Atomistic insights into the screening and role of oxygen in enhancing the Li+ conductivity of Li7P3S11−xOx solid-state electrolytes

Hanghui Liu ab, Zhenhua Yang *ab, Qun Wang ab, Xianyou Wang c and Xingqiang Shi d
aKey Laboratory of Materials Design and Preparation Technology of Hunan Province, School of Materials Science and Engineering, Xiangtan University, Xiangtan 411105, Hunan, China. E-mail: yangzhenhua@xtu.edu.cn
bKey Laboratory of Low Dimensional Materials & Application Technology (Ministry of Education), School of Materials Science and Engineering, Xiangtan University, Xiangtan 411105, Hunan, China
cNational Local Joint Engineering Laboratory for Key Materials of New Energy Storage Battery, National Base for International Science & Technology Cooperation, Hunan Province Key Laboratory of Electrochemical Energy Storage & Conversion, School of Chemistry, Xiangtan University, Xiangtan 411105, Hunan, China
dDepartment of Physics, Southern University of Science and Technology, Shenzhen 518055, China

Received 27th September 2019 , Accepted 15th November 2019

First published on 16th November 2019


Herein, we implement first-principles calculations to design Li7P3S11−xOx at an atomic scale, aiming to obtain stable Li7P3S11−xOx-type solid electrolyte materials with good Li+ conductivity. After searching for chemical potentials, Li2O2 is expected to be the potential raw material, and it can afford the most favorable growth environment for the synthesis of Li7P3S11−xOx (x = 0.25, 0.50, 0.75 and 1). Among these compounds, it is found that Li7P3S10.25O0.75 exhibits the most desirable Li+ conductivity of 109 mS cm−1 at 300 K, which is far higher than that of Li7P3S11 (50 mS cm−1 at 300 K). By structural analysis, it is demonstrated that the Li diffusion pathway in Li7P3S10.25O0.75 is significantly broadened relative to that in Li7P3S11 (71.38 Å3vs. 69.48 Å3), which breaks the bottleneck during Li diffusion. Moreover, the resistance of Li ion diffusion in Li7P3S10.25O0.75 decreases due to the balance of interactions between Li and its neighbouring atoms at the transition state, which induces a much lesser energy barrier of Li7P3S10.25O0.75 than that of Li7P3S11 (0.20 eV vs. 0.31 eV). Moreover, introducing Li vacancies is unlikely to alter the essence of the inherent superionic conductivity of Li7P3S10.25O0.75. Furthermore, Li7P3S10.25O0.75 can maintain good thermal stability and similar electrochemical stability to Li7P3S11. This study successfully clarifies the role of oxygen in enhancing the Li+ conductivity of Li7P3S11−xOx. Moreover, it affords a new strategy to design other solid-state electrolytes with good Li+ conductivity.


Introduction

Lithium-ion batteries (LIBs) have been extensively used in portable appliances and electric vehicles due to their excellent storage properties.1,2 Conventional LIBs mainly consist of electrodes, electrolyte materials and separators.3–6 Among these components, electrolyte materials play a key role in the development of LIBs.7 Up to now, organic liquid and polymer electrolytes have been most widely used due to their excellent Li-ion conductivity.8 Unfortunately, they both suffer from liquid leakage, highly toxicity and flammability, thus causing a safety hazard.9–12 The development of inorganic solid electrolytes, such as sulfides and oxides, is the essential way to improve the safety performance of LIBs.13,14 Oxide-based solid electrolytes15–22 have excellent electrochemical stability, but they have high interface impedance, which results in the failure of Li+ transport.23 In contrast, lithium ions can transfer faster in sulfide-based solid electrolytes than oxide-based solid electrolytes due to the lower electronegativity of sulfur than that of oxygen.24 Moreover, sulfide-based solid electrolytes have larger transmission channels for lithium ions than oxide-based solid electrolytes due to the larger radius of sulfur,25 which induces higher ionic conductivity for sulfide-based solid electrolytes.26 Furthermore, sulfide-based solid electrolytes have better ductility, and they can be closely contacted with electrode materials.27 Therefore, sulfide-based solid electrolytes, such as Li2S–P2S5,28,29 Li2S–P2S5–P2O5,30 Li2S–P2S5–P2S331 and Li2S–GeS2–P2S532 systems, have wider application prospects.27,33 Among sulfide-based solid electrolytes, Li7P3S11 is a promising next-generation solid-state electrolyte.34 It is theoretically reported that Li7P3S11 has the lowest diffusion barrier (0.30 eV) due to its body centered cubic (bcc) anion accumulation structure.26 However, Li7P3S11 has a large band gap of 3.50 eV,35 which induces poor electronic conductivity.36 Moreover, Li7P3S11 has good compatibility with lithium metal and a wide electrochemical window.37 Consequently, several electrochemical properties have been observed when Li7P3S11 acts as a solid electrolyte in LIBs.33,34 However, it is still far from meeting the increasing demands of electric vehicles and energy storage systems. More importantly, the ionic conductivity needs to be further improved. It is well known that ionic doping is one strategy to improve the conductivity of materials.38,39 Impressively, an enhancement in ionic conductivity has been achieved by the moderate substitution of oxygen for sulfur in Li10SiP2S12 solid-state electrolytes (from 1.6 × 10−3 S cm−1 to 3.1 × 10−3 S cm−1).40 As for Li7P3S11, considerable progress has been made in oxygen doping. Experimentally, Li7P3S11−xOx has been successfully fabricated using a mixture of 1 mol% P2S3 and 3 mol% P2O5 at ambient temperature.30 Furthermore, its ionic conductivity can reach 4.6 × 10−3 S cm−1, which is higher than that of Li7P3S11 (3.2 × 10−3 S cm−1). However, the oxygen doping mechanism including thermodynamic properties and Li-ion diffusion dynamics in Li7P3S11−xOx has still not been established. For Li7P3S11−xOx, it is difficult to experimentally justify the quantitative relationship among formation energy, ionic conductivity, band gap, and oxygen doping concentration (x) at an atomic scale. Therefore, we performed first-principles calculations to clarify the inherent physical and chemical behaviour of oxygen in Li7P3S11−xOx. Furthermore, we attempted to obtain Li7P3S11−xOx with poor electronic conductivity, excellent ionic conductivity, good thermal stability and electrochemical stability. We hope that our results can afford important theoretical guidance to design Li7P3S11-based solid electrolytes. Moreover, this method might be feasible to design other solid electrolytes.

