A facile strategy to achieve high conduction and excellent chemical stability of lithium solid electrolytes

Abstract

Lithium solid electrolytes have shown their potential for high-energy density batteries. The use of solid electrolytes will also be able to overcome safety issues associated with conventional carbonate-based electrolytes. However, achieving the combination of high ionic conductivity and excellent electrochemical stability in lithium solid electrolytes is still a major challenge. Herein we report a facile strategy to achieve high conduction and excellent electrochemical stability by the substitution of Cl for O based on the concept of bottleneck size and binding energy. The ionic conductivities of Li_10.42Si_1.5P_1.5Cl_0.08O_11.92 and Li_10.42Ge_1.5P_1.5Cl_0.08O_11.92 are 1.03 × 10⁻⁵ S cm⁻¹ and 3.7 × 10⁻⁵ S cm⁻¹ at 27 °C, respectively, which are 13 orders of magnitude higher than that of the pure Li₃PO₄, and 1 order of magnitude higher than that of the pristine Li_10.5Si_1.5P_1.5O₁₂. The electrochemical stability with metallic lithium is up to 9 V vs. Li⁺/Li, which is one of the widest electrochemical windows of solid electrolytes. This research also addresses the crystal structure, lithium ion migration mechanism, and battery performance.

---

Introduction

Although lithium-ion batteries are now widely used in many electronic products, safety, energy density, and service life are still issues, which continue to plague lithium-ion batteries for the potential of mass market of electric vehicles and large-scale energy storage.¹ There are two ways to increase the energy density, one of which is to increase the specific capacity and the other one is to increase the voltage potential of the cathode which is limited by current liquid electrolytes. Applying solid electrolytes to lithium batteries will eliminate the conventional voltage limitation (4.5 V vs. Li⁺/Li) of currently used carbonate-based electrolytes and open up the possibility of high-voltage operation with 5 V-class cathode materials.² On the other hand, metallic lithium batteries exhibit the highest theoretical energy densities. Among many possible systems, Li–S and Li–O₂ batteries are very attractive as candidates for next-generation high-energy density batteries. The use of solid electrolytes could potentially suppress the formation of lithium dendrites, that are caused by uneven current distributions at the metal–electrolyte interface during cycling.³

A number of works reported new solid electrolytes, such as sulfide-based superionic conductors.⁴ Recently, a breakthrough has been achieved in inorganic sulfide-based electrolyte with extremely high room temperature conductivity of 10⁻² S cm⁻¹.^5,6 Generally, sulfides have higher ionic conductivity than oxides but lower air and moisture stability than oxides. The hypersensitivity of the sulfides to air and moisture requires sophistically and tedious treatment procedures under dry inert gas atmosphere, which increase their processing cost.⁷ Moreover, the sulfides suffer inferior electrochemical stability due to reacting continuously with Li anode.⁸ In an effort to overcome inferior electrochemical stability, the lithium phosphorous oxynitride (LPON) electrolyte developed at Oak Ridge National Laboratories (ORNL) exemplifies the extremely good stability with Li metal anode. The ORNL group reports thin film batteries employing the LPON electrolyte, LiCoO₂ cathode, and Li metal anode are capable of over 20 000 charge–discharge cycles with 0.001% capacity loss per cycle. Unfortunately LPON has relatively poor ionic conductivity of ca. 2 × 10⁻⁶ S cm⁻¹.⁹

Although the advantages of non-flammable solid electrolytes are widely acknowledged, their low ionic conductivities or low chemical and electrochemical stabilities prevent them from practical applications. It is known that the ionic conductivity is given by:

