Impedance spectra were analyzed by fitting the data using ZView electrochemical impedance spectroscopy (EIS) software (Version 3; available at [(https://www.scribner.com/software/68-general-electrochemistr376-zview-for-windows/)]). The Al/Li-Nb-O/Cu cells exhibited stable electrochemical performance, with no signs of lithium dendrite formation, instability, non-ideal behavior, or short-circuiting under low electric fields.
The structural properties, including the crystallinity of the thin films, were examined using X-ray diffraction (XRD) analysis. No distinct peaks were observed in the XRD patterns of the Li-Nb-O films (Fig. 3), confirming their amorphous structure. This observation is consistent with the expected amorphous nature of sputter-deposited thin films at room temperature, as reported in literature. During sputtering at low power, or without substrate heating or biasing, the low-energy particle impacts promotes growth along low-energy planes, yielding films with poor crystallinity. Recently, Rahn et al. reported that sputtered LiNbO thin films remained amorphous after annealing up to 350 °C and started to crystallize only at 500 °C. In this study, crystalline structure formation was achieved either by depositing thin films with substrate heating or through post-annealing; however, both methods resulted in lithium loss. Consequently, room-temperature deposition was chosen as the preferred approach. Previous studies have reported that single-crystal LiNbO exhibits low ionic conductivity (~ 10 S cm at 500 K), which limits its suitability as a solid-state electrolyte in LIBs. Amorphous LiNbO thin films are anticipated to exhibit higher ionic conductivity, primarily due to their inherent structural disorder and the absence of grain boundaries, which commonly impede ion transport in the crystalline lattice. Moreover, amorphous thin-film electrolytes provide excellent surface compatibility with electrodes, effectively reducing interfacial resistance. This reduction is crucial for mitigating short-circuit risks and enhancing safety of Li-ion batteries and capacitors. It is emphasized that amorphous layers are of great importance in various scientific and technological fields. They are preferred for application in transparent electronics, thin-film transistors, solar cells, memory devices, gravitational-wave detectors, and corrosion-resistant coatings.
Figure 4 presents FESEM images of the Li-Nb-O thin films. The surface morphology reveals a granular structure with the fine grains. Grain growth is evidently enhanced with increasing DC power, primarily due to two factors. First, higher DC power results in a greater number of target atoms being sputtered by Ar ion bombardment. Sputtered atoms with adequate kinetic energy can diffuse to favorable sites, facilitating effective atomic bonding and promoting grain growth. Secondly, the increase in substrate temperature due to high-energy bombardment promotes grain growth with well-defined boundaries.
Energy-dispersive X-ray (EDX) spectroscopy was conducted to analyze the elemental distribution and chemical composition of the Li-Nb-O thin films. The EDX detector cannot identify Li due to the inherently low intensity of its Kα X-ray emission line. As shown in Fig. 4, the EDX mapping results demonstrate a homogeneous distribution of the constituent elements, including niobium (Nb) and oxygen (O), across the films. This result indicates the absence of significant Nb agglomeration on surface, thereby promoting uniform material formation. Such uniformity is an essential criterion for consistent electrochemical performance and mechanical stability. Figure 5 displays the selected EDX spectrum for the Li-Nb-O sample deposited at a DC poswer of 80 W. Si peak is attributed to the glass substrate. The Nb-to-O atomic ratio was determined from the EDX measurements, yielding values in the range of 0.28-0.44, as presented in the inset table of Fig. 5. A systematic increase in the Nb/O ratio with increasing DC sputtering power was observed, reflecting the progressive increase in Nb content.
XPS analysis was used to characterize the elemental composition and oxidation states of the constituent species in Li-Nb-O films deposited at various DC powers. Figure 6a displays the XPS survey scan spectra of the films, confirming the presence of Li, Nb, O, and C, in agreement with previously reported results. No impurities were detected except carbon, attributed to surface contamination; the C 1s peak at 284.5 eV was used as an internal standard to calibrate all XPS spectra for charge compensation. XPS is a surface-sensitive method that quantifies elemental composition in the near-surface region of a sample. The lithium content, as one of the most critical parameters affecting ionic conductivity, was experimentally evaluated by XPS analysis. However, due to the weak Nb signal in XPS spectra, the Nb atomic ratio was further determined using EDX analysis. The combined application of EDX and XPS enabled a quantitative assessment of the film composition.