Methodology

Computational details

First-principles calculations within the density functional theory (DFT) framework have been implemented by employing the Vienna ab initio simulation package.41,42 The generalized gradient approximation (GGA) with the Perdew–Burke–Ernzerhof (PBE) method43 was applied to treat the exchange–correlation energy. Projector augmented wave (PAW) method44 was used to describe the core electrons. The valence electronic states of 2s1, 3s23p3, 3s23p4 and 2s22P4 were taken into account as the valence states for Li, P, S and O, respectively. The cut-off energy was set to 350 eV for Li–P–S system and 520 eV for Li–P–S–O system, respectively. The 4 × 4 × 4 grid of k-points sampling was adopted, and all atoms were relaxed to their equilibrium positions with a force convergence criterion of 0.03 eV Å−1. The electronic band gap was calculated by PBE, which yields the band gap of Li7P3S11 (Fig. S1 in the ESI). Oxygen-doped Li7P3S11 was simulated using the 84-atom 1 × 2 × 1 supercells (Li28P12S44) with periodic boundary conditions. The substitutional defect was simulated by replacing S with O atoms. To obtain oxygen-doping concentration on the effect of Li7P3S11, the crystal models of Li28P12S43O, Li28P12S42O2, Li28P12S41O3 and Li28P12S40O4 were constructed, which correspond to Li7P3S10.75O0.25, Li7P3S10.50O0.50, Li7P3S10.25O0.75 and Li7P3S10O.

Defect formation energies

Impurity formation energy (Ef(X)) was implemented to judge the difficulty of oxygen doping in Li7P3S11. Moreover, it can be obtained using the following equation:45,46
 
image file: c9cp05329h-t1.tif(1)
where Etot(X) and Etot(bulk) represent the total energies of the impurity complexes and perfect bulk material, respectively, ni is the number of elements i, ni > 0 (ni < 0) represents the case that atoms are incorporated into (removed from) Li7P3S11; μi is the chemical potential of element i, and Δμi indicates the relative chemical potential under different chemical growth environments.

μ Li, μP and μS can be obtained by calculating the total energies per Li, P and S in their bulk phases. For the calculation of μO, O2 molecule is placed in a cubic box with a lattice length of 15 Å. The calculated total energy of O2 is −8.72 eV, which is in excellent agreement with the reported value in literature (−8.78 eV).47

The crystal structures of competing phases can be obtained from inorganic crystal structure database (ICSD), and formation enthalpies (Tables S1 and S2, ESI) were first calculated by DFT, and then the chemical potential limits analysis program (CPLAP)48–50 was implemented to treat the relative chemical potential (Δμi) of the stability region.

Climbing image nudged elastic band method

The energy barriers of Li-ion diffusion in Li7P3S11 and oxygen-doped Li7P3S11 were theoretically studied by climbing-image nudged elastic band method (CI-NEB).51–53 The Li+ vacancy migration simulations were performed using 1 × 2 × 1 supercells. Moreover, compensating background charge was adopted for the overall charge neutrality.

Ab initio molecular dynamics simulations

To further simulate Li+ conductivity at different temperatures, a more realistic approach called ab initio molecular dynamics (AIMD) simulations was performed in the constant volume (NVT) ensemble. Moreover, single Li+ vacancy was introduced with a compensating background charge.54,55 Cut-off energies of 280 eV and 400 eV were set for perfect and defect structures, respectively. In addition, a minimal Γ-centered 1 × 1 × 1 k-point mesh was adopted to reduce the computational cost. The time step of molecular dynamics was 2 fs, and the diffusivity of Li ion can be defined as follows:56
 
image file: c9cp05329h-t2.tif(2)
where d denotes the dimensionality factor of diffusion path (d = 3 in this case), t denotes the time duration and Δr(t) denotes the average mean square displacement (MSD).

Accordingly, using the obtained diffusivity D(T), the ionic conductivity can be derived from Nernst–Einstein relation as follows:56

 
image file: c9cp05329h-t3.tif(3)
where ρ is the molar density of diffusing ions in the unit cell, z equals +1, F is the Faraday constant, and R is the gas constant.

Results and discussion

Bulk properties

Li7P3S11 crystal belongs to the triclinic space group P[1 with combining macron]. The structure consists of PS4 tetrahedra, and some of the PS4 tetrahedra are linked together to form P2S7 ditetrahedra by sharing the vertex S atom. Li atoms are located between PS43− and P2S74−, and the P2S7 ditetrahedra has good bulkiness, which provides diffusion channel for Li ion diffusion.57 The primitive cell is composed of two Li7P3S11 molecular units with 42 atoms. There are different atomic positions in Li7P3S11 molecular unit: seven for Li atoms, eleven for S atoms and three for P atoms. Moreover, the optimized lattice parameters are as follows: a = 12.68 Å, b = 6.29 Å, c = 12.53 Å and α = 103.05°, β = 113.64°, γ = 73.45°, which is in good agreement with the experimental results (a = 12.50 Å, b = 6.03 Å, c = 12.53 Å and α = 102.85°, β = 113.20°, γ = 74.47°).35,58 The calculated band gap is 2.66 eV, which is close to the reported value54 (2.60 eV). After performing a structure search (Fig. S2–S17, ESI), the optimum structures of Li7P3S10.75O0.25, Li7P3S10.50O0.50, Li7P3S10.25O0.75 and Li7P3S10O were identified (Fig. S18, ESI). These compounds were fully relaxed and the corresponding lattice constants are summarized in Table 1. Due to the smaller radius of O2− (0.70 Å) than that of S2− (1.05 Å),59 it is noted that their crystal volumes show gradual trend of shrinkage with increasing oxygen-doping concentration (x), and they are accompanied by increasing lattice distortion.
Table 1 Lattice parameters of Li7P3S10.75O0.25, Li7P3S10.50O0.50, Li7P3S10.25O0.75 and Li7P3S10O
Compounds a (Å) b (Å) c (Å) α (°) β (°) γ (°) V3)
Li7P3S11 12.68 12.58 12.53 102.92 113.71 73.33 1740.38
Li7P3S10.75O0.25 12.73 12.44 12.52 103.16 113.43 74.34 1734.77
Li7P3S10.50O0.50 12.63 12.33 12.49 102.79 112.79 75.03 1716.52
Li7P3S10.25O0.75 12.63 12.27 12.44 102.30 112.69 75.36 1706.11
Li7P3S10O 12.62 12.24 12.30 102.40 113.09 75.76 1678.69


