Tran Viet Toanab,
Luu Tuan Anhab,
Nguyen Thi Minh Nguyetbc,
Tran Anh Tuabd and
Nguyen Huu Huy Phuc*abd
aFaculty of Materials Technology, Ho Chi Minh City University of Technology (HCMUT), 268 Ly Thuong Kiet Str., Dist. 10, Ho Chi Minh City, Vietnam. E-mail: nhhphuc@hcmut.edu.vn
bVietnam National University Ho Chi Minh City, Linh Trung Ward, Thu Duc Dist., Ho Chi Minh City, Vietnam
cVNU-HCM Key Laboratory for Material Technologies, Ho Chi Minh City University of Technology (HCMUT), 268 Ly Thuong Kiet Str., Dist. 10, Ho Chi Minh City, Vietnam
dNational Key Laboratory of Polymer and Composite Materials – Ho Chi Minh City, 268 Ly Thuong Kiet, District 10, Ho Chi Minh City, Vietnam
First published on 21st May 2024
Argyrodite-type solid electrolytes of Li6PS5Cl doped with multivalent cations (Mg2+, Ba2+, Zn2+, Al3+, Y3+) were prepared via a mechanochemical synthesis method. The lattice constant (a0), interplanar spacing (d220, d311, d222), and micro-strain (ε) showed that the doping elements were incorporated into the crystal structure of Li6PS5Cl. The lattice constant and interplanar spacing of the doped samples were smaller than those of Li6PS5Cl. The prepared samples exhibited a positive lattice strain, and the substituted samples exhibited higher strains than Li6PS5Cl. The doped samples exhibited higher ionic conductivity than Li6PS5Cl at 25 °C. Li5.94Al0.02PS5Cl exhibited the highest σDC of approximately 2.36 × 10−3 S cm−1 at 25 °C. The charge carrier movement at the grain boundary changing from long-range diffusion in Li6PS5Cl to short-range diffusion in Li5.94Al0.02PS5Cl enhanced the conductivity.
The ionic conductivity of Li6PS5X (X = Cl, Br, and I) at room temperature ranges from 10−6 to approximately 1–2 × 10−3 S cm−1 and could be improved by the aliovalent substitution of S2−, P5+, and Li+. Li5.5PS4.5Cl1.5 exhibited a high conductivity of 9.4 × 10−3 S cm−1 at 25 °C,7 whereas Li5.3PS4.3Cl1.0Br0.7 exhibited an ionic conductivity of 16.6 × 10−3 S cm−1 at 30 °C.8 High-entropy multicationic substituted Li6.5[P0.25Si0.25Ge0.25Sb0.25]S5I exhibited a high ionic conductivity of approximately 13 × 10−3 S cm−1 at room temperature,9 and Li6.2Si0.2P0.8S5Cl0.5Br0.5 exhibited a high ionic conductivity of 5.12 × 10−3 S cm−1 at room temperature.10 The aliovalent substitution of Li+ by a multivalent cation improved the ionic conductivity of Li6PS5Cl. The ionic conductivity of Li5.7Ca0.15PS5Cl at 25 °C was approximately 5.2 × 10−3 S cm−1, which exceeded that of Li6PS5Cl (3.1 × 10−3 S cm−1).11 Li5.4Al0.2PS5Br exhibited a room temperature ionic conductivity of 2.4 × 10−3 S cm−1, which exceeded that of Li6PS5Br (1.0 × 10−3 S cm−1).12 These SEs were prepared through solid-state reactions at a high temperature. Therefore, maintaining low oxygen and humidity concentrations will be a barrier to the mass production of these substances because sulfide-based SEs react with oxygen in a dry atmosphere at approximately 270 °C.13 Thus, mechanochemical synthesis, which occurs at room temperature, is a good option, in addition to solid-state reactions at high temperatures. The ionic conductivity of mechanochemically synthesized Li6PS5Cl at room temperature was slightly enhanced because of a multivalent cation at the grain boundary.14
This study enhanced the ionic conductivity of mechanochemically synthesized Li6PS5Cl by the aliovalent substitution of Li+ with multivalent cations (Mg2+, Ba2+, Zn2+, Al3+, Y3+). Data obtained from alternating current (AC) impedance spectroscopy was analyzed using conductivity isotherms and the dielectric constant and dielectric loss. The ionic conductivity of Li5.94Al0.02PS5Cl was approximately 2.36 × 10−3 S cm−1 at 25 °C, which was approximately twice that of Li6PS5Cl. Furthermore, this value exceeded the reported ionic conductivity of argyrodite SEs prepared without heat treatment. The activation energy of direct current and Li ion migration suggested that ion movement at the grain boundary was a critical process in the prepared samples.