σ = Zenμ (1)

where, Ze is the ionic charge, n is the carrier concentration and μ is the mobility of the ion. The mobility of the ions mainly depends on the activation energy for ion conduction.¹⁰ To migrate from one site to next available one, the lithium ions have to pass through a bottleneck consisting of four oxygen ions. The bottleneck size is the predominant factor of the activation energy.¹¹ As predicted by Inaguma et al.,¹² the size of the bottleneck mostly influences the activation energy of the ionic conduction by varying the repulsion potential between oxygen and lithium at the bottleneck site. Ion mobility is physically governed by the bottleneck size and chemically by the binding energy between mobile ions and the network anions. The suitable bottleneck size should be favor of ions conduction, while the covalent bonding between the mobile ion and network anion should be weak as possible.¹³ The present paper reports a facile method to improve ionic conductivity of oxides while maintaining their chemical stability with air and moisture and electrochemical stability with Li metal anode. To demonstrate the concept discussed above, the solid solutions of Li₃PO₄ and Li₄Si/GeO₄ are selected as prototype material by considering their good stability, while the halide elements (F, Cl, Br) are selected to partly substitute O ions to adjust the bottleneck size and binding energy, in order to improve the ionic conductivity, considering the higher electronegativity of F than O, and larger ionic radii of Cl and Br than O. Two issues will be addressed in this work: (1) is substitution of halide elements for oxygen valid strategy to enhance the ionic conductivity of oxides? (2) Dose the halide-oxides have the expected electrochemical stability with Li metal? This research also addresses the crystal structure, lithium ions migration mechanism, and battery performance of the halide-oxides materials.

Experimental

Synthesis of halide-substituted solid solutions of Li₃PO₄ and Li₄Si/GeO₄ [Li_10.5-xSi/Ge_1.5P_1.5Cl_xO_12−x (0.05 ≤ x≤ 1.0)]

Li_10.5-xSi/Ge_1.5P_1.5Cl_xO_12−x (0.05 ≤ x≤ 1.0) was prepared by melt-casting method. All the raw materials were received from Sigma, and the stoichiometric amounts of Li₂CO₃ (99%, with 5 wt% excess to compensate for the loss of lithium during melting), SiO₂ (99%), GeO₂ (99.999%), NH₄H₂PO₄ (98.5%), LiCl (99%), LiF (99%), and LiBr (99%) were weighed and ball-milled using zirconia balls for about 2 h in industrious ethanol in air. The mixed materials were pre-calcined at 700 °C for 2 h, subsequently, the calcined powder was melted at 1200–1250 °C, the sheet with thickness of about 1 mm was obtained after casting the melt using a rolling machine. For some specific samples, the F and Br were added to study their effects on ionic conductivity. Although the materials are not air-sensitive, to avoid possible interference from absorbed moisture, all the casted sheets were immediately transferred into an argon-filled glove box.

Electrochemical characterization

For the ionic conductivity measurements, the sheets were cut to small pellet and the conducting silver paste was painted on both sides of the small sheet and sealed in a Swagelok cell. Electrochemical impedance spectroscopy (EIS) measurements were carried out using a Solartron 1260 + 1287 System, applying 100 mV in the frequency range 1 MHz–10 Hz, without any pressure applied. Activation energy was measured at the temperature range from 27 °C to 110 °C.

Cyclic voltammogram (CV) of the samples made of Ag\Li_10.42Si/Ge_1.5P_1.5Cl_0.08O_11.92 sheet\Li was measured using a linear sweep voltammetry between a voltage range from −0.5 V to 9 V vs. Li⁺/Li at a 1 mV s⁻¹ scan rate.

Full cells using Li_10.42Si/Ge_1.5P_1.5Cl_0.08O_11.92 as the electrolytes, 0.3Li₂MnO₃·0.7LiMn_1.5Ni_0.5O₄ as the cathode and lithium metal as the anode were assembled in 2025 coin cells. The cathode powder was mixed with super P conductive carbon (TIMCAL Ltd.) and polyvinylidene fluoride (PVDF, Sigma) at a weight ratio of 8 : 1 :  1 in an N-methylpyrrolildone (NMP, Sigma) solvent to form uniform slurries and then coated on Al foils. To reduce interfacial impedance between the cathodes and electrolytes, a very little drop of 1 M LiPF₆ in EC–DEC (1 : 1 by weight) as a buffer between solid electrolyte and cathode layer. Charge–discharge test was performed at a potential range between 2.0 and 5.0 V vs. Li⁺/Li at 0.1 C rate.