Analysis of structural stability in Li7P3S11−xOx (x = 0.25, 0.50, 0.75 and 1)

Next, we turn to investigate the structural stability of Li7P3S11−xOx (x = 0.25, 0.50, 0.75 and 1). First, the relative chemical potential ΔμiμLi, ΔμP and ΔμS) is implemented to determine the preferential environment for the formation of Li7P3S11. Moreover, the allowed range of Δμi is then determined by the following constraints:
 
μLi + 3ΔμP + 11ΔμS ≤ ΔHf(Li7P3S11)(4)
 
ΔμLi ≤ 0, ΔμP ≤ 0, ΔμS ≤ 0(5)
 
μLi + ΔμS ≤ ΔHf(Li2S)(6)
 
μP + 5ΔμS ≤ ΔHf(P2S5)(7)
 
μP + 3ΔμS ≤ ΔHf(P4S3)(8)
 
μP + 7ΔμS ≤ ΔHf(P4S7)(9)
 
μP + 9ΔμS ≤ ΔHf(P4S9)(10)
where ΔHf is formation enthalpy. By using these requirements, the range of Δμi for the formation of Li7P3S11 is obtained, as visualized in Fig. 1. It is noted that Li7P3S11 allows for a wide range of relative chemical potentials. The polygon A–B–C–D–E–F is a region in the ΔμLivs. ΔμP plane in which Li7P3S11 is stable. Outside this region, the precipitation of other phases, such as Li2S, P2S5, P4S7, P4S9 and P4S3, occurs. Point A, bound by Li2S and elemental phosphorus, indicates a phosphorus-rich condition: ΔμLi, ΔμP and ΔμS are −1.78 eV, 0 eV and −0.49 eV, respectively. Point B and point C both represent sulphur-rich condition, and they correspond to the equilibrium of Li7P3S11 with Li2S and P2S5. At point D and point E, P–S secondary phases possibly form. Point D is bound by P2S5, P4S7 and P4S9, whereas point E is bound by P4S7 and P4S3. Similarly, point F is bound by P4S3 and elemental phosphorus.

image file: c9cp05329h-f1.tif
Fig. 1 Accessible region of chemical potential for Li7P3S11 spanned by ΔμLi and ΔμP, where the gray shaded region indicates the stable region of Li7P3S11.

Furthermore, in order to introduce oxygen easily into Li7P3S11, we turn to optimize formation conditions of Li7P3S11−xOx (x = 0.25, 0.50, 0.75 and 1) via formation energies. Here, ΔμLi, ΔμP and ΔμS originate from the formation conditions of Li7P3S11 since Li7P3S10.75O0.25, Li7P3S10.50O0.50 and Li7P3S10.25O0.75 must maintain similar crystal structures to that of Li7P3S11. Therefore, how to select relative chemical potential of oxygen (ΔμO) is a key issue. Accordingly, the relative chemical potential of oxygen (ΔμO), together with values of ΔμiμLi, ΔμP, ΔμS) at points A–F, was considered (for details see Fig. S19–S24, ESI). After searching for the relative chemical potential of oxygen (ΔμO; Table S3 in ESI), an optimal condition (ΔμLi = −1.78 eV, ΔμP = 0 eV, ΔμS = −0.49 eV, ΔμO = −1.66 eV) can be obtained to synthesize Li7P3S11−xOx due to the low formation energy. In other words, Li2O2 is expected to be the most favorable raw material that can afford the growth environment for the synthesis of Li7P3S11−xOx. Then, we presented the formation energies of Li7P3S11−xOx (x = 0.25, 0.50, 0.75 and 1) under optimal conditions (Fig. 2). As seen from Fig. 2, they show a thermodynamically stable state. Moreover, it gets easier and easier to form Li7P3S11−xOx with increasing oxygen-doping concentration (x) due to the decreasing formation energies. Since both Li3PO4 and P2O5 substitutions in Li7P3S11 have been carried out in experiments,30,60 we have presented their formation energies as a function of oxygen-doping concentration (x) under these growth environments, thus aiming to compare their level of difficulty for oxygen doping. For Li3PO4 substitution, the formation energies of Li7P3S11−xOx (x = 0.25, 0.50, 0.75 and 1) are positive, which indicates that Li7P3S11−xOx exhibits a thermodynamically unstable state. Notably, Li7P3S10.75O0.25 exhibits a slightly positive formation energy (0.21 eV), which implies that its thermodynamically metastable structure is likely to be observed in experiment. Moreover, this phenomenon has been confirmed in experiment.60 Unfortunately, Li7P3S11−xOx (x = 0.50, 0.75 and 1) is actually very difficult to form due to its positive formation energy. This explains why Li7P3S10.75O0.25 exhibits the highest conductivity among Li7P3S10.75O0.25, Li7P3S10.50O0.50, Li7P3S10.25O0.75 and Li7P3S10O. For P2O5 substitution, the formation energies of Li7P3S11−xOx (x = 0.25, 0.50, 0.75 and 1) are all negative, which indicates that Li7P3S11−xOx can maintain a thermodynamically stable character under this growth condition. In other words, P2O5 substitution is a more effective method to introduce oxygen in Li7P3S11 relative to Li3PO4.


image file: c9cp05329h-f2.tif
Fig. 2 The formation energies of Li7P3S10.75O0.25, Li7P3S10.50O0.50, Li7P3S10.25O0.75 and Li7P3S11O under different growth conditions (Li3PO4, P2O5 and Li2O2 substitutions).