The structure of the samples was characterized by X-ray diffraction (XRD; X8, Bruker), SEM (S4800, Hitachi) and EDS (ULTIM MAX, Oxford Instrument). The samples were prepared in an Ar-filled glove box and loaded into an air-tight sample holder for characterization.
The samples for resistivity measurements were prepared by uniaxially cold pressing the powder under a pressure of 330 MPa to form a pellet (thickness = 1.2–1.4 mm; diameter = 10 mm), as reported.15 AC impedance spectroscopy was conducted using a potentiostat (PGSTAT302N, Autolab, Herisau, Switzerland) from 9 MHz to 10 Hz. The samples for the impedance measurements were prepared by uniaxially pressing the sample (approximately 160 mg) into pellets (approximately 10.0 mm in diameter) under a pressure of 330 MPa at room temperature. The pellet was placed in a holder made of polycarbonate with two stainless steel rods as blocking electrodes. Thereafter, the cell was placed in an N2 flow in a glass tube for temperature dependence measurements. The temperature was gradually increased from room temperature to 110 °C and held at each temperature for 1 h prior to the impedance measurements. The AC applied voltage was 100 mV.
nλ = 2dsinθ. | (1) |
(2) |
The Nelson–Riley equation was employed to determine the lattice constant (a0).
(3) |
The a0 was determined from the linear fit equation, which was derived from a plot with a as the y-axis and as the x-axis.
The Williamson–Hall (WH) analysis is a simple method for estimating the lattice strain (ε) by analyzing the X-ray data and considering the peak width as a function of 2θ.
(4) |
Table 1 lists the values of the lattice constant (a0), interplanar spacing (d220, d311, d222), and micro-strain (ε). The a0 of Li6PS5Cl was 9.8401 Å and was consistent with the reported 9.8397(4) Å of Li6PS5Cl prepared via the mechanochemical method.16 The a0 of Li5.99Ca0.005PS5Cl was practically similar to that of Li6PS5Cl, indicating that the doping amount was too small to influence the measurement result. The a0 of Li5.94Ca0.03PS5Cl was 9.8346 Å, which was smaller than that of Li6PS5Cl. The a0 values of Li5.94M0.06/nnPS5Cl (Mn = Mg2+, Ca2+, Ba2+, Zn2+, Al3+, Y3+) were close to each other. Thus, the multivalent cation doping reduced the a0 of Li6PS5Cl; this observation was consistent with reported results.11,12 The values of the interplanar spacing (d220, d311, d222) of the substituted samples were smaller than those of Li6PS5Cl. In crystals, cations are surrounded by anions and vice versa such that the electrostatic interaction among oppositely charged ions strengthens the crystal structure. The substitution of Li ions with multivalent cations resulted in the formation of vacancies. A multivalent cation has a higher positive charge density than a Li ion; therefore, the electrostatic attraction with negative ions will be stronger. Thus, the lattice constant and interplanar spacing will be reduced. The lattice strain (ε) represents the displacement of unit cells about their normal positions. All the prepared samples exhibited a positive lattice strain, and the substituted samples exhibited higher strains than Li6PS5Cl. SEM-EDS results of Li5.8Ca0.1PS5Cl is shown in Fig. 1c. The SEs are in the form of particles with a size of several hundred nanometer. EDS results indicated that Ca was well dispersed in the prepared powder sample. The results suggest that the multivalent cations were successfully incorporated into the crystal structure of Li6PS5Cl.
