Hydrothermal Synthesis of Lithium Lanthanum Titanate

Abstract

Lithium lanthanum titanate (LLTO) is a very promising material due to its ability to conduct lithium ions. It has many potential applications in the field of lithium batteries and sensors. Typical synthesis methods include solid-state reaction and sol–gel synthesis. We report a novel solvothermal synthesis method that produces almost single-phase LLTO samples at significantly reduced costs. The samples thus obtained were investigated by X-ray diffraction (XRD), scanning electron microscopy (SEM), electrical impedance spectroscopy (EIS), and chemical analysis. The results obtained for the newly synthesized samples were compared with results obtained from samples prepared using the solid-state reaction method. The XRD data show the formation of orthorhombic LLTO for the solvothermal synthesis, tetragonal LLTO for the hydrothermal synthesis, and cubic LLTO for the solid-state reaction. Additionally, XRD showed that the solid-state reaction of LLTO is a multi-stage process in which intermediary compounds such as La₂Ti₂O₇ are formed. The bulk ionic conductivity of the LLTO samples produced through the solvothermal and hydrothermal processes is estimated at 10⁻⁴ S/cm, and the grain boundary conductivity is estimated at 10⁻⁶ S/cm.

1. Introduction

One of the main obstacles encountered during the development of all solid-state lithium batteries is the synthesis of a solid electrolyte [1,2,3]. An ideal electrolyte must present a set of properties such as high ionic conductivity, good interface stability with the electrodes, and high mechanical strength, and, for some applications, elasticity; it should also be non-toxic and non-flammable. It should be produced using non-critical materials [4,5]. LLTO is a perovskite-type oxide that belongs to the class of lithium super-ionic conductors (LiSICON) and meets the above criteria. The composition of the material is Li_3xLa_(2/3−x)□_(1/3−2x)TiO₃, where the □ sign denotes the lithium/lanthanum vacancies. The material presents complex mechanisms that control the ionic conductivity [3,6,7,8,9,10,11,12]. Some of the material parameters that influence conductivity are the grain structure (grain boundaries), the domains structure, the lattice parameters, the crystallization system, the crystallographic structure, and the interplay between the concentrations of Li⁺ ions and the vacancies. These parameters are interconnected. For example, increasing the Li⁺ concentration increases the number of available charge carriers but depletes the number of vacancies, which reduces the mobility of the charge carriers. The highest ionic conductivity is observed for the Li_0.35La_0.55TiO₃ composition, where x ≈ 0.11 [6,13]. This material was therefore selected as the aim of our synthesis experiments.

Bulk LLTO samples are typically obtained following two main paths: using a the solid-state reaction method and using sol–gel method [14,15,16,17]. Both methods (including their respective variations) present specific advantages and disadvantages. The solid-state reaction is typically a slow process. For the synthesis of LLTO, the solid-state reaction method requires high reaction temperatures (in the range of 1100–1300 °C), long dwell time frames (extending to 10–12 h), and intermediary grinding stages. The samples obtained present good crystallinity and are usually single-phase samples [18,19,20]. Compared with the solid-state reaction, the sol–gel synthesis achieves a more homogeneous mixture of the starting materials (due to the formation of the gel from a liquid phase), which leads to lower requirements for the heat treatment temperatures (900–1150 °C) and shorter preparation times. However, the crystallinity of the samples is lower and the number (and sometimes the quantity) of secondary phases generated is increased [5,14,15,16,21].

Recently [22], flash sintering was successfully used for the preparation of single-phase, high-density Li_0.5La_0.5TiO₃. The starting reactants used for the experiment were mixed in stoichiometric ratios with citric acid and ethylene glycol. Similarly to the stages of the sol–gel method, the authors obtained a resin containing the metallic ions in the required ratio. The resin was heat treated, and a white, amorphous powder was obtained. The powder was pressed into shape and had electrodes applied to its surface. The sample thus formed was simultaneously subjected to a heat treatment (T ≈ 1000 °C) and an electric field (in the range of 80–120 V/cm). The duration of the process was remarkably short (45 s after the onset of the flash).

High density was also achieved recently [23] by controlling the crystallinity of the TiO₂ precursor. The stability of TiO₂ is linked to the crystallization system in which its polymorphs are found. The reactivity is increased when TiO₂ is synthesized in the form of rutile nanoparticles coated with anatase and brookite.

Table 1. Comparative table of various synthesis methods reported in the literature.

Method	Calcination	Sintering	Crystallographic System	Conductivity (σ) (S/cm)	Reference
Solid-state reaction		1250 °C, 1300 °C, 1350 °C	Li poor—orthorhombic Li rich—tetragonal	1.2∙10−3 to 1.8∙10−4 bulk Li rich, tetragonal—high σ Li poor, orthorhombic—low σ	[3]
Solid-state	800 °C/4 h, then 1150 °C/12 h (×2)	1350 °C/6 h	Tetragonal	1∙10−3 bulk 2∙10−5 total	[6]
Flash sintering	550 °C/4 h	1250 °C/45 s, electric field strength E = 80–120 V/cm	Cubic	0.5∙10−3 bulk 3.4 to 5.9∙10−7 grain boundary	[22]
Solid-state reaction, TiO2 nanoparticles	1000–1200 °C/8 h	1150–1300 °C/12 h	Cubic/tetragonal	~10−3–10−4 bulk ~10−4–10−6 grain boundary	[23]
Sol–gel	900 °C/6 h	900 °C/1 h	Cubic with superlattice formation	~1.2 to 1.6∙10−3 bulk 1.5 to 4.4∙10−4 grain boundary	[10]
Sol–gel	800–1100 °C	1250 °C	Tetragonal/tetragonal + orthorhombic (depending on heat treatment temperature)	14.1∙10−3 bulk	[16]

provides an overview of the synthesis methods employed for obtaining LLTO samples and of the corresponding ionic conductivity values.