Selection of Li7P3S11−xOx (x = 0, 0.25, 0.50, 0.75 and 1)

It is believed that poor electronic conductivity and excellent ionic conductivity are essential conditions for solid electrolytes. Therefore, we first investigated the quantitative relationship between energy barriers and band gaps of Li7P3S11−xOx (x = 0, 0.25, 0.50, 0.75 and 1), aiming to identify oxygen-doped Li7P3S11 with good electrochemical performance. Since Li ions mainly migrate along the b crystallographic direction in pure Li7P3S11,35,54 we only focus on the b direction to discuss the case that Li ions migrate in both pure and oxygen-doped Li7P3S11. Here, the well-established climbing image nudged-elastic-band (CI-NEB) technique is employed to obtain the energy barriers of Li ions in Li7P3S10.75O0.25, Li7P3S10.50O0.50, Li7P3S10.25O0.75 and Li7P3S10O, along with Li7P3S11 (Fig. S25 and S26, ESI). Moreover, the energy barriers and band gaps of Li7P3S11−xOx (x = 0, 0.25, 0.50, 0.75 and 1) are presented in Fig. 3. Notably, the energy barrier of Li7P3S11 is 0.31 eV, which is close to the previous value (0.30 eV).35 By comparison, it is found that energy barriers of Li7P3S11−xOx gradually decrease when the oxygen-doping concentration (x) varied from 0 to 0.75. However, the energy barriers of Li7P3S11−xOx increased if the oxygen-doping concentration (x) kept on increasing. To a large extent, this is due to a collapse of diffusion channel induced by the largest lattice distortion of Li7P3S10O. Moreover, Li7P3S10.25O0.75 has the lowest Li ion diffusion barrier of 0.20 eV among these five compounds (see Table S4 and Fig. S27, S28, ESI), which is far lower than the energy barrier of Li-ion diffusion in Li7P3S11 (0.31 eV). In other words, oxygen doping plays a key role in decreasing the energy barrier. However, the band gaps of Li7P3S11−xOx would fluctuate with the increase in oxygen-doping concentration (x). In other words, oxygen doping has less effect on the band gaps of Li7P3S11−xOx, and they can still exhibit good electrical insulation after oxygen doping.
image file: c9cp05329h-f3.tif
Fig. 3 The calculated energy barriers and band gaps of Li7P3S11 and Li7P3S11−xOx (x = 0.25, 0.50, 0.75 and 1).

Exploration of oxygen regulation on the crystal structure and electronic structure of Li7P3S10.25O0.75

Furthermore, due to the lowest Li-ion diffusion barrier in Li7P3S10.25O0.75, we considered Li7P3S10.25O0.75 as a research object, together with Li7P3S11 as a reference, aiming to probe into the role of oxygen in Li7P3S10.25O0.75 by the analysis of crystal and electronic structures. We first verified that both Li7P3S11 and Li7P3S10.25O0.75 have excellent thermal stabilities by AIMD using the NVT ensemble and a Nose–Hoover thermostat at 400 K (Fig. S29, ESI). Next, we turned to compare the atomic configurations of P2S7 ditetrahedra in Li7P3S11 and Li7P3S10.25O0.75 to further clarify the effects of oxygen doping on the crystal structure of diffusion channel for Li ion in Li7P3S10.25O0.75. It is evident that local distortion occurs in Li7P3S10.25O0.75 with O substitution for S. As compared with the pristine crystal, the shortest S–S distance (S2–S3) drastically increased. Here, S2–S3 distance in Li7P3S10.25O0.75 was elongated by 0.33 Å. Accordingly, the Li diffusion bottleneck of the narrow channel was broken. Next, in order to further understand why diffusion channel is broadened, we turn to investigate the structural change of clusters, which consist of S1, S2, S4, S5 and P before and after oxygen doping. Considering that Li–O bond occurs in the Li7P3S10.25O0.75, the nearest Li atom is included (Fig. 4c and d). As oxygen is doped into Li7P3S11, the P–S4 bond length and S4–P–S5 bond angle slightly change by −0.01 Å and 0.06°, respectively. Moreover, the O–P–S5 bond angle (110.42°) is smaller than the S1–P–S5 bond angle (112.86°), the O–P bond length is shorter than S1–P bond (1.54 Å vs. 2.01 Å), and S2–P bond length is slightly shortened (from 2.05 Å to 2.02 Å). To keep the structure stable, the O–P–S2 bond angle has to become larger than the S1–P–S2 bond angle; moreover, the S2–P–S4 bond angle increases from 89.74° to 92.44°. As a result, Li diffusion pathway in Li7P3S10.25O0.75 is significantly broadened relative to that in Li7P3S11 (71.38 Å3vs. 69.48 Å3, which is obtained by the open source Zeo++ software46,61,62), and the bottleneck is broken during Li diffusion.
image file: c9cp05329h-f4.tif
Fig. 4 Local atomic structures of (a) P2S7 ditetrahedra and (c) the cluster consist of S1, S2, S4, S5, P and Li in Li7P3S11 together with (b) P2S7 ditetrahedra and (d) the cluster consist of O, S2, S4, S5, P and Li in Li7P3S10.25O0.75.

Owing to the largest structure distortion occurring in the cluster consisting of S1, S2 and P, we investigated the electronic structure of O–P–S2 cluster, along with S1–P–S2 cluster in Li7P3S11 for comparison, as shown in Fig. 5, thus aiming to obtain the regulation rule at an electronic scale by oxygen doping. To clarify change rule of atomic bonds in clusters, the partial density of states (PDOS) of S1–P–S2 and O–P–S2 clusters were calculated and presented in Fig. 5. For the O–P–S2 cluster, a more obscure overlap of O-2p and P-3p orbitals can be observed, and the O–P bond exhibits a stronger ionic character relative to the S1–P bond. As for the O–P bond, the Bader charges of O and P are −1.48 |e| and 1.94 |e|, respectively. Compared with the S1–P bond (−0.84 |e| for S1, 1.20 |e| for P), it is obvious that the O atom captured more electrons from the bonding P atom due to the higher electronegativity of O than S. As a result, hybridization between P-3p and S2-3p at −3.12 eV appears. Moreover, the Bader charge of S2 atom changed from −0.89 |e| to −0.88 |e|, respectively. In other words, the P–S2 bond in O–P–S2 cluster showed a slightly more covalent character than that in S1–P–S2. In particular, with oxygen doping, the orbitals near the CBM and VBM did not move to lower or higher energy-region, respectively, and the band gap was 2.63 eV, which is close to the band gap value of Li7P3S11. In other words, Li7P3S10.25O0.75 continues to possess its insulation character like Li7P3S11. Consequently, the analysis of total electron charge densities was implemented to gain a better visual understanding of the bonding character. It is illustrated that S2 is polarized slightly toward the nearest P atom, resulting in more charge aggregation regions in the P–S2 bond with the incorporation of oxygen. Therefore, the covalent interaction of the P–S2 bond increases slightly, which is consistent with the results discussed above (Fig. 6).


image file: c9cp05329h-f5.tif
Fig. 5 The partial density of states and Bader charges for the S1–P–S2 and O–P–S2 clusters, respectively.

image file: c9cp05329h-f6.tif
Fig. 6 Total electron charge densities for (a) S1–P–S2 cluster and (b) O–P–S2 cluster.