Lattice constant a0 (Å) | d220 (Å) | d311 (Å) | d222 (Å) | Microstrain ε | |
---|---|---|---|---|---|
Li6PS5Cl | 9.8401 | 3.4790 | 2.9669 | 2.8406 | 0.00330 |
Li5.99Ca0.005PS5Cl | 9.8400 | 3.4790 | 2.9669 | 2.8406 | 0.00418 |
Li5.94Ca0.03PS5Cl | 9.8346 | 3.4781 | 2.9661 | 2.8399 | 0.00433 |
Li5.8Ca0.1PS5Cl | 9.8341 | 3.4779 | 2.9660 | 2.8397 | 0.00390 |
Li5.94Ba0.03PS5Cl | 9.8347 | 3.4771 | 2.9653 | 2.8390 | 0.00460 |
Li5.94Zn0.03PS5Cl | 9.8348 | 3.4771 | 2.9653 | 2.8391 | 0.00458 |
Li5.94Mg0.03PS5Cl | 9.8356 | 3.4774 | 2.9655 | 2.8393 | 0.00488 |
Li5.94Al0.02PS5Cl | 9.8354 | 3.4778 | 2.9667 | 2.8404 | 0.00418 |
Li5.94Y0.02PS5Cl | 9.8349 | 3.4772 | 2.9653 | 2.8391 | 0.00445 |
Fig. 2a–d show the frequency dependence of the real part of conductivity, σ′, of Li6PS5Cl, Li5.94Ca0.03PS5Cl, Li5.94Mg0.03PS5Cl, and Li5.94Al0.02PS5Cl obtained from 10 Hz to 9 MHz at different temperatures, respectively, using conductivity isotherms. Fig. S1a–c† show the conductivity isotherms of Li5.96Ca0.002PS5Cl, Li5.9Ca0.05PS5Cl, and Li5.8Ca0.1PS5Cl, respectively. Furthermore, Fig. S2a–c† show the conductivity isotherms of Li5.94Ba0.03PS5Cl, Li5.94Zn0.03PS5Cl, and Li5.94Y0.02PS5Cl, respectively. The isotherms of all the samples comprised three regions: electrode polarization, a plateau-like region, and polarization conductivity at low, intermediate, and high frequencies, respectively. The high-frequency polarization conductivity could be explained by the power law behavior, σω ∝ ωn, where n is a fractional exponent (0 ≤ n ≤ 1) and is associated with the interaction between the ions and environment.17 The accumulation of ions at blocking electrodes caused electrode polarization.18 The isotherms were analyzed using the Jonscher power law to understand the ion dynamics of the prepared electrolytes. The conductivity in the intermediate- and high-frequency regions followed the Jonscher power law equation.
σ = σDC + Aωn, | (5) |
→ σ − σDC = Aωn → log10(σ − σDC) = nlog10ω + log10A, | (6) |
Fig. 2 Conductivity isotherms of (a) Li6PS5Cl; (b) Li5.94Ca0.03PS5Cl; (c) Li5.94Mg0.03PS5Cl and (d) Li5.94Al0.02PS5Cl measured from 10 Hz to 9 MHz. |
Fig. 3a and c show the temperature dependence of the ionic conductivity, σDC, of the SEs of Li6−2xCaxPS5Cl and Li5.94M0.06/nnPS5Cl (Mn = Mg2+, Ba2+, Zn2+, Al3+, Y3+), respectively. The log10(σDC) satisfied a practically linear dependence on an inverse temperature; therefore, it followed the Arrhenius equation, σ = σ0exp(−Ea,DC/(kBT)). The DC activation energy, Ea,DC, was calculated and is shown in Table 2. The Ea,DC of Li6PS5Cl was approximately 16 kJ mol−1. The Ea,DC of the doped samples exceeded that of Li6PS5Cl in the range of 28–31 kJ mol−1. Furthermore, the Ea,DC of Li6−3xAlxPS5Br (x = 0.1, 0.15, 0.2, 0.25, 0.3) has been reported to exceed that of Li6PS5Br.12 Thus, the aliovalent substitution of Li ions in argyrodite-type SEs led to an increase in Ea,DC. The ionic conductivity of Li6PS5Cl at 25 °C was approximately 1.25 × 10−3 S cm−1. Fig. 3b and d show the ionic conductivity (σDC) of the SEs of Li6−2xCaxPS5Cl and Li5.94M0.06/nnPS5Cl (Mn = Mg2+, Ba2+, Zn2+, Al3+, Y3+) at 25 °C and 50 °C, respectively. At 25 °C, all the doped samples exhibited ionic conductivities higher than that of Li6PS5Cl. Li5.94Al0.02PS5Cl exhibited the highest σDC (approximately 2.36 × 10−3 S cm−1) at 25 °C. The σDC of Li6PS5Cl and Li5.94Al0.02PS5Cl were 2.15 × 10−3 and 6.00 × 10−3 S cm−1 at 50 °C, respectively. The σDC and Ea,DC results confirmed that the multivalent cation was successfully incorporated into the crystal structure of Li6PS5Cl.
