Introduction

Among the electrochemical power sources, one of the leading positions belongs to the lithium-ion batteries (LIBs). Compared with most other kinds of batteries, LIBs have a higher output voltage and higher energy storage density. Of special interest is development of all-solid-state lithium-ion batteries using solid lithium-conducting electrolytes instead of traditional liquid organic or polymer electrolytes. The major advantages of all-solid-state lithium-ion batteries are simple design and, the main thing, safety. According to several works [1], promising solid electrolytes for medium- and low-temperature all-solid-state lithium-ion batteries are LLZ (Li7La3Zr2O12, σ300 K = 2.1 × 10−4 S × cm−1) [2], LISICON (Li14ZnGe4O16, σ300 K = 10−7 S × cm−1) [3], NASICON-like Li1 + хAlхTi2 − х(PO4)3 (σ300 K = 7.0 × 10−4 S × cm−1) [4], and glass-ceramic Li1.6Al0.6Ge1.4(PO4)3 (σ300 K = 4.0 × 10−4 S × cm−1) [5]. Unfortunately, most of the above materials have high Li+ ion conductivity but are unstable in contact with metal lithium. Binary compounds in the system Li2O-ZrO2 are free from this disadvantage [6, 7] but their conductivity is too small. Therefore, investigations focused on raising the ionic conductivity of lithium zirconates are of great importance.

Among lithium zirconates, the maximum value of conductivity (10−4–10−5 S × cm−1 at 570 К) is observed for Li8ZrO6 [8, 9].

The first crystal structure and transport properties of Li8ZrO6 and Li7LO6 (L = Nb, Ta), which have close resemblance, were studied in work [8]. All of these compounds have layered structure derivative of CdI2. Crystal lattice of both Li8ZrO6 and Li7LO6 consists of layers containing ions of Zr4+ (L5+) and 2 of 8 Li+ cations in octahedral positions, the remaining 6 Li+ ions occupy half of tetrahedral sites between layers. Composition of layers may be described as [(Li2Zr)O6]n in case of Li8ZrO6 or as [(Li□L)O6]n in case of Li7LO6. As one can see, Li7LO6 compounds contain structural lithium vacancies (□) in octahedral positions; in Li8ZrO6 such vacancies are absent. Due to the presence of the lithium vacancies, lithium ion conductivity of Li7LO6 is considerably higher as compared to Li8ZrO6, which has ordered crystal structure. Therefore, one can expect that the introduction of additives creating vacancies in the lithium sublattice will increase Li+ ion conductivity of Li8ZrO6. Actually, the increase of conductivity is observed at substitution of Li+ by Mg2+ or Sr2+ according to Eq. 1 [10] and at substitution of Zr4+ by Nb5+ or V5+ according to Eq. 2 [11, 12].

$$ 2{Li}^{+}\to Mg{(Sr)}^{2+}+{V}_{Li} $$
(1)
$$ {Li}^{+}+{Zr}^{4+}\to Nb{(V)}^{5+}+{V}_{Li} $$
(2)

Introduction of additives of another type such as substitution of Zr4+ by triple-charged cations Al3+, Y3+, and Sc3+ with charge compensation by interstitial Li+ ions is not followed by conductivity rise [13,14,15]. It is of interest that a considerable increase of conductivity takes place at isovalent substitution of Zr4+ by Ce4+ [16].

In this paper, the values of conductivity in the solid solutions Li8 − 2xMexZrO6 (Me = Mg, Sr), Li8 − xZr1 − xAxO6 (A = V, Nb), and Li8Zr1 − xCexO6 are compared, and temperature dependences of conductivity of these solid solutions are analyzed. A number of samples were examined by Raman spectroscopy and nuclear magnetic resonance (NMR) spectroscopy. Based on the data obtained, conclusions about factors determining high conductivity of the materials under consideration are made.

Experimental

According to work [11], the range of solid solutions based on Li8ZrO6 in the system Li8 − xZr1 − xVxO6 lies at x < 0.05. The data of the present study show that single-phase region boundary in this system corresponds to x ≈ 0.02. Furthermore, Li7.98Zr0.98V0.02O6 solid solution has the highest conductivity in Li8 − xZr1 − xVxO6 system. So, to compare the values of conductivity in other systems with heterovalent substitutions, we chose compositions with the same value x = 0.02, that is, with the equal concentration of lithium vacancies: Li7.960.02Me0.02ZrO6 (Me = Mg, Sr) and Li7.980.02Zr0.98A0.02O6 (А = V, Nb). All these compositions lie in the single-phase regions of the corresponding systems and have conductivity close to maximal value. A sample containing niobium was synthesized and studied in this work. The data on conductivity of Mg- and Sr-containing solid solutions are taken from several works [17, 18]. In the system Li8Zr1 − xCexO6 as well as in the systems with heterovalent dopants, a sample with x = 0.02 was studied.

