Effect of Sr and Nb co-doped on the total conductivity of Li0.5La0.5TiO3 ceramic electrolytes

The solid electrolyte Li0.5La0.5TiO3 is widely used in solid-state batteries due to its high grain conductivity at room temperature (RT). However, the ever-increasing requirement for high ionic conductivity necessitates the improvement of its total conductivity. In this study, tuning the total conductivity of perovskite-type La2/3-xLi3xTiO3 (LLTO) membranes through a co-doping strategy was proposed and systematically investigated, Here, an Li0.5−y+x Sr x La0.5−x Ti1−y Nb y O3 (LLSTN, x = 0, 0.04, 0.06, 0.08, and 0.1 mol%) ceramic solid electrolyte was prepared via the conventional solid-state reaction method. The introduction of Sr2+ and Nb5+ increases the lithium-ion vacancies and transforms the Li0.5La0.5TiO3 crystal structure from tetragonal to cubic. On the other hand, the lattice constant becomes larger, causing the migration channel of the lithium ions to become larger. Meanwhile, with the increase of Nb5+ doping amount, lithium lanthanum niobate forms between grains, inhibiting grain growth and it helps to reduce the resistance of lithium ion migration at grain boundaries. The total conductivity of sample Li0.5La0.42Sr0.08Ti0.92Nb0.08O3 reaches 5.10 × 10−5S·cm−1 at RT, which is about six times higher than that of the undoped sample, and the activation energy is 0.28 eV.


Introduction
All-solid-state lithium-ion batteries (ASSLBs) have the same working principle as traditional lithium-ion batteries, but liquid electrolytes and separators are replaced by solid electrolytes.ASSLBs are widely used in energy storage devices, sensors, and electrochemical devices.Inorganic solid electrolytes (ISEs) have attracted considerable attention due to their non-flammability, excellent thermal stability, and good mechanical properties [1][2][3][4][5].The use of inorganic solid electrolytes (ISEs) can replace flammable liquid electrolytes and is considered an important method to simultaneously improve the energy density and safety of lithium batteries.Various inorganic solid electrolytes have developed rapidly in the past decade, such as NASICON-type, Halide electrolyte, Garnet electrolyte, Sulfide electrolytes, perovskites, and their inorganic polymer composites.Despite there have been encouraging advancements, the progress in commercializing the ASSBs has been hindered [6][7][8].Sulfide electrolytes have a poor air stability and a narrow electrochemical window, and halide electrolytes have a poor electrochemical performance at room temperature (RT).On the other hand, oxide electrolytes have a wide electrochemical window (up to 5 V) and also show excellent electrochemical stability [9].The Li 3x La 2/3−x TiO 3 (LLTO) electrolyte is a common oxide solid electrolyte that has a high grain conductivity and a remarkable air stability at RT.The research community believes that LLTO has great potential for application within the field of solid electrolytes [10].The impedance of LLTO originates from both the crystal grains and the grain boundaries.The grain boundary impedance is usually 1-2 orders of magnitude greater than that of the grains.Therefore, ion diffusion at the grain boundaries has a key role in the whole diffusion process [11].Both optimizing the crystal structure and doping LLTO with suitable elements are effective strategies to increase its ionic conductivity.Zhang increased the LLTO lattice size by doping it with Sr 2+ , while Sr doping leads to a transition from the tetragonal to the cubic structure, which results in an increase in electrical conductivity [12].
In addition, it is experimentally demonstrated that doping Nb 5+ on the B site is helpful in reducing the amount and size of pores in LLTO-based ceramic pellets [13].
In this study, Li 0.5−y+x Sr x La 0.5−x Ti 1−y Nb y O 3 samples (LLSTN, x = 0, 0.04, 0.06, 0.08, and 0.1 mol%) were prepared via the solid-state method.This paper investigates Sr and Nb dopants effect on the microstructure and lattice structure of LLTO.when x = 0.08, the total conductivity reached as high as 5.10 × 10 −5 S•cm −1 at room temperature, being 6 times of that with x = 0.08.