The hydrothermal method (including the solvent-based variant) could be a promising synthesis alternative, and to the best of our knowledge, the employment of the hydrothermal method to the synthesis of LLTO is still little explored. There is, however, extensive literature in the hydrothermal synthesis of other Li-based materials. LiMn₂O₄, LiMnNiO₄, and Li₄Ti₅O₁₂ are a few examples of materials that were successfully synthesized using the hydrothermal method. Two synthesis directions are observed. The first method consists of a two-stage hydrothermal process. During the first stage, some of the precursors (usually Mn₂O₃) are obtained through a hydrothermal process. In the second stage, these precursors are mixed with a material that represents a Li⁺ ion source (lithium acetate dihydrate—CH₃COOLi∙2H₂O). The mixture is then heat treated to form the targeted material [24,25,26,27,28]. For the second method, the sample is obtained during a single-stage hydrothermal process. The Li⁺ ion source typically consist of lithium hydroxide or lithium nitrate [29,30,31,32,33,34,35,36,37]. With this setup, the Li⁺ ion source participates directly in the hydrothermal process. Typical reaction conditions include relatively low temperatures (120–180 °C [33] up to 280 °C [34]) and long dwell times (on the order of 36 h). The greatest hindrance for the hydrothermal synthesis of lithium-containing compounds is the tendency of the Li⁺ ions to disperse into the liquid phase and therefore be absent from the ceramic sample. Both synthesis directions (single-stage and two-stage processes) were tested in our laboratory. The samples thus obtained present similar characteristics concerning the crystallographic structure, the phase purity, and the ionic conductivity. The successful synthesis of LLTO by this method could allow the fabrication of samples with nanostructured organization. The properties of the nanostructured LLTO could be different from the properties encountered in the bulk samples. Nanostructured LLTO also has the potential to be integrated within a hybrid, ceramic–organic electrolyte. Lastly, the hydrothermal synthesis could yield LLTO samples at reduced time and energy costs. Therefore, the objective of this research was to investigate the feasibility of preparing Li_0.35La_0.55TiO₃ using a hydrothermal-based process. A secondary objective was to test the properties of the LLTO samples obtained following the hydrothermal-based synthesis and to compare the results with data obtained for LLTO samples prepared using a conventional approach (solid-state reaction). Finally, this study could open new research directions, such as optimization of the synthesis parameters and integration of the newly synthesized LLTO samples into hybrid electrolytes.

2. Materials and Methods

2.1. Preparation of the LLTO Perovskite

The single-stage process was a solvothermal synthesis. The two-stage process was a hydrothermal synthesis. Samples were also prepared with the solid-state reaction.

The starting materials used for the solvothermal synthesis of the LLTO perovskite were Li₂CO₃, 99.99%; La(NO₃)₃∙6H₂O, ≥ 95% purity; and Ti-{O-CH-(CH₃)₂}₄, ≥98% (TTL). The solvent used during the solvothermal reaction was CH₃-CH₂-OH, 99.8% purity. For the hydrothermal experiments, the starting materials were La(NO₃)₃ aqueous solution (c_La = 29 g/L), TiOCl₂ aqueous solution (c_Ti = 50.8 g/L), and Li₂CO₃. The solvent used was distilled water. The precipitation agent was KOH. For the solid-state reaction synthesis, the starting materials were TiO₂; La₂O₃, 99.99%; and Li₂CO₃. The chemicals were acquired from Sigma-Aldrich, Darmstadt, Germany (La₂O₃, Li₂CO₃), VWR Chemicals, Lutterworth, England (La(NO₃)₃∙6H₂O), and Thermo Scientific, Bend, OR, United States of America (TTL) and were used directly, without any further purification. TiOCl₂ and TiO₂ were prepared in the laboratory.

For the single-stage reaction, stoichiometric quantities of La(NO₃)₃∙6H₂O and TTL were weighed and dissolved in ethanol. Li₂CO₃ was added so that the required Li mass was exceeded from 10 wt% up to 27 wt%. The excess Li was added in order to compensate for losses during the solvothermal process and subsequent heat treatment. Li₂CO₃ is not soluble in ethanol under ambient conditions; therefore, initially, the Li₂CO₃ particles remained in suspension and possibly formed nucleation centers. The initial suspension was loaded in an autoclave under Ar atmosphere. The autoclave temperature was 200 °C. The pressure was varied from 75 to 106 atm. The dwell time varied from 3 to 6 h. The color of the solution changed following the completion of the solvothermal process. The initial solution was colorless, and the suspended powder particles were white. After the solvothermal process, both the solution and the powder acquired a yellow tint, which indicates that the synthesis reaction has occurred. The solution was filtered following the solvothermal process. The resulting green powder was dried, pressed into pellets, and heat treated in air. With this process, the untreated powder presents a very fine structure at a macroscopic scale. This property leads to further complications for obtaining the pellets. The pellets were formed by compressing the fine powder in a cylindrical die with a 20 mm diameter under a maximum pressure of 16 MPa applied uniaxially. If the pressure is exceeded, the pellet becomes lodged into the die. The thickness of the pellets after pressing was typically 1.7 mm. The heat treatment consisted of a single stage at a temperature of 1350 °C for a dwell time of 6 h. The pellets were allowed to cool naturally to room temperature. The pellets shrank during the heat treatment. They lost approximately 40% of their diameter and 20% of their thickness.

For the two-stage method, precursors for La and Ti were precipitated separately from solutions of La(NO₃)₃ and TiOCl₂, respectively. As mentioned at the beginning of this section, the concentrations of the metal ions in the two initial solutions were c₁ = 29 g/L for La and c₂ = 50.8 g/L for Ti. These concentrations were then used to calculate the volumes of solution that contain the stoichiometric quantities of metals required for preparing the Li_0.35La_0.55TiO₃ composition. The La(NO₃)₃ solution underwent a precipitation reaction with KOH. The KOH solution had a molar concentration of 1 M. It was added to the La(NO₃)₃ solution by dripping under magnetic stirring. The pH of the La(NO₃)₃ solution was 5.22 before the precipitation. The precipitation was stopped when the pH of the mixture reached 9.9. The TiOCl₂ solution underwent a similar precipitation. The molar concentration of the KOH solution was 5 M. The initial pH of the TiOCl₂ solution was 0.26. The precipitation was stopped when the pH of the solution reached 10.04. The two resulting precipitates (for La and Ti, respectively) were mixed and subsequently loaded into an autoclave and used for the hydrothermal reaction. The reaction conditions were P = 100 atm, T = 200 °C, and t = 2 h. The powder obtained following the hydrothermal process was filtered and washed to neutral pH. The slurry was then mixed with distilled water (leaving about 1 inch of water above the height level of the solid fraction), and a stoichiometric (i.e., without excess Li) amount of Li₂CO₃ was added to the solution under magnetic stirring. The water was the then removed by freeze-drying. The obtained green powder was pressed into 20 mm pellets under a pressure of 159 MPa, applied uniaxially. The pellets were then subjected to a heat treatment at 1350 °C for 6 h.