To further understand the hopping process of the Li ion in Li7P3S10.25O0.75, along with Li7P3S11 as a reference, we thoroughly investigated Li ion diffusion in Li7P3S11 and Li7P3S10.25O0.75. The complete migration process can be divided into two stages: Path ① and Path ②, as shown in Fig. S30 and S31 (ESI). Considering its higher energy barrier for Li-ion diffusion, we focus on Path ① for Li7P3S11 and Path ② for Li7P3S10.25O0.75. Moreover, we explored the changes in local crystal structures when Li ion transfers from the initial position, via the saddle point, to the final position (Fig. 7). To clearly depict the interaction between Li ion and its surrounding S atoms at these three positions, we only presented the local structures, which consist of Li ion and its surrounding S atoms (Fig. 8). In Li7P3S11, it is noted that atomic distances among the initial Li ion and its neighbouring five sulfur atoms (S6, S7, S8, S9, S10) vary from 2.46 Å to 2.95 Å, which are all larger than the ionic bond length of Li–S (1.13 Å). In other words, the interatomic interaction between Li and S was quite weak. As a result, Li ion can transfer freely from the initial position to the final position. As the Li ion hops to the transition state, the number of neighbouring sulfur atoms is reduced to three with the Li–S (S7, S10 and S11) distances of 2.43 Å, 2.50 Å and 2.41 Å. Moreover, Li ion is inclined to be situated between S10 and S11. In the final structure, the sulfur atoms around Li ion are denoted as S7, S10, S11, S12 and S13, respectively, and their corresponding Li–S distances are 2.54 Å, 2.91 Å, 2.49 Å, 2.44 Å and 2.57 Å, respectively. In this hopping process, the final structure is more stable than the initial structure, implying that the hopping process is thermodynamically spontaneous. Similarly, for Li7P3S10.25O0.75, one Li ion is surrounded by five sulfur atoms (S6, S14, S15, S16, S17) in the initial structure, and the corresponding Li–S distances vary from 2.41 Å to 2.92 Å. In the transition structure, it is found that Li–S distances vary from 2.30 Å to 2.47 Å. As a result, the Li–S interaction is still weak. Most importantly, it is noted that the Li ion is situated inside the triangle formed by S atoms, and the weak Li–S interaction can counteract each other, thus forming the balance state between Li and S atoms. Accordingly, the resistance of Li ion diffusion in Li7P3S10.25O0.75 is relatively small, which induces the lower energy barrier of Li-ion diffusion compared to Li7P3S11. In the final structure, the Li–S distances (S6, S7, S8, S9, and S10) correspond to 2.43 Å, 2.47 Å, 2.57 Å, 2.54 Å and 3.01 Å, respectively.


image file: c9cp05329h-f7.tif
Fig. 7 Li-ion migration pathways and the relevant atomic annotations for (a) path ① of Li7P3S11 and (b) path ② of Li7P3S10.25O0.75, together with the local atomic structure around the initial, transitional, final Li ions in (c) the path ① of Li7P3S11 and (d) the path ② of Li7P3S10.25O0.75.

image file: c9cp05329h-f8.tif
Fig. 8 Diffusion of Li ion in (a) Li7P3S11 and (b) Li7P3S10.25O0.75.

Li+ conductivity in Li7P3S10.25O0.75

Next, we turn to perform AIMD simulations to investigate Li+ diffusion kinetics in Li7P3S10.25O0.75 at different temperatures. Li+ conductivities of Li7P3S10.25O0.75 and Li7P3S11 were first determined (the mean square displacements were listed in Fig. S32 and S33, ESI). It is observed that the Li+ conductivity of Li7P3S11 at 300 K is fitted to 50 mS cm−1, which is in good agreement with the previously reported value (57 mS cm−1) by Ong et al.54 Notably, the predicted Li+ conductivity of Li7P3S10.25O0.75 at 300 K increases to 109 mS cm−1, which indicates that oxygen doping is an effective approach to enhance the Li+ conductivity of Li7P3S11.

Based on our calculated ionic conductivities (σ) as a function of temperature, we presented the Arrhenius plots for Li7P3S11 and Li7P3S10.25O0.75 (see Fig. 9) according to the following relation63

 
image file: c9cp05329h-t4.tif(11)
where e, CLi, l, v0 and k are the electronic charge, Li ion concentration, hopping distance and Boltzmann's constant, respectively, and Ea and T indicate activation energy and temperature, respectively. As shown in Fig. 9, compared with Li7P3S11, it is observed that ionic conductivity (σ) in Li7P3S10.25O0.75 is much larger when the temperature is lower than 721 K. Moreover, Li7P3S10.25O0.75 has a lower activation energy than Li7P3S11 (142 meV vs. 176 meV). In other words, the introduction of appropriate oxygen dopant can remarkably increase the ionic conductivity of Li7P3S11 over its usual operating temperature range.


image file: c9cp05329h-f9.tif
Fig. 9 Arrhenius plots for (a) Li7P3S11 and (b) Li7P3S10.25O0.75.

In order to intuitively describe the dynamic behaviour of Li ions, we used the van Hove function (G)46,54,56 to analyze and compare the motion of Li ion in Li7P3S11 and Li7P3S10.25O0.75. The van Hove function can be divided into self-part (Gs(t,r)) and distinct part (Gd(t,r)) (see Methodology in the ESI). Gs(t,r) can characterize the motion information of a particle moving off its initial location by a distance (r) after time (t). While Gd(t,r) can determine the movement of the remaining particles after time (t).