Li2CO3 (high purity) and V2O5, Nb2O5, CeO2, and ZrO2 (analytical grade) were used as starting materials for the solid-state synthesis of Li8Zr0.98Ce0.02O6, L7.98Zr0.98Nb0.02O6, and Li7.98Zr0.98V0.02O6. Zirconium, niobium, and cerium oxides were pre-dried at 1270 K to a constant weight. The starting materials were mixed in a stoichiometric ratio, except for Li2CO3, which was added with 5–8 wt% surplus. The syntheses were carried out in glass carbon crucibles under vacuum in the hermetic heatproof steel reactor. The synthesis was carried out by stepwise increasing the temperature from 573 to 1253 K. Then, the samples were annealed in the air at 873 K for 6 h to remove traces of carbon.

The crystal structure of the products obtained was determined by the X-ray diffraction (XRD) after each stage of the synthesis. The XRD analysis was carried out using Rigaku D-MAX-2200 V diffractometer in CuKα-radiation, 2θ = 10°–80°. XRD patterns of compounds were compared with database PDF-2 [19] and literature data.

Chemical analysis of synthesized samples was performed using inductively coupled plasma atomic emission spectrometer OPTIMA 4300 DV (Perkin Elmer, USA).

Raman spectra of the polycrystalline samples were collected with Raman microscope-spectrometer RENISHAW U-1000 (Ar+-radiation, λ = 514 nm) at 50–4000 сm−1.

Thermal analysis of the materials obtained was performed using synchronous thermal analyzer STA 449 F1 Jupiter over the temperature range 300–870 K in alundum crucibles at the rate of heating 10 K min−1 in the air. The mass of the samples was 10–20 mg. The data obtained were processed by means of NETZSCH Proteus software.

NMR experiments were carried out using an updated Bruker SXP 4-100 pulsed NMR spectrometer at resonance frequency 194.37 MHz in the temperature range 300–670 K. NMR spectra and the spin-lattice relaxation time of 7Li nuclei were measured by the Fourier transformation of the signal of the free induction decay.

The samples for conductivity measuring with diameter 6 mm and thickness 1–2 mm were obtained by pressing the powders of particle size < 50 μm in the steel die at 70 MPa. Then the samples were covered by the powder of the same composition and sintered under helium in the heatproof steel reactor for 5 h at 1179 К. The density of the samples calculated from the geometrical parameters and mass was 90–95% of the theoretical one.

For AC impedance studies, both sides of the sintered samples were painted with silver paste. The impedance measurements were performed in the air at temperatures ranging from 25 to 600 °C, using an immittance meter E7-25 (MNIPI) in the frequency range from 20 Hz to 1 MHz. The results were fitted by using special software program (EQUIVCRT.COM). All of the conductivity measurements were performed on two sets of identical samples to check the reproducibility of the results.

Electron conductivity was measured at direct current using blocking Ag [15] electrodes in the temperature range 470 to 870 K. For this purpose, direct voltage which was less than solid electrolyte decomposition voltage (20–40 mV) was applied to Ag│solid electrolyte│Ag cell. After a time, the current acquired a fixed value. Under these conditions, ionic current is blocked and the charge is carried by electrons and holes. Knowing the current value and applied voltage, we can calculate the electron conductivity.

Results and discussion

Ionic conductivity and DSC

Figure 1 shows XRD patterns of the samples studied. It is seen that in all cases, diffraction diagram corresponds to the Li8ZrO6 phase (PDF2 No. 26-0867 card-index data).

Fig. 1
figure 1

XRD patterns of (1) Li8ZrO6 PDF2 # 26-0867 and (2) undoped Li8ZrO6 phase and (3) Li7.98Zr0.98V0.02O6, (4) Li7.98Zr0.98Nb0.02O6, (5) Li8Zr0.98Ce0.02O6, (6) Li7.96Sr0.02ZrO6, and (7) Li7.96Mg0.02ZrO6 solid solutions

