Giordano, Y. Shao-Horn, Chem. Rev. 116 (2016)
knowledged. The solid-state NMR facility for advanced material science 140.
at the Radboud University is greatly acknowledged for supporting this [31] J.P. Yue, M. Yan, Y.X. Yin, Y.G. Guo, Adv. Funct. Mater. 28 (2018) 16.
research. The assistance of Frans Ooms, Michel Steenvoorden, Bert [32] X.W. Yu, A. Manthiram, Energy Environ. Sci. 11 (2018) 527.
[33] B.B. Wu, S.Y. Wang, W.J. Evans, D.Z. Deng, J.H. Yang, J. Xiao, J. Mater. Chem. A 4
Zwart and Kees Goubitz is gratefully acknowledged. (2016) 15266.
[34] J. van den Broek, S. Afyon, J.L.M. Rupp, Adv. Energy Mater. 6 (2016) 11.
Appendix A. Supplementary data [35] W.D. Richards, L.J. Miara, Y. Wang, J.C. Kim, G. Ceder, Chem. Mater. 28 (2016)
266.
[36] C. Ma, K. Chen, C.D. Liang, C.W. Nan, R. Ishikawa, K. More, M.F. Chi, Energy
Supplementary data to this article can be found online at https:// Environ. Sci. 7 (2014) 1638.
doi.org/10.1016/j.jpowsour.2018.11.029. [37] K. Takada, N. Ohta, L.Q. Zhang, X.X. Xu, B.T. Hang, T. Ohnishi, M. Osada, T. Sasaki,
Solid State Ionics 225 (2012) 594.
[38] A. Sakuda, A. Hayashi, M. Tatsumisago, Chem. Mater. 22 (2010) 949.
References [39] S. Wenzel, S.J. Sedlmaier, C. Dietrich, W.G. Zeier, J. Janek, Solid State Ionics 318
(2018) 102.
[1] P.K. Nayak, E.M. Erickson, F. Schipper, T.R. Penki, N. Munichandraiah, [40] V. Epp, M. Wilkening, World Scientific: Hong Kong, 2015.
P. Adelhelm, H. Sclar, F. Amalraj, B. Markovsky, D. Aurbach, Adv. Energy Mater. 8 [41] C. Yu, S. Ganapathy, E.R.H.V. Eck, H. Wang, S. Basak, Z. Li, M. Wagemaker, Nat.
(2018) 16. Commun. 8 (2017) 1086.
[2] W.D. Li, B.H. Song, A. Manthiram, Chem. Soc. Rev. 46 (2017) 3006. [42] C. Yu, S. Ganapathy, N.J.J. de Klerk, I. Roslon, E.R.H. van Eck, A.P.M. Kentgens,
[3] D. Andre, H. Hain, P. Lamp, F. Maglia, B. Stiaszny, J. Mater. Chem. A 5 (2017) M. Wagemaker, J. Am. Chem. Soc. 138 (2016) 11192.
17174. [43] C. Yu, L. van Eijck, S. Ganapathy, M. Wagemaker, Electrochim. Acta 215 (2016) 93.
[4] J.B. Goodenough, J. Solid State Electrochem. 16 (2012) 2019. [44] M. Wagemaker, A.P.M. Kentgens, F.M. Mulder, Nature 418 (2002) 397.
[5] R.C. Xu, S.Z. Zhang, X.L. Wang, Y. Xia, X.H. Xia, J.B. Wu, C.D. Gu, J.P. Tu, Chem. [45] A.C. Larson, R.B. Von Dreele, Los Alamos National Laboratory Report, LAUR, 2004,
Eur J. 24 (2018) 6007. p. 86.
[6] P.K. Nayak, L.T. Yang, W. Brehm, P. Adelhelm, Angew. Chem. Int. Ed. 57 (2018) [46] A.R. Denton, N.W. Ashcroft, Phys. Rev. A 43 (1991) 3161.
102. [47] S.M. Haile, D.L. West, J. Campbell, J. Mater. Res. 13 (1998) 1576.
[7] Y.J. Liu, P. He, H.S. Zhou, Adv. Energy Mater. 8 (2018) 22. [48] J.T.S. Irvine, D.C. Sinclair, A.R. West, Adv. Mater. 2 (1990) 132.