Experimental
The samples with x = 0, 0.04, 0.06, 0.08, and 0.1 mol% are denoted as LLSTN1, LLSTN2, LLSTN3, LLSTN4, and LLSTN5, respectively.All LLSTN samples were synthesized via the solid-state method.The raw materials used in the experiment were Li 2 CO 3 (99.9%),TiO 2 (99.9%),La 2 O 3 (99.9%),Nb 2 O 5 (99.9%), and SrCO 3 (99.9%).The oxide powders were weighed according to the stoichiometric proportions of LLSTN.An excess of Li 2 CO 3 (10 wt%) was added to compensate for the lithium loss during high-temperature calcination and sintering.Alcohol was used as the grinding medium, and a QM-3SP2 planetary ball mill was used for ball milling for 26 h.The powders were dried in 50 °C.The powders were then calcined at 900 °C for 3 h in air, and they were subjected to a second ball milling process.The calcined powders were then ball milled for 24 h and dried again.The dried powders were then mixed with 5 wt% paraffin wax and pressed into tablets of f 12 mm × 1.2 mm at 10 MPa.The ceramic sheets were finally sintered at 1200 °C for 6 h in air.After polishing both sides of the pellets by an emery paper (No. 2000), the samples were washed in ethanol via an ultrasonic cleaner and dried at 70 °C.Silver paste was sputtered on both sides of the pellet surface.The Ag/LLSTN electrolyte /Ag symmetric cells were fabricated by deposition of Ag film on both sides of the LLSTN electrolyte.The Ag film on each side of the membrane was deposited at 25 mA for 60 s through magnetron sputtering.Then, heat the Ag/ LLSTN electrolyte /Ag symmetric cells at 700 °C for 30 min to tightly bond the silver electrode with the electrolyte.The crystal structure of the ceramic sheets (after polished by 2000 mesh sandpaper) were characterized via x-ray diffraction (XRD).The microstructure of sintered ceramic pellets was observed by field emission scanning electron microscopy (SEM).The valence state of Ti and O element was analyzed via x-ray photoelectron spectroscopy (XPS).Impedance data measurement was conducted using an electrochemical workstation and the impedance diagram can be fitted through equivalent circuits using Zview software (Scribner Associates Inc.).The CHI660E (Shanghai Chenhua Instruments Limited) electrochemical workstation can also be used to measure the constant potential polarization curve of electrolytes (the relationship between current and time).