Temperature dependences of conductivity for Li8ZrO6 and solid solutions are similar (Fig. 2). Low- and high-temperature linear regions are separated at 680–710 K by a transition section, which is characterized by a sharp increase in conductivity with increasing temperature. Electron conductivity value of both undoped Li8ZrO6 and solid solutions in all cases was by three orders of magnitude less than ionic one. Therefore, further on when discussing the synthesized material conductivity, we mean their ionic conductivity. DSC measurements of Li8ZrO6 show the existence of endothermic peak at 680 K (Fig. 3). The enthalpy of effect is − 23.6 J g−1. DSС curves of the doped samples also show endothermic peaks in the range 680–700 K. In all cases, the temperature of thermal effect is close to the temperature of the drastic change in conductivity. It gives reason to suggest that at 680–710 K, the materials examined undergo a phase transition followed by lithium sublattice disorder [10]. However, there is another probable interpretation of Li8ZrO6 and its solid solutions’ conductivity behavior. The fact is that binary compounds of the Li2O–MxOy systems with the high content of lithium oxide, such as Li5MO4 (M = Al, Ga [20], Fe [21], Tl [22]), Li6ZnO4 [20], Li7LO6 (L = Bi, Sb, Ta) [23], and also Li8ZrO6 [24], are highly hygroscopic and interact with ambient air moisture. For the first time, influence of this effect on conductivity was found for Li5AlO4. The first conductivity measurements of Li5AlO4 as well as of Li5GaO4 and Li6ZnO4 showed [20] that at heating up to ~ 680 K, the conductivity of these materials increases jump-like achieving 10−1 S × cm−1 but activation energy decreases sharply. At first, such a behavior as in the case of Li8ZrO6 was explained by lithium sublattice disorder [23]. However, later, it was shown through the example of Li5AlO4 that the jump-like increase of conductivity in this case is due to the melting of LiOH, which appears as a result of interaction between Li5AlO4 and atmospheric moisture [25, 26]. If one compares temperature dependences of conductivity for Li5AlO4, Li5FeO4, Li5TlO4, Li8ZrO6, and its solid solutions (Fig. 2) and Li7LO6 (L = Bi, Sb, Ta) [23], it can readily be seen that in all cases, lgσ − 1/T curves demonstrate the same peculiarities. First, one observes a sharp increase in conductivity and decrease in activation energy, and this at similar temperatures which are close to the melting point of LiOH (680–710 K). Second, in all cases, DTA or DSC measurements indicate an endothermic peak in the same temperature region. Third, the presence of a jump in conductivity and its temperature does not depend on introduction of additives, their nature, and concentration. The assumption about the absence of phase transition in Li8ZrO6 near 700 K is supported by the fact that high-temperature XRD study does not show a change of the crystal structure, which was to be expected at disorder of cation sublattice. The X-ray pattern obtained at 893 K [27] is identical to that obtained at room temperature [28].

Fig. 2
figure 2

Arrhenius plots for (1) Li8ZrO6 and (2) Li8Zr0.98Ce0.02O6, and (3) Li7.96Sr0.02ZrO6, (4) Li7.98Zr0.98Nb0.02O6, (5) Li7.96Mg0.02ZrO6, and (6) Li7.98Zr0.98V0.02O6 solid solutions

Fig. 3
figure 3

DSC plot of Li8ZrO6: heating rate 10 K min−1

Raman spectroscopy

Table 1 shows the observed bands of the Raman spectra for Li8ZrO6 and some of the solid solutions synthesized in this work as compared to the literature data. One can see that the spectra of Li8ZrO6 and its solid solutions contain a number of extra bands, which were not observed for Li8ZrO6 in work [23]. However, most of these extra bands may be referred to Li2CO3, LiOH, and LiOH·H2O [29,30,31]. The formation of LiOH as a result of interaction between Li8ZrO6 and water vapor according to the reaction

$$ {Li}_8{ZrO}_6+2{H}_2O\to {Li}_4{ZrO}_4+4 LiOH $$
(3)

was experimentally proved in work [24]. It seems that the formation of LiOH·H2O and Li2CO3 is a result of the further interaction of LiOH and atmosphere moisture and CO2. Therefore, it is reasonable to assume that the anomalous behavior of conductivity of Li8ZrO6 and its solid solutions may be due to the formation of LiOH, which melts at elevated temperatures as in the case of Li5AlO4. As concerns Li7LO6 (L = Nb, Ta, Sb, Bi), the authors [23] do not find in the samples studied any interaction with atmosphere products, but do not register any change in crystal lattice at temperatures above 700 K either. Therefore, the point of sharp increase in conductivity of the materials under discussion when heated to ~ 700 K remains unclear.

Table 1 Observed absorption bands (cm−1) in the Raman spectra for Li8ZrO6 and solid solutions Li7.98Zr0.98V0.02O6, Li7.98Zr0.98Nb0.02O6, and Li8Zr0.98Ce0.02O6 as compared to literature data

NMR spectroscopy

7Li NMR experiments were performed for Li7.98Zr0.98V0.02O6 solid solution in the temperature range 300–750 K (Fig. 4). Modeling of the spectrum shape shows that spectral line can be described as superposition of two lines which correspond to different structural positions of lithium Li1 and Li2. Such positions in solid solutions with Li8ZrO6 structure are represented by octahedral positions in layers and tetrahedral ones between layers. 7Li NMR spectra for undoped Li8ZrO6 are similar [9].