Solid-state samples were obtained by mixing stoichiometric amounts of Li₂CO₃ (accounting for 10% excess Li), La₂O₃, and TiO₂. The powders were ground, pressed into pellets, and heat treated. The pellets were produced on the same 20 mm diameter die, by uniaxial pressing to a pressure of 32 MPa. The heat treatment consisted of three calcination stages and one sintering stage. The calcination stages were performed at 800 °C for 4 h (one stage) and 1150 °C for 12 h (two stages). The sintering stage was carried out at a temperature of 1350 °C for 6 h (one stage). The pellets were re-ground between the heating stages.

The synthesis conditions described above generated the best samples in terms of sample crystallinity, phase purity, and ionic conductivity. Other synthesis experiments were carried out using different methods, different compositions of the starting materials, and different reaction conditions. These experiments yielded samples with reduced crystallinity, reduced phase purity, increased porosity, and consequently lower ionic conductivity.

Table 2. Examples of synthesis experiments.

Method	Starting Materials	Parameters to Obtain the Initial Powder	Heat Treatment	Result
Sol–gel	TTL + La(NO3)3∙6H2O + LiNO3	Drying the gel at 120 °C Calcination at 800 °C/1 h	1000 °C/10 h	Low crystallinity, high porosity, low conductivity (~10−5 S/cm bulk conductivity), but also lowest Li losses.
Solid-state reaction	TiO2 + La(OH)3 + LiOH∙H2O + Li2CO3	Calcination at 800 °C/4 h Calcination at 1150 °C/12 h (one stage)	1150 °C/12 h	High crystallinity, but some Al-based secondary phases are observed in small quantities (the high reactivity of LiOH causes reactions with the Al2O3 crucible).
Solid-state reaction	TiO2 + La2O3 + Li2CO3	Calcination at 800 °C/4 h Calcination at 1150 °C/12 h (two stages)	1350 °C/6 h	High crystallinity, high phase purity (no contamination), highest observed conductivity.
Hydrothermal	TiOCl2(aq.) + La(NO3)3(aq.) + LiCl∙H2O	P = 90 bar, T = 200 °C, t = 2 h	-	Total loss of Li into the solution.
Hydrothermal	Step 1: TiOCl2(aq.) + La(NO3)3(aq.) Step 2: precursor obtained from step 1 + Li2CO3	Step 1: P = 100 bar, T = 200 °C, t = 2 h Step 2: freeze-dry	1350 °C/6 h	Good crystallinity, low amounts of secondary phases, low conductivity despite the low Li losses.
Solvothermal	TTL + La(NO3)3∙6H2O + LiNO3	P = 92 bar, T = 200 °C, t = 3 h	1350 °C/6 h	Significant Li loss, very low phase purity (68 wt% to 54 wt% LLTO and, respectively, 32 wt% to 46 wt% La4Ti9O24).
Solvothermal	TTL + La2O3 + Li2CO3	P = 85 bar, T = 200 °C, t = 20 h	1350 °C/6 h	The majority phase for this sample is Li2La2Ti3O10. La2Ti2O7 and LLTO are formed as secondary phases. The LLTO fraction is approximately 20 wt%.
Solvothermal	TTL + La(NO3)3∙6H2O + LiOH + Li2CO3	P = 85 bar, T = 200 °C, t = 3 h	1350 °C/6 h	Significant Li losses into the solution.
Solvothermal	TTL + La(NO3)3∙6H2O + Li2CO3 (10 wt% excess Li)	P = 75 to 106 bar, T = 200 °C, t = 3 to 6 h, respectively	1350 °C/6 h	Identical samples are obtained through the specified pressure and dwell time ranges. The samples present good crystallinity, high phase purity, moderate Li losses, but low conductivity.
Solvothermal	TTL + La(NO3)3∙6H2O + Li2CO3 (27 wt% excess Li)	P = 85 bar, T = 200 °C, t = 20 h	1350 °C/6 h	Increased Li losses into the solution compared with the previous case, but otherwise, the samples are comparable.

presents the review of the synthesis parameters and some of the synthesis experiments that did not produce samples with the required characteristics.

2.2. Investigation Techniques

The sample microstructure was analyzed by scanning electron microscopy (SEM) using a FEI company Quanta 250 device equipped with an energy-dispersive X-ray spectroscopy (EDX) detector with a 30 mm² area. The SEM and the EDX detector were sourced from the same manufacturer, the FEI company, Eindhoven, The Netherlands. The samples were coated with a thin film of Au by sputtering. The film thickness was approximately 10 nm. Particle size distribution was determined from the SEM images using the ImageJ software, version 154. The chemical compositions of the samples were investigated by EDX in order to determine the La: Ti atomic ratio. Inductively coupled plasma–optical emission spectroscopy (ICP-OES) measurements were carried out on an Agilent 725 device, Agilent Technologies Incorporated, Palo Alto, CA, USA. ICP-OES was used for estimating the Li concentration.