Structural characterization

X-ray diffraction (Shimadzu XRD-6000 and 7000 Cu-Kα) was employed to detect crystallographic structure of the samples in air with no seal. Lattice constants and structures were analyzed by Rietveld refinement of the powder XRD data using GSAS software. The structure was drawn by VESTA software.

Results and discussion

Fig. 1 represents typical Nyquist plots of Li_10.42Si_1.5P_1.5Cl_0.08O_11.92 (square symbol) and Li_10.42Ge_1.5P_1.5Cl_0.08O_11.92 (circle symbol) measured at 27 °C. It shows a depressed rather than ideal semicircle, inclined rather than vertical capacitive tail. The appearance of the tail at low frequencies in the case of ionically blocking electrodes is an indication that the pellet is ionically conducting in nature.¹⁴ Close examination of the Nyquist plot, there is only one semicircle, suggesting that the grain and grain boundary resistance be unable separated. The total conductivity of the pellet at 27 °C is calculated by the equation:where R_t, d, S are the total resistance, the thickness of sample, the electrode area, respectively. For the sample (d = 1.0 mm, S = 28.3 mm²), the total conductivity of Li_10.42Si_1.5P_1.5Cl_0.08O_11.92 at 27 °C is calculated to be 1.03 × 10⁻⁵ S cm⁻¹, which is 13 orders of magnitude higher than that of the pure Li₃PO₄ (10⁻¹⁸ S cm⁻¹),¹⁵ and 1 order of magnitude higher than that of the pristine Li_10.5Si_1.5P_1.5O₁₂ (10⁻⁶ S cm⁻¹),^4,16 and the room-temperature conductivity of its counterpart Li_10.42Ge_1.5P_1.5Cl_0.08O_11.92 is much higher with 3.7 × 10⁻⁵ S cm⁻¹.

Fig. 1 Typical Nyquist plots of Li_10.42Si_1.5P_1.5Cl_0.08O_11.92 (square symbol) and Li_10.42Ge_1.5P_1.5Cl_0.08O_11.92 (circle symbol) measured at 27 °C.

Fig. 2 shows the Arrhenius plots of Li_10.42Si/Ge_1.5P_1.5Cl_0.08O_11.92 in the range between 27 °C and 110 °C. The activation energies for two types of solid electrolytes were calculated using the Arrhenius equation:

σ = Aexp(−E_a/kT) (3)

where A represents the pre-exponential factor, k is the Boltzmann constant, and T is the absolute temperature. The activation energy E_Si (Li_10.42Si_1.5P_1.5Cl_0.08O_11.92) and E_Ge (Li_10.42Ge_1.5P_1.5Cl_0.08O_11.92) for the conduction was calculated to be 0.44 eV and 0.39 eV, respectively, which is much lower than that of the pristine Li_10.5Si_1.5P_1.5O₁₂ (≈0.53 eV).¹⁶

Fig. 2 Arrhenius plots of Li_10.42Si/Ge_1.5P_1.5Cl_0.08O_11.92 in the range between 27 °C and 110 °C.

Fig. 3 shows the effect of halide substitution on the room-temperature conductivity of Li_10.5−xSi_1.5P_1.5Cl/F_xO_12−x (0.05 ≤ x≤ 1.0). The square symbol displays the conductivity variation with substitution of Cl for O. A maximum conductivity σ_Li ≈ 1.03 × 10⁻⁵ S cm⁻¹ was found for x = 0.08. The solid circle symbol displays the conductivity variation with substitution of F for O. The maximum conductivity σ_Li ≈ 7.6 × 10⁻⁶ S cm⁻¹ was lower than their Cl counterpart, while the conductivities of Li_10.4Si_1.5P_1.5Cl_0.05F_0.05O_11.9 and Li₁₀Si_1.5P_1.5Cl_0.25F_0.25O_11.5 (triangular symbol) are comparable with their Cl counterpart Li_10.4Si_1.5P_1.5Cl_0.1O_11.9 and Li₁₀Si_1.5P_1.5Cl_0.5O_11.5. Whereas, the conductivities of Li_10.4Si_1.5P_1.5Cl_0.05Br_0.05O_11.9 is much lower (≈7.4 × 10⁻⁶ S cm⁻¹, not shown in this paper).