Fig. 4
figure 4

7Li NMR spectra as a function of temperature for Li7.98Zr0.98V0.02O6

Figure 5 shows 7Li NMR line width of the central transition as function of temperature. As one can see, at temperature increase, the width of 7Li NMR line is gradually reduced. Such behavior of NMR spectra is typical for systems with fast ion transport. Sharp narrowing of NMR line takes place when characteristic frequency of diffusion τd−1 approaches the width of the NMR line of rigid lattice. For Li7.98Zr0.98V0.02O6 solid solution, this temperature area is ~ 450–600 K (Fig. 5). Earlier, close temperature interval of NMR line narrowing was observed for undoped Li8ZrO6 [9]. It should be pointed out that for Li7LO6, which has a similar crystal structure but contains structural lithium vacancies close to 7LiNMR, the line is observed at essentially lower temperatures 200–400 K [23], which correlates with higher Li+ conductivity of Li7LO6 as compared to Li8ZrO6.

Fig. 5
figure 5

7Li NMR line width (full width at half-height) as a function of temperature for Li7.98Zr0.98V0.02O6

Activation energy for jumps of Li+ ions in solid solution Li7.98Zr0.98V0.02O6 calculated from NMR data at low temperatures is equal to 44 kJ × mol−1, whereas the value obtained from the temperature dependence of conductivity is more than twice as large (99 kJ × mol−1). It suggests that the total mobility of lithium is to a great degree due to the local jumps of Li+.

The results of the transport property study of materials under discussion are summarized in Table 2. One can see that introduction of dopants in all cases results in conductivity increase and activation energy lowering as compared to undoped Li8ZrO6. In case of systems with two- and fifth-charged dopants, it is due to formation of lithium vacancies. Unlike Li7LO6, charge carriers in Li8ZrO6 are thermal activated defects rather than structural vacancies. At substitution of Li+ or Zr4+ ions by M2+ or A5+ ions respectively, the concentration of lithium vacancies drastically increases but activation energy decreases by a value of defect formation. Both factors lead to the increase of conductivity. The growth of conductivity at isovalent substitution of Zr4+ by Ce4+ seems to be a result of geometrical changes. As ion radius of Ce4+ is essentially larger compared to Zr4+ (0.87 and 0.76 Å respectively) [32], the substitution of Ce4+ for Zr4+ should lead to broadening of lithium migration channels and lowering of the average value of potential barrier which separates an occupied Li+ position in the crystal lattice from a vacant one.

Table 2 Ionic conductivities and the activation energies for Li8ZrO6 and solid solutions Li7.98Zr0.98A0.02O6 (A = V, Nb), Li7.96Me0.02ZrO6 (Me = Mg, Sr), and Li8Zr1 − xCexO6

It should be noted, however, that conductivity increase of Li8ZrO6 at introduction of dopants in low-temperature region in all cases is not very large. Maximal value of conductivity is observed for Li7.98Zr0.98V0.02O6 and is equal to 8.9·10−6 S × cm−1 at 476 К (Table 2). That is only six times as large as compared to undoped Li8ZrO6. Rather low value of conductivity is the result of low solubility of the dopants and, as a consequence, low concentration of lithium vacancies. In Li7LO6, half of the octahedral Li+ positions are vacant, so its lithium ion conductivity is essentially higher and for Li7BiO6 achieves 3 × 10−3 S × cm−1 at 476 К [23].

In the high-temperature region, the conductivity of Li8ZrO6 and solid solutions based on it drastically increases and at 870 K achieves 10−1 S × cm−1 (Table 2). However, the reasons for such a behavior are not well established. A possible solution to this problem could be a precise neutron diffraction study of crystal structure in the temperature region of existence of supposed disordered form of Li8ZrO6 or an investigation into the dependence of conductivity as function of time at ~ 800 K at a change of dry atmosphere into moisture and vice versa as it was done in [33].

Conclusion

Temperature dependences of conductivity and Raman spectra of solid solutions based on lithium hexaoxozirconate Li7.96Me0.02ZrO6 (Me = Mg, Sr), Li7.98Zr0.98A0.02O6 (А = V, Nb), and Li7.98Zr0.98Ce0.02O6 are investigated. Ionic conductivity at doping increases as compared to Li8ZrO6; however, below ~ 680 K, it remains rather low and significantly smaller as compared to Li7LO6 (L = Nb, Ta, Sb, Bi), which have the similar crystal structure but contain structural lithium vacancies.

Conductivity of both undoped Li8ZrO6 and solid solutions shows a jump-like rise with increasing temperature at 680–710 K. Earlier, such a behavior was explained by a structural phase transition with disordering of the lithium sublattice. The data of Raman spectroscopy, however, shows that the melting of lithium hydroxide formed due to interaction of samples with atmospheric moisture may be another reason for a sharp growth of conductivity near 700 K.

The results of NMR experiments show that at low temperatures, the movement of lithium cations in the crystal lattice takes place via exchange between octahedral and tetrahedral positions.