Effect of Sm³⁺ Substitutions on the Lithium Ionic Conduction and Relaxation Dynamics of Li_5+2xLa₃Nb_2−xSm_xO₁₂ Ceramics

Abstract

In the present work, we studied the effects of substitutional Sm³⁺ ions on the ionic conduction properties of Li_5+2xLa₃Nb_2−xSmxO₁₂ (LLN-Sm) ceramics with x = 0.0—0.6. The investigated final ceramics, prepared by solid state reaction, were sintered at 1000 °C for 12 h. XRD investigations showed the formation of the cubic garnet phase for all of the studied samples. The ionic conductivity was found to increase with Sm³⁺ content, with the highest value of 7.04 × 10⁻⁵ S/cm for the Li_5+2xLa₃Nb_2−xSm_xO₁₂ sample compared to 7.49 × 10⁻⁶ S/cm for the pure LLN sample, both at RT. Lithium ion mobilities of LLN-Sm garnets at different temperatures were estimated. Considerable enhancement of mobility, the main factor leading to ionic conductivity improvement, was obtained for samples with Sm³⁺ substitutions. Relaxation processes were studied by the electric modulus, and the corresponding activation energy was found to be very similar to the ionic conduction process.

1. Introduction

Rechargeable lithium ion batteries are major energy reservoirs/depots in our daily life, where they are used in various electronic devices. However, the current batteries suffer safety and capacity loss issues due to the used liquid electrolytes [1,2,3]. These drawbacks could be eliminated by using solid lithium electrolytes as an alternative. However, the necessary range of the ionic conductivity for practical applications, 10⁻³—10⁻² S/cm, is difficult to achieve in most solid lithium ionic conductors. This is one of the most challenging hurdles in the development of all solid-state batteries [1,2,3]. Therefore, different inorganic crystalline and glassy lithium ion conducting materials, such as Li₇P₃S₁₁ [4], Li₁₀GeP₂S₁₂ [5], Li₂S-B₂S₃ [6], Li_1+xAl_xGe_2−x(PO₄)₃ [7], Li_1+xAl_xTi_2−x(PO₄)₃ [8,9] and Li_3xLa_2/3−3xTiO₃ [10], have been investigated. Among these materials, lithium-conducting sulfides show high ionic conductivity in the 10⁻³—10⁻² S/cm range at RT, but they require handling under inert gas conditions due to their sensitivity to moisture [4,5,6]. On the other hand, lithium ion conductors based on phosphate or oxide ceramics exhibit high ionic conductivity in the 10⁻⁵—10⁻³ S/cm range and are mostly stable in normal atmosphere. They could be mass produced using simple methods, such as solid state reaction and sol–gel techniques [7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22]. A detailed review on various crystalline and glassy solid lithium ion conductors was recently published by Campanella et al. [23]