Fig. 3 Effect of halide substitution on the room-temperature conductivity of Li_10.5−xSi_1.5P_1.5Cl/F_xO_12−x (0.05 ≤ x≤ 1.0).

It is worth noting that the activation energy versus Cl content displays an inverted peak with the conductivity (Fig. 4). The lowest activation energy observed in the composition of Li_10.42Si_1.5P_1.5Cl_0.08O_11.92 that had the highest ionic conductivity.

Fig. 4 The plots of room-temperature conductivity and activation energy versus Cl content.

A high conductivity is necessary but not a sufficient property to make an electrolyte to be useful in practical terms. A wide electrochemical window is also an essential parameter to ensure good performance in rechargeable lithium batteries. The electrochemical stability window has been evaluated by running a cyclic voltammogram of cells, in which Li and Ag serve as reference/counter and working electrodes, respectively. Fig. 5 displays the current–potential curve of Li_10.42Si/Ge_1.5P_1.5Cl_0.08O_11.92. First, the potential was swept from open voltage (OCV) to −0.5 V (cathodic sweep) and then from −0.5 V to 9 V (anodic sweep). The cathodic current corresponding to lithium deposition, and the anodic current corresponding to lithium dissolution. No obvious current was observed except the ones due to the lithium deposition and dissolution, indicating that the Li_10.42Si/Ge_1.5P_1.5Cl_0.08O_11.92 electrolytes have a wide electrochemical window up to 9 V vs. Li⁺/Li. Li_10.42Si/Ge_1.5P_1.5Cl_0.08O_11.92 would be an alternative for stable protective solid electrolyte interphase and buffer film taking into account their excellent electrochemical stability with lithium metal and promising ionic conductivities.

Fig. 5 The current–potential curves of Li_10.42Si/Ge_1.5P_1.5Cl_0.08O_11.92.

All XRD patterns of Li_10.5−xSi_1.5P_1.5Cl_xO_12−x (0.05 ≤ x≤ 0.5) (Fig. S1†) show that the matrix of Cl substitution is in the solid solution with structure of γ-Li₃PO₄ (Powder Diffraction File 15-0760, space group Pnma). The orthorhombic structure of γ-Li₃PO₄ widely used as a host for preparing lithium solid electrolytes. In particular, the LPON, LISICON and thio-LISICON electrolyte materials based on γ-Li₃PO₄ structure have been well developed.^17–19 Two different structures have been recognized in the system of solid solutions between Li₃PO₄ and Li₄SiO₄: the Li₄SiO₄-type phase and γ-Li₃PO₄-type phase with respective formulae of Li_4−xSi_1−xP_xO₄ (0 ≤ x≤ 0.12) and Li_3+ySi_yP_1−yO₄ (0 ≤ y≤ 0.42).²⁰ In the Li₄SiO₄-type phase, vacancies can be introduced into the normal Li⁺ sites in the Li₄SiO₄ structure, while in the γ-Li₃PO₄-type phase, interstitial Li⁺ can be introduced into the γ-Li₃PO₄ structure. In the system, x = y = 0.5, it shows γ-Li₃PO₄-type phase judged from the XRD patterns (Fig. S1†), indicating an interstitial ion migration mechamism.

The crystal structure of Cl-substituted Li_10.5−xSi_1.5P_1.5Cl_xO_12−x (0.05 ≤ x≤ 0.5) was analyzed by the Rietveld method.²¹ The structure model of orthorhombic γ-Li₃PO₄ was adapted as the initial structure model with space group of Pnma (no. 62).²² Typical observed, calculated, and difference patterns for Rietveld refinement of Li₁₀Si_1.5P_1.5Cl_0.5O_11.5 were shown in Fig. S2.† Table S1† lists the crystallographic data and details of the structure refinement. The resultant R-values reached R_wp = 0.0787, R_p = 0.0638. The Li1, Li2, Li3, P, Si, O1, O2, O3, Cl sites were the special positions of 8d, 4c, 8d, 4c, 4c, 8d, 4c, 4c, and 8d sites (the multiplicity and Wyckoff letter), respectively. The lattice constants were refined to be a = 5.012(7) Å, b = 6.116(6) Å, c = 10.599(8) Å.