The crystallographic characteristics of the samples were investigated using a Bruker D8 Advance, BRUKER AXS GmbH, Karlsruhe, Germany, diffractometer operating in reflection geometry using the ϴ-ϴ configuration. The X-ray tube of the diffractometer uses a Cu anode. The device is equipped with a Cu Kβ filter. The phase identification was performed using the EVA software, version 5, 2019 and the ICDD PDF-5+ 2024 database. The experimental diffraction patterns were fitted using the Rietveld code with the FullProf software, version April 2023. The experimental diffraction peaks were fitted using the Thompson–Cox–Hastings pseudo-Voight function. The pseudo-Voight function consists of a weighted sum of Gaussian and Lorentzian functions. The full width at half maximum (FWHM) parameters of these functions depends on the diffraction angle according to the following expressions:

$${FWHM}_{Gauss}^2=H_G^2=U{tan}^2\left(\theta \right)+Vtan\left(\theta \right)+W$$

$${FWHM}_{Lorentz}=H_L=Xtan\left(\theta \right)+{\displaystyle \frac{Y}{cos\left(\theta \right)}}$$

where U, V, and W are fitting parameters that describe the instrumental broadening, respectively, and X and Y are refinable parameters used for the description of the sample-induced peak broadening. The instrumental contribution to the width of the diffraction peaks (the U, V, and W parameters) was determined by measuring an Al₂O₃ reference sample, which does not provide any peak broadening associated with crystallite size and/or strain effects. The Al₂O₃ etalon and the LLTO samples were measured under identical conditions. During the refinement, the X parameter was used to describe isotropic strain effects. The Y parameter describes isotropic size-broadening effects. For samples with tetragonal or orthorhombic symmetry, an anisotropic correction was added. The anisotropy direction was chosen along the c axis due to the observation of the fact that the width of the (0 0 1) reflection exceeds the width of any other reflection. The apparent size is calculated according to the following equation:

$$App.size={\displaystyle \frac{1}{{\beta }_{size}}}$$

where β_size is calculated from the size-dependent contributions to the FWHM.

The initial structural data were provided by the matching ICDD PDF-5 database entries. The background was modeled as a linear interpolation between a set of manually selected background points, where the height of the points is a refinable parameter. The quantitative phase analysis is performed according to the following equation:

$$W_i={\displaystyle \frac{\frac{S_i{Z}_i{M}_i{V}_i}{t_i}}{{\sum }_n\frac{S_n{Z}_n{M}_n{V}_n}{t_n}}}$$

where W_i is the weight fraction of phase i from a system of n phases; S represents the scale factor; Z is the number of formula units per unit cell; M is the molar mass of the formula unit; V is the cell volume; and t is the Brindley particle-absorption contrast factor.

Electrical impedance spectroscopy (EIS) measurements were carried out using a Metrohm Autolab PGSTAT 128 N equipment, Metrohm AG, Utrecht, The Netherlands. The samples were prepared for EIS by polishing. The last polishing step was carried out with sandpaper of the 4000 P grit size, Buehler, Lake Bluff, IL, United States of America. Au electrodes were deposited on the polished surfaces by sputtering. The measurements were performed at room temperature. The applied signal had a sinusoidal shape and an amplitude of 0.2 V RMS. The frequency range of the measurements was 1 Hz–1 MHz. Data were collected and fitted using the Nova 2.1 software. The semicircles observed on the Nyquist plot, characteristic of conduction mechanisms within grains and at grain boundaries, respectively, were fitted separately with the equation for a semicircle. The semicircle represents a model circuit that consists of a resistor (R_s) connected in series to a parallel R_P–CPE group (R_P denotes a resistor and CPE denotes a constant phase element). The fit returns values for the circuit elements. R_P is then used for extracting the conductivity value using the following equation:

$$\sigma ={\displaystyle \frac{l}{R_{P}S}}$$

where l and S represent the thickness and the cross-sectional area of the sample, respectively.

3. Results

3.1. XRD Results

The green powders obtained following both the solvothermal and the hydrothermal methods were generally amorphous, regardless of the synthesis parameters. The X-ray diffraction pattern for these materials did not present any Bragg reflections characteristic of long-range ordering. Very broad features could be observed that indicate the formation of amorphous materials. Some speculations could be made on the nature of the amorphous powder. As mentioned in the previous sections, the pellets shrink during the heat treatment that follows the solvothermal synthesis. They lose as much as 70% of their volume and 25% of their mass. The entire Li mass, including the excess Li, of Li_0.35La_0.55TiO₃ accounted for only about 1.5% of the total sample mass. It is known that Li losses occur through evaporation during the heat treatments. However, even the complete loss of Li could not be sufficient to explain the observed losses of mass. It could therefore be inferred that the amorphous powder still contains carbonates, and possibly some organic residues, caused by the presence of TTL in the autoclave.

For the green powder produced following the two-stage method, a single diffraction peak was observed at 2ϴ ≈ 25.5°. This peak could be tentatively explained as an indication of the presence of anatase TiO₂.

The samples become crystalline following the heat treatment. Figure 1 shows representative XRD patterns for the powders obtained following the solvothermal/hydrothermal processes prior to the heat treatment (green powder) and following the last stage of the heat treatment (calcinated/sintered powder). Figure 1a shows the entire XRD pattern of the five samples. Figure 1b shows a detailed image of the XRD patterns corresponding to the sintered samples in the 2ϴ range of 45° to 50°.

In the case of the sintered powder generated from the solvothermal method, two phases were obtained. The main phase corresponded to the LLTO perovskite, as identified by the 01-070-6718 database entry. It accounted for approximately 95.7% of the sample mass. The secondary phase was identified as the rutile TiO₂ according to the 04-002-8295 database entry. The quantitative phase analysis shows that the TiO₂ mass concentration was approximately 4.3%. The main phase crystallized in an orthorhombic system with the Pmmm space group. The lattice parameters of the main phase were as follows: a = 3.876(1) Å, b = 3.866(1) Å, and c = 7.782(1) Å, respectively. Errors are estimated by comparing results from different fit models of the same sample. This approach is used throughout the entire paper. The formation of the La-rich layer superlattice was clearly observed by the presence of the characteristic (0 0 1) reflection at a diffraction angle of 2ϴ ≈ 11.4°. No significant variation in the lattice parameters was observed with the modification of the reaction parameters for the solvothermal (single-stage) reaction. However, as shown in Table 2, modifications of the starting materials can lead to the formation of other phases. The discussion will remain focused on the mixture of starting materials that was observed to generate the LLTO phase.

Table 3. Variations in lattice parameters with the reaction conditions for LLTO samples obtained through the solvothermal method.