The atomic percentages of the constituent elements are summarized in the table presented in Fig. 6b. Examination of the relative Li and Nb content in the film surface reveals how maintaining a relatively high constant RF power on the Li target, while varying DC power applied to Nb target, affects the chemical composition. To clarity regarding the compositional region of the Li-Nb-O ternary system, the Li-to-Nb content ratio was calculated, yielding values in the range of 2.31-1.36. This indicates the successful achievement of a Li-rich compositional region. As shown in Fig. 6b, the Li/Nb atomic ratio exhibits a general decreasing trend with increasing DC power. This result is in agreement with the corresponding enhancement in Nb content.
Figure 6c presents the high-resolution Nb 3d core-level spectrum of the Li-Nb-O film deposited at a DC power of 80 W. This sample was selected for detailed study because films deposited at lower DC powers contain reduced niobium content, resulting in significantly weaker Nb 3d signal intensity. The spectrum exhibits a Nb 3d-Nb 3d doublet, with peak centers located at binding energies of 209.87 eV and 207.16 eV, respectively. These values with the spin-orbit splitting of approximately 2.7 eV are characteristic of Nb oxidation state and align well with values reported in the literature. In this study, the Nb 4s peak was not observed in the XPS spectra. This is due to the Li/Nb ratio exceeding 1.0, confirming the formation of a Li-rich phase.
Figure 6d displays the Li 1s core-level spectra of the films, each characterized by a single sharp and symmetric peak centered at ~ 54.78 eV. This indicates the monovalent chemical state of Li ions in all samples. The comparable intensities of the Li 1s XPS peaks observed across the films can be ascribed to the application of constant RF power to the Li target during sputtering.
Figure 7 presents the O 1s core-level spectra of th Li-Nb-O thin films. The O 1s peaks were deconvoluted to 530.61, 531.45, and 532.67 eV, corresponding to Nb-oxygen, Li-O, and Li-OH bonds, respectively. The presence of hydroxyl species is attributed to the absorption of moisture during exposure to the ambient atmosphere. Figure 7f depicts the variation in relative quantities of Nb-O and Li-O bonds as a function of DC sputtering power. The integrated area of the 530.61 eV peak gradually increased (3.19-11.51%) with rising DC power (20-80 W), indicating enhanced Nb-O bond formation. This enhancement is attributed to the higher Nb concentration achieved at increased sputtering power. In contrast, the fraction of O associated with Li-O bonds decreased from 79.81 to 70.52% over the same DC power range. This result suggests a greater contribution from weakly bonded Li ions, which predominantly govern ionic conductivity in disordered solids.
Impedance spectroscopy was conducted on the Al/Li-Nb-O/Cu configuration to evaluate the effect of co-sputtering parameters on carrier transport and charge storage. The schematic and cross-sectional FESEM image of this structure are shown in Fig. 8a, b, respectively. The contacts act as blocking electrodes for Li ions under an alternating electric field. The impedance response depends primarily on material type, synthesis method, and measurement temperature. Other factors, such as crystal structure, chemical composition, and grain boundary distribution, also influence the overall impedance behavior.
Figure 9a presents the a.c. conductivity of Li-Nb-O thin films measured at room temperature. Electrical conductivity in solid-state electrolytes is a thermally-activated process, governed by the migration of weakly bound ions under an applied electric field. Frequency dependence of σ(ω) for all Li-Nb-O films follows Jonscher's power law:
Here, σ is the dc conductivity, A is a pre-exponential factor related to polarizability, ω is the angular frequency, and n is an experimentally determined fitting parameter. The parameter n reflects the interaction degree between charge carriers and surrounding lattice. The exponent "n" is frequency-independent but can vary with the type of mobile ion, ion concentration, and system temperature. The universal dynamic response (UDR) in disordered solids is described by Jonscher's law.