Garnet lithium ion conductors Li₅La₃M₂O₁₂, Li₆ALa₂M₂O₁₂, and Li₇La₃Zr₂O₁₂ (M = Ta and Nb, A = alkaline earths) and their derivative compounds are expected to be good electrolytes in solid-state batteries [11,12,13,14,15,16,17,18,19,20,21,22]. The ionic conductivity value of these materials and their chemically modified derivatives is in the 10⁻⁶—10⁻⁴ S/cm range at RT. Specific conductivity values depend on the crystal structure, the composition, and the lithium content and are characterized by a small conductivity contribution from the grain boundary [11,12,13,14,15,16,17,18,19,20,21,22]. These characteristics led to extensive research work in the past decade in order to explore the ionic conduction mechanism and improve the ionic conductivity of these materials. It is now established that chemical substitutions of divalent and/or trivalent cations on the La³⁺ and M⁵⁺ sites, respectively, lead to increased lithium content and, subsequently, increased conductivity. In the past few years, further research interest was devoted to the Li₇La₃Zr₂O₁₂ compound, as it shows the highest ionic conductivity of garnet-like materials [17]. This material could crystallize in the tetragonal or cubic phases, where the cubic phase shows a higher conductivity of 3 × 10⁻⁴ S/cm at RT [17]. The high conductive cubic phase of Li₇La₃Zr₂O₁₂ could be stabilized by appropriate chemical doping, such as Ga³⁺ or Al³⁺ on Li⁺ sites or by Nb⁵⁺ or Ta⁵⁺ on Zr⁴⁺ sites [24,25]. A comprehensive review on garnet-type electrolytes was recently published [26]. Although most research work on lithium garnet materials is directed towards Li₇La₃Zr₂O₁₂, high ionic conductivity values in the 10⁻⁴ S/cm range could also be obtained in doped Li₅La₃Nb₂O₁₂ (LLN) and Li₅La₃Ta₂O₁₂ (LLT) materials [21,27,28,29,30]. In our recent work on Gd³⁺−substituted Li_5+2xLa₃Nb_2−xGd_xO₁₂, the ionic conductivity was found to increase to 1.12 × 10⁻⁴ S/cm for x = 0.5, which is about two orders of magnitude higher than the un-doped LLN material [27]. Similarly, Sc³⁺ doping in Li_5+2xLa₃Nb_2−xSc_xO₁₂ garnets leads to ionic conductivity of 3.7 × 10⁻⁴ S/cm for x = 0.625 [28]. Moreover, recent Pr doping in Li_5+xLa₃Nb_{2 − x}Pr_xO₁₂ samples showed a high conductivity value of 5.6 × 10⁻⁴ S/cm for x = 0.8 composition [29]. These results suggest that the effect of doping in LLN and LLT garnet materials is promising, and further studies are necessary. In this work, we study the effect of Nb⁵⁺ site substitutions by Sm⁺³ on the ionic conduction of Li_5+2xLa₃Nb_2−xSm_xO₁₂. Sm⁺³ substitution is expected to increase lithium content in the materials, leading to enhanced ionic conductivity.

2. Materials and Methods

Li_5+2xLa₃Nb_2−xSm_xO₁₂ samples (with x = 0.25, 0.4, 0.5, 0.6; hereafter abbreviated LLN-Sm25, LLN-Sm40, LLN-Sm50, and LLN-Sm60, respectively) were synthesized by mechanical milling and solid-state reaction techniques. High purity Li₂CO₃, La₂O₃, Nb₂O₃ and Sm₂O₃ were used with 10 wt% excess of Li₂CO₃ added for compensation of lithium loss at high temperatures. The mixed powder underwent three cycles of alternate ball milling and calcination at different periods and temperatures. It was first ball milled using tungsten carbide pots and balls for 12 h with a 250 rpm rotation speed. The acquired powder was calcined at 700 °C for 12 h, then ball milled again under the same conditions. A second calcination step was performed at 900 °C for 12 h followed by final ball milling at 400 rpm for 3 h. The obtained powder was pressed under 2.5 ton in a 10 mm diameter die to obtain pellets with a 1–2 mm thickness. The sintering of the samples was performed at 1000 °C for 12 h in alumina crucibles with pellets impeded in the mother powder to limit the Li₂O loss. Structural characterization of the LLN-Sm samples was studied by XRD measurements in the 0 ≤ 2θ ≤ 80 range using monochromated radiation (λ = 1.5406 Å). The electrical properties of the LLN-Sm samples were studied by impedance spectroscopy at different temperatures under flowing nitrogen gas using the Novocontrol concept 50 system.

3. Results

The XRD patterns of the LLN-Sm samples are shown in Figure 1. The observed patterns in Figure 1 agree with the standard XRD database of Li₅La₃Nb₂O₁₂ (PDF#45-0109), indicating the formation of the cubic garnet structure. Minor peaks observed in the XRD patterns of the x = 0.6 composition are attributed to impurity. Nyquist impedance plots of the LLN-Sm25 ceramics are shown in Figure 2 at selected temperatures as a representative example. A single semicircle with a decreasing radius as the temperature increases was observed. The estimated resistance from the intercept of the semicircle with the Z’ axis is assigned to the total resistance of the materials, where it was not possible to separate grain and grain boundary contributions in the current impedance data.