Fig. 6 shows the lattice constants and volume of unit cell of pure Li₃PO₄ and Cl-substituted Li_10.5−xSi_1.5P_1.5Cl_xO_12−x solid solutions. The lattice constants and volume of unit cell of Cl-substituted Li_10.5−xSi_1.5P_1.5Cl_xO_12−x solid solutions are much larger than the pure Li₃PO₄,²² while the lattice constants and volume of unit cell vary very slightly when Cl concentration increases further, since the ionic radius of Cl⁻ (1.81 Å) is much larger than that of O²⁻ (1.4 Å).¹¹ There are no simple correlation between lattice constants and ionic radii.²³ Addressing the concept of bottleneck size and binding energy between mobile ions and the network anions, substitution of Cl for O increases the lattice constants and volume of unit cell, would enlarge the bottleneck size, then decrease the energy barriers (activation energy) and enhance the ionic conductivity. On the other hand, the ionicity of Li–Cl is smaller than that of Li–O due to the lower electronegativity of Cl⁻ (3.16) than O²⁻ (3.44). As the increase of Cl concentration, the Li⁺ cations would be rigidly bonded to the Cl⁻ anions, which behaves as an obstacle for the ions migration, and increases the activation energy and decreases the ionic conductivity. Hence there should be an optimized Cl concentration with considerably large bottleneck size for Li⁺ ions migration and relatively small binding energy between Li–O/Cl, which may lead to the minimum activation energy and maximum ionic conductivity (Cl = 0.08). This has been demonstrated in Fig. 4.

Fig. 6 The lattice constants and volume of unit cell of pure Li₃PO₄ and Cl-substituted Li_10.5−xSi_1.5P_1.5Cl_xO_12−x solid solutions.

Fig. 7(a) illustrates the unit cell of Li₁₀Si_1.5P_1.5Cl_0.5O_11.5. In an ideal crystal of γ-Li₃PO₄, the Li ions are located at two different crystallographically sites indicated with Li1 and Li2, using the Wyckoff labels, 8d and 4c, respectively. The d site accounts for 8 equivalent atomic sites, and the c site accounts for 4 equivalent atomic sites per unit cell. The O ions fully occupy three different sites designated as O1(8d), O2(4c) and O3(4c), and the P ions fully occupy the 4c sites. Since in an ideal crystal, all the lithium cations fully occupy their positions and are rigidly bonded to the framework, which creates no prerequisites for ion migration. In the real crystal of Li₁₀Si_1.5P_1.5Cl_0.5O_11.5, Si and P ions share the same 4c sites due to partially substitution of Si for P (Fig. 7(b)). Noted that one site can just accommodate one atom, the P and Si ions cannot appear at 4c site at the same time, designating as white and pink spheres for P ions, and white and yellow spheres for Si ions in Fig. 7(b), indicating that they are occupied by P or Si exclusively. Cl and O ions share the same 8d sites due to partially substitution of Cl for O (Fig. 7(b)), designating as white and green spheres for Cl ions, and white and red spheres for O1 ions in Fig. 7(b), indicating that they are occupied by Cl or O exclusively. It is difficult to determine the exact positions of Cl ions due to their minor content and volatilization during preparation procedure. In the present work, we refine the crystal structure according to O1 sites occupied by Cl ions, and getting a reasonable agreement factors R_wp and R_p. However, the case that other O sites were occupied by Cl ions is also possible. 1.33 interstitial Li ions were introduced into unit per cell in the system of Li₁₀Si_1.5P_1.5Cl_0.5O_11.5. There are two crystallographically inequivalent Li interstitial sites I and II, and four sites equivalent to I site and eight sites equivalent to II site.²⁴ We firstly refine the crystal structure according to I site occupied by interstitial Li ions. However, it is unreasonable with the distance of 0.79 Å between interstitial Li ions and the neighbor O ions (not shown in this paper). Therefore, we supposed that the 1.33 interstitial Li ions occupy the interstitial II sites, designating as Li3 in Fig. 7 (white and blue spheres).