Results and discussion
The XRD patterns of each sample sintered at 1200 °C for 6 h are shown in figure 1(a).The diffraction peaks match well with those of the cubic phase (PDF#46-0466), while a few peaks match those of the tetragonal phase (PDF#087-0935).Some diffraction peaks corresponding to Li 0.037 La 0.31 NbO 3 (LLNO) can also be seen.Sr doping leads to a transition of the LLTO structure from tetragonal to cubic, and the two phases coexist when this transition is incomplete.As the doping amount increases, the intensity of the peak corresponding to the tetragonal phase gradually decreases.Figure 1  ceramics in the 2q range of 32°-34°.The diffraction peaks of samples LLSTN3 and LLSTN4 in this range are slightly shifted toward a lower angle, indicating the expansion of the lattice.This is because the introduced Sr 2+ ions occupy the crystallographic A-sites, which means that the Sr 2+ ions partially replace the La 3+ ions.Since the ionic radius of Sr 2+ (1.44 Å) is larger than that of La 3+ (1.36 Å) and the ionic radius of Nb 5+ (0.64 Å) is larger than that of Ti 4+ (0.604 Å) [14], when Sr 2+ and Nb 2+ are successfully incorporated into the LLTO lattice, the diffraction peak of 33°is slightly shifted toward a lower angle.The LLSTN sample diffraction peak is refined using a cubic phase with a PM-3M space group and a tetragonal phase with a P4/MMM space group.The refined results are shown in figure 2 and table 1.The parameters obtained from the Rietveld refinement of the XRD pattens are listed in table 1, and the low values of the Rietveld refinement parameters (R P , R wp ) confirm the accuracy of the fit.The structure of the samples after the introduction of Sr 2+ and Nb 5+ belongs to two space groups, namely PM-3M and P4/MMM, with a slight increase in the lattice constant with increasing doping amount.The increase in the lattice constant makes the lithium-ion migration channel larger.The ratios of the proportions of the cubic and tetragonal phases for each sample are also presented in table 1, and it can be seen that the proportion of the cubic phase increases with increasing doping amount, the cubic phase content of samples LLSTN3 and LLSTN4 is high.It can also be clearly seen from figure 2(f) that changes in the doping amount of elements will cause changes in the content of cubic and tetragonal phases in LLSTN.The higher conductivity of the cubic phase compared with the tetragonal phase has been demonstrated in a previous study [15].
In this article, we adopted the mutual-compensating Li-loss (MCLL) sintering method proposed by Wang et al [16].MCLL is an efficient method is proposed to effectively control the Li 2 O atmosphere during the sintering process.Therefore, it can be seen from figure 3 that the microstructure of the ceramic electrolytes prepared via this method is dense compared with that reported in [13,17], as well as other works.Figure 3(a)-(e) show SEM images of LLSTN1, LLSTN2, LLSTN3, LLSTN4, and LLSTN5, respectively.The average grain size is shown in figure 3(f).The grains of sample LLSTN1 are mostly lamellar and disordered, which results in more voids being generated between the grains.From the figure 3(f), it can be seen that the introduction of doping elements can reduce grain size.This may be due to the low melting point of Nb 2 O 5 , the introduction of an appropriate amount of Nb 2 O 5 can lower the energy barrier for the diffusion of ceramic particles and facilitate the densification of the ceramics [18].The grains have a square shape and are tightly packed.However, there are several impurities between the grains, and these impurities can pin the grain boundaries and inhibit grain growth.Considering the high number of intergranular impurities in sample LLSTN2 and LLSTN5, Energy Dispersive Spectrometer (EDS) analysis was conducted on samples LLSTN2 and LLSTN5, the results are shown in figure 4. The EDS spectra reveal that these fine impurities do not contain the Sr element; indeed, there are only three elements in the samples: O, Nb, and La.This finding combined with the XRD results indicates that these impurities may be Li 0.037 La 0.31 NbO 3 (LLNO), the calculated quantities of these elements are basically the same as those of LLNO).LLNO is a lithium-ion conductor, a research has shown there is no the significant Laenrichment at the grain boundary in LLNO, and it is beneficial lithium-ion migration, which in turn increases the conductivity [17,19,20].
Figure 5(a) shows the XPS spectrum of Ti.There are two main peaks: Ti 2p3/2 at 457 eV and Ti 2p1/2 at 463 eV.Upon doping with Sr 2+ and Nb 5+ , the Ti 2p3/2 peak shifts toward a lower binding energy, and the chemical environment around the Ti element changes [21], indicating the successful incorporation of Nb 5+ into the LLTO lattice.Figures 5(b)-(f) show the XPS spectra of oxygen.The O1s scanA peak is located at around 529 eV, which corresponds to lattice oxygen.The O1s scanB peak is located at around 531 eV, which corresponds to vacancy oxygen.In figure 5(b), the ratio of the area of the O1s scanB peak to that of the O1s scanA peak is about 0.6, which indicates that there is more lattice oxygen in LLSTN1.Upon increasing the Sr 2+ and Nb 5+ amounts, the area of the O1s scanB peak gradually increases, and the ratios of the areas of the O1s scanB and scanA peaks where L is the thickness of the ceramic sample, A is the electrode area of the sample, and R is the impedance of the sample.Figure 6(e) shows the equivalent circuit used for the fitting [24], where R b and R gb represent the grain impedance and grain boundary impedance, respectively, and CPE gb represents the constant phase element of the grain boundary response at high frequencies.The straight line at low frequencies represents the diffusion of lithium ions, which is considered as a semi-infinite diffusion process and is modeled with Warburg elements.
plots marked in (a)], (c) temperature Arrhenius plot of the conductivity of the LLSTN ceramics, (d) relationship between the ionic conductivity and the activation energy of the LLSTN ceramics at RT, and (e) equivalent circuit for the impedance spectroscopy of the LLSTN ceramics.
From table 2 alongside figure 6(a) and (b), it can be seen that the grain impedance increases with increasing doping amount, but the grain boundary impedance is greatly decreased to all sample, thus the total impedance decreases.Among the investigated samples, sample LLSTN4 has the lowest total impedance R total = 2798 Ω and the highest total conductivity σ total = 5.10 × 10 −5 S•cm −1 .However, as the doping amount increases, the conductivity decreases.Excessive quantities of Sr and Nb cannot enter the lattice, which prevents the migration  of Li ions and also causes a reduction in the number of A-site vacancies; thus, the overall ionic conductivity decreases [12].Therefore, the conductivity of sample LLSTN5 decreases.Figure 6(c) shows the impedance values of the electrolyte measured in the temperature (T) range of 25 °C-100 °C for all samples.The logarithm of the obtained conductivity is plotted as a function of 1000/T and then linearly fitted.The figure shows that all curves have the same trend.The activation energy (E a ) is then calculated using the Arrhenius equation [22]: where σ 0 is a pre-factor, and K is the Boltzmann constant.The Arrhenius equation indicates that the relaxation of the electrolyte lattice increases with increasing temperature and the ion mobility increases, so the activation energy decreases [25].The activation energies obtained for each sample by fitting the corresponding data are listed in table 2. It can be seen that sample LLSTN4 has the lowest E a (0.28 eV).
Decrease in E a indicates that more ions migrate or that the migration of ions is promoted.Based on the analysis of the results in table 1 and figure 6, it can be seen that the cubic phase content in LLSTN3 and LLSTN4 is is composed of an alternate stacking of La-rich and La-poor layers in the c-axis [26].The diffusion of Li ions in the La-poor layer is not affected, but is hindered by La cations in the La-rich layer.Hu et al have demonstrated that the conductivity of Li ions in cubic structures [figure 7 (a)] is higher than that in tetragonal structures.So the lithium ion conductivity of samples LLSTN3 and LLSTN4 is higher than that of other samples, and their activation energy is low.In samples LLSTN3 and LLSTN4, the content of cubic structure is similar, but the lattice  Table 2. Electrochemical properties of the LLSTN samples(R b is the grain impedance, R gb is grain boundary impedance, σ b is grain conductivity, σ gb is grain boundary conductivity, σ total is total conductivity, E a is activation energy and σ e is electron conductivity).parameters of tetragonal structure are different [27].The LLSTN4 has larger lattice parameters (a = b = 3.8952 Å, c = 7.7791 Å), thus increasing the bottleneck of lithium-ion migration (The bottleneck for lithium transfer consists of four O atoms in adjacent four Ti octahedra).So overall, the total Li ion conductivity of sample LLSTN4 is high (σ total = 5.10 × 10 −5 S•cm −1 ) and the activation energy is low (E a = 0.28 eV).