The variation of the total lithium ionic conductivity of the LLN-Sm ceramics with temperature is depicted in Figure 3, and the conductivity is fitted by the Arrhenius relation:

$${\sigma }_{dc}= {\sigma }_o exp \left(- \frac{\Delta E}{k T}\right).$$

(1)

In

Table 1. The dc conductivity values for Li_5+2xLa₃Nb_2−xSm_xO₁₂ garnets at 300 K. ΔE_σdc, ΔE_μ and $\Delta E_m$ are the activation energies for conduction, mobility and relaxation process, respectively.

X	$${\text{σ}}_{\mathrm{dc}}(\mathbf{s}/\mathbf{cm})$$ at 300 K	ΔEσdc(eV)	ΔEμ(eV)	$$\mathbf{\Delta }{\mathbf{E}}_{\mathbf{m}}\left(\mathbf{eV}\right)$$
0.00	7.49 × 10−6	0.60	0.52	0.58
0.25	1.35 × 10−5	0.51	0.39	0.49
0.40	7.04 × 10−5	0.47	0.40	0.47
0.50	4.31 × 10−5	0.47	0.35	0.47
0.60	3.04 × 10−5	0.47	0.34	0.47

, we summarize the values of the ionic conductivity at 300 K and the corresponding activation energy ∆E of the LLN-Sm ceramics. It is observed from the data in Figure 3 and Table 1 that the ionic conductivity of the LLN-Sm ceramics increases with Sm³⁺ doping. The composition with x = 0.4, i.e., Li_5.6La₃Nb_1.6Sm_0.4O₁₂, has the highest conductivity of 7.04 × 10⁻⁵ S/cm compared to 7.26 × 10⁻⁶ S/cm for the un-doped LLN. These results indicate that considerable lithium conductivity enhancement is achieved by Sm³⁺ substitutions in LLN garnets. Pinzaru and Thangadurai previously studied the ionic conduction in Li_5+2xLa₃Nb_2−xSm_xO₁₂ garnets with 0 ≤ x ≤ 0.7, and the highest conductivity of 5.84 × 10^-5 S/cm was obtained for x = 0.3 composition at RT [30]. This is slightly lower than the highest conductivity obtained in this study, despite the fact that they sintered the samples at a higher temperature of 1100 °C. The activation energy ∆E has a value of 0.47 eV for the LLN-Sm40, LLN-Sm50, and LLN-Sm60 samples, whereas a slightly larger value of 0.51 eV is obtained for the LLN-Sm25 sample. These values are noticeably smaller than that of 0.60 eV for the pure LLN garnets. The enhanced lithium ionic conductivity in Sm⁺³-substituted LLN ceramics may be attributed to increasing either the concentration or the mobility of the mobile Li⁺ ions.

It is expected that the substitution of trivalent Sm³⁺ cations on the pentavalent Nb⁵⁺ sites in Li_5+2xLa₃Nb_2−xSm_xO₁₂ will lead to an increase in Li⁺ content per unit formula from 5Li in pure LLN to (5 + 2x)Li for x content of the Sm³⁺-substituted LLN-Sm garnets. This increase in Li⁺ ions content may lead to increased conductivity of LLN-Sm materials. Another factor that affects the ionic conductivity is the mobility of Li⁺ ions. Sm³⁺ substitutions on Nb⁵⁺ sites may lead to the creation of more vacant sites available for the mobile Li⁺ ions, thereby leading to enhanced mobility and conductivity. Here, from the analysis of the conductivity spectra, we estimated the mobility values of mobile Li⁺ ions at different temperatures in the LLN-Sm garnets.