Fig. 7 (a) Ball and stick drawing of unit cell of Li₁₀Si_1.5P_1.5Cl_0.5O_11.5. (b) Enlarged view of the selected P/SiO/Cl₄ polyhedra. Li1 ions: black spheres, Li2 ions: orange spheres, Li3 ions: blue spheres, P ions: pink spheres, Si ions: yellow spheres, O ions: red spheres, Cl ions: green spheres.

Interstitial migration of the lithium ion were showing in dimension 2a × 2b × 2c supercells in Fig. 8. The hypothesis of interstitial mechanism described here for ions diffusion within the type II voids suggest that possible mechanism for interstitial Li ion diffusion in all three crystallographic directions (Fig. 8(a)). We thus considered the possibly efficient diffusion path along the three crystallographic directions, by selecting shorter Li–Li distance to hop which indicating smaller migration barrier²⁴ tabulated in Table 1. Fig. 8(b) described diffusion of an interstitial Li ion Li3(0.76472, 0.71260, 0.31232) to its equivalent site Li3(0.26472, 0.71260, 0.31232) along a axis. During the motion process, an interstitial Li ion Li3(8d) kicks out and replaces a neighboring Li1(8d) while the “kicked-out” Li1(8d) migrates to next neighboring Li3(8d) site, then kicks out Li1(8d) to take its equivalent site Li3(0.26472, 0.71260, 0.31232). It takes four steps from one interstitial site to its equivalent site, and the shortest Li–Li distance is 1.76 Å, and the largest Li–Li distance is 2.37 Å. The diffusion path along a axis is (Fig. 8(b)):

Li3(8d) → Li1(8d) → Li3(8d) → Li1(8d) → Li3(8d) (4)

Fig. 8 (a) Polyhedra drawing of supercell of Li₁₀Si_1.5P_1.5Cl_0.5O_11.5 having dimension 2a × 2b × 2c. (b)–(d) Lithium ions diffusion path along a–c axis, (e) projection of Li sites of supercell of Li₁₀Si_1.5P_1.5Cl_0.5O_11.5 on the b × c plane, (f) lithium ions diffusion path along a × b × c axes.

Table 1 Interstitial migration step of Li₁₀Si_1.5P_1.5Cl_0.5O_11.5 solid solution

Net direction	Step	Net distance/Å
a	Li3(0.76472, 0.71260, 0.31232)	1.76, 2.37, 1.76, 2.37
	Li1(0.60328, 0.75030, 0.33631)
	Li3(0.51472, 0.78740, 0.43768)
	Li1(0.35328, 0.74970, 0.41369)
	Li3(0.26472, 0.71260, 0.31232)
b	Li3(0.51472, 0.28740, 0.43768)	2.0, 2.0
	Li2(0.66414, 0.37500, 0.47482)
	Li3(0.51472, 0.46260, 0.43768)
c	Li3(0.48528, 0.53740, 0.56232)	2.0, 2.82, 2.37, 2.0, 2.82, 2.37
	Li2(0.33586, 0.62500, 0.52518)
	Li1(0.35328, 0.50030, 0.41369)
	Li3(0.26472, 0.53740, 0.31232)
	Li2(0.41414, 0.62500, 0.27518)
	Li1(0.39672, 0.50030, 0.16369)
	Li3(0.48528, 0.53740, 0.06232)

---

The diffusion paths along b and c axis are similar with that along a axis, both are zigzag motions. The diffusion path along b axis is (Fig. 8(c)):

Li3(8d) → Li2(4c) → Li3(8d) (5)

The diffusion path along c axis is (Fig. 8(d)):

Li3(8d) → Li2(4c) → Li1(8d) → Li3(8d) → Li2(4c) → Li1(8d) → Li3(8d) (6)

The proposed diffusion paths show a slight anisotropy along the three crystallographic directions. Based on the diffusion steps from one interstitial site to its equivalent site and the Li–Li distance, it can be deduced that the diffusion path along b axis is the most efficient and major motion path of this solid solution. Fig. 8(e) shows the projection of Li sites of supercell of Li₁₀Si_1.5P_1.5Cl_0.5O_11.5 on the b × c plane, which can visually deliver a zigzag diffusion path along b axis. Fig. 8(f) shows the 3D-zigzag network of lithium ions diffusion path along a × b × c axes. The diffusion along b axis combined with the diffusion along a × c axes.