Here, we presented Gs(t,r) and Gd(t,r) for only Li7P3S11 and Li7P3S10.25O0.75 at 400 K (Fig. 10). As shown in Fig. 10, two persistent peaks were observed between 3.5 Å and 4.5 Å and between 6 Å and 7 Å in Gd(t,r) for these two structures. Compared with the previous results of Li7P3S11 treated at 600 K,54 a much more inactive movement of Li ions in Li7P3S11 can be observed at 400 K. It is noteworthy that obvious peaks for r < 1 Å only occur after 30 ps in Li7P3S11, while obvious peaks for r < 1 Å appear earlier and they arise after 20 ps in Li7P3S10.25O0.75. This indicates the previous occurrence of highly correlated ionic motion in Li7P3S10.25O0.75. However, Gs(t,r) of Li7P3S11 exhibits pronounced peaks for r < 2 Å before 50 ps. By comparison of Gs(t,r) between Li7P3S11 and Li7P3S10.25O0.75, it is found that the peak for r < 2 Å in Li7P3S10.25O0.75 disappears much earlier (37 ps vs. 50 ps). Moreover, additional peaks can be observed for the larger distance values (r) in Li7P3S10.25O0.75. In other words, Li ions in Li7P3S10.25O0.75 tend to diffuse faster and are farther away from their initial positions than those in Li7P3S11.


image file: c9cp05329h-f10.tif
Fig. 10 Plots of the distinct parts (Gd(t,r)) of the van Hove correlation function for (a) Li7P3S11 and (c) Li7P3S10.25O0.75, together with the self-parts (Gs(t,r)) for (b) Li7P3S11 and (d) Li7P3S10.25O0.75 at 400 K.

Lastly, we further elaborated the influence of Li vacancy defects in Li7P3S10.25O0.75, and the Li+ conductivities in Li7−xP3S10.25O0.75 (x = 0, 0.25, 0.50 and 0.75) were investigated. Moreover, the optimal crystal structures of Li7−xP3S10.25O0.75 (x = 0, 0.25, 0.50 and 0.75) were obtained using a structural search (Fig. S34–S49 in the ESI). Then, considering the cost of simulated time, the conductivities of Li7−xP3S10.25O0.75 at 600 K are calculated and illustrated in Fig. S50 and S51 (ESI). Thus, it is found that introducing Li vacancies has almost no effect on the Li+ conductivity of Li7−xP3S10.25O0.75. Therefore, the presence of Li vacancies is unlikely to alter the essence of inherent superionic conductor of Li7−xP3S10.25O0.75.

Electrochemical stability of Li7P3S11 and Li7P3S10.25O0.75

In addition to excellent Li+ conductivity, maintaining good electrochemical stability is an important aspect for the solid-state electrolytes.

Accordingly, the Li grand potential phase stability plots for Li7P3S11 and Li7P3S10.25O0.75 were performed using pymatgen.56 The decomposition reactions for every voltage plateau are listed in the ESI (Tables S5 and S6). Fig. 11 shows that the reduction products of Li7P3S11 at 0 V are Li2S and Li3P, which is in excellent agreement with the experimental results.64 Moreover, the voltage range (2.28–2.31 V) without Li uptake and loss can be observed, indicating the intrinsic electrochemical stability range. Furthermore, both Li3PS4 and P2S5 prefer to form in this range. Fortunately, these two compounds possess electronic insulation and ionic conductivity, which can act as a passivation layer to prevent the further decomposition of the electrolyte. Similarly, the electrochemical stability of Li7P3S10.25O0.75 was assessed. It was found that the dopant oxygen is incorporated into Li3PO4 in the stable range. Notably, the framework structure of Li3PO4 was identical to that of Li3PS4. Therefore, the production of Li3PO4 did not affect the electrochemical stability range. As a result, Li7P3S10.25O0.75 can still maintain a similar electrochemical stability to Li7P3S11.


image file: c9cp05329h-f11.tif
Fig. 11 Voltage profiles of (a) Li7P3S11 and (b) Li7P3S10.25O0.75. Only the phase equilibria in the stable range against Li uptake and loss were marked for concise.

Conclusions

In conclusion, our results first reveal that crystal volumes of Li7P3S11−xOx (x = 0, 0.25, 0.50, 0.75 and 1) decrease gradually with the increase of oxygen-doping concentration (x) and that the most significant lattice deformation occurs on Li7P3S10O. Moreover, ΔμiμLi = −1.78 eV, ΔμP = 0 eV, ΔμS = −0.49 eV, ΔμO = −1.66 eV) is an optimal condition to synthetize Li7P3S11−xOx. In other words, Li2O2 is expected to be the most favorable raw material to introduce oxygen into Li7P3S11. Notably, the band gaps of Li7P3S11−xOx (x = 0.25, 0.50, 0.75 and 1) show very small fluctuations with increasing oxygen-doping concentration. Encouragingly, Li7P3S10.25O0.75 has a much lower energy barrier of 0.20 eV, which is far lower than that of Li7P3S11 (0.31 eV). Moreover, it is demonstrated that the shortest S–S distance in diffusion channel increases obviously with oxygen introduction (3.96 Å vs. 3.63 Å). Accordingly, Li diffusion pathway in Li7P3S10.25O0.75 is significantly broadened relative to Li7P3S11 (71.38 Å3vs. 69.48 Å3), which breaks the bottleneck during Li diffusion. Furthermore, the resistance of Li ion diffusion in Li7P3S10.25O0.75 decreases due to the balance of interaction between Li and its neighboring atoms at the transition state. Then, it is confirmed that Li+ conductivity is much larger when the temperature is lower than 721 K in Li7P3S10.25O0.75. Moreover, Li7P3S10.25O0.75 has lower activation energy than Li7P3S11 (142 meV vs. 176 meV). Next, it is found that introducing Li vacancies has almost no effect on the Li+ conductivities of Li7−xP3S10.25O0.75, which indicates that Li vacancies are unlikely to alter the essence of the inherent superionic conductor of Li7P3S10.25O0.75. Thus, it is verified that both Li7P3S11 and Li7P3S10.25O0.75 have excellent thermal stabilities at 400 K. Furthermore, it is noted that Li7P3S10.25O0.75 can provide a similar electrochemical stability range to Li7P3S11. As a result, Li7P3S10.25O0.75 still can maintain similar electrochemical stability to Li7P3S11. Therefore, Li7P3S10.25O0.75 is expected to be a much better candidate than Li7P3S11 for high-performance solid electrolytes.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This study is financially supported by the National Natural Science Foundation of China (Grant No. 21573187), Hunan Provincial Natural Science Foundation of China (Grant No. 2017JJ2246), Start-up Funds for Doctor Supported by Xiangtan University (Grant No. 12QDZ02), Opening Foundation of Key Laboratory of Materials Design and Preparation Technology of Hunan Province (Grant No. KF20140701), and the Financial Support from the Shenzhen Fundamental Research Foundation (Grant No. JCY20170817105007999).