Starting Materials	Pressure (bar)	Temperature (°C)	Time (h)	Heat Treatment (°C/h)	Lattice Parameters (Å)
TTL + La(NO3)3∙6H2O + Li2CO3 (10 wt% excess Li)	75	200	3	1350 °C/6 h	a = 3.875 b = 3.865 c = 7.779
TTL + La(NO3)3∙6H2O + Li2CO3 (10 wt% excess Li)	106	200	6	1350 °C/6 h	a = 3.876 b = 3.866 c = 7.782
TTL + La(NO3)3∙6H2O + Li2CO3 (27 wt% excess Li)	85	200	20	1350 °C/6 h	a = 3.874 b = 3.863 c = 7.786
TTL + La(NO3)3∙6H2O + Li2CO3 (10 wt% excess Li)	114	200	3	1350 °C/6 h	a = 3.874 b = 3.864 c = 7.786
TTL + La(NO3)3∙6H2O + LiNO3 (10 wt% excess Li) *	92	200	3	1350 °C/6 h	a = 3.875 b = 3.863 c = 7.786

* The sample contains significant amounts of La₄Ti₉O₂₄; however, the LLTO phase is formed, and it still retains the same orthorhombic structure and essentially the same lattice parameters as the other samples produced by the solvothermal method.

shows the values of the lattice parameters observed in the case of various solvothermal synthesis experiments. The average size of the coherently scattering domains (we refer to this parameter as crystallite size) is 145 nm.

Figure 2 shows the diffraction pattern and subsequent Rietveld analysis for the solvothermal sample. The examination of the diffraction pattern indicates that the calculated peak intensities do not perfectly match the observed intensities. The observed intensity is smaller than the calculated intensity for the (0 0 1) reflection, which appears at 11.4°. For the (1 0 2), (0 1 2), (1 1 0), and (1 1 2) reflections, respectively, the observed intensity is greater than the calculated intensity. This could be an indication that the structural model is not yet entirely accurate, but it could also indicate that LLTO exists as a mixture of phases. The best agreement factors were achieved considering a model that contains approximately 83% orthorhombic LLTO, 3% cubic LLTO, 10% orthorhombic La_0.67TiO₃, and 4% TiO₂. For this model, the obtained weighted profile factor for points with Bragg contributions, corrected for background, was R_wp = 16.6, and the goodness of fit parameter was χ² = 3.35. Assuming contributions from only the orthorhombic LLTO phase and TiO₂, the same agreement factors are R_wp = 18.7 and χ² = 4.69. It should be noted that accurate phase identification might not be possible with our setup, due to the strong overlap of the reflections.

For the sample prepared following the two-stage hydrothermal method, the main phase is the LLTO perovskite phase. Figure 3 shows the XRD pattern and corresponding Rietveld refinement of the sintered hydrothermal sample. The LLTO phase is crystallized in a tetragonal system, although it could also exist as a mixture between the tetragonal and the cubic phases. The Rietveld refinement was carried out in the latter model. The space group characteristic of the tetragonal cell is P4/mmm, and the lattice parameters are a = 3.870(2) Å and c = 7.770(1) Å. The lattice parameter of the cubic cell is a = 3.875(1) Å, and the characteristic space group is Pm−3m. The majority phase is the tetragonal phase.

It accounts for 64 wt% of the mass of the sample. The cubic phase represents 29 wt% of the total mass. The phase composition values should be interpreted with care since the peak intensities are overlapped, and the structural models of the LLTO phases may not yet be sufficiently accurate to determine a unique solution. The remaining 7 wt% is explained by the presence of a secondary phase, namely Li₂La₂Ti₃O₁₀, irrespective of the model employed. The superlattice peak is observed, but the intensity is low. The average crystallite size of the tetragonal phase is approximately 100 nm.

The dynamics of the solid-state reaction of LLTO was tentatively explored. The formation of the LLTO phase by solid-state reaction is known to be a multi-stage process [18,22]. This process was observed experimentally, in agreement with the literature data.

Three calcination stages, interrupted by intermediate grinding of the powder, are required for obtaining the LLTO phase. The solid-state reaction does not appear to generate the LLTO phase directly. Instead, the initial mixture of oxides appears to produce some intermediary phases. These intermediary phases act as precursors. The pellets have to be reground in order to re-homogenize the composition and to allow the precursors to continue the chemical reaction toward the final, stable structure of LLTO.

The XRD patterns collected at the intermediary steps are represented in Figure 4. Examination of the diffraction data shows the formation of at least two types of phases. These are the La₂Ti₂O₇ (ICDD—PDF entry 00-072-064) phase and the LLTO phase. A possible third phase could be Li_1.7Ti_1.6O₄. This third phase is potentially observed at the third calcination stage; however, it presents a single, unique Bragg reflection at a 2ϴ angle of ~18.4°, corresponding to the (1 1 1) plane. The other reflections of the Li_1.7Ti_1.6O₄ phase are overlapping diffraction peaks corresponding to the LLTO and La₂Ti₂O₇ phases. The identification of the Li_1.7Ti_1.6O₄ phase is only tentative; however, the presence of this phase does not contradict the model that describes the dynamics of the solid-state reaction. The phase composition of the samples (at the intermediary steps) is presented following the chronological order of the heat-treatment stages, starting with the second stage.

At the second calcination stage, the sample contains mostly La₂Ti₂O₇, and only a small amount of the sample is formed in the LLTO phase. Accurate identification of the crystallographic system corresponding to the LLTO phase is impossible at this stage due to the low intensity of the observed reflections.

At the third calcination stage, the sample contains the three types of phases mentioned above (LLTO, La₂Ti₂O₇, and Li_1.7Ti_1.6O₄). The LLTO phase itself appears to contain a mixture between the tetragonal and the cubic polymorphs. This configuration produced the best fit. As the solid-state reaction proceeds, the quantity of La₂Ti₂O₇ decreases and the quantity of the LLTO phase increases.