The conductivity spectra of the LLN-Sm40 ceramics are shown in Figure 4 at selected temperatures. A frequency-independent dc conductivity is observed at low temperatures and frequencies, whereas ${\sigma }^{\prime }\left(\omega \right)$ increases with frequency at the high-frequency side. The conductivity data were analyzed by the following formula [31,32]:

$${\sigma }^{\prime }\left(\omega \right)= {\sigma }_{dc} \left[1+{\left(\frac{\omega }{{\omega }_H}\right)}^n\right].$$

(2)

The dc conductivity σ_dc, the mobility µ, and the hopping frequency ω_H are correlated according to the following relation:

$${\sigma }_{dc}=e{n}_c\mu =\frac{n_c{e}^2\gamma {\lambda }^2}{kT}{\omega }_H.$$

(3)

The mobility is then calculated by

$$\mu =\left(\frac{e \gamma {\lambda }^2}{k}\right).\left(\frac{{\omega }_H}{T}\right).$$

(4)

In the above equations, n_c represents the exponent of the dispersion region of the conductivity, γ = 1/6 is a geometrical factor for 3D ion hopping, and λ = 1.7 Å is the hopping distance [33]. The fitting results of the conductivity spectra are shown as solid curves in Figure 4. Based on the estimated values of the hopping frequency, the mobility of lithium ions can be calculated by Equation (4).

The temperature variation of µ is shown in Figure 5 for the LLN-Sm samples. It is clear that the mobility of lithium ions increases considerably with Sm⁺³ substitutions. As a quantitative comparison at 220 K, we notice from Figure 3 that conductivity increases by a factor of ~60 for the LLN-Sm40 compared to the pure LLN sample. At the same temperature, Li⁺ ion mobility in the LLN-Sm40 sample increased by a factor of ~28 compared to LLN garnets. These results indicate that the enhanced conductivity in Sm-substituted LLN garnets is due to the increased mobility of Li⁺ ions, and the concentration of mobile Li⁺ ions plays minor role in the conductivity enhancement. It is suggested that Sm³⁺ substitutions in LLN garnets will force Li⁺ ions at 24d sites to leave their positions and move to occupy 48 g/96 h octahedral sites [33,34,35]. This process will allow more vacant 24d sites, which are essential for 3D diffusion process through the 24d–96h–48g–96h–24d chain pathway, as illustrated in Figure 6 [27,33,34,35,36].

The relaxation dynamics of the LLN-Sm ceramics are studied by the complex electric modulus $M^{*}\left(\omega \right)$, which can be expressed as follows [37]:

$$M^{*}\left(\omega \right)=j\omega C_o{Z}^{*}\left(\omega \right)$$

(5)

where C_o = A/l is the capacitance of a free sample cell. The variation in the imaginary part M’’ against the frequency is shown in Figure 7a for the x = 0.3 composition as an example. A relaxation peak is observed, and the relaxation time ${\tau }_M$ is calculated from the frequency at peak maximum f_max. The Arrhenius variation of ${\tau }_M$ of the LLT-Sm ceramics is presented in Figure 7b, and the values of its activation energy E_m are summarized in Table 1. The values of E_m are in good agreement with the activation energy of the ionic conduction process.

4. Conclusions

Sm³⁺-substituted Li_5+2xLa₃Nb_2−xSm_xO₁₂ (LLN-Sm) ceramics with x = 0–0.6 compositions were successfully synthesized by solid-state reaction and conventional sintering techniques. The LLN-Sm ceramics crystallized in the cubic garnet structure, and minor impurities appeared for the x = 0.5 and 0.6 compositions. The ionic conductivity was found to increase for the Sm³⁺-substituted samples compared to pure LLN material. The highest conductivity was obtained for the Li_5.8La₃Nb_1.6Sm_0.4O₁₂ sample with a value of 7.04 × 10⁻⁵ S/cm at RT. The mobility of Li⁺ ions was estimated from the analysis of the real part of the conductivity, and it was noticed that the mobility increased considerably for the Sm³⁺-substituted samples. It can be concluded that Sm³⁺ substitutions in LLN garnets force more Li⁺ ions to evacuate their 24d tetrahedral sites and move to 48g/96h octahedral sites. This process facilitates Li⁺ ions diffusions through the 24d–96h–48g–96h–24d chain pathway. The relaxation properties the of LLN–Sm garnets were studied by electric modulus, and the relaxation time was estimated. The relaxation process was activated with the same activation energy of the conduction process.