[8] Z.H. Gao, H.B. Sun, L. Fu, F.L. Ye, Y. Zhang, W. Luo, Y.H. Huang, Adv. Mater. 30 [49] S. Boulineau, M. Courty, J.M. Tarascon, V. Viallet, Solid State Ionics 221 (2012) 1.
(2018) 1705702. [50] Y. Kato, S. Hori, T. Saito, K. Suzuki, M. Hirayama, A. Mitsui, M. Yonemura, H. Iba,
[9] W. Chen, T.Y. Lei, C.Y. Wu, M. Deng, C.H. Gong, K. Hu, Y.C. Ma, L.P. Dai, W.Q. Lv, R. Kanno, Nat. Energy 1 (2016) 16030.
W.D. He, X.J. Liu, J. Xiong, C.L. Yan, Adv. Energy Mater. 8 (2018) 1702348. [51] R. Mercier, J.P. Malugani, B. Fahys, G. Robert, Solid State Ionics 5 (1981) 663.
[10] J.G. Kim, B. Son, S. Mukherjee, N. Schuppert, A. Bates, O. Kwon, M.J. Choi, [52] Y. Haven, Recl. Trav. Chim. Pay B 69 (1950) 1471.
H.Y. Chung, S. Park, J. Power Sources 282 (2015) 299. [53] Y. Wang, W.D. Richards, S.P. Ong, L.J. Miara, J.C. Kim, Y. Mo, G. Ceder, Nat. Mater.
[11] T. Minami, M. Tatsumisago, M. Wakihara, C. Iwakura, S. Kohjiya, I. Tanaka, 14 (2015) 1026.
Springer Science & Business Media: Tokyo, 2006. [54] S.P. Ong, Y.F. Mo, W.D. Richards, L. Miara, H.S. Lee, G. Ceder, Energy Environ. Sci.
[12] W.J. Kwon, H. Kim, K.N. Jung, W. Cho, S.H. Kim, J.W. Lee, M.S. Park, J. Mater. 6 (2013) 148.
Chem. A 5 (2017) 6257. [55] V. Epp, O. Gun, H.J. Deiseroth, M. Wilkening, J. Phys. Chem. Lett. 4 (2013) 2118.
[13] E.E. Jay, M.J.D. Rushton, A. Chroneos, R.W. Grimes, J.A. Kilner, Phys. Chem. Chem. [56] P. Heitjans, J. Kärger, Springer-Verlag: Berlin Heidelberg, 2005.
Phys. 17 (2015) 178. [57] M. Wilkening, P. Heitjans, ChemPhysChem 13 (2012) 53.
[14] D. Rettenwander, G.J. Redhammer, M. Guin, A. Benisek, H. Krueger, O. Guillon, [58] A. Kuhn, S. Narayanan, L. Spencer, G. Goward, V. Thangadurai, M. Wilkening, Phys.
M. Wilkening, F. Tietz, J. Fleig, Chem. Mater. 30 (2018) 1776. Rev. B 83 (2011) 094302.
[15] D. Safanama, N. Sharma, R.P. Rao, H.E.A. Brand, S. Adams, J. Mater. Chem. A 4 [59] S. Wenzel, S.J. Sedlmaier, C. Dietrich, W.G. Zeier, J. Janek, Solid State Ionics 318
(2016) 7718. (2018) 102.
[16] G.F. Ortiz, M.C. Lopez, P. Lavela, C. Vidal-Abarca, J.L. Tirado, Solid State Ionics 262 [60] M. Nagao, A. Hayashi, M. Tatsumisago, Electrochim. Acta 56 (2011) 6055.
(2014) 573. [61] T. Kobayashi, Y. Imade, D. Shishihara, K. Homma, M. Nagao, R. Watanabe,
[17] K. Fu, Y.H. Gong, S.M. Xu, Y.Z. Zhu, Y.J. Li, J.Q. Dai, C.W. Wang, B.Y. Liu, G. Pastel, T. Yokoi, A. Yamada, R. Kanno, T. Tatsumi, J. Power Sources 182 (2008) 621.