A significant increase in σ(ω) was observed as the DC sputtering power increased from 20 to 30 W, followed by a slight increase between 40 and 80 W. Ionic transport primarily depends on factors such as ionic polarizability, Li-O binding strength, Li ion contribution, hydration radius, and vacancy density. Additionally, Li-Li Coulombic interactions and ion-electrode interactions play critical roles in overall transport behavior. Here, enhanced ionic conductivity is directly correlated with the increased number of sputtered Nb atoms due to higher DC sputtering power.
With increasing Nb concentration, Nb atoms exhibit a strong tendency to occupy Li sites. This behavior is primarily attributed to the similar ionic radii of Nb (0.64 Å) and Li (0.76 Å), as well as the difference in bond strengths within the Li-Nb-O system. The Nb-O covalent bond (E ~ 567.31 kJ/mol) is considerably stronger than the Li-O ionic bond (E ~ 392.50 kJ/mol), favoring Nb substitution at Li sites and the formation of Li vacancies. Substitution of monovalent Li by pentavalent Nb disrupts charge neutrality, promoting the formation of additional Li vacancies. These vacancies serve as diffusion sites, facilitating Li-ion transport within the Li-Nb-O framework. Moreover, the presence of Li vacancies reduces Li-Li electrostatic repulsion, thereby further enhancing ionic mobility. XPS analysis confirms predominant Nb-O bond formation and a reduction in the concentration of bonded Li ions with increasing DC power.
Measuring dc conductivity of electrolyte films is challenging due to the ionic polarization at low frequencies and bulk ionic transport at high frequencies. The σ value can approximated as σ(ω) at the frequency of peak in loss spectrum. This approximation agrees well with the results obtaine from fitting Nyquist plots (Z''-Z') using an equivalent circuit model. The room-temperature σ values for the co-sputtered Li-Nb-O thin films are listed in Table 2. The maximum σ value obtained was 1.33 × 10 S cm, comparable to previously reported results for LiNbO (Table 3). This value is even higher than that reported for material synthesized by other methods. For example, Wang et al. recently reported a σ of 6.39 × 10 S cm for Li-Nb-O thin films grown by ALD at a deposition temperature of 235 °C. This indicates the potential of co-sputtering to produce thin-film electrolytes with improved ionic conductivity. In this study, Li-Nb-O films provide moderate σ value, surpassing comparable amorphous systems such as Li-Ta-O (σ ~ 10 S cm) and Li-Al-O (σ ~ 2.85 × 10 S cm).
Figure 9b shows the capacitance-frequency curves of Li-Nb-O based MIM capacitors at room temperature. The low-frequency capacitance of the samples is presented in Table 2. The highest C value (625 nF cm) was achieved for film prepared at a DC power of 30 W. This behavior is directly related to the ionic conduction mechanism. Incorporation of Li ions into niobium oxide induces a complex process that enhances charge storage, associated with the electrode polarization effect. This is attributed to the formation of electric double layers (EDLs), resulting from the ion accumulation at the blocking electrolyte/electrode interfaces.
In the low-frequency range, Li ions are able to respond to external electric field and undergo long-range diffusion, leading to resistive behavior. With increasing frequency, sharp decline in capacitance is attributed to reduced effective ionic polarization. This behavior is due to the inability of Li ions to follow the rapidly oscillating electric field. At high frequencies, the plateau observed in C suggests that induced dipoles fail to respond to the electric field. This is related to the limited time available for dipolar rotation or ionic displacement.
The dielectric constant (k) of a metal-insulator-metal (MIM) capacitor can be calculated using the following formula:
Here, C represents the capacitance per unit area, ɛ is the vacuum permittivity, and d is the thickness of the dielectric layer. The obtained k values for the Li-Nb-O thin films are shown in Fig. 9d. These results indicate that all layers can be classified as ion-conducting dielectric. The maximum dielectric constant (k) was measured to be 141.21, significantly exceeding the values reported for materials synthesized via alternative methods, as summarized in Table 4. Notably, the Li-Nb-O thin films exhibit a higher dielectric constant compared to pure NbO (k ~ 47), confirming the contribution of the electric double layer (EDL) effect. Similarly, Nila Pal et al. reported an increase in dielectric constant of AlO from 8.8 to 35 upon the incorporation of Li ions into the film using a sol-gel process.