References

  1. Z. Shadike, Y. N. Zhou, L. L. Chen, Q. Wu, J. L. Yue, N. Zhang, X. Q. Yang, L. Gu, X. S. Liu, S. Q. Shi and Z. W. Fu, Nat. Commun., 2017, 8, 566 CrossRef PubMed.
  2. Y. Wang, D. Lu, M. Bowden, P. Z. El Khoury, K. S. Han, Z. D. Deng, J. Xiao, J.-G. Zhang and J. Liu, Chem. Mater., 2018, 30, 990–997 CrossRef CAS.
  3. M. Armand and J.-M. Tarascon, Nature, 2008, 451, 652–657 CrossRef CAS PubMed.
  4. Y. Tang, C. Wang, J. Zhou, Y. Bi, Y. Liu, D. Wang, S. Shi and G. Li, J. Power Sources, 2013, 227, 199–203 CrossRef CAS.
  5. S. Shi, H. Zhang, X. Ke, C. Ouyang, M. Lei and L. Chen, Phys. Lett. A, 2009, 373, 4096–4100 CrossRef CAS.
  6. Y. Li, D. Wu, Z. Zhou, C. R. Cabrera and Z. Chen, J. Phys. Chem. Lett., 2012, 3, 2221–2227 CrossRef CAS PubMed.
  7. J. B. Goodenough and Y. Kim, Chem. Mater., 2010, 22, 587–603 CrossRef CAS.
  8. K. Xu, Chem. Rev., 2004, 104, 4303–4418 CrossRef CAS PubMed.
  9. H. Jia, H. Onishi, R. Wagner, M. Winter and I. Cekic-Laskovic, ACS Appl. Mater. Interfaces, 2018, 10, 42348–42355 CrossRef CAS.
  10. S. Röser, A. Lerchen, L. Ibing, X. Cao, J. Kasnatscheew, F. Glorius, M. Winter and R. Wagner, Chem. Mater., 2017, 29, 7733–7739 CrossRef.
  11. N. Membreño, K. Park, J. B. Goodenough and K. J. Stevenson, Chem. Mater., 2015, 27, 3332–3340 CrossRef.
  12. B. Xu, H. Duan, H. Liu, C. A. Wang and S. Zhong, ACS Appl. Mater. Interfaces, 2017, 9, 21077–21082 CrossRef CAS PubMed.
  13. K. H. Park, Q. Bai, D. H. Kim, D. Y. Oh, Y. Zhu, Y. Mo and Y. S. Jung, Adv. Energy Mater., 2018, 8, 1800035 CrossRef.
  14. F. Wu, Y. Zheng, L. Li, G. Tan, R. Chen and S. Chen, J. Phys. Chem. C, 2013, 117, 19280–19287 CAS.
  15. M. M. Mahmoud, Y. Cui, M. Rohde, C. Ziebert, G. Link and H. J. Seifert, Materials, 2016, 9, 506 CrossRef PubMed.
  16. M. Gellert, K. I. Gries, C. Yada, F. Rosciano, K. Volz and B. Roling, J. Phys. Chem. C, 2012, 116, 22675–22678 CrossRef CAS.
  17. V. Thangadurai, H. Kaack and W. J. F. Weppner, J. Am. Ceram. Soc., 2003, 86, 437–440 CrossRef CAS.
  18. V. Thangadurai and W. Weppner, Adv. Funct. Mater., 2005, 15, 107–112 CrossRef CAS.
  19. R. Murugan, V. Thangadurai and W. Weppner, Angew. Chem., Int. Ed., 2007, 46, 7778–7781 CrossRef CAS PubMed.
  20. H. Buschmann, J. Dolle, S. Berendts, A. Kuhn, P. Bottke, M. Wilkening, P. Heitjans, A. Senyshyn, H. Ehrenberg, A. Lotnyk, V. Duppel, L. Kienle and J. Janek, Phys. Chem. Chem. Phys., 2011, 13, 19378–19392 RSC.
  21. N. C. Rosero-Navarro, T. Yamashita, A. Miura, M. Higuchi, K. Tadanaga and J. W. Stevenson, J. Am. Ceram. Soc., 2017, 100, 276–285 CrossRef CAS.
  22. Y. Y. Sun, M. L. Agiorgousis, P. Zhang and S. Zhang, Nano Lett., 2015, 15, 581–585 CrossRef CAS PubMed.
  23. K. Takada, N. Ohta and Y. Tateyama, J. Inorg. Organomet. Polym., 2014, 25, 205–213 CrossRef.
  24. R. C. Xu, X. H. Xia, Z. J. Yao, X. L. Wang, C. D. Gu and J. P. Tu, Electrochim. Acta, 2016, 219, 235–240 CrossRef CAS.
  25. R. P. Iczkowski and J. L. Margrave, J. Am. Chem. Soc., 1961, 83, 3547–3551 CrossRef CAS.
  26. Y. Wang, W. D. Richards, S. P. Ong, L. J. Miara, J. C. Kim, Y. Mo and G. Ceder, Nat. Mater., 2015, 14, 1026–1031 CrossRef CAS PubMed.
  27. K. Aso, A. Hayashi and M. Tatsumisago, Inorg. Chem., 2011, 50, 10820–10824 CrossRef CAS PubMed.
  28. A. Hayashi, S. Hama, H. Morimoto, M. Tatsumisago and T. Minami, J. Am. Ceram. Soc., 2001, 84, 477–479 CrossRef CAS.
  29. Y. Onodera, K. Mori, T. Otomo, A. C. Hannon, M. Sugiyama and T. Fukunaga, IOP Conf. Ser.: Mater. Sci. Eng., 2011, 18, 022012 Search PubMed.
  30. K. Minami, A. Hayashi, S. Ujiie and M. Tatsumisago, Solid State Ionics, 2011, 192, 122–125 CrossRef CAS.
  31. K. Minami, A. Hayashi, S. Ujiie and M. Tatsumisago, J. Power Sources, 2009, 189, 651–654 CrossRef CAS.
  32. R. Kanno and M. Murayama, J. Electrochem. Soc., 2001, 148, A742–A746 CrossRef CAS.