H. Xie, Y.G. Yao, Y.F. Mo, E. Wachsman, L.B. Hu, Chem. Mater. 29 (2017) 8037. [62] J. Auvergniot, A. Cassel, J.B. Ledeuil, V. Viallet, V. Seznec, R. Dedryvere, Chem.
[18] C.K. Chan, T. Yang, J.M. Weller, Electrochim. Acta 253 (2017) 268. Mater. 29 (2017) 3883.
[19] V. Thangadurai, S. Narayanan, D. Pinzaru, Chem. Soc. Rev. 43 (2014) 4714. [63] R. Koerver, F. Walther, I. Aygun, J. Sann, C. Dietrich, W.G. Zeier, J. Janek, J. Mater.
[20] I.H. Chu, H. Nguyen, S. Hy, Y.C. Lin, Z.B. Wang, Z.H. Xu, Z. Deng, Y.S. Meng, Chem. A 5 (2017) 22750.
S.P. Ong, ACS Appl. Mater. Interfaces 8 (2016) 7843. [64] R. Koerver, I. Aygun, T. Leichtweiss, C. Dietrich, W.B. Zhang, J.O. Binder,
[21] N. Kamaya, K. Homma, Y. Yamakawa, M. Hirayama, R. Kanno, M. Yonemura, P. Hartmann, W.G. Zeier, J. Janek, Chem. Mater. 29 (2017) 5574.
T. Kamiyama, Y. Kato, S. Hama, K. Kawamoto, A. Mitsui, Nat. Mater. 10 (2011) [65] J. Auvergniot, A. Cassel, D. Foix, V. Viallet, V. Seznec, R. Dedryvere, Solid State
682. Ionics 300 (2017) 78.
[22] K. Xu, Chem. Rev. 114 (2014) 11503. [66] C. Yu, S. Ganapathy, E.R.H. van Eck, L. van Eijck, S. Basak, Y.Y. Liu, L. Zhang,
[23] Z.X. Zhang, L. Zhang, Y.Y. Liu, C. Yu, X.L. Yan, B. Xu, L.M. Wang, J. Alloy. Comp. H.W. Zandbergen, M. Wagemaker, J. Mater. Chem. A 5 (2017) 21178.
747 (2018) 227. [67] Y.B. Zhang, R.J. Chen, T. Liu, Y. Shen, Y.H. Lin, C.W. Nan, ACS Appl. Mater.
[24] P.R. Rayavarapu, N. Sharma, V.K. Peterson, S. Adams, J. Solid State Electrochem. Interfaces 9 (2017) 28542.
16 (2012) 1807. [68] M. Eom, S. Son, C. Park, S. Noh, W.T. Nichols, D. Shin, Electrochim. Acta 230
[25] S.T. Kong, H.J. Deiseroth, C. Reiner, O. Gun, E. Neumann, C. Ritter, D. Zahn, Chem. (2017) 279.
Eur J. 16 (2010) 2198. [69] J. Yue, F.D. Han, X.L. Fan, X.Y. Zhu, Z.H. Ma, J. Yang, C.S. Wang, ACS Nano 11
[26] H.-J. Deiseroth, S.-T. Kong, H. Eckert, J. Vannahme, C. Reiner, T. Zaiss, (2017) 4885.
M. Schlosser, Angew. Chem. Int. Ed. 47 (2008) 755. [70] Y. Yang, G.Y. Zheng, S. Misra, J. Nelson, M.F. Toney, Y. Gui, J. Am. Chem. Soc. 134
[27] N.J.J. de Klerk, T. Roslon, M. Wagemaker, Chem. Mater. 28 (2016) 7955. (2012) 15387.
[28] R.P. Rao, S. Adams, Phys. Status Solidi a-App. Mater. Sci. 208 (2011) 1804. [71] S. Ganapathy, E.R.H. van Eck, A.P.M. Kentgens, F.M. Mulder, M. Wagemaker,
[29] M.A. Kraft, S.P. Culver, M. Calderon, F. Bocher, T. Krauskopf, A. Senyshyn, Chem. Eur J. 17 (2011) 14811.
C. Dietrich, A. Zevalkink, J. Janek, W.G. Zeier, J. Am. Chem. Soc. 139 (2017)

36

You might also like