The cathode material 0.3Li₂MnO₃·0.7LiMn_1.5Ni_0.5O₄ which proves to be able to stably cycle in a wide potential window of 2.0–5.0 V has been adopted to evaluate the electrochemical performance of Li_10.42Si_1.5P_1.5Cl_0.08O_11.92 and Li_10.42Ge_1.5P_1.5Cl_0.08O_11.92. Fig. 9(a) and (b) show the charge–discharge curves of the hybrid electrolyte Li_10.42Si_1.5P_1.5Cl_0.08O_11.92 and Li_10.42Ge_1.5P_1.5Cl_0.08O_11.92 cells at 0.1 C rate. The charge and discharge capacities of Li_10.42Si_1.5P_1.5Cl_0.08O_11.92 cell at the 1st cycle were 114.5 mA h g⁻¹ and 114.1 mA h g⁻¹, respectively, delivering almost 100% coulombic efficiency. The initial charge capacity of Li_10.42Ge_1.5P_1.5Cl_0.08O_11.92 cell was 330.1 mA h g⁻¹, much higher than the theoretical capacity of 0.3Li₂MnO₃·0.7LiMn_1.5Ni_0.5O₄ (213 mA h g⁻¹), whereas the initial discharge capacity was 133.2 mA h g⁻¹. The charge and discharge capacities after 100 cycles were 81.6 mA h g⁻¹ and 81.9 mA h g⁻¹, respectively, with 100% coulombic efficiency for Li_10.42Si_1.5P_1.5Cl_0.08O_11.92 cells (Fig. 9(c)). However, the discharge capacity was 11.2 mA h g⁻¹ for Li_10.42Ge_1.5P_1.5Cl_0.08O_11.92 cells after 23 cycles (Fig. 9(d)). The cycle performance of Li_10.42Si_1.5P_1.5Cl_0.08O_11.92 cells was much better than that of Li_10.42Ge_1.5P_1.5Cl_0.08O_11.92 cells.

Fig. 9 (a) The charge–discharge curves of Li_10.42Si_1.5P_1.5Cl_0.08O_11.92 and (b) Li_10.42Ge_1.5P_1.5Cl_0.08O_11.92 cells at 0.1 C rate. (c) Cycle performance of Li_10.42Si_1.5P_1.5Cl_0.08O_11.92 cells. (d) Cycle performance of Li_10.42Ge_1.5P_1.5Cl_0.08O_11.92 cells.

Conclusions

To achieving the combination of high ionic conductivity and excellent electrochemical stability in lithium solid electrolytes still remains a major challenge. A facile strategy was proposed to achieve reasonably high conduction and excellent electrochemical stability by the substitution of Cl for O based on the concept of bottleneck size and binding energy. New electrolytes, Li_10.42Si_1.5P_1.5Cl_0.08O_11.92 and Li_10.42Ge_1.5P_1.5Cl_0.08O_11.92, were synthesized that showed promising ionic conductivity 1.03 × 10⁻⁵ S cm⁻¹ and 3.7 × 10⁻⁵ S cm⁻¹ at 27 °C respectively, 13 orders of magnitude higher than that of the pure Li₃PO₄, and 1 order of magnitude higher than that of the pristine Li_10.5Si_1.5P_1.5O₁₂, and superior electrochemical stability with metallic lithium up to 9 V vs. Li⁺/Li, one of the solid electrolytes with most wide electrochemical window. These series of new electrolytes are orthorhombic structure of γ-Li₃PO₄ with interstitial 3D-zigzag ions migration mechanism along b axis. The charge and discharge capacities of Li_10.42Si_1.5P_1.5Cl_0.08O_11.92 cell at the 1st cycle were 114.5 mA h g⁻¹ and 114.1 mA h g⁻¹, respectively, with almost 100% coulombic efficiency. The coulombic efficiency was almost 100% after 100 cycles for Li_10.42Si_1.5P_1.5Cl_0.08O_11.92 cells. This strategy holds promising potential to develop new solid electrolytes combined with high ionic conductivity and excellent electrochemical stability thus advance lithium ion battery technology by enabling the use of metallic lithium anode.