Sample
Figure 8 shows the Potentiostatic polarization curve for each sample under a DC polarization of 2 V over a period of 1800s.Using a silver electrode to block the migration of lithium ions, only electrons can pass under a voltage applied over a long period of time, and the electronic conductivity of the electrolyte can be measured using this method [25,28].It can be seen from figure 6 that the current of each sample does not initially reach a stable value when the voltage is applied, and the initial currents of samples LLSTN3 and LLSTN4 are small.After 1200 s, the currents stabilize at about 1 nA.The electronic conductivity (σ e ) of the electrolyte is calculated from the current values in the steady state of the electrolyte according to [28][29][30]: where D is the thickness of the electrolyte, I is the current after stabilization, and U is the applied voltage.The electronic conductivities obtained from all the calculations are listed in table 2. It can be seen from table 2 that the electronic conductivities of the LLSTN samples are all on the order of 10 −9 , which is four orders of magnitude lower than the ionic conductivity, indicating that the electrolytes prepared using the proposed method are all good ionic conductors.

Conclusions
This article proposes a co-doping strategy to introduce Sr and Nb into Li 0.5 La 0.5 TiO 3 .The effect of Sr and Nb codoping on the crystal structure, microstructure, and electrical properties of Li 0.5 La 0.5 TiO 3 was investigated.Among the investigated samples, the conductivity of sample LLSTN4 reached 5.10 × 10 −5 S•cm −1 , which is about 6 times higher than that of the undoped sample.Additionally, the activation energy of this sample is about 0.28 eV, and its electronic conductivity is 4.27 × 10 −9 S•cm −1 .What'smore, it was found that Ta and Nb doping is effective to increase conductivity and decrease activation energy, which is attributed to an increase in cubic phase content, Li 0.037 La 0.31 NbO 3 (LLNO) inhibits grain growth and reduces grain boundary impedance,larger Li-ion transmission channels and more Li-ion vacancies.
(b) shows a magnified view of the XRD patterns of the LLSTN

are 2 .
1 and 3.8 in LLSTN4 and LLSTN5, respectively.On the one hand, this is because the substitution of Ti 4+ with Nb 5+ causes a change in the Ti-O bond length.On the other hand, the lattice mismatch at the LLTO domain boundaries causes a lack of oxygen in the lattice[22].

Figure 6 (
a) shows the impedance spectra of the different samples at RT in the frequency range of 0.1 Hz-1 MHz.The experimental data were fitted using the ZView software, and the equivalent circuits are shown in figure 6(e).All the curves show two arcs; the first arc corresponds to the grain boundary impedance at low frequencies, as shown in figure 6(a), while the second arc corresponds to the grain impedance at high frequencies, as shown in figure 6(b).The conductivity (σ) of the samples can be calculated according to the following equation [23]:

Figure 6 .
Figure 6.(a) Grain boundary impedance plots of the LLSTN samples at RT, (b) grain impedance plots of the LLSTN samples at RT [the corresponding enlarged impedance.

Table 1 .
Rietveld refinement parameters of the LLSTN samples.