After the last heat treatment, the sample contains almost exclusively LLTO. The La₂Ti₂O₇ is no longer observed. However, LLTO is again observed as a mixture of polymorphs (possibly cubic and orthorhombic). The main LLTO phase has cubic symmetry. The characteristic space group is again Pm−3m. The lattice cell size is a = 3.870(1) Å. The (2 0 0) reflection at about 46.9 degrees does not show the splitting characteristic for the lower symmetry system (Figure 1b). The secondary LLTO phase is orthorhombic. The Bragg reflections characteristic of the orthorhombic phase are very broad and low in intensity. This indicates the low crystallinity of the secondary phase. It also makes the accurate identification of the phase nearly impossible. If the orthorhombic phase is confirmed, the phase crystallizes with the Pmmm space group and lattice parameters of approximately a = 3.887 Å, b = 3.874 Å, and c = 7.805 Å. The average crystallite sizes are 130 nm for the cubic phase and 25 nm for the orthorhombic phase. It can be observed that the orthorhombic phase is visible only as a shoulder to the peaks of the main phase (Figure 1b). The orthorhombic phase accounts for approximately 26 wt% of the sample mass. Interestingly, the LLTO phase shows the formation of the superstructure at the third calcination stage; however, no superstructure is observed after the sintering stage.

Table 4. XRD results.

Sample	Main LLTO Phase: System, Space Group, Lattice (Å)	Secondary LLTO Phase: System, Space Group, Lattice (Å)	Phase Composition
Solvothermal (single stage)	orthorhombic Pmmm a = 3.876(1) b = 3.865(1) c = 7.783(1)	cubic Pm−3m a = 3.88	83% orthorhombic LLTO 3% cubic LLTO 10% La0.67TiO3 4% TiO2
Hydrothermal (two stages)	tetragonal P4/mmm a = 3.870(2) c = 7.770(1)	cubic Pm−3m a = 3.875(1)	64% tetragonal LLTO 29% cubic LLTO 7% Li2La2Ti3O10
Solid-state reaction (last heat treatment)	cubic Pm−3m a = 3.870(1)	orthorhombic Pmmm a = 3.887 b = 3.874 c = 7.805	72% cubic LLTO 26% orthorhombic LLTO 2% other impurities

summarizes the phase compositions and cell parameters of the samples produced following the three synthesis methods. The examination of the diffraction peak at about 47 degrees (Figure 1b) appears to be a good indication of the crystalline symmetry of the main phase. For the cubic phase, only the (2 0 0) reflection is observed at this angle range. For the tetragonal LLTO phase, both the (0 0 4) and the (2 0 0) reflections are observed separately. This is obvious, since for a tetragonal system a ≠ c, and for this system, due to the formation of the La layer superstructure, c_{superstructure} ≈ 2a_cubic.

Finally, for an orthorhombic LLTO system, a ≠ b ≠ c, and therefore, the equivalent reflections are split yet again. For the orthorhombic LLTO structure, at about 47°, the observed reflections are generated from the (0 0 4), (2 0 0), and (0 2 0) planes. This trend continues at higher angles. At approximately 48.5°, no reflection is observed for the cubic system, but the (1 1 3) and (2 0 1) reflections are observed in the tetragonal system. Respectively, the (1 1 3), (2 0 1), and (0 2 1) peaks are observed for the orthorhombic system.

When samples produced by different methods are compared, the superstructure peak is clearly observed for the solvothermal method, barely observed for the hydrothermal method, and not observed for the solid-state method. The symmetry and the phase composition of the samples seem to change. The orthorhombic structure is clearly visible for samples prepared by the solvothermal method. The characteristic peaks at 2ϴ ≈ 47° are well defined. Also, peaks are observed at 48.5°. For the samples prepared by the hydrothermal method, the characteristic peaks at 47° are less clearly separated, but they remain nevertheless visible. This suggests the formation of a tetragonal phase. The experimental data were initially fit with an orthorhombic system, and it was observed that the lattice parameters a and b converge. This could further indicate the formation of a tetragonal system. The diffraction peaks at 48.5° are still visible. For the sample obtained by solid-state reaction, the reflection at 47° shows a well-defined, narrow peak, and no reflection is visible at 48.5°, which indicates the formation of a main phase with cubic symmetry. This is surprising, since the sample was not quenched, although quenching is reported in the literature to be required for obtaining a cubic structure [17].

It could be inferred that the sample obtained by the solid-state reaction is a high-crystalline-symmetry sample (possibly cubic or tetragonal with some orthorhombic traces). The sample obtained by the solvothermal method is a lower-symmetry sample (orthorhombic). Finally, the sample prepared by the hydrothermal method is intermediary in terms of symmetry to the other samples (tetragonal or a tetragonal/cubic mixture). This observation is consistent with the preparation methods, namely, the ways in which Li participates in the reactions. For the solvothermal method, the Li precursor is loaded into the autoclave. The Li losses in the solution are high; therefore, the Li content of the ceramic is low. This condition is known to favor the formation of an orthorhombic phase [3,17,23]. On the other hand, the Li precursor is added after the hydrothermal is already complete for the two stage-process. There are, apparently, lower Li losses into the solution (compared with the solvothermal method), so the Li concentration in the ceramic is higher, thus favoring the formation of a higher-symmetry phase. Finally, the situation is similar in the case of the sample obtained by the solid-state reaction, namely, the concentration of Li is relatively high.

3.2. SEM Results

Figure 5 shows representative SEM images obtained for the three types of samples (obtained by the solvothermal method, the hydrothermal method, and the solid-state reaction, respectively). In the case of the samples prepared by the solvothermal/hydrothermal methods, SEM investigations were performed on both the powder obtained immediately following the synthesis in the autoclave (green powder) and the powder that underwent the last high-temperature treatment (sintered powder).

The green powder produced by the single-stage solvothermal method presents spherical structures with an average diameter of 2.5–3 µm. The size distribution is non-homogeneous. The diameters range from sub-micrometer values to 6 µm. Figure 5b shows the particle size distribution of the respective spherical particles. After sintering, the samples show the formation of a porous structure with interconnected grains. The edges of the grains appear well defined and seem to lack sharp corners. The size distribution of the particles ranges from 1 to 7.5 µm, with an estimated average particle size of 2.5 µm. It can be observed on the SEM image in Figure 6 (which is presented in the next section) that for some regions of the sample, the gray value is 0 (complete black). These regions are attributed to voids. The total void area represents approximately 5.6 μm², 2.09% of the total area represented in Figure 6 (≈268 μm²). Therefore, it could be inferred that the area density of the sample represents 97.9% of the area density of an ideal sintered sample (no porosity).