Li6PS5Cl | Li5.96Ca0.002PS5Cl | Li5.94Ca0.03PS5Cl | Li5.9Ca0.05PS5Cl | Li5.8Ca0.1PS5Cl | |
---|---|---|---|---|---|
Ea,DC/kJ mol−1 | 16 | 26 | 29 | 28 | 31 |
Ea,DC/eV | 0.16 | 0.26 | 0.29 | 0.28 | 0.31 |
Ea,m/kJ mol−1 | 19 | 24 | 28 | 31 | 34 |
Ea,m/eV | 0.19 | 0.24 | 0.28 | 0.31 | 0.34 |
τ0,m/s | 5.47 × 10−7 | 2.56 × 10−9 | 6.31 × 10−9 | 1.57 × 10−9 | 6.79 × 10−9 |
Li5.94Ba0.03PS5Cl | Li5.94Zn0.03PS5Cl | Li5.94Mg0.03PS5Cl | Li5.94Al0.02PS5Cl | Li5.94Y0.02PS5Cl | |
---|---|---|---|---|---|
Ea,DC/kJ mol−1 | 27 | 28 | 29 | 29 | 29 |
Ea,DC/eV | 0.27 | 0.28 | 0.29 | 0.29 | 0.29 |
Ea,m/kJ mol−1 | 31 | 31 | 31 | 33 | 30 |
Ea,m/eV | 0.31 | 0.31 | 0.31 | 0.33 | 0.30 |
τ0,m/s | 8.66 × 10−9 | 1.39 × 10−9 | 4.42 × 10−9 | 2.26 × 10−9 | 1.64 × 10−9 |
Fig. 4a–d show the frequency dependence of the real part of permittivity, ε′, of Li6PS5Cl, Li5.94Ca0.03PS5Cl, Li5.94Mg0.03PS5Cl, and Li5.94Al0.02PS5Cl obtained from 10 Hz to 9 MHz at different temperatures, respectively. Fig. S3a–c† show the frequency dependence of the ε′ of Li5.96Ca0.002PS5Cl, Li5.9Ca0.05PS5Cl, and Li5.8Ca0.1PS5Cl, respectively. Furthermore, Fig. S4a–c† show the frequency dependence of the ε′ of Li5.94Ba0.03PS5Cl, Li5.94Zn0.03PS5Cl, and Li5.94Y0.02PS5Cl, respectively. The increase in the plots in the low-frequency region was attributable to the electrode–electrolyte interface polarization due to the accumulation of ions near the electrode. This led to the formation of a space-charged layer that blocked the electric field and enhanced the electrical polarization. The ε′ of all the samples increased with an increase in temperature, indicating that charge carrier movement was thermally activated. The ε′ reflects the amount of energy stored in the form of polarization when an electric field is applied.20 In most ion-conducting materials, ε′ decreases with an increase in frequency.21 The plot of Li5.94Mg0.03PS5Cl at room temperature continuously decreased in the intermediate- and high-frequency regions; however, the plots of the other samples obtained at room temperature exhibited maxima at 105–106 Hz. Maxima were observed in all the plots at 50 °C or above. The change in the shape of the plots of Li5.94Mg0.03PS5Cl suggests that a change occurred in the microstructure, and this process was temperature-dependent.