Acknowledgements

This work was supported by SERC 2011 Public Sector Research Funding (PSF) Grant R-265-000-424-305.

Notes and references

1. J. B. Goodenough and Y. Kim, Chem. Mater., 2010, 22, 587.
2. Z. C. Zhang, L. B. Hu, H. Wu, W. Weng, M. Koh, P. C. Redfern, L. A. Curtiss and K. Amine, Energy Environ. Sci., 2013, 6, 1806.
3. L. Suo, Y. S. Hu, H. Li, M. Armand and L. Q. Chen, Nat. Commun., 2012, 4, 1481.
4. R. Kanno and M. Maruyama, J. Electrochem. Soc., 2001, 148, A742.
5. N. Kamaya, K. Homma, Y. Yamakawa, M. Hirayama, R. Kanno, M. Yonemura, T. Kamiyama, Y. Kato, S. Hama, K. Kawamoto and A. Mitsui, Nat. Mater., 2011, 10, 682.
6. Y. Seino, T. Ota, K. Takada, A. Hayashi and M. Tatsumisago, Energy Environ. Sci., 2014, 7, 627.
7. G. Sahu, Z. Lin, J. Li, Z. Liu, N. Dudney and C. D. Liang, Energy Environ. Sci., 2014, 7, 1053.
8. J. L. Schaefer, Y. Lu, S. Moganty, P. Agarwal, N. Jayaprakash and L. A. Archer, Appl. Nanosci., 2012, 2, 91.
9. W. C. West, J. F. Whitacre and J. R. Lim, J. Power Sources, 2004, 126, 134.
10. Y. Inagum and M. Itoh, Solid State Ionics, 1996, 86, 257.
11. O. Bohnke, C. Bohnke, J. O. P. Ahmed, C. Lopez, H. Duroy, F. Berre and J. L. Fourquet, Chem. Mater., 2001, 13, 1593.
12. Y. Inaguma, Y. Matsui, J. Yu, Y. J. Shan, T. Nakamura and M. Itoh, J. Phys. Chem. Solids, 1997, 58(6), 843.
13. H. Y. Hong, Mater. Res. Bull., 1978, 13, 117.
14. J. T. S. Irvine, D. C. Sinclair and A. R. West, Adv. Mater., 1990, 2, 132.
15. B. Wang, B. C. Chakoumakos, B. C. Sales, B. S. Kwak and J. B. Bates, J. Solid State Chem., 1995, 115, 313.
16. Y. W. Hu, I. D. Raistrick and R. A. Huggins, J. Electrochem. Soc., 1977, 124, 1240.
17. J. B. Bates, N. J. Dudney, G. R. Gruzalski, R. A. Zuhr, A. Choudhury, D. F. Luck and J. D. Robertson, Solid State Ionics, 1992, 647, 53.
18. P. G. Bruce and A. R. West, J. Electrochem. Soc., 1983, 130, 662.
19. R. Kanno, H. Kawamoto and Y. M. Irie, Solid State Ionics, 2000, 130, 97.
20. A. Khorassani, G. Izquierdo and A. R. West, Mater. Res. Bull., 1981, 16, 1561.
21. H. Xie, J. A. Alonso, Y. T. Li, M. T. Fern and J. B. Goodenough, Chem. Mater., 2011, 23, 3587.
22. W. Baur and H. T. Ohta, J. Solid State Chem., 1982, 44, 50.
23. R. Ubic and G. Subodh, J. Alloys Compd., 2009, 488, 374.
24. Y. A. Du and N. A. W. Holzwarth, J. Electrochem. Soc., 2007, 154, A999.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4ra11287c

---