By contrast, the powders produced by the two-stage hydrothermal method show different characteristics. The green powder is obtained as a homogeneous mixture of agglomerated, cubic-shaped structures. The related sintered powder shows cubic-shaped grains with well-defined edges and sharp corners. The width of the grains ranges from 1 µm to 9.5 µm. The average grain size is on the order of 2–3 µm. The porosity of the samples obtained by the two-stage method is reduced compared with the porosity of the samples obtained by the single-stage method. Similarly, the morphology of the samples obtained through the solid-state reaction shows cubic-shaped grains, sharp edges, and sharp corners, with average sizes in the range of 2–3 µm and reduced porosity.

3.3. Chemical Composition

EDX analysis of the green powder prepared by the single-stage method shows two types of regions. In the first type, the chemical composition is consistent with the formation of a precursor for the LLTO phase. Here, the atomic ratio of La and Ti is La:Ti = 0.55:1. For the second type of region, the concentration of Ti is higher than the expected value. Typical values of the atomic ratio for this region are La:Ti = 0.29:1.

Figure 6 shows representative EDX spectra acquired for the LLTO sample prepared following the solvothermal method after the high-temperature heat-treatment stage.

In the case of the corresponding sintered powder, the images obtained using backscattered electrons show the formation of two chemically distinct phases (Figure 6a,b). The main phase produces the highest gray values. EDX data acquired on grains of the main phase (Figure 6a, point 1) indicate the formation of the LLTO oxide (Li_0.35La_0.55TiO₃) with the observed La:Ti atomic ratio of 0.55:1. For the secondary phase (Figure 6b, point 2), the EDX spectrum shows high-intensity peaks corresponding to the Ti Kα₁ and Kα₂ lines. The La Lα₁, Lα₂, Lβ₁, and Lβ₂ spectral lines were also observed, but at lower intensities. This secondary phase could contain TiO₂ (which is observed by XRD). An analysis of the area covered by the two phases was carried out. If we imagine that a slab of the sample is cut with parallel surfaces, such that grains of different phases have the same thickness, then the phase composition of the sample could be estimated knowing the area distribution and the density of the phases. For this evaluation, the values of the crystallographic densities were used, assuming that the main phase is orthorhombic LLTO (ρ_LLTO = 5.58 g/cm³) and that the second phase is rutile TiO₂ (ρ_TiO2 = 4.25 g/cm³). Under these assumptions, the phase composition is 92.05 wt% LLTO and 7.95 wt% TiO₂. The results are very close to the phase composition determined by XRD, where typically, the TiO₂ phase was found in the composition range of 4.5–6 wt%. The chemical composition of the main phase was also confirmed by ICP-OES. By this method, the observed ratio of the metallic species is La:Ti = 0.52:1 and Li:Ti = 0.08:1 after sintering. It is observed that the Li content of the sample is lower than the content of the aimed composition (Li:Ti = 0.35:1). The difference is explained by the Li losses into the solution and by the Li losses during the heat treatment.

For the case of the two-stage process, the Li losses into the solution are minimized. The precursor for Li does not participate directly in the hydrothermal process, and Li losses through freeze-drying are expected to be negligible. Indeed, the ICP-OES chemical analysis of the green powder indicates that the Li:Ti ratio is 0.41:1. EDX data obtained for the green powder shows a non-uniform composition. In some regions, the sample shows a close composition to the expected value for LLTO (observed La:Ti = 0.54:1). Then, individual crystallites that contain almost exclusively Ti are observed. These crystallites are linked to the formation of TiO₂ (also possibly identified by XRD). By contrast, the corresponding sintered powder shows a very homogeneous composition (EDX). The average composition is La:Ti = 0.53:1, with individual points at which the value of La:Ti = 0.55:1 is observed.

For the solid-state synthesis, the observed composition at the end of the sintering stage is La:Ti = 0.55:1 (EDX), La:Ti = 0.62:1, and Li:Ti = 0.18:1 (ICP-OES), respectively. The EDX spectra corresponding to the results discussed in this section can be found in the Supplementary Materials.

3.4. EIS Investigations

The EIS investigations were conducted on three samples corresponding to the three synthesis methods discussed in this paper. The Nyquist plots corrected for the sample geometry are shown in Figure 7. A typical Nyquist diagram for an ionic conductor presents a linear section at low frequencies, associated with the blocking effect of the electrodes, and two arcs. The low-frequency arc is associated with the conductivity mechanisms at the grain boundaries. The high-frequency arc is associated with the bulk conductivity of the material. The linear section is observed below 40 Hz for the single-stage sample, below 10 Hz for the two-stage sample, and below 57 Hz for the sample produced by solid-state synthesis.

The low-frequency arc is clearly visible for all samples. The frequency (along this arc) at which the imaginary part of the impedance (Z″) reaches the peak absolute value is ν_top ≈ 110 Hz for the solvothermal sample, ν_top = 159 Hz for the hydrothermal sample, and ν_top = 404 Hz for the sample obtained by solid-state reaction. The high-frequency arc is not completely visible for any sample. The diagrams corresponding to the solvothermal and hydrothermal samples present an inflection point at about 272 kHz and 251 kHz, respectively.

At higher frequencies (i.e., beyond the inflection point), the slope of the Nyquist diagram decreases for the hydrothermal sample and even changes its sign for the solvothermal sample. This could indicate the emergence of the high-frequency arc ascribed to the bulk conductivity of the sample. For the sample obtained by solid-state reaction, the behavior is similar. The inflection point (which actually represents the minimum absolute value of Z″) is observed at approximately 443 kHz. The emergence of the high-frequency arc is much more pronounced for this sample; however, the complete arc is not observed. The maximum frequency experimentally available on our setup is 1 MHz, which is relatively low compared with the frequency values typically employed in the study of LLTO (1 MHz–10 GHz [9,12,17,21,22]). The lack of observation of the high-frequency arc could be explained by the fact that the arc is located at frequency domains that cannot be explored with our current setup. Because the complete arc is not available for data processing, the reliability of the fit is low. Conductivity values are only estimated, and the results have to be interpreted with care.