Fig. 4 Frequency dependent of the real part of permittivity, ε′, of (a) Li6PS5Cl; (b) Li5.94Ca0.03PS5Cl; (c) Li5.94Mg0.03PS5Cl and (d) Li5.94Al0.02PS5Cl measured from 10 Hz to 9 MHz. |
Fig. 5a–d show the frequency dependence of the loss factor, tanδ, of Li6PS5Cl, Li5.94Ca0.03PS5Cl, Li5.94Mg0.03PS5Cl, and Li5.94Al0.02PS5Cl obtained from 10 Hz to 9 MHz at different temperatures, respectively. Fig. S5a–c† show the frequency dependence of the tanδ of Li5.96Ca0.002PS5Cl, Li5.9Ca0.05PS5Cl, and Li5.8Ca0.1PS5Cl, respectively. Fig. S6a–c† show the frequency dependence of the tanδ of Li5.94Ba0.03PS5Cl, Li5.94Zn0.03PS5Cl, and Li5.94Y0.02PS5Cl, respectively. Two peaks assignable to ion migration at the grain boundary and bulk were observed in all the plots in the low- and high-frequency regions. The peak in the low-frequency region shifted toward high-frequency with an increase in temperature. Generally, the grain boundary migration resistivity was expected to decrease with an increase in temperature because Li+ diffused from the bulk to the grain boundary. Thus, the intensity of the peak corresponding to the grain boundary resistivity in the loss factor decreased with an increase in temperature. The peak in the high-frequency region was not fully observed at 50 °C or above. The maximum value of the peak in the high-frequency region increased with an increase in the temperature, indicating that the number of charge carriers increased because of thermal activation. The migration energy (Ea,m) and migration characteristic time (τ0,m) of the Li+ moving at the grain boundary could be derived from the temperature dependence of the peak position at low-frequency in the tanδ using the Arrhenius equation, τm = τ0,mexp(−Ea,m/(kBT)). Table 2 lists the obtained Ea,m and τ0,m values. The Ea,m of the samples was similar to the Ea,DC, implying that the ion migration at the grain boundary was a critical process. The τ0,m of Li6PS5Cl was approximately 5.47 × 10−7 s. The τ0,m of the doped samples was approximately 10−9 s, which was almost 102 times faster than that of Li6PS5Cl. Thus, the Li ion moving at the grain boundary changed from long-range diffusion in Li6PS5Cl to short-range diffusion in the doped samples. The results suggested that the substitution of the Li ion with multivalent cations resulted in vacancy formation, which was the new hopping position for the Li ion. The addition of multivalent ions to Li6PS5Cl enhanced the Li ion mobility with a gradual decrease in the migration time; however, the migration energy increased as reflected in the Ea,m. Multivalent cations have a higher positive charge than Li+; thus, Li ions are repelled from their vicinity. From there, an Li ion can be trapped near the multivalent ion site, leading to a high migration barrier. It has been reported that the short inter-cage jump is a critical process in Li5.4Al0.2PS5Br, and this process was associated with an increase in activation energy.12 In addition, the change from short- to long-range diffusion was associated with a decrease and increase in activation energy and migration time, respectively.22 Thus, the results in this section were consistent with the reported results and ionic conductivity of the samples.
Fig. 5 Frequency dependent of the loss factor, tanδ, of (a) Li6PS5Cl; (b) Li5.94Ca0.03PS5Cl; (c) Li5.94Mg0.03PS5Cl and (d) Li5.94Al0.02PS5Cl measured from 10 Hz to 9 MHz. |
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4ra02621g |
This journal is © The Royal Society of Chemistry 2024 |