  33. M. Nagao, H. Kitaura, A. Hayashi and M. Tatsumisago, J. Power Sources, 2009, 189, 672–675 CrossRef CAS.
  34. X. Yao, D. Liu, C. Wang, P. Long, G. Peng, Y. S. Hu, H. Li, L. Chen and X. Xu, Nano Lett., 2016, 16, 7148–7154 CrossRef CAS PubMed.
  35. K. Xiong, R. C. Longo, S. Kc, W. Wang and K. Cho, Comput. Mater. Sci., 2014, 90, 44–49 CrossRef CAS.
  36. A. Hayashi and M. Tatsumisago, Electron. Mater. Lett., 2012, 8, 199–207 CrossRef CAS.
  37. M. Tatsumisago, M. Nagao and A. Hayashi, J. Asian Ceram. Soc., 2018, 1, 17–25 CrossRef.
  38. L. F. Wan, J. T. Incorvati, K. R. Poeppelmeier and D. Prendergast, Chem. Mater., 2016, 28, 6900–6908 CrossRef CAS.
  39. X. Wang, R. Xiao, H. Li and L. Chen, Phys. Rev. Lett., 2017, 118, 195901 CrossRef PubMed.
  40. K.-H. Kim and S. W. Martin, Chem. Mater., 2019, 31, 3984–3991 CrossRef CAS.
  41. G. Kresse and D. Joubert, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 1758–1775 CrossRef CAS.
  42. R. Car and M. Parrinello, Phys. Rev. Lett., 1985, 55, 2471–2474 CrossRef CAS PubMed.
  43. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865–3868 CrossRef CAS.
  44. P. E. Blöchl, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 50, 17953–17979 CrossRef PubMed.
  45. H. Jiang and D. A. Stewart, ACS Appl. Mater. Interfaces, 2017, 9, 16296–16304 CrossRef CAS PubMed.
  46. Z. Zhu, I.-H. Chu, Z. Deng and S. P. Ong, Chem. Mater., 2015, 27, 8318–8325 CrossRef CAS.
  47. H. Xu, D. Lee, J. He, S. B. Sinnott, V. Gopalan, V. Dierolf and S. R. Phillpot, Phys. Rev. B: Condens. Matter Mater. Phys., 2008, 78, 174103 CrossRef.
  48. J. Buckeridge, D. O. Scanlon, A. Walsh and C. R. A. Catlow, Comput. Phys. Commun., 2014, 185, 330–338 CrossRef CAS.
  49. F. H. Taylor, J. Buckeridge and C. R. A. Catlow, Chem. Mater., 2016, 28, 8210–8220 CrossRef CAS.
  50. D. O. Scanlon, J. Buckeridge, C. R. A. Catlow and G. W. Watson, J. Mater. Chem. C, 2014, 2, 3429–3438 RSC.
  51. G. Henkelman, B. P. Uberuaga and H. Jónsson, J. Chem. Phys., 2000, 113, 9901–9904 CrossRef CAS.
  52. Z. Yang, S. Zhao, Y. Pan, X. Wang, H. Liu, Q. Wang, Z. Zhang, B. Deng, C. Guo and X. Shi, ACS Appl. Mater. Interfaces, 2018, 10, 3142–3151 CrossRef CAS PubMed.
  53. S. Shi, J. Gao, Y. Liu, Y. Zhao, Q. Wu, W. Ju, C. Ouyang and R. Xiao, Chin. Phys. B, 2016, 25, 018212 CrossRef.
  54. I. H. Chu, H. Nguyen, S. Hy, Y. C. Lin, Z. Wang, Z. Xu, Z. Deng, Y. S. Meng and S. P. Ong, ACS Appl. Mater. Interfaces, 2016, 8, 7843–7853 CrossRef CAS PubMed.
  55. Z. Deng, B. Radhakrishnan and S. P. Ong, Chem. Mater., 2015, 27, 3749–3755 CrossRef CAS.
  56. Z. Deng, Z. Zhu, I.-H. Chu and S. P. Ong, Chem. Mater., 2016, 29, 281–288 CrossRef.
  57. M. Murakami, K. Shimoda, S. Shiotani, A. Mitsui, K. Ohara, Y. Onodera, H. Arai, Y. Uchimoto and Z. Ogumi, J. Phys. Chem. C, 2015, 119, 24248–24254 CrossRef CAS.
  58. H. Yamane, M. Shibata, Y. Shimane, T. Junke, Y. Seino, S. Adams, K. Minami, A. Hayashi and M. Tatsumisago, Solid State Ionics, 2007, 178, 1163–1167 CrossRef CAS.
  59. R. T. Sanderson, J. Am. Chem. Soc., 1983, 105, 2259–2261 CrossRef CAS.
  60. B. Huang, X. Yao, Z. Huang, Y. Guan, Y. Jin and X. Xu, J. Power Sources, 2015, 284, 206–211 CrossRef CAS.
  61. T. F. Willems, C. H. Rycroft, M. Kazi, J. C. Meza and M. Haranczyk, Microporous Mesoporous Mater., 2012, 149, 134–141 CrossRef CAS.
  62. R. L. Martin, B. Smit and M. Haranczyk, J. Chem. Inf. Model., 2012, 52, 308–318 CrossRef CAS PubMed.
  63. J. Kang, H. Chung, C. Doh, B. Kang and B. Han, J. Power Sources, 2015, 293, 11–16 CrossRef CAS.
  64. S. Wenzel, D. A. Weber, T. Leichtweiss, M. R. Busche, J. Sann and J. Janek, Solid State Ionics, 2016, 286, 24–33 CrossRef CAS.

Footnote

Electronic supplementary information (ESI) available. See DOI: 10.1039/c9cp05329h

This journal is © the Owner Societies 2019