Table 5. Estimation of grain conductivity and grain boundary conductivity.

Sample	σgrain (S/cm)	σgrain boundary (S/cm)
Solvothermal method (1 stage)	7.48∙10−4	3.66∙10−6
Hydrothermal method (2 stages)	~10−4	2.80∙10−6
Solid-state synthesis	1.44∙10−3	1.93∙10−5

shows the results of the conductivity estimations.

The different types of available LLTO phases have different conductivity characteristics. At the simplest explanation level, the conductivity characteristics could be correlated with the material structure based on the following observations. Generally, the bulk conductivity increases with the symmetry of the material; thus, the highest bulk ionic conductivity is observed for cubic LLTO, followed by the conductivity of tetragonal LLTO. Finally, orthorhombic LLTO shows the lowest conductivity. Conductivity is anisotropic. The material forms a layered structure consisting of alternations of La-rich (Li-poor) atomic planes and La-poor (but Li- and vacancy-rich) atomic planes along the c axis. Conductivity occurs along the direction of the La-poor layers (along the a and b axes), but it becomes hindered on the direction perpendicular to the La-rich planes (along the c axis). For the cubic phase (when the layered structure is not formed), the conductivity is isotropic, and therefore, overall, larger than the conductivity of the tetragonal phase. Orthorhombic LLTO again tends to generate the layered structure discussed before. The orthorhombic phase appears at low Li concentrations, and therefore, the charge carrier density is below optimum values, which further lowers the bulk conductivity. There are further parameters that dictate the bulk conductivity, such as the lengths of the atomic bonds (the size of the gap between the oxygen tetrahedra, which represents the conduction channel), the size of the unit cell, and the domain structure, and then, the parameters that define the conductivity at the grain boundaries have to be taken into account. An exhaustive discussion of the effects of these parameters is beyond the scope of this work. However, it can be mentioned that increasing the number of grain boundaries has the effect of lowering the conductivity.

Some speculations can be made on the behavior of the samples presented in this study. The resistivity of the sample is proportional to the diameter of the arc.

The single-stage solvothermal sample crystallizes in an orthorhombic system, and the Li content of the sample is low. These characteristics lower the bulk conductivity. The microstructure of the sample shows some porosity, and the sample contains TiO₂, which is an insulator. These characteristics are expected to lower the grain boundary conductivity.

The sample prepared by the two-stage hydrothermal method presents large grains and low porosity, similarly to the solid-state reaction sample; therefore, higher grain boundary conductivity is expected but not observed. The reasons for the lower-than-expected grain boundary conductivity are not entirely clear, but the presence of the Li₂La₂Ti₃O₁₀ secondary phase, the coexistence of the tetragonal and the cubic LLTO polymorphs, and possibly the relative orientation of the grains could lead to the formation of a complex grain boundary structure that might not favor high conductivity.

The overall conductivity of the solid-state sample is higher that the conductivity of the other two samples. This is expected, because the sample contains densely packed grains (as evidenced by SEM), and this should improve grain boundary conductivity. The sample is almost single-phase (as evidenced by XRD). The absence of insulating impurities could also be expected to improve conductivity. The main phase is apparently cubic (according the XRD data). This could increase the bulk conductivity of the sample. The Li content of the sample is relatively high (although below optimum according to the ICP-OES data), so high bulk conductivity could be expected. However, the presence of the secondary, orthorhombic, LLTO phase could limit both components of the conductivity (by increasing the number of grain boundaries and by redistributing the available Li⁺ charge carriers between two phases).

4. Conclusions

Lithium super-ionic-conductor perovskite samples of the Li_3xLa_2/3−xTiO₃ type were successfully synthesized using two methods based on the hydrothermal technique. Compared with the conventional solid-state reaction method, the hydrothermal-based methods provides some advantages: the samples are obtained following a faster procedure, the number of high-temperature treatment events is decreased to one, and the hydrothermal/solvothermal methods open new directions for future research.

From the synthesis perspective, the two hydrothermal-based methods vary in the type of solvent used, the type of precipitation agent used, and the method of controlling the Li reactions. The hydrothermal method developed for this study consists of two stages. In the first stage, a La-Ti-(oxide/hydroxide) precursor is prepared using distilled water as a solvent and KOH as the precipitation agent. Lithium is added as Li₂CO₃ to the precursor at a second stage. The solvothermal method consists of a single stage. A Li-La-Ti-(oxide/hydroxide) precursor is prepared using ethanol as a solvent, and no precipitation agent is required.

The presence of Li during the solvothermal process has a strong impact on the properties of the sample at the end of the synthesis process (after the heat treatment).

Everything else being equal (i.e., under the same heat treatment), the samples prepared by the solvothermal method show the lowest Li content. They are also characterized by some porosity and the presence of small amounts of TiO₂ as a secondary product. The main LLTO phase is orthorhombic. The formation of the orthorhombic phase is known to occur at low Li concentrations [3]. According to the literature [23], the presence of TiO₂ (identified by XRD and suggested by EDX) and the porosity observed by SEM could indicate an incomplete calcination reaction. The Li losses in the solution are the most significant limitation of the method.

By contrast, the samples prepared by the hydrothermal method show a high Li content and low porosity. However, they still contain small amounts of impurities. The main LLTO phase appears to be tetragonal, but a secondary, cubic LLTO phase could also be formed.

The ionic conductivity of the samples obtained by the two hydrothermal-based methods is lower than the ionic conductivity of the control material (LLTO synthesized by the solid-state reaction). This result is expected for the solvothermal samples due to the low Li content, orthorhombic crystal symmetry, the porosity of the sample, and the presence of TiO₂. The lower conductivity could also be explained for the hydrothermal samples by the crystallization of LLTO in a tetragonal system and the presence of other impurities.