Dielectric loss, the ratio of dissipated to stored energy, can be calculated in terms of the real (ε') and imaginary (ε'') components of the permittivity as follows:
Figure 9 c presents room-temperature loss tangent plots; corresponding values at 100 Hz are listed in Table 2. The Li-Nb-O thin films with low ionic conductivity exhibit a pronounced relaxation peak. In contrast, higher ionic conductivity leads to a suppression of this feature, yielding a nearly flat loss profile with minimal dielectric loss value.
The loss peak delineates the low- and high-frequency regimes, reflecting resonance between the applied electric field and the characteristic dipolar relaxation frequency. At the resonant condition (ωτ = 1), power transfer from the external source to the dipolar system is maximized. This results in peak energy dissipation and the corresponding thermal response. The absence of a loss peak arises from dominance of electrode polarization over dipolar relaxation, known as the masking effect. The observed decrease in the dielectric loss is attributed to the enhanced Li ion mobility. The accelerated response of Li-Li dipoles under applied alternating field leads to a reduced relaxation time. Sharma et al. reported that a zero relaxation time leads to the dielectric loss approaching zero (ɛ'' = 0 for ωτ = 0) .
To investigate the thermally-activated diffusion of lithium ions, impedance spectroscopy was performed over the temperature range of 27-110 °C. Figure 10a, b display the Nyquist plots of Li-Nb-O films deposited at DC powers of 20 W and 30 W, respectively. To extract accurate information from the impedance diagrams, modeling and fitting of Nyquist plots were carried out using ZView software. The corresponding equivalent circuit used to model the system is shown in Fig. 8c. In this circuit, R represents the ohmic resistance of the electrical contacts. Typically, Nyquist plots exhibit a semicircle in the high-frequency region and an inclined line at low frequencies. The high-frequency semicircular arc reflects the bulk response of the films and is modeled by a bulk resistance (R) in parallel with a constant phase element (CPE). At low frequencies, the inclined line is attributed to dominant ionic conduction. This feature is represented in the equivalent circuit by a constant phase element (CPE), which captures stray capacitive behavior arising from ion accumulation at the electrode/electrolyte interface.
The ionic conductivity (σ) of Li-Nb-O films at various temperatures can be calculated from the fitted bulk resistance (R) using the following equation:
Here, L denotes the film thickness, corresponding to the length of conduction pathway, and S represents the contact area between the electrolyte and the electrodes. As shown in Fig. 10c, d, the exponential increase in ionic conductivity with rising temperature follows the Arrhenius relation:
Here, σ denotes the ionic conductivity, σ₀ is the pre-exponential factor, Eₐ represents the activation energy, k is the Boltzmann constant, and T is absolute temperature. The activation energy (Eₐ) can be extracted from the linear fit of log(σ) versus 1/T. The Li-Nb-O films deposited at a DC power of 20 W exhibits an activation energy of 0.53 eV, which is comparable to reported values ranging from 0.4 to 0.65 eV for lithium niobate in previous studies. For films deposited at a DC power of 30 W, the activation energy was determined to be 0.38 eV. The observed decrease in E with increasing DC power is attributed to the substitution of Li ions by Nb, leading to an increased concentration of Li vacancies. These vacancies facilitate the lower-energy transport of Li ions by creating favorable diffusion pathways and reducing Coulombic repulsion between adjacent Li ions.
Figure 11a, b exhibit a noticeable enhancement in capacitance with rising temperature. This is attributed to improved electrode polarization caused by thermally-activated processes. The optimized Li-Nb-O based MIM capacitors (DC: 30 W; RF: 100 W) exhibited a maximum low-frequency capacitance of 1.3 μF cm, corresponding to a dielectric constant of 293 at 110 °C.
Figure 11c, d shows the loss spectra of the Li-Nb-O films at various temperatures. As the temperature increases, the relaxation peak shifts to higher frequencies, concomitant with a reduction in peak intensity. This indicates a reduction in relaxation time due to enhanced